1. Immunology and Inflammation
Download icon

The HIV-1 latent reservoir is largely sensitive to circulating T cells

  1. Joanna A Warren
  2. Shuntai Zhou
  3. Yinyan Xu
  4. Matthew J Moeser
  5. Daniel R MacMillan
  6. Olivia Council
  7. Jennifer Kirchherr
  8. Julia M Sung
  9. Nadia R Roan
  10. Adaora A Adimora
  11. Sarah Joseph
  12. JoAnn D Kuruc
  13. Cynthia L Gay
  14. David M Margolis
  15. Nancie Archin
  16. Zabrina L Brumme
  17. Ronald Swanstrom
  18. Nilu Goonetilleke  Is a corresponding author
  1. Department of Microbiology and Immunology, University of North Carolina, United States
  2. Lineberger Comprehensive Cancer Center, University of North Carolina, United States
  3. UNC Center For AIDS Research, University of North Carolina, United States
  4. British Columbia Centre for Excellence in HIV/AIDS, Canada
  5. Department of Medicine, University of North Carolina, United States
  6. UNC HIV Cure Center, University of North Carolina, United States
  7. Department of Urology, University of California San Francisco, United States
  8. Gladstone Institute of Virology and Immunology, United States
  9. Faculty of Health Sciences, Simon Fraser University, Canada
Research Article
  • Cited 4
  • Views 1,704
  • Annotations
Cite this article as: eLife 2020;9:e57246 doi: 10.7554/eLife.57246

Abstract

HIV-1-specific CD8+ T cells are an important component of HIV-1 curative strategies. Viral variants in the HIV-1 reservoir may limit the capacity of T cells to detect and clear virus-infected cells. We investigated the patterns of T cell escape variants in the replication-competent reservoir of 25 persons living with HIV-1 (PLWH) durably suppressed on antiretroviral therapy (ART). We identified all reactive T cell epitopes in the HIV-1 proteome for each participant and sequenced HIV-1 outgrowth viruses from resting CD4+ T cells. All non-synonymous mutations in reactive T cell epitopes were tested for their effect on the size of the T cell response, with a≥50% loss defined as an escape mutation. The majority (68%) of T cell epitopes harbored no detectable escape mutations. These findings suggest that circulating T cells in PLWH on ART could contribute to control of rebound and could be targeted for boosting in curative strategies.

Introduction

Antiretroviral therapy (ART) has transformed human immunodeficiency virus type 1 (HIV-1) from a fatal disease to a chronic condition. However, ART must be taken life-long as interruption of therapy in people living with HIV-1 (PLWH) mostly results in viral rebound within weeks (Davey et al., 1999; Crowell et al., 2019; Bar et al., 2016; Buzon et al., 2014). This rebound results from cells harboring HIV-1 DNA that is integrated into the host genome. While >95% of proviral DNA is replication-incompetent, the remaining fraction that we define here as the ‘HIV-1 reservoir’, retains the capacity to produce infectious virus particles, either stochastically (Chun et al., 1997; Finzi, 1997; Siliciano et al., 2003; Eisele and Siliciano, 2012), or through interventions to induce coordinated reactivation (Archin et al., 2012; Søgaard et al., 2015; Elliott et al., 2013; Elliott et al., 2015). The largest and best characterized HIV-1 reservoir resides in circulating resting CD4+ T cells. The inducible form of the HIV-1 reservoir is present in blood cells at a low frequency (~1 in 106 resting CD4+ T cells), but is highly stable, with an estimated half-life of 44 months (Crooks et al., 2015; Siliciano and Siliciano, 2015; Siliciano and Siliciano, 2004). Natural decay of the HIV-1 reservoir would therefore take >70 years (Crooks et al., 2015; Siliciano and Siliciano, 2015; Siliciano and Siliciano, 2004).

Strategies to allow PLWH to cease ART without viral rebound aim to either eliminate all HIV-1 infected cells harboring persistent, replication-competent virus (HIV-1 eradication), or to achieve a state of durable HIV-1 suppression without rebound (HIV-1 remission). Both approaches may harness HIV-1-specific CD8+ T cells to achieve reservoir reduction or elimination. T cells recognize infected cells through T cell receptors (TCR) that bind viral peptides presented on human leukocyte antigen (HLA) on the surface of the infected cell (Townsend and McMichael, 1985; Townsend et al., 1984; Townsend et al., 1986). On peptide recognition, CD8+ T cells can produce cytokines or release lytic molecules killing the infected cell. There is significant evidence that CD8+ T cells contribute to the control of untreated HIV-1, supported by studies of simian immunodeficiency virus (SIV) in non-human primates (Pandrea et al., 2011; Barouch et al., 2002; Schmitz et al., 2005; McBrien et al., 2020). First, certain HLA-I types, such as B*57:01/:03 and B*27:05, are overrepresented in HIV-1 elite virus controllers, defined as PLWH who exhibit sustained HIV-1 control below 50 copies/ml without ART. This suggests that specific subpopulations of CD8+ T cells are capable of controlling HIV-1 replication in the absence of therapy. Second, HIV-1-specific CD8+ T cells exert significant immune pressure on HIV-1, resulting in the accumulation of non-synonymous mutations in and around reactive T cell epitopes that can confer escape from CD8+ T cell recognition and killing (Borrow et al., 1997; Goulder et al., 1997; Phillips et al., 1991; Price et al., 1997). HIV-1 escape from CD8+ T cells, which is typically measured as a ≥ 50% decrease in the magnitude of the T cell response, can occur within days to weeks of infection (Borrow et al., 1997; Price et al., 1997; Koup et al., 1994; Goonetilleke et al., 2009). Both T cell immunodominance (relative strength of a T cell response) and viral entropy (sequence conservation of HIV-1 at a population level) are major determinants of the rate of virus escape in untreated infection (Liu et al., 2013; Ferrari et al., 2011; Barton et al., 2016). Notably, virus escape is ongoing in untreated HIV-1 infection (Allen et al., 2005), with new virus variants emerging over time in response to newly induced T cell responses, shifts in T cell clonotypes, or to compensate for fitness costs of earlier mutations. Lastly, when T cell selection pressure is removed, as in the case of virus transmission to recipients with HLA alleles different from the donor, reversion of mutations can occur (Leslie et al., 2004; Carlson et al., 2014).

ART is highly effective in inhibiting viral replication with multiple studies reporting that HIV-1 does not undergo directional selection during suppressive ART (Kearney et al., 2014; Kearney et al., 2016; Rosenbloom et al., 2017; Van Zyl et al., 2017). Recent studies have reported that the majority of proviruses, both total HIV-1 DNA and replication-competent viruses, that persist during ART, reflect viruses circulating during the year prior to ART initiation (Brodin et al., 2016; Abrahams et al., 2019; Pankau et al., 2020). These studies suggest that ART initiation is the major driver of latency formation, including the formation of the replication-competent stable HIV-1 reservoir. Proviruses that persist during ART would therefore include HIV-1 variants that have escaped from T cell selection over the entire course of untreated infection (Phillips et al., 1991; Deng et al., 2015). ART also impacts the HIV-1-specific T cell response, with reduced HIV-1 loads resulting in lower antigen drive and over time, yielding improved proliferation and a less exhausted T cell phenotype (Rehr et al., 2008; Rutishauser et al., 2017) (reviewed in Warren et al., 2019). We have recently shown that HIV-1-specific T cell responses are detectable ex vivo and are highly stable in PLWH on durable ART suppression (Xu et al., 2019), with others showing that ex vivo isolated CD8+ T cells can suppress HIV-1 superinfected CD4+ T cells as well as in vitro reactivated reservoir virus (Sung et al., 2015; Shan et al., 2012).

Collectively, these data suggest that HIV-1-specific T cell responses on ART should reflect those detected prior to the onset of ART. Likewise, the persistent HIV-1 reservoir should reflect the burden of virus escape mutants circulating pre-ART. Of note, a recent study reported that up to 98% of proviral HIV-1 sequences harbored mutations in immunoprevalent viral epitopes, though unmutated targets were also identified (Deng et al., 2015). A high burden of HIV-1 escape variants in the reservoir could limit approaches that aim to increase anti-HIV-1 immunity (Ondondo et al., 2016; Borducchi et al., 2016; Mothe et al., 2015b; Achenbach et al., 2015; Hanke, 2014).

Here, we present the first study reporting within-host HIV-1 escape frequencies in PLWH on ART. A non-selective approach was taken, specifically sequencing HIV-1 outgrowth viruses (OGVs) from the reservoir induced from resting CD4+ T cells, and mapping circulating T cell responses against the entire HIV-1 proteome in 25 PLWH on ART. These strategies enabled us to empirically determine the extent and patterns of pre-ART escape within the stable HIV-1 reservoir. We observed that the majority, 68%, of T cell epitopes in the HIV-1 reservoir did not harbor escape variants and were recognized by circulating T cells. Furthermore, analysis of ‘escape’ epitopes (the remaining 32%) showed that escape was significantly lower in regions of HIV-1 that are more conserved. We conclude that the majority of circulating HIV-1-specific T cells in PLWH on ART can recognize the stable HIV-1 reservoir.

Results

Study cohort

HIV-1-specific T cell responses and the size of the HIV-1 reservoir were measured in 25 PLWH (ages 25–66, mean 46; male = 16, female = 9) durably suppressed by combination ART (viral load <50 copies/ml for 1.6–20.1 years, mean 7.2 years). Of the 25 participants, four initiated ART during acute HIV-1 infection, while the other 21 initiated ART during chronic HIV-1 infection (see Materials and Methods for definitions). Most participants (n = 24) were HLA-typed. Characteristics of the cohort are summarized in Table 1 and full clinical details and IRB approvals are provided in Materials and Methods.

Table 1
Demographic and clinical data of participants (n = 25).
ParticipantSexRace*EthnicityStatus at diagnosisAge§Years suppressedHiv-1 RNA**Nadir CD4Mean IUPM [n]††HLA type
00231MCNHA669.661.5E+072770.839 [1]A*11:01, A*24:02, B*35:TDS‡‡‡, B*35:02, C*04:01, C*04:01
00250MCNHA487.446.8E+054030.318 [3]A*01:01, A*31:01, B*27:03, B*44:02, C*02:02, C*06:02
00531MCNHC5214.873.8E+05811.372 [5]A*02:01, A*02:01, B*07:02, B*44:02, C*03:04, C*07:02
00546MCNHC5620.09§§N/A1950.453 [2]A*02:01, A*32:01, B*15:01, B*40:01, C*02:02, C*03:04
00673MCNHC498.053.2E+051680.790 [4]A*24:02, A*30:02, B*15:17, B*18:01, C*05:01, C*07:01
00674MCNHC576.256.8E+043382.064 [5]A*02:01, A*03:01, B*07:TDVB§§§, B*40:01, C*03:04, C*07:02
00720MCNHC252.964.0E+051680.499 [4]A*01:01, A*02:01, B*07:02, B*18:01, C*07:02, C*12:03
00728MCNHC306.001.8E+04354¶¶0.676 [6]A*02:01, A*03:01, B*07:02, B*40:02, C*02:02, C*07:02
00749MAANHC254.10N/A4020.365 [2]A*03:01, A*30:02, B*15:03, B*40:01, C*02:10, C*03:04
00773MCNHA462.012.4E+042340.233 [1]A*02:01, A*11:01, B*18:01, B*51:01, C*01:02, C*02:02
00833MAANHC368.167.9E+03358***0.048 [1]A*02:02, A*74:01, B*15:16, B*82:01, C*03:02, C*14:02
00834MAANHC252.152.1E+043650.479 [1]A*02:05, A*30:02, B*51:01, B*58:01, C*04:01, C*16:01
00848FCNHC444.781.4E+048000.022 [1]A*01:01, A*30:02, B*08:01, B*18:01, C*05:01, C*07:01
00870MCNHC545.285.3E+04154†††1.400 [1]A*11:01, A*24:02, B*35:01, B*55:01, C*03:03, C*04:01
00879FAANHC541.604.5E+026120.093 [1]A*03:01, A*33:01, B*14:02, B*58:02, C*06:02, C*08:02
00926FAANHC383.831.3E+055080.262 [1]A*02:02, A*02:05, B*:39:10, B*45:01, C*12:03, C*16:01
00929FAANHC611.771.9E+045280.268 [1]A*01:01, A*74:AB****, B*15:MJMN¶¶¶, B*81:AB****, C*02:10, C*18:01
00930MCNHC532.706.4E+051742.084 [1]A*01:01, A*11:01, B*08:01, B*40:01, C*03:04, C*07:01
00937FAANHC455.848.8E+042601.276 [2]A*03:01, A*30:01, B*35:01, B*52:01, C*04:01, C*16:01
00941FCNHC4714.586.6E+03121.50E-08‡‡[1]N/A
01014FCHC5219.851.2E+06160.338 [1]A*23:01, A*74:01, B*07:02, B*41:02, C*15:05, C*17:03
01086FAANHC305.81N/A3200.0529 [1]A*02:05, A*03:01, B*07:06, B*53:01, C*04:01, C*07:02
01095FCNHC546.561.1E+05882.528 [1]A*03:01, A*31:01, B*40:01, B*44:02, C*03:04, C*05:01
10430MCNHC599.478.1E+04781.367 [5]A*02:01, A*30:01, B*13:02, B*51:01, C*06:02, C*14:02
20296MCNHA356.902.5E+044271.50E-08‡‡ [1]A*29:02, A*31:01, B*08:01, B*44:03, C*07:01, C*16:01
Mean:467.228.6E+052930.775
Range:25–661.6–20.104.5E+02-
1.5E+07
12–8001.50E-08–2.528
  1. * C: White/Caucasian, AA: Black/African American.

