Adaptation of an R5 Simian-Human Immunodeficiency Virus Encoding an HIV Clade A Envelope with or without Ablation of Adaptive Host Immunity: Differential Selection of Viral Mutants

In this study, we constructed a simian-human immunodeficiency virus carrying an R5 Kenyan HIV-1 clade A env (SHIV-A). To bypass host immunity, SHIV-A was rapidly passaged in naive macaques or animals depleted of both CD8⁺ and B cells. Next-generation sequencing identified different mutations that resulted from optimization of viral replicative fitness either in the absence of adaptive immunity or due to pressure from adaptive immune responses.

ABSTRACT

Simian-human immunodeficiency virus (SHIV) infection in rhesus macaques (RMs) resembles human immunodeficiency virus type 1 (HIV-1) infection in humans and serves as a tool to evaluate candidate AIDS vaccines. HIV-1 clade A (HIV-A) predominates in parts of Africa. We constructed an R5 clade A SHIV (SHIV-A; strain SHIV-KNH1144) carrying env from a Kenyan HIV-A. SHIV-A underwent rapid serial passage through six RMs. To allow unbridled replication without adaptive immunity, we simultaneously ablated CD8⁺ and B cells with cytotoxic monoclonal antibodies in the next RM, resulting in extremely high viremia and CD4⁺ T-cell loss. Infected blood was then transferred into two non-immune-depleted RMs, where progeny SHIV-A showed increased replicative capacity and caused AIDS. We reisolated SHIV-KNH1144p4, which was replication competent in peripheral blood mononuclear cells (PBMC) of all RMs tested. Next-generation sequencing of early- and late-passage SHIV-A strains identified mutations that arose due to “fitness” virus optimization in the former and mutations exhibiting signatures typical for adaptive host immunity in the latter. “Fitness” mutations are best described as mutations that allow for better fit of the HIV-A Env with SIV-derived virion building blocks or host proteins and mutations in noncoding regions that accelerate virus replication, all of which result in the outgrowth of virus variants in the absence of adaptive T-cell and antibody-mediated host immunity.

IMPORTANCE In this study, we constructed a simian-human immunodeficiency virus carrying an R5 Kenyan HIV-1 clade A env (SHIV-A). To bypass host immunity, SHIV-A was rapidly passaged in naive macaques or animals depleted of both CD8⁺ and B cells. Next-generation sequencing identified different mutations that resulted from optimization of viral replicative fitness either in the absence of adaptive immunity or due to pressure from adaptive immune responses.

INTRODUCTION

Despite intense research effort, there is still no safe, effective vaccine against human immunodeficiency virus type 1 (HIV-1)/AIDS. HIV-1 strains circulating worldwide are highly diversified and geographically unevenly distributed. The main group of HIV-1 strains, group M, is comprised of clades A to K, with certain clades predominating in different parts of the world. As there is evidence of differential antibody (Ab) responses among HIV-1 clades (1), defining the genetic diversity of HIV-1 strains circulating in a given region and developing vaccines targeting the major clades therein is of great interest. HIV-1 clade A (HIV-A) accounts for the majority of strains circulating in Kenya (2); it is also prevalent in Central and East African countries (2–6). HIV-A is found in Eastern Europe and Central Asia as well (3, 7, 8). As HIV-A is associated with a high heterosexual transmission rate compared to other clades (5), it is not surprising that the prevalence of HIV-A is increasing. Together, these findings imply a need to develop an effective anti-HIV-A vaccine.

Nonhuman primate (NHP) models for HIV-1 infection are essential for the preclinical evaluation of candidate AIDS vaccines. HIV-1 infects and replicates only in humans and chimpanzees but does not cause significant disease progression in most of the latter. Simian immunodeficiency virus (SIV) infection of rhesus macaques (RMs) has become a widely used NHP model. SIV can infect RMs via mucosal routes and causes AIDS-like disease. However, SIV and HIV-1 envelopes are so divergent that anti-Env antibodies do not cross-react. To overcome such limitations, simian-human immunodeficiency virus strains (SHIVs) have been generated. Almost all SHIVs are chimeras that are built from the SIVmac239 backbone, aside from newly described SHIV strains constructed with an SIVmac251-derived backbone (9), and carry HIV-1 env, vpu, tat, and rev. SHIV infection of RMs has been widely used to evaluate HIV-1 Env-based interventions, including active and passive immunization.

