Clonally expanded CD4⁺ T cells can produce infectious HIV-1 in vivo

Clinical Summary.

The patient was a 58-y-old African-American man diagnosed with an HIV-1 infection in May 2000. We characterized the virus present in samples obtained from this patient over 13 y, from the time cART was initiated until his death from metastatic squamous cell carcinoma. At the time of his initial HIV-1 diagnosis, his CD4⁺ T-cell count was 16 cells/µL, and plasma HIV-1 RNA was 5.4 log₁₀ copies/mL. Analysis of HIV genetic variation using single genome sequencing (SGS) of the HIV pro–pol region (7) revealed that the virus present at the time of diagnosis was genetically diverse (nucleotide average pairwise distance of 1.4%; Fig. S1). cART was started, and plasma HIV-1 RNA decreased to <50 copies/mL within 4 mo, with typical decay kinetics (8), and remained <50 copies/mL for 5 y, with transient viremic periods due to nonadherence to medication. Despite a good virological response to cART, immune recovery was incomplete (1), and the CD4⁺ T-cell count never exceeded 350 cells/µL (Fig. S1).

About 12 y after cART was initiated, plasma HIV-1 RNA increased to 100–200 copies/mL (Fig. 1). SGS of the viral RNA in plasma revealed two distinct populations: a genetically diverse group of drug-resistant viruses and a second population of identical viruses that lacked drug resistance mutations (Fig. 1 and Fig. S1). At the time of the increase in viremia, the patient was diagnosed with human papilloma virus (HPV)-negative squamous cell carcinoma of the tongue. The cART regimen was changed to optimize efficacy against the drug-resistant variants, and viremia decreased to 50–100 copies/mL but did not become undetectable in commercial assays (Fig. 1 and Fig. S1). The reduction in viremia was associated with a decline of the drug-resistant viral variants to below the limit of detection, but the population of identical viruses that did not carry resistance mutations persisted for more than 8 mo at a level similar to that seen before the change in cART (Fig. 1B). During the subsequent period when the patient was treated with radiation and chemotherapy, the size of the primary tumor decreased, the level of plasma HIV-1 RNA declined gradually to less than 40 copies/mL, and the proportion of virus without resistance mutations increased to >90% (Fig. 1 A and C).

Clonal Expansion of the Cell Lineage Responsible for Plasma Viremia.

We previously showed that these identical viral sequences were produced by a clonally expanded cell containing a provirus, which was integrated into a region of the human genome that cannot be mapped to a unique location; thus, the location is ambiguous (3) and the provirus was designated AMBI-1. AMBI-1 viral RNA was first detected (as 1 of 19 sequences obtained) in a plasma sample taken when plasma HIV-1 RNA was <50 copies/mL at year 9.5 of cART, 2 y before viremia increased to a level that was detectable by routine clinical HIV-1 RNA assay. AMBI-1 viral RNA was not detected in several samples taken earlier, between 0.2 and 7.8 y after the initiation of cART. At the time of the increase in viremia, after 12 y of cART, the cells that contained the AMBI-1 provirus comprised one of the most highly expanded HIV-infected clones detected in this patient; analysis of peripheral blood mononuclear cells (PBMCs) taken after 12.1 y of cART (about 11% of the PBMCs were CD4⁺ T cells) revealed that 13% of the pro–pol sequences were AMBI-1 (Fig. 1C). We had previously estimated, based on integration site analysis, that AMBI-1 constituted a smaller fraction of the proviral DNA (at least 3.2%) (3). These data imply that a substantial fraction of the proviruses detected in the integration site analysis were defective and may have had large internal deletions, or hypermutations, which would have prevented their detection by PCR for pro–pol DNA. Quantitative analysis of PBMCs taken after 12.1 y of cART revealed 209 HIV-1 DNA copies/million PBMC; thus, we estimate that there were ∼9 million cells containing the AMBI-1 provirus in the patient at the time of the viral rebound at year 12 (Fig. S2). AMBI-1 proviruses were not detected in PBMC obtained after 3.6 or 7.8 y on cART (Fig. 1C and Dataset S1), suggesting that extensive expansion of this clone occurred after 7.8 y on therapy.

To obtain the full-length sequence of the AMBI-1 integrated provirus, we selectively PCR-amplified two overlapping DNA fragments from CD8-depleted CD4⁺ T cells (12.1 y on cART), using primers that matched the flanking host and internal HIV-1 sequences (Fig. 2A and SI Materials and Methods). Sequence analysis showed that the reading frames for all viral genes were open and that there were no obvious disabling mutations; env sequence analyses predicted that AMBI-1 was CCR5-tropic (15% false-positive rate by Geno2Pheno; GENAFOR).

