HIV integration provides a survival advantage to in vitro infected primary human CD4 + T cells

Since proviral insertional mutagenesis of STAT3 has been associated with clonal expansion, we first wanted to see if HIV integration is associated with T cell persistence over a longer period in culture. Frozen aliquots of primary human CD4 + CD45RA + T cells from three different HIV-negative donors (Donors 3, 4, and 5) were obtained from STEMCELL Technologies Inc. and thawed in RPMI medium containing IL-2. No live participants were included in this study. After 4 days of stimulation with anti-CD3/CD28 beads, each donor sample was split into replicates, with Donor 3 and 5 being split into three replicates each and Donor 4 into six replicates. The activated T cells were transduced with a replication incompetent, VSV G-protein pseudotyped HIV vector, with a gfp reporter located in place of nef [47], at an MOI of ~1 gfp transducing-unit per cell (TU/cell). T cells from each donor were cultured for extended periods until there were no longer enough viable cells for collection (Fig 1A). IL-2 was added every three days and anti-CD3/CD28 beads added just after the cultures reached maximum expansion, as their number began to decline. Cultures were split as necessary and a portion at each split was cryopreserved for future analysis. Throughout the time course, cells from the different donors and replicates were assessed by flow cytometry for Gfp expression as a reporter for HIV integration and expression.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Proviral integration associated with HIV+ cell survival.

A) Experimental design. Primary human CD4 + CD45RA + T cells from three different donors were infected with NL43ΔEnvΔNef-GFP pseudotyped with VSV-G at an MOI of 1 GFP TU/cell. Cells were stimulated with anti-CD3/CD28 beads as they began to die off after each expansion period. We acknowledge that BioRender was used for depiction of an HIV virion. Created in BioRender. Murimi-worstell, D. (2025) https://BioRender.com/wgmt54w. B) GFP expression of each replicate from the 3 donors over time assessed by flow cytometry. C) Bulk HIV provirus content was assessed by LTR-specific qPCR at the time points shown using ACH-2 cells as a single-copy HIV standard and CCR5 as a cellular DNA control. The provirus presence was assessed until day 136.

https://doi.org/10.1371/journal.ppat.1013087.g001

Over time following infection, the percentage of Gfp-positive (GFP+) cells in each culture declined, an observation consistent with prior work by others [47] but differing from our previous experience using similar culturing methods [43]. This decline supports the idea that production of viral proteins or Gfp is harmful to infected T cells [48,49] or that the HIV provirus was being silenced over time [50]. In three of the six replicates from Donor 4, a resurgence in Gfp expression was detected starting around day 80, and two of those replicates surpassed 50% of the cells expressing Gfp (and therefore were HIV+) on day 126, indicating a growth advantage in the infected relative to uninfected cells (Fig 1B).

To test whether GFP+ cell reduction was due to proviral silencing, qPCR was performed to measure the number of HIV proviruses per cell. ACH2 cells were used as a standard as this cell line has been described to have close to 1 provirus per cell despite having a low level of ongoing viral replication which leads to amplification in highly passaged cultures [51–53]. A standard curve of one HIV vector plasmid per one genome of DNA was used to determine the number of HIV proviruses in our ACH2 cell DNA. Primers were targeted to the HIV LTR to detect proviral DNA and amplification was normalized to CCR5, a host gene with two copies per diploid genome in primary cells [54]. We independently determined that our ACH2 cell DNA contained about 2 proviruses per cell (S1 Fig).

Our analysis showed a reduction in HIV provirus content over time in all Donor 4 replicates, reflecting a loss of infected cells rather than of provirus silencing (Fig 1C). Additionally, the subsequent increases in Gfp expression in replicates 2, 3 and 4 were accompanied by increases in provirus per cell, reflecting preferential growth or survival of an initially small fraction of infected cells (Fig 1C).

These data suggest a growth or survival advantage of an initially small fraction of the HIV infected and Gfp expressing cells, as the HIV/GFP+ cells made up increasingly large proportions of the cells that survived until day 126. To distinguish the three Donor 4 replicates with resurgent HIV/GFP+ populations, they will be referred to as “GFP-High” and the other three replicates without a visible resurgent population will be referred to as “GFP-Low.”

Resurgent HIV+ cells contain clonal proviruses within intron 1 of STAT3

As our laboratory has previously described [43], HIV integration within STAT3 can be reproducibly associated with clonal expansion of primary human CD4 + T cells infected and grown in vitro. We therefore hypothesized that the HIV/GFP+ cells in the GFP-High replicates would also contain proviruses integrated in STAT3. To test this hypothesis, DNA was extracted from the Donor 4 day 126 samples and ISA was performed using a modification of the protocol of Wells et. al. [44].

The day 126 DNA samples were sheared using an enzymatic reaction to create random, unique, double strand breaks in the genomic DNA, as previously described [55]. Following a method like that previously described [44,55], a linker was ligated to the DNA fragments, which were amplified and barcoded for Illumina sequencing. Sequence data were analyzed using an analytical pipeline that trims each read of linker and proviral sequences leaving only the host sequence, which was then aligned to the hg19 reference. Random shearing of the host sequences leaves each with a unique terminal breakpoint, which will almost always be different for each starting DNA molecule. Sequences with identical integration sites and different breakpoints must have come from different descendants of a single infected cell, which subsequently underwent clonal expansion. The number of different breakpoints for a given integration site is a measure of the relative size of each such clone.

ISA of all the Donor 4 replicates at day 126 identified integration sites throughout the genome (Fig 2A). All replicates that exhibited clonal GFP+ expansion contained proviral Integrations within STAT3. In individual replicates, proviruses within CKAP2L, DCUN1D4, GSDMD, and CCDC77 were also associated with clonal expansion at day 126.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Integration in STAT3 is associated with GFP-High replicates.