    NH: Non-Hispanic; H: Hispanic.

  2. A: Initiation of ART during acute HIV-1 infection, prior to establishment of setpoint viremia; C: Initiating of ART during chronic HIV-1 infection.

    Durable suppression defined as time from ≤ 40 copies/ml to date of T cell measurement.

  3. ** HIV-1 RNA at start of combination ART.

    †† IUPM (infectious units per million). Mean of IUPM measurements over 2 years from date of T cell mapping.

  4. ‡‡ Below the limit of detection.

    §§00546 on monotherapy from 1985 to 1996, initiated combination ART in 1996. Years suppressed calculated from data of initiation of cART.

  5. ¶¶ CD4 count 261 on 8/17/17 (after cell collection date).

    ***Lowest CD4 value, CD4 nadir before ART initiation is 567.

  6. †††Lowest CD4 value, CD4 nadir before ART initiation is 157.

    ‡‡‡B*35:01/B*35:42.

  7. §§§ B*07:02/B*07:61/B*07:16.

    ¶¶¶ B*15:03/B*15:103/B*15:220.

  8. **** B*81:01/B*81:02.

HIV-1 outgrowth viruses in PLWH on ART show no directional selection over time

The HIV-1 reservoir in resting CD4+ T cells was measured using quantitative viral outgrowth assays (QVOAs) (Lee et al., 2017; Siliciano and Siliciano, 2005; Trumble et al., 2017; Ho et al., 2013; Massanella and Richman, 2016) for all 25 participants (Figure 1). Longitudinal IUPM measurements were obtained for 10 participants (Supplementary file 1). From these data, we calculated the mean of IUPM measurements (mean number of IUPM = 2/participant, range 1–6) made in the 2 years prior to HIV-1 T cell epitope mapping described below. Within the cohort, measurements ranged from below the limit of detection (participants 00941 and 20296, see Material and Methods for definition of limit of detection) to 2.528 IUPM (Table 1, Supplementary file 1).

Schema of empirical testing of T cell escape in the HIV-1 reservoir.

Top left panel. The quantitative viral outgrowth assay (QVOA) provided a minimal estimate of the size of the replication-competent reservoir virus from total resting CD4+ T cells of PLWH on ART (n = 25). Supernatants from HIV-1 p24+ wells harboring outgrowth virus (OGV) were sequenced by either PacBio (n = 22) or MiSeq (n = 1). Top right panel. Ex vivo IFN-γ ELISpot assays identified reactive T cell epitopes in clade B HIV-1 proteome in each participant. Lower panel. Sequencing and T cell data were combined to identify putative virus escape mutations by overlaying reactive T cell reactive epitopes (red horizontal line) on highlighter plots of each participant’s outgrowth sequences (n = 23). Peptides spanning all non-synonymous changes between the clade B-defined epitope and outgrowth sequences (orange boxes) were synthesized. In addition, peptides matching OGVs in participants in which > 40% of viruses contained the same non-synonymous mutation generating a ‘stripe’ on the highlighter plot were also synthesized (orange box, no overhead red line). Variant peptides were examined for their impact on T cell recognition in ex vivo IFN-γ ELISpot. Virus escape was defined as a ≥ 50% difference in magnitude between variants, either Clade B sequence or OGV variants (vertical, red dotted line).

OGVs were then sequenced. Primer ID sequencing (Lee et al., 2017; Emery et al., 2017), which allows for simultaneous in-depth sequencing of multiple HIV-1 regions was performed in one participant (participant 00673, mean of 3,700 Primer ID template consensus sequences/epitope). This approach was replaced with PacBio SMRT sequencing of the p24+ wells at endpoint dilution (i.e. average one virus-infected cell/well) to enable nearly, full-length sequencing of the HIV-1 genome. PacBio sequencing provided long reads (two reactions spanned almost the entire length of HIV-1) which identified sites of virus diversity within and outside epitopes that were reactive in initial T cell mapping (see Materials and Methods describing ‘stripes’). Note that given the initial effective population size in each well was, on average, n = 1 (Rouzine et al., 2001), in vitro mutations occurred stochastically, with only 4–5 mutations expected across the virus genome over the 15 day QVOA culture (Song et al., 2012; Brown and Richman, 1997). Mutations arising in culture were therefore not expected to bias results.

Phylogenetic analysis of the 22 participants for whom near full-length sequencing was available (mean = 16 sequences per participant, range 1–71) confirmed that sequences from each participant formed monophyletic clades (Figure 2A and B). Participants were infected with HIV-1 clade B, except for one participant (participant 00926) who was infected with HIV-1 clade G. Available HIV-1 sequences from participants treated during acute infection (n = 3) showed greater genetic similarity (maximum pairwise diversity 0.0007) than sequences from participants treated during chronic infection (n = 19) (maximum pairwise diversity 0.0088) (Mann-Whitney two tailed test, p=0.003, Supplementary file 2). This is consistent with reports in which most HIV-1 infections are established from a single HIV-1 variant, and therefore the early onset of ART, which halts viral evolution, results in a smaller and more homogenous HIV-1 reservoir (Kearney et al., 2014; Keele et al., 2008; Haaland et al., 2009; Derdeyn et al., 2004; Abrahams et al., 2009).

Neighbor-joining phylogenetic trees of sequenced outgrowth virus.

The phylogenetic trees display 5’ (A) and 3’ (B) half genome sequences generated by PacBio sequencing of outgrowth virus from the quantitative viral outgrowth assay from n = 22 PLWH on ART. Each color indicates outgrowth virus sequences from a different PLWH on ART. Black stars denote participants who initiated ART during acute infection. All participants were infected with a clade B HIV-1 virus, except for one participant (red star), who was infected with a clade G HIV-1 virus. The genetic distance is shown by scale bar, 2 substitutions out of 100 bp. Trees are midpoint-rooted.

For seven participants, OGVs from longitudinal samples were available for sequencing (up to 4.4 years, 2–4 time points/participant) (Figure 3). OGV sequences from different timepoints for each participant were interspersed within-host phylogenetic trees, suggesting no major directional selection over time while participants remained suppressed on ART (Figure 3, Figure 3—figure supplement 1). Note, the single long-branched sequence in participant 10430's phylogeny met quality scores for sequencing and checks for hypermutations without evidence of mislabeling and was therefore retained. Overall, these longitudinal data are consistent with observations that ART effectively halts viral replication and subsequent evolution (Kearney et al., 2014; Nettles, 2005). Given the lack of directional selection, longitudinal sequences for each participant were pooled, increasing the mean sequencing depth of OGVs from 16 to 25 sequences (range: 1–71) per participant (Figure 3—figure supplement 2). In general, identical OGV sequences (see definition in Materials and Methods) were identified only infrequently; however, the 5’ and 3’ half genome sequences from two participants (participants 00250 and 00749, Figure 3A and E) exhibited >20% clonality, consistent with reports that in that the HIV-1 reservoir is in part maintained by clonal expansion (Supplementary file 3; Bui et al., 2017).

Figure 3 with 2 supplements see all
Neighbor-joining within-host phylogenetic trees inferred from longitudinal outgrowth virus sequences.

Phylogenetic trees displaying 3’ half genome sequences of longitudinal outgrowth virus sequences from seven participants (A–G). Each color represents a different time point in which outgrowth virus was sequenced for each participant. Sequences from different time points for each participant are distributed throughout the phylogeny. The genetic distance is shown by scale bar (indicated with an asterisk), 1 substitution out of 1000 bp. Trees are midpoint-rooted. (A) 00250 treated during acute infection and (E) 00749 have > 20% clonal sequences.

The HIV-1 reservoir retains susceptibility to HLA-associated T cell responses

Bioinformatics approaches have been useful in identifying HIV-1 polymorphisms that are statistically significantly enriched among individuals expressing specific HLA class I alleles, identifying these as putative sites of viral escape (Carlson et al., 2012). Many of these HLA-associated polymorphisms have been experimentally verified as immune escape mutations, and lists of these HLA-associated polymorphisms are particularly comprehensive for HIV-1 subtype B (Carlson et al., 2012). We examined the OGV sequences recovered from 21 clade B-infected participants for whom both HIV-1 sequencing and HLA typing had been performed for the presence of known HLA-associated polymorphisms. For each HIV-1 sequence, each HLA-associated viral site (total number of HLA-associated sites n = 2,819, Supplementary file 4) was classified as either ‘nonadapted’, ‘adapted’, or ‘possibly adapted’. 'Nonadapted' viral sites exhibited the specific HIV-1 residue predicted to be susceptible to the restricting HLA, ‘adapted’ sites exhibited the specific HIV-1 residue predicted to confer escape from the restricting HLA, and 'possibly adapted' sites exhibited any residue other than the 'nonadapted' form, supporting it as a possible escape variant. Minimum and maximum bounds of HLA-associated adaptation in each participant's reservoir were estimated by computing the total proportion of HLA-associated sites exhibiting 'adapted' forms (minimum bound), and those exhibiting either 'adapted' or 'possibly adapted' forms (maximum bound) (Figure 4A, Supplementary file 4; Carlson et al., 2012). Within participants, the proportion of HLA-associated sites exhibiting the HLA-adapted form ranged from 9–33% (adapted) and 32–66% (adapted + possibly adapted), and varied markedly by HIV-1 protein (Carlson et al., 2012). Importantly, our observation that, across all HIV-1 proteins, the minimum and maximum bounds for within-host escape burden ranged from 20% (adapted only) to 48% (adapted + possibly adapted) meant that between 52 and 80% of HLA-associated sites in the HIV-1 reservoir were predicted to retain susceptibility to host HLA-restricted T cell responses.

Figure 4 with 2 supplements see all
Quantifying T cell escape in the HIV-1 reservoir.

(A) HLA associations in OGVs: The proportion of polymorphisms harboring HLA-adapted (blue bar) and HLA-adapted + possibly HLA-adapted polymorphisms (light purple) in PLWH on ART (Clade B infected with full-length sequencing, n = 22). Results are summarized as total, as well as by individual HIV-1 protein. (B) T cell mapping: Ex vivo IFN-γ ELISpot was performed to determine T cell responses (n = 166), shown by protein, against the clade B HIV-1 proteome in 25 PLWH on ART. (C) Escape testing: Variants detected in OGVs of participant 00749 were tested ex vivo using IFN-γ ELISpot alongside the original reactive clade B 18-mer and optimal peptide. Bars show the T cell magnitude reported as spot forming units per million (SFU/M) PBMCs. Escape was defined as a ≥ 50% decrease (red vertical dotted line) in the mean magnitude of triplicate wells compared to the peptide which elicited the greatest T cell response. The left side shows the proportion of each variant in total OGVs (n = 50). (D) Cohort-level escape: T cell escape was examined in participants for which OGV sequence data were available (n = 23 participants, n = 151 epitopes). The proportion of empirically confirmed T cell epitopes that harbored escape variant(s) (n = 49, white) or no escape variant(s) (n = 102, black) in OGV. Results are summarized in total, as well as by individual HIV-1 protein. (E) Escape by participant: The proportion of reactive T cell epitopes that harbored escape variant(s) (white) and no escape variant(s) (black) per participant (n = 23). Gray bars indicate participants (n = 2) where epitopes (n = 15) were not tested for escape due to limited OGVs. (F) Within-epitope escape: Stacked bar graph showing the proportion of individual OGVs that conferred (gray –escape) or did not confer (blue – no escape) T cell escape in each epitope for which escape was observed (n = 49). Red dotted line is the average (27%) frequency of OGVs that conferred T cell escape across all epitopes. Data are ordered in ascending order of no escape OGVs, grouped by whether the participants initiated ART in either acute or chronic HIV-1 infection. Arrows identify immunoprevalent T cell epitopes, in which the majority of OGVs do not harbor escape variants. (G) Shannon entropy: A Shannon entropy score was calculated for reactive T cell epitopes for which sequence data were available (n = 151, 23 participants). Entropy was higher in epitopes which contained T cell escape variant(s) (n = 49, median = 0.29) than epitopes with no escape variant(s) (n = 102, median = 0.23) in the OGVs (Mann Whitney two-tailed test, p=0.009). (H) Escape in vaccine immunogen: The proportion of reactive T cell epitopes in each participant (n = 25) that are targeted by the HIVconsvX immunogen (total = 59) and harbored one or more escape variants (white), or no escape variants (black) in OGVs. Gray bars indicate participants (n = 2) where epitopes in the immunogen could not be examined for escape due to limited OGVs.