To be biologically relevant, NHP models need to reflect the salient features of HIV-1 transmission in humans. Mucosally transmitted HIV-1 strains detected during acute infection are almost exclusively R5 tropic and have a tier 2 neutralization phenotype. Given the biological characteristics of viruses in humans, SHIVs used for preclinical AIDS vaccine evaluations should be R5 tropic and show some degree of neutralization sensitivity, preferably a tier 2 phenotype. Furthermore, they should not demonstrate overwhelming, rapid disease progression in NHPs like the earlier SHIVs, such as SHIV89.6P, did (10). The earliest SHIV strains constructed used env genes of laboratory-adapted HIV-1, resulting in X4 SHIVs. Some of these either were nonpathogenic or acquired acute pathogenicity upon prolonged replication in RMs or after serial passage (reviewed in reference 11). The next-generation SHIV carried dualtropic HIV-1 89.6 env. However, adaptation of the parental SHIV89.6 to RMs resulted in the highly virulent SHIV89.6P, which acted as an X4 virus in vivo and irreversibly destroyed memory and naive CD4⁺ T cells within 2 weeks. Newer SHIVs have since been constructed encoding R5 env genes (reviewed in references 11 and 12). We have generated a panel of R5 SHIVs, some of which contain recently transmitted env genes of Zambian HIV-1 clade C isolates. The resulting clade C SHIVs (SHIV-Cs) have been used successfully to test the efficacy of passive and active immunization strategies (13, 14). One tier 2 SHIV-C, SHIV-1157ipd3N4 (15), has been used to assess the relative transmissibility of an exclusively R5 virus through different mucosal routes in RMs. Unlike SIV, SHIV-1157ipd3N4 reflected the relative risks of HIV-1 acquisition among humans following different modes of sexual exposure (16). These data reflect the biological relevance of R5 SHIVs when used for mucosal challenge studies.

A number of R5 SHIVs carrying env genes of different clades have been generated. The most frequently used clade B SHIV, SHIV_SF162P3, is a tier 2 virus; there is also a tier 1 version (SHIV_SF162P4). SHIVs expressing HIV-1 clade B transmission/founder env have also been developed by inoculating RMs with cocktails of SHIV variants (17). Recently, a tier 2 SHIV carrying a clade E env was adapted to RMs (18). Three groups have reported constructing SHIVs carrying clade A envs. Himathongkham et al. (19) constructed three SHIV-A strains, none of which were replication-competent in RM lymphoid cells. The second group systematically analyzed the role of RM CD4 in restricting efficient replication of SHIV-As (20–22). Significant differences in the CD4 amino acid sequences between humans and macaques induced Env mutations during SHIV-A adaptation in vitro and altered the sensitivity of progeny viruses to neutralizing monoclonal antibodies (MAbs). It is interesting to note that introduction of an SIV vif allele into the SHIV-A construct resulted in better viral replication kinetics in lymphocytes of pig-tailed macaques (22). The last group generated SHIV-A variants using the backbone of an SIVmac251-derived clone, SIVmac766, and modified Env residue 375 to improve the suboptimal binding efficiency to rhesus CD4 (9).

Here we report the construction of SHIV-KNH1144, a chimera carrying env of the primary isolate HIV-A KNH1144 from Kenya (2). The parental infectious molecular SHIV-A clone underwent rapid serial passage through six naive RMs. We obtained the biological isolate, SHIV-KNH1144p1, from the sixth recipient. However, this early-passage virus had suboptimal replication kinetics. Therefore, we decided on a novel adaptation strategy: to allow unrestricted virus replication in the absence of adaptive host immunity. To achieve this, we simultaneously ablated CD8⁺ and B cells with cytotoxic MAbs. Although Hatziioannou et al. have used ablation of CD8⁺ cells for viral adaptation to macaques (23), our group is the first to our knowledge to use the combination of anti-CD8 and anti-CD20 MAbs to temporarily block the generation of both adaptive T-cell and antibody-mediated antiviral immune responses to optimize the host milieu for viral replication and adaptation. This strategy was previously used to adapt SHIV-E to RMs (18). Single-agent anti-CD20 MAb has not been used by others for the purpose of SHIV adaption, although this MAb has been employed by Mao et al. to assess the influence of B cells on acute SHIV infection (24).