A) Integration site analysis was performed on all six Day 126 Donor 4 replicates and is shown as the number of different breakpoints for each site (Y axis) vs the chromosomal location of the integration site within the whole human genome (X-axis). Integration sites with large numbers of breakpoints that contained additional sites in immediate proximity to one another were combined for a true representation of clonality [56] and are indicated with the relative orientation of the provirus at their chromosomal location. B) Relative frequency of integration sites in the indicated genes in the different replicates at Day 17 and Day 126. Left Y-axis measures percentage of GFP positive cells. Green represents GFP+ cells and light grey represents GFP- cells. The right Y-axis measures percentage of total breakpoints. Percentage of integrations within intergenic regions are in dark grey. Integrations within genes are colored light blue, and integrations within genes associated with clonal expansion are indicated by the color shown in the key for each individual gene. C) Location, orientation, and frequency of proviral integration sites within STAT3 of the replicates. D) Top: structure of STAT3 and location of primers (red arrows) used to confirm the integration sites shown by the black arrows within intron 1. Bottom: gel analysis of products of PCR using the STAT3 and LTR primers to confirm integration sites identified in replicates 2, 3, and 4 by ISA. Un, uninfected T cells; NTC, no template control.

https://doi.org/10.1371/journal.ppat.1013087.g002

Proviral integration patterns and GFP percentages were compared to day 17. As shown in Fig 1, there was initial decrease in GFP+ cells over time indicating selection against HIV/GFP+ cells under our culturing condition, until a resurgence in GFP-High replicates. We also observed a reduction in proviral integrations within genic regions over time, except in the GFP-High replicates where integration within specific genes made up a large portion of integration sites (clonal2B). Excluding the GFP-High replicates, we saw selection for integration within genic regions fall from about 60% to closer to 20%, corresponding to the reduction in GFP-expressing cells, which declined from around 50% to 1% (Fig 2B). Increased integrations in genic regions at initial infection followed by selection against those genic integrations over time supports findings from prior studies [38]. In the GFP-High replicates, the increase in GFP-percentage corresponded to increases in proviral integration within specific genes (Fig 2B), providing evidence that HIV integration within specific genes was associated with a survival advantage.

The GFP-High populations in replicates 2 and 4 contained clonal proviruses within STAT3, which comprised 24.3% and 8.9% of the total breakpoints, respectively (Fig 2B). The proviruses within STAT3 were all in the same orientation of transcription as the gene. Each of these two replicates also contained additional clonal integration sites (Table 1 and Fig 2A and 2B). Replicate 2 contained a clonally amplified provirus within CKAP2L, but in the opposite transcriptional orientation. Replicate 4 contained two additional genes with clonally amplified integration sites, DCUN1D4 and GSDMD, which made up 23.2% and 26.3% of total integration sites, respectively (Table 1 and Fig 2B). The amplified provirus within DCUN1D4 was in the same orientation as translation but the one in GSDMD was not (Table 1). However, the GSDMD provirus was in the same orientation as and just upstream of an annotated anti-sense lncRNA of unknown function, LOC10030756. Cells containing these additional proviruses were amplified to a similar extent as one in STAT3 and could also be contributing to the resurgence of HIV+ cells, although their possible joint contribution has not been determined.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Clonal proviral integration sites in donor 4 replicates at day 126 after infection.

https://doi.org/10.1371/journal.ppat.1013087.t001

Additional clones of cells with proviral integration sites within STAT3 were also identified in replicates 1 and 3, both of which contained fewer HIV+ cells at day 126 than replicates 2 and 4. Both replicates 1 and 3 contained proviruses within STAT3, which made up just under 2% of total integrations in each (Table 1 and Fig 2B). Additionally, however, replicate 3 contained 18.4% of total integrations within CCDC77, suggesting that clonal expansion of cells with a provirus at this site may also have played a role in the 13% GFP+ cells (Table 1 and Fig 2B).

STAT3 integration sites identified in the four replicates were in the same orientation of transcription as the STAT3 host and most were in intron 1 just before the start of exon 2, the first coding exon (Fig 2C). HIV provirus integration within this region of STAT3 has previously been described in cases of HIV-associated T cell and B cell lymphomas [16,17]. Replicate 3 contained integration sites within introns 1 and 3 (Table 1 and Fig 2C), the latter being the location of some of the previously described integrations influencing clonal expansion of T cells grown in vitro and potentially leading to a shortened protein product via N-terminal truncation [43].

PCR was performed on all replicates to confirm the presence of HIV proviral integration sites identified by ISA. Primers complementary to different locations within intron 1 of STAT3 and the HIV 3’LTR identified integration sites within the GFP-High replicates 2, 3 and 4 (Fig 2D). Together with the ISA data, proviruses integrated in intron 1 of STAT3 were associated with resurgent GFP/HIV+ populations within these three Donor 4 replicates. Proviral integration within STAT3 was not detected in replicate 1, most likely because there were only 3.5% HIV/GFP+ cells at time of DNA collection. Detection of proviruses in STAT3 associating with clonal expansion implies that HIV integration within the first intron of STAT3 can contribute to the survival, outgrowth, and persistence of HIV+ cells in the culturing conditions shown in Fig 1A.

Enrichment of STAT3 expression in Donor 4 replicates with increased HIV/GFP expression

To understand the influence of proviral integration on expression of STAT3 and its downstream target genes, bulk RNAseq was performed at day 126 on all GFP-High and GFP-Low replicates from Donor 4. The GFP-High replicates, which had greater than 10% GFP+ cells on day 126, expressed higher levels of STAT3 than the GFP-Low replicates, with less than 10% of cells expressing gfp (Fig 3A). Replicate 2 showed the highest expression of STAT3, and the highest percentage of integration sites within STAT3, but did not have the highest percentage of GFP+ cells. This result hints that integration within genes other than STAT3, or possibly other epigenetic effects, may also contribute to HIV/GFP + T cell survival and persistence.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Upregulated STAT3 signature in GFP-High replicates.