HIV-1-specific T cell responses are detectable in PLWH on ART

To empirically assess the level of T cell escape, we mapped HIV-1-specific T cell responses in the 25 PLWH on ART. T cell responses were quantified by ex vivo IFN-γ ELISpot using a matrix method containing overlapping peptides spanning the consensus sequence of the entire clade B HIV-1 proteome (Supplementary file 5; Goonetilleke et al., 2009). While HIV-1-specific T cell responses remain detectable in PLWH on ART, the overall magnitude of the T cell response is ~10 fold lower than in untreated HIV-1 infection, which may result in the underestimation in the breadth of T cell responses (Xu et al., 2019; Addo et al., 2003). To increase ELISpot assay sensitivity, we added a higher number of PBMCs, 4 × 105 cells, to each well. After the initial matrix ELISpot, individual 18 amino acid (18-mer) peptides were tested to confirm reactive T cell responses. In cases in which overlapping 18-mer peptides were both reactive, we assumed responses were driven by a single reactive T cell epitope and only the higher magnitude response was designated reactive. Across the 25 participants, a total of 166 reactive T cell epitopes were identified (see Methods for positivity criteria). HLA typing was used to predict optimal CD8+ T cell epitopes, which were tested alongside 18-mer peptides. Optimal T cell epitopes, defined as inducing a T cell response ≥ than the corresponding 18-mer, were confirmed for 25/166 epitopes (Supplementary files 5 and 6).

The mean summed magnitude of HIV-1-specific T cell responses was 1,033 SFU/M (range: 156–2,856 SFU/M), and the mean T cell breadth was seven reactive epitopes (range: 1–19 epitopes) per participant (Table 2). Predictably, and consistent with reports in untreated HIV-1 infection, the magnitude of the mean summed HIV-1-specific T cell response correlated positively with the breadth of reactive epitopes targeted (r = 0.57, Spearman Rank two-tailed, p=0.003, 25 pairs, Supplementary file 7; Addo et al., 2003). As noted above, while the magnitude of the HIV-1-specific T cell response was weaker compared to untreated HIV-1 infection, the pattern of HIV-1 proteins targeted by T cells was similar to previous reports of untreated infection (Figure 4B; Xu et al., 2019; Addo et al., 2003). For example, the highest magnitude T cell response within a participant (immunodominant T cell response) most frequently targeted Gag and Nef, whereas the most frequently targeted protein across the cohort (immunoprevalent T cell response) was Gag (Figure 4—figure supplement 1; Addo et al., 2003). No differences in either T cell magnitude or breadth were observed between women and men, or between participants treated during acute versus chronic HIV-1 infection (Supplementary file 7). Neither T cell magnitude nor breadth exhibited any association with measured HIV-1 IUPM (Supplementary file 8; Xu et al., 2019).

Table 2
HIV-1-Specific T cell reponses (n = 25 participants).
ParticipantSummed
magnitude
(106/PMBC)
Immunodominant
T cell response
Total T cell
breadth
# of epitopes with
escape variant(s)
in outgrowth virus
Escape in
immunodominant
epitope (Y/N)
002312856Vif153→170133Y
002501531Env765→782104Y
00531390Gag357→37411Y
00546895Gag260→27741N
006732313Nef73→9053N
006741394Env838→85573Y
007201947Pol521→538188Y
00728439Gag204→22170N
00749418Pol481→49863N
007731174Pol273→29030N
00833333Gag164→18140N
00834725Pol273→29042N
00848276Nef89→10630N
00870363Nef65→8260N
00879500Gag292→30951N
00926234Gag180→19770N
00929798Gag180→19772N
009302182Nef81→98197Y
00937297Pol257→27453N
009412216Nef81→9811N/A*N/A*
010141463Nef65→8294N
01086258Gag17→3431Y
010952337Pol313→33042N
10430157Nef97→11411Y
20296331Gag204→2214N/A*N/A*
Total16649
(32% - 49/151)
Mean:103372
Range:157–28561–190–8
  1. *No outgrowth virus sequence data available to test for the presence of an escape variant in the sequenced outgrowth virus.

Escape is not observed in the majority of T cell epitopes in PLWH on ART

To empirically quantify T cell escape in this cohort, OGV variants were tested for their effect on T cell response (Figure 4C). In participants for which > 1 OGV sequences were obtained (n = 23 participants and collectively 151 epitopes), peptide/s were synthesized to match each participant’s OGV autologous virus sequence/s. This included any viral variants that differed from the clade B peptide used to originally map the respective reactive T cell epitope. In total, 115 of 151 epitopes contained variant OGVs.

All variant peptides were tested ex vivo in an IFN-γ ELISpot assay alongside the originally tested 18mer clade B consensus peptide, and if identified, the optimal epitopes. Escape was defined as a ≥ 50% decrease in the average magnitude of triplicate wells compared to the peptide that elicited the greatest T cell response, whether autologous peptide or clade B consensus peptide (Figure 4C, Supplementary file 6). We would note that a range of escape thresholds, ranging from 50–80% (Borrow et al., 1997; Goonetilleke et al., 2009; Radebe et al., 2015), have been applied in previous HIV-1 studies. In this study, our goal was to maximize sensitivity to detect escape by identifying all outgrowth sequences in our dataset with the potential to compromise T cell immunity; therefore, the 50% threshold was applied.

Most variant peptides did not impact T cell magnitude. In total, virus escape was confirmed in the replication-competent HIV-1 reservoir in 49 of 151 (32%) HIV-1 T cell epitopes (Figure 4D) with escape occurring more frequently in epitopes in Pol, Env and Nef (25%, 23%, 21%) proteins (Figure 4D).

As described above, participants recognized an average of 7 T cell epitopes, ranging from 1 to 19. In each participant, one third (average = 34%) of reactive epitopes harbored virus escape mutations (Figure 4E, Table 2). In six participants, no T cell escape was observed in any reactive T cell epitope (i.e. 0% escape in 3–7 epitopes, additional DNA sequencing in Figure 4—figure supplement 2), whereas in the two participants who responded to only one epitope, virus escape was confirmed in both cases (i.e. 100% escape) (Figure 4E, Table 2).

At each epitope, only a minority of outgrowth viruses confer T cell escape

We next examined each of the 49 epitopes in which escape was observed to quantify the fraction of sequenced OGVs that conferred escape that is the 'depth' of escape (Figure 4F). On average for each epitope, 27% of OGV sequences tested produced a ≥ 50% loss in T cell response (Figure 4F, Supplementary file 6). The range of within-epitope escape was broad, starting at 1–2% to 100% of all OGVs. Note, that an escape depth of 100% (which was observed for n = 5 epitopes) occurred when all OGVs were identical but induced a lower T cell response than the originally tested consensus clade B epitope. In these cases, we assumed that the primary T cell response was induced against a consensus epitope no longer detectable in the reservoir. No differences in the depth of virus escape was observed between participants treated in acute or chronic infection (Figure 4F, Mann-Whitney two-tailed test, p=0.216). Our mapping included multiple, well-characterized immunoprevalent epitopes that also harbored a mix of escape and non-escape OGVs, but overall, mostly non-escape (Figure 4F, Supplementary file 5).

Epitope entropy but not T cell immunodominance is associated with escape in PLWH on ART

We hypothesized that escape, as observed in studies of untreated HIV-1 infection, would be associated with less conserved regions of HIV-1. Shannon entropy is used to measure the level of HIV-1 population diversity at each position along the HIV-1 proteome (Shannon, 1948). Here, epitope entropy was calculated as previously described (Liu et al., 2013), as the mean entropy of each position within the epitope. High entropy values correspond to sites with more population diversity, that is sites that are less conserved (Allen et al., 2005; Shannon, 1948). As expected, escape was more commonly observed in regions of higher Shannon entropy (Mann-Whitney two-tailed test, p=0.009) (Figure 4G).

While our previous work in untreated HIV-1 infection reported that immunodominance is a major determinant in the time to viral escape, immunodominance of circulating T cell responses in PLWH on ART did not associate with the presence or absence of escape within the cognate epitope (Mann-Whitney two-tailed test, p=0.971, Figure 4—figure supplement 1). This may be because effective ART in this study cohort prevented directional selection and immunodominance hierarchies subsequent to the initial escape event may have shifted, whether in untreated infection or at the time of ART initiation.

Conserved HIV-1 vaccine immunogens preferentially target non-escaped T cell epitopes

Our and others’ observations that T cell escape occurs more frequently in higher entropy (less conserved) epitopes (Liu et al., 2013; Ferrari et al., 2011; Barton et al., 2016) suggests that vaccine-induced HIV-1-specific T cells targeting low entropy (more conserved) regions may contribute to better HIV-1 control. This may be due to HIV-1-specific T cells maintaining selection pressure on regions of the virus where escape is likely to incur a fitness cost. Several conserved-element vaccine immunogens have been designed which connect the most conserved parts of the viral proteome, either directly in-frame or with linker sequences to form artificial conserved element immunogens. Vaccination with these immunogens aims to focus T cell responses against highly conserved epitopes that are generally subdominant in natural infection (Barouch and Korber, 2010; Rolland et al., 2007; Mothe et al., 2015a; Borthwick et al., 2017; Gorse et al., 2008). We interrogated four different conserved vaccine immunogen designs, HIVconsv (Létourneau et al., 2007), HIVconsvX (Ondondo et al., 2016), HIVACAT T cell Immunogen (HTI) (Mothe et al., 2015a), and the p24 conserved element (p24 CE) (Kulkarni et al., 2013), using our dataset to examine (1) whether the immunogen contained epitopes targeted by circulating T cell responses in PLWH on ART and (2) whether the epitopes in the immunogen contained escape variant(s) found in OGVs recovered from of our participants.

The 151 epitopes that were targeted by T cells in our study participants were differentially covered by these four immunogens. The HIVconsX immunogen contained the largest number of circulating T cell responses observed in our dataset (59/151; 39%; mean of 3 epitopes/participant, range 0–6) while the p24 CE vaccine contained the least (16/151; 11%; mean of 1 epitope/participant, range 0–4) (Figure 4H, Supplementary file 9). Nevertheless, escape frequency was on average twofold lower in reactive T cell epitopes (12–29%) found within the immunogens relative to those excluded from immunogen designs (33–44%). For example, the overall proportion of reactive T cell epitopes within the HIVconsX immunogen that contained escape variants was 15.3% (9/59 epitopes, average 0 escape epitopes/participant, range 0–2), whereas the proportion of reactive T cell epitopes outside of the HIVconsX immunogen that contained escape variants was almost >2.5-fold higher (43.5%, average two escape epitopes/participant). This difference was statistically significant (p=0.001, Pearson’s χ2 test with Yates correction, Supplementary file 9). Similarly, the other three conserved immunogen vaccines also targeted T cell epitopes with a low rate of escape (Létourneau et al., 2007; Mothe et al., 2015a; Kulkarni et al., 2013). This suggests that current conserved immunogen designs are likely to be immunogenic in PLWH on ART and are more likely to target T cell epitopes that do not harbor pre-ART escape.