SHIV-A replicated to extremely high levels for several weeks, which resulted in CD4⁺ T-cell depletion requiring necropsy. Infected blood was passaged into two nonimmunodepleted RMs, where progeny SHIV-A demonstrated remarkably increased replicative capacity as well as disease progression to AIDS. We reisolated a passage 4 virus, SHIV-KNH1144p4, which replicated in peripheral blood mononuclear cells (PBMC) of all randomly selected RM donors tested. To understand the mechanisms involved in successful adaptation, we compared passage 1, passage 2, and passage 4 SHIV-As by next-generation sequencing. Interestingly, novel mutations were identified that became fixed in viruses where adaptive host immunity had little influence. These mutations were mapped and linked to the “fitness” of viral replication. In contrast, SHIV-KNH1144p4 had mutations that were consistent with immune selection pressure exerted by cell-mediated and neutralizing Ab-mediated mechanisms.

RESULTS

Construction and characterization of parental SHIV-KNH1144.

To screen for intact R5 HIV-A env clones, we first inserted multiple HIV-A env genes into expression plasmids. The resulting constructs were cotransfected into 293T/17 cells with HIV-1 ΔEN plasmid, a proviral construct with env and nef deleted and encoding green fluorescent protein (GFP) in lieu of nef to produce pseudotyped viruses. The latter were used to infect CEM.NKR.CCR5 cells and tested for GFP expression, which could be observed only if the pseudotyped viruses carried infectious HIV-1 Env. This was the case for HIV-A KNH1144 env, which was the only one that gave rise to an infectious SHIV-A, although we had tested a number of other HIV-A env genes (data not shown).

Next, the infectious KNH1144 env was inserted into the proviral backbone of SHIV-1157ipd3N4 by swapping env genes (Fig. 1). SHIV-1157ipd3N4 is a mucosally transmissible, pathogenic R5 tier 2 SHIV-C; it contains a duplication of the NF-κB binding site in the U3 region of the long terminal repeats (LTRs) to increase replicative capacity and response to stimulation by cytokines, especially tumor necrosis factor alpha (TNF-α) (15, 25). Since the HIV-1 LTR has two to four NF-κB sites while the SIVmac239 LTR has just one, the engineered LTR of SHIV-1157ipd3N4 is more akin to that of HIV-1. An infectious virus stock of parental SHIV-KNH1144 was produced by transfecting 293T cells, using cell-free filtered supernatant to infect RM PBMC depleted of CD8⁺ cells, and collecting cell-free supernatants. The resulting virus was tested for coreceptor usage in a panel of different cell lines expressing various coreceptors. SHIV-KNH1144 was able to replicate only in cells expressing CD4 and CCR5 (data not shown).

FIG 1.

Strategy for cloning of the parental clade A SHIV-KNH1144 infectious molecular clone (IMC). SHIV-1157ipd3N4 (15) was used as the backbone. The fragment of HIV-KNH1144 between the KpnI (K) and BamHI (B) sites was amplified and swapped with the corresponding region of the proviral backbone. TM, transmembrane region; NN, two NF-κB sites present in the 3′ LTR. This duplication will be copied into the 5′ LTR during the retroviral life cycle.

When SHIV-KNH1144 was tested in a panel of PBMC cultures obtained from randomly selected RM donors, it replicated in only a few of the donor cultures (data not shown). To generate an NHP/SHIV model that represents the Kenyan HIV-A epidemic, the parental SHIV-KNH1144 proviral clone clearly needed to undergo adaptation to the new RM host.

Initial rapid passage of SHIV-KNH1144 in naive RMs.

