HIV Dynamics & Evolution

22nd International
HIV Dynamics
& Evolution
Promoting discussion between HIV specialists
May 13-16, 2015
Hungarian Academy of Sciences • Budapest, Hungary
cme.ucsd.edu/hivdynamics
22nd International
HIV Dynamics
& Evolution
Promoting discussion between HIV specialists
May 13-16, 2015
Hungarian Academy of Sciences • Budapest, Hungary
Visit us at
facebook.com/ucsdcme
twitter.com/UCSanDiegoCME
Supporters
We would like to thank the following companies & organizations for their
generous support.
The Hungarian Academy of Sciences
*******************
Merck
*******************
Gilead
*******************
Microsoft
Viruses Journal
PROGRAM Wednesday, May 13, 2015 Page Location: Art’otel 5:00 pm Registration 7:00 pm Welcome Reception/Dinner Thursday, May 14, 2015 Location: Hungarian Academy of Sciences ORIGIN AND GLOBAL EVOLUTION OF THE EPIDEMICS Session Chair: Michael Worobey and Ahidjo Ayouba 9:00 am Welcome and Introductions 9:15 am ADAPTATIONS FACILITATING CROSS‐SPECIES TRANSMISSION AND EMERGENCE OF SIVSM IN RHESUS MACAQUES, *Alison Hill 1 9:35 am CORRELATION BETWEEN HISTORICAL LACK OF MALE CIRCUMCISION AND THE EPIDEMIC EMERGENCE OF HIV‐2, *Joao Sousa 2 9:55 am PHYLOGENETICS AND PHYLOGEOGRAPHY OF HIV‐1 SUBTYPE G ENV REVEAL COMPLEX EVOLUTIONARY PATTERNS AROUND THE EPICENTRE OF THE HIV‐1 EPIDEMIC *Jeffrey R. Dorfman 3
10:15 am ORIGIN OF HIV‐1 GROUP O AND P IN WESTERN LOWLAND GORILLAS *Mirela D'arc 4 10:30 am Break WITHIN‐HOST DIVERSITY AND EVOLUTION #1 Session Chairs: Ron Swanstrom and Richard Neher 11:00 am USING DEEP SEQUENCING TO REVEAL COMPARTMENTALIZATION OF HIV‐1 POPULATIONS WITHIN THE BODY, *Ronald Swanstrom 5 11:20 am SEQUENCE SPACE EXPLORATION AND POPULATION DYNAMICS OF HIV‐1 QUASISPECIES *Richard Neher 6 11:40 am PARALLEL EVOLUTION OF HIV‐1 IN A LONG TERM EXPERIMENT, *Frederic Bertels 7 12:00 pm PHYLOGENETIC RECONSTRUCTION OF VIRAL QUASISPECIES DYNAMICS *Veronika Boskova 8 12:15 pm RECOMBINATION FACILITATES SURVIVAL OF LATENT HIV‐1 LINEAGES IN PRODUCTIVELY INFECTED CELLS, *Taina Immonen 9 12:30 pm Lunch PROGRAM WITHIN‐HOST DIVERSITY AND EVOLUTION #2 Session Chairs: Jan Albert and Grace McCormack 2:00 pm WHOLE‐GENOME DEEP SEQUENCING OF LONGITUDINAL SAMPLES FROM HIV‐1 PATIENTS FOLLOWED FROM EARLY INTO CHRONIC INFECTION, *Jan Albert 2:20 pm TOWARDS EMPIRICALLY‐DERIVED SEQUENCE SPACE AND FITNESS LANDSCAPES OCCUPIED BY RNA VIRUSES, *Marco Vignuzzi 2:40 pm EXTREME HETEROGENEITY IN GENETIC DIVERSITY AND MOLECULAR EVOLUTION FOUND IN CHRONIC HCV INFECTION, *Jayna Raghwani 3:00 pm PACBIO SINGLE MOLECULE REAL TIME SEQUENCING OF THE HCV ENVELOPE DIVERSITY FROM EARLY ACUTE INFECTION TO CHRONICITY, *Cynthia Ho 3:15 pm DYNAMICS OF CTL ESCAPE DURING EARLY ACUTE HIV‐1 INFECTION REVEALED BY DENSE TEMPORAL SAMPLING AND TARGETED DEEP SEQUENCING, *Sivan Leviyang 3:30 pm Break IMMUNE ESCAPE AND VACCINE RESEARCH Session Chairs: James Mullins and Morgane Rolland 3:50 pm NO EVIDENCE OF STRONGER NEUTRALIZING ANTIBODY RESPONSES IN RV144 BREAKTHROUGH VACCINE RECIPIENTS TWO TO THREE YEARS AFTER HIV INFECTION *Morgane Rolland 4:10 pm ASSESSING THE PROPENSITY OF HIV‐1 TO EVOLVE ANTIBODY ESCAPE VARIANTS DURING FREE VIRUS AND CELL‐CELL TRANSMISSION, *Carsten Magnus 4:25 pm THE ROLE OF TRANSMITTED AND DE NOVO CELLULAR IMMUNE ESCAPE IN HIV DISEASE PROGRESSION, *Jonathan Carlson 4:45 pm INFLUENCE OF RECOMBINATION ON ACQUISITION AND REVERSION OF IMMUNE ESCAPE AND COMPENSATORY MUTATIONS IN HIV‐1, *Helen Alexander 5:00 pm THE CONSERVED ELEMENTS (CE) APPROACH TO HIV VACCINE DESIGN, *James Mullins 5:20 pm Posters 7:00 pm Dinner at Hungarian Academy of Sciences Page 10 11 12
13 14 15 16 17 18
19 PROGRAM Friday, May 15, 2015 Page Location: Hungarian Academy of Sciences LATENCY, RESERVOIRS AND CURE Session Chairs: Alan Perelson and Mary Kearney 9:00 am CARD‐SGS REVEALS PERSISTENT EXPRESSION OF HIV‐1 RNA IN CLONALLY EXPANDED HIV‐INFECTED CELLS DURING ART, *Mary Kearney 20 9:20 am PERSISTENT AND PROLIFERATING HIV‐INFECTED CELLS: TISSUE‐DISTRIBUTION AND EXPRESSION OF VIRAL RNA AND PROTEIN, *James Mullins 21 9:40 am MODELING THE EFFECTS OF VORINOSTAT IN VIVO REVEALS BOTH TRANSIENT AND DELAYED ACTIVATION OF HIV TRANSCRIPTION IN LATENTLY INFECTED CELLS *Alan Perelson 22 10:00 am TARGETING CCR5 RECEPTOR USING GENE THERAPY WILL NOT BE SUFFICIENT TO CURE HIV, *Aridaman Pandit 23 10:15 am EFFECT OF THE LATENT RESERVOIR ON THE EVOLUTION OF HIV AT THE WITHIN‐AND BETWEEN‐HOST LEVELS, *Hilje Doekes 24 10:30 am Break WITHIN‐HOST DYNAMICS Session Chairs: Rob De Boer and Sebastian Bonhoeffer 11:00 am HOW RAPIDLY ARE HIV‐1 INFECTED CELLS KILLED?, *Rob De Boer 25 11:20 am HIV CONTROL AFTER TREATMENT CESSATION: MATHEMATICAL MODEL PREDICTIONS *Jessica Conway 26 11:40 am MODELLING ONGOING REPLICATION OF HIV IN DRUG SANCTUARIES, *Helen Fryer 27
12:00 pm A ROBUST SCALING LAW ESTIMATES THE REQUIRED DURATION OF TREATMENT WITH HIV‐1 MUTAGENS, *Vipul Gupta 28 12:15 pm TARGETING HIV LATENCY, *Feng Fu 29 12:30 pm Lunch PHYLOGENETIC INSIGHTS INTO EPIDEMIC DYNAMICS Chairs: Tanja Stadler and Tulio Oliveira 2:00 pm DETECTION OF TRANSMISSION CLUSTERS IN HIV PHYLOGENIES, *Tanja Stadler 30 2:20 pm PHYLOGEOGRAPHIC METHODS GIVE INSIGHT INTO THE EPIDEMIOLOGICAL DYNAMICS BETWEEN HIV‐1 RISK GROUPS, *Denise Kühnert 31 PROGRAM 2:35 pm 2:50 pm BAYESIAN INFERENCE OF TRANSMISSION PATTERNS FROM TIMED PHYLOGENIES *Caroline Colijn NEAR FULL LENGTH HIV‐1 SEQUENCING TO UNDERSTAND HIV PHYLODYNAMICS IN AFRICA IN REAL TIME, *Tulio de Oliveira Page 32 33 3:10 pm USING PHYLOGENETIC TOOLS TO ESTIMATE ASPECTS OF HIV TRANSMISSION DYNAMICS IN GENERALIZED HIV EPIDEMICS: FINDINGS FROM THE PANGEA_HIV METHODS COMPARISON EXERCISE, *Oliver Ratmann 34 3:30 pm Break THERAPY, PREVENTION, AND EPIDEMIOLOGY Session Chairs: Annemie Vandamme and Oliver Laeyendecker 4:00 pm HIV‐1 TRANSMITTED DRUG RESISTANCE: NEW INSIGHTS INTO THE TRANSMISSIBILITY OF SDRMS, *Ana Abecasis 35 4:15 pm THE IMPORTANCE OF PROTEIN STABILITY FOR THE EVOLUTION OF HIV DRUG RESISTANCE, *Abayomi Olabode 36 4:30 pm DIRECT COMPARISON OF SEROLOGIC ASSAYS AND VIRAL DIVERSITY MEASURES FOR IDENTIFYING INDIVIDUALS WITH RECENT HIV INFECTION, *Oliver Laeyendecker 37 4:50 pm NGS COMBINED WITH PHYLOGENETIC ANALYSIS TO DETECT SUPER‐INFECTION IN HIV‐ 1 INFECTED INTRAVENOUS DRUG USERS, *Simona Paraschiv 38 5:10 pm RECONCILING NAMED PARTNER AND GENETIC PARTNER HIV‐1 TRANSMISSION NETWORKS IN NEW YORK CITY, *Joel Wertheim 39 5:30 pm Posters 7:00 pm Dinner Cruise (Sign‐ups required) 10:00 pm Saturday, May 16, 2015 Location: Hungarian Academy of Sciences HIV VIRULENCE AND CONTROL Session Chairs: Andrew Leigh Brown and Jonathan Carlson 9:00 am ESTIMATING THE RESPECTIVE CONTRIBUTIONS OF HUMAN AND VIRAL GENETIC VARIATION TO HIV CONTROL, *István Bartha 40 9:20 am ESTIMATING THE BETWEEN‐HOST HERITABILITY OF VIRAL TRAITS: OLD‐SCHOOL PARENT‐OFFSPRING VERSUS MODERN PHYLOGENETICS, *Gabriel Leventhal 41 9:40 am DETERMINING THE VIRAL GENETIC BASIS OF HIV VIRULENCE USING WHOLE GENOME SEQUENCING, *Chris Wymant 42 PROGRAM Page 10:00 am IMPORTANCE OF TRAIT SELECTION IN ESTIMATING VIRAL CONTRIBUTION TO VIRULENCE IN HIV INFECTION, *Venelin Mitov 10:15 am MODELING HIV VIRULENCE EVOLUTION IN THE CONTEXT OF IMMUNE ESCAPE *C. H. van Dorp 10:30 am Break BIOINFORMATICS SOFTWARE AND ALGORITHMS FOR HIV/VIRAL RESEARCH Session Chairs: Sergei Kosakovsky Pond and Erik Volz 11:00 am FULL‐LENGTH ENVELOPE ANALYSIS (FLEA): A PIPELINE AND WEB SERVICE FOR FULL‐ LENGTH ENV SEQUENCE ALIGNMENT, ANALYSIS, AND VISUALIZATION, *Kemal Eren 11:15 am SELECTION OF ANTIGEN SWARMS FROM NEUTRALIZING ANTIBODY DEVELOPMENT AGAINST HIV‐1, *Peter Hraber 11:30 am CLUSTER RECONSTRUCTION IN SIMULATED TRANSMISSION CHAINS *Manon Ragonnet‐Cronin 11:45 am PHYLODYNAMIC INFERENCE OF CONTACT NETWORK STRUCTURE, *David Rasmussen 12:00 pm QUANTITATIVE METHOD FOR THE DELINEATION OF HIV‐1 SPECIFIC ANTIBODY EPITOPE REACTIVITY IN POLYCLONAL PATIENT SERA, *Nathanael Hoze 12:15 pm Final Adjournment 43 44 45 46 47 48 49 POSTER ABSTRACTS 1 STRUCTURED OBSERVATIONS REVEAL SLOW HIV‐1 CTL ESCAPE, *H. Roberts Page 53 2 NINE YEARS EXPERIENCE OF HIV‐ DRUG RESISTANCE MONITORING IN SOUTH AFRICA, 2006–2014 *G. Jacobs 54 3 THE RETROVIRUS INTEGRATION DATABASE (RID), *W. Shao 55 4 A SEQUENTIAL MONTE CARLO APPROACH FOR JOINT ESTIMATION OF VIRAL PHYLOGENY AND
EPIDEMIOLOGICAL DYNAMICS, *A. Smith 56 5 THERAPY STRATEGIES IN HIV INFECTION USING MATHEMATICAL MODELLING,*A. Boianelli 57 6 PREDICTED BINDING AFFINITIES OF HIV‐1C VIF, VPR AND VPU EPITOPES DURING PRIMARY INFECTION 58 AGAINST HOST MHC CLASS I HLA‐A AND HLA‐B MOLECULES, *R. Rossenkhan 7 SUBTYPE‐SPECIFIC STRUCTURAL CHARACTERISTICS AND MOLECULAR DYNAMICS OF GLYCOSYLATED HIV‐1 GP120 PROTEINS, *N. Wood 59 8 HIV‐1 GROUP M DIVERSITY: NEW INSIGHTS ON THE EVOLUTION OF THE VIRUS, *M. Tongo 60 9 EVOLUTIONARY RATES OF HIV‐1 ACCESSORY GENES FROM FULL‐LENGTH DATASETS ACROSS SUBTYPES *G. Yebra 61 10 THE CONTRIBUTION OF ANGOLA FOR THE EARLY SPREAD OF THE HIV‐1 EPIDEMIC, *A. Pineda‐Peña 62 11 COMPOSITE SEQUENCE‐STRUCTURE STABILITY MODELS AS SCREENING TOOLS FOR IDENTIFYING MUTATIONAL TARGETS FOR HIV DRUG AND VACCINE DEVELOPMENT, *S. Manocheewa 63 12 LONG‐RANGE HIV GENOTYPING USING PROVIRAL DNA FOR ANALYSIS OF HIV DRUG RESISTANCE AND HIV TRANSMISSION DYNAMICS, *V. Novitsky 64 13 MODEST IMPROVEMENTS TO HIV TREATMENT AND CARE COULD PREVENT HALF OF ALL NEW HIV INFECTIONS AMONG MEN HAVING SEX WITH MEN: A PHYLOGENETIC STUDY, *O. Ratmann 65 14 WILL HIV VANISH OR EVOLVE RESISTANCE FACING MASS ANTI‐RETROVIRAL TREATMENT? *S. Alizon 66 15 ACCURATE DETECTION OF MINOR MUTATIONS WITH DEFINED SAMPLING DEPTH AND ERROR CUTOFF USING NEXT GENERATION SEQUENCING, *S.Zhou 67 16 LINKAGE OF FEMALE FOUNDER HIV‐1 POPULATIONS TO TRANSMITTING MALE’S BLOOD AND SEMINAL VIRUSES, *C. Williams 68 17 COMPARISON OF MAJOR AND MINOR VIRAL SNPS IDENTIFIED BY SINGLE TEMPLATE SANGER AND PYROSEQUENCING IN EARLY HIV‐1 INFECTION, *J. Mullins 69 18 FITNESS‐BALANCED ESCAPE DETERMINES RESOLUTION OF DYNAMIC FOUNDER VIRUS ESCAPE PROCESSESS IN HIV‐1 INFECTION, *J. Mullins 70 19 DEEP SEQUENCING ANALYSIS OF HIV‐1 TRANSMISSION AND SUBSEQUENT EVOLUTION IN SIX TRANSMISSION PAIRS, *J. Mullins 71 POSTER ABSTRACTS 20 FINE STRUCTURE GENETIC ANALYSIS OF INTRA‐PATIENT HIV POPULATIONS PRIOR TO ANTIRETROVIRAL THERAPY USING NEXT GENERATION SEQUENCING, *J. Hattori 72 21 WEBMUTCOR: A TOOL TO DISCOVER INTERACTIONS BETWEEN DRUG RESISTANCE MUTATIONS AND CYTOTOXIC‐T‐LYMPHOCYTE ESCAPE MUTATIONS IN THE HIV‐1 POL REGION, *W. Smidt 73 22 COMPARING ESTIMATED TIME OF HIV‐1 INFECTION OBTAINED BY BAYESIAN ANALYSIS AND THE BED ASSAY, *M. M. Lunar 74 23 EXPLORING TRANSMISSION DYNAMICS OF HIV IN RURAL KWAZULU‐NATAL, USING PHYLOGENETICS *T. de Oliveira 75 24 INTERTYPE AND INTER SPECIES GENOMIC DIVERSITY OF HIV‐1 TAT, *I. Khandaker 76 25 SPATIO‐TEMPORAL EVOLUTION OF HIV‐1 TAT IN SUBTYPES B AND C, *C. Roy 77 26 DENTIFYING WITHIN‐HOST HIV‐1 SUBPOPULATIONS BY COSEGREGATION OF PROFILE HIDDEN MARKOV MODEL UPDATE VECTORS, *P. Edlefsen 78 27 POPULATION‐LEVEL EVOLUTION OF HIV‐1 ENV STRUCTURE OVER THREE DECADES; A MULTIDIMENSIONAL APPROACH TO STUDY THE DYNAMICS OF COMPLEX PHENOTYPES, *H. Haim 79 28 REASSESMENT OF MOLECULAR EVOLUTION OF CRIMEAN–CONGO HEMORRHAGIC FEVER VIRUS BASED ON COMPLETE S AND M SEGMENT SEQUENCES, *M. Stanojevic 80 29 THE CHANGING FACE OF THE HIV‐1 SUBTYPE C EPIDEMIC IN SOUTH AFRICA – DETECTION OF UNIQUE SUBTYPES AND RECOMBINANT FORMS, *S. Engelbrecht 81 ADAPTATIONS FACILITATING CROSS‐SPECIES TRANSMISSION AND EMERGENCE OF SIVSM IN RHESUS MACAQUES *A. Hill (1), S. Ita (2), M. Mangano (3), V. Hirsch (4), R. Desrosiers (5), D. Evans (6), R. Newman (7) 1. Harvard University/Boston College 2. Harvard University/Boston College 3. Boston College 4. NIAID 5. University of Miami 6. University of Wisconsin 7. Broad Institute The distribution of lentiviruses among primates reflects a history of interspecies transmission and emergence of new virus‐host relationships. An understudied element of emergence is the degree to which viruses must adapt to the genetic environment of new host species, and how adaptations to the new host initially affect viral fitness. SIVmac emerged as the result of a cross‐species transmission of SIVsm into rhesus macaques, and comparing cohorts of SIVmac‐ and SIVsm‐infected macaques provides an opportunity to examine a lentivirus evolving during the early stages of emergence. Using archived samples from four cohorts of macaques, we compared evolution of macaque‐adapted viruses (SIVmac239, SIVmac251) to incompletely‐adapted, “emerging” viruses (SIVsmE543‐3, SIVsmE660). Longitudinal samples included the inoculum for each cohort, as well as acute and chronic time‐points for each animal. Samples were processed for deep sequencing, and consensus sequences of complete viral coding regions were assembled de novo. Based on comparison of viral populations in all four cohorts, we identified nine candidate species‐specific changes in six viral genes, reflecting adaptation of SIVsm to the rhesus macaque host; these included known adaptations to overcome restriction by macaque TRIM5α. Each of the candidate adaptations was introduced singly or in combinations into SIVsmE543‐3 (forward mutations, reflecting adaptation to the macaque host) and SIVmac239 (reversions to the ancestral residue). These were then tested in a multiplex, deep‐sequencing based fitness assay, in which changes in the frequencies of mutant and parental sequences (among sequencing reads) were used to calculate differences in relative fitness. The assay was validated using a known escape mutation to rhesus TRIM5α, which demonstrated a clear fitness advantage over the wild‐type SIVsmE543‐3 in a rhesus T cell line expressing restrictive TRIM5 alleles. Together, these studies represent a novel approach to the identification and characterization of viral determinants of cross‐species transmission. 1 CORRELATION BETWEEN HISTORICAL LACK OF MALE CIRCUMCISION AND THE EPIDEMIC EMERGENCE OF HIV‐2 *J. Sousa (1), M. Temudo (2), V. Müller (3), A. Vandamme (4) 1. KU Leuven ‐ University of Leuven, Department of Microbiology and Immunology, Rega Institute for Medical Research, Clinical and Epidemiological Virology, B‐3000 Leuven, Belgium; Centro de Malária e outras Doenças Tropicais and Unidade de Microbiologia Médica, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisboa, Portugal; Email: [email protected] Tel: +3216332160. Fax: +3216332131. Miderbroedersstraat 10, B‐3000 Leuven, Belgium. 2. Department of Agriculture, Environment and Development, Tropical Research Institute (IICT), Lisboa, Portugal 3. Institute of Biology, Eötvös Loránd University, Budapest, Hungary; Parmenides Center for the Conceptual Foundations of Science, Pullach/Munich, Germany 4. KU Leuven ‐ University of Leuven, Department of Microbiology and Immunology, Rega Institute for Medical Research, Clinical and Epidemiological Virology, B‐3000 Leuven, Belgium; Centro de Malária e outras Doenças Tropicais and Unidade de Microbiologia Médica, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisboa, Portugal Background: Epidemic HIV‐2 emerged in humans twice (groups A and B), both circa 1930. Its closest ancestors are SIVsmm infecting sooty mangabeys from southwest Ivory Coast. The earliest serological surveys of HIV‐2 in West Africa (1987–91) show a patchy spread with only Ivory Coast and Guinea‐Bissau having mature epidemics by then. Methods: We estimated past male circumcision rates of 209 West African ethnic groups, based on an ethnographic literature review. Uncertainty was incorporated by defining plausible ranges of parameters (e.g. timing of introduction, proportion circumcised). We generated 1000 sets of past circumcision rates per city using Latin Hypercube Sampling with different parameter combinations. We explored the correlation between early HIV‐2 prevalence and estimated circumcision rate (both logit‐transformed). A credible confidence interval for the correlation was constructed using 500 bootstrap replicates. Results: Demographic and Health Surveys show that male circumcision is nowadays almost universal throughout the region. However, our ethnographic review reveals that, in early 20th century,many ethnic groups did not practice it, or did it late in life. Our estimates for circumcision rates in1950 are, for Bissau: 80–85%; Abidjan: 50–57%; Bouaké: 56–63%; Monrovia: 79–84%; the major cities of Senegal, Gambia, Guinea, and Sierra Leone had >90%. HIV‐2 prevalence in 1987–91 and circumcision rate in 1950 were correlated (r=‐0.463; 95% CI: ‐0.726–‐0.213; p=0.017). In addition, southwestern Ivory Coast and neighboring Liberian areas were inhabited by non‐circumcising ethnic groups (in 1930) heavily exposed to SIVsmm through bushmeat. Conclusions: The differential HIV‐2 spread between West African countries correlates with different historical circumcision rates in their main cities. We suggest HIV‐2 only formed early substantial foci in low circumcision rate cities. Lack of male circumcision in rural areas exposed to bushmeat may have had a role in boosting the chances of successful SIV zoonosis. 2 PHYLOGENETICS AND PHYLOGEOGRAPHY OF HIV‐1 SUBTYPE G ENV REVEAL COMPLEX EVOLUTIONARY PATTERNS AROUND THE EPICENTRE OF THE HIV‐1 EPIDEMIC M. Tongo (1), R. Essomba (2), F. Nindo (3), F. Abrahams (4), A. Nanfack (5), J. Fokam (6), D. Takou (7), J. Torimiro (8), E. Mpoudi‐Ngole (9), W. Burgers (10), D. Martin (11), *J. Dorfman (12) 1. ICGEB and Univ of Cape Town Div of Immunology 2. ICGEB and Univ of Cape Town Div of Immunology 3. Univ of Cape Town Computational Biology Group 4. ICGEB 5. CIRCB, Yaoundé, Cameroon 6. CIRCB, Yaoundé, Cameroon 7. CIRCB, Yaoundé, Cameroon 8. CIRCB, Yaoundé, Cameroon 9. Institute of Medical Research and Study of Medicinal plants, Yaoundé, Cameroon 10. Univ of Cape Town, Div of Medical Virology 11. Univ of Cape Town, Computational Biology Group 12. ICGEB and Univ of Cape Town Div of Immunology The HIV‐1 subtype G lineage has played a central role in the evolutionary complexity of the HIV‐1M epidemic in central/west Africa. Although originally considered non‐recombinant, it has been proposed that the ancestral subtype G progenitor was the recombinant progeny of a virus related to circulating recombinant form (CRF) 02_AG and an uncharacterised HIV‐1M lineage, with most of its envelope gene (env) derived from the uncharacterised parental lineage. Using Bayesian and maximum likelihood methods, we analysed new subtype G env sequences sampled from 8 individuals in Yaoundé, Cameroon during 2007‐2010, together with all publically available full‐length env sequences that had known sampling dates and locations, and were identified as subtype G. We inferred from our analysis that the most recent common ancestor (MRCA) of the analysed subtype G env sequences most likely occurred in Nigeria in ~1962 (HPD interval 1954.6‐1970.4), nominally older than previous estimates. Furthermore, the analysis showed that subtype G env phylogeny has a complex structure including seven distinct lineages, each likely dating back to the 1970s. We detected multiple introductions of subtype G env into Spain/Portugal, but only a single introduction into Cuba. CRF06_cpx env sequences clustered only within the subtype G clade, while CRF25_cpx env sequences were found at the root outside the subtype G clade. This suggests that the CRF06_cpx env could plausibly have been derived through recombination from a subtype G parent, while the CRF25_cpx env was likely derived from an unsampled and possibly extinct HIV‐1M lineage related to the subtype G MRCA. Overall, our analysis involves the most diverse assemblage of HIV‐1M subtype G env sequences yet studied and sheds substantial light on the evolutionary and spatio‐temporal origins of subtype G env sequences that evolved at the epicentre of the HIV‐1 epidemic, and are currently circulating in many different parts of the world. 3 ORIGIN OF HIV‐1 GROUP O AND P IN WESTERN LOWLAND GORILLAS *D. Mirela (1), A. Ayouba (2), A. Esteban (3), G. Learn (4), V. Boué (5), F. Liegeois (6), L. Etienne (7), N. Tagg (8), F. Leendertz (9), C. Boesch (10), N. Madinda (11), M. Robbins (12), M. Gray (13), A. Cournil (14), M. Ooms (15), M. Letko (16), V. Simon (17), P. Sharp (18), B. Hahn (19), E. Delaporte (20), E. Mpoudi Ngole (21), M. Peeters (22) 1. UMI 233, Institut de Recherche pour le Développement (IRD) and University of Montpellier 1, 34394 Montpellier, France; Laboratory of Human Virology, Universidade Federal do Rio de Janeiro, 21949‐570 Rio de Janeiro, Brazil HIV‐1 groups M, N, O and P, each resulted from an independent cross‐species transmission event of SIVs infecting African apes. While groups M and N have been traced to distinct chimpanzee communities in south‐central and south‐eastern Cameroon, the ape reservoirs of groups O and P remained unknown.Fecal samples from western lowland gorillas (n=2,611) in southern Cameroon and northern Gabon, eastern lowland gorillas (n=103) in the Democratic Republic of Congo (DRC) and mountain gorillas (n=218) from Uganda and DRC were screened for SIVgor specific antibodies and nucleic acids. SIVgor was only identified at four sites in southern Cameroon with prevalence rates ranging from 0.8% to 22%. Partial and full‐length SIVgor sequences revealed extensive genetic diversity and indicated that a single chimpanzee‐to‐gorilla transmission is at the origin of the SIVgor lineage. Full‐length genome sequences identified two new SIVgor strains from south‐west Cameroon closely related to HIV‐1 P across the entire genome and one SIVgor strain from central Cameroon was very closely related to HIV‐1 group O across most of its genome. Functional analyses identified APOBEC3G as a likely barrier for chimpanzee‐to‐gorilla, but not gorilla‐to‐human, virus transmission. SIVgor is thus at the origin of HIV‐1 groups O and P in humans. Group P has only been detected in two individuals, but group O has spread extensively throughout west central Africa and is estimated to have infected around 100,000 people. Thus, both chimpanzees and gorillas harbor viruses that are capable of crossing the species barrier to humans and causing major disease outbreaks. 4 USING DEEP SEQUENCING TO REVEAL COMPARTMENTALIZATION OF HIV‐1 POPULATIONS WITHIN THE BODY S. Joseph (1), S. Zhou (2), *R. Swanstrom (3) UNC Chapel Hill Deep sequencing has the potential to identify minor variants within a population. However, it can also be used to identify minor variants between populations. We have developed an amplicon spanning the V1 to V3 region of the HIV‐1 env gene and used it for paired‐end sequencing with the MiSeq platform. Paired‐end reads allow reading of V1 and V2 and the linked C2 and V3 region, although the reads do not overlap within C2 leaving a small gap. We have also used the Primer ID template tag to enumerate the exact number of viral genomes sampled, and by accounting for PCR resampling we are able to build template consensus sequences that reduce the overall error rate to 1 in 10,000 nucleotides (essentially the error of RT in the first cDNA synthesis step). We have used this approach to compare viral populations in the blood and CSF from subjects across a wide range of disease states. Because of its enhanced sampling the deep sequencing approach has greater sensitivity of detecting minor variants compared to single genome amplification (SGA) in head‐to‐head comparisons. Compartmentalized variants are readily detected in a significant number of subjects across disease states. Thus deep sequencing provides a new look at the compartmentalization of viral populations, identifying independently replicating pools of virus, with a dramatically increased level of sensitivity. 