Discussion

In this study, we employed two unbiased approaches to comprehensively examine the landscape of escape in the HIV-1 reservoir in 25 PLWH durably suppressed on ART. First, we examined the frequency of published HLA-associated polymorphisms in HIV-1 subtype B OGV sequences. Overall, 20% of HLA-associated sites in the HIV-1 reservoir harbored the known HLA-adapted viral form, suggesting potential viral escape from T cell responses. This number increased to 48% if both known and putative HLA-adapted forms were included in the calculation. These estimates are consistent with substantial immune escape in the reservoir, but nevertheless support the notion that the majority of HLA-associated sites in the reservoir maintain susceptibility to host T cell responses. Next, in all participants, we mapped HIV-1-specific T cell epitopes across the full HIV-1 proteome. Across the cohort, the mean breadth of HIV-1 epitopes targeted per person was 7. T cell targeting of HIV-1 proteins was consistent with studies of untreated infection, with T cells most often targeting Gag, Pol, and Nef (Addo et al., 2003). No differences were observed between treatment initiated in acute (n = 3) or chronic infection (n = 22). We then systematically examined whether non-synonymous mutations in reactive epitopes impacted T cell recognition and functional responses. Consistent with the overall HLA-adaptation estimates from the bioinformatics analysis, we experimentally confirmed that 32% (49/151) of all reactive T cell epitopes in HIV-1 OGVs contained mutations that conferred T cell escape, and that conversely, 68% of circulating HIV-specific T cell responses were able to fully recognize the autologous replication-competent-reservoir virus.

The minimal empirically defined T cell escape frequency of 32% observed in PLWH on ART in this study is consistent with studies of T cell escape in untreated HIV-1 infection. A 30% escape frequency was observed in primary untreated HIV-1 infection, in which HIV-1-specific T cell responses were comprehensively mapped against the autologous transmitted/founder virus in participants, and T cell reactivity was assessed against mutant viruses that emerged over the first 6 months of infection (Liu et al., 2013). In that study, similar to ours, CD8+ T cell escape occurred more frequently in epitopes of higher Shannon entropy (Liu et al., 2013; Ganusov et al., 2011; Bansal et al., 2005). A similar escape frequency of 33% was also observed in a large cross-sectional study of chronically infected HIV-1 Clade C untreated individuals (n >2,000), in which the T cell escape frequency was determined by examining participant HLA-associated polymorphisms in immunoprevalent HIV-1 CD8+ T cell epitopes (Carlson et al., 2012). In PLWH durably suppressed on ART, HIV-1-specific T cell responses are stable and maintain immunodominance hierarchies over years (Xu et al., 2019). Both the stability of the HIV-1-specific T cell response and the comparable level of population-level virus escape in treated and untreated individuals is consistent with studies of HIV-1 reservoir formation and kinetics. These studies showed that the HIV-1 reservoir is mostly formed at the time of ART initiation (Brodin et al., 2016; Abrahams et al., 2019; Pankau et al., 2020) and its subsequent rate of decay is very slow (Crooks et al., 2015; Siliciano and Siliciano, 2015; Siliciano and Siliciano, 2004), with only limited variation in the number of proviruses, or the types of proviral defects observed in PLWH on ART over time (Lu et al., 2018). More recently, longitudinal analyses of near full-length proviral sequences in PLWH on ART showed no enrichment in escaped epitopes over time suggesting that HIV-1-specific T cells do not significantly alter the provirus landscape in people durably suppressed (Antar et al., 2020). Furthermore, a recent study reported that brief analytical treatment interruption (ATI) (median of 5 weeks) does not significantly impact the composition of virus populations (phylogenetic analyses sequencing of HIV-1 env) or size (total DNA, cell-associated RNA, IUPM) of the latent HIV-1 reservoir (Salantes et al., 2018). Collectively, these studies suggest a model in which circulating T cells, as well as virus escape variants in infected cells at the time of ART initiation, are maintained throughout durable viral suppression, and during brief periods of ATI.

Our observations contrast somewhat with those of Deng and colleagues who reported a predominance of CD8+ T cell-resistant viruses in proviral HIV-1 DNA (Deng et al., 2015). Deng et al did not empirically investigate proteome-wide patterns of escape (i.e. escape in 32% of reactive T cell epitopes) in the replication-competent HIV-1 reservoir, as reported here. Rather, their study focused on predicting the ‘proportion’ of escape within epitopes in Gag. They mostly used sequencing and bioinformatics approaches to conclude that, in epitopes in which virus variation was observed, in individuals treated in chronic infection, most (>98%) proviruses contained variants previously reported to confer T cell escape. Our empirical testing, which focused on T cell escape in the persistent HIV-1 reservoir, did not yield the same high proportion of escape variants reported in that study. Indeed, the average within-epitope escape frequency in 49 empirically tested epitopes was much lower at 27% (Supplementary file 5). This dataset included escape variants that can emerge early after infection such as the Gag T242N escape in Gag 240–249, restricted by B*58:01 and B57:01/3 (Leslie et al., 2004; Brumme et al., 2009). Deng et al also noted a higher within-epitope escape in the DNA reservoir in individuals treated in chronic versus acute infection. While only a few participants in this study were treated in acute infection, we observed no difference in within-epitope escape frequencies between participants treated in chronic or acute HIV-1 infection. Together this suggests that the differences in the studies did not arise from the regions of HIV-1 studied, or the differences in stage of HIV-1 infection at time of ART initiation. Rather, we propose that differences in within-epitope escape frequencies between studies derived from two sources. First, our empirical approach showed clearly that HIV-1 variants often do not escape T cell recognition and therefore prediction methods may overestimate the level of escape. Indeed, Deng et al did include examples where virus variants predicted to confer escape did not decrease T cell responses following experimental testing. Second, the HIV template, HIV DNA versus OGV RNA, used for virus sequencing largely differed between studies.

Our observations that epitopes mostly harbor a mix of escape and non-escape variants are consistent with previous studies in untreated infection reporting that T cell escape is rarely all-or-nothing, but rather occurs at different rates (days to years) with multiple variants emerging over time (Goonetilleke et al., 2009; Liu et al., 2013; Asquith et al., 2006). These changes reflect both ongoing compensatory changes to maintain viral fitness and selection pressure from different T cell clonotypes (Brockman et al., 2010; Ladell et al., 2013; Ritchie et al., 2014), and progressive dysregulation of the immune response. While the HIV-1 reservoir is mostly formed at ART initiation, almost 30% of sequences enter the HIV-1 reservoir much earlier. The mixture of escape and non-escape OGVs observed in this study likely represent OGVs at different stages of T cell escape entering the reservoir at different stages of infection.

There are limitations of this study. First, detection of escape variants may have been limited by the sequencing depth obtained from OGVs from QVOAs. In addition, different stimuli can activate different viruses, and even the negative wells from the QVOA can generate new viruses after new stimulation, contributing to an underestimation of, or a bias in the replication-competent viruses present in the culture (Ho et al., 2013Hosmane et al., 2017). Applying a rule of three calculation, our average depth of sequencing (25 sequences/participant) provided 95% confidence that any undetected variant comprised <12% of the population. While sampling biases were minimized by our choice of sequencing methods (nearly full-length and Primer ID), which have a low probability of introducing PCR mutations (Zhou et al., 2015; Salazar-Gonzalez et al., 2008), we cannot exclude that a low-frequency variants not detected by sequencing may have conferred escape (Abrahams et al., 2019). It is notable that that participant OGVs sequenced over time did not exhibit directional selection suggesting that the QVOA method recovers outgrowth virus that is representative of all variants (escaped and non-escaped) in the HIV-1 reservoir. Second, our bioinformatics analysis of HLA-associated polymorphisms in HIV-1 reservoir sequences simply identifies viral variants that are enriched among individuals expressing a particular HLA allele, as defined using population-level analyses (Carlson et al., 2012), and thus provides only a crude estimate of adaptation burden. This is because these lists of HLA-associated polymorphisms do not capture every escape variant that occurs in vivo; moreover, not every HLA-associated viral site defined at the population level is necessarily targeted by every individual expressing the relevant HLA allele. It is notable, however, that the results of the bioinformatics approach were consistent with our empirical studies, suggesting that the estimates made correctly captured the range of escape frequencies within this cohort. Third, in our empirical studies, we cannot exclude that the use of clade B consensus peptides underestimated the total breadth of the T cell response (Goonetilleke et al., 2009; Altfeld et al., 2003). However, our approach of synthesizing autologous peptides at reactive epitopes, and experimentally testing all ‘stripes’ (non-synonymous changes in >40% of sequences, see Materials and Methods) in our OGVs to identify additional epitopes to study, minimizes this limitation. Fourth, our empirical approach did not detect all mechanisms of escape such as possible immune dysfunction and processing mutations. To fully assess loss of CD8+ T cell responses due to immune dysfunction, longitudinal mapping and sequencing studies pre -and post-ART are needed. While no previously documented processing mutations were detected in our dataset, we anticipate the bioinformatics analysis which enumerated all HLA-associated polymorphisms regardless of their location with respect to known (or novel) T cell epitopes, will have captured some processing mutations. Lastly, we did not confirm the CD4 or CD8 restriction of all reactive and escaped T cell epitopes in this study. However, both consistent with previous studies (McNeil et al., 2001; Appay et al., 2002; Betts et al., 2001; Riou et al., 2012) and the optimal epitope mapping performed here, it is likely that the bulk of T cell responses identified in this study were CD8-restricted.

In summary, empirical testing supported by bioinformatics analysis, showed that 68% of circulating HIV-1 specific T cells across our cohort were able to recognize the replication-competent virus from the latent reservoir. Moreover, T cells in most individuals targeted one or more epitopes with no detectable escape and within epitopes, on average, only a minority of OGVs conferred escape. Treatment interruption studies have shown that circulating HIV-1 specific T cell responses are insufficient to fully suppress virus rebound in most individuals. Recent work has shown recombinant viruses rapidly emerge following HIV rebound (Vibholm et al., 2019; Lu et al., 2018; Cohen et al., 2018). Our study, supported by animal models (Pandrea et al., 2011) and case reports investigating viral rebound (Smith et al., 2015), suggest that circulating HIV-1-specific T cells very likely afford some contribution in control of virus rebound. However, T cell immunity is rapidly overwhelmed, in part due to recombination-mediated escape (Ritchie et al., 2014; Streeck et al., 2008). Altogether, this suggests that HIV-1 cure strategies that include efforts to augment T cell immunity in PLWH on ART should include immunogen designs that target existing T cell epitopes, particularly those with low entropy which, as we show here, are less likely to harbor escape in the replication-competent reservoir. Ultimately, success of these therapies will likely be dependent on the ability to exert very rapid virus control and to limit the emergence of viral recombinants.

Materials and methods

Study cohort

Request a detailed protocol

Participants were enrolled through either the UNC Chapel Hill HIV-1 Clinical Trials Unit, the Women’s Interagency HIV-1 Study (WIHS) UNC Chapel Hill Site, or the University of California San Francisco (UCSF) site. All participants were adults (>18 years of age) with documented HIV-1 infection. Infection was confirmed by either a positive HIV test by licensed ELISA test and confirmed by western Blot (WB), multispot HIV assay, HIV-1 RNA, or two documented plasma HIV-1 RNA >1000 copies/ml.

Enrollment criteria included a stable ART regimen for ≥12 months prior to consent/enrollment. Stable ART was defined as ≥nucleoside/nucleotide reverse transcriptase inhibitors plus a non-nucleoside reverse transcriptase inhibitor, integrase inhibitors, or a protease inhibitor without interruption (missing doses for no more than 2 consecutive days or no more than 4 cumulative days) in the 24 weeks immediately prior to consent/enrollment. Other potent fully suppressive ART combinations were allowed on a case-by-case basis, after clinician review. Prior changes in or elimination of medicines for easier dosing, intolerance, or other reasons were permitted if an alternative suppressive regimen was maintained. Second, plasma HIV-1 RNA was required to be <50 copies/mL for ≥12 months prior to enrolment as defined by a minimum of two tests (limit of detection determined by assay employed). Third, within 60 days of consent/enrollment, plasma HIV-1 RNA was required to be <50 copies/ml and hemoglobin required to be ≥12 g/L. Treatment during acute HIV-1 infection was defined as treatment within 30 days of either HIV-1 PCR positive but HIV-1 antigen negative, or a HIV-1 antigen discordant test.

Class I HLA typing was performed on 24/25 participants the UNC Center for AIDS Research Core Laboratory. Study characteristics of the cohort are summarized in Table 1.

Study approval/IRB approvals

Request a detailed protocol

Participants were enrolled with the following IRB approved studies: (1) CID 0819 - Apheresis Procedures to Obtain Leukocytes for Research Studies from HIV Positive Participants (08–1575), (2) The UNC Women’s Interagency HIV Study (WIHS) (12-1660). Review and implementation of all protocols utilized for the collection of samples for this this analysis were approved by the University of North Carolina at Chapel Hill Biomedical Institutional Review Board (IRB) and the University of California at San Francisco IRB. All participants provided written informed consent. All experimental protocols were approved by local Institutional Biomedical Review Boards (ethics numbers: 14–0741, 11–0228, and 13–3613, 15–1626) and performed in accordance with the relevant guidelines. Study characteristics of the cohorts are summarized in Table 1.