Infectivity of AMBI-1 Provirus.

To determine whether the AMBI-1 provirus was infectious, we cotransfected the two overlapping amplified fragments into 293T cells using lipofectamine 2000 (Fig. 2B). Supernatants from the transfection were used to infect CD8-depleted lymphoblasts from a healthy HIV-negative donor. As shown in Fig. 2A, HIV-1 p24 antigen levels increased exponentially, and sequencing of the corresponding viral RNA showed that it was identical to AMBI-1. The same transfection supernatants were serially diluted and used to infect TZM-bl indicator cells, in the presence and absence of 300 nm efavirenz, a nonnucleoside RT inhibitor. Luciferase activity above background was detected in cultures in the absence of efavirenz but not in its presence, again demonstrating the presence of replication-competent HIV-1 (Table S1).

Table S1.

Infection of TZM-bl cells with viral supernatants

		2-Fold diluations
Template			Virus dilution in plate
	A
No drug	B		1:2	1:4	1:8	1:16	1:32	1:64	1:128	1:256	1:512	Cell control
	C
	D
300 nM EFV	E		1:2	1:4	1:8	1:16	1:32	1:64	1:128	1:256	1:512	Cell control
	F
	G
	H
			pNL4-3 Plasmid T1 RLU Minus Cell Control
		Average	377.62	572.45	692.80	736.23	772.31	566.81	368.27	216.19	108.36
	A
No drug	B		375.52	530.12	649.92	708.90	721.49	531.50	371.36	201.55	105.19
	C		363.75	571.90	696.14	747.73	794.75	585.42	389.93	222.46	113.21
	D		393.58	615.35	732.36	752.07	800.70	583.52	343.51	224.55	106.68
300 nM EFV	E		−1.09	0.74	0.95	0.83	0.74	0.86	0.66	0.43	0.25
	F		−1.09	0.04	0.40	0.21	0.02	0.14	0.05	0.04	0.06
	G		−1.09	−0.16	0.60	0.01	0.01	0.13	0.02	−0.02	0.02
			Ambiclone T1 RLU Minus Cell Control
		Average:	36.38	30.34	14.53	6.78	2.78	1.60	0.75	0.23	0.06
	A
No drug	B		29.97	29.22	15.57	7.29	2.29	1.45	0.57	0.20	0.02
	C		41.45	31.68	14.52	7.16	3.28	1.33	0.86	0.32	−0.02
	D		37.71	30.12	13.49	5.89	2.78	2.03	0.83	0.17	0.19
300 nM EFV	E		−0.30	0.05	−0.05	0.05	−0.01	−0.08	−0.03	−0.08	−0.11
	F		−0.03	0.20	0.26	0.14	−0.02	−0.04	−0.03	0.06	0.09
	G		−0.26	−0.01	0.05	−0.02	−0.17	−0.16	−0.14	−0.03	−0.10
	H

Open in a new tab

Infection of TZM-bl cells with serially diluted HEK293T AMBI-1 T1 supernatants was performed as described in SI Materials and Methods. Triplicate infections were done with each of the supernatant dilutions in the presence and absence of 300 nm of efavirenz. The data from each infection are shown as relative light units (RLUs) minus the uninfected cell control (sham infection). Virus dilution are color coded: 300 nM Efavirenz (light orange); no drug control (grey). Virus production (average RLU): light green.

Recovery of Replication-Competent HIV from Clonally Expanded Cells.