5 SEQUENCE SPACE EXPLORATION AND POPULATION DYNAMICS OF HIV‐1 QUASISPECIES F. Zanini (1), J. Albert (2), *R. Neher (3) 1. MPI for Developmental Biology, Tuebingen, Germany 2. Department of Microbiology, Tumor and Cell Biology, Karolinska Institutet, Stockholm, Sweden 3. MPI for Developmental Biology, Tuebingen, Germany During an untreated chronic infection, HIV‐1 accumulates many mutations and expands into a genetically diverse population. Using longitudinal deep sequencing data on 11 untreated patients, we characterized the genome wide dynamics and composition of HIV‐1 quasispecies. We show that HIV‐1 populations explore sequence space in a stereotypic and predictable manner. By analyzing the accumulation of minor variants, we link conservation in cross‐sectional sequence alignments to fitness costs of mutations within patients. Through extrapolation of mutation accumulation from conserved to neutral sites, we estimate the in vivo mutation rate matrix of HIV‐1 and find close agreement with previous in vitro measurements (Abram et al. 2010). While low frequency variation is similar among patients, substitutions and adaptation are mostly patient specific. We further show that the dynamics of single nucleotide variants is dominated by hitch‐hiking and linked selection. Variants closer than 300 bp are not efficiently shuffled by recombination and influence each other. Standard neutral coalescent models are inconsistent with the observed diversity and dynamics. Alternative null models builton the premise of abundant fitness variation, however, describe the data well. 6 PARALLEL EVOLUTION OF HIV‐1 IN A LONG TERM EXPERIMENT *F. Bertels (1), C. Leemann (2), K. Metzner (3), R. Regoes (4) 1. Department for Environmental Systems Sciences, ETH Zurich, Zurich 2. Division of Infectious Diseases and Hospital Epidemiology, University Hospital Zurich and Insitute of Medical Virology, University of Zurich, Zurich 3. Division of Infectious Diseases and Hospital Epidemiology, University Hospital Zurich and Insitute of Medical Virology, University of Zurich, Zurich 4. Department for Environmental Systems Sciences, ETH Zurich, Zurich Observing the evolution of organisms in a constant environment over long periods of time has been a successful strategy to investigate fundamental evolutionary questions. A particularly interesting question is to which extent evolution is predictable or repeatable. This has been studied for various bacteria and viruses. For HIV‐1 parallel evolution has been observed under drug treatment in multiple studies, but the repeatability of HIV‐1 evolution in the absence of drugs has not been studied systematically yet. To investigate HIV‐1 parallel evolution in a tightly controlled experiment, we passaged HIV‐1 NL4‐3 for almost one year in two different human T‐cell lines (MT2 and MT4) in two replicates each. For each of the four replicate lines, we sequenced the entire HIV‐1 genome at five longitudinal time points by next‐generation sequencing to track the evolution of HIV‐1. Over the course of our experiment we observed an astonishing amount of parallel evolution: about 50% of the identified mutations emerged in more than one population. Additionally when we attempted to reconstruct the phylogenetic tree from the consensus sequences, the genealogy could not be recovered. Instead the sequences clustered by the environment they evolved in. This pronounced parallelism confounds any inference drawn from viral sequences that is based on the assumption of neutral evolution. This has far reaching consequences especially for the application of phylogenetic methods. In particular this makes it very challenging to correctly identify transmission pairs from HIV‐1 phylogenies for the extent of parallel evolution we observe in our experiment. 7 PHYLOGENETIC RECONSTRUCTION OF VIRAL QUASISPECIES DYNAMICS *V. Boskova (1), T. Stadler (2) Department of Biosystems Science and Engineering, ETH Zurich, Basel, Switzerland; SIB Swiss Institute of Bioinformatics, Basel, Switzerland Especially in fast evolving and reproducing populations such as RNA viruses, the population of sequences present in one host at a time is often very rich but also very repetitive. Deep‐sequencing approaches allow for quantification of sequences and their diversity. The amount of sequences from such sequencing efforts represents a computational overload for current phylodynamic and phylogenetic model implementations in full Bayesian framework. Heuristic approaches aimed at reducing the computational burden apply the inference models only to the unique sequences, i.e. ignoring frequencies of the different sequences and instead assuming each one occurs only once, or to only a random subsample of the full dataset. We set out to investigate these heuristics in terms of how much loss of information on dynamic properties of the process occurs. Based on the identified drawbacks of the heuristics, we propose a new tool for efficient reconstruction of viral epidemiological and evolutionary dynamics from full quasispecies datasets. The framework involves first reconstructing the haplotypes and their frequencies within the quasispecies from the raw reads. Phylogenetic analyses are then performed on the haplotype alignment and the frequency information. Use of such complete datasets should lead to a more complete picture of pathogen dynamics, insight into transmission bottlenecks, and in more reliable parameter estimation. 8 RECOMBINATION FACILITATES SURVIVAL OF LATENT HIV‐1 LINEAGES IN PRODUCTIVELY INFECTED CELLS *T. Immonen (1), J. Conway (2), E. Romero‐Severson (3), A. Perelson (4), T. Leitner (5) 1. Theoretical Biology and Biophysics, Los Alamos National Laboratory, Los Alamos, New Mexico, United States of America 2. Department of Mathematics, Pennsylvania State University, University Park, Pennsylvania, United States of America 3. Theoretical Biology and Biophysics, Los Alamos National Laboratory, Los Alamos, New Mexico, United States of America 4. Theoretical Biology and Biophysics, Los Alamos National Laboratory, Los Alamos, New Mexico, United States of America 5. Theoretical Biology and Biophysics, Los Alamos National Laboratory, Los Alamos, New Mexico, United States of America HIV‐1 is subject to immune pressure exerted by the host, giving variants that escape the immune response an advantage. Virus released from activated latent cells compete against variants that have continually evolved and adapted to host immune pressure. Nevertheless, there is increasing evidence that latent virus lineages survive in patient plasma despite their reduced fitness due to long‐term immune memory. We investigated the survival of latent lineages by simulating within‐host HIV‐1 sequence evolution and the cycling of viral lineages in and out of the latent reservoir. Our model incorporates a detailed mutation process including nucleotide substitutions, recombination, latent reservoir dynamics, diversifying selection pressure driven by the immune response, and purifying selection pressure due to the accumulation of deleterious mutations. We evaluated the ability of our model to capture sequence evolution in vivo by comparing our simulated sequences to HIV‐1 sequence data from 16 HIV‐infected untreated patients. Empirical sequence divergence and diversity measures were qualitatively and quantitatively similar to those of our simulated HIV‐1 populations, suggesting that our model invokes realistic trends of HIV‐1 genetic evolution. Moreover, reconstructed phylogenies of simulated and patient HIV‐1 populations showed similar topological structures. We found that the key mechanism allowing latent lineages to survive in the productively infected cell population was recombination with already successfully replicating virus. Recombination increased the survival probability of latent lineages approximately 20‐fold. Prevalence of latent lineages in the productively infected cells was observed in only 2% of simulations when we ignored recombination, while the proportion increased to 38% of simulations when we allowed recombination. Comparing simulations with low and high prevalence of latent forms in productively infected cells further showed that latency reduced sequence divergence and increased diversity. * Tel: 505‐606‐0074; Fax: 505‐665‐3493; E‐mail: [email protected] 9 WHOLE‐GENOME DEEP SEQUENCING OF LONGITUDINAL SAMPLES FROM HIV‐1 PATIENTS FOLLOWED FROM EARLY INTO CHRONIC INFECTION F. Zanini (1), J. Brodin (2), L. Thebo (3), C. Lanz (4), R. Neher (5),*J. Albert (6) 1. Max Plank Institute for Developmental Biology, Tuebingen, Germany 2. Department of Microbiology, Tumor and Cell Biology, Karolinska Institute, Stockholm, Sweden 3. Department of Microbiology, Tumor and Cell Biology, Karolinska Institute, Stockholm, Sweden 4. Max Plank Institute for Developmental Biology, Tuebingen, Germany 5. Max Plank Institute for Developmental Biology, Tuebingen, Germany 6. Department of Microbiology, Tumor and Cell Biology, Karolinska Institute, Stockholm, Sweden We extracted viral RNA from frozen serum samples from 11 untreated HIV patients, with approximately 8 time points per patient, spanning at least five years from an established time of infection. We developed a protocol for whole‐genome sequencing using six overlapping primer sets. Primers were designed to ensure unbiased and sensitive amplification of unknown HIV templates representing all major subtypes. We optimized the library preparation protocol to obtain long inserts starting from less than 1 ng of DNA and sequenced all samples on the Illumina MiSeq platform, obtaining around 10,000x coverage. We are able to call minor alleles as rare as 0.2% in the viral population and preserve linkage information over 500 bp. This deep whole genome data set covering many years in 11 patients provides a comprehensive portrait of HIV‐1 intra‐patient evolution that should be suitable for many different types of analyses. We are planning to release the dataset soon and will in addition to raw reads provide pre‐analyzed data such as consensus sequences, minor single nucleotide variant frequencies and tools to extract haplotypes up to 500 bp in length in most regions of the HIV‐1 genome. 10 TOWARDS EMPIRICALLY‐DERIVED SEQUENCE SPACE AND FITNESS LANDSCAPES OCCUPIED BY RNA VIRUSES G. Moratorio (1), A. Bordería (2), R. Henningsson (3), H. Blanc (4), M. Fontes (5), *M. Vignuzzi (6) 1. Institut Pasteur, Viral Populations and Pathogenesis, Paris, France 2. Institut Pasteur, Viral Populations and Pathogenesis, Paris, France 3. Institut Pasteur, Viral Populations and Pathogenesis, Paris, France 4. Institut Pasteur, Viral Populations and Pathogenesis, Paris, France 5. Department of Mathematical Science, Lund University, Sweden and Institut Pasteur, Viral Populations and Pathogenesis, Paris, France 6. Institut Pasteur, Viral Populations and Pathogenesis, Paris, France The very high dimensionality of the sequence space explored by even simple organisms makes it challenging to study and understand, and further complicates generating empirical fitness landscapes occupied by them. Due to their high mutation rates, RNA viruses rapidly generate swarms of closely related genotypes as they attempt to explore sequence space, and constitute excellent models to study populations evolve in fitness landscapes. We are attempting to generate empirically‐derived fitness landscapes by altering the sequence space of a model RNA virus, Coxsackie virus B3, and examining its position and evolution on fitness peaks. Based on the potential effect of point mutations, the six synonymous codons for Leucine and for Serine were classified into three groups (‘1 to Stop’, only one point mutation required to introduce a stop codon; ‘More‐V’, more volatile, one mutation changes the amino‐acid to one with different properties, and ‘Less‐V’, less volatile, higher likelihood to be a silent mutation or to maintain close physico‐chemical properties). By reverse genetics, we replaced all 100 Leu/Ser codons within the P1 region (structural protein) by one of these three groups to generate Synthetic Synonymous (SynSyn) viruses ‐ to present the same amino acid sequence but different nucleotide sequences. The evolvability of the SynSyn viruses were evaluated by deep sequencing tissue culture samples passaged under normal and mutagenic conditions, as well as in a mouse model. Highly quantitative fitness values were also obtained for each sample. We show how a combination of biological knowledge and mathematical tools (PCA, Isomap, etc.) can unveil structure in sequence space and provide a lower‐dimensional representation preserving structure while removing noise. This low‐dimensional representation is useful for visualization, but also as a basis for further analysis. Incorporating fitness values attained from competition assays, we reconstruct empirical fitness landscapes by means of interpolation. The validity of the empirical fitness landscape model is investigated experimentally by predicting the fitness of a sample, given its position in sequence space. Additionally, we show that SynSyn viruses with different 'starting points' in sequence space on the same fitness landscape undertake different evolutionary trajectories. Our data constitutes a first step in accurately characterizing virus evolution in terms of sequence space and fitness landscapes. 11 EXTREME HETEROGENEITY IN GENETIC DIVERSITY AND MOLECULAR EVOLUTION FOUND IN CHRONIC HCV INFECTION *J. Raghwani (1), R. Rose (2), I. Sheridan (3), P. Lemey (4), M. Suchard (5), P. Farci (6), P. Klenerman (7), O. Pybus (8) 1. Department of Zoology, University of Oxford, UK 2. BioInfoExperts, Thibodeaux, LA, USA 3. Peter Medawar Building for Pathogen Research, University of Oxford, Oxford, UK 4. Rega Institute, KU Leuven – University of Leuven, Leuven, Belgium 5. Departments of Biomathematics, Biostatistics, Human Genetics, University of California, Los Angeles, CA 90095, USA 6. Laboratory of Infectious Diseases, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD 20892, USA 7. Peter Medawar Building for Pathogen Research, University of Oxford, Oxford, UK 8. Department of Zoology, University of Oxford, Oxford, UK In contrast to most fast‐evolving RNA viruses, the hepatitis C virus (HCV) can cause both acute and chronic infection in humans, with viral clearance occurring in 15 to 20% of cases. However, in the majority of HCV patients that become chronically infected, the liver is expected to develop cirrhosis, cancer, and other related diseases. In recent years there has been great progress made in treatment of HCV, with the newly approved direct‐acting antiviral drugs being reported to clear the virus successfully in 70% of patients. However, despite the clinical and scientific achievements, our understanding of HCV replication behaviour and thus of the within‐host evolution is limited, especially compared to HIV. In this study, we undertake the first comprehensive analysis of HCV evolutionary dynamics during chronic infection by quantifying viral diversity and divergence through time. We investigate more than 4000 viral gene sequences obtained from 15 HCV patients, which have been sampled longitudinally over several years. These results are compared to those of 9 well‐studied HIV subjects, which indicate key differences in these two chronically infectious RNA viruses. Notably, a significant degree of heterogeneity is observed in the molecular evolution and population genetic diversity in HCV, both among patients and over time, strongly suggesting the presence of complex replication dynamics. As a consequence, we suggest a novel mechanism by which HCV establishes chronic infection that can explain apparent paradoxes in the natural history of this virus. 12 PACBIO SINGLE MOLECULE REAL TIME SEQUENCING OF THE HCV ENVELOPE DIVERSITY FROM EARLY ACUTE INFECTION TO CHRONICITY C. Ho (1), J. Raghwani (2), S. Koekkoek (3), M. de Jong (4), O. Pybus (5), R. Molenkamp (6), J. Schinkel (7) 1. AMC 2. University of Oxford 3. AMC 4. AMC 5. University of Oxford 6. AMC 7. AMC Deep sequencing has revolutionized the study of heterogeneous RNA virus populations, but for phylogenetic studies longer sequence length and low error rates are desired. The Pacific Biosciences Single Molecule, Real Time (SMRT) sequencing provides long reads and circular consensus sequences (CCS) improves accuracy. We investigated the hepatitis C virus (HCV) envelope (E1E2, 1680 bp) evolution in five subjects with incident infection who progressed to chronicity using Pacbio sequencing. The five subjects were infected with closely related HCV genotype 4d variants and coinfected with HIV‐1. Four subjects were men who have sex with men (MSM) and the 5th subject was the female partner of one of the MSM. Fifty samples, collected between 2001‐2013, were SMRT sequenced. The sequencing error at 7 CCS full passes was 0.37% with insertions as the main type of error (0.24%), followed by deletions (0.11%). Mismatches were surprisingly low (0.02%). The median coverage at 7 full passes was 612 CCS reads/sample (range 149‐935). Prior to phylogenetic analysis, insertions with respect to a sample specific reference sequence were removed. Neighbor Joining phylogenies revealed a close phylogenetic relationship between the four MSM at early time points, and the transmission event from one MSM to the female subject. Intra‐host phylogenies of reads sampled early during infection imply that a single founder virus established infection in all five subjects This finding was supported by the low genetic diversity observed at these early time points. The increase in diversity coincided with progression to chronicity and the emergence of multiple co‐existing lineages. Changes in the genetic diversity during chronic infection corresponded with a non‐ladder like phylogeny. SMRT sequencing is able to combine great coverage with long reads and can provide rich insights into HCV dynamics from transmission bottlenecks to long‐term chronic infection. 13 DYNAMICS OF CTL ESCAPE DURING EARLY ACUTE HIV‐1 INFECTION REVEALED BY DENSE TEMPORAL SAMPLING AND TARGETED DEEP SEQUENCING *S. Leviyang (1), G. Kijak (2), E. Sanders‐Buell (3), M. Eller (4), N. Goonetilleke (5), M. Rolland (6), M. (7), R. Thomas (8), E. Harbolick (9), M. Bose (10), P. Pham (11), C. Oropeza (12), K. Poltavee (13), A. O’Sullivan (14), R. Ribeiro (15), A. Perelson (16), G. Shaw (17), L. Eller (18), R. O’Connell (19), N. Michael (20), M. Robb (21), S. Tovanabutra (22), J. Kim (23) 1. Georgetown University Background HIV‐1 evolution associated with CTL pressure has been typically observed after peak viral load (VL), however, sampling limitations of CTL response and viral escape may confound our understanding of this complex interaction. Earlier and deeper characterization of viral sequences at locations targeted by CTL may clarify the factors shaping peak VL and viral decline, with implications for pathogenesis and vaccine design. Methods Using data from the RV217 cohort that include twice‐weekly sampling starting 4‐8 days prior to peak VL and continuing past VL downslope, and HIV‐1 targeted deep sequencing, we examined initial viral diversity at multiple time points in regions targeted by CTL. This data was used to reconstruct the dynamics of CTL escape and changes in the growth rates of wild type and CTL‐escape variants. Results We found that CTL escape can emerge at peak VL, with CTL‐mediated selection initiating ~1 week prior to peak VL. We measured CTL escape rates of ~0.9 day‐1, exceeding current estimates of ~0.4 day‐1 (Goonetilleke,2009), and supporting a potent early CTL response. We observed differences in the slope of VL decline depending on the presence or absence of CTL escape, suggesting that CTL escape contributes to the VL profile. The growth rates of CTL‐escape variants were relatively high, ~0.7 day‐1, and did not differ between participants who exhibited CTL escape near peak VL compared to individuals that developed escape later during VL decline, suggesting that target‐cell limitation was not responsible for VL control during those times. Conclusions Using a novel combination of early detection of infection, frequent sampling, and deep sequencing, we showed that CTL response can occur earlier and with greater potency than previously observed, and plays a role in shaping the VL profile during early acute infection. This information is important for modeling of early HIV‐1 pathogenesis and vaccine design. 14 NO EVIDENCE OF STRONGER NEUTRALIZING ANTIBODY RESPONSES IN RV144 BREAKTHROUGH VACCINE RECIPIENTS TWO TO THREE YEARS AFTER HIV INFECTION S. Krebs (1), S. Tovanabutra (2), G. Donofrio (3), E. Sanders‐Buell (4), S. Miller (5), M. Bose (6), K. Poltavee (7), H. Zhao (8), K. Wong (9), A. O'Sullivan (10), J. Lee (11), B. Ahani (12), S. Muhammad (13), S. LePore (14), C. Oropeza (15), A. Chenine (16), V. Polonis (17), S. Rerks‐Ngarm (18), R. Paris (19), P. Edlefsen (20), P. Gilbert (21), M. Robb (22), N. Michael (23), J. Mullins (24), J. Kim (25), *M. Rolland (26) MHRP/HJF, Bethesda, MD, USA The RV144 vaccine efficacy trial showed a reduction in HIV infections that associated with stronger binding ‐but not neutralizing‐ antibody responses against Env‐V2. Here we investigated the impact of vaccination on HIV evolution and neutralizing antibody (NAb) responses elicited after infection.Env genes from HIV breakthrough infections were sequenced following endpoint‐dilution PCR amplification from plasma. Treatment‐naive sera samples were tested against 35 viruses (HIV‐1 subtypes A/B/C/D/CRF01_AE/CRF02_AG) using a high‐throughput robotic microneutralization assay.Neutralization assays performed using sera from 31 vaccine and 49 placebo recipients about three years post‐HIV diagnosis showed that one vaccinee and eight placebo recipients neutralized >70% of the panel, noting that this vaccinee had only received one of the six RV144 immunizations. Analysis of 41 subjects tested at later times (>1086 days) revealed a trend toward less breadth in vaccinees compared to placebo recipients (median: V=34, P=49, p=0.12). Next, we analyzed 2,247 env sequences obtained from 28 vaccine and 45 placebo recipients at HIV diagnosis and six months later. Intra‐host env diversity tended to be lower among sequences from vaccinees at HIV diagnosis (p=0.07), but not six months later (p=0.52). There was no evidence for vaccine/placebo differences in the diversification of env (p=0.50) or in the number of selected sites (p≥0.34). In addition, the divergence calculated from the vaccine insert sequences showed no vaccine/placebo difference (p≥0.38).In summary, we found no evidence that vaccination intensified the development of NAb responses in breakthrough vaccinees, suggesting that vaccine‐induced immune responses were too weak to exert a sustained genetic pressure on env or to prime for stronger NAb responses post‐infection. The limited NAb responses among vaccinees could reflect that the vaccine selected out intrinsically good humoral responders, i.e., those experiencing vaccine breakthrough were poor responders to vaccination and may also be impaired in their ability to mount Nabs upon natural infection. 