PBMC isolation

Request a detailed protocol

PBMC were isolated from apheresis product by centrifugation (×1200 g for 15 min at room temperature) on a Ficoll-Paque density gradient (GE Healthcare Life Sciences Ficoll-Paque Plus). Briefly, the apheresis product was diluted 1:2.5 in 2% FBS/PBS and 30 ml diluted product was overlaid onto 15 ml of Ficoll-Plaque in a 50 ml Falcon tube prior to centrifugation. PBMCs were harvested, then washed three times in 2% FBS/PBS. PBMC were counted and then frozen (Goonetilleke et al., 2006).

Quantitative viral outgrowth assay (QVOA) and infectious units per million (IUPM)

Request a detailed protocol

QVOAs were performed as previously described (Crooks et al., 2015). Briefly,~5 × 107 million resting CD4+ T cells, defined as CD4+ CD45+ CD3+ CD69 CD25 CD8 CD14 HLA-DR, were isolated via negative selection from PBMC (Keedy et al., 2009). CD4+ T cells were maximally stimulated in limiting dilutions with phytohemagglutinin (PHA) (Remel-PHA; Thermo Scientific), IL-2, and irradiated CD8-depleted PBMC from a seronegative donor, which were added twice to cultures to support virus propagation. The CD8-depleted PBMC were obtained from seronegative donors previously screened to ensure adequate CCR5 expression. Culture supernatants were collected on days 15 and 19 and assayed for p24 expression by ELISA (ABL, Rockville, MD). Only cultures that contained an equivalent or greater level of p24 antigen on day 19 compared with day 15 were scored as positive. A maximum likelihood method was used on day 15 data to estimate the frequency of latent HIV infection, reported as bias-corrected Infectious Units Per Million (IUPM) (Supplementary file 1; Trumble et al., 2017). For 10 participants, longitudinal IUPM measurements were available. IUPM values measured in the 2 years prior to initial T cell mapping were averaged (range of IUPM values: 1–6, mean number of IUPM values/participant: 2). The limit of detection was defined as ≤1 p24 positive well (range 12–36 cultures, mean 18 cultures) at the highest input of cells (2.5 million cells).

Replication-competent reservoir HIV-1 virus sequencing

Request a detailed protocol

Supernatants from end point dilutions of p24 positive wells from the QVOA were sequenced for replication-competent virus. Twenty-two participants had supernatants sequenced by nearly full-length sequencing, and one participant had supernatant sequenced by Primer ID. For seven participants, longitudinal p24 supernantants were available (2–4 time points over 4.4 years). In two participants, no sequencing was possible as no p24 supernatants were identified in the QVOA; in these assays IUPM was calculated as below the threshold of detection.

Primer ID sequencing for target regions

Request a detailed protocol

A multiplexed Primer ID library that covered partial gag, integrase coding domain, V1/V2 region, partial gp41 and partial nef region was constructed using the previous protocol (Zhou et al., 2015; Dennis et al., 2018). In brief, viral RNA was extracted from culture supernatants using QIAmp viral mini kit (Qiagen). cDNA was synthesized using a cDNA primer mixture targeting the above regions with a block of random nucleotides in each cDNA primer serving as the Primer ID, and SuperScript III reverse transcriptase (ThermoFisher). After two rounds of purification of cDNA, we amplified the cDNA and attached Illumina adaptor sequences to the PCR products. Gel-purified libraries were pooled and sequenced using MiSeq 300 bp pair-end sequencing (Illumina). The TCS pipeline (https://github.com/SwanstromLab/PID) was used for de-mutiplexing of the sequenced regions and construct Primer ID Template Consensus Sequences (TCSs) for each sequenced viral template.

Nearly full-length sequencing

Request a detailed protocol

Viral RNA was converted to cDNA using Superscript III reverse transcriptase and an oligo(dT) primer. The 3’ and 5’ half genomes were amplified in separate PCR reactions using barcoded primers, and the PCR products were pooled, and gel purified. The SMARTbell Template Prep Kit (PacBio) was used to add adaptors to amplicons, and amplicons were then submitted for PacBio sequencing (movie time of 10 hr). Sequences were demultiplexed by barcode, and then analyzed using the PacBio Long Amplicon Analysis (LAA) package. Sequences were aligned separately using MUSCLE v3.8.31 and in-house scripts were used to examine the frequencies of mutations (Edgar, 2004a; Edgar, 2004b). Neighbor-joining phylogenetic trees were built using Ninja v1.2.2 (http://wheelerlab.org/software/ninja/), and approximate maximum likelihood trees were built using FastTree 2.1 (Price et al., 2010).

After aligning using MUSCLE and trimming of positions with indels, we performed two-by-two comparisons of the sequences at 5HG and 3HG and calculated the percent of comparisons with 0 to 3 nucleotide differences among the total comparisons. We chose one nucleotide difference as the clonality cut-off as we estimated that the QVOA and the sequencing assay might generate two substitutions per HIV genome. Using this cut-off, the percent of clonality was defined as the percentage of two-by-two comparisons yielding 0 to 1 nucleotide differences among all the comparisons. The Ruby module viral_seq was used to calculate the pair-wise diversity of sequences (https://rubygems.org/gems/viral_seq).

PacBio sequencing proficiency was confirmed by the strong correlation between participant IUPM and the number of full-length sequences obtained, both of which are determined by the number of p24+ ELISA wells in the QVOA assay (r = 0.755, p<0.0001), Spearman Rank, 24 pairs, (Figure 3—figure supplement 2).

Droplet digital polymerase chain reaction (ddPCR)

Request a detailed protocol

The concentration of HIV-1 DNA copies in each sample was determined using the following primers and probes spanning HXB2 coordinates 684–810 (Malnati et al., 2008): LTRgagF: 5’-TCTCGACGCAGGACTCG-3’, LTRgagR: 5’- TACTGACGCTCTCGCACC- 3’, and LTRgagProbe: 5’/56-FAM/CTCTCTCCT/ZEN/TCTAGCCTC/31ABkFQ/- 3’. Cell concentration was determined using the following RPP30 primers and probes: RPP30-F: 5’- GATTTGGACCTGCGAGCG-3’, RPP30-R: 5’-GCGGCTGTCTCCACAAGT-3’, RPP30-Probe: 5’-/56-FAM/CTGACCTGA/ZEN/AGGCTCT/31ABkFQ/- 3’ (Hindson et al., 2011). LTRgag ddPCR reactions contained the following: 10 μl ddPCR Supermix for Probes (no dUTP) (Bio-Rad), 1 μl each primer (LTRgagF and LTRgagR) at a concentration of 20 μM, 0.35 μl of LTRgag probe at a concentration of 20 μM, 6.65 μl DEPC-treated water (Thermo Fisher Scientific), and 3 μl of undiluted DNA for a total volume of 22 μl. RPP30 ddPCR reactions contained the following: 10 μl ddPCR Supermix for Probes (no dUTP) (Bio-Rad), 1 μl each primer (RPP30-F and RPP30-R) at a concentration of 20 μM, 0.35 μl of RPP30 probe at a concentration of 20 μM, 8.65 μl DEPC-treated water (Thermo Fisher Scientific), and 1 μl of DNA diluted 1:100 in DEPC-treated water. 8E5 DNA was used as a positive control and no-template controls were included in each run. Samples were run in duplicate and the values averaged.

Droplets were generated using the QX200 Automated Droplet Generator (Bio-Rad), in a final volume of 40 μl. Following droplet generation, plates were sealed with a pierceable foil seal and immediately subjected to the following PCR conditions on a C1000 Touch Thermal Cycler with 96-Deep Well Reaction Module (Bio-Rad): 95°C for 10 min, then 45 cycles of 95°C for 30 s followed by 60°C for 1 min, a final extension of 98°C for 10 min and a 4°C hold. Following thermal cycling, plates were read on the QX200 Droplet Reader (Bio-Rad) with the following set-up: Experiment = Rare Event Detection, Mix = ddPCR Supermix for Probes (no dUTP), Target 1 = FAM, Target 2 = VIC.

Analysis of ddPCR data used QuantaSoft (Bio-Rad) with thresholds for the detection of each target set using the negative controls. The same threshold was used for all reactions of a particular target (either LTRgag or RPP30) in a given run. The concentration of HIV-1 copies / μl of sample was calculated as follows: Quantasoft LTRgag concentration ×(total volume of PCR reaction/volume of DNA added). The concentration of cells/μl of sample was calculated as follows: Quantasoft RPP30 concentration x (total volume of PCR reaction/volume of DNA added) x dilution factor x 0.5.

DNA extraction

Request a detailed protocol

Total DNA was extracted from purified resting CD4+ T cells using the DNeasy Blood and Tissue Kit (Qiagen), following the manufacturer’s instructions with the following modifications. DNA was eluted from the column using 200 μl of DEPC-treated water (Thermo Fisher Scientific). Following a one-minute centrifugation at 6000 x g, the eluate was re-applied to the membrane, incubated for 1 minat room temperature, and re-eluted. DNA was stored at −20°C until use.

Amplification of gag and nef

Request a detailed protocol

A nested PCR approach was used to generate partial gag and nef amplicons. For amplification of partial gag amplicons, the first-round PCR primers were Gag_F_Outer_1: ‘5- GGACGGCGACTGGTGAGTACG −3’ (HXB2 coordinates 732–752) and Gag_R_Outer: 5’ – CCAATTCCCCCTATCATTTTTGGTTTCC −3’ (HXB2 coordinates 2404–2377). Each reaction contained 14.8 μl of DEPC-treated water, 2 μl of 10 × High Fidelity Buffer, 0.8 μl of 50 μM MgSO4, 0.4 μl of 10 μM dNTP mix, 0.4 μl of forward and reverse primer at 20 μM and 5U of Platinum Taq DNA Polymerase High Fidelity (Thermo Fisher Scientific). One copy of HIV-1 DNA was added to each reaction contained, based on ddPCR estimates. A no template control was included in each run. First round thermal cycling conditions were as follows: an initial denaturation of 2 min at 98°C, followed by 35 cycles of: 15 s at 98°C for denaturation, 30 s at 56°C for annealing, and 2 min 30 s at 72°C for extension, followed by a final extension at 72°C for 10 min and a 4°C hold. Second round PCR primers were as follows: Gag_F_Inner_2: 5’ – CTTTTGACTAGCGGAGGCTAGAAGG −3’ (HXB2 coordinates: 759–783) and Gag_R_Inner_3: 5’- GAGCTTCCCTTAGCTGACCCTCTAC −3’ (HXB2 coordinates: 2319–2295). Second round reactions were set up as described above, and 1 μl of first-round PCR product. Second round thermal cycling conditions were same as those described above, however, with an annealing temperature of 55°C.

A nested PCR approach was also used to generate nef amplicons. First round PCR primers were: U5-623F: 5’ -AAATCTCTAGCAGTGGCGCCCGAACAG −3’ (HXB2 coordinates 623–634) and U5-601R: 5’-TGAGGGATCTCTAGTTACCAGAGTC-3’ (HXB2 coordinates 9686–9662). Reaction components were identical to those described above. One copy of HIV-1 DNA was added to each reaction and a no-template control was included in each run. First round thermal cycling conditions were as follows: an initial denaturation of 2 min at 92°C, followed by 10 cycles of 10 s at 92°C for denaturation, 30 s at 60°C for annealing, 10 min at 68°C for extension, then 20 cycles of 10 s at 92°C for denaturation, 30 s at 55°C for annealing, 10 min at 68°C for extension, followed by a final extension of 10 min at 68°C, and a 4°C hold. The second-round primers were as follows: 4653F: 5’-CCCTACAATCCCCAAAGTCAAGGAG-3’ and OFM19: 5’- GCACTCAAGGCAAGCTTTATTGAGGCTTA-3’. Each reaction consisted of 34.8 μl of DEPC-treated water (Thermo Fisher Scientific), 5 μl of 10 × High Fidelity Buffer, 2 μl of 50 μM MgSO4, 5 μl of 0.2 μM forward and reverse primers and 5 U of Platinum Taq DNA Polymerase High Fidelity (Thermo Fisher Scientific), and 2 μl of first round PCR product. Second round thermal cycling conditions were as follows: 2 min at 94°C for the initial denaturation, then 10 cycles of 15 s at 94°C for denaturation, 30 s at 55°C for annealing, and 8 min at 68°C for extension, then 25 cycles of 15 s at 94°C for denaturation, 30 s at 55°C for annealing, and 8 min at 68°C for extension with an additional 20 s of extension time added each cycle; a final 7 min at 68°C for extension, followed by a hold at 4°C. Second round PCR products were screened for positives on a 0.8% agarose gel stained with SYBR Safe DNA Gel Stain (Thermo Fisher Scientific), and visualized with a UV gel imager. Amplicons were sequenced using Sanger sequencing.