To determine whether AMBI-1 could be directly recovered from PBMCs, we performed an infectious virus recovery assay on cells obtained after 12.2 y of cART (SI Materials and Methods). Serial dilutions of purified total CD4⁺ T cells (CD4⁺ cells were selected without regard to their activation state) were phytohemagglutinin (PHA) treated, cocultured, in replicates, with γ-irradiated feeder cells and CD8-depleted allogeneic blasts from healthy HIV-negative donors for 28 d, and the culture supernatants were assayed weekly for virus production by measuring p24 antigen. As shown in Fig. 2C, infectious virus was recovered from multiple coculture replicates; none of the p24-negative wells contained HIV-1 RNA. The titer of infectious HIV-1 in the starting sample, calculated by a maximum likelihood estimate (9), was 9.1 infectious units per million CD4⁺ T cells. Thus, positive cocultures containing 3 × 10⁴ or 1 × 10⁵ donor CD4⁺ T cells were likely to have arisen from a single patient-derived cell that produced infectious virus. SGS analysis of p6-RT sequences obtained from virus-positive wells that had been seeded with 3 × 10⁵, 1 × 10⁵, and 3 × 10⁴ CD4⁺ T cells showed that there was a single viral variant in each well. As shown in Fig. 2C, 5 of the 11 positive coculture wells contained sequences identical to AMBI-1, including 2 wells with 100,000 cells and a well with 30,000. Full-length sequencing of one of the isolates (derived from the well containing 30,000 cells) showed that the virus was identical to AMBI-1 and not a mixture or a recombinant. Culture supernatant from this well was used to infect CD8-depleted lymphoblasts from a healthy HIV-negative donor, resulting in rapid increases in HIV-1 p24 antigen (Fig. 2D). Sequence analysis of the HIV-1 RNA from this viral passage was identical to AMBI-1. Analysis of the HIV-1 sequences in the supernatants from the other positive wells revealed that a second variant, referred to as outgrowth 1 (OG-1), was also present in multiple wells (Fig. 2C), suggesting that there is a second clonally expanded cell in this patient that can produce infectious virus, although integration site analysis will be needed to confirm that OG-1 was present in a clonally expanded cell. OG-1 was transiently detected in rebound viremia during a treatment interruption (Fig. S1B) ∼6 y before the sample used for the in vitro virus recovery assay. Approximately half of the proviruses that were induced produced AMBI-1. We estimated that there were ∼250 AMBI-1 proviruses per million CD4⁺ cells (discussed above) and a total of ∼9 million AMBI-1–infected cells in the infected individual (Fig. S2). Thus, only a small fraction (<5%) of the cells that carry the AMBI-1 provirus produced virus in this assay. It is possible that this calculation may underestimate the abundance of AMBI-1 virus-producing cells present in the infected individual. p24 accumulation in AMBI-1 containing wells was slower than in some of the other wells (Fig. 2 B and C), and it is possible that some of the AMBI-1–positive wells might not have been detected after 28 d in culture in our assay.

Wide Anatomic Distribution of Clonally Expanded Cells and Enrichment in Tumors.

The level of AMBI-1 RNA in plasma declined after treatment of the patient’s cancer and then increased with its recurrence (Fig. 1), suggesting a connection between the extent of the tumor and the abundance of the clone carrying the AMBI-1 provirus. We examined samples obtained at autopsy from the region of the primary squamous cell carcinoma and some of the metastatic lesions (Fig. 3A). These samples showed infiltration by CD4⁺ lymphoid cells (Fig. 3 B and C). HIV-1–infected cells were detected in all of the lymphoid tissues and metastatic lesions tested (summarized in Fig. 3D, Table S2, and Dataset S1); the levels of HIV-1 DNA present in cancer tissue (∼12–20 copies per million cells) were, in general, about 10-fold lower than that detected in lymphoid tissue (Table S2). However, the AMBI-1 provirus and the integration site were detected in DNA derived from both tumor and lymphoid tissues (Table S2 and Dataset S1) and were present as a significantly larger fraction of all of the HIV-1 DNA sequences in the cancer metastases than in lymphoid tissue (Fig. 3D; P = 0.001). Other clonal populations of infected cells, as well as proviruses encoding the replication competent variant OG-1, were detected in both tumor and lymphoid tissues (Fig. S3).

Experimental Design.

We conducted a virologic, immunologic, and anatomic analysis of samples from a single HIV-infected patient from the time of initiating antiretroviral therapy to death 13 y later. The patient was enrolled in research studies of HIV replication (97-I-0082 and 08-I-0221) approved by the intramural National Institute of Allergy and Infectious Diseases (NIAID) Institutional Review Board, FWA00005897, at the National Institutes of Health (NIH) Clinical Center (Bethesda). The study participant underwent an informed consent process and provided written consent for research studies, including genetic testing.

Single Genome Sequencing.

Viral RNA was extracted from plasma and was used to synthesize cDNA with a primer in pol (HXB2 position 3500). Genomic DNA (gDNA) was extracted directly from PBMCs. The cDNA or gDNA was end-point diluted and PCR amplified, and single-genome sequences were obtained from an 1,100-bp region that encodes a segment from p6 through the first 660 nt of RT as previously described (7). Frozen pieces of tissue from autopsy-derived samples were disrupted with disposable pestles and lysed using proteinase K, and gDNA was extracted with DNeasy Blood & Tissue Kit (Qiagen). Isolated gDNA was then end-point diluted and subjected to single genome sequencing applying the same cycling conditions previously described (7) but with primers amplifying the env-U3 region (HXB2 positions from 8694 to 9486). Primer sequences are given in Table S3. PCR conditions were as follows: 94 °C for 2 min, 94 °C for 30 s, 50 °C for 30 s, 72 °C for 1 min 30 s (44 cycles), 72 °C for 3 min, 4 °C hold (14); 94 °C for 2 min, 94 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min (15), and 72 °C for 3 min, 4 °C hold (second step).