15 ASSESSING THE PROPENSITY OF HIV‐1 TO EVOLVE ANTIBODY ESCAPE VARIANTS DURING FREE VIRUS AND CELL‐CELL TRANSMISSION *C. Magnus (1), L. Reh (2), J. Weber (3), T. Uhr (4), P. Rusert (5), A. Trkola (6) Institute of Medical Virology, University of Zurich, Zurich, Switzerland The human immunodeficiency virus type 1 (HIV‐1) can infect target cells via two distinct routes: (i) free virus transmission: free virions can enter and infect target cells and (ii) cell‐cell transmission: virions can be transmitted directly from infected to uninfected cells via narrow range connections between cells, such as the virological synapse. Cell‐cell transmission has been shown to be less sensitive to neutralizing antibodies. Here we investigate if this increases the chances of viral escape variants to evolve as a response to selective pressure induced by neutralizing antibodies. To study whether one transmission route favors the occurrence of viral escape variants, we measured virus inhibition by broadly neutralizing antibodies (bnAbs) in experimental settings that either only allows for free virus or cell‐cell transmission to occur. Based on the information on the inhibitory potential in the two transmission modes, we developed a model that allows us to compare the probabilities that an escape variant arises in either transmission setting for each antibody‐virus pair measured. We further determined the mutant selection windows for bnAbs PGT121 and PG128 using a neutralization escape mutant and the matching wildtype virus.We find strong evidence that the cell‐cell transmission route serves as a rescue pathway for virus transmission because escape variants can evolve over substantially wider antibody concentration ranges via this route. Knowledge on which infection route is more prone to evolution of escape variants opens possibilities to tailor future intervention strategies that directly target both transmission routes reducing the chances for escape mutants to occur. 16 THE ROLE OF TRANSMITTED AND DE NOVO CELLULAR IMMUNE ESCAPE IN HIV DISEASE PROGRESSION J. Carlson (1), V. Du (2), N. Pfeifer (3), C. Brumme (4), M. Schaefer (5), R. Shapiro (6), S. Frater (7), S. Mallal (8), M. John (9), T. Ndung'u (10), P. Goulder (11), T. Allen (12), E. Hunter (13), P. Goepfert (14) 1. Microsoft Research, 2. U of Alabama at Birmingham, 3. Max Plank Institute 4. U. of British Columbia, 5. Emory, 6. Beth Israel Deaconess MC, 7. Oxford, 8. Vanderbilt 9. Murdoch U., 10. U. of KwaZulu‐Natal, 11. Oxford, 12. Ragon Institute, 13. Emory 14. U. of Alabama at Birmingham The extent of HIV intra‐ and inter‐host adaptation to the HLA‐mediated immune response remains a central challenge for HIV vaccine design, as transmitted and de novo HLA escape mutations may compromise vaccine efficacy and reduce the level of protection conferred by certain HLA alleles. One strategy is to vaccinate against escaped epitopes, though it is unclear whether such epitopes can elicit effective immune responses. We developed a probabilistic model of sequence evolution, trained on >4,000 HIV sequences with matched HLA types, that yields a natural metric of the extent of HLA‐specific adaptation. Adaptation of an individual’s autologous sequence to their HLA alleles was strongly associated with VL and CD4 counts in both cross‐sectional and longitudinal early infection cohorts, confirming the role autologous adaptation plays in disease progression and validating our models. Within transmission pairs, the extent to which the donors’ Gag, Pol and Nef sequences were “pre‐adapted” to the recipients’ HLA alleles predicted recipient VL 24 months post infection (Spearman’s rho =0.27, p=0.004, N=113) and time to CD4<250 (HR = 12.7, p = 0.003, N=46). Within chronically‐infected populations, the extent to which circulating viruses were adapted to an individual’s HLA alleles (which estimates the average risk of escape mutation transmission) combined with autologous (within host) adaption to predict set point VL. Experimentally, lower rates of interferon‐γ ELISPOT responses to autologous peptides were observed during acute infection when epitopes in the founder virus were adapted to recipient HLAs. Moreover, primary CTL responses to adapted epitopes had lower antigen sensitivity and reduced cytotoxicity compared to responses in HLA matched individuals whose autologous epitopes were unadapted, despite similar HLA binding and cytokine polyfunctionality profiles among adapted and non‐adapted epitopes. Thus, transmitted escape appears to elicit ineffectual CTL responses and combines with de novo escape to largely shape the dynamics of disease progression. This observation argues that many CTL escape variants are universal across individuals with similar HLA repertoires and suggests that vaccination strategies that target such escapes must achieve a quality of response that is not observed in natural infection. 17 INFLUENCE OF RECOMBINATION ON ACQUISITION AND REVERSION OF IMMUNE ESCAPE AND COMPENSATORY MUTATIONS IN HIV‐1 P. Nagaraja (1), *H. Alexander (2), S. Bonhoeffer (3), N. Dixit (4) 1. Indian Institute of Science 2. ETH Zurich 3. ETH Zurich 4. Indian Institute of Science When HIV is transmitted to a new host, it must adapt to a different immune response. Specifically, the virus must acquire escape mutations to epitopes newly targeted by cytotoxic T lymphocytes (CTLs) in the new host, along with compensatory mutations to alleviate the cost of escape. Simultaneously, it must revert such mutations acquired in the previous host at epitopes that are no longer targeted. While high mutation and turnover rates drive acquisition of individual mutations, recombination affects associations between mutations and thus influences the rate of multi‐locus adaptation. Since recombination can both bring together and break up adaptive combinations, it is unclear under what circumstances its net effect on adaptation is accelerating or decelerating. Using a detailed model of HIV dynamics and evolution in the presence of CTL pressure, we predict the influence of recombination on the waiting time until first appearance of the viral strain carrying all requisite mutations. We find that depending on the underlying fitness landscape – determined by both cost of escape and extent of immune killing – adaptation proceeds via distinct dominant pathways associated with different net effects of recombination. Specifically, recombination tends to decelerate adaptation when a purely uphill fitness trajectory is accessible at each epitope, but accelerate adaptation when a fitness valley must be crossed. Recombination thus invariably accelerates reversion, but can either hasten or delay escape. The greatest delay occurs when multiple epitopes are targeted, escape and compensatory mutations associated with each epitope are located far apart, and escape mutations alone carry low cost. The potentially large effect of recombination could have implications for interpreting data on the timing of observed immune escape. 18 THE CONSERVED ELEMENTS (CE) APPROACH TO HIV VACCINE DESIGN *J. Mullins (1), V. Kulkarni (2), S. Manocheewa (3), X. Hu (4), P. Munson (5), D. Fuller (6), B. Felber (7), G. Pavlakis (8) University of Washington For many years HIV vaccine efforts focused on delivery systems and vectors, rather than on which viral proteins to include. Recently, however, greater effort has been placed on immunogen design, first exploring the use of central state and then variation inclusive immunogens. Most recently, we and others have proposed the use of subgenic segments of viral proteins composed of only conserved regions of the viral proteome as immunogens in order to increase the breadth of strain recognition and redirect immune responses to the otherwise subdominant epitopes that have been associated with better immune control of infection. We engineered DNA‐based immunogens encoding conserved elements (CE) of HIV‐1 selected on the basis of stringent conservation, functional importance, broad HLA‐coverage and association with viral control. DNA vectors were developed to express 7 collinearly arranged CE from p24gag found in >98% of HIV‐1 group M sequences. These sequences were primarily located at p24 multimerization interfaces. By analogy to HIV, similar vectors were developed for SIV p27gag. Naïve and infected macaques with low or undetectable viral loads were immunized with DNA encoding CE alone or in prime‐boost regimens using in addition DNA encoding the complete Gag. All CE DNA‐vaccinated macaques developed robust CE–
specific T cell and humoral immune responses. In contrast, upon repeated vaccination with HIV or SIV full‐length gag DNA, only half of the animals developed T cell immunity targeting any of the CE. However, gag DNA vaccination significantly boosted the magnitude and breadth of preexisting CE T cell and antibody responses. CE immunogens also induced qualitatively different responses, with CE alone inducing multifunctional responses. Thus CE vaccines induce broad responses to vulnerable sites of the virus while avoiding “decoy” targets that divert effective T cell responses towards less protective epitopes. Our prime‐boost approach provides a novel strategy to increase the magnitude and breadth of cellular and humoral immunity. 19 CARD‐SGS REVEALS PERSISTENT EXPRESSION OF HIV‐1 RNA IN CLONALLY EXPANDED HIV‐INFECTED CELLS DURING ART J. Spindler (1), A. Wiegand (2), N. Johnson (3), W. Shao (4), F. Hong (5), A. Cillo (6), E. Halves (7), E. Fyne (8), J. Coffin (9), J. Mellors (10), *M. Kearney (11) 1. HIV Drug Resistance Program, National Cancer Institute, Frederick, MD 2. HIV Drug Resistance Program, National Cancer Institute, Frederick, MD 3. HIV Drug Resistance Program, National Cancer Institute, Frederick, MD 4. Advanced Biomedical Computing Center, Leidos, Frederick, MD 5. Department of Medicine, University of Pittsburgh, Pittsburgh, PA 6. Department of Medicine, University of Pittsburgh, Pittsburgh, PA 7. Department of Medicine, University of Pittsburgh, Pittsburgh, PA 8. Department of Medicine, University of Pittsburgh, Pittsburgh, PA 9. Department of Molecular Biology and Microbiology, Tufts University, Boston MA 10. 3Department of Medicine, University of Pittsburgh, Pittsburgh, PA 11. HIV Drug Resistance Program, National Cancer Institute, Frederick, MD Background: Little is known about the expression of HIV‐1 proviruses in cell populations that persist during ART. Single‐genome sequencing (SGS) of HIV‐1 cell‐associated RNA and DNA (CARD) can characterize the diversity and distribution of the expressing and total infected cell populations. Here, we used CARD‐SGS to assess the genetics of HIV‐1 in longitudinal blood samples from patients on suppressive ART. Methods: HIV‐1 CARD was extracted from 4‐8 aliquots of <106 PBMCs from longitudinal samples obtained 2‐4 weeks apart from two patients on long‐term suppressive ART. Genomic DNA and cDNA were each diluted to <1 HIV‐1 molecule/PCR reaction, and the p6gag/pro/pol region was amplified and sequenced. Because both patients initiated ART with high HIV diversity, identical RNA sequences from the same aliquot of PBMCs were assumed to be derived from the same cell, whereas identical sequences in different aliquots were assumed to be derived from clonally expanded cells. The expressed population was compared phylogenetically to the total proviral population. Results: The genetic diversity of both intracellular RNA and DNA was high (1‐2% APD) with more G to A hypermutants detected in the DNA (23% vs. 10%). Phylogenetic analyses revealed that intracellular DNA and RNA populations intermingled and that clonally expanded, infected cells frequently express HIV‐1 RNA. Analysis of longitudinal samples showed that the same HIV‐1 RNA sequences were identified 2‐4 weeks apart, indicating that clonally expanded cell populations can persist during ART despite expression of unspliced HIV‐1 RNA. Conclusions: By characterizing HIV‐1 proviral expression in infected cells that persist during ART, we found that diverse proviruses express HIV‐1 RNA, including those in clonally expanded cells, and that HIV‐1 RNA expression persists in infected cells for at least for 2‐4 weeks. Additional studies are needed to determine if persistent HIV‐1 expression results from stochastic reactivation of latent proviruses or from continuous transcription. 20 PERSISTENT AND PROLIFERATING HIV‐INFECTED CELLS: TISSUE‐DISTRIBUTION AND EXPRESSION OF VIRAL RNA AND PROTEIN *J. Mullins (1), R. Fox (2), M. Bull (3), S. McLaughlin (4), D. Westfall (5), J. Herbeck (6), L. Frenkel (7) University of Washington Over the past few years we have shown that the decline of HIV infected cells during ART is substantially offset by the increasing representation of infected cells bearing identical proviruses. We and others have also shown that the proliferation of infected cells that can persist for more than 17 years are responsible for this phenomenon, and we have further shown that these infected cells are substantially more likely than random to be inserted within genes associated with regulation of the cell cycle and cancer. The relative abundance of replication competent versus replication defective proviruses in proliferating cells is low, but we have shown that even replication defective proviruses can produce virions and readily detectable low‐level viremias during ART (virus “blips”). In the current study we show that proliferating HIV‐infected cells were present at differing levels in the tissues of different subjects that died while on ART, with up to ~75% of all HIV‐infected cells in the body being derived from a single infected cell harboring a defective provirus. These proliferating infected cells were also found throughout the body, although at differing levels. Likewise, low levels of viral RNA and protein were detected in most tissues of each subject. Furthermore, viral protein production was found to be common in T cells and to a lesser extent, macrophages, in gut biopsies of individuals on suppressive ART. These studies are helping to unravel that apparently extensive role of proliferating infected cells to persisting HIV infection, the infectious HIV reservoir, and continuing HIV antigenic stimulation under ART. 21 MODELING THE EFFECTS OF VORINOSTAT IN VIVO REVEALS BOTH TRANSIENT AND DELAYED ACTIVATION OF HIV TRANSCRIPTION IN LATENTLY INFECTED CELLS R. Ke (1), J. Elliot (2), S. Lewin (3), *A. Perelson (4) 1. Los Alamos National Lan 2. Monash University, Me;bourne 3. Monash University, Melbourne 4. Los Alamos National Laboratory Recent efforts to cure human immunodeficiency virus type‐1 (HIV‐1) infection have focused on developing latency reversing agents to eradicate the latent reservoir. The histone deacetylase inhibitor, vorinostat, has been shown to activate HIV RNA transcription in latently infected cells and alter host cell gene transcription. However, it is not clear how latent cells respond dynamically to vorinostat treatment and what impact the treatment has on reservoir size in vivo. We have constructed viral dynamic models of latency that incorporate the impact of vorinostat treatment, and fitted the models to data collected from a recent clinical trial. The results show that HIV transcription may be activated transiently during the first few days of treatment and that there is a delay in the sustained activation of HIV transcription, which varies from patient to patient and may depend on the long term impact of vorinostat on host gene expression. Parameter estimation suggests that HIV transcription in latently infected cells activated by vorinostat occurs at lower levels than in productively infected cells. Furthermore, the estimated loss rate of transcriptionally activated cells remains close to baseline in most patients, suggesting vorinostat treatment does not induce killing and reduce the latent reservoir in vivo. This work represents the first model to study the dynamics of the HIV latent reservoir under treatment with a latency reversing agent. 22 TARGETING CCR5 RECEPTOR USING GENE THERAPY WILL NOT BE SUFFICIENT TO CURE HIV *A. Pandit (1), R. de Boer (2) Theoretical Biology, Utrecht University, The Netherlands Highly active antiretroviral therapy (HAART), a cocktail of antiretroviral drugs, has successfully transformed Human immunodeficiency virus type 1 (HIV‐1) from a detrimental pathogen to a manageable chronic infection. In spite of its effectiveness, HAART therapy is marred by high cost, toxicity, drug resistance, and the need to be administered lifelong because the virus can persist for years inside the host. Gene therapy offers an opportunity to create a population of genetically modified host cells which are less susceptible to viral entry by modifying the viral entry co‐receptor CCR5. Abstaining viral entry promises to control HIV infection within patients and can exempt them from lifelong use of HAART. Using mathematical modeling, we show that gene therapy proposed to target CCR5 co‐receptor will fail to suppress HIV‐1 infection. We propose that cells modified using CCR5 gene therapy should incorporate suicidal genes that are selectively expressed upon HIV‐1 infection to curb HIV‐1 transmission. This combination will ensure an effective suppression of HIV‐1 viral load and an increase in the T cell count. 23 EFFECT OF THE LATENT RESERVOIR ON THE EVOLUTION OF HIV AT THE WITHIN‐ AND BETWEEN‐HOST LEVELS *H. Doekes (1), C. Fraser (2), K. Lythgoe (3) 1. Theoretical Biology and Bioinformatics, Utrecht University, the Netherlands 2. Department of Infectious Disease Epidemiology, Imperial College London, United Kingdom 3. Department of Infectious Disease Epidemiology, Imperial College London, United Kingdom. Current address: Division of Biosciences, University of Exeter at Tremough, United Kingdom The existence of long‐lived reservoirs of latently infected CD4+ T cells is the major barrier to curing HIV, and has been extensively studied in this light. However, the effect of these reservoirs on the evolutionary dynamics of the virus has received remarkably little attention. Here, we present a within‐host quasispecies model that incorporates a long‐lived reservoir, which we then nest into an epidemiological model of HIV dynamics. For biologically plausible parameter values, we find that the presence of a latent reservoir can severely delay evolutionary dynamics within a single host, with longer delays associated with larger relative reservoir sizes and/or homeostatic proliferation of cells within the reservoir. These delays can fundamentally change the dynamics of the virus at the epidemiological scale. In particular, the delay in within‐host evolutionary dynamics can be sufficient for the virus to evolve intermediate levels of virulence that maximize transmission, as is observed, and not the very high levels of virulence that previous models have predicted. In conclusion, we argue that the latent reservoir has important, but hitherto unappreciated, roles in both within‐ and between‐host viral evolution. Correspondence address: Hilje M. Doekes Hugo R. Kruytgebouw, room Z527 Padualaan 8 3584 CH Utrecht The Netherlands Phone: +31 (0)30 253 3688 Fax: +31 (0)30 253 2837 E‐mail: [email protected] 24 HOW RAPIDLY ARE HIV‐1 INFECTED CELLS KILLED? S. Gadhamsetty (1), *R. De Boer (2) Theoretical Biology, Utrecht University Despite strong genetic and evolutionary evidence that CD8+ T cells play an important role in controlling HIV‐1 infection, several experiments suggest that the rate at which productively infected cells are killed by CTL is very slow. We review alternative explanations for these experimental observations using a general model with target cells, two stages of infected cells, and CTLs. The dominant eigenvalue of the linear subsystem of infected cells cells in the eclipse phase, I, and virus producing cells, P, can be used to confirm that the downslope of the viral load during ART, δ, will reflect the length of the eclipse phase when killing is fast [1]. We also confirm the recent finding that the total killing rate is determined by the replication rate of the subsystem of I and P in the chronic steady state [2]. Hence if the breadth of the immune response is large, each CTL clone makes a small contribution, explaining why the rate at which an immune escape variant takes over the viral quasi species can be very small—despite strong immune control [2]. Since the rate at which the viral load increases following CD8+ T cell depletion provides a measure of the viral replication rate at the previous chronic steady state, we finally use the dominant eigenvalue, to estimate the killing rate required to balance this rate of replication. [1] Klenerman P, Phillips RE, Rinaldo CR, Wahl LM, Ogg G, May RM, McMichael AJ, Nowak MA. Cytotoxic T lymphocytes and viral turnover in HIV type 1 infection. Proc Natl Acad Sci U S A. 1996 93(26):15323‐8. [2] Van Deutekom HW, Wijnker G, de Boer RJ. The rate of immune escape vanishes when multiple immune responses control an HIV infection. J Immunol. 2013 191(6):3277‐86. Correspondence Address: Theoretical Biology, Utrecht University Padualaan 8, 3584 CH Utrecht, Netherlands Email: [email protected] Phone +31 30 253 7560 Fax +31 30 253 2837 25 HIV CONTROL AFTER TREATMENT CESSATION: MATHEMATICAL MODEL PREDICTIONS *J. Conway (1), A. Perelson (2) 1. Pennsylvania State University 2. Los Alamos National Laboratory Antiretroviral therapy (ART increases both the length and quality of life of HIV‐infected individuals, but is not a cure.A recent report from the VISCONTI study suggests that ART, initiated during primary infection, may induce post‐treatment control (PTC) of HIV infection, i.e., maintain HIV RNA < 50 copies/ml, after treatment termination. Our investigation explores the hypothesis that ART initiated during primary infection permits PTC by limiting the size of the latent reservoir, which if small enough at treatment termination, may allow the adaptive immune response to prevent viral rebound and control infection. We model within host HIV dynamics using differential equations to capture interactions between target cells, HIV‐infected cells, latently HIV‐infected cells, virus, and adaptive immune responses. Analysis of our model reveals a range in adaptive immune response strengths where a patient may either show viral rebound or PTC (bistability). Our model predicts that in this regime of bistability, there is a threshold in latent reservoir sizes at treatment termination below which PTC can be achieved. Using data on latent reservoir sizes in patients treated during primary infection, we also predict population‐level viral rebound times for non‐controllers consistent with observations. 26 MODELLING ONGOING REPLICATION OF HIV IN DRUG SANCTUARIES *H. Fryer (1), R. Lorenzo‐Redondo (2), C. Fletcher (3), T. Schacker (4), A. Haase (5), S. Wolinsky (6), A. McLean (7) 1. Oxford University 2. Northwestern University 3. University of Nebraska 4. University of Minnesota 5. University of Minnesota 6. Northwestern University 7. Oxford University HIV‐1 viremia persists at low levels after many years of clinically effective antiretroviral therapy. Furthermore, drug concentrations vary across anatomical site, and new evidence for the diversification of HIV‐1 during therapy is now emerging. A coherent explanation for these observations is that there is persistent replication in areas with low drug penetration, however this has been challenged on the basis that ongoing replication during therapy would inevitably lead to the emergence of drug resistance. In reality, modern drug combinations succumb to drug resistance only very rarely. We developed a spatial within‐host model of HIV‐1 and fitted it to viral dynamics data during therapy. Using this model we address the following questions. Can drug sanctuaries harbour ongoing replication of HIV during therapy without the emergence of drug resistance? Where, and how large, could the ongoing replication space be? What impact could improvements to the penetration of drugs have on viral dynamics and resistance? We demonstrate that the combination of spatial heterogeneity to drug penetration and spatial effects to mixing can result in the ongoing replication of drug sensitive virus. The location of regions of ongoing replication should be identifiable by reduced changes in infection measures over long time periods compared to those observed in regions with high drug penetration such as the blood. Our data spanning the first 6 months of therapy are consistent with ongoing replication occurring in a small fraction of the lymphatic tissue. However, improvements to the sensitivity and duration of HIV dynamics studies are needed to test this theory further. We predict that improvements to the penetration of drugs would could halt ongoing replication. 