To avoid generation of immune escape mutants, we rapidly passaged the virus through six naive RMs; infected blood was transferred at peak viremia from one infected animal to the next recipient (Fig. 2A). Before passage, we measured viral RNA (vRNA) loads to ensure that the copy number was ≥10⁴ copies/ml, a value that in our experience is associated with persistent systemic infection and subsequent seroconversion in RMs. This condition was fulfilled for all transfers via infected blood. Transfer from the second recipient, RFn-9, involved cell-free virus isolated by cocultivation (Fig. 2A, green arrow). Virus was expanded briefly in RM PBMC and administered intravenously (i.v.) to recipient 3, RDd-11. Peak vRNA loads in the first 6 recipients showed a tendency to increase with each new viral passage; in the last recipient, RHf-12, the peak vRNA load was >10⁷ copies/ml (Fig. 2B, bottom right panel). Nevertheless, viral replication was eventually controlled in the recipients, some of which had only occasional, low-level vRNA blips beyond the acute infection (Fig. 2B). Furthermore, none of the RMs showed any decreases in CD4⁺ T-cell counts (data not shown). Together, these findings were compatible with insufficient adaptation of SHIV-KNH1144 progeny viruses to the new host species. We therefore decided to reisolate virus, termed SHIV-KNH1144p1, from peak viremia at week 2 from the last recipient, RHf-12, to characterize the virus and use it for a novel adaptation approach.

FIG 2.

Strategy used to adapt SHIV-KNH1144 in rhesus macaques (RMs). (A) Serial passage of SHIV-KNH1144 in RMs. SHIV-KNH1144 was first passaged rapidly through a series of 6 naive RMs (RRp-9 through RHf-12). SHIV-KNH1144p1 was isolated from RHf-12 at week 2 postinoculation by cocultivation of RM PBMC. SHIV-KNH1144p1 was passaged further through RTp-9 (boxed), which was immunodepleted of CD8⁺ and CD20⁺ cells. Virus concentrated from plasma from RTp-9 at week 4 was designated SHIV-KNH1144p2 and used for next-generation sequencing. Afterwards, 2 naive RMs (RZy-11 and RZs-7) were inoculated via transfer of blood from RTp-9 at week 8 postinoculation, the time of necropsy. The biological isolate SHIV-KNH1144p4 was isolated from RZy-11 at week 36 postinoculation by RM PBMC coculture. w, week postinoculation. (B) Viral RNA (vRNA) loads of the first series of RMs in the rapid passage.

Ablation of adaptive host immunity and NK function to optimize SHIV-A adaptation.

Recently, we described the adaptation of a SHIV-E to RMs by employing a strategy to allow unbridled virus replication in RMs by transiently inhibiting adaptive host immunity along with components of innate immunity (18). This same strategy was used to continue adapting SHIV-A in the next RM, RTp-9. Using a cytotoxic anti-CD8α MAb (cM-T807), we ablated CD8⁺ T cells and NK cells, most of which express the CD8α marker in RMs. Simultaneously, we decided to prevent the formation of anti-SHIV Ab responses by ablating B cells with the anti-CD20 MAb Rituxan. After the double-immune depletion, animal RTp-9 was inoculated with SHIV-KNH1144p1 i.v. (Fig. 2A). Removal of these host defenses indeed resulted in sustained, very high vRNA loads of >10⁸ vRNA copies/ml for 8 weeks (Fig. 3A). Of note, the expected compensatory rise of CD4⁺ T cells to make up for the loss of the CD8⁺ cell fraction was only temporary, and the absolute CD4⁺ T-cell numbers declined significantly toward the end of the disease course (Fig. 3B). RTp-9 developed AIDS-related opportunistic infections and persistent diarrhea at week 8, necessitating necropsy. Blood was collected before necropsy, and virus from RTp-9 was designated SHIV-KNH1144p2. Immunohistochemistry revealed disseminated cytomegalovirus infection in multiple organs, including interstitial pneumonitis and enteritis.

FIG 3.

Virus passage in the immune-ablated RM RTp-9 depleted of CD8⁺ and B cells with cytotoxic MAbs. (A) RTp-9 was given doses of MAbs against CD8 (blue arrows) and CD20 (purple arrows) to destroy CD8⁺ T cells and NK cells as well as B cells, respectively. SHIV-KNH1144p1 was inoculated i.v. repeatedly (red arrows). vRNA loads and CD8⁺ T-cell counts are shown for RTp-9. The area under the curve (AUC) for vRNA loads is shown in gray. (B) Percentage of CD3⁻CD19⁺ B cells (purple line) and CD4⁺ T-cell counts (dashed black line) in RTp-9. (C) AUC for vRNA load levels of RTp-9 over 8 weeks. (D) AUC for vRNA loads of each of the six RMs within the 2-week intervals between virus inoculation and transfer of infected blood to the next RM. (E) Sum of the AUC values for the initial 6 RMs (indicated by colors ranging from blue to red; panel D was compared to the AUC of the immunodepleted RM RTp-9 [gray]). 13×, 13-fold increase.