Sequences were aligned using Clustal W, G > A hypermutants were identified with Hypermut 2.0 (16), and neighbor-joining trees were constructed in MEGA 6.0 (14) with HXB2 HIV-1 as the outgroup. Phylogenetic structure was tested by bootstrap analysis (1,000 replicates); branches with bootstrap support ≥80% are indicated in figures with an asterisk (16).

Specific Provirus Sequencing.

To amplify and sequence specific proviruses from CD8-depleted CD4⁺ T cells, ad hoc primer sets were designed based on sequences in the human genome near the integration site. Complementary primers spanning different regions of the HIV-1 genome were used. A nested PCR was performed using Phusion hot start flex polymerase (NEB), following the manufacturer’s instructions; annealing temperatures were calculated with the NEB Tm calculator available online (tmcalculator.neb.com). The full-length AMBI-1 genome was isolated using two amplicons that overlapped in the IN/vif region. PCR products were isolated by gel electrophoresis and directly sequenced. Primer sequences are listed in Table S3. The PCR products were used to transfect HEK293T cells to generate infectious virus (see Transfection Analysis).

Quantitation of HIV DNA and RNA.

HIV-1 DNA copy numbers were determined, in DNA extracted from PBMCs, using HIV-1 integrase as the target for PCR amplification, as previously reported (17). To determine HIV-1 DNA levels in lymphoid and tumor tissue, samples were collected in 2 mL SPEX SamplePrep (Metuchen) tubes with 1.4-mm acid-washed zirconium beads. With the samples on dry ice, 1.0 mL TriReagent (Molecular Research Center) was added, the tubes were loaded into a Precellys 24 (Bertin Technologies) tissue homogenizer, homogenized twice at 6,000 rpm for 30 s, frozen for 3 min on dry ice, and then homogenized twice more at 6,000 rpm for 30 s. RNA was isolated by adding 0.1 mL 1-bromo-3-chloropropane (Molecular Research Center), followed by vortexing or homogenization for 15 s, and then centrifugation at 14,000 rpm for 15 min at 4 °C. The aqueous phase, which contained the RNA, was collected and transferred to a new tube containing 240 ng glycogen (Roche Life Sciences). The RNA was precipitated with 0.5 mL 2-propanol, and the pellet was washed with 70% ethanol. DNA was extracted from the remaining organic phase by adding 0.5 mL 4 M guanidine thiocyanate, 1 M Tris base, and 50 mM sodium citrate and repeating the phase separation steps described above. The aqueous phase with the DNA was transferred to a new tube containing 240 ng glycogen and precipitated with 0.4 mL 2-propanol, the DNA was collected, and the pellet was washed with 70% (vol/vol) ethanol. HIV gag DNA copy numbers were determined using a multiplexed qPCR assay, as described previously (18). Samples were tested undiluted and diluted 1:10 with spiked-in RCAS (replication competent avian sarcoma virus) controls incorporated in the multiplexed assay to monitor any inhibition of the assay. Results were normalized to diploid genome cell equivalents of DNA, using quantitative PCR for the CCR5 gene as previously described (19). Results were reported as cell associated HIV-1 copies per 10⁶ cell equivalents (Table S2). No HIV RNA was detected in any of the autopsy tissue samples; however, this probably reflects a prolonged delay in obtaining the autopsy.

Isolation of Integration Sites.

Integration site analysis was performed as previously described (3). Briefly, DNA was isolated from Ficoll-purified PBMCs (5–10 × 10⁶), negatively selected CD4⁺ T cells, or autopsy samples. gDNA was fragmented by random shearing into 300- to 500-bp fragments. Linker-mediated nested PCR was performed to amplify the human genomic regions and the linked viral sequences from both the 5′ and 3′ long terminal repeats (LTRs). Paired end-sequencing was carried out using the MiSEq. 2 × 150-bp paired end kit (Illumina). The sequences of the host-viral junctions and the host DNA breakpoints were determined. The host DNA sequences were then mapped to human genome (hg19) with BLAT. A stringent filter was used to check quality of recovered integration sites. Sequences that have exactly the same integration site but different host DNA breakpoints came from different cells; this test identifies proviruses in clonally amplified cells (3).