27 A ROBUST SCALING LAW ESTIMATES THE REQUIRED DURATION OF TREATMENT WITH HIV‐1 MUTAGENS *V. Gupta (1), N. Dixit (2), . (3) Indian Institute of Science Chemical mutagens that can increase the mutation rate of HIV‐1 beyond its error threshold present a promising treatment strategy that may be less susceptible to resistance than current antiretroviral drugs. Several mutagens are under development, including one that entered clinical trials. Successful treatment with a mutagen requires accurate estimates of the error threshold of HIV‐1, µc, and the time, t, it takes for error catastrophe to occur once the mutation rate, µ, exceeds µc. Recently, we employed stochastic simulations that closely mimic patient data of the within‐host evolution of HIV‐1 and predicted that µc was 2‐6 fold above the natural mutation rate of HIV‐1. In the present study, we applied the simulations to estimate t. We found, interestingly, that t obeys a scaling law: t»0.6/(µ‐µc). This scaling law, intriguingly, was robust to variations in parameter values including the population size, recombination rate, and genome length, although µc depended on these parameters. We examined whether the scaling law arose also from the quasispecies theory. In the quasispecies theory, the dominant eigenvector of the value matrix, W, whose entries define the rates at which genomes are produced from the erroneous replication of other genomes, yields the quasispecies. The spectral gap, or the difference between the two largest eigenvalues, of W defines t. Computing these latter quantities, we found that the above scaling law was not obeyed by the quasispecies theory, identifying a point of departure between the quasispecies theory and population genetics‐based models. Finally, we applied the scaling law to estimate the required duration of treatment with HIV‐1 mutagens. We predict thus that when a mutagen increases µ to 5µc, 8‐11 years of treatment would be necessary. The transition to error catastrophe in HIV‐1 thus appears to be a slow process, requiring many years of treatment. 28 TARGETING HIV LATENCY *F. Fu (1), . (2), S. Bonhoeffer (3) ETH Zurich HIV latency is the barrier for a cure, yet it is recently thought to offer a potential therapeutic target for improving health outcomes of HIV patients. Here we provide a mathematical framework, explicitly taking into account the role of cellular latency and reactivation in viral rebound if treatment is interrupted. In particular, we focus on 'activation‐kill' approaches aimed at eliminating the reservoir of latently infected cells. Outcomes of such an "anti‐latency" therapy strongly depend on the duration of and the efficacy of treatment. A minimal period of anti‐latency treatment is required to achieve desired clinical outcomes (i.e., greater clearance rates and longer period of no viral rebound). Surprisingly, the anti‐latency treatment, if not administered for a sufficiently large period, is detrimental and can actually speed up viral rebound. Our proof‐of‐principle modeling offers closed‐form solutions that are understandable to clinicians and experimentalists and can be used to improve risk management of targeting HIV latency. 29 DETECTION OF TRANSMISSION CLUSTERS IN HIV PHYLOGENIES *T. Stadler (1), L. Villandre (2), J. Barido‐Sottani (3) 1. ETH Zürich 2. McGill University 3. ETH Zürich Genetic sequencing data is frequently used to reconstruct phylogenies on the between‐host HIV level. These phylogenies contain information about the underlying host contact network and epidemiological dynamics within that population. One aim is to identify transmission clusters from phylogenetic trees. These transmission clusters are typically interpreted as fast transmission chains in highly connected regions in the host contact network. Identified transmission clusters thus allow us to obtain insight into the host population structure, and to quantify the transmission dynamics within the transmission clusters. A range of heuristics is currently used to identify transmission clusters from phylogenies, but it is not clear how well these inferred transmission clusters correspond to highly connected parts in the host population network (so‐called network communities). Based on our simulation study, we find poor correspondence between inferred phylogenetic transmission clusters and network communities. We propose a model‐based transmission cluster inference tool, rather than previously‐used heuristics, to improve the accuracy of phylogenetic transmission cluster inference. 30 PHYLOGEOGRAPHIC METHODS GIVE INSIGHT INTO THE EPIDEMIOLOGICAL DYNAMICS BETWEEN HIV‐1 RISK GROUPS *D. Kühnert (1), T. Stadler (2), T. Vaughan (3), A. Drummond (4) 1. ETH Zürich 2. ETH Zürich 3. University of Auckland 4. University of Auckland The between host dynamics of HIV depend on a number of factors, one of which is the risk group of the infected hosts. HIV risk groups, such as heterosexuals (HET), men having sex with men (MSM) or injecting drug users (IDU), can be modelled as discrete demes in which infected hosts reside. This way one can measure differences between within‐deme and between‐deme transmission. We developed and implemented a birth–death–sampling process with migration and/or infection among discrete demes, in the Bayesian MCMC framework BEAST 2. Similarly to the recently introduced birth–
death skyline model (1), this multi‐type birth–death model allows the estimation of epidemiological parameters, such as the effective reproduction number R and also allows epidemiological rates to change over time. We analyse a previously published set of HIV‐1 sequences from Latvia (2) from hosts belonging to either of the risk groups HET or IDU. Our results confirm the hypothesis that the reproduction number within the IDU risk group was higher than within HET hosts, and that more transmission occurred from IDU to HET hosts than vice versa (3). However, transmission from HET to IDU hosts does not appear to be negligible, as previously found. (1) Stadler, T., Kühnert, D., Bonhoeffer, S., and Drummond, A. J. (2013). “Birth‐death skyline plot reveals temporal changes of epidemic spread in HIV and hepatitis C virus (HCV).” Proc Natl Acad Sci USA, 110(1): 228–33. (2) Balode D, Skar H, Mild M, Kolupajeva T, Ferdats A, Rozentale B, Leitner T, Albert J. 2012 Phylogenetic analysis of the Latvian HIV‐1 epidemic. AIDS Res. Hum. Retroviruses 28, 928 – 932. (doi:10.1089/AID. 2011.0310) (3) Stadler T, Bonhoeffer S. Uncovering epidemiological dynamics in heterogeneous host populations using phylogenetic methods. Phil Trans R Soc Lond B Biol Sci. 2013 Mar 19;368(1614):20120198. PubMed PMID:23382421. 31 BAYESIAN INFERENCE OF TRANSMISSION PATTERNS FROM TIMED PHYLOGENIES *C. Colijn (1), M. Kendall (2), X. Didelot (3) Imperial College London The question of how well genomic data can capture transmission events patterns remains a challenge despite significant advances in phylodynamics. This is particularly true for pathogens like HIV, with variable infectivity over a long course of infection, and substantial within‐host diversity. Reconstructing individual transmission events from HIV sequence data would have a significant impact on our ability to use sequence data to inform estimates of recent incidence and changes in transmission. Here we present a novel Bayesian approach to infer transmission events from genomic data while considering within‐host diversity. We infer a time‐labeled phylogeny using Bayesian evolutionary analysis by sampling trees (BEAST), and then infer transmission among both sampled and unsampled cases via a Monte Carlo Markov chain. We find that under a realistic model of within‐host evolution, reconstructions of simulated outbreaks contain substantial uncertainty even when genomic data reflect a high substitution rate. Despite that, posterior estimates of patterns such as recent changes in incidence are reasonable. We discuss the ability of the method to estimate the fraction of transmissions occurring during the acute stage of infection, and the fraction of transmissions occurring between different groups within the population. We conclude with a discussion of the flexibility of our framework to incorporate additional data, and the role of the sampling fraction. 32 NEAR FULL LENGTH HIV‐1 SEQUENCING TO UNDERSTAND HIV PHYLODYNAMICS IN AFRICA IN REAL TIME S. Danaviah (1), J. Manasa (2), E. Wilkinson (3), S. Pillay (4), Z. Sibisi (5), S. Msweli (6), D. Pillay (7), *T. de Oliveira (8) Africa Centre for Health and Population Studies, UKZN, South Africa Background: HIV transmission continues in Africa at alarming rates despite biological and behavioural interventions. Understanding the drivers of HIV transmission and evolution and translating the results into effective interventions is a key component of halting the epidemic. Recent technological advancement in complete genome sequencing has expanded the breadth and speed of genomic analyses currently possible. We have constructed a high‐throughput genomics and bioinformatics pipeline that has successfully generated high quality complete HIV genomes in a hyper‐endemic region of South Africa (SA), through the PANGEA_HIV Consortium Methods: HIV RNA was extracted from plasma from patients failing antiretroviral therapy, within the Africa Centre (AC) research area and 4 overlapping regions spanning the 9.7kb complete HIV genome were amplified in a one‐step RT‐PCR strategy optimised for subtype C virus. Pooled amplicons were sequenced on an Illumina MiSeq. Fragments were quality controlled with SMALT software and assembled using two independent strategies (de novo and mapping to reference) in Geneious. Resulting consensus sequences were aligned against published HIV complete genomes from South Africa (n=300). Bayesian and maximum likelihood trees with branch support were reconstructed in PhyML and MrBayes. Results: Amplification success rate of complete genomes, on samples with viral loads >10,000 c/ml was 77.3%. Near complete HIV genomes were generated for 433/473 samples sequenced thus far, with all nine open reading frames, the U5 / partial R region of the 5’ LTR and partial U3 of the 3’ LTR represented. Coverage of the HIV genome averaged 99.9% with a mean depth of coverage of 15 539 times. Phylogenetic reconstruction confirmed that the AC strains were all HIV‐1 subtype C where 246/473 sequences clustered with other complete genomes from SA. The discrete AC clusters (n=22) suggested multiple independent introductions of subtype C into the surveillance area and onward transmission within the population. Conclusions: This is the first report, to the best of our knowledge, of a high‐throughput complete HIV genome sequencing and analysis pipeline in Africa. The genetic diversity of HIV variants in this population is high and is mediated primarily by multiple introductions of HIV. Interventions therefore must be cognizant of the dynamics that drive these independent introductions in order to impact on going HIV transmission. 33 USING PHYLOGENETIC TOOLS TO ESTIMATE ASPECTS OF HIV TRANSMISSION DYNAMICS IN GENERALIZED HIV EPIDEMICS: FINDINGS FROM THE PANGEA_HIV METHODS COMPARISON EXERCISE *O. Ratmann (3), E. Hodcroft (4), M. Pickles (5), A. Cori (6), M. Hall (7), S. Lycett (8), D. Bezemer (9), C. Colijn (10), M. Kendall (11), X. Didelot (12), G. Plazotta (13), A. Poon (14), J. Joy (15), G. Leventhal (16), D. Rasmussen (17), D. Kuehnert (18), T. Stadler (19), M. Karcher (20), G. Baele (21), M. Suchard (22), P. Lemey (23), V. Minin (24), E. Volz (25), B. Dearlove (26), S. Frost (27), A. Gavryushkina (28), D. Welch (29), A. Drummond (30), A. Rambaut (31), A. Leigh Brown (32), C. Fraser (33) 3. Imperial College London, 4. University of Edinburgh, 5. Imperial College London 6. Imperial College London, 7. University of Edinburgh, 8. University of Edinburgh 9. HIV Monitoring Foundation Netherlands, 10. Imperial College London, 11. Imperial College London, 12. Imperial College London, 13. Imperial College London, 14. University of British Columbia 15. BC Centre of Excellence HIV/AIDS, 16. ETH Zuerich, 17. ETH Zuerich, 18. ETH Zuerich, 19. ETH Zuerich 20. University of Washington, 21. University of Leuven, 22. UCLA, 23. University of Leuven 24. University of Washington, 25. Imperial College London, 26. University of Cambridge 27. University of Cambridge, 28. University of Auckland, 29. University of Auckland, 30. University of Auckland, 31. University of Edinburgh, 32. University of Edinburgh, 33. Imperial College London PANGEA_HIV (Phylogenetics and Networks for Generalised HIV Epidemics in Africa) will generate >10,000 full‐genome HIV sequences from sub‐Saharan Africa to better characterize these epidemics for HIV intervention trials. To understand the accuracy and reliability of existing phylogenetic tools, the PANGEA_HIV methodology working group conducted a methods comparison exercise.Two agent‐based epidemiological models capturing generalized HIV transmission dynamics in a village‐like, and regional population were developed, and used to simulate transmission trees from different epidemic scenarios. Gag, pol and env subtype C sequences were generated along the transmission tree output for a subset of infected individuals, and associated with demographic and clinical variables of these individuals. Simulations were designed to test the accuracy and power of different phylogenetic methods in estimating 1) recent changes in HIV incidence and 2) the proportion of HIV infections arising from transmissions from individuals in acute HIV infection. Simulations controlled for different sequence sampling strategies and the proportion of HIV infections originating from outside the sampling area. By December 2014, 27 external participants in 7 research groups had analysed the training data sets. The true transmission dynamics were revealed and discussed at a workshop in London on December 2nd 2014. Most teams was able to classify epidemics correctly into increasing, stable, and decreasing, but faced methodological challenges in quantifying changes in HIV incidence. Almost all groups used innovative, scalable tree reconstruction methods. The final testing data sets will be released early in 2015 with final results due May‐1 2015. 34 HIV‐1 TRANSMITTED DRUG RESISTANCE: NEW INSIGHTS INTO THE TRANSMISSIBILITY OF SDRMS R. Winand (1), K. Theys (2), M. Eusebio (3), J. Aerts (4), R. Camacho (5), P. Gomes (6), A. Vandamme (7), *A. Abecasis (8) 8. Centro de Investigação Interdisciplinar Egas Moniz (CiiEM), Instituto Superior de Ciências da Saúde Sul, Campus Universitário, Quinta da Granja, Caparica, Portugal Surveillance drug resistance mutations (SDRMs) in drug naïve patients (DN) are typically used to measure HIV‐1 Transmitted Drug Resistance (TDR). We tested here how SDRMs in patients failing treatment (TR), the original source of TDR, contribute to assessing TDR, transmissibility and transmission source of SDRMs. The prevalence of SDRMs in DN and TR patients was retrospectively measured for 3554 HIV‐1 subtype B infected patients. The transmission ratio (prevalence in DN/prevalence in TR) of each SDRM was calculated and analyzed by robust linear regression with outlier detection to interpret transmissibility. Prevalence of SDRMs in DN and TR were linearly correlated, but some SDRMs were classified as outliers – above (protease: D30N, N88D/S, L90M, reverse transcriptase: G190A/S/E) or below (RT: M184I/V) expectations. The normalized regression slope was 0.073 for PI, 0.084 for NRTI, and 0.116 for NNRTI. We present an innovative and simple approach to investigate the transmissibility of SDRMs by determining individual mutation transmission ratios and using linear regression to describe the relationship between their prevalence in TR and DN. The significant linear correlation between prevalence of SDRMs in DN and in TR indicates that the latter can be useful to predict levels of TDR. Higher transmission ratios and outliers above the regression line indicate more onwards transmission among DN and/or higher persistence of such SDRMs, while the opposite indicates lower transmission among DN and/or lower persistence. Our results emphasize the importance of monitoring SDRMs in TR in order to gain further insight into the dynamics of the transmission of SDRMs. 35 THE IMPORTANCE OF PROTEIN STABILITY FOR THE EVOLUTION OF HIV DRUG RESISTANCE *A. Olabode (1), D. Robertson (2), S. Lovell (3) Computational and Evolutionary Biology, Faculty of Life Sciences, The University of Manchester, Michael Smith Building, Manchester M13 9PT, United Kingdom Many HIV‐1 drug resistance‐associated mutations have been reported. In some cases a single mutation is capable of conferring resistance to the virus. However, in the majority multiple mutations at different locations are required and these are thought to be due to direct epistatic or co‐evolutionary changes in context of protein structure. Mutations are known to cause different degrees of conformational changes to protein structures. Therefore, understanding the evolutionary pathways that lead to the acquisition of drug resistance (and how it affects the structure and function of the virus) is important for predicting resistance and drug regime choices. Here we analyse protease data from the HIV Drug Resistance Database and using pol subtype B sequence alignments reconstruct the probable evolutionary pathways that lead to drug resistance using likelihood methods. We integrated this information with protein structure data, in order to characterise the energy changes along these evolutionary pathways. We find that there is a distribution of “enabling” mutations that contribute to protein stability. Most of these enabling mutations were found to act in concert, stabilizing the viral protein structure before the emergence of drug resistance‐associated mutations. At least four protease resistance stabilising mutations appear to be a prerequisite for drug resistance as opposed to a direct coevolutionary mechanism. The maintenance of protein structure stability is thus a major constraining factor to the evolution of HIV‐1. As a consequence drug resistant strains of HIV‐1 need to follow an evolutionary pathway in which enabling mutations occur to stabilize the viral structure. This understanding of evolutionary constraints on the viral protein will provide information useful for prediction of future evolution of the virus, resistance to drug therapy and better drug targets. 36 DIRECT COMPARISON OF SEROLOGIC ASSAYS AND VIRAL DIVERSITY MEASURES FOR IDENTIFYING INDIVIDUALS WITH RECENT HIV INFECTION *O. Laeyendecker (1), A. Redd (2), S. Wendel (3), Q. Abdool Karim (4), V. Naranbhai (5), N. Garrett (6), S. S. Abdool Karim (7), T. Quinn (8), S. Eshleman (9) 1. NIAID 2. NIAID 3. NIAID 4. CAPRISA & Columbia Univ 5. CAPRISA 6. CAPRISA 7. CAPRISA & Columbia Univ 8. NIAID 9. JHU Background: Identifying individuals recently infected with HIV from cross‐sectional surveys is important to measure population level incidence and identify groups currently experiencing high rates of infection. Serologic assays are currently being used and measures of HIV sequence diversity are being proposed to identify recently infected individuals. Methods: Samples were obtained from women in the CAPRISA 004 trial (paired samples from 76 seroconverters). The mean time from HIV seroconversion to sample collection was 71 days for the first sample (interquartile range [IQR]: 64 to 79 days) and 774 days for the second sample (IQR 495 to 857 days). Samples were analyzed with two serologic tests, the BED capture enzyme immunoassay (BED‐CEIA) and the BioRad‐Avidity assay. Sequence data generated from the p24 (HXB nt 1493 to 1735) and gp41 (HXB nt 7858‐8260) by 454 pyrosequencing. Three sequence based diversity measures were generated; viral diversity (mean pairwise distance), complexity (the number of distinct reads divided total read), and Shannon entropy. Receiver operating characteristics analyses (ROC) were run using <1 year post infection as the definition of recent infection, and the under the curve (AUC) was calculated to determine the discriminatory capacity for each method. Results:The BED‐CEIA and BioRad‐Avidity assay had AUCs of 0.964 and 0.967, respectively. The mean number of sequence reads obtained per sample was 2451 for p24 and 3626 for gp41. For p24, the AUCs for diversity, complexity and entropy were 0.749, 0.792 and 0.810. For gp41, the AUCs for diversity, complexity and entropy were 0.737, 0.786 and 0.654. Both serologic assays had significantly higher AUCs (p<0.0001) than any of the diversity measures. Conclusion:Serologic assays were superior to viral diversity measures for distinguishing between individuals with recent vs. non‐recent HIV infection. 37 NGS COMBINED WITH PHYLOGENETIC ANALYSIS TO DETECT SUPER‐INFECTION IN HIV‐1 INFECTED INTRAVENOUS DRUG USERS S. Paraschiv (2), E. Radu (3), L. Banica (4), B. Popescu (5), I. Nicolae (6), I. Niculescu (7), A. Abagiu (8), D. Otelea (9) 2. Molecular Diagnostics Laboratory, National Institute for Infectious Diseases “Matei Bals”, Bucharest, Romania 3. University of Medicine and Pharmacy “Carol Davila”, Bucharest, Romania 4. Molecular Diagnostics Laboratory, National Institute for Infectious Diseases “Matei Bals”, Bucharest, Romania 5. University of Medicine and Pharmacy “Carol Davila”, Bucharest, Romania 6. Molecular Diagnostics Laboratory, National Institute for Infectious Diseases “Matei Bals”, Bucharest, Romania 7. Clinical Department, National Institute for Infectious Diseases “Matei Bals”, Bucharest, Romania 8. Clinical Department, National Institute for Infectious Diseases “Matei Bals”, Bucharest, Romania 9. Molecular Diagnostics Laboratory, National Institute for Infectious Diseases “Matei Bals”, Bucharest, Romania Introduction: A HIV‐1 sub‐epidemic started recently in Romania among intravenous drug users (IDUs). It is characterised by the co‐circulation of local F1 subtype strains and CRF14_BG strains. HIV‐1 is one of the fastest mutating organisms, displaying a remarkable inter‐ and intra‐host genetic heterogeneity. Due to the high HIV transmission risk IDUs are exposed to, super‐infections are to be expected in these patients. Methods: We analyzed strains from 19 IDUs and 20 heterosexuals newly diagnosed with HIV‐1. The genomic region targeted was V2‐V3 loop of env gene. NGS was performed on the GS‐Junior instrument (Roche). Read mapping, error correction and viral variant reconstruction were performed using QuRe v.0.99971. Phylogenetic analysis was performed with the ML method and FastTree software. Intra‐host genetic diversity was calculated using the maximum‐composite‐likelihood method (Mega6). Geno2pheno coreceptor algorithm was used for tropism prediction. Results: We identified 5 super–infections in IDUs. Phylogenetic analysis indicated that the reconstructed viral variants clustered into 2 well supported groups belonging to F1 and CRF14_BG strains. In these patients the majority of the viral variants belonged to the CRF14_BG cluster, suggesting that the recombinant form might have replicative advantage. The CRF14_BG variants were exclusively CXCR4 tropic; F1 variants were CCR5 tropic. CXCR4 tropism at baseline predicts faster disease progression. No super‐infections were identified in the heterosexuals infected with F1 strains. The overall viral diversity in the IDU group was higher. Conclusion: NGS and phylogenetic analysis can provide important insights into the intra‐host sub‐population structures. The viral populations were more diverse in IDUs than in heterosexuals; super‐infections were identified. 