Transmission of highly virulent CXCR4 tropic HIV-1 through the mucosal route in an individual with a wild-type CCR5 genotype

Abstract Nearly all transmitted/founder (T/F) HIV-1 are CCR5 (R5)-tropic. While previous evidence suggested that CXCR4 (X4)-tropic HIV-1 are transmissible, detection was not at the earliest stages of acute infection. Here, we identified an X4-tropic T/F HIV-1 in a participant in acute infection cohort. Coreceptor assays demonstrated that this T/F virus is strictly CXCR4 tropic. The participant experienced significantly faster CD4 depletion compared with R5 virus infected participants in the same cohort. Naïve and central memory CD4 subsets declined faster than effector and transitional memory subsets. All CD4 subsets, including naïve, were productively infected. Increased CD4 + T cell activation was observed over time. This X4-tropic T/F virus is resistant to broadly neutralizing antibodies (bNAbs) targeting V1/V2 and V3 regions. These findings demonstrate that X4-tropic HIV-1 is transmissible through the mucosal route in people with the wild-type CCR5 genotype and have implications for understanding the transmissibility and immunopathogenesis of X4-tropic HIV-1.


Introduction
The vast majority of T/F HIV-1 are CCR5 tropic 1,2 .While the biological mechanisms underlying the transmission advantage of R5 tropic HIV-1 remain incompletely understood, previous evidence suggested that X4 tropic HIV-1 are transmissible.The best evidence comes from the identi cation of X4 viruses in people homozygous for CCR5 delta 32 allele [3][4][5] .These studies suggested that clinical HIV-1 infection could be established and maintained by the CXCR4 coreceptor.Moreover, the identi cation of X4 viruses in recently diagnosed HIV-1 infections also indicated that X4 viruses are transmissible, although the chance of transmission is low 6,7 .However, one caveat of the previous studies is that viral identi cation was not at the earliest stage of acute HIV infection.Therefore, the exact phenotype of the transmitted virus was unable to be characterized.Because HIV-1 could also use alternative coreceptors other than CCR5 and CXCR4, as assessed by in vitro assays, to establish clinical infection 5,8 , the possibility that viral transmission in people homozygous for CCR5 delta 32 was mediated by alternative coreceptors could not be completely excluded.To our knowledge, a T/F HIV-1 exclusively using CXCR4 has not been identi ed so far and it remains unclear whether X4 tropic HIV-1 without capacity to use CCR5 can be transmitted through the mucosal route in people with a wild-type CCR5 genotype.
We identi ed a participant (40700) who was infected by an X4 tropic T/F HIV-1 in the RV217 Thailand cohort, a longitudinal study of individuals at high risk of HIV-1 infection who were followed while still uninfected with twice weekly HIV-1 RNA testing 9 .The risk factor for HIV-1 transmission was men who have sex with men (MSM).Participant 40700 had a wild-type CCR5 genotype and expressed normal levels of CCR5 on CD4 + T cells.An unusually fast CD4 depletion was observed in participant 40700 upon HIV-1 transmission.We comprehensively characterized the genetic and phenotypic property related to the immunopathogenesis of this highly pathogenic T/F virus.