The area under the curve (AUC) of the vRNA load in each RM during the 2-week period before infected blood was transferred to the next RM was calculated for each of the six RMs during the initial rapid serial virus passage (Fig. 2A and 3D). Peak vRNA loads in these 6 recipients showed a trend of increasing with each passage; in the last recipient, RHf-12, it surpassed 10⁷ copies/ml (Fig. 2B and 3D). However, viral replication was significantly higher in the immunodepleted RM RTp-9 (Fig. 3A and C), and the vRNA AUC for this animal was ∼13 times greater than the sum of all AUC values for first 6 RMs during the rapid passage at 2-week intervals (Fig. 3E).

Virulence of progeny SHIV-A in two nonimmunodepleted RMs.

To test whether viral passage through the immunodepleted host had increased the virulence of our SHIV-A, the blood collected from RM RTp-9 before necropsy was transferred into two immunocompetent recipients, RZy-11 and RZs-7 (Fig. 2A and 4A and B). The vRNA load patterns of these two RMs were markedly higher throughout the period of observation (Fig. 4A) than those of the first six RMs used during the initial virus passages (Fig. 2B). During the chronic phase of infection, viral loads were now sustained, and animal RZy-11 developed AIDS as defined by absolute CD4⁺ T-cell depletion (<200 cells/μl) (Fig. 4B) as well as opportunistic infections. It had to be euthanized due to disseminated Mycobacterium avium complex infection. Virus was reisolated from RM RZy-11 at week 36 postinoculation and termed SHIV-KNH1144p4 (Fig. 2A and 4A). Clearly, virus transferred from the double-immune-depleted RM RTp-9 had become more aggressive, as shown by improved replication kinetics in both immunocompetent recipients, and was pathogenic.

FIG 4.

Virus passage in immunocompetent animals RZs-7 and RZy-11. (A) vRNA loads in RZs-7 and RZy-11. RZs-7 and RZy-11 were inoculated i.v. with 10 ml of blood from RTp-9 collected at necropsy. SHIV-KNH1144p4 was isolated 36 weeks after infection. (B) Peripheral CD4⁺ T-cell counts of RZs-7 (green line) and RZy-11 (dashed line). The dotted line (200 CD4⁺ T cells/μl) indicates severe depletion compatible with AIDS. (C) Histopathology of colon, lung, mesenteric lymph node, and liver in RM RZy-11 by H&E staining. Top left, the colonic lamina propria is markedly expanded by sheets of epithelioid macrophages, small numbers of lymphocytes, and plasma cells with loss of crypts (scale bar = 50 μm). Inset, acid-fast stain showing numerous intracytoplasmic acid-fast bacilli in macrophages (opportunistic infection). Top right, pulmonary granuloma composed of numerous epithelioid macrophages interspersed with lymphocytes and a multinucleated giant cell (arrowhead) (scale bar = 50 μm). Bottom left, mesenteric lymph node is multifocally infiltrated by numerous epithelioid macrophages (scale bar = 100 μm). Bottom right, the hepatic parenchyma contains a discrete granuloma composed of centrally located numerous epithelioid macrophages rimmed by small numbers of lymphocytes (scale bar = 20 μm).

Replication of parental SHIV-KNH1144 and adapted SHIV-KNH1144p4 in PBMC from naive RM donors.

To assess whether the serial initial virus passages in the six RMs followed by the prolonged, high-level replication in the double-immune-depleted RM RTp-9 had resulted in better SHIV-A adaptation to the RM species, we compared the ability of the parental SHIV-KNH114 versus the late-passage SHIV-KNH1144p4 isolate to replicate in PBMC cultures of 10 randomly selected RM donors. While the parental virus did not replicate appreciably in most PBMC cultures, SHIV-KNH1144p4 was able to replicate in all 10 RM PBMC cultures (Fig. 5A), although the levels of p27 in the culture supernatants varied from donor to donor as expected for cells from an outbred RM population.

FIG 5.