38 RECONCILING NAMED PARTNER AND GENETIC PARTNER HIV‐1 TRANSMISSION NETWORKS IN NEW YORK CITY *J. Wertheim (1), S. Kosakovsky Pond (2), K. Scheffler (3), D. Smith (4), S. Mehta (5), S. Shah (6), L. Forgione (7), L. Torian (8) 1. University of California, San Diego 2. University of California, San Diego 3. University of California, San Diego 4. University of California, San Diego 5. University of California, San Diego 6. New York City Department of Health and Mental Hygiene 7. New York City Department of Health and Mental Hygiene 8. New York City Department of Health and Mental Hygiene BACKGROUND: The New York City Department of Health and Mental Hygiene (DOH) interviews persons with newly diagnosed HIV infection (index cases) and elicits partners, who are notified of exposure and offered HIV testing. When resistance testing is ordered by a physician with whom the case or positive partner has initiated care, the viral nucleotide sequence is reported to surveillance. METHODS: Between 2006 and 2012, DOH interviewed 756 index cases with genotypes; these cases named 810 HIV+ partners with genotypes, for a total of 1,342 cases linked to named partners (189 index cases were also named by other index cases). Using pol sequences, we identified index and named partners who were closest relatives in a maximum likelihood phylogeny. We then estimated the Tamura‐Nei 93 (TN93) genetic distance between each pair of index cases and named partners. We designated viral sequence pairs that fell below a validated distance cutoff of 1.75% as genetic links. Our data made it possible to construct two networks: the network of cases and their named partners (the named partner network; N=1,342) and the network of clusters of persons genetically linked by their TN93 distance (the genetic partner network; N=862 [63% of the 1332 cases]). We examined the degree of overlap between the two networks. We used logistic regression to assess the variables associated with the index case successfully naming at least one genetically linked partner RESULTS: 449 of 756 (59%) index cases named partners who were also genetically linked (Table). Heterosexual female index cases were more likely to be genetically linked to a named partner (77%) than men who have sex with men (42%, OR=0.21, 95% CI 0.14, 0.32) and male injecting drug users (23%, OR=0.20, 95% CI 0.09, 0.43). Black index cases were less likely than whites and Hispanics to name a genetically linked partner (53%, OR=0.46, 95% CI 0.25, 0.87). In the named partner network, 743 out of 1,369 (55%) cases were genetically linked to a named partner, whereas in the genetic network, 743 out of 889 (84%) persons were genetically linked to a named partner. CONCLUSIONS: Construction of genetic transmission networks can supplement partner naming by identifying previously unknown parts of a potential transmission network, i.e., unnamed partners. If real‐time genotyping coupled with network analysis can be implemented, it can be used to interdict ongoing transmission and to improve epidemic control. 39 ESTIMATING THE RESPECTIVE CONTRIBUTIONS OF HUMAN AND VIRAL GENETIC VARIATION TO HIV CONTROL *I. Bartha (1), P. McLaren (2), C. Brumme (3), R. Harrigan (4), A. Telenti (5), J. Fellay (6) 1. École Polytechnique Fédérale de Lausanne, School of Life Sciences, Lausanne, Switzerland 2. École Polytechnique Fédérale de Lausanne, School of Life Sciences, Lausanne, Switzerland 3. BC Centre for Excellence in HIV/AIDS, Vancouver, Canada 4. BC Centre for Excellence in HIV/AIDS, Vancouver, Canada 5. J. Craig Venter Institute, La Jolla, USA 6. École Polytechnique Fédérale de Lausanne, School of Life Sciences, Lausanne, Switzerland Viral load setpoint is a major correlate of HIV disease progression. Genome‐wide association studies have identified common human polymorphisms that together explain no more than 15% of its phenotypic variance. Here we present a joint assessment of the respective contributions of human and viral variation to setpoint. Human genotype data across the Major Histocompatibility Complex (MHC) region, full‐length consensus HIV sequences and setpoint viral load results were available for 1034 treatment naïve HIV‐1 infected individuals. Heritability (h2) estimation was carried out with GCTA using three kernel matrices representing: 1) the human relatedness across the MHC, 2) the viral phylogeny, and 3) the sample‐specific noise. The human kernel was estimated from 27 common polymorphisms selected by LASSO. Phylogenetic trees were inferred from the viral sequences using RAxML. The viral kernel was derived from the phylogenetic trees by taking the branch length of the shared ancestry. Estimating the host heritability of viral load using the host kernel alone yielded a median estimate of h2=8% with an interquartile range (IQR) of 1% across 15 bootstrap replicates of the samples. The estimates of the viral heritability drawn from 30 bootstrapped viral trees had a median of 29% (IQR=10%). Combining the host and viral relatedness matrices showed a comparable viral heritability of 26% (IQR=9%) but a decreased host contribution of 4% (IQR=0%). This is the first estimate of the combined and respective contributions of the host and the viral genomes to the observed variability of HIV viral load. We showed that both the pathogen and host genomes have detectable impacts on the clinical outcome of infection, which are however not independent. These results suggest that the main predictor of clinical outcome is the viral genotype and that host determination of viral load is largely dependent on its ability to select viral variants. 40 ESTIMATING THE BETWEEN‐HOST HERITABILITY OF VIRAL TRAITS: OLD‐SCHOOL PARENT‐OFFSPRING VERSUS MODERN PHYLOGENETICS *G. Leventhal (1), S. Bonhoeffer (2) ETH Zurich ETH Zurich The heritability of a trait is one of the most widely used tools to quantify how fast a trait will evolve in a population. As a result, many different methods have been proposed to measure heritability in real populations. With the rise of the availability of genetic data, traditional methods such as parent‐offspring regression or sibling analysis have been superseded by phylogeny‐based methods to estimate heritability. However, as with all phylodynamic models, tree‐based methods for heritability estimation require an underlying model that describes how the trait evolved along the tree. Using set‐point viral load in HIV as an example, I will show that tree‐based estimates of heritability are very sensitive to model mis‐specification such as the absence of selection, both within‐ and between hosts. In contrast, estimates from parent‐offspring (i.e. donor‐recipient) regression are more robust to such misspecification. The difficulty of obtaining good parent‐offspring pairs must thus be weighed against potentially strongly biased estimates of heritability when the between‐ and within‐host host selective forces are unclear. 41 DETERMINING THE VIRAL GENETIC BASIS OF HIV VIRULENCE USING WHOLE GENOME SEQUENCING *C. Wymant (1), F. Blanquart (2), M. Cornelissen (3), L. Gras (4), A. Gall (5), P. Kellam (6) 1. Imperial College London 2. Imperial College London 3. University of Amsterdam 4. HIV Monitoring Foundation 5. Wellcome Trust Sanger Institute 6. Wellcome Trust Sanger Institute The severity of HIV infection varies considerably between individuals. Set‐point viral load (SPVL), a predictor of clinical outcomes such as time to progression to AIDS, varies by several orders of magnitude. While estimates of the extent of SPVL heritability vary widely (6% to 59%), the fact that it is heritable shows that it is in part controlled by the virus itself. We investigate how, and how much, using whole viral genome data and clinical follow up (including SPVL and CD4 measurements) for around 700 patients from the Netherlands and the UK, sampled between 1985 and 2014. The patients were identified as seroconverters, and the samples were taken less than two years after the first positive test and before ART.We present a proposed pipeline for processing the genetic data and searching for features therein, be they single‐nucleotide polymorphisms or more complicated properties, that correlate with infection severity. Early results will be discussed, including greater SPVL for individuals infected with two different strains, estimates of moderate‐to‐high heritability with greater confidence, and more precise dating of the origin of HIV in Europe.(Note: numerical results will be added to the abstract before the abstract booklet is finalised.) 42 IMPORTANCE OF TRAIT SELECTION IN ESTIMATING VIRAL CONTRIBUTION TO VIRULENCE IN HIV INFECTION *V. Mitov (1), T. Stadler (2) 1. Programme Doctorate at D‐BSSE 2. Assistant Professor at the Department of Biosystems Science and Engineering It has been observed that HIV infected patients have a very variable time to AIDS. The inverse of this time span is called virulence. It is controversially discussed to what extend the viral genotype and to what extend the patient’s immune system and other environmental factors influence virulence. In order to quantify the contribution of the virus to virulence, the so‐called heritability of virulence is estimated. Since the full transmission chain is often unknown, phylogenetic methods to estimate heritability from HIV genetic sequencing data have been developed. A number of phylogenetic methods have been applied to different cohorts yielding contradicting conclusions. We argue that a main reason for previous discrepancy is that the applied methods ignore selection on virulence. We present a new phylogenetic tool to estimate heritability of virulence with explicitly taking into account selection. The method is validated on simulated datasets, revealing that ignoring selection on virulence can lead to an underestimation of heritability. Applying the method to HIV genetic sequencing data from Switzerland and the UK, we observe that the HIV genotype has a significant impact on virulence, with an estimated heritability above 0.2. 43 MODELING HIV VIRULENCE EVOLUTION IN THE CONTEXT OF IMMUNE ESCAPE *C. van Dorp (1), M. van Boven (2), R. de Boer (3) 1. Utrecht University 2. RIVM 3. Utrecht University A pathogen like HIV evolves rapidly under multiple levels of selection, and has to cope with a heterogeneous host population. Although these aspects have been studied before, the true nature of host‐heterogeneity has not been addressed to our satisfaction. During (untreated) infection, HIV evades cellular immune responses and because of the massive polymorphism of the Human Leukocyte Antigen, the targets of these responses (epitopes) differ strongly between hosts. Supported by data, it has been suggested that HIV has evolved virulence levels that are optimal for transmission. Some models indeed predict this, but others caution that this mode of adaptation is not self‐evident, mostly due to the short‐sightedness of evolution during the infection of an individual host. Several theories have been proposed to better explain the evolution of virulence, and we aim to contribute to these attempts. We are developing a model of HIV's evolutionary dynamics that is highly detailed and realistic, and captures the interesting features of host‐heterogeneity, immune escape and compensatory mutations, and selection on multiple levels. We hypothesize that these properties might be sufficient to explain HIV's observed virulence distribution. 44 FULL‐LENGTH ENVELOPE ANALYSIS (FLEA): A PIPELINE AND WEB SERVICE FOR FULL‐LENGTH ENV SEQUENCE ALIGNMENT, ANALYSIS, AND VISUALIZATION *K. Eren (1), B. Murrell (2), M. Laird (3), C. Ignacio (4), E. Paxinos (5), D. Smith (6), S. L. Kosakovsky Pond (7) 1. University of California, San Diego 2. University of California, San Diego 3. Pacific Biosciences 4. University of California, San Diego 5. Pacific Biosciences 6. University of California, San Diego 7. University of California, San Diego We have recently developed an approach to deeply sequence full‐length env from HIV RNA, using Pacific Biosciences’ single molecule, real‐time sequencing platform. This approach produces ±8000 high‐quality, full‐length env sequences in a single run, enabling accurate analyses of HIV populations and how they evolve over time. To take advantage of these sequences, we developed the Full‐Length Envelope Analyzer (FLEA) which is publicly available as a web service at datamonkey.org. The central step in the FLEA pipeline is building a large multiple‐sequence alignment. We build this alignment through an intermediate clustering step, in which we collapse the sequence space into a smaller set of “master” sequences with preserved reading frame, align these, and then perform pairwise, codon‐aware, frame‐correcting alignment of each original read to the nearest master sequence to generate a full multiple‐sequence alignment. Phylogenies are inferred, and we use the HyPhy package for evolutionary analyses and summarizing residue and putative glycosylation variants. The pipeline is publicly available as part of the Datamonkey web service: users may upload their sequence files to our cluster for remote analysis. An interactive report containing publication‐quality plots and visualizations using the D3 library is provided, including: plots of evolutionary dynamics, a visualization of gene‐wide dN/dS rates, an alignment browser, and phylogenetic tree browser. We demonstrate the sequencing protocol and analysis pipeline on a donor who was deeply sequenced at 6 time points, providing perhaps the most comprehensive picture of HIV dynamics and evolution yet. 45 SELECTION OF ANTIGEN SWARMS FROM NEUTRALIZING ANTIBODY DEVELOPMENT AGAINST HIV‐1 *P. Hraber (1), B. Korber (2), E. Giorgi (3), T. Bhattacharya (4), S. Gnanakaran (5), A. Lapedes (6), G. Learn (7), E. Kreider (8), Y. Li (9), G. Shaw (10), B. Hahn (11), D. Montefiori (12), S. Alam (13), M. Bonsignori (14), M. Moody (15), H. Liao (16), F. Gao (17), B. Haynes (18) *LANL University of Pennsylvania Duke University HIV vaccine efforts inspired by studying broadly neutralizing antibody (bnAb) ontogeny characterize the coevolution of virus and B‐cell repertoire during development of neutralization breadth. Previous ontogeny studies have revealed that viral diversification precedes neutralization breadth, which suggests antigenic diversity may be necessary to acquire breadth. While bnAb ontogeny studies sequence hundreds to thousands of virus variants from one donor, it is feasible to construct Envs for detailed analysis for only a fraction of these. To address this problem, we present a method to select a set of variants that represents antigenic diversity among longitudinal sequences. It uses loss of the transmitted/founder (TF) virus to identify sites that change rapidly. In subjects sequenced over the first 3‐5 years after infection by a single virus, while developing plasma neutralization breadth, 80‐100% TF loss in any time‐point identified Env positions as selected sites. An algorithm then selected Envs to represent all recurrent mutations in the selected sites. Using 398 single‐genome amplification (SGA)‐derived Envs that spanned three years of infection, TF loss identified 35 sites under putative immune selection in the African individual CH505. Encouragingly, these sites corresponded to two autologous antibody epitopes, CH103 and DH227. The algorithm identified 54 Envs that represent all recurrent mutations in selected sites. The Envs were well dispersed throughout the phylogeny and represented the development of binding and neutralization in a set of 124 previously hand‐picked Envs. The algorithm chooses sequence sets with more recurrent mutations and less redundancy than would be chosen randomly or by hand. Thus, the algorithm objectively provides a minimal, manageable number of Envs to represent diversity in natural infection, to help study virus‐antibody coevolution. Designating this minimal representation of antigenic diversity as an “antigen swarm”, the results suggested a swarm vaccine concept, an immunization strategy comprised of serial administration of increasingly diverse antigens. 46 CLUSTER RECONSTRUCTION IN SIMULATED TRANSMISSION CHAINS *M. Ragonnet‐Cronin (1), E. Hodcroft (2), S. Lycett (3), M. Hall (4), O. Ratmann (5), C. Fraser (6), A. Leigh Brown (7) 1. University of Edinburgh 2. University of Edinburgh 3. University of Edinburgh 4. University of Edinburgh 5. Imperial College London 6. Imperial College London 7. University of Edinburgh Inferring HIV transmission from phylogenetic analysis of HIV sequences is common but it is unclear how well phylogenies approximate true transmission chains. As part of the PANGEA project, the Discrete Spatial PhyloSimulator was used to simulate a heterosexual HIV epidemic. The network included high and low risk households and a small group of sex workers. Risk of transmission remained constant throughout an individual’s lifespan. A transmission phylogeny congruent with the transmission network was generated and pol sequences were simulated along the phylogeny using πBUSS. Sampling took place 53 years into the epidemic at sampling fractions of 100%, 60% (consistent with high coverage such as in the UK) and 20% were analysed. Phylogenetic trees were reconstructed in RaxML and dated using Least Squares Dating. Sequences were considered phylogenetically clustered if they shared a common ancestor in the phylogeny within the last 5 years. 4662 individuals were sampled, clustering into 641 clusters sized two and above in the true network. The mean number of onward transmissions was 2 with 17% of transmissions occurring within the first year and 32% within the first 2 years post‐infection. At 100% coverage, 86.7% of the phylogenetically clustered sequences represented nodes clustered in the true network. This high proportion was maintained at 60% coverage (86.4%) and dropped only slightly (81.7%) for the 20% sampled fraction. However the majority of these connections represent association with the same transmission chain rather than true transmission links and this effect is amplified as the sampling fraction decreases. We are currently investigating the effect of thinning reconstructed transmission networks to resolve the true sequence of transmission events within clusters. 47 PHYLODYNAMIC INFERENCE OF CONTACT NETWORK STRUCTURE *D. Rasmussen (1), T. Stadler (2) ETH Zurich We present a new modeling framework for understanding how sexual contact networks shape the phylogenic history of sexually transmitted pathogens like HIV. This framework is built upon a phylodynamic model that statistically describes how host network structure influences the shape of pathogen phylogenies by extending earlier coalescent models for infectious diseases to include local contact network structure. These phylodynamic models allow us to explore how network properties such as clustering, contact heterogeneity and assortative mixing shape tree structure. Moreover, this framework allows for network properties to be directly estimated from phylogenies using likelihood‐based inference in a way that takes into account how network structure and epidemiological dynamics interact to shape a pathogen’s phylogenetic history. We end by exploring how much we can infer about network structure from phylogenies simulated over a range of sampling densities characteristic of different HIV epidemics. 48 QUANTITATIVE METHOD FOR THE DELINEATION OF HIV‐1 SPECIFIC ANTIBODY EPITOPE REACTIVITY IN POLYCLONAL PATIENT SERA *N. Hoze (1), P. Rusert (2), C. Magnus (3), A. Trkola (4), R. Regoes (5) 1. Institute of Integrative Biology, ETH Zurich, Zurich, Switzerland 2. University of Zurich, Institute of Medical Virology 3. University of Zurich, Institute of Medical Virology 4. University of Zurich, Institute of Medical Virology 5. Institute of Integrative Biology, ETH Zurich, Zurich, Switzerland Upon HIV‐1 infection, the adaptive immune system generates a polyclonal antibody response. Determining the composition of epitope reactivities in polyclonal plasmas, especially of elite neutralizers, is of fundamental importance for identifying broad and potent antibodies and for development of anti‐HIV vaccines. Here we aimed at improving existing computational methods for the delineation of antibodies in plasma samples. Our method relies on the determination of neutralization fingerprints, which are derived by integrating data of inhibitory concentrations (IC50) of defined monoclonal antibodies against a diverse set of HIV‐1 isolates. In the past, these fingerprints have been successfully used to determine the plasma frequency of neutralizing antibodies directed against different antigenic regions (Georgiev et al, Science 2013). A drawback of the existing method is, that it solely relies on the ranks of the IC50 measurements for delineation, and thus is only able to extract the relative ratios of antibody reactivities and not the specific concentrations in a plasma sample. Here we developed a quantitative method that incorporates an extended fingerprint, making use of all the quantitative information contained in the neutralization profiles of each antibody‐virus pair. We model the combined effect of different antibodies by either Loewe additivity or Bliss independence, depending on whether antibodies target the same or different sites on the viral spike, respectively. We tested the performance of our method on simulated sera and found that our method predicts the relative ratios of monoclonal antibodies in these sera more accurately than the method of Georgiev et al. Additionally, our method is able to retrieve not only relative ratios of antibodies in a plasma sample, but also their concentrations. Our study makes an important further step towards exploiting the maximum available information for delineating patient plasmas. 49 50 POSTERS 51 52 1 STRUCTURED OBSERVATIONS REVEAL SLOW HIV‐1 CTL ESCAPE H. Roberts (1), J. Hurst (2), N. Robinson (3), H. Brown (4), P. Flanagan (5), L. Vass (6), S. Fidler (7), J. Weber (8), A. Babiker (9), R. Phillips (10), A. McLean (11), J. Frater (12) 1. Nuffield Department of Clinical Medicine, University of Oxford 2. Nuffield Department of Clinical Medicine, University of Oxford 3. Nuffield Department of Clinical Medicine, University of Oxford 4. Nuffield Department of Clinical Medicine, University of Oxford 5. Nuffield Department of Clinical Medicine, University of Oxford 6. Nuffield Department of Clinical Medicine, University of Oxford 7. Division of Medicine, Wright Fleming Institute, Imperial College, London 8. Division of Medicine, Wright Fleming Institute, Imperial College, London 9. Medical Research Council Clinical Trials Unit, London 10. Nuffield Department of Clinical Medicine and Oxford NIHR Comprehensive Biomedical Research Centre and The Oxford Martin School, University of Oxford 11. Department of Zoology and Oxford Martin School, University of Oxford 12. Nuffield Department of Clinical Medicine and Oxford NIHR Comprehensive Biomedical Research Centre and The Oxford Martin School, University of Oxford The existence of viral variants that escape from the selection pressures imposed by cytotoxic T‐lymphocytes (CTLs) in HIV‐1 infection is well documented, but it is unclear when they arise, with reported measures of the time to escape in individuals ranging from days to years. A study of participants enrolled in the SPARTAC (Short Pulse Anti‐Retroviral Therapy at HIV Seroconversion) clinical trial allowed direct observation of the evolution of CTL escape variants in 125 adults with primary HIV‐1 infection observed for up to three years. Patient HLA‐type, longitudinal CD8+ T‐cell responses measured by IFN‐γ ELISpot and longitudinal HIV‐1 gag, pol and nef sequence data were used to study the timing and prevalence of CTL escape in the participants whilst untreated. Results showed that sequence variation within CTL epitopes at the first time point (within six months of the estimated date of seroconversion) was consistent with most mutations being transmitted in the infecting viral strain rather than with escape arising within the first few weeks of infection. Escape arose throughout the first three years of infection, but slowly and steadily ‐ even in an HLA matched host who mounted a measurable, specific, CD8+ T‐cell response the average time before the targeted epitope evolved an escape mutation was longer than two years. Survival analysis of time to escape found that possession of a protective HLA type significantly reduced time to first escape in a patient (p = 0.01), however there was no direct evidence linking early escape to progression in these patients. Having accounted for how our definition of escape may have affected our results, we conclude that escape in these patients is less frequent than reported elsewhere because studying data from a trial cohort has allowed us to pay proper attention to those patients in whom nothing much happens. 