Results
Identi cation and characterization of the X4 tropic T/F virus in participant 40700 Participant 40700 was identi ed in the RV217 Thailand cohort and was infected by a CRF01_AE HIV-1.
The rst large volume blood draw was 15 days from the rst HIV-1 RNA-positive test and 23 days from the last HIV-1 RNA-negative test.Analysis of near full-length viral genomes obtained by single genome ampli cation (SGA) identi ed a homogenous viral lineage at day 15 (Fig. 1a).Genetic analysis using the Poisson-Fitter tool 10 showed that the frequency distribution of the Hamming Distance followed Poisson distribution and the sequences exhibited a star-like phylogeny (Fig. 1b).The estimated days from HIV-1 transmission (based on the Poisson-Fitter tool) was 26.These results demonstrated that participant 40700 was identi ed during acute HIV infection (AHI) which was established by a single T/F virus.
Coreceptor prediction using Geno2Pheno showed a high likelihood of CXCR4 usage of the 40700 T/F virus (FPR = 0.1%) 11 .This observation prompted us to phenotypically characterize the coreceptor usage of this T/F virus.Coreceptor inhibition assay in TZM-bl cell line showed that the entry activity of the 40700 T/F pseudovirus can be completely blocked by the CXCR4 inhibitor AMD3100 while the CCR5 inhibitor Maraviroc showed no inhibition at the highest concentration (100 µM) (Extended Data Fig. 1a).Coreceptor assay in NP-2 cell lines demonstrated that the 40700 T/F pseudovirus can infect the NP-2 CXCR4 cell line with high e ciency while no infectivity was observed in the NP-2 CCR5 cell line (Extended Data Fig. 1b).These data suggested that the 40700 T/F virus uses CXCR4 exclusively.
To better characterize the phenotype of the 40700 T/F virus, we constructed a full-length infectious molecular clone (IMC) (Extended Data Fig. 2).A viral replication assay in puri ed CD4 + T cells from three healthy donors showed that the 40700 T/F IMC was replication competent in primary CD4 + T cells (Fig. 1c).Overall, its replication kinetics was comparable to the X4 tropic strain NL4.3 and an R5-tropic T/F virus CH058 (Fig. 1c).We then determined the coreceptor usage of the 40700 IMC in a panel of NP-2 cell lines expressing CCR5, CXCR4, as well as another seven alternative coreceptors (Fig. 1d).Consistent with the results of the pseudovirus, the 40700 IMC replicated in the NP-2 CXCR4 cell line with high e ciency.No viral replication was detected in the NP-2 CCR5 cell line (Fig. 1d).The 40700 IMC could also infect CCR3, APJ and FPRL1 cell lines with low e ciency (Fig. 1d).
A CCR5 genotyping showed that participant 40700 did not have the CCR5 delta32 mutation (Methods).
We analyzed the CCR5 and CXCR4 expression on CD4 + T cells of 40700 isolated from day 20 (from the rst positive test for HIV-1 RNA) (Extended Data Fig. 3).The CCR5 and CXCR4 expression in 40700 showed similar pattern as previously observed in two participants infected by R5 viruses in the same cohort 12 .The CCR5 expression was high on the effector memory (EM) and transitional memory (TM) CD4 subsets, low on the central memory (CM) subset, and was undetectable on the naïve subset.The CXCR4 expression showed a reciprocal pattern (Extended Data Fig. 3).Taken together, these data demonstrated that participant 40700 was infected by a CXCR4 tropic T/F virus lacking demonstrable CCR5 tropism, and the transmission of X4 virus in 40700 could not be explained by unusual expression of the CCR5 and CXCR4 coreceptors.