HLA-associated polymorphisms methods

Request a detailed protocol

HLA-associated polymorphisms have previously been identified across the HIV-1 subtype B proteome via statistical association (Carlson et al., 2012). Briefly, HLA-adapted forms denote viral residues that are enriched among individuals expressing the restricting HLA (i.e. the inferred escape form), whereas HLA-nonadapted forms denote viral residues that are underrepresented among individuals expressing the restricting HLA (i.e. the inferred susceptible form) (Carlson et al., 2012). For each HIV-1 reservoir sequence, we identified all HIV codons under selection by one or more host HLA alleles (as defined in Carlson et al., 2012) and classified the autologous HIV residue as 'nonadapted', ‘adapted’ or 'possibly adapted' (where the latter extended to any variant other than the 'nonadapted' form). For each participant, a minimum bound of HLA-associated adaptation in the reservoir was estimated by computing the total proportion of HLA-associated sites exhibiting the 'adapted' form, while a maximum escape bound was estimated by including both 'adapted' and 'possibly adapted' forms. HLA-associated polymorphisms were analyzed at the HLA subtype (4-digit) level, where ambiguous allele calls were assigned to the most common allele in the list. Note that, for this analysis, Pol residue numbering begins at position 1 of protease.

Peptides

18-mer peptides overlapping by 10 amino acids were synthesized (Sigma-Genosys, US) to match the clade B HIV-1 consensus sequence (https://www.hiv.lanl.gov/). The corresponding 18-mer autologous outgrowth reservoir virus peptide/s for all previously defined reactive 18-mers from the initial mapping ELISpot were also synthesized. In addition, optimal CD8+ T cell peptides predicted through LANL’s epitope location finder (‘ELF’) software based on HLA-prediction software (https://www.hiv.lanl.gov/content/sequence/ELF/epitope_analyzer.html), and 18-mer peptides spanning stripes (see definition below) were synthesized. To cover variation in autologous OGV, a total of 656 peptides were synthesized.

T cell mapping against HIV-1 clade B consensus sequence

Request a detailed protocol

Clade B consensus peptides (n = 386) were arranged into pools in a matrix format using the Peptide Portal program (SCHARP), adapting code from the Deconvolute this! Program (Precopio et al., 2008). Each peptide was repeated 3 × in the matrix peptide plate. Cryopreserved PBMCs were thawed and rested overnight before being placed in the ELISpot plates (Merck, Millipore) at 4 × 105 cells per well. Peptides were added to the plate with a final concentration of 4 μg/ml and incubated for 18–20 hr at 37°C, 5% CO2. Coating, development (MabTech), and reading of ELISpot plates (AID Reader) has been described previously (Goonetilleke et al., 2006). Data from positive pools were then deconvoluted (Roederer and Koup, 2003) to identify candidate epitope peptides. Putative positive 18-mer peptide–specific T cell responses were confirmed in triplicate in a follow-up IFN-γ ELISpot with either 2- or 4 × 105 cells per well and a peptide concentration of 10 μg/ml, with 4 to 6 negative control wells (media only) and at least one positive control well (10 μg/ml PHA; Sigma-Aldrich) together. Positive T cell responses were defined as ≥12.5 SFU per million, ≥4 × background and no zero values in any replicate of antigen-stimulated wells.

Evaluating T cell escape in epitopes previously defined as reactive in clade B consensus mapping

View detailed protocol

Testing OGV variants for T cell escape (replicates, controls, positivity criteria) was the same as for the initial mapping studies with the exception that peptides were tested at 10 µM. Escape variants were defined as eliciting a 50% decrease in the mean magnitude compared to the peptide that elicited the highest magnitude T cell response in IFN-γ ELISpot. Optimally defined CD8+ epitopes (8–11 aa) were defined as equal to, or greater than the T cell magnitude elicited by the 18-mer peptide (Carlson et al., 2012).

Examining virus stripes and processing mutations

Request a detailed protocol

We observed clustered mutations in OGV sequence data outside of mapped T cell epitopes. We hypothesized that these regions may reflect pre-ART escape against T cell responses not detected using our matrix mapping approach. Regions where > 40% of the sequences contain non-synonymous mutations in the same position in the proteome were defined as ‘stripes’. Twenty-four stripes were identified across participants. 18-mer peptides spanning stripes were synthesized and tested individually in ELISpot. As forecast, some T cell responses were detected. In total T cells recognized epitopes at 11/24 stripes, resulting in a total of 166 reactive T cell epitopes in 25 participants. Prediction software and HLA typing was used to screen virus sequences for the presence of antigen processing mutations (https://www.hiv.lanl.gov/content/immunology/variants/variant_search.html?db=ctl). No antigen processing mutations associated with reactive T cell epitopes were detected in the cohort.

Entropy calculations

Request a detailed protocol

A Shannon entropy score (Shannon, 1948) was calculated for all reactive epitopes in this study, 18-mer and all experimentally confirmed optimal epitopes (Supplementary file 4). For each reactive epitope, the LANL sequence database was used to identify submitted Clade B sequences. Between 701 and 1936 sequences were analyzed per epitope. The Shannon Entropy-Two tool from the LANL database was used to identify the variation in protein sequence alignments (reactive epitope sequence vs. all known sequences), and an entropy score was calculated by averaging the entropy score for each amino acid position (https://www.hiv.lanl.gov/content/sequence/ENTROPY/entropy.html).

Statistics

All statistical analyses were performed using Graph Pad Prism Version 8. Data were assumed to be non-normally distributed and as appropriate, two-tailed tests applied. A p value of ≤ 0.01 was considered significant.

Data availability

Sequencing data have been deposited in Gen Bank under PRJNA666896, MT307344-MT308415 and MW054719-MW054856 All data generated (raw) or analyzed during this study are included in the manuscript and supporting files (supplementary files).

The following data sets were generated

References

    1. Bansal A
    2. Gough E
    3. Sabbaj S
    4. Ritter D
    5. Yusim K
    6. Sfakianos G
    7. Aldrovandi G
    8. Kaslow RA
    9. Wilson CM
    10. Mulligan MJ
    11. Kilby JM
    12. Goepfert PA
    (2005)
    CD8 T-cell responses in early HIV-1 infection are skewed towards high entropy peptides
    Aids 19:241–250.
  1. Conference
    1. Elliott J
    2. Solomon A
    3. Wightman F
    4. Smith M
    5. Palmer S
    6. Prince M
    7. Watson J
    8. Hoy J
    9. McMahon M
    10. Lewin S
    (2013)
    The safety and effect of multiple doses of vorinostat on HIV transcription in HIV-Infected patients receiving combination antiretroviral therapy
    Conference On Retroviruses and Opportunistic Infections.

Decision letter

  1. Julie Overbaugh
    Reviewing Editor; Fred Hutchinson Cancer Research Center, United States
  2. Tadatsugu Taniguchi
    Senior Editor; Institute of Industrial Science, The University of Tokyo, Japan
  3. Steven Deeks
    Reviewer; University of California, San Francisco, United States
  4. Sarah Rowland-Jones
    Reviewer

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This work provides a detailed study to investigate the question of whether or not HIV harboured in the latent reservoir of patients well-controlled on HIV therapy can be recognized by their circulating T cells. In a carefully-performed study of 25 subjects, the authors use viral outgrowth culture and sequence analysis, followed by functional T cell assays, to show that most of the known epitopes remain subject to T cell recognition. This is an important result for HIV "cure" strategies, for which T cell recognition is likely to be an important component.

Decision letter after peer review:

Thank you for submitting your article "The HIV-1 latent reservoir is largely sensitive to circulating T cells" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Tadatsugu Taniguchi as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Steven Deeks (Reviewer #1); Sarah Rowland-Jones (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

This manuscript describes a detailed study to investigate the question of whether or not HIV harboured in the latent reservoir of patients well-controlled on ART can be recognised by their circulating T cells. In a carefully-performed study of 25 subjects, the authors use viral outgrowth culture and sequence analysis, followed by functional T cell assays, to show that over ⅔ of known B clade epitopes remain subject to T cell recognition. This is an important result for HIV "cure" strategies, for which T cell recognition is likely to be an important component.

Summary:

Warren et al. report that the majority of detectable CD8+ T cell responses are able to recognize the latent viral reservoir in people living with HIV on antiretroviral therapy. The authors employ sequencing of outgrowth virus and epitope mapping of autologous CD8+ T cell responses across both consensus and observed variant sequences spanning the HIV proteome to assess mutational escape. On a per-person level, six of 23 persons had no observed escape mutations and two had complete escape, with the remaining 15 having escape from some but not all T cell responses, as defined by sequencing of outgrowth viruses and IFN-γ ELISpot assays. The work provides additional information for CD8+ T cell-based strategies to reduce or eliminate replication-competent proviral reservoirs, and is consistent with reports of T cell dysfunction as an additional mechanism of inadequate viral control and reports of CD8+ T cell contributions to limited viral control during and after cessation of ART.

Revisions for this paper:

1) It may be possible for virus mutations to arise during viral outgrowth, including reversion of mutations that incur fitness costs, which could have significant impacts on the study conclusions. Methods exist to determine full length proviral sequencing from the latent reservoir. OGV can be compared to matched proviral DNA sequencing from ex vivo resting CD4 T cells to address this. This seems critical to interpretation of the data.

2) Outgrowth in the absence of CD8 T cell pressure, as happens in the outgrowth assay, may preferentially disfavor escaped variants with fitness costs, which limits the ability for OGV sequencing alone to provide an accurate measurement of escape. Again, this speaks to the need for full length sequencing from the reservoir.

3) It is unclear why confirmation of CD8+ T cell responses using optimal epitopes was only achieved for a minority of the detected responses, which raises concerns about the majority of unconfirmed responses and whether they are suitable for inclusion in the analysis.

4) The use of ELISpot assays does not take into account flanking region mutations that are well known to alter antigen processing and serve as a mechanism of escape. Although this is perhaps beyond the scope of the study, it should be addressed as a caveat.

5) The authors acknowledge that their method for testing whether a variant epitope peptide escapes recognition or not is not completely comprehensive, and would miss – for example – variants that affect antigen processing or binding affinity (particularly the off-rate, as shown in reference 20 for the immunodominant B27 p24 epitope). Could they estimate – perhaps using data from the Molecular Immunology database? – the proportion of escape variants that might be missed using their methods?

6) The authors study four patients who initiated ART during acute HIV infection – the rest began therapy in chronic infection. One might predict that the length of the period of infection before initiating ART would influence the likelihood of developing escape mutations – do the authors have any data to support this hypothesis?

Revisions expected in follow-up work:

One reviewer felt that sequence data was needed to support this story's outlined in points 1 and 2, but other reviewers felt that this could be fairly limited and focused on the sequencing of the QVOA "non-escape" subjects. This more reasonable sequencing effort would help address the concerns of the one reviewer without requiring large amounts of sequence data.

https://doi.org/10.7554/eLife.57246.sa1

Author response

Revisions for this paper:

1) It may be possible for virus mutations to arise during viral outgrowth, including reversion of mutations that incur fitness costs, which could have significant impacts on the study conclusions. Methods exist to determine full length proviral sequencing from the latent reservoir. OGV can be compared to matched proviral DNA sequencing from ex vivo resting CD4 T cells to address this. This seems critical to interpretation of the data.

One reviewer felt that sequence data was needed to support this story's outlined in points 1 and 2, but other reviewers felt that this could be fairly limited and focused on the sequencing of the QVOA "non-escape" subjects. This more reasonable sequencing effort would help address the concerns of the one reviewer without requiring large amounts of sequence data.

We believe that both our assay design and published data are incompatible with in vitro reversion. In our study, sequencing was performed on end-point dilution of virus i.e. on average, a single virus-infected cell. A numbered detailed response is provided below. In summary, while we have performed additional proviral DNA sequencing to address the reviewer’s concerns (detailed in #3) we do not consider it critical for interpretation of these data.