Transfection of Amplified Viral DNA.

HEK293T cells were cotransfected (using Lipofectamine 2000) with two overlapping AMBI-1 DNA fragments generated by PCR (10 µg of each). Forty-eight hours later, supernatants from transfected HEK293T cells (referred to as T1 supernatants) were harvested. T1 supernatants were used to infect CD8-depleted lymphoblasts obtained from a healthy HIV-negative donor. The cells were isolated using Ficoll-hypaque density-gradient centrifugation. Cells were incubated in 60 mL T-cell growth medium [phenol red-free RPMI 1640, 10% (vol/vol) heat-inactivated FBS, 60 U/mL penicillin, 60 µg/mL streptomycin, 100 U/mL recombinant human-IL-2, and T-cell conditioned medium] + 0.5 µg/mL PHA for 4 d. On day 4, blasts were CD8 depleted using BD IMag CD8 depletion beads, and two million CD8-depleted blasts were resuspended in medium with 400 µL T1 supernatant and cultured for 2 h. Cells were washed in RPMI and cultured in new medium. Supernatants were sampled at 0, 4, 7, 10, and 14 d after infection and assayed for p24 antigen following the manufacturer’s instruction (Alliance HIV-1 p24 ELISA Kit; Perkin-Elmer). HIV-1 RNA was extracted from the day 7 supernatant, and the pro-RT region was amplified by bulk one-step RT-PCR and bidirectionally sequenced for comparison with the sequence of AMBI-1. To exclude DNA carryover from transfection, the remaining RNA was DNase-treated, reextracted, and subjected to cDNA synthesis, and the pro-RT region was amplified by bulk PCR with or without RT, using primers described for SGS in Table S3.

Infectivity in TZM-bl Cells.

AMBI-1 infectivity was determined using the TZM-bl reporter cell line. TZM-bl cells were infected with serial dilutions of T1 supernatants in the presence or absence of 300 nM efavirenz and were assayed, after 48 h, for luciferase activity. Studies were done in triplicate, and results are shown in Table S1. The luciferase activity of uninfected cells (sham infection) was measured, and this background activity was subtracted from the data.

Ex Vivo Recovery of Infectious Virus.

Virus culture assays were performed as previously described (9, 20) with the modifications described below. Cryopreserved PBMCs were enriched for CD4⁺ T cells by negative selection, serially diluted from 1,000,000 to 10,000 cells per well, and seeded in individual wells in sets of 6. Cells were stimulated with PHA in the presence of a 10-fold excess of γ-irradiated allogeneic PBMCs from healthy HIV-negative donors. Replication-competent viruses were amplified by adding CD8-depleted lymphoblasts from healthy HIV-negative donors. Virus cultures were maintained for 4 wk; supernatants were collected weekly and used to measure HIV-1 p24 antigen by ELISA. The p24-negative wells were tested for the presence of HIV-1 RNA using Roche Taqman v2.0. The frequency of HIV-1–infected cells was determined by a maximum likelihood method as previously described (9). HIV-1 RNA was extracted from the supernatants and subjected to bulk RT-PCR or to p6-RT SGS (as described above), yielding at least 20 sequences per supernatant. Sequences were then aligned with MEGA v6.0 (14) and subjected to phylogenetic analysis. The complete AMBI-1 genome was recovered from the supernatant and the sequence compared with the sequence of the AMBI-1 provirus. RT-PCR was performed with SuperScriptIII and OligoDT (Life Technologies), and cDNA was then amplified as described above to produce three overlapping fragments spanning U5-RT, RT-env, and env-U3 (Table S3). PCR products were subjected to direct sequencing.

Passage of Infectious Virus.

Supernatant from a culture well (replicate 5 of 30,000 patient cells per well) from AMBI-1 was recovered as described above and was used to infect CD8-depleted lymphoblasts from a healthy HIV-negative donor. HIV-1 p24 antigen was measured in supernatants collected on days 1, 4, 7, 10, 14, and 17. HIV-1 RNA in the supernatant from days 7, 10, and 14 was extracted, and the pro-RT region was amplified by bulk RT-PCR and sequenced for comparison with AMBI-1 sequence.

Statistical Analysis.

Differences in numbers of copies of HIV proviruses in lymphoid and tumor tissues were analyzed using Fisher exact test statistics.