A. Bulk RNA-seq of unsorted donor 4 replicates was performed at day 126 post-transduction. Replicates 2,3, and 4 were designated as “GFP-High” and Reps 1,5, and 6 as “GFP-Low.” A) STAT3 expression in GFP-high (blue) and GFP-low (red) Donor 4 replicates in transcripts per million (TPM). B) Coverage of transcripts across STAT3 in each replicate of Donor 4. C) Principal component analysis using DESeq2 (56) of GFP-High replicates compared to GFP-Low replicates. D) Volcano plot of GFP-High gene expression compared to GFP-Low gene expression. Genes that are differentially expressed are labeled in red. Parameters for differential expression are log2 fold change > 0.5 and adjusted p-value < 0.05. E) IL6-JAK-STAT3 Signaling hallmark gene set, from the Molecular Signatures Database (MSigDB), analyzed by gene set enrichment analysis [57] normalized enrichment score (NES) = 1.55, FDR q-value = 0.02. F) Heat map of gene rank of the leading-edge genes in the Hallmark IL6/JAK/STAT3 signaling gene set.

https://doi.org/10.1371/journal.ppat.1013087.g003

Previous studies looking at HIV-1 proviral integrations sites in vivo, have identified that the HIV LTR can act as a promoter and drive the expression of host genes [37,45,46]. To determine if STAT3 overexpression was driven by the HIV LTR, we compared the number of reads from exons 2–24, downstream of the integration site, to those from exon 1, upstream of the integration site (Fig 3B). The GFP-High replicates showed an increase in the ratio of exon 2 reads to exon 1 reads compared to the GFP-Low replicates (S2 Fig). This increase provides evidence that STAT3 overexpression was driven by the HIV LTR. Taken together, these results imply that LTR driven chimeric transcripts of STAT3 were driving STAT3 overexpression in the GFP-High replicates, thus providing additional evidence that insertional mutagenesis of STAT3 led to the growth advantage of the HIV/GFP+ cells.

To explore the effects of STAT3 expression in resurgent GFP+ cells, differential gene expression analysis was performed using DESeq2 [58]. For this analysis, the Donor 4 replicates were divided into GFP-High (Reps 2, 3, and 4) and GFP-Low (Reps 1, 5, and 6) groups, and data for each group were pooled for analysis of differentially expressed genes (Fig 3C). Overall, there were only 213 genes differentially expressed (adjusted p < 0.05) between the two groups (Fig 3D). Notably, STAT3 was one of the most differentially expressed upregulated genes (Fig 3D). Additionally, downstream targets of STAT3, such as SOCS3 [59,60] and SGK1 [61], were also upregulated (Fig 3D). All three of these STAT3-related genes are associated with anti-apoptotic pathways [62–65].

GZMB, encoding granzyme B, a serine protease associated with cytotoxic T cells, was identified as one of the most highly differentially expressed genes in replicates 2 and 3 (S3 Fig), in addition to GZMH and NKG7, two other markers of cytotoxic T cells (Fig 3D). Previous studies have identified STAT3 as a transcription factor for GZMB [66–68], suggesting that insertional mutagenesis of STAT3 in our system led to an increase in its expression. The upregulation of these markers of cytotoxic T cells is of particular interest, since their expression has been identified as being associated with persistent HIV-infected CD4 + cells from PWH, which were enriched in expression of a Th1 cytotoxic cell phenotype [41,42].

To further compare GFP-High replicates to the GFP-Low replicates, gene set enrichment analysis (GSEA) [57] was performed. GSEA looks for enrichment in predetermined sets of genes between the groups being compared. Utilizing the 50-hallmark gene sets from the molecular signatures database (MSigDB), GFP-High replicates were significantly enriched in the hallmark gene set IL-6/JAK/STAT3 signaling (Fig 3E and 3F). Upregulation of STAT3 and STAT3-related genes and enrichment of the STAT3 signaling pathway (NES 1.55, q value 0.02) provides evidence that the LTR-driven STAT3 plays an active role in shaping cellular gene expression leading to preferential growth and/or survival of the affected cells from the Donor 4 GFP-High replicates.

GFP-High replicate 2 has largest increase in protein expression and phosphorylation of STAT3

To confirm that upregulation of the STAT3 transcription signature corresponds to upregulation at the protein level, cells frozen at earlier time points were thawed and cultured until day 126 and analyzed by flow cytometry for Gfp and STAT3 expression and activation. GFP-High replicates contained distinct populations of GFP+ cells with higher expression of STAT3 compared to the GFP- cells, whereas the GFP-Low sorted replicates did not contain these overexpressing STAT3 populations (Fig 4A). Replicate 2 showed the largest populations of STAT3 overexpressing cells with Replicates 3 and 4 having smaller populations. The increased protein expression reflects the increase in STAT3 transcription between the GFP-High replicates (Figs 3A and S2). Moreover, the staining for STAT3 identifies the number of cells overexpressing STAT3, by increased fluorescence intensity, which quantifies the number of HIV/GFP+ cells where insertional mutagenesis of STAT3 is driving survival. Thus, replicate 2 had the largest percentage of integration within STAT3 (Table 1 and Fig 2A and 2B), the highest expression of STAT3 transcripts (Fig 3A and 3B), and the largest population of overexpressing STAT3 protein (Fig 4A and 4B). Likewise, insertional mutagenesis of STAT3 played a smaller role in the resurgent HIV/GFP+ cells in Replicate 3 and 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Increased STAT3 expression and activation in GFP-High Replicates.