1) We are sequencing outgrowth virus from, on average, a single infected cell seeded in 2.5 million resting CD4 T cells. The culture period is 15 days. The effective population size in each well of the QVOA is n=1. Such a low effective population size means that the first mutation occurring in wells is stochastic or random. Deterministic models which assume a large effective population size do not apply here (reviewed in (1)). As one mutation occurs over a HIV cycle, in a 15-day culture we estimate that 4-5 cycles will occur. Therefore 5 mutations, initiated by a random mutation, are expected over the culture period.

Putting this information together, if reversion were to occur to a level to skew our study conclusions, then:

i) The first or second “stochastic” mutation would need to occur at the same or closely located positions in a ~10,000 bp HIV genome in not one, but multiple QVOA wells.

ii) Further, to impact our results this stochastic mutation would need to consistently dominate the viral population in no more than 4 virus cycles in multiple QVOA wells. iii) Further still, the mean T cell breadth was 7 in this study i.e. 7 T cell epitopes across the entire HIV proteome. Therefore, to skew our results five mutations generated over 15 days would need to induce non-synonymous reverting changes in multiple epitopes as well as dominate the viral population.

We would hope the reviewers would agree that the likelihood of i-iii occurring is extremely low to impossible.

This scenario, that is in vitro reversion arising for a single infectious molecular clone (IMC) was tested in Song et al. (Feng Gao, Duke). The authors infected CD4 T cells with an IMC (note, the effective population size would be >>> than endpoint dilution). Cells were cultured for approximately 15 days and virus sequenced using single genome amplification as used in this study. See Figure 8 from Song et al., 2012 for the highlighter plot. SGA sequencing (n=47) showed rare mutations (average = 1.3 nucleotide changes per genome, (range 0-4) in 3’ SGA half (4685bp) and no evidence of directional selection. This paper is now cited in the manuscript with text that discusses the low probability of in vitro reversion.

2) in vivo studies of reversion also suggest that in vitro reversion in QVOA is highly unlikely. In transmission studies including mother-to-child studies in which the recipient lacks the selecting HLA, documented reversion has required months to years (2-6). Reversion to consensus in acute infection studies lacking T cell epitope data has been described to occur more rapidly, though still over many weeks to months (7, 8). Please note that in contrast to QVOA, in acute infection the effective population size of HIV is very high bringing in deterministic models of virus mutation which include the effects of recombination where more rapid virus evolution would be predicted.

3) We hope that #1 and #2 above, make clear why we do not consider additional generation of proviral sequencing necessary for interpretation of data in this paper. However, we agree that inclusion of proviral sequencing is very interesting and further rounds out this paper. Note, our additional data are restricted to n=2 participants with sequencing of genomic regions, not full-length sequencing. This reflects the extraordinary cost of DNA sequencing approaches, cell limitations and genuine technical difficulty of these approaches. For example, generation of these data took 11 weeks and cost $15,000 in reagents.

In brief, we isolated DNA from 5 million resting memory CD4+ T cells (50million PBMC) and sequenced near-full nef and gag regions in 00728 and 00720. Longitudinal reservoir sequencing was available for both participants providing good numbers of comparator outgrowth viruses to address the reviewer’s concerns. The highlighter plot shown in Figure 4—figure supplement 2A shows clearly that the intact DNA sequences (no stop codon) show very high sequence identity with OGVs with no evidence of reversion at or around T cell epitopes (gray boxes). Please note that while identical gene-level sequences were observed, the parent outgrowth viruses were not clonal. As such, the detection of identical DNA and outgrowth virus gene sequences does not represent a dominant clonal sequence, that could skew data.

In summary, consistent with modeling estimates explained in #1 above, there was no evidence of reversion or other selection occurring in HIV DNA not detectable OGVs. These data are discussed in the manuscript, new data included in Figure 4—figure supplement 2 and the Materials and methods.

2) Outgrowth in the absence of CD8 T cell pressure, as happens in the outgrowth assay, may preferentially disfavor escaped variants with fitness costs, which limits the ability for OGV sequencing alone to provide an accurate measurement of escape. Again, this speaks to the need for full length sequencing from the reservoir.

Please see our response above. Given there is on average, a single virus-infected cell in a well, there are no a priori variants that are favored and stochastic, not deterministic, mutation governs the changes in vitro. A random mutation in cycle one that has a fitness benefit would need to sweep through the population in 4-5 virus cycles. Given the consistency of our sequencing data across wells at T cell epitopes, this favorable fitness mutation would also need to randomly occur and then overtake the virus population in 4 cycles in the majority of independent cultures in our assay system. This would need to occur at multiple T cell epitopes. We believe the probably of such events occurring is near impossible. In summary, the absence of CD8 T cell selection pressure has very little effect of the virus population undergoing stochastic changes in this short-term culture.

3) It is unclear why confirmation of CD8+ T cell responses using optimal epitopes was only achieved for a minority of the detected responses, which raises concerns about the majority of unconfirmed responses and whether they are suitable for inclusion in the analysis.

We are sorry to appear so contrary, but we must also strongly disagree with the reviewer’s comment here. A documented T cell response to an 18mer peptide is not “unconfirmed” and all T cell data presented are appropriate for analysis. Our response below is divided into sections A and B. “A” details why a documented T cell responses to a longer peptide is valid and “B” details why in this study did not subsequently map all optimal epitopes.

A) There are >20 years of literature from scores of laboratories to show that longer peptides can be used to successfully map T cell responses and to describe T cell escape. To our knowledge, there are no reports in the literature of a measured T cell response to a longer peptide being a false positive, not also detectable by a shorter peptide. The biology here is that the core epitope of the longer peptides can still sit in the HLA pocket and overhangs do not significantly interfere with T cell recognition. Studies using longer peptides span infectious disease through to cancer, T cell mapping in mouse (9), NHP (10) and humans (11). In HIV, the Walker-Harvard (11) , Goulder-Oxford (12, 13), McElrath-FHCRC and Brander-IrsiCaixa (14) have all routinely mapped HIV specific T cell responses using longer overlapping peptides; this includes clinical testing of T cell vaccines (15, 16). Indeed, the majority of T cell responses in HIV research are reported in terms of responses to overlapping longer (15-20mer) peptide pools spanning a HIV protein, often Gag (17). We and others have used longer peptide to document HIV escape in multiple studies (18-20). Last, we would note that a high proportion of published studies in HIV that examine virus escape, no longer empirically T cell responses and simply rely on prediction methods.

Our Materials and methods section clearly details the rigorous criteria applied for assessing a T cell response and mapping epitopes i) all peptides are tested independently in quadruplicate and only considered reactive when 4x greater than the mean of 6 mock wells) ii) all reactive peptides were then independently re-tested again in quadruplicate in all assays assessing variants. We went further and then generated the corresponding autologous peptides, again testing both the clade B peptide alongside another – this compares with vast majority of HIV studies that map using only consensus peptide sets. Across our dataset we did not observe any discordant data in independent testing. As we note in our paper, our study is the largest proteome-wide mapping study in PLWH on ART to date. We provide all T cell data in an attached spreadsheet to the study as a resource for the field.

B) HLA diversity prevents optimal epitope mapping of every new T cell response. To date, thousands of HLA I alleles have been identified and the binding motifs are only known for a small subset of HLA alleles. This is consistent with previous studies in which no optimal epitope was known/predicted in over 60% of mapped reactive HIV peptides (19-21). To empirically identify the optimal epitope and then define the HLA restriction firstly, 8- 9- and 10- peptides overlapping by 7-9 amino acids need to be synthesized for each reactive longer peptide. Once the shortest reactive epitope has been defined then restriction should then be confirmed using antigen presentation from a cell line expressing a single HLA. To determine optimal CD8+ T cell epitopes for all CD8+ T cell responses in our studies, we estimate that we would need to generate 3,000-4,000 autologous peptides ($60-80K in peptide costs alone). Testing would require additional large blood draws of all our study participants.

To generate the most informative and detailed dataset while maintaining reasonable costs, we used bioinformatics approaches combine with HLA typing to predict the optimal epitope for common HLA alleles in the human population. Of 166 epitopes, bioinformatic algorithms predicted >120 optimal epitopes. Empirical testing showed that only 20% of these predictions were correct, demonstrating that bioinformatics approaches are not perfect.

Summary: With respect, we feel the reviewer is holding us to an unrealistic and more important, an unnecessary standard. We reiterate that all T cell responses described in this paper were confirmed and comply with the most rigorous standards for measurement in our field.

4) The use of ELISpot assays does not take into account flanking region mutations that are well known to alter antigen processing and serve as a mechanism of escape. Although this is perhaps beyond the scope of the study, it should be addressed as a caveat.

5) The authors acknowledge that their method for testing whether a variant epitope peptide escapes recognition or not is not completely comprehensive, and would miss – for example – variants that affect antigen processing or binding affinity (particularly the off-rate, as shown in reference 20 for the immunodominant B27 p24 epitope). Could they estimate – perhaps using data from the Molecular Immunology database? – the proportion of escape variants that might be missed using their methods?

We agree that our empirical method will miss flanking variants that affect antigen processing or binding affinity and is noted as a limitation in our Discussion. As per the reviewer’s suggestion we looked for flanking / processing mutations to reactive epitopes on the LANL database and found none detailed.

The bioinformatic approach developed by Dr Brumme, specifically “possible adaptations” extends to processing and binding affinity ns changes previously associated with specific HLA in untreated infection. Consistent with this, the predicted level of escape from “possibly adapted” was higher (48%) than our empirical testing identifying variants, including binding affinity variants, we may have missed. We have revised the Discussion to emphasize this point.

6) The authors study four patients who initiated ART during acute HIV infection – the rest began therapy in chronic infection. One might predict that the length of the period of infection before initiating ART would influence the likelihood of developing escape mutations – do the authors have any data to support this hypothesis?

We agree with the reviewer that length of untreated infection will influence escape prevalence, consistent with previous studies report that early treatment restricts the virus population diversity at an early stage and limits the emergence of new immune-escape variants (reviewed in doi: 10.1097/COH.0000000000000120).

In our cohort, no difference in the likelihood of developing escape mutations was observed between participants treated in acute (n=3 with sufficient OGV data) or chronic HIV-1 infection (Mann-Whitney two tailed test, p=0.675). This is noted in the manuscript, however we do not consider we have sufficient study numbers to further emphasize this point.

References

A. J. Brown, D. D. Richman, HIV-1: gambling on the evolution of drug resistance? Nature medicine 3, 268-271 (1997).

A. J. Leslie et al., HIV evolution: CTL escape mutation and reversion after transmission. Nature medicine 10, 282-289 (2004).

S. J. Kent, C. S. Fernandez, C. J. Dale, M. P. Davenport, Reversion of immune escape HIV variants upon transmission: insights into effective viral immunity. Trends in microbiology 13, 243-246 (2005).

H. Crawford et al., Compensatory mutation partially restores fitness and delays reversion of escape mutation within the immunodominant HLA-B*5703-restricted Gag epitope in chronic human immunodeficiency virus type 1 infection. Journal of virology 81, 8346-8351 (2007).

K. Gounder et al., High frequency of transmitted HIV-1 Gag HLA class I-driven immune escape variants but minimal immune selection over the first year of clade C infection. PloS one 10, e0119886 (2015).

C. F. Thobakgale et al., Impact of HLA in mother and child on disease progression of pediatric human immunodeficiency virus type 1 infection. Journal of virology 83, 10234-10244 (2009).

T. M. Allen et al., Selective escape from CD8+ T-cell responses represents a major driving force of human immunodeficiency virus type 1 (HIV-1) sequence diversity and reveals constraints on HIV-1 evolution. Journal of virology 79, 13239-13249 (2005).

V. Novitsky et al., Timing constraints of in vivo gag mutations during primary HIV-1 subtype C infection. PloS one 4, e7727 (2009).

B. Ondondo et al., Novel Conserved-region T-cell Mosaic Vaccine With High Global HIV-1 Coverage Is Recognized by Protective Responses in Untreated Infection. Molecular therapy : the journal of the American Society of Gene Therapy 24, 832-842 (2016).

A. T. Jones et al., HIV-1 vaccination by needle-free oral injection induces strong mucosal immunity and protects against SHIV challenge. Nature communications 10, 798 (2019).

M. M. Addo et al., Comprehensive epitope analysis of human immunodeficiency virus type 1 (HIV-1)-specific T-cell responses directed against the entire expressed HIV-1 genome demonstrate broadly directed responses, but no correlation to viral load. Journal of virology 77, 2081-2092 (2003).

P. Kiepiela et al., CD8+ T-cell responses to different HIV proteins have discordant associations with viral load. Nature medicine 13, 46-53 (2007).