Replication of adapted SHIV-KNH1144p4 in naive RM PBMC and coreceptor usage. (A) Replication of parental SHIV-KNH1144 and SHIV-KNH1144p4 in PBMC from 10 randomly selected naive RMs. The relatively lower level of replication in PBMC from donors 31334 and 31335 is shown at an appropriately lower scale in ng/ml (inset). (B) SHIV-KNH1144p4 was tested for usage of various coreceptors (CCR2, CCR3, CXCR4, CCR5, GPR15/BOB, and CXCR6/BONZO/STRL33) expressed on engineered U87.CD4 or Ghost cells.

Coreceptor usage of SHIV-KNH1144p4.

To determine coreceptor usage, SHIV-KNH1144p4 was incubated with cell lines expressing various coreceptors, including CCR1, CCR2, CCR3, CXCR4, CCR5, GPR15/BOB, or CXCR6/BONZO/STRL33. Like the parental SHIV-KNH1144, SHIV-KNH1144p4 productively infected only cells expressing CCR5 (Fig. 5B). In addition, SHIV-KNH1144p1 and SHIV-KNH1144p4 were also tested in CEMx174-GFP cells and found to be negative for X4 tropism (data not shown). We conclude that the passaged biological isolate SHIV-KNH1144p4 is still exclusively R5 tropic after adaptation. Importantly, unlike what we had observed during SHIV-E adaptation (18), this isolate did not undergo expansion of its coreceptor usage during the extended, high-level replication in the immunodepleted host.

Susceptibility of SHIV-A strains to broadly neutralizing Abs.

To evaluate the susceptibility of the newly generated SHIV-A strains to neutralization, the parental SHIV-KNH1144, SHIV-KNH1144p1, and SHIV-KNH1144p4 were tested against a panel of broadly neutralizing MAbs (BNAbs) and a reference polyclonal IgG preparation (Table 1). For BNAb PGT121, 50% inhibition of the tier 2 SHIV_SF162P3 was seen at 0.005 μg/ml, whereas our SHIV-KNH1144p4 required 1.35 μg/ml (26). Our SHIV-A was also more resistant to VRC01 (2.17 μg/ml, versus 0.10 μg/ml for SHIV_SF162P3 [27]). In addition, the adapted SHIV-KNH1144p4 was categorized as a tier 2 virus using a standard anti-HIV-1 serum panel as well as anti-HIV-C IgG preparations (HIVIG-C) (Table 2).

TABLE 1.

Sensitivity of SHIV-A strains to neutralization by BNAbs and anti-HIV clade C IgG

MAb	Specificity	SHIV-KNH1144		SHIV-KNH1144p1		SHIV-KNH1144p4
		ID50 in TZM-bl cellsa	Maximum % neutralization	ID50 in TZM-bl cells	Maximum % neutralization	ID50 in TZM-bl Cells	Maximum % neutralization
VRC01	CD4bs	0.72	98	0.42	98	2.17	98
3BNC117	CD4bs	1.85	92	2.85	81	2.36	82
CH31	CD4bs	0.71	99	0.61	99	0.67	99
PG9	V2 apex	0.39	85	0.1	94	0.28	78
PG16	V2 apex	0.12	79	0.06	75	0.01	70
CH01	V2 apex	>25	30	>25	35	>25	31
PGT121	V3 glycan	0.9	95	0.54	98	1.35	96
10-1074	V3 glycan	0.95	97	0.38	100	2.19	99
10E8	MPER	<0.01	98	<0.01	100	1.21	92
PGT151	gp120/gp41 glycan	0.01	97	<0.005	100	0.06	97
HIVIG-Cb	Polyclonal	44	94	50	97	376	69

^aValues are the antibody concentration, in micrograms per milliliter, at which relative luminescence units were reduced by 50% compared to values for virus control wells (ID₅₀).

^bHIVIG-C, anti-HIV clade C IgG.

TABLE 2.

Neutralization tiers of SHIV-A strains

Virus	ID50 in TZM-bl Cellsa							Neutralization tier
	I230b	P534b	S230bb	707010117b	707010627b	HIVIG-Cc	GMTd
SHIV-KNH1144	135	233	298	195	3,242	44	359	1B
SHIV-KNH1144p1	76	167	251	433	2,528	50	322	1B
SHIV-KNH1144p4	10	25	66	24	10	376	21	2
Q23.17	85	10	84	136	10	12	40	1B
MW965.26	1,413	22,473	11,236	11,830	20,748	2	9,738	1A

^aValues are the sample dilution (or antibody concentration in micrograms per milliliter for HIVIG-C) at which relative luminescence units were reduced by 50% compared to values for virus control wells (ID₅₀).