Rapid CD4 + T cell depletion in participant 40700
Participant 40700 experienced signi cantly faster CD4 + T cell decline compared to infections established by R5 virus in the RV217 Thailand cohort (Fig. 2a).The CD4 count dropped from 707 cells/µL to 130 cells/µL during the rst 223 days of infection before ART initiation (40700 was on and off ART since day 266 as described in Methods) (Fig. 2b).The rate of CD4 decline was 1.6 cells/µL/day in 40700, while the median rate of CD4 decline in the R5 group was 0.27 cells/µL/day (Fig. 2a).By analyzing the mean and variance of the participant speci c CD4 slopes with a Linear Mixed Effect (LME) model, we were able to con rm that 40700 had a statistically signi cant rate of CD4 + T cell decline compared with the R5 group (P < 0.001; normal distribution test) (Fig. 2a).The CD8 T cell count was relatively stabilized after AHI and the plasma viral load (VL) was sustained above 10 5 copies/mL (Fig. 2b).Analysis of the dynamics of each CD4 subset showed that the naïve subset declined faster (0.97 cells/µL/day) than the memory subsets (Fig. 2c).Among the memory CD4 subsets, the central memory (CM) subset declined faster (0.42 cells/µL/day) than effector memory (EM) and transitional memory (TM) subsets (0.17 and 0.07 cells/ µL/day, respectively) (Fig. 2c).These data suggested that the rapid CD4 depletion in 40700 was mainly due to the loss of naïve and CM CD4 + T cells.
Broad CD4 subset targeting of the 40700 T/F virus CXCR4 tropic HIV-1 are considered to have a broader CD4 subset targeting than R5 viruses 13 .In X4-tropic SHIV infected macaques, the resting naïve CD4 + T cells were productively infected and rapidly depleted 14 .The strict X4 tropism of the 40700 T/F virus provides a unique opportunity to determine the CD4 subset preference and pathogenesis of X4 virus in natural HIV-1 infection.The faster depletion of the naïve and CM CD4 subsets in 40700 indicated that they could be preferentially infected due to relatively high levels of CXCR4 expression (Extended Data Fig. 3).Quanti cation of total and integrated HIV-1 DNA at day 78 showed that both the total and integrated HIV-1 DNA were detectable in all CD4 subsets (Fig. 3a).The CM CD4 subset contained the highest level of both the total and integrated HIV-1 DNA (Fig. 3a).Possibly due to the relatively low level of CXCR4 expression, the EM and TM subsets contained lower amount of total HIV-1 DNA compared with the CM and naïve subsets (Fig. 3a).In the naïve subset, the level of total HIV-1 DNA was 9.1-fold higher than the integrated DNA, while in the memory subsets, the total HIV-1 DNA was approximately 2-fold higher than the integrated DNA (Fig. 3a).This observation indicated that while the viruses could enter the naïve CD4 + T cells with high e ciency, the integration process could be defective or delayed due to the resting status of the naïve cells.
To determine which CD4 subsets were productively infected in 40700, we quanti ed the cell-associated HIV-1 RNA in longitudinal samples by amplifying both the pol region and the tat/rev transcript (Fig. 3b).Because the tat/rev transcript was generated at relatively late stage of virus life cycle, it could more accurately re ect the productive infection status.The results showed that all CD4 subsets were productively infected (Fig. 3b).In contrast, in RV217 Thailand participants who only harbored R5 viruses, the cell-associated HIV-1 RNA were undetectable in naïve CD4 subset as demonstrated in our recent study 12 .Quanti cation of the pol region and the tat/rev transcript showed similar dynamics (Fig. 3b).At earlier time points, the CM CD4 subset contained relatively high level of cell-associated HIV-1 RNA (Fig. 3b).The naïve and TM subsets contained relatively low level of cell-associated RNA initially but achieved similar level as the CM subset at the last time point.The EM subset had the lowest level of cellassociated RNA at the last time point (Fig. 3b).These results suggested that the CM CD4 subset was preferentially targeted by the 40700 T/F virus during early infection.