53 2 NINE YEARS EXPERIENCE OF HIV‐ DRUG RESISTANCE MONITORING IN SOUTH AFRICA, 2006 ‐ 2014 G. Jacobs (1), M. Claassen (2), G. Van Zyl (3), W. Preiser (4), S. Engelbrecht (5) Division Of Medical Virology, University of Stellenbosch, South Africa HIV‐1 has been a major health problem for more than 20 years in South Africa, with an approximate 6.3 million infected individuals currently alive of whom 2.7 million are receiving antiretroviral therapy. At the Division of Medical Virology, Tygerberg Academic Hospital, Stellenbosch University and NHLS, we established an in‐house assay to detect HIV‐1 resistance associated mutations. Briefly, a 1.2 kb HIV‐1 pol fragment containing the major known PR and RT mutation sites are amplified and sequenced through Sanger Sequencing and analysed for resistance mutations with the Stanford HIV database. Since the implementation of the South African national ART rollout programme in 2004 we have seen marked increases in the number of requests not only for viral load tests to monitor ART, but also for genotypic HIV‐1 drug resistance tests. More than 7000 HIV‐1 sequences were generated and analysed from 2006 to 2014. In 2014 alone, 1582 samples were received from 145 different clinical sites across the country including adults, children and infants. Many of these samples are from patients failing first‐line ART. The most common NRTI mutations observed amongst our study population are M184V, K65R and Y115F, while the most common NNRTI mutations are V106M and Y181C. Approximately 97.6% of HIV‐1 pol sequences analysed belong to HIV‐1 subtype C. However, at least 1.2% of HIV‐1 sequences remain unassigned and difficult to characterise. We recently described BC URFs in the Western Cape Province of South Africa. In addition, we have now identified a possible new previously uncharacterised HIV‐1 CD CRF amongst at least 23 patients from our 2011 – 2014 cohort samples. The newly identified CRFs and URFs warrant further investigation as they have implications for diagnostic assays, viral molecular epidemiology and vaccine design. We will continue to record our observations to help strengthen and improve therapy outcomes. 54 3 THE RETROVIRUS INTEGRATION DATABASE (RID) W. Shao (1), J. Shan (2), M. Kearney (3), X. Wu (4), B. Luke (5), J. Coffin (6), S. Hughes (7) 1. Leidos Biomedical Research, Inc 2. Leidos Biomedical Research, Inc 3. HIV Drug Resistance Program, NCI 4. Leidos Biomedical Research, Inc 5. Leidos Biomedical Research, Inc 6. Tufts Univeristy 7. HIV Drug Resistance Program, NCI Retrovirus replication requires that the virus integrate a DNA copy of its genome into the host chromosomal DNA. Although there are numerous published studies that describe the distribution of retrovirus integration sites, there is no large publicly available centralized database that contains the available integration site information. It has been known for some time that, in specific cases, the site of integration of alpha, beta and gamma retroviruses can lead to tumor formation in animals, and it is now clear that certain HIV integration sites can enhance the growth and/or persistence of infected cells. Currently, most of the retrovirus integration site information is found in supplementary materials, which makes retrieving it for meta‐analyses difficult. Thus, a comprehensive database that includes information about integration sites is critically needed. We have built the NCI Retrovirus Integration Database (RID) to record integration site information for all retroviruses, including HIV‐1. RID is an in‐progress MySql based relational database. Briefly, it has tables to store host, virus and subtypes, sample/patient and tissue and demographics information without (for integration sites in humans) revealing personally identifiable information. Chromosome, integration site, associated genes, exon/intron information, provirus orientation, and the references from which the information was collected are provided on the database. Additionally, we built several tools into the database to facilitate mapping of the integration sites to USCS genome browser, to plot the integration site patterns on a chromosome, and to display provirus LTRs in their inserted genome sequence for PCR/probe design. We also created a robust, user friendly website that allows users to query the database and analyze the data dynamically. Currently, the database is mostly populated with HIV‐1 data. Other retroviruses, including endogenous proviruses, will be included. In conclusion, we have created a relational database to store comprehensive retrovirus integration information. This information will facilitate the retrieval and analysis of the published integration datasets. The database will be available for public use shortly. For a link, see http://www.retrovirus.info. 55 4 A SEQUENTIAL MONTE CARLO APPROACH FOR JOINT ESTIMATION OF VIRAL PHYLOGENY AND
EPIDEMIOLOGICAL DYNAMICS A. Smith (1), E. Ionides (2), A. King (3) 1. Department of Bioinformatics, University of Michigan, Ann Arbor, Michigan, USA. 2. Department of Statistics, University of Michigan, Ann Arbor, Michigan, USA. 3. Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, Michigan, USA. Center for the Study of Complex Systems, University of Michigan, Ann Arbor, Michigan, USA. Department of Mathematics, University of Michigan, Ann Arbor, Michigan, USA. A major challenge in the field of phylodynamics is how best to extract information on disease dynamics from patterns in viral gene sequences. Because the molecular evolution of HIV occurs on a similar timescale as disease dynamics, it is possible to infer fundamental epidemiological parameters from viral sequence data. A key component in this inference process is the link between models of disease dynamics and viral gene sequences: the viral phylogeny is the structure that forms this bridge.Current inference methods condition on phylogenetic reconstructions, and therefore operate in two disjoint steps: (1) estimation of the viral phylogeny based on some generic model of sequence evolution and (2) fitting of epidemiological models to properties of the estimated phylogeny. This approach may lead to biases of unknown magnitude because in general the model used to estimate the phylogeny and the model of disease dynamics disagree in their assumptions.Here, we develop a novel method for fitting models of disease dynamics to heterochronous viral sequences in a single procedure. The key innovation of our approach is coupling of phylogenetic techniques with partially observed Markov process models. By placing the inference problem in this framework, we are able to jointly estimate the phylogeny and the parameters of the dynamic model. We use sequential Monte Carlo to estimate the likelihood by integrating over the space of unobserved transmission trees, and iterated filtering to obtain maximum likelihood estimates of parameters. This approach comes at the cost of higher computational effort, but has the advantage of increased flexibility in consideration of different mechanistic models of disease.We apply this method to datasets of HIV‐1 subtype B partial pol sequences from covariate‐defined subpopulations of men who have sex with men in Detroit, MI, USA. We show that this approach allows for comparison of different dynamic models of disease and estimation of parameters relating to stage‐specific transmission, the influence of HAART, and transmission between subpopulations. 56 5 THERAPY STRATEGIES IN HIV INFECTION USING MATHEMATICAL MODELLING A. Boianelli (1), E. Hernández‐Vargas (2) Systems Medicine of Infectious Diseases, Helmholtz Centre for Infection Research, Inhoffenstraße 7, 38124 Braunschweig, Germany Patients infected with HIV receive prolonged combined antiretroviral treatment (ART), which increases the CD4+ T cell counts in blood and reduces the viral load to undetectable plasma levels. Nevertheless, the appearance of multi‐drug resistant genotypes may reduce the susceptibility to ARTs. Current HIV treatment guidelines seem poorly supported as practitioners have not achieved a consensus on the optimal time to switch antiretroviral treatments. A widely accepted strategy (which we refer to as ‘‘switch on virological failure’’) is to continue the current therapy until the viral load exceeds a fixed level (e.g. 1000–500 copies/ml). On the one hand, early switching risks poor adherence to a new drug regimen and prematurely exhausting the limited number of remaining salvage therapies. On the other hand, switching drugs too late allows the accumulation of multi‐drug resistant strains. Here, we propose different mathematical models to contribute to the discussion. Computational results suggest that in the absence of other practical constraints, switching rapidly between therapies is relevant to delay the viral escape. However, simulation scenarios when the therapy is not initiated during the first 3 years postinfection, switching in response to virological failure could outperform proactive switching strategies (alternate between two regimens every 3 months). In case that proactive switching is opted, the switching time between therapies should not be larger than 100 days. Further clinical evidences are needed to support these predictions. Future directions will focus on the application of a stochastic branching process to elucidate the key dynamics of CD4+ T cell memory reservoirs and evaluate the extinction probability for each reservoir subset using different treatment strategies. References: Hernandez‐Vargas, E.A., Colaneri, P., Middleton, R. Optimal therapy scheduling for a simplified HIV infection model. Automatica, 49, 2013 Haering, M., Hördt, A., Meyer‐Hermann, M., & Hernandez‐Vargas, E. A. Computational Study to Determine When to Initiate and Alternate Therapy in HIV Infection. BioMed Research International, 2014. 57 6 PREDICTED BINDING AFFINITIES OF HIV‐1C VIF, VPR AND VPU EPITOPES DURING PRIMARY INFECTION AGAINST HOST MHC CLASS I HLA‐A AND HLA‐B MOLECULES. R. Rossenkhan (1), I. MacLeod (2), T. Sebunya (3), R. Musonda (4), B. Gashe (5), V. Novitsky (6), M. Essex (7) 1. 1Department of Immunology and Infectious Diseases, Harvard T.H.Chan School of Public Health,, Boston, MA 02115, USA; 2Botswana Harvard AIDS Institute, Gaborone, Botswana; 3Department of Biological Sciences, University of Botswana, Gaborone, Botswana 2. 1Department of Immunology and Infectious Diseases, Harvard T.H.Chan School of Public Health,, Boston, MA 02115, USA; 2Botswana Harvard AIDS Institute, Gaborone, Botswana 3. 3Department of Biological Sciences, University of Botswana, Gaborone, Botswana 4. 1Department of Immunology and Infectious Diseases, Harvard T.H.Chan School of Public Health,, Boston, MA 02115, USA; 2Botswana Harvard AIDS Institute, Gaborone, Botswana 5. 3Department of Biological Sciences, University of Botswana, Gaborone, Botswana 6. 1Department of Immunology and Infectious Diseases, Harvard T.H.Chan School of Public Health,, Boston, MA 02115, USA; 2Botswana Harvard AIDS Institute, Gaborone, Botswana 7. 1Department of Immunology and Infectious Diseases, Harvard T.H.Chan School of Public Health,, Boston, MA 02115, USA; 2Botswana Harvard AIDS Institute, Gaborone, Botswana HIV‐1C is the most globally prevalent HIV subtype. Viral proteins Vif, Vpr, and Vpu interact with the host cell environment, and disable host antiretroviral defences. Studying these viral proteins during primary HIV‐1C infection could provide insights into viral post‐transmission dynamics and may help improve understanding of HIV‐1C pathogenesis in early infection. We evaluated predicted binding affinities of HIV‐1C Vif, Vpr and Vpu epitopes against host MHC class I HLA‐A and HLA‐B molecules for 17 HIV‐infected individuals using NetMHCpan 2.8. The HIV‐1C Vif, Vpr and Vpu for each patient were represented by MRCA reconstructed from earliest time point viral quasispecies (median 59 days p/s). Vif, Vpr and Vpu peptides in the susceptible form (non‐HLA adapted) were compared to the corresponding escape epitopes (HLA‐adapted) over the first 500 days post‐seroconversion. Previously described CTL epitopes across HIV‐1 Vif, Vpr and Vpu were derived from the LANL database and analyzed in the context of MHC class I HLA alleles restriction. We demonstrated that four known HLA‐associated regions developed one or more mutations during the follow‐up period. Mutations in anchor residues of restricted 9‐mer epitopes were hypothesized to occur more frequently, and would reduce binding affinity to a greater degree than mutations in non‐anchor residues. We included all possible epitope variations containing an escape mutation that altered HLA binding affinity, to avoid introducing bias by selecting only the epitopes with the strongest impact on HLA binding. We determined that primary infection mutations in HIV‐1C Vif and Vpr were more likely to occur in epitope anchor residues that resulted in significantly decreased epitope binding (predicted affinity). Dynamics of HIV‐1C accessory proteins during early stage infection highlight the need for further quantitative work on the impact of CTL escape mutations in HIV‐1C accessory proteins. 58 7 SUBTYPE‐SPECIFIC STRUCTURAL CHARACTERISTICS AND MOLECULAR DYNAMICS OF GLYCOSYLATED HIV‐1 GP120 PROTEINS N. Wood (1), O. Grant (2), E. Fadda (3), R. Woods (4), S. Travers (5) 1. UCT Computational Biology Group, Institute of Infectious Disease and Molecular Medicine, University of Cape Town Health Sciences Campus, Anzio Road, Observatory 7925, South Africa; Fax: +27 21 406 6068 2. Complex Carbohydrate Research Centre, University of Georgia, Athens, GA 30602, USA 3. Department of Chemistry; National University of Ireland, Maynooth; Maynooth; Ireland 4. Complex Carbohydrate Research Centre, University of Georgia, Athens, GA 30602, USA 5. South African National Bioinformatics Institute (SANBI), University of Western Cape, Private Bag X17, Bellville, 7535, South Africa Recent studies have shown that the human immune system can produce broadly cross‐neutralising (BCN) antibodies capable of neutralising a large spectrum of HIV strains. For many of these BCN antibodies, carbohydrates on the surface of the HIV‐1 gp120 glycoprotein play a key role, since their epitopes comprise either entirely, or partially, of carbohydrates (glycans). Using molecular dynamics (MD), we previously have shown that the presence of N‐linked glycans has a significant effect on the dynamics of the gp120 glycoprotein. However, these studies were limited due to the challenges associated with building 3D structures for densely glycosylated glycoproteins.Here, we present an approach that first explores the most populated rotamers of the Asn‐GlcNAc linkage and then, if necessary, adapts the carbohydrate structure to its environment by iteratively rotating the interglycosidic linkages within normal bounds. This automated process increases the number of carbohydrates that can be attached to a glycoprotein without the need for manual adjustment or energy minimisation. We have applied this method to investigate the structural differences in HIV‐1 gp120 subtype glycosylation profiles, using reference sequences of subtypes A1, A2, CRF01_AE, B, C, and D as model targets. Our results suggest that the high density of glycans on the surface of HIV‐1 gp120 causes mutual exclusion to occur, which may result in associated, and distinctive, glycan‐protein or glycan‐glycan interactions. Using MD we further present selected simulations that illustrate how different carbohydrate distributions influence the dynamics of the gp120 glycoprotein model, which may play a role in both coreceptor usage and in forming carbohydrate‐dependent epitopes.Our approach will facilitate the relatively unexplored use of MD simulations to investigate the changes in the dynamics of the HIV‐1 gp120 glycoprotein in different glycosylated states, and this, in turn, may help guide the development of engineered glycoproteins as immunogens for eliciting BCN antibody responses. 59 8 HIV‐1 GROUP M DIVERSITY: NEW INSIGHTS ON THE EVOLUTION OF THE VIRUS M. Tongo (1), J. Dorfman (2), D. Martin (3) 1. Division of Immunology, IDM, University of Cape Town 2. Division of Immunology, IDM, University of Cape Town 3. Bioinformatique Node, IDM, University of Cape Town The currently classified HIV subtypes and CRFs are representative of the viruses primarily responsible for the AIDS epidemic. They do not give a complete view of HIV‐1M diversity. There are presently many sequences that are too divergent to be placed within any existing subtype or CRF grouping, and certain recombinant viruses contain sequences of indeterminate origin, providing further evidence that HIV diversity is not fully represented under the current classification system. In an effort to understand the extent of HIV‐1M diversity, we analysed a set of 471 genomic sequences representing the full diversity of each HIV‐1M subtype. A blinded fully exploratory screen for recombination using RDP4 was performed with recombinant viruses being decomposed into their constituent parts. Recombination‐aware maximum likelihood (ML) phylogenetic analyses were then used to identify rare lineages, including those parental to some CRFs. To understand the differences between the established genome structures of these rare lineages and those described during our exploratory recombination screen, a re‐analysis of presently described CRFs was undertaken and ML methods were used to test whether segments of recombinationally derived sequence were from parental viruses falling within the existing subtype classification system or from novel divergent virus lineages. Phylogenetic analyses of mostly recombination‐free HIV‐1M sequences indicated that many parental sequences of CRFs including 06_, 13_, 18_, 25_, 27_and 49_cpx are not classifiable within the currently defined HIV‐1M subtypes. Furthermore, most of their recombinationally derived genome segments are apparently from parental viruses within novel divergent lineages. Interestedly, one of the supposedly subtype G–derived sequence within 27_cpx appears to have instead been derived by an ancestral subtype G from a virus in the 27_cpx lineage. This implies that there is potentially a far more diverse pool of HIV‐1M sequences circulating amongst humans than the classified subtypes and CRFs might suggest. 60 9 EVOLUTIONARY RATES OF HIV‐1 ACCESSORY GENES FROM FULL‐LENGTH DATASETS ACROSS SUBTYPES G. Yebra (1), A. Leigh Brown (2) University of Edinburgh Background: An increasing number of HIV full length genomes (FLG) are becoming available derived from next‐generation sequencing techniques. However, most molecular evolutionary studies focus on HIV structural genes and little is known about accessory genes. Methods: Nearly FLG (9 subtypes A1, 15 D) from historical samples taken in Uganda in 1986, together with East African FLG from GenBank (21 A1, 1985‐2011 / 37 D, 1990‐2011), were analysed with BEAST to estimate evolutionary rates of accessory genes (vif, vpr, vpu, nef) and compare them to structural genes (gag, pol, env). Also, 2 other FLG datasets with balanced sampling times were analysed: Southern Africa subtype C (n=100, 1989‐2011), and global subtype B (n=106, 1982‐2009). Results: We found variation in the evolutionary rates of accessory genes across subtypes, but on average gag presented a rate 1.3‐times higher than pol (t=3.1; p=0.02) and for env it was 1.5‐times higher than for gag (t=2.5; p=0.04) and 1.9‐times than for pol (t=4.0; p=0.007). After a Bonferroni correction for multiple t‐tests, only the comparison env versus pol remained significant (p=0.021). Vif (mean rate=2.17E‐3 [95%CI 1.4‐2.9E‐3] subtitutions/site/year) presented a similar rate to gag (2.52E‐3 [2.0‐3.0E‐3]) and pol (1.90E‐3 [1.5‐2.3E‐3]) whereas vpr (3.09E‐3 [1.2‐4.9E‐3]), vpu (3.63E‐3 [2.5‐4.7E‐3]) and nef (3.39E‐3 [2.2‐4.5E‐3]) were closer to env (3.68E‐3 [2.3‐5.0E‐3]) but varied greatly across the datasets. The estimations for the shortest genes (vpr (291nt) and vpu (247nt)), had large confidence intervals which represents a limitation for their evolutionary analysis. Conclusions: We analysed the evolutionary rates of the HIV accessory genes in 4 datasets corresponding to different subtypes and epidemics. According to our results, vif presents a rate more similar to that of gag and pol meanwhile vpr, vpu and nef are more similar to env. This information could be valuable when analysing small coding regions derived from FLG. Evolutionary rates for each dataset × 10‐3 (substitution/site/year) Gene Length A1 B C D Gag 1500nt 2.26 (1.80‐2.74) 2.61 (2.24‐2.96) 2.94 (2.53‐3.38) 2.26 (1.77‐2.80) Pol 1500nt 1.83 (1.52‐2.17) 2.18 (1.83‐2.51) 1.95 (1.79‐2.14) 1.62 (1.30‐1.96) Env 2500nt 2.54 (2.25‐2.87) 4.28 (3.73‐4.85) 4.39 (3.98‐4.84) 3.50 (2.81‐4.28) Vif 580nt 2.04 (1.58‐2.57) 2.79 (2.23‐3.30) 2.19 (1.60‐2.85) 1.65 (1.01‐2.33) Vpr 291nt 1.97 (1.33‐2.67) 3.27 (2.54‐4.00) 2.49 (1.85‐3.20) 4.63 (2.34‐7.16) Vpu 247nt 3.88 (2.79‐4.87) 3.14 (2.56‐3.73) 4.47 (3.23‐5.86) 3.02 (1.85‐4.35) Nef 618nt 4.29 (3.69‐4.89) 3.27 (2.83‐3.75) 3.47 (2.75‐4.32) 2.51 (1.85‐3.30) 61 10 THE CONTRIBUTION OF ANGOLA FOR THE EARLY SPREAD OF THE HIV‐1 EPIDEMIC A. Pineda‐Peña (1), J. Varanda (2), K. Theys (3), I. Bártolo (4), T. Leitner (5), N. Taveira (6), A. Vandamme (7), A. Abecasis (8) 1. 1. Global Health and Tropical Medicine, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisbon, Portugal 2. International Public Health and Biostatistics Unit and Microbiology Unit, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisbon, Portugal 3. Molecular Biology and Immunology Department, Fundación Instituto de Inmunología de Colombia (FIDIC) and Basic Sciences Department, Universidad del Rosario, Bogotá, Colombia. 4. KU Leuven ‐ University of Leuven, Department of Microbiology and Immunology, Rega Institute for Medical Research, Clinical and Epidemiological Virology, B‐3000 Leuven, Belgium Background: Angola borders and has long‐term links with Democratic Republic of Congo (DRC) and Republic of Congo (RC) as well as high levels of HIV‐1 genetic diversity. Therefore, we aimed to investigate the role of this country in the initial spread of the HIV‐1 pandemic. Methods: env and pol sequences and city locations from DRC, RC and Angola were retrieved from published articles and Los Alamos HIV database. We created three random datasets for env and two for pol based on time span, geography and viral diversity. Each env dataset comprised 148 sequences: 72 from DRC, 36 from RC and 36 from Angola, whereas each pol dataset included 134, 50 and 70 sequences, respectively. A non‐reversible phylogeographic model implemented in BEAST was used to infer a time‐calibrated phylogenetic history. The Bayesian Stochastic Search Variable Selection and robust counting analyses were performed to estimate the most significant pathways of viral dispersal. Results: The pandemic was originated in Kinshasa and the most recent common ancestor of group M was ~1918 (95% Bayesian Credible interval [1900‐1937]) using env, and 1910 [1886‐1931] using pol. Noteworthy, the earliest HIV‐1 introduction to Angola was in ~1936 [1917‐1954] using env and ~1937 [1920‐1954] using pol, whereas the earliest introduction to Brazzaville was around 1943 [1925‐1960] and ~1940 [1925‐1955], respectively. 