^bStandard serum panel.

^cHIVIG-C, anti-HIV clade C IgG.

^dGMT, geometric mean antibody titer.

Next-generation sequencing of SHIV-KNH1144p1, SHIV-KNH1144p2, and SHIV-KNH1144p4.

We sought to determine the mutations that were selected as parental SHIV-A was subjected to the different adaptation protocols. During the rapid serial passages, host adaptive immune responses were mostly avoided since virus was transferred at week 2 postinoculation, the time of peak viremia. Typically, no antiviral Ab responses will yet have formed, and cell-mediated immune responses are only starting to be generated at this early time point. We reasoned that high replication levels at week 2 would allow selection of progeny virions with better replicative fitness in the new host species. Furthermore, the newly inserted R5 HIV-A envelope might also undergo mutations to allow for a better fit with the SIVmac239 building blocks within the context of the chimeric virion. We therefore anticipated that the early SHIV-KNH1144p1 isolate might reveal mutations associated with better replication capacity, or fitness, of the newly created hybrid virus in the RM host environment. We therefore coined the term “fitness mutations” for genomic changes that occurred in the absence of adaptive host immune responses.

The evolution of SHIV-KNH1144p2 also occurred in the absence of host adaptive immune responses due to the double-immune depletion of CD8⁺ cells and B cells. Any emerging mutations in this isolate can likewise be attributed to improved fitness of the virus, and these are referred to as “fitness mutations” as well.

RNA was isolated from SHIV-KNH1144p1, SHIV-KNH1144p2, and SHIV-KNH11p4 isolates and analyzed by deep sequencing. Ideally, the sequencing would be performed on virus isolated from plasma samples collected from viremic animals. Unfortunately, this could be done only for SHIV-KNH1144p2 from animal RTp-9 (plasma samples from other animals were lost). Sequencing reads were aligned to the parental SHIV-KNH1144 genome, and nucleic acid mutations were used to predict changes to the amino acid sequence. The locations having greater than 5% mutated reads are shown according to the position in the parental SHIV-KNH1144 genome (Fig. 6). The locations having >80% of the mutated reads are marked in red, blue, or green (Fig. 6) and represent fitness mutations (F), cytotoxic-T-lymphocyte (CTL)-related mutations (C), and Ab-related mutations (A), respectively. These prevalent mutations are also listed in Fig. 7 with their locations as well as nucleotide and amino acid substitutions. Most mutated reads resulted in amino acid substitutions (Fig. 7). Mutations not represented in at least 80% of the reads are shown in black.

FIG 6.

Frequency of mutations in the SHIV-KNH1144p1, SHIV-KNH1144p2, and SHIV-KNH1144p4 genomes. Sequence reads for each isolate were aligned to the parental SHIV-KNH1144. The height of each bar represents the percentage of sequence reads for a given mutation. Mutated reads at a prevalence of >5% are shown in the corresponding positions in the genome. Mutations identified in the SHIV-KNH1144p1 and SHIV-KNH1144p2 isolates emerged during adaptation in RMs without pressure from host adaptive immunity and are attributed to optimization of viral replication parameters (“fitness mutations”) in the new species; those observed at ≥80% prevalence are termed F1 to F9 and shown in red. Mutations observed in the biological isolate SHIV-KNH1144p4 occurred in the presence of host adaptive immunity, including CTL and Ab responses. Some fitness mutations had become fixed (F1 to F4 and F8) and are shown in red. Mutations appearing at ≥80% frequency are indicated in red (fitness mutations), blue (CTL-related mutations), or green (Ab-related mutations). Unique mutations appearing in the immunodepleted RM RTp-9 are highlighted yellow with a black box; F8 is highlighted yellow with no black box as it appears in the final isolate. Asterisks indicate that these mutations were also found in highly viremic RMs during the course of SIVmac239 infection (28).

FIG 7.