Heightened CD4 + T cell activation over time
Immune activation is correlated with CD4 decline in HIV-1 infection 15,16 .The rapid CD4 depletion in 40700 prompted us to determine the level of CD4 + T cell activation over time.Quanti cation of the expression of T cell activation markers showed that after acute infection, coincident with a brief drop of plasma VL (Fig. 2b), there was a temporary decrease of CD4 activation and proliferation (Fig. 4).
However, instead of achieving a steady-state as observed in most HIV infected people 15 , the level of CD4 activation increased after this brief decline.Among different CD4 subsets, the TM subset expressed the highest level of activation and proliferation markers (Fig. 4).The naïve CD4 subset was negative for all investigated markers except for CD38 (Fig. 4).The lack of Ki-67 expression in the naïve subset suggested that as previously observed in X4-tropic SHIV infected macaques 14 , the naïve CD4 + T cells which were productively infected and depleted in 40700 were largely quiescent.The level of CD4 activation and proliferation decreased upon ART.However, the expression of PD-1 did not decrease immediately after treatment (Fig. 4).
The 40700 T/F virus is resistant to bNAbs targeting the V1/V2 and V3 regions X4 tropic HIV-1 from different subtypes tend to be more resistant to broadly neutralizing antibodies (bNAbs) targeting the V3 region 12,[17][18][19] .Determination of neutralization susceptibility to a panel of bNAbs showed that while the 40700 T/F virus can be neutralized by bNAbs targeting the CD4 binding site and MPER, it was completely resistant to all investigated bNAbs targeting the V1/V2 and V3 regions (Fig. 5a).Investigation of the kinetics of autologous neutralization showed that participant 40700 developed relatively low level of autologous neutralization activity compared with two infections established by R5 T/F viruses (Fig. 5b).
Analysis of V3 amino acid sequence identi ed that the conserved V3 N301 glycan site is lost in 40700 due to the T303I substitution (Extended Data Fig. 4a).Indeed, substitutions at the N301 glycan exist in nearly all X4 viruses identi ed in the RV217 Thailand cohort 12 .As most of the CRF01_AE viruses, the N332 glycan site shifted to the N334 position in 40700 (Extended Data Fig. 4a).In comparison to the CRF01_AE consensus sequence, the 40700 T/F virus has two positively charged amino acid substitutions in V3, including the Q313R mutation at the V3 crown (Extended Data Fig. 4a).Moreover, the 40700 T/F virus has a longer V2 loop and an additional glycan stie in V2 region compared with most R5 T/F viruses in the same cohort (Extended Data Fig. 4b-c).These ndings identi ed unique genetic feature and glycan arrangement in V1/V2 and V3 regions of the 40700 T/F virus.
Investigation of longitudinal envelope evolution in 40700 showed that the viral population diversi ed from the T/F virus by accumulation of point mutations (Fig. 6).Interestingly, during early viral evolution, the variable loops were relatively conserved while most of the xed or predominant mutations emerged in the conserved regions (Fig. 6).This could be associated with relatively low level of autologous neutralization response in 40700.

Discussion
The current study demonstrates that HIV-1 with strict X4 tropism can be transmitted through the mucosal route in a person with wild-type CCR5 genotype and can cause rapid CD4 depletion.This nding, together with the fact that the 40700 T/F virus is resistant to bNAbs targeting the V1/V2 and V3 regions, necessitate the need to monitor the transmission and spread of highly pathogenic X4 HIV-1 which tend to be resistant to bNAbs targeting the variable loops 12,[17][18][19] .
While it has been well established that R5 tropic HIV-1 have transmission advantage, the underlying mechanisms remain poorly understood.A long-term question is why X4 viruses are less transmissible while CXCR4 is expressed on a broader range of CD4 + T cells than CCR5.The current study, together with our recent ndings 12 , provide evidence that the different CD4 subset preference could be an important determinant underlying the distinct transmissibility of R5 and X4 HIV-1.In our recent study, tracking coreceptor switch of the T/F virus demonstrated that upon the origin of the earliest X4 virus, the viral population diverged in vivo in terms of CD4 subset targeting 12 .While the R5 population remained predominant in the EM and TM CD4 subsets, the emerging X4 viruses lost advantage in the EM and TM subsets while gained advantage in the CM and naïve subsets.In participant 40700, the CM CD4 subset was preferentially infected during early infection while the EM subset was less infected by this X4 tropic T/F virus.Because the EM CD4 + T cells are more abundant in mucosal tissues than the CM and naïve CD4 + T cells [20][21][22] , the replication advantage of R5 virus in the EM CD4 + T cells could provide an advantage of mucosal transmission.Moreover, the EM and TM cells are likely to have higher viral burst size than the naïve and CM cells due to their more activated status, and thus release more virions into the plasma.Indeed, the R5 viruses remained predominant in plasma in all participants harboring X4 variants in the RV217 Thailand cohort 12 .This could be another reason for the transmission advantage of R5 tropic HIV-1.Our ndings also indicate that the EM and TM CD4 subsets could play a more important role in mediating mucosal HIV-1 transmission than CM and naïve subsets (otherwise, the X4 virus would have a higher chance to be transmitted given their advantage in the CM and naïve subsets).A better understanding of the CD4 subset preference in determining the transmissibility of HIV-1 could open new possibilities for preventing HIV-1 transmission (for example, by downregulation of CCR5 expression on the EM and TM CD4 subsets).
While the altered CD4 subset preference towards the naïve and CM cells may compromise the transmissibility of X4 tropic HIV-1, it could enhance the pathogenicity.In 40700, the rapid CD4 depletion was mainly due to the loss of naïve and CM CD4 subsets.In line with this nding, our recent study on coreceptor switch showed that upon the origin of the X4 viruses, the CM and naïve CD4 subsets declined faster than the EM and TM subsets 12 .Similar as previously observed in macaques infected by X4-tropic SHIV 14 , the resting naïve CD4 + T cells were productively infected and depleted in 40700.The exponentially increased viral load in the naïve subset was also similar as previously observed in the macaque model 14 .The mechanisms for productive infection and depletion of the resting naïve CD4 + T cells in vivo require further investigation.In addition to direct infection of the naïve CD4 + T cells, the high viral burden in the CM CD4 subset, which is essential for CD4 hemostasis as shown in SIV infected natural hosts and HIV-1 infected people [23][24][25][26] , could be another reason for the enhanced pathogenicity of X4 tropic HIV-1.In nonprogressive SIV infection, a low viral burden in the CM CD4 subset and the lack of chronic immune activation are considered as two hallmarks [26][27][28] .While in most HIV-1 infected people, the level of CD4 activation could achieve a steady state 15 , the level of CD4 activation increased in 40700 after acute infection.Further efforts are needed to determine whether a broader range of CD4 subset infection could lead to an enhanced immune activation in HIV-1 infection.
The current study has several limitations.First, we only focused on a single participant.Second, because samples from the tissues such as lymph nodes were not available, the observations were based on cells from peripheral blood.Future research using animal model could lead to a better understanding of the immunopathogenesis of this X4 tropic T/F HIV-1.