A2) Donor 4 replicates on day 126 were stained for STAT3 expression and cell viability (using Zombie Violet^Tm). The day 126 cells were analyzed by flow cytometry for cell viability and viable cells were then sorted for Gfp expression. STAT3 expression was then assessed in the sorted GFP+ and GFP- cells. B) Median fluorescence Intensity (MFI) was compared between GFP+ and GFP- cells from respective replicates. C) Day 126 Donor 4 GFP-High replicate cells were stained for pSTAT3 (Y705) expression and cell viability. The day 126 cells were gated on cell viability then analyzed by flow cytometry. Viable cells were then gated for Gfp expression and pSTAT3 expression was assessed in the GFP+ and GFP- cells. D) MFI compared between GFP+ and GFP- cells from respective replicates.

https://doi.org/10.1371/journal.ppat.1013087.g004

To see if the increased STAT3 was active in the GFP-High replicates, they were stained with antibody specific for phosphorylation at tyrosine 705 (Y705) of STAT3 (pSTAT3), a modification that allows STAT3 to dimerize, enter the nucleus, and carry out its role as a transcription factor. The replicates were analyzed for pSTAT3 (Y705) by flow cytometry and showed an increase in the percentage of pSTAT3-positive populations in the GFP+ cells compared to the GFP- cells (Fig 4C). However, only the GFP+ cells in replicates 2 and 3 had increased pSTAT3 expression, measured by fluorescence intensity, compared to the GFP- cells (Fig 4D), while a difference in pSTAT3 expression between GFP+ and GFP- cells was not apparent in replicate 4. Replicate 4 had a smaller percentage of HIV/GFP+ cells over expressing STAT3 and there was no difference between the GFP+ and GFP- cells, indicating some difference, perhaps integration within a different gene, in the mechanism leading to clonal expansion and survival in this replicate.

The flow cytometry data identified STAT3 overexpressing populations within the HIV/GFP+ fraction of the three GFP-High replicates. Further interrogation identified increases in phosphorylation of STAT3 within the GFP+ cells of replicates 2 and 3. Increased abundance of STAT3 might allow for the increase in activated STAT3 and therefore drive upregulation of genes that use STAT3 as a transcription factor. We conclude that overexpression and activation of STAT3 in the GFP-High replicates led to enrichment in STAT3 signaling and upregulation of downstream targets of STAT3, thereby contributing to the resurgence and survival of the HIV/GFP+ cells in the GFP-High replicates.

STAT3 signaling is associated with persistence of replicate 2 and 3 HIV/GFP+ cells

To further investigate the increase of GFP+ cells in these populations at later time points, bulk RNAseq was performed on sorted GFP+ and GFP- cells from each Donor 4 GFP-High replicate and differential gene expression analysis was performed using DEseq2. Pooled GFP+ data were compared to the GFP- cells sorted from the GFP-High replicates (Fig 5A) resulting in 273 differentially expressed genes (adjusted p < 0.05) (Fig 5B). Of these differentially expressed genes, again, STAT3 overexpression due to HIV proviral insertional mutagenesis was a driving factor in the HIV/GFP+ cell survival advantage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. STAT3 signaling plays the largest role in Replicate 2 and 3 HIV/GFP+ cell survival.

Donor 4 GFP-High replicates on Day 126 were sorted for GFP expression and viability and the two fractions were subjected to RNAseq. A) Principal component analysis using DESeq2 of sorted GFP-High cells. B) Volcano plot of sorted GFP-High replicates 2, 3, and 4. Data from GFP+ cells were pooled and compared to that from pooled XGFP- cells. Pooled data from the 3 replicates are shown. C) Expression of apoptosis-associated genes compared between GFP+ and GFP- cells from GFP-High replicates. D) Comparison of expression of cytotoxic genes. E) Comparison of expression of some housekeeping genes.

https://doi.org/10.1371/journal.ppat.1013087.g005

To better understand what the differentially expressed genes reveal about the GFP+ cells, Gene Ontology (S4A Fig) and Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway (S4B Fig) analyses were performed. Both analyses identified upregulation of genes associated with DNA replication, cell division and cell cycle. These results suggest that the HIV/GFP+ persistent cells were undergoing a higher level of cell division than the GFP- negative cells. This conclusion was further supported by GSEA, which also showed significant enrichment in cell cycle associated gene sets, E2F Targets and G2M Checkpoint (S4C Fig), consistent with greater HIV/GFP+ cell proliferation and expansion.

Despite STAT3 being a differentially expressed gene, its previously identified downstream targets were not identified as such in the pooled data from the three GFP+ cells from the GFP-High replicates. Reduced STAT3 overexpression and activation in replicate 4 GFP+ cells compared to the other GFP-High replicate GFP+ cells (Fig 4A and 4C), might explain why the downstream targets of STAT3 were not identified as genes differentially expressed between GFP+ and GFP- cells. Examination of the differentially expressed genes of interest identified in Fig 3D shows that that targets of STAT3 were upregulated in replicates 2 and 3 GFP+ cells but to a lesser extent or not at all in replicate 4. SOCS3 and SGK1 were both increased in expression only in the two replicates with increased STAT3 expression and activation (Fig 5C). Further suggesting that increases in these anti-apoptotic genes were contributing to T cell persistence. Additionally, STAT3-related cytotoxic factors, GZMB and GZMB-related genes, were also upregulated in GFP+ cells from replicates 2 and 3 but not replicate 4 (Fig 5D). GZMB was upregulated in GFP+ cells when analyzed by flow cytometry (S3 Fig). All three of the GFP-High GFP+ replicates contained a higher percentage of GZMB+ cells and a higher mean fluorescence intensity than their matching GFP- cells. GFP+ cells in replicate 2 showed the largest increase in GZMB compared to GFP- cells (S3 Fig). Control genes showed no difference (Fig 5E).