P. Kiepiela et al., Dominant influence of HLA-B in mediating the potential co-evolution of HIV and HLA. Nature 432, 769-775 (2004).

B. Mothe et al., Definition of the viral targets of protective HIV-1-specific T cell responses. Journal of translational medicine 9, 208 (2011).

A. Guimaraes-Walker et al., Lessons from IAVI-006, a phase I clinical trial to evaluate the safety and immunogenicity of the pTHr.HIVA DNA and MVA.HIVA vaccines in a prime-boost strategy to induce HIV-1 specific T-cell responses in healthy volunteers. Vaccine 26, 6671-6677 (2008).

N. Goonetilleke et al., Induction of multifunctional human immunodeficiency virus type 1 (HIV-1)-specific T cells capable of proliferation in healthy subjects by using a prime-boost regimen of DNA- and modified vaccinia virus Ankara-vectored vaccines expressing HIV-1 Gag coupled to CD8+ T-cell epitopes. Journal of virology 80, 4717-4728 (2006).

A. Masemola et al., Hierarchical targeting of subtype C human immunodeficiency virus type 1 proteins by CD8+ T cells: correlation with viral load. Journal of virology 78, 3233-3243 (2004).

V. V. Ganusov et al., Fitness costs and diversity of the cytotoxic T lymphocyte (CTL) response determine the rate of CTL escape during acute and chronic phases of HIV infection. Journal of virology 85, 10518-10528 (2011).

N. Goonetilleke et al., The first T cell response to transmitted/founder virus contributes to the control of acute viremia in HIV-1 infection. The Journal of experimental medicine 206, 1253-1272 (2009).

M. K. Liu et al., Vertical T cell immunodominance and epitope entropy determine HIV-1 escape. The Journal of clinical investigation 123, 380-393 (2013).

M. Altfeld et al., Enhanced detection of human immunodeficiency virus type 1-specific T-cell responses to highly variable regions by using peptides based on autologous virus sequences. Journal of virology 77, 7330-7340 (2003).

https://doi.org/10.7554/eLife.57246.sa2

Article and author information

Author details

  1. Joanna A Warren

    Department of Microbiology and Immunology, University of North Carolina, Chapel Hill, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0595-0390
  2. Shuntai Zhou

    1. Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, United States
    2. UNC Center For AIDS Research, University of North Carolina, Chapel Hill, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing - original draft
    Competing interests
    No competing interests declared
  3. Yinyan Xu

    Department of Microbiology and Immunology, University of North Carolina, Chapel Hill, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  4. Matthew J Moeser

    1. Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, United States
    2. UNC Center For AIDS Research, University of North Carolina, Chapel Hill, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1368-5171
  5. Daniel R MacMillan

    British Columbia Centre for Excellence in HIV/AIDS, Vancouver, Canada
    Contribution
    Formal analysis
    Competing interests
    No competing interests declared
  6. Olivia Council

    1. Department of Microbiology and Immunology, University of North Carolina, Chapel Hill, United States
    2. Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, United States
    Contribution
    Data curation
    Competing interests
    No competing interests declared
  7. Jennifer Kirchherr

    Department of Medicine, University of North Carolina, Chapel Hill, United States
    Contribution
    Data curation, Investigation
    Competing interests
    No competing interests declared
  8. Julia M Sung

    1. Department of Medicine, University of North Carolina, Chapel Hill, United States
    2. UNC HIV Cure Center, University of North Carolina, Chapel Hill, United States
    Contribution
    Resources, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  9. Nadia R Roan

    1. Department of Urology, University of California San Francisco, San Francisco, United States
    2. Gladstone Institute of Virology and Immunology, San Francisco, United States
    Contribution
    Resources, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5464-1976
  10. Adaora A Adimora

    Department of Medicine, University of North Carolina, Chapel Hill, United States
    Contribution
    Resources, Data curation
    Competing interests
    No competing interests declared
  11. Sarah Joseph

    1. Department of Microbiology and Immunology, University of North Carolina, Chapel Hill, United States
    2. UNC HIV Cure Center, University of North Carolina, Chapel Hill, United States
    Contribution
    Data curation
    Competing interests
    No competing interests declared
  12. JoAnn D Kuruc

    1. Department of Medicine, University of North Carolina, Chapel Hill, United States
    2. UNC HIV Cure Center, University of North Carolina, Chapel Hill, United States
    Contribution
    Resources, Data curation, Funding acquisition
    Competing interests
    No competing interests declared
  13. Cynthia L Gay

    1. Department of Medicine, University of North Carolina, Chapel Hill, United States
    2. UNC HIV Cure Center, University of North Carolina, Chapel Hill, United States
    Contribution
    Resources, Data curation
    Competing interests
    No competing interests declared
  14. David M Margolis

    1. Department of Microbiology and Immunology, University of North Carolina, Chapel Hill, United States
    2. UNC Center For AIDS Research, University of North Carolina, Chapel Hill, United States
    3. Department of Medicine, University of North Carolina, Chapel Hill, United States
    4. UNC HIV Cure Center, University of North Carolina, Chapel Hill, United States
    Contribution
    Resources, Formal analysis, Supervision, Funding acquisition, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  15. Nancie Archin

    1. Department of Medicine, University of North Carolina, Chapel Hill, United States
    2. UNC HIV Cure Center, University of North Carolina, Chapel Hill, United States
    Contribution
    Resources, Data curation, Supervision, Writing - review and editing
    Competing interests
    No competing interests declared
  16. Zabrina L Brumme

    1. British Columbia Centre for Excellence in HIV/AIDS, Vancouver, Canada
    2. Faculty of Health Sciences, Simon Fraser University, Burnaby, Canada
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8157-1037
  17. Ronald Swanstrom

    1. Department of Microbiology and Immunology, University of North Carolina, Chapel Hill, United States
    2. Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, United States
    3. UNC Center For AIDS Research, University of North Carolina, Chapel Hill, United States
    4. UNC HIV Cure Center, University of North Carolina, Chapel Hill, United States
    Contribution
    Data curation, Supervision, Writing - review and editing
    Competing interests
    UNC is pursuing IP protection for Primer ID, and Ronald Swanstrom is listed as a coinventor and has received nominal royalties.
  18. Nilu Goonetilleke

    1. Department of Microbiology and Immunology, University of North Carolina, Chapel Hill, United States
    2. Department of Medicine, University of North Carolina, Chapel Hill, United States
    3. UNC HIV Cure Center, University of North Carolina, Chapel Hill, United States
    Contribution
    Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Writing - original draft, Writing - review and editing
    For correspondence
    nilu_goonetilleke@med.unc.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2278-1656

Funding

National Institute of Allergy and Infectious Diseases (P30 AI50410)

  • Ronald Swanstrom

National Institute of Allergy and Infectious Diseases (U01 AI103390)

  • Adaora A Adimora

National Institute of Allergy and Infectious Diseases (U01 AI034989)

  • Nadia Roan

National Institute of Allergy and Infectious Diseases (U01 AI131310)

  • Joanna A Warren
  • Shuntai Zhou
  • Yinyan Xu
  • Nilu Goonetilleke

National Institute of Allergy and Infectious Diseases (R01 AI140970)

  • Ronald Swanstrom

National Institute of Allergy and Infectious Diseases (UM1AI126617)

  • Daniel R MacMillan
  • Zabrina L Brumme

National Institute of Allergy and Infectious Diseases (1UM1AI126619)

  • Joanna A Warren
  • Shuntai Zhou
  • Julia M Sung
  • JoAnn D Kuruc
  • Cynthia L Gay
  • David M Margolis
  • Nancie Archin
  • Nilu Goonetilleke

Canadian Institutes of Health Research (PJT-148612)

  • Daniel R MacMillan
  • Zabrina L Brumme

Canadian Institutes of Health Research (PJT-159625)

  • Daniel R MacMillan
  • Zabrina L Brumme

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank all the volunteers who participated in this study and all colleagues in respective labs who facilitated this study, including Catalina Ramirez, Erin Stuelke, Katherine Sholtis James, Brigitte Allard, Caroline Baker, John Schmitz and team, and the UCSF clinical team. We thank Genevieve Clutton for critical review of the manuscript. This study was funded by the University of North Carolina (UNC) Center for AIDS Research (P30 AI50410), the UNC Women’s Interagency HIV Study (WIHS) (U01 AI103390), the University of California San Francisco WIHS (U01 AI034989), the National Institute of Allergy and Infectious Disease (U01 AI131310, R01 AI140970), with major contribution from CARE, a Martin Delaney Collaboratory of the National Institute of Allergy and Infectious Diseases, National Institute of Neurological Disorders And Stroke, National Institute On Drug Abuse and the National Institute Of Mental Health of the National Institutes of Health, (1UM1AI126619). Support was also provided by the BELIEVE Martin Delaney Collaboratory (UM1AI126617).

Ethics

Human subjects: Participants were enrolled with the following IRB approved studies: (1) CID 0819 - Apheresis Procedures to Obtain Leukocytes for Research Studies from HIV Positive Participants (08-1575), (2) The UNC Women's Interagency HIV Study (WIHS) (12-1660). Review and implementation of all protocols utilized for the collection of samples for this this analysis were approved by the University of North Carolina at Chapel Hill Biomedical Institutional Review Board (IRB) and the University of California at San Francisco IRB. All participants provided written informed consent. All experimental protocols were approved by local Institutional Biomedical Review Boards (ethics numbers: 14-0741, 11-0228, and 13-3613, 15-1626) and performed in accordance with the relevant guidelines.

Senior Editor

  1. Tadatsugu Taniguchi, Institute of Industrial Science, The University of Tokyo, Japan

Reviewing Editor

  1. Julie Overbaugh, Fred Hutchinson Cancer Research Center, United States

Reviewers

  1. Steven Deeks, University of California, San Francisco, United States
  2. Sarah Rowland-Jones

Publication history

  1. Received: March 26, 2020
  2. Accepted: September 24, 2020
  3. Accepted Manuscript published: October 6, 2020 (version 1)
  4. Version of Record published: October 28, 2020 (version 2)

Copyright

© 2020, Warren et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,704
    Page views
  • 248
    Downloads
  • 4
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Immunology and Inflammation
    Hannah M Isles et al.
    Research Article

    Neutrophils are rapidly recruited to inflammatory sites where their coordinated migration forms clusters, a process termed neutrophil swarming. The factors that modulate early stages of neutrophil swarming are not fully understood, requiring the development of new in vivo models. Using transgenic zebrafish larvae to study endogenous neutrophil migration in a tissue damage model, we demonstrate that neutrophil swarming is a conserved process in zebrafish immunity, sharing essential features with mammalian systems. We show that neutrophil swarms initially develop around an individual pioneer neutrophil. We observed the violent release of extracellular cytoplasmic and nuclear fragments by the pioneer and early swarming neutrophils. By combining in vitro and in vivo approaches to study essential components of neutrophil extracellular traps (NETs), we provide in-depth characterisation and high-resolution imaging of the composition and morphology of these release events. Using a photoconversion approach to track neutrophils within developing swarms, we identify that the fate of swarm-initiating pioneer neutrophils involves extracellular chromatin release and that the key NET components gasdermin, neutrophil elastase, and myeloperoxidase are required for the swarming process. Together our findings demonstrate that release of cellular components by pioneer neutrophils is an initial step in neutrophil swarming at sites of tissue injury.

    1. Computational and Systems Biology
    2. Immunology and Inflammation
    Margaret A Myers et al.
    Research Article

    Influenza viruses cause a significant amount of morbidity and mortality. Understanding host immune control efficacy and how different factors influence lung injury and disease severity are critical. We established and validated dynamical connections between viral loads, infected cells, CD8+ T cells, lung injury, inflammation, and disease severity using an integrative mathematical model-experiment exchange. Our results showed that the dynamics of inflammation and virus-inflicted lung injury are distinct and nonlinearly related to disease severity, and that these two pathologic measurements can be independently predicted using the model-derived infected cell dynamics. Our findings further indicated that the relative CD8+ T cell dynamics paralleled the percent of the lung that had resolved with the rate of CD8+ T cell-mediated clearance rapidly accelerating by over 48,000 times in 2 days. This complimented our analyses showing a negative correlation between the efficacy of innate and adaptive immune-mediated infected cell clearance, and that infection duration was driven by CD8+ T cell magnitude rather than efficacy and could be significantly prolonged if the ratio of CD8+ T cells to infected cells was sufficiently low. These links between important pathogen kinetics and host pathology enhance our ability to forecast disease progression, potential complications, and therapeutic efficacy.