Substitutions in SHIV-KNH1144p1, SHIV-KNH1144p2, and SHIV-KNH1144p4 compared with the parental SHIV-KNH1144. Footnote symbols: a, position in the parental SHIV-KNH1144 genome; b, convergent evolution resulting in optimized SIVmac239 sequences as described in reference 28; c, F4-1 and F4-2, the G→a mutation affects overlapping reading frames (underlined); an amino acid change occurred only in Rev. N/A, not applicable; C1, constant region 1; MSD, membrane-spanning domain; V2 or V3, variable loop 2 or 3; RT, reverse transcriptase; CTL, cytotoxic T lymphocyte; NHR, N-terminal heptad repeat. Mutated nucleotides are in lowercase letters.

Remarkably, even after the short adaptation in RMs, four mutations (F1 to F4) are highly represented in the early-stage isolate, SHIV-KNH1144p1 (Fig. 6, top panel, and Fig. 7). We were surprised to find 100% frequency of F1, a mutation found in the primer binding site (PBS). This mutation became fixed early, as it was present in all subsequent virus populations as well (Fig. 6, middle and bottom panels, and Fig. 7). An identical mutation in the SIVmac239 PBS was found earlier in highly viremic RMs during the course of SIVmac239 infection (28). F2 was localized to the constant region 1 (C1) of gp120 (Fig. 7). This mutation also was found in 100% of the reads. It is not likely to be selected by humoral immune responses given that the virus was rapidly passaged every 2 weeks. F3, located between the membrane-spanning domains 2 and 3 (MSD2 and MSD3) (Fig. 7), is known to lie within the endocytic signal region of gp41; this region interacts with Gag protein during the virus maturation process (26, 29). It appears that F3 was selected to allow for better fit of the newly inserted HIV-A Env into the chimeric particle that contains SIVmac239 Gag. F4, located at position 8594, is found within two open reading frames (ORFs) (Fig. 7); however, the first reading frame located in MSD3 of gp41 did not result in an amino acid change. In contrast, the same mutation induced an amino acid change in the second exon of Rev from E to K. None of these four mutations showed signs of selection from adaptive host immune responses, as expected for the rapid passage. Rather, we believe that these mutations have been selected because they gave the progeny virus a selective advantage due to improved fitness of virus replication in RMs.

Mutations F1 to F4 were also present in the SHIV-KNH1144p2 isolate, although F4 was present at a slightly lower frequency of 78% (Fig. 6, middle panel). In addition to these, five new mutations appeared at high prevalence in this isolate. A deletion of 6 amino acids (F5) was observed in Vpr, resulting in the removal of two amino acids from the protein sequence. F6 and F7 are located in the gp120 V4 and C4, respectively, but were not influenced by humoral immune responses since RM RTp-9 was depleted of B cells with the anti-CD20 MAb. F8 was found at 100% frequency in the gp41 N-terminal heptad repeat but again was not selected due to Ab-mediated responses. Another nucleic acid substitution in a noncoding region was found in the 3′ LTR (F9).

Several additional mutations appeared in the late, fully adapted isolate SHIV-KNH1144p4. This was not unexpected, as the virus was isolated at week 36 postinoculation from RM RZy-11, which had not been subjected to immune ablation. Interestingly, the initial “fitness” mutations F1 to F4 as well as F8 have become fixed and are represented at 100% frequency (Fig. 6, bottom panel, and Fig. 7). F5 and F9 have disappeared while F6 and F7 are no longer found at high prevalence in this isolate. In addition, mutations unique to this isolate that are highly prevalent could be assigned to either selection pressure asserted by CTLs (C1 to C5) or Ab-mediated selection pressures (A1 to A5, located in Env).

Two new mutations appeared in Env, but curiously, neither of the mutations, located at position 6557 (F10) or 7106 (F11), resulted in amino acid changes (Fig. 6, bottom panel, and Fig. 7). Consequently, neither of these mutations could be attributed to selective pressures exerted by either the cellular or humoral arm of the immune system. Such mutations are more likely of a “fitness” nature and may affect RNA processing or other, as-yet-unknown virion properties. Of note, neither of these mutations is in the region of the Rev-responsive element (RRE).

Mutations that may be related to CTL selection pressure (C1 to C5) were found in Gag, reverse transcriptase, integrase, and Nef (Fig