30% of all HIV‐1 migrations out of Kinshasa in the late 1930’s to late 1950’s were directed to Luanda, as many as to Brazzaville. Other significant links with Angola included mainly the province of Cabinda but with fewer migrations. Conclusions: We show that a large amount of the initial migrations out of Kinshasa were directed to Angola, occurring as early as to Brazzaville. This study indicates a pivotal role of Angola in the early spread of the HIV‐1 epidemic, thus sharing the epicenter with DRC and RC. 62 11 COMPOSITE SEQUENCE‐STRUCTURE STABILITY MODELS AS SCREENING TOOLS FOR IDENTIFYING MUTATIONAL TARGETS FOR HIV DRUG AND VACCINE DEVELOPMENT S. Manocheewa (1), J. Mittler (2), R. Samudrala (3), J. Mullins (4) University of Washington Background. Rapid evolution and high sequence diversity enable Human Immunodeficiency Virus (HIV) populations to acquire mutations to escape antiretroviral drugs and host immune responses, and thus are major obstacles for the control of the pandemic. One strategy to overcome this problem is to focus drugs and vaccines on regions of the viral genome in which mutations are likely to cripple function through destabilization of viral proteins. Studies relying on sequence conservation alone have had only limited success in determining critically important regions. Results. We tested the ability of two structure‐based computational models to assign sites in the HIV‐1 CA that would be refractory to mutational change. These models predicted that about 20% of all 4,142 theoretically possible single amino acid variants would destabilize the mature CA. Such destabilizing mutations were rarely found in a database of 5811 HIV‐1 CA coding sequences, with none being present at a frequency greater than 2%. More importantly, we found that 90% of variants with high instability scores (from a set of 184 CA variants whose replication fitness or infectivity has been studied in vitro) had aberrant capsid structures and reduced viral infectivity. Conclusions. We showed that a joint scoring metric that takes into account both sequence conservation and protein structure stability, performed better at identifying deleterious mutations than sequence conservation or structure stability information alone. This metric may be useful for identifying targeting sites for drug and vaccine development. 63 12 LONG‐RANGE HIV GENOTYPING USING PROVIRAL DNA FOR ANALYSIS OF HIV DRUG RESISTANCE AND HIV TRANSMISSION DYNAMICS V. Novitsky (1), M. Zahralban‐Steele (2), M. McLane (3), S. Moyo (4), E. Widenfelt (5), S. Gaseitsiwe (6), M. Essex (7) 1. Harvard School of Public Health 2. Harvard School of Public Health 3. Harvard School of Public Health 4. Botswana‐Harvard AIDS Institute 5. Botswana‐Harvard AIDS Institute 6. Botswana‐Harvard AIDS Institute 7. Harvard School of Public Health HIV genotyping is a critical tool for antiviral drug‐resistance testing and has been used in research on HIV transmission clusters and virus transmission dynamics. The protocol developed by Gall et al. (J Clin Microbiol 2012;50:3838‐44) has enabled high‐throughput genotyping of near full‐length HIV‐1 genomes in individuals infected with multiple HIV‐1 subtypes. The protocol by Gall et al. targets viral RNA, and is very robust and reproducible when HIV‐1 RNA load is high (e.g., above 10,000 cps/ml). However, specimens with levels of HIV‐1 RNA below 1,000 cps/ml present a substantial challenge, and few of those samples can be genotyped. We developed a robust methodology for long‐range HIV genotyping using proviral DNA. This approach allows genotyping in HIV‐infected subjects with suppressed viral load, such as individuals on ART. Using proviral DNA as a template, long‐range HIV genotyping was successfully performed in a BCPP cohort with a high proportion of virologically suppressed individuals, with a success rate of about 90%. Spanning the HIV‐1 pol and env genes, this long‐range HIV genotyping technique enables analysis of drug resistance (both transmitted and acquired) for all major groups of ARVs, including protease inhibitors, NRTI, NNRTI, integrase strand transfer inhibitors, and virus entry inhibitors. The comprehensive phylogenic analysis is enhanced by substantial sequence length and a large number of informative sites, as two amplicons span over ~7,120 bp and cover about 80% of the unique HIV‐1 genome sequence, leading to greater refinement of viral linkage in HIV prevention studies. Long‐range HIV genotyping using proviral DNA could facilitate studies in populations on ART when amplification from viral RNA is unsuccessful due to low levels of HIV‐1 RNA load. The method is cost‐effective (under $150 per subject by Sanger sequencing), and has the potential to help enable the scale‐up of public health HIV prevention interventions. 64 13 MODEST IMPROVEMENTS TO HIV TREATMENT AND CARE COULD PREVENT HALF OF ALL NEW HIV INFECTIONS AMONG MEN HAVING SEX WITH MEN: A PHYLOGENETIC STUDY O. Ratmann (1), A. van Sighem (2), D. Bezemer (3), A. Gavryushkina (4), P. Reiss (5), C. Fraser (6) 1. Imperial College London 2. Stichting HIV Monitoring Foundation 3. Stichting HIV Monitoring Foundation 4. University of Auckland 5. Stichting HIV Monitoring Foundation 6. Imperial College London To better target HIV prevention efforts amongst men having sex with men (MSM), it is critical to quantify the proportion of HIV transmissions that originate throughout the HIV infection and care cascade, from undiagnosed to treated individuals. We conducted a combined analysis on molecular genetic and clinical data from HIV infected individuals in the Dutch ATHENA cohort between 1996 and 2013. Using viral evolutionary analyses, we determined potential transmitters to 601 recipient MSM that were diagnosed with recent HIV infection up to December 2010. Using clinical data, we associated treatment cascade stages with potential transmission intervals. 1,660 person‐years of potential transmission intervals were associated with phylogenetic evidence for direct HIV‐1 transmission. Between 96/07‐10/12, the estimated proportion of transmissions from undiagnosed men was 69% (95% confidence interval: 61‐78%), 24% (18%‐29%) from diagnosed, untreated men, and 7% (3%‐12%) from men that initiated ART. ART is highly effective in preventing HIV transmission amongst men, with a relative transmission risk of 6% (2%‐9%) compared to diagnosed untreated men with CD4 > 500 cells/mm3. Of the transmissions from undiagnosed men, at least 41% (36%‐45.7%) are estimated to happen within the first year of HIV infection. Therefore, annual, universal HIV testing could only have averted an estimated 23% (17%‐28%) of all infections between 09/07‐10/12. Oral pre‐exposure prophylaxis (PrEP), offered to and taken by the 33% of all uninfected men with a last negative test, could have averted an estimated 27% (17%‐39%) of all of these infections. This proportion is estimated to rise above 50% if 50% of all men would have used oral PrEP, in combination with expanded ART provision. Phylogenetic analysis indicates that half of the infections amongst Dutch MSM between 09/07‐10/12 could have been averted with a feasible combination prevention strategy that centers on the use of oral PreP by 50% of all men. 65 14 WILL HIV VANISH OR EVOLVE RESISTANCE FACING MASS ANTI‐RETROVIRAL TREATMENT? E. Geidelberg (1), S. Alizon (2) 1. Erasmus Mundus Master Programme in Evolutionary Biology 2. CNRS, Montpellier, France HIV drug resistance is a major clinical issue and several studies have focused on the factors that can lead to its evolution over the course of an infection. In comparison, the epidemiology of HIV drug resistance has attracted less attention, arguably because the rates of transmitted drug resistance (TDR) we observe are often low. However, until recently, high treatment coverage had mostly been witnessed in countries with high income economy, where individuals are followed over time and second or third line drugs are available. With the extension of the access to highly active antiretroviral therapy (HAART) to low income countries, one can ask whether TDR will remain so rare. To tackle this question, we developped a so‐called 'nested model', which describes both within‐host dynamics and epidemiological dynamics. More precisely, we build a stochastic within‐host model, the main outcome of which is the time to fixation of the drug resistant mutant. This result is then incorporated into an epidemiological model. Our main result is that the evolution of resistance at the between‐host level depends mostly on within‐host parameters (such as the mutation rate or the cost of resistance) rather than on between‐host parameters (such as the heterogeneity in numbers of partners in the population). Finally, all our epidemiological models yield two classes of outcomes with either a mostly resistant epidemics or a mostly sensitive epidemics. 66 15 ACCURATE DETECTION OF MINOR MUTATIONS WITH DEFINED SAMPLING DEPTH AND ERROR CUTOFF USING NEXT GENERATION SEQUENCING S. Zhou (1), L. Ping (2), C. Mccann (3), R. Swanstrom (4) 1. University of North Carolina at Chapel Hill, Chapel Hill, NC, USA Knowing sampling depth and accurate assessment of sequencing errors is critical for viral population studies using next generation sequencing (NGS). We have developed an improved approach of tagging viral RNA templates with a block of degenerate nucleotides (Primer ID) combined with MiSeq sequencing. We used 8E5 cell supernatants as a source of homogeneous virion RNA templates to measure the residual error rate using the Primer ID approach to be around 1 in 10,000 consensus nucleotides. With the error rate measured from the 8E5 RNA control and with the Poisson distribution we can identify the cutoff to accurately identify drug resistance mutations at low abundance. In addition, we generated a MiSeq library covering 5 regions of HIV‐1 genome from one plasma sample of an HIV‐1 subtype C infected subject by using a mixture of cDNA primers (with Primer ID) and a mixture of forward PCR primers in one reaction. We used 4,000 copies of RNA templates estimated by the viral load test. The 5 regions were p17 (HXB2 numbering 790‐1303), protease/RT (HXB2 numbering 2166‐2635), intergrase/vif (HXB2 numbering 4993‐5495), gp120 (HXB2 numbering 6207‐6572), and gp41/nef (HXB2 numbering 8257‐8767). The total sequenced regions were around 2,000 base pairs in length and the average number of consensus sequences (sampling depth) was 790 (range: 392‐1281). We built neighbor‐joining trees and calculated Pi, Tajima’s D score, and ds/dn for the 5 regions sequenced. We also found that the viral population in this subject had two minor drug resistance mutations at protease coding domain D30N (0.4%) and M46I (0.2%), both of which were above the cutoff defined by the error rate from the 8E5 control and Poisson distribution. Conclusions: Primer ID combined with NGS provides an approach to detect minority mutations with defined sampling depth of templates and accurate abundance cutoff to account for errors. 67 16 LINKAGE OF FEMALE FOUNDER HIV‐1 POPULATIONS TO TRANSMITTING MALE’S BLOOD AND SEMINAL VIRUSES C. Williams (1), M. Campbell (2), R. Payant (3), K. Yuhas (4), R. Coombs (5), J. Hughes (6), J. Mullins (7), J. Lingappa (8), L. Frenkel (9) 1. Seattle Children's Research Institute 2. University of Washington 3. Seattle Children's Research Institute 4. University of Washington 5. University of Washington 6. University of Washington 7. University of Washington 8. University of Washington 9. University of Washington, Seattle Children's Research Institute Background: Few studies have compared HIV‐1 founder populations to those of the transmitting partner’s blood and genital viruses. This study compared the genetic characteristics of transmitting males’ blood and genital viruses to the females’ founder virus. Methods: Twenty‐nine serodiscordant heterosexual couples from East and Southern Africa were studied following sequence‐confirmed male‐to‐female HIV‐1 transmission. Blood plasma RNA, seminal DNA and RNA, and cervical nucleic acids from specimens collected within six months of transmission had single‐template‐derived env gp160 amplicons from each tissue were sequenced. Maximum‐likelihood trees were generated and genetic diversity and divergence from the imputed female founder to their partner’s viruses were compared. Results: Analyses of 8 couples with a median=13 sequences (range 0‐36) from both blood and genital tissues. Seven seroconverting females had low median genetic diversity across tissues (median=.4% IQR 0.3‐0.5), consistent with a single founder, while the eighth woman had two founder strains (median=2.0%). The genetic diversity of transmitter blood plasma RNA, seminal plasma RNA and cell‐associated semen DNA were similar, with medians of 2.0% (IQR 1.5‐3.1), 1.4% (IQR 0.7‐1.8) and 2.2% (IQR 1.5‐3.2), respectively, and 2.8% (IQR 1.4‐3.3) across all tissues. The imputed MRCA of the seroconverting female (taken as the founder virus) was nearest to the transmitting male partner’s seminal RNA in 4, seminal DNA in 1 and blood plasma RNA in 3. However, due to the lack of consistent tissue compartmentalization, tissue sources were not clearly assigned. Conclusions: Consistent with a genetic bottleneck, the founding female viral populations appear to have arisen from a single viral variant, with one exception. Female founder viruses were not consistently more closely linked to her male partner’s genital versus blood‐derived viruses, most likely due to the absence of significant compartmentalization of male’s genital and blood viruses. These findings suggest that analysis of blood viruses would suffice for vaccine design. 68 17 COMPARISON OF MAJOR AND MINOR VIRAL SNPS IDENTIFIED BY SINGLE TEMPLATE SANGER AND PYROSEQUENCING IN EARLY HIV‐1 INFECTION S. Iyer (1), E. Casey (2), H. Bouzek (3), M. Kim (4), W. Deng (5), B. Larsen (6), H. Zhao (7), R. Bumgarner (8), M. Rolland (9), J. Mullins (10) 1. University of Washington 2. University of Washington 3. University of Washington 4. University of Washington 5. University of Washington 6. University of Washington 7. University of Washington 8. University of Washington 9. Walter Reed Army Institute of Research 10. University of Washington Massively‐parallel sequencing technologies, such as 454‐pyrosequencing, allow the identification of sequence variants in populations at lower levels than consensus sequencing and most single‐template Sanger sequencing experiments, but there is little data that comprehensively compares results from these methods. Single nucleotide polymorphisms (SNPs) were assessed using both Sanger and 454‐pyrosequencing on plasma sample obtained during early HIV‐1 infection from 32 subjects participating in the STEP vaccine trial. Pyrosequences encompassing 50% of the viral genome (gag, env, nef), were compared to a median of five individually‐amplified full‐length viral genomes sequenced using Sanger technology. The consensus nucleotide sequences were identical in 27 of the subjects; among the remaining five subjects, disagreements were found in <1% of the sites evaluated (out of nearly 117,000 sites across all subjects). The majority of the SNPs observed only in pyrosequences represented <2% of the subject’s viral sequence population. Over 89% of all minor SNPs observed only in pyrosequences were found at a frequency below the detection threshold of Sanger sequencing for that subject. When only genomic positions with at least the same number of reads as the mean number of HIV templates sequenced were considered, the number of minor SNPs found only in pyrosequences was reduced by 50%, and the number of SNPs observed by both technologies was reduced by 15%. In summary, 454‐pyrosequencing did not reveal the presence of additional founder viruses present at low levels in the population. These results provide guidance regarding the design, utility and limitations of sequencing of variable template sources, and emphasize parameters for improving the interpretation of massively‐parallel sequencing data to address important questions regarding target sequence evolution. 69 18 FITNESS‐BALANCED ESCAPE DETERMINES RESOLUTION OF DYNAMIC FOUNDER VIRUS ESCAPE PROCESSESS IN HIV‐1 INFECTION J. Sunshine (1), B. Larsen (2), B. Maust (3), E. Casey (4), W. Deng (5), L. Chen (6), D. Westfall (7), M. Kim (8), H. Zhao (9), S. Ghorai (10), E. Lanxon‐Cookson (11), M. Rolland (12), A. Collier (13), J. Maenza (14), J. Mullins (15), N. Frahm (16) 1. Fred Hutchinson Cancer Research Center 2. University of Washington 3. University of Washington 4. University of Washington 5. University of Washington 6. University of Washington 7. University of Washington 8. University of Washington 9. University of Washington 10. University of Washington 11. University of Washington 12. Walter Reed Army Research Institute 13. University of Washington 14. University of Washington 15. University of Washington 16. Fred Hutchinson Cancer Research Center Investigations into host‐virus interactions during primary HIV‐1 infection may inform the design of future vaccine immunogens. To understand the interplay between host cytotoxic T‐lymphocyte (CTL) responses and the mechanisms by which HIV‐1 evades them, we studied viral evolutionary patterns associated with host Gag CTL responses in six linked transmission pairs. HIV‐1 sequences corresponding to full‐length p17 and p24 gag were generated by 454 pyrosequencing for all pairs near the time of transmission and seroconverting partners were followed for a median of 847 days post infection. T‐cell responses were screened by IFNg/IL‐2 FluoroSpot using autologous peptide sets reflecting any Gag variant present in at least 5% of sequence reads in the individual’s viral population. We detected a median of 2 responses in the transmitting partners (range 1‐4, 11 total) and 5 in the seroconverting partners (range 1‐5, 20 total), with 13 of the 20 Gag‐responses in the seroconverters evolving over the study period. CTL escape processes were highly dynamic and subject‐specific, while reversion processes did not appear to play a strong role. When multiple escape mutations developed in a targeted epitope, the variant that ultimately became fixed in the viral population had the least fitness cost to the virus. When multiple mutations within an epitope achieved fitness‐balanced escape, these dynamic escape processes did not resolve, but were instead maintained in the viral population. Additional mutations not detected in the viral population were engineered to confer escape, and were shown to incur high fitness costs, suggesting that functional constraints limit the available sites tolerable to escape mutations. These results further our understanding of the impact of CTL escape and reversion from the founder virus in HIV and contribute to the identification of immunogenic Gag regions most vulnerable to a targeted T‐cell attack. 70 19 DEEP SEQUENCING ANALYSIS OF HIV‐1 TRANSMISSION AND SUBSEQUENT EVOLUTION IN SIX TRANSMISSION PAIRS B. Larsen (1), E. Casey (2), B. Maust (3), J. Sunshine (4), W. Deng (5), L. Chen (6), D. Westfall (7), M. Kim (8), H. Zhao (9), T. Sibley (10), B. Hall (11), N. Frahm (12), J. Maenza (13), A. Collier (14), J. Mullins (15) 1. University of Washington 2. University of Washington 3. University of Washington 4. Fred Hutchinson Cancer Research Center 5. University of Washington 6. University of Washington 7. University of Washington 8. University of Washington 9. University of Washington 10. University of Washington 11. University of Washington 12. Fred Hutchinson Cancer Research Center 13. University of Washington 14. University of Washington 15. University of Washington A single founder virus initiates fulminate in most HIV‐1 infections. What leads to the emergence of this variant is important to vaccine development. We studied the emergence of founder virus(es) through study of early HIV‐1 populations in six seroconverting MSM and their infecting partners. Massively parallel 454/pyrosequencing was performed over ~half of the viral genome (4.7kb) corresponding to most of the gag, pol, and env genes. A median of 145 viral genome templates were sequenced from the blood plasma a median of 36 days following infection from the seroconverting partner (SP), as were 255 templates from the transmitting partners (TP) a median of 41 days after transmission. An average of eight time points were sequenced in the SP (follow‐up period 350‐1250 days). The five SP with single founder variants had a total of 269 minor variants, 32 of which were found at some frequency in the TP’s viral population. Six percent of AA within gag that evolved in the SP were within peptides recognized by CTL in the TP, while 47% were located in peptides recognized by the SP. We found little relationship between sites that evolved during follow‐up and the database frequency of those amino acids, suggesting a limited role of reversions towards an ancestral viral state in this early period post‐transmission. Prior analyses of these transmission pairs using Sanger sequencing of ~10 full‐length viral genomes found that founder viruses were not the most abundant variant in the host but were not atypical either. Together, our results are consistent with the hypothesis that the founder quickly evolves from the transmitted strain(s) in the new host. 71 20 FINE STRUCTURE GENETIC ANALYSIS OF INTRA‐PATIENT HIV POPULATIONS PRIOR TO ANTIRETROVIRAL THERAPY USING NEXT GENERATION SEQUENCING J. Hattori (1), V. Boltz (2), W. Shao (3), M. Kearney (4), F. Maldarelli (5) 1. HIV Drug Resistance Program, National Cancer Institute, Frederick, MD 2. HIV Drug Resistance Program, National Cancer Institute, Frederick, MD 3. Leidos Biomedical Research, Frederick, MD 4. HIV Drug Resistance Program, National Cancer Institute, Frederick, MD 5. HIV Drug Resistance Program, National Cancer Institute, Frederick, MD HIV infection proceeds in vivo with the development of genetically diverse populations. Measuring the genetic characteristics of HIV populations, such as diversity, replicating population size, and recombination is essential to understand population behavior, and the spread of new variants in response to immune or drug selection. Previously, we and others used single genome sequencing (SGS) analysis of HIV obtained from plasma of infected untreated individuals to characterize population structure. By analyzing HIV pro‐pol, we previously reported relatively high replicating populations sizes (>1X104 infectious units/patient) using allele based approaches (Maldarelli et al., J. Virol. 