Study participants
All study participants were from the RV217 Thailand acute infection cohort 9 .Participant 40700 became infected around the time of entry and the risk factor for HIV-1 transmission was men who have sex with men (MSM).The subtype of the 40700 T/F virus is CRF01_AE.Participant 40700 initiated ART on day 266 (from the 1 st positive test for HIV-1 RNA) and stopped ART on day 282.ART was restarted on day 388.Written consent was provided by all participants.The study was approved by the local ethics review boards, the Walter Reed Army Institute of Research, and the institutional review boards of the University of Maryland School of Medicine.

Pseudovirus preparation and titration
The pseudovirus stocks were prepared as previously described 30 .In brief, 2 μg of env clone was cotransfected with 4 μg of the pNL4.3-ΔEnv-vpr+-luc+into 293T cells in a T25 ask using the FuGENE6 transfection reagent (Promega).The cells were cultured at 37 o C for 6 hours before the medium was completely replaced with fresh medium.The culture supernatants containing the pseudoviruses were harvested at 72 hours post transfection, aliquoted and stored at −80°C until use.The infectious titers (TCID 50 ) of the pseudovirus stocks were determined on TZM-bl cells.

Generation of infectious molecular clone (IMC) of the 40700 T/F virus
To obtain the full-length genome of the 40700 T/F virus, three overlapping fragments were ampli ed using plasma or PBMC samples collected at day 20 (from the 1 st positive test for HIV-1 RNA) (Extended Data Fig. 2).The 5' half and 3' half viral genomes were ampli ed using day 20 plasma sample as described above.The 5' LTR-gag fragment was ampli ed using viral DNA extracted from the day 20 PBMCs with the forward primer 5'-TGGAAGGGCTAATTTACTCCAAGAAAAG-3' (nt 1-28) and the reverse primer 5'-TCTGATAATGCTGWRAACATGGGTAT-3' (nt 1294-1319).The full-length 40700 T/F virus was inferred as the consensus sequence of the three overlapping fragments (Extended Data Fig. 2).The 40700 T/F sequence was chemically synthesized and cloned into the vector pUC57-Brick (Genescript).
To generate the viral stock of the 40700 T/F IMC, 6 μg of IMC was transfected into 293T cells in a T25 ask using the FuGENE6 transfection reagent (Promega).The cells were cultured at 37 o C for 6 hours before the medium was replaced by fresh medium.The culture supernatants were harvested at 72 hours post transfection, aliquoted and stored at −80°C until use.The infectious titer (TCID 50 ) of the viral stock was determined in TZM-bl cells.