Identification of STAT3 as a shared differentially expressed gene further supports the role of STAT3 in cell survival and/or proliferation. STAT3 activity is most commonly associated with IL-6 or IL-10 signaling [69] but also plays roles in other pathways such as response to TCR engagement [70]. The primary CD4 + T cell cultures were stimulated by TCR signaling through anti-CD3/CD28 beads, as conclusion supported by enrichment in cell cycle and proliferation in gene ontology, pathway, and gene set enrichment analyses (S4 Fig). However, only replicates 2 and 3, which had increased STAT3 activation, showed a clear activated STAT3 expression signature (Fig 4C and 4D). Taken together, the data suggest that upregulation of activated STAT3 can influence anti-apoptotic gene signatures, which might contribute to primary T cell persistence in culture.

Considering the role of STAT3 as a transcription factor in T cell development, we utilized RNAseq data to analyze subtypes of the T cells. The patterns of gene expression of markers of T cell subtypes implied that the T cells grown in culture did not seem to be influenced by STAT3 expression (S5A Fig). In general, STAT3 expression is most associated with the Th17 subtype [70]. However, gene signatures more associated with Th2 subtypes were associated with both the GFP/HIV+ and GFP- groups. This T cell subtype signature has been recorded with T cells that are activated in culture with TCR stimulation but do not receive a secondary signal, such as IL6 for Th17 [71]. This result is consistent with the use, in our model, of CD3 and CD28 crosslinking to mimic TCR stimulation without a secondary signal. Therefore, our persistent cells at day 126 expressed a Th2-like signature instead of the Th17 signature associated with STAT3 expression in vivo.

STAT3 has also been implicated in T cell phenotype development from naïve to memory [27,60]. RNA sequencing profiles of the GFP/HIV + GFP-High replicates did not show many stark changes to a memory or naïve phenotype compared to the GFP- cells (S5B Fig). Flow cytometric analysis of our GFP-High replicates was performed by staining for CD45RO, a marker of memory T cells [72,73], as well as CCR7 which is expressed on central memory (T_CM) and absent on effector memory (T_EM) T cells [72,73]. The results revealed that the GFP+ cells in replicates 2 and 3, which had larger clonal populations with STAT3 integrations, had a greater increase in both T_CM and T_EM differentiation than the GFP- cells from the same replicate (S5C Fig). Replicate 4 GFP/HIV+ cells only demonstrated increase in T_EM (S5C Fig). Taken together, STAT3 may be driving differentiation of primary T cells into a memory phenotype, but as these T cells are grown in vitro with no secondary signals further exploration would be needed to support these findings.

Resurgent HIV+ donor 4 cells do not contain unique STAT3 polymorphisms

We saw an increased STAT3 expression in three replicates of donor 4 associated with increased GFP + T cells but we did not see this effect in cells from the other two donors transduced in vitro. Therefore, we asked whether Donor 4 had any specific polymorphisms that may have influenced STAT3 activity. Utilizing RNAseq data from the Donor 4 replicates, STAT3 variants were identified by comparison to the hg38 genome (S1 Table). Of note, there were two T to C polymorphisms in all replicates located in exon 24, which is an untranslated region of STAT3. While more polymorphisms were detected in expressed introns in replicate 2, all the variants were also found in other replicates, therefore there were no polymorphisms that were specific to STAT3 in resurgent replicates influencing cell growth or survival and there were no modifications to the active sites, SH2 domain or DNA binding region, of STAT3. However, polymorphisms in the 3’UTR can alter the function of STAT3 transcripts through effects on post-transcriptional processing. While these changes are not obvious alterations in coding exons they could lead to alterations in processes, such as mRNA stability, of STAT3 that contributed to survival of T cell clones with proviruses in STAT3.

Integration in genes other than STAT3 may contribute to HIV/GFP+ cells survival advantage

Flow cytometry, shown in Fig 4A, identified STAT3 overexpression in only a portion GFP+ cells in replicates 3 and 4. To investigate the potential of proviruses at other integration sites to contribute to HIV/GFP+ survival, day 126 cells were sorted for Gfp expression and cell viability. DNA and RNA were extracted from both the GFP+ and GFP- populations of each GFP-High replicate for analysis by qPCR and bulk RNA sequencing. qPCR was performed to assess the number of proviruses present in both the GFP+ and GFP- cells. Quantification of HIV proviruses allows us to understand if the effects of proviral integration work synergistically with multiple integrations per cell or separately with cells containing single proviruses at different locations. Using the same qPCR approach as Fig 1C, GFP+ cells in replicates 2 and 3 were found to contain around 2 HIV proviruses per cell, and those in replicate 2 contained closer to 3 proviruses per cell (Fig 6A). These results imply that, in replicates 3 and 4, two proviruses were influencing individual cell survival. On the other hand, replicate 2 HIV/GFP+ cells contained an additional provirus per cell and multiple proviruses in STAT3 (Fig 2C), suggesting that a large fraction of these cells contained proviruses integrated at additional sites along with one in STAT3. FACS analysis showed that most replicate 2 cells were overexpressing STAT3, implicating it as the main contributor to HIV/GFP+ cell persistence (Fig 4A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Increased STAT3 expression and activation in GFP-High Replicates.