87:10313; 2013); as we described, such measurements represent minimum estimates, and limited by sample size, which, for SGS, is typically in the range of 20‐50 sequences/sample. New approaches to directly measure population parameters would be useful to provide more accurate measurements of population parameters. Next generation sequencing provides greater sampling of populations, and, combined with the use of primer identification strategies (Jabara, et al., PNAS 108:20166; 2011) and rigorous bioinformatic processing to remove potential PCR recombination errors, yields large and reliable datasets suitable for detailed analysis. We used paired end sequencing (Illumina) to obtain sequences from c. 250 nt region of HIV pol from longitudinal plasma samples obtained from infected individuals (N=3) prior to introduction of antiretroviral therapy. A total of 9239 sequences were obtained with a median 787 sequences/time point (range 98‐3751 sequences). HIV diversity (percent average pairwise difference) ranged from 2.3‐3.2%. Using these larger sample sizes to determine population sizes, we measured effective population sizes of 6.6X104 – 3.9X105 infectious units/patient). Detailed linkage analyses revealed specific polymorphic sites with high degree of linkage, suggesting complex population structure. Properly generated, next generation sequencing datasets represent powerful tools to investigate HIV populations in vivo. 72 21 WEBMUTCOR: A TOOL TO DISCOVER INTERACTIONS BETWEEN DRUG RESISTANCE MUTATIONS AND CYTOTOXIC‐T‐LYMPHOCYTE ESCAPE MUTATIONS IN THE HIV‐1 POL REGION W. Smidt (1) 1. University of Pretoria Correlated mutations are important in the understanding the dynamics mutation selection in the HIV Pol region due to drug resistance mutations. Beside providing cumulative resistance to antiretrovirals, minor mutations in HIV protease and reverse transcriptase often provide a mitigation factor to potential fitness impacts of drug resistance mutations. Furthermore, the Pol region can acquire mutations allowing escape from host cytotoxic‐T‐lymphocyte (CTL) responses together with corresponding mutations mitigating potential fitness impacts. However, the detection of correlation between two mutations can yield false positives due to correlations of the mutations in question with auxiliary mutations. To circumvent this issue, a tool has been developed and demonstrated here that can accurately and rapidly measure correlations between mutations in the HIV Pol region while taking into account the contribution of auxiliary mutations. A web interface to this tool called WebMutCor has been developed that can be freely accessed by the public. This tool allows the generation of graphs summarizing the interaction networks between mutations as well as providing statistical support for correlated mutations. Using this tool, the interactions between known drug resistance mutations and CTL escape mutations were discovered, shedding light on the dynamics between selection of drug resistance and CTL escape during HIV infection. 73 22 COMPARING ESTIMATED TIME OF HIV‐1 INFECTION OBTAINED BY BAYESIAN ANALYSIS AND THE BED ASSAY M. M. Lunar (1), A. Vandamme (2), J. Tomažic (3), P. Karner (4), T. D. Vovko (5), B. Pecavar (6), G. Volcanšek (7), M. Poljak (8), A. B. Abecasis (9) 1. Institute of Microbiology and Immunology, Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia 2. Clinical and Epidemiological Virology, Rega Institute for Medical Research, Department of Microbiology and Immunology, KU Leuven, Leuven, Belgium; Microbiology Unit and Global Health and Tropical Medicine, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisbon, Portugal 3. Department of Infectious Diseases, University Medical Center Ljubljana, Ljubljana, Slovenia 4. Department of Infectious Diseases, University Medical Center Ljubljana, Ljubljana, Slovenia 5. Department of Infectious Diseases, University Medical Center Ljubljana, Ljubljana, Slovenia 6. Department of Infectious Diseases, University Medical Center Ljubljana, Ljubljana, Slovenia 7. Department of Infectious Diseases, University Medical Center Ljubljana, Ljubljana, Slovenia 8. Institute of Microbiology and Immunology, Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia 9. International Public Health and Biostatistics Unit and Global Health and Tropical Medicine, Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisbon, Portugal Different methods allow us to estimate the point in history when HIV was transmitted between individuals. The purpose of this study was to compare the estimates obtained in a Bayesian phylodynamic approach and the AwareTM BEDTM EIA HIV‐1 Incidence Test (BED test) (Calypte Biomedical Corporation, Portland, Oregon). The BED assay, used to determine whether an infection was acquired recently, employs a 155‐day time interval following infection for characterization of a recent infection. For this reason the time of infection for individuals defined as being recently infected was thus estimated to be within a 155‐day interval prior to the time of sampling. Times to the most recent common ancestor (tMRCAs) were determined by the Monte Carlo Markov chain method available in the BEAST package v1.7.1, using a relaxed clock model with uncorrelated lognormal distribution and the Bayesian skyline coalescent model. These methods were applied for 3 different groups of patients for whom their source of infection was known: 2 pairs of patients, among which at least one of the two patients was determined as recently infected; and one trio of patients, for which all were determined as recently infected. We found that the tMRCAs obtained using the Bayesian approach were estimated earlier in time compared to estimates obtained by the BED assay. The median tMRCA dates estimated by Bayesian analysis were 0.5, 1.1 and 1.1 years before the date of infection estimated by the BED assay for transmission pair 1, pair 2 and the trio, respectively. The estimated tMRCA corresponds to the time of origin of the strain that gave origin to the infections of that cluster. This strain should therefore have been originated before the estimated time of infection of any of the patients in our cluster, somewhere along the branch that leads to that cluster and inside the body of the patient who transmitted that strain to one of the patients in the cluster. As such, our results indicate that the tMRCA estimates are consistent with the BED assay results. 74 23 EXPLORING TRANSMISSION DYNAMICS OF HIV IN RURAL KWAZULU‐NATAL, USING PHYLOGENETICS E. Wilkinson (1), F. Tanser (2), S. Danaviah (3), J. Manasa (4), D. Pillay (5), T. de Oliveira (6) Africa Centre for Health and Population Studies, UKZN, South Africa Background, Despite antiretroviral rollout, HIV incidence remains high in South Africa. We have observed significant social and geographical heterogeneity in HIV incidence in rural KwaZulu‐Natal (Tanser et al. Science 2013). We sought to utilise phylogenetics to better understand the drivers of ongoing transmission in this hyperendemic population. Further, we explore the utility of sampling a fraction of infections to discern changes in incidence, through phylodynamic approaches. Methodology, Viral load was measured in on 2,420 dried blood spots (DBS) testing positive in the population‐based surveillance of 2011 and 2012. 804partial HIV‐1 pol gene sequences were obtained from DBS samples with a viral load (VL) >10,000. Sequences were analysed against all reference strains from South Africa and manually edited prior to phylogenetic inference. The inferred phylogenies were analysed to identify clusters of low genetic diversity (≤ 0.03) corresponding to transmission clusters/pairs, defined by branch support > 95%. Clinical, demographic, socio‐economic and geographic characteristics of infected individuals were analysed against identified clusters to discern traits associated with transmission. Molecular clock and coalescent analyses were performed in a Bayesian framework order to identify the origin of the clusters and the rate of expansion of the epidemic. Results, A total of 74 transmission clusters were identified. The mean size of clusters was 2.39 (variation 2 ‐ 13). The mean age of women in clusters was 31.55 years and of men, 35.91 years. 24.8% of individuals within the genotypic clusters were recently infected though they comprise approximately 5 % of the cohort. GIS results support the scenario of ‘hotspots’ of transmission, while coalescent analyses provide further evidence that HIV incidence is decreasing in this population. No drug resistance clusters were identified, in spite of 5% of the sequences presenting at least one drug resistance mutation. Conclusions,These results demonstrate the underling complex nature of the dynamics of HIV transmission and that acutely infected patients may disproportionately contribute towards transmissions in the era of increase ARVs. The use of genotypic data coupled with detailed patient information can be used to identify and characterise HIV transmission events. However, no drug resistance transmission cluster has been identified. The use of genotypic data analysed within a phylodynamic framework also reflects the decreasing trends in incidence likely due to the scale‐up of treatment coverage in recent years. 75 24 INTERTYPE AND INTER SPECIES GENOMIC DIVERSITY OF HIV‐1 TAT I. Khandaker (1), C. Roy (2), H. Oshitani (3) 1. Department Of Virology, Tohoku University Graduate School Of Medicine, 2‐1 Seiryomachi, Aoba Ward, Sendai City, Japan‐9808575, Tel: 81‐22‐717‐8210, Fax: 81‐22‐717‐8211 2. Department Of Virology, Tohoku University Graduate School Of Medicine, 2‐1 Seiryomachi, Aoba Ward, Sendai City, Japan‐9808575, Tel: 81‐22‐717‐8210, Fax: 81‐22‐717‐8211 3. Department Of Virology, Tohoku University Graduate School Of Medicine, 2‐1 Seiryomachi, Aoba Ward, Sendai City, Japan‐9808575, Tel: 81‐22‐717‐8210, Fax: 81‐22‐717‐8211 Intra‐ and ‐interspecies genomic diversity are the important determinants for the HIV‐1 viral evolution. Our study aims to investigate the genomic diversity of HIV‐1 Tat in different types, circulating recombinant forms (CRFs), and other species like Chimpanzee (CPZ). A total of 1991 sequences of different species and types of HIV‐1 Tat were obtained from the Los Alamos National Laboratory (LANL) compendium. Evolutionary distance, genomic diversity and phylogenetic tree analyses were performed. The mean overall amino acid divergence (%) among the analyzed sequences was 69%. Average evolutionary divergence over sequence pairs within the groups showed that CPZ group was highly divergent, and M group with CRFs showed almost similar divergence to type O. Evolutionary divergence over sequence pairs between groups revealed that group M with CRFs were closer to N group compared with others groups such as O, P, and CPZ. The mean (± SE) evolutionary diversity within Intra‐ and ‐interspecies level were 0.224 (±0.020) and 0.081 (± 0.013), respectively with a coefficient value of 0.266. The phylogenetic tree demonstrates that CRFs Tat sequences were well‐ immersed with their origin except CRF01_AE, CRF02_AG, CRF15_01B, CRF33_01B, CRF34_01B, CRF36_cpx, CRF48_01B, CRF52_01B, CRF53_01B, CRF54_01B, and CRF55_01B, which together formed a separate cluster. Interestingly, recently evolved CPZ strains were close to types O and P. Moreover, CRFs of subtypes B and C evolved rapidly with a high genetic divergence rate. In summary, our study findings on HIV‐1 Tat genomic diversity contribute to better understanding on HIV‐1 evolution. 76 25 SPATIO‐TEMPORAL EVOLUTION OF HIV‐1 TAT IN SUBTYPES B AND C C. Roy (1), I. Khandaker (2), H. Oshitani (3) 1. Department Of Virology, Tohoku University Graduate School Of Medicine, 2‐1 Seiryomachi, Aoba Ward, Sendai City, Japan‐9808575, Tel: 81‐22‐717‐8210, Fax: 81‐22‐717‐8211 2. Department Of Virology, Tohoku University Graduate School Of Medicine, 2‐1 Seiryomachi, Aoba Ward, Sendai City, Japan‐9808575, Tel: 81‐22‐717‐8210, Fax: 81‐22‐717‐8211 3. Department Of Virology, Tohoku University Graduate School Of Medicine, 2‐1 Seiryomachi, Aoba Ward, Sendai City, Japan‐9808575, Tel: 81‐22‐717‐8210, Fax: 81‐22‐717‐8211 IV‐1 Tat is an internal regulatory protein having versatile role in viral transcription. The full‐length Tat is a 101‐amino acid protein encoded by two exons: exon‐1 and exon‐2. Here we studied the spatio‐temporal evolution of HIV‐1 Tat. We curated 493 and 280 full‐length HIV‐1 Tat sequences of globally circulating HIV‐1 subtypes B and C strains, respectively, deposited in Los Alamos National Laboratory web database. The data sets contained sequences from a total 45 countries of six different continents. We performed mutation, phylogenetic and selection pressure analyses using the bioinformatics tools. In mutation analysis, we found higher global diversity in subtype B Tat sequences than subtype C. Compared to subtype C, there was a high level of divergence in the auxiliary domain (residues 58 to 69) of Tat exon‐1 and also in exon‐2. In phylogenetic analysis, both subtypes B and C were found diffusely distributed and intermixed among globally circulating HIV‐1 variants indicating that divergent HIV‐1 strains from different parts of the world have been slowly but continuously emerging since its origin. In selection pressure analysis, we found following notable positively selected sites among both subtypes: Thr40, Ala58, His59, Asn61, Ser62, Thr64, His65, Ala67, Ser68, Leu69 and Ser70 in exon‐1 and Ser75, Pro77, Asp80, Pro81, and Ser87 in exon‐2. Notably, numerous positively selected sites were found in the Tat‐transactivation responsive RNA (TAR) interaction region. In exon‐1, the mean estimated nucleotide substitution rates for subtypes B and C were 1.53x10‐3 (95% HPD Interval: 1.09 x10‐3 to 2.08x10‐3) and 2.14x10‐3 (95% HPD Interval: 1.35 x10‐3 to 2.91x10‐3) per site per year, respectively. Taken together, our study results provide novel insight on the spatio‐temporal evolutionary dynamics HIV‐1 Tat. Amino acid substitutions found in our study accomplish in developing successful Tat based gene therapy, Tat‐TAR interaction inhibitors and vaccines. 77 26 DENTIFYING WITHIN‐HOST HIV‐1 SUBPOPULATIONS BY COSEGREGATION OF PROFILE HIDDEN MARKOV MODEL UPDATE VECTORS P. Edlefsen (1) 1. Fred Hutch Profile Hidden Markov Models (PHMMs) are statistical models of biological sequences related by common descent in a star‐like phylogeny. While the popular HMMER and SAM software packages for PHMMs do not implement true maximum likelihood estimation, the Profillic software does. A feature of applying true maximum likelihood methods for estimating the parameters of PHMMs is that at convergence, the Baum‐Welch update contributions from each sequence mutually cancel. The Baum‐Welch expectation maximization update procedure can be viewed as an "online" algorithm, with the geometric interpretation that at each iteration the current Profile HMM parameters are modified by adding the update vectors contributed by each sequence. Using these sequence‐specific update vectors (called "Alignment Profiles"), I show that subpopulations of HIV‐1 within‐host evolution (e.g. from multiple founders) can be automatically identified by clustering and analysis of cosegregation. I argue that HIV‐1 within‐host dynamics are better‐modeled by networks of speciating and recombining quasispecies than by trees of individual sequences, and that Profile HMMs are a good candidate for representing the quasispecies. 78 27 POPULATION‐LEVEL EVOLUTION OF HIV‐1 ENV STRUCTURE OVER THREE DECADES; A MULTIDIMENSIONAL APPROACH TO STUDY THE DYNAMICS OF COMPLEX PHENOTYPES Y. O’Malley (1), O. DeLeon (2), H. Salimi (3), N. Eichelberger (4), R. Puliadi (5), J. Stapleton (6), H. Haim (7) 1The University of Iowa, Carver College of Medicine, Microbiology, Iowa City, IA To characterize the changes that occurred in structure of the HIV‐1 envelope glycoproteins (Envs) over the past 3 decades we initiated a comprehensive population‐level and longitudinal study. A total of 996 Envs were isolated from plasma samples of 145 individuals provided between 1985 and 2012 in two HIV clinics. Structural properties were examined using a cell‐based ELISA system that measures binding of probes to the membrane‐bound Env trimer. Assay output is a continuous variable that reflects epitope integrity and exposure. The propensity of each epitope to diversify within an individual was determined by variability in probe binding to different Envs isolated from the same plasma sample. This was corrected for the total genetic distance between all Envs isolated from that sample. The average value for each epitope (from 194 plasma samples) was designated its Diversification Index (DI). Different epitopes have significantly different DIs. Importantly, the DI of each epitope correlates very well with its diversification in longitudinal samples and with its variability in the entire tested population. Thus, the propensity of structural features to diversify in the infected individual is conserved and likely a major determinant of epitope status in the population. To examine evolution of complex sets of epitopes each Env was assigned a feature vector describing its pattern of binding to (n) different probes. The n‐dimensional vectors of all Envs were used for measuring inter‐isolate distances. To correct for the propensity of each epitope to diversify all features (dimensions) were weighted according to their DIs. Distances between isolates thus represent probabilities of transition between complex states. High occupancy of specific structural states was observed that was greater than that predicted by calculations based on feature distribution. Several longitudinal samples demonstrated drift/shift patterns toward higher occupancy states. Increased dimensionality of these studies will likely improve models for immunogen/therapeutic design. *Hillel Haim, MD, PhD Department of Microbiology Carver College of Medicine The University of Iowa 51 Newton Rd, 3‐770 BSB Iowa City Iowa 52242 Tel: 319‐335‐9989 Fax: 319‐335‐9006 Email: Hillel‐[email protected] 79 28 REASSESMENT OF MOLECULAR EVOLUTION OF CRIMEAN–CONGO HEMORRHAGIC FEVER VIRUS BASED ON COMPLETE S AND M SEGMENT SEQUENCES M. Stanojevic (1), V. Nikolic (2), M. Siljic (3), G. Stamenkovic (4), A. Gligic (5) 1. University of Belgrade Faculty of Medicine, Belgrade, Serbia 2. University of Belgrade Faculty of Medicine, Belgrade, Serbia 3. University of Belgrade Faculty of Medicine, Belgrade, Serbia 4. University of Belgrade, Institute for Biological Research "Siniša Stankovic", Belgrade, Serbia 5. Institute of Virology, Vaccines and Sera Torlak, Belgrade, Serbia Crimean‐Congo hemorrhagic fever virus (CCHFV) is an arbovirus belonging to the genus Nairovirus of the Bunyaviridae family. It is a negative‐sense ssRNA virus with tri‐partite genome (segments small (S), medium (M) and large (L), the causative agent of Crimean–Congo hemorrhagic fever (CCHF), a tick‐borne infection known to be present in wide geographic area of Eurasia and throughout Africa. In Serbia, the disease is known to be present since 1954, with several epidemics (1970, 1988, 1989, 1995, 2001), whereas in 1977 the first CCHFV isolate was obtained. High level of CCHFV genetic diversity may be attributed to the long duration of viral co‐evolution with tick reservoirs and vertebrate hosts. Previous studies of the time of CCHFV lineages divergence were based on varying number of viral sequences and gave differing estimates of the time of the most recent common ancestor (tMRCA), ranging from several hundred to several thousand years ago. We present a phylogenetic study exploring temporal phylogeny and phylogeography of CCHFV based on all publicly available sequences, comprising 76 complete coding S and 64 complete coding M segment sequences, from 22 countries, collected in the time period 1956‐2014. Quartet puzzling analysis revealed strong phylogenetic signal in both datasets, whereas screening for recombination identified 4 M segment recombinants., that were excluded from further analysis. General topology of the S segment phylogenetic tree was congruent to previously described 6 major clades, while clustering was significantly different in the M segment, implying possible reassortment. tMRCA estimate for the S segment clades was dated 1460 years ago (95% HPD 615 – 2568), whereas for the M to 2451 years ago (95% HPD 219 – 3308), however, these estimates may be significantly hampered by the fact that we detected very weak temporal structure in the sequence data. Phylogeographic analysis placed the origin of both S and M segment sequences to Africa. 80 29 THE CHANGING FACE OF THE HIV‐1 SUBTYPE C EPIDEMIC IN SOUTH AFRICA – DETECTION OF UNIQUE SUBTYPES AND RECOMBINANT FORMS. S. Engelbrecht (2), M. Claassen (3), G. Jacobs (4), G. van Zyl (5), W. Preiser (6) 2. Stellenbosch University / NHLS 3. NHLS 4. Stellenbosch University 5. Stellenbosch University / NHLS 6. Stellenbosch University / NHLS Background: South Africa is experiencing one of the most devastating HIV‐1 epidemics in the world with more than 6.3 million people infected, the majority with HIV‐1 subtype C. An estimated 2.7 million people are currently receiving antiretroviral treatment. The NHLS virology laboratory at Tygerberg Hospital screened almost 7000 patient samples for HIV‐1 genotypic drug resistance testing between 2006 and 2014. Aim: The aim of this study was to investigate HIV‐1 subtype diversity in these samples obtained from 168 different locations in Gauteng, KwaZulu‐Natal, Free State, Western and Eastern Cape Provinces. Methods: Viral RNA was isolated, the protease (PR) and partial reverse transcriptase (RT) regions of the pol gene were amplified by “in‐house” RT‐PCR and directly sequenced. Online subtyping tools, RIP, REGA v3, jpHMM, SCUEAL and COMET, were used for preliminary subtyping. Multiple sequence alignments were done and subtypes were inferred using Maximum Likelihood (ML) phylogenetic analysis. Results: In total we have analysed 6868 sequences. Although the majority was subtype C, we have detected 192 non‐C subtypes (2.8%). Samples originated from Gauteng, Eastern Cape, Western Cape and the Free State Provinces. Eighty one pure polsubtypes were detected: subtypes A (n=34), B (n=26), D (n=4), G (n=14) and one each of subtypes F2, H and J. Unique recombinant forms detected, included subtypes AC (n=7), BC (n=4), CD (n=23), CG (n=7), and CH (n=8) as well as a spectrum of other recombinants. An increasing number of complex recombinants (n=21) were also detected. Conclusions: The detection of multiple subtypes and unique and complex recombinant forms in South Africa is of concern and indicates a complex and evolving epidemic. It is possible that these forms in the future might give rise to new circulating recombinant forms. It is of the utmost importance that we continue to monitor HIV‐1 diversity in South Africa. 81 AUTHOR INDEX AUTHOR Ana Abecasis Jan Albert Helen Alexander Samuel Alizon Ahidjo Ayouba Istvan Bartha Frederic Bertels Alessandro Boianelli Veronika Boskova Jonathan Carlson John Coffin Caroline Colijn Jessica Conway Mirela D’arc Rob De Boer Tulio De Oliveira Hilje Doeks Jeffrey Dorfman Paul Edlefsen Susan Engelbrecht Kemal Eren Christophe Fraser Simon Frost Helen Fryer Feng Fu Astrid Gall Vipul Gupta Hillel Haim Joshua Herbeck Alison Hill Cynthia Ho Emma Hodcroft Nathanael Hoze Peter Hraber Taina Immonen Graeme Jacobs Mary Kearney Irona Khandaker Denise Kühnert Oliver Laeyendecker Sivan Leviyang Gabriel Leventhal Andrew Leigh Brown Maja Lunar Katrina Lythgoe Carsten Magnus Siriphan Manocheewa Karin Metzner PAGE(S) 35 6, 10 18 66 4 40 7 57 8 17 20, 55 32 9, 26 4 25 33, 75 24 3, 60 78 81 45 24, 34, 47, 65 34 27 29 42 28 79 21 1 13 34, 47 49 46 9 54 20, 55, 72 76 31 37 14 41 34, 47, 61 74 24 16, 49 63 7 E‐MAIL [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] hillel‐[email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]‐lj.si [email protected] [email protected] [email protected] [email protected] AUTHOR INDEX Venelin Mitov Viktor Mueller James Mullins Richard Neher Vladimir Novitsky Aridaman Pandit Simona Paraschiv Alan Perelson Andrea Pineda Sergei Pond Art Poon Bogdan Popescu Abayomi Olabode Eugen Radu Jayna Raghwani Manon Ragonnet David Rasmussen Oliver Ratmann Ramon Lorenzo Redondo Roland Regoes Hannah Roberts Morgane Rolland Raabya Rossenkhan Chandra Nath Roy Wei Shao Werner Smidt Alex Smith Joao Sousa Tanja Stadler Maja Stanojevic Ronald Swanstrom Aime Marcel Simon Tongo Passo Anne‐Mieke Vandamme C.H. Van Dorp Marco Vignuzzi Joel Wertheim Corey Williams‐Wietzikoski Natasha Wood Chris Wymant Gonzalo Yebra Fabio Zanini Shuntai Zhou 43 2 19, 21, 63, 68, 69, 70, 71 6 64 23 38 9, 14, 22, 26 62 39, 45 34 38 36 38 12, 13 47 48 34, 47, 65 27 7, 49 53 14, 15, 69, 70 58 77 55 73 56 2 30, 31, 34, 43, 48 80 5, 67 60 2, 35 44 11 39 68 59 42 61 6, 10 67 [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] SAVE THE DATE 23rd Annual HIV Dynamics & Evolution April 24‐27, 2016 Woods Hole, Massachusetts 
`