Determination of coreceptor usage
Coreceptor usage of the 40700 T/F pseudovirus was determined by both coreceptor inhibition assay in TZM-bl cell line and entry assay in NP-2 cell lines.For inhibition assay, TZM-bl cells were seeded in a 96well plate at a density of 1 × 10 5 cells per well.The next day, the cells were pre-treated with different concentrations of the CCR5 inhibitor Maraviroc or the CXCR4 inhibitor AMD3100 at 37 o C for 1 hour.The treated cells were infected with approximately 500 TICD 50 of the pseudovirus.The infected cells were lysed at day 3 after infection.The infectivity in each well was determined by measuring the relative luciferase units (RLU) in the cell lysates using the Britelite plus system (PerkinElmer).The percentage of inhibition was determined by comparing the infectivity with positive control wells without drug inhibition.
For entry assay, NP-2 cell lines expressing CCR5 or CXCR4 were seeded in a 96-well plate at a density of 1 × 10 5 cells per well.The next day, the cells were infected with approximately 200 TCID 50 of each pseudovirus (MOI = 0.002).After 6 hours of incubation at 37°C, the infected cells were washed twice with the culture medium and cultured at 37°C for three days.At 72 hours post infection, the infected cells were lysed, and the infectivity was determined by measuring the relative luciferase units (RLU) in the cell lysates using the Britelite plus system (PerkinElmer).Viral infectivity was considered positive if the RLU value was at least 5-fold higher than the background RLU value in the NP-2 parental cell line.All experiments were performed in triplicate.
Coreceptor usage of the 40700 IMC was determined in NP-2 cell lines expressing CCR5, CXCR4 and a panel of alternative coreceptors (CCR3, APJ, FPRL1, CCR8, GPR15, CCR2b and CCR1).NP-2 cells were seeded in a 96-well plate one day before infection at a density of 1 × 10 5 cells per well.Approximately 100 TCID 50 of viral stock was used for infection (MOI = 0.001).After 4 hours of incubation at 37°C, the infected cells were washed three times with the culture medium and cultured at 37°C for ve days.The p24 concentrations in the culture supernatants were measured on day 5 post infection (PerkinElmer).

Determination of viral replication capacity in primary CD4 + T cells
Puri ed CD4 + T cells from healthy donors were stimulated with 1 µg/mL soluble anti-CD3 (clone OKT3, eBioscience) and 1 µg/mL soluble anti-CD28 (clone CD28.2, eBioscience) in the presence of 50 IU/mL IL-2 (PeproTech) for three days.The stimulated CD4 + T cells were infected by the 40700 T/F IMC at 37 o C for 4 hours (MOI = 0.001).The cells were washed three times after infection and were cultured at 37 o C for 7 days.The culture supernatants were collected every day and the viral growth kinetics was determined by measuring the p24 concentration in the supernatant (PerkinElmer).All infections were performed in triplicate.
The CCR5 and CXCR4 antibodies were titrated to determine the optimal concentration.To determine the non-speci c staining, cells was cold-inhibited by a 100-fold excess of the unlabeled CCR5 or CXCR4 antibody (the same clone as the staining antibody) mixed with the respective labeled antibody.A uorescence minus one (FMO) staining was also determined for CCR5/CXCR4 staining.The highest concentration of the labeled antibody with which the cold inhibition showed virtually overlapping staining with the FMO was used to determine the levels of CCR5 and CXCR4 expression on each CD4 subset.