DNA and RNA were extracted from GFP-High replicates sorted for cell viability and GFP expression on day 126. A) qPCR was performed to estimate provirus copy number per cell using primers and probes for LTR and normalized to host CCR5 sequences. ACH-2 cells were used as standard and day 126 uninfected cells as negative control. B-E) Bulk RNA-seq counts were used to assess transcriptional activation related to the provirus insertion. Sorted GFP+ samples are colored green and GFP- samples are colored in grey. B) Expression of genes in the indicated replicates with same sense proviruses associated with clonal expansion. Gene expression from the RNAseq analysis was compared between GFP+ and GFP- samples. C) Sequence coverage of STAT3 as well the genes DCUN1D4 and CCDC77, which have a provirus integrated in the same transcriptional sense as the host gene in replicates 4 and 3, respectively. The peak heights are all normalized relative to exon 1. Translational start sites are identified on the gene maps and integration sites are shown by black arrows. D) Expression of genes with provirus integration in opposite orientation to a host gene associated with clonal expansion. Integration in replicate 4 is in same orientation as LOC100310756 but between exons 2 and 3 and in opposite orientation within GSDMD. E) Top panels. Sequencing coverage of CKAP2L, which had an opposite sense orientation integration in exon 5 of replicate 2. The translation start site is identified in the map and the integration site is identified with an arrow. Bottom Panels. Sequencing coverage of GSDMD, which had an opposite sense orientation integration in replicate 4. The color of the coverage graphs represents the predominant transcripts that align with the gene. Green represents transcripts that have the same sense alignment with the gene, blue are anti-sense transcripts to the overall gene, and the light purple represents transcripts of both same sense and anti-sense to overlapping genes. LOC100310756 transcripts are colored blue as the transcript is anti-sense to GSDMD.

https://doi.org/10.1371/journal.ppat.1013087.g006

These results implied that the clonally expanded cells with a provirus in STAT3 also had a second provirus in another gene, possibly contributing to their expansion. To understand the role of integration sites associated with clonal expansion in HIV/GFP+ cell persistence, we used bulk RNAseq analysis on the FACS sorted GFP+ and GFP- populations for each GFP-High replicate to investigate transcriptional changes in genes with proviruses showing large numbers of breakpoints. Increased expression of STAT3 was identified in the GFP+ relative to GFP- cell population in all three GFP-High replicates (Fig 6B). Clonal integration in the same orientation as gene transcription as well as increased expression of CCDC77 was observed only in GFP+ replicate 3 cells, while increased integration in and increased expression of DCUN1D4 were seen in in GFP+ replicate 4 cells (Fig 6B). Increases in these transcripts were observed just downstream the location of proviral integration, providing evidence that the increased expression was driven by insertional mutagenesis (Fig 6C). Unfortunately, little is known about the roles of CCDC77 and DCUN1D4 in cell function, making it difficult to infer how HIV insertional activation of expression of these genes might be affecting cell persistence.

In the case of replicate 2, GFP+ cells contained clonal HIV provirus integration in CKAP2L (Fig 2A and 2B and Table 1). However, no difference was seen in CKAP2L expression between replicate 2 and the other GFP-High replicates (Fig 6D). Examination of expression in IGV also revealed significant increases in levels of CKAPL transcripts downstream of the site of HIV integration same orientation as the provirus but, in this case, in opposite orientation to the gene (Fig 6E, blue). Altogether, these data suggest that HIV/GFP+ cells in replicate 2 contain multiple proviruses influencing host RNA expression. Therefore, the effects of multiple proviruses integrated in the HIV/GFP+ cells in this replicate could contribute to their survival through unique mechanisms.

Another expanded GFP+ clone also contained a provirus antisense to the host gene, in this case, in the second intron of GSDMD, encoding gasdermin D, in replicate 4 (Fig 6E). This provirus was also upstream of and in the same sense as LOC100310756, a lncRNA of unknown function (Fig 6E). As with the other proviruses in the clonally expanded GFP+ cells, the GSDMD provirus was associated with greatly increased expression of LOC100310756, but, again, transcription of the overlapping GSDMD was not significantly affected

Several studies have found that early activation of the HIV protease by some non-nucleoside reverse transcriptase inhibitors, including efavirenz (EFV), can induce GSDMD mediated pyroptotic cell death in T cells and macrophages by premature activation of HIV protease leading to cleavage of CARD8 and driving caspase 1 activation and inflammasome production [74–76]. To see if replicate 4 clones, showed resistance to this cell death pathway due to integrations within GSDMD, we cultured early (Day 31) and late (Day 114) time points of replicate 4 cells with DMSO, EFV, or lopinavir (LPV). LPV is an HIV protease inhibitor, which has been shown to prevent this pyroptotic cell death pathway [74,76]. After 72 hrs of culturing cells with respective anti-retrovirals, late time point replicate 4 had little reduction, around 10%, in GFP/HIV+ cells, whereas the early time point cells lost about 50% of GFP/HIV+ cells (S6 Fig). The limited reduction in the late time point replicate 4 HIV+ cells, suggests that proviral integrations in GSDMD and/or integration upstream of LOC100310756, influenced T cell survival. The early time points were before the clonal population dominated the GFP/HIV+ cell populations Fig 2B and 2C), further supporting that the proviral integrations within this genic region could be influencing survival. While the level of GSDMD transcript did not seem to be greatly affected, translation of GSDMD may have been affected by the antisense RNA or by the increase in L0C100310756, also antisense to the GSDMD gene locus. These results are consistent with an effect on CARD8 inflammasome mediated cell death, but further studies would be needed to validate these findings.

Persistent HIV+ cells share many characteristics with HIV-associated T Cell lymphomas

As previously discussed, Mellors et. al. [17] described several HIV + T cell lymphomas with clonal provirus integration at sites in STAT3 intron 1, near those in the GFP-High replicates described here. Our next step was to understand how persistent HIV/GFP+ cells, especially from replicates 2, 3, and 4, compared to the HIV + T cell lymphomas. Publicly available bulk RNAseq data from the two high-grade T cell lymphomas [17], were available for comparison with the infected GFP-sorted primary T cells from Donor 4. DEseq2 was used for differential gene expression analysis. Principal component analysis showed that the lymphomas differed greatly in PC1 from both the HIV/GFP+ and GFP- sorted T cells infected and grown in vitro (Fig 7A). These transcriptional differences could be due to a variety of factors such as heterogeneity of tumor cell populations and further events, including additional mutations or HIV proviruses at other integration sites [17] in the samples compared. Nevertheless, to understand if the GFP-High replicates model these HIV + T cell lymphomas, we compared the pattern of differentially expressed genes in GFP-High replicates relative to the GFP-Low replicates compared to the lymphomas (Fig 7B). STAT3 and its downstream targets SOCS3 and SGK1 were upregulated in the Replicate 2 and 3 GFP+ cells compared to their corresponding GFP- cells. Cytotoxic T cell markers such as GZMB, GZMH and NKG7, were also upregulated. Mellors et. al. [17] also noted the increased expression of STAT3, GZMB and TNFRSF8 (CD30) as important markers of the HIV+ lymphomas.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Similarity of STAT3 dysregulation and cytotoxic phenotype in GFP-High GFP + Cells to HIV-driven T cell lymphoma.

Bulk RNAseq data from GFP-sorted Donor 4 day 126 samples were compared to data from HIV+ high-grade T cell lymphomas [17] using DESeq2. A) Principal component analysis. B) Heatmap of differentially expressed genes of GFP-High replicates compared to GFP-Low replicates from Fig 3 and high-grade T cell lymphoma. Not all differentially expressed genes are shown. Genes associated with STAT3 signature and cytotoxic phenotype are highlighted. C) Bulk RNA-seq data from Donor 4 replicates and high-grade T cell lymphomas compared to ALK- anaplastic large cell lymphoma related genes. Gene list generated from [31].

https://doi.org/10.1371/journal.ppat.1013087.g007

Increased STAT3 activity, GZMB expression, and CD30 expression are markers of anaplastic large cell lymphomas (ALCL) of T cell origin, which PWH are of increased risk of developing [20]. Mellors et. al. [17] have also shown clonal proviral integrations within STAT3 from DNA extracted from formalin-fixed paraffin-embedded samples of T cell ALCL. ALCLs are distinguished by CD30 expression and usually driven by mutations in the receptor tyrosine kinase ALK [77]. However, ALK-negative ALCLs, which are driven by STAT3 dysregulation and express cytotoxic factors, are also the type of ALCL described with clonal HIV integrations. Using markers of ALK- ALCL we compared the GFP-sorted Donor 4 replicates with the high-grade T cell lymphomas to see if any of the Donor 4 replicates might model HIV-driven lymphoma (Fig 7C). This comparison confirmed that all samples lacked ALK expression. However, the upregulation of the T cell activation and ALCL marker CD30 (TNFRSF8) was present in all samples, including GFP + , GFP- and tumor samples. While many of the genes associated with ALCL were expressed in the Donor 4 replicates, previously mentioned genes, TNFRSF8, STAT3, GZMB, and PRF1 stood out as upregulated in the resurgent replicate 2 HIV/GFP+ cells and high-grade T cell lymphoma samples. Additionally, clusterin (CLU) expression was also upregulated in both the GFP+ samples, particularly in replicates 1, 2 and 3, and the lymphomas. Clusterin is an anti-apoptotic cell survival marker [78] that has been characterized in oncogenesis [79]. These comparisons of the GFP-High replicates to the HIV+ high-grade T cell lymphoma samples suggest that the GFP-High replicates 2 and 3 model at least some oncogenic pathways and T cell phenotypes contributing to the transformation of the HIV + T cells in vivo.

Primary human CD4 + T cell in vitro culture

Primary - CD45RA + CD4 + T cells from normal human donors 3, 4 and 5 were obtained from STEMCELL Technologies, inc. (Cells from donors 1 and 2 were described in [43]). Our study did not include any live participants. Cells were cultured in RPMI 1640 (Gibco) supplemented with 10% Fetal Bovine Serum (Gibco) and 1% Pen/Strep (Gibco) and recombinant IL-2 (30 U/mL) (STEMCELL Technologies). Cells were stimulated as needed using anti-CD3/CD28 stimulation beads, Dynabeads^Tm Human T-ActivatorCD4/CD8 for T Cell Expansion and Activation, (Gibco, Cat#11131D) for 3–4 days and magnetic Dynabeads^Tm removed. T cells were maintained at about 5x10⁵ per ml of medium. Recombinant IL-2 was added every 3 days (30 U/mL) and T cells were stimulated with anti-CD3/CD28 beads whenever the count started to decrease (approximately every 10–14days) (Fig 1A). when the cell count exceeded 5x10⁵ cells/mL new medium was added but old medium was not discarded, unless cell counts began to drop below 5x10⁵ cells/mL.

HIV transduction

After initial stimulation with anti-CD3/CD28 beads, primary T cells were infected with a replication incompetent NL4–3 ΔEnv ΔNef EGFP vector [47], with EGFP in place of Nef. The vector was pseudotyped with VSV-G. Spinoculation was performed at 800xg for 90minutes at 37^oC with 1μg/mL of polybrene to increase transduction efficiency. Cells were infected at a MOI of 1 gfp-transducing unit per cell (TU/cell). Virus-containing medium was removed immediately after spinoculation and replaced with fresh IL-2-containing RPMI media.

GFP nomenclature

Throughout this paper the nomenclature for the gfp reporter varies to follow standard naming of retroviral gene and protein and to differentiate genotypes from phenotypes. The use of gfp refers to the gene within the HIV vector, Gfp will refer to the protein. Finally, GFP + , GFP- and HIV/GFP+ will refer to the phenotype of the cells that either express or do not express Gfp.

Flow cytometry and FACS

Primary T cells were analyzed for cell viability using propidium iodide (PI) (Invitrogen, Cat#: P3566) or Zombie Violet (Biolegend, Cat# 423114) staining. GFP phenotype was used to identify cells with active HIV expression. Cell viability and Gfp expression were assessed by flow c cytometry using an LSRII (BD Biosciences) or Attune NxT (Thermo Fisher) analyzer. FACS was used to sort live cells into GFP+ and GFP- using a FACSAria II Cell Sorter (BD Biosciences) and Bigfoot Spectral Cell Sorter (ThermoFisher). To assess intracellular protein expression, T cells were fixed and permeabilized using Cyto-Fast Fix/Perm Buffer Set (Biolegend, Cat#: 426803), then stained with monoclonal antibodies for STAT3 (Abcam, Cat#: ab300101) and GZMB (Biolegend, Cat#: 372219). Isotype control for STAT3 staining (Abcam, Cat#: ab199093) and isotype control for GZMB (Biolegend, Cat#: 400130). For pSTAT3 staining, T cells were fixed and permeabilized using the True-Phos Perm Buffer Set (Biolegend, Cat#: 425401) then stained using monoclonal antibody for pSTAT3 (Y705) (Biolegend, Cat#: 651008) and isotype control (Biolegend, Cat#: 400130). Protein expression was assessed as above. For analysis of T cell phenotype, T cells were stained with antibodies for CD45RO (Biolegend, Cat#: 304215) and CCR7 (Biolegend, Cat#: 353203) and isotype controls (Biolegend, Cat#: 400331 and 400212) used to set population gates.

Provirus quantification (qPCR)

Genomic DNA was extracted and prepared using DNeasy Blood & Tissue Kit (Qiagen, Cat#: 69504). HIV proviral LTR sequence primers (Fwd: ATGCTGCATATAAGCAGCTGC, Rev: GAGGGATCTCTAGTTACCAGAG) were used in comparison to sequence primers for the single-copy CCR5 gene (Fwd: CAAAAAGAAGGTCTTCATTACACC, Rev: CCTGTGCCTCTTCTTCTCATTTCG) to quantify proviruses per cell. A standard curve of HIV provirus per cell was used based on mass calculations of the HIV vector plasmid diluted in cellular DNA (Raji cells) (S1 Fig). This standard curve was used to determine that there were about 2 proviruses per cell.in DNA from the ACH-2 cell line, which originally contained a single HIV provirus, but has been shown to accumulate more proviral copies during passage [51,92] (Courtesy of the Kearney Lab, NCI) and used to provide a standard curve for quantifying proviral DNA in the donor 4 samples. qPCR reactions were performed using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, Cat#: 1725274) and read out using aCFX Connect Real-Time System (Bio-Rad).

Integration site analysis (ISA)

DNA was extracted from cells using a DNeasy Blood & Tissue Kit (Qiagen, Cat#: 69504). Linker-mediated PCR and library preparation followed the Wells et al [44] protocol. 1ug of DNA from each sample underwent fragmentation, end repair and linker ligation using the NEB Next Ultra II FS DNA Library Prep Kit for Illumina (New England Biolabs, Cat#: E7805S). Linker-mediated PCR was performed followed by nested PCR which was added in the Illumina adapters for sequencing. The primers used were as described by the Wells et al [44] protocol except that they lacked the 8-nuecliotide inline-index sequence. Paired-end sequencing was performed on the Illumina MiSeq V3 and reads of 150 NT in length were obtained.

Using the Illumina adapters, which contained both 5’ and 3’ indices, samples were demultiplexed. Each sample underwent pre-alignment filtering and trimming, removing sequences using cutadapt [93] to remove linker and HIV LTR sequences. After trimming, sequences less than 20 NT were discarded. Reads were then aligned to the hg19 genome using Hisat2 [94]. Each alignment was annotated and unique sites, and number of breakpoints were determined utilizing python. The ISA Pipeline can be found on GitHub. For visualization and analysis, unique integration sites that were located close to one another were added together to get an estimation of clonality of integration within specific sites.

RNAseq analysis

RNA was prepared using RNeasy Mini Kit (Qiagen, Cat#: 74104) on both unsorted and sorted cells. A cDNA library was prepared from unsorted cells using TruSeq mRNA library preparation (Illumina) and single-end sequences (150 bp read length were determined using a HiSeq (Illumina) sequencer. Sorted samples had less RNA, therefore libraries were prepped using TruSeq Stranded Total RNA with Ribo-Zero (Illumina) and analyzed using the NovaSeq6000 (Illumina) with a 150 bp single end read length.

Transcriptome visualization

RNAseq reads were aligned to the hg38 reference genome using hisat2 [94]. The stranded, aligned sequences were used for visualization using the Integrated Genome Viewer (IGV) software [95,96]. IGV coverage plots were color coded based on the orientation of aligned reads.

Differential gene expression and gene set enrichment analysis

HTseq [97,98] was used to count hg38-aligned reads for each sample. Transcript counts were used to calculate TPM and used for differential gene expression by DESeq2 (56), a well characterized pipeline for normalizing gene counts and determining differentially expressed genes. Genes that had a log2Fold change greater than 0.5 and adjusted Pvalue less than 0.05 were considered differentially expressed.

Gene set enrichment analysis (GSEA) [57] was used on normalized gene counts from DESeq2 output. The normalized counts were compared between the groups. The number of permutations was 1000 and the permutation type was gene_set, as there were less than seven replicates per group. The gene sets were derived from the Molecular Signatures Database (MSigDB). Hallmark gene sets, groups of 50 pre-defined genes whose expression is positively or negatively associated with specific cell processes, were selected for analysis. To determine the significance of a gene set, the false discovery rate (FDR) q-value was used to adjust for the number of permutations, thus accounting for the likelihood that a gene set is significantly enriched given the number of permutations. FDR q-values less than 0.05 were considered significant.

STAT3 variant comparison

Hg38 aligned RNAseq reads were sorted and indexed. Bcftools [99] was used to identify variants compared to the hg38 consensus at individual bases in the STAT3 gene. A quality score greater than 90 was used to identify STAT3 variants from donor 4 RNAseq reads in S1 Table.