Quanti cation of cell associated HIV-1 RNA
To quantify HIV-1 RNA, RNA was extracted from sorted CD4 + T cells using the RNeasy Mini kit (Qiagen).A total of 8.5 µL extracted RNA was subjected to one-step RT-PCR using the Superscript III one-step RT-PCR system (Invitrogen).To amplify the pol region, the one-step RT-PCR was performed using the forward primer Pol F1 5'-TACAGTGCAGGGGAAAGAATA-3' (nt 4809-4829) and the reverse primer Pol R1 5'-CTTCTTGGCACTACTTTTATGTCAC-3' (nt 4993-5017).The PCR conditions were as follow: a reverse transcription step at 50 o C for 1h; A denaturing step at 94 o C for 2 min; 16 cycles of a denaturing step at 94 o C for 15 sec, an annealing step at 55 o C for 30 sec, an extension step at 68 o C for 1 min, and one cycle of an additional extension at 68 o C for 5 min.The rst round PCR products were diluted 10-fold and a total of 6.4 µL of diluted PCR products were used for the real-time PCR using the forward primer Pol F1, the reverse primer Pol R2 5'-CTGCCCCTTCACCTTTCC-3' (nt 4957-4974), and the probe Pol Famzen: 5'-/56-FAM/TTTCGGGTT/ZEN/TATTACAGGGACAGCAG/3IABkFQ/-3' (nt 4896-4921).The real-time PCR was performed on the QuantStudio 3 Real-Time PCR Systems (Thermo Fisher Scienti c) using the following conditions: A denaturing step at 94 o C for 4 min, 45 cycles of a denaturing step at 94 o C for 3 sec, an annealing and extension step at 60 o C for 20 sec.To amplify the tat/rev transcript, the one-step RT-PCR was performed using the forward primer Tat1.4AE 5'-TGGCAG GAAGAAGCGGAAG (nt 5971-5989) and the reverse primer Rev AE-ter 5'-TGTCTCTGYCTTGCTCKCCACC-3' (nt 8433-8454).The real-time PCR was performed using the forward primer Tat2 AE-bis 5'-GTAAGGATCATCAAAATCCTVTACCARAGCA-3' (nt 6015-6045), the reverse primer Rev AE-ter 5'-TGTCTCTGYCTTGCTCKCCACC-3' (nt 8433-8454), and the probe Tat-Rev AE 5'-/56-FAM/TT CYT TCG G/ZEN/G CCT GTC GGG TTC C/3IABkFQ/-3' (nt 8399-8421).
The PCR conditions were the same as for amplifying the pol region.The copy number of the input RNA was determined by using the RNA standard generated by in vitro transcription.In brief, the amplicon region (the CRF01_AE consensus sequence was used in the current study) was cloned into the pUC57 vector downstream of the T7 promoter.The DNA fragment containing the amplicon was PCR ampli ed and the RNA was generated by in vitro transcription using the MEGAscript T7 Transcription Kit (Invitrogen).

Neutralization assay
The neutralization activity of plasma samples and monoclonal antibodies (mAbs) was determined by using a luciferase reporter system in TZM-bl cells.Plasma samples were heat inactivated at 56°C for 45 minutes.The inactivated plasma was diluted at a 1:3 serial dilution starting from 1:20.The mAbs were diluted at a 1:3 serial dilution from a starting concentration of 25 μg/mL.The virus stocks were diluted to a concentration that achieved approximately 150,000 RLU in the TZM-bl cells (or at least 10 times above the background RLU of the cells control).The serial diluted plasma samples or mAbs were then incubated with the viruses for 1 hour at 37°C in duplicate before the TZM-bl cells were added.The 50% inhibitory dose (ID 50 ) was determined as the dilution at which the relative luminescence units (RLUs) were reduced by 50% in comparison to the RLUs in the virus control wells after subtraction of the background RLUs in cell control wells.

Determination of the rate of CD4 + T cell decline
The rate of CD4 decline was determined using a linear mixed effect model (LME).The LME model was hierarchical in the sense that it estimated a population speci c slope and intercept with time, as well as subject-speci c slopes and intercepts.The longitudinal CD4 data from the earliest available time point to the last available time point before ART initiation was used for analysis.To determine whether participant 40700 had a statistically signi cant rate of CD4 decline, we calculated the empirical best linear unbiased prediction (EBLUP) for the rate of CD4 decline.A normal distribution test was then conducted to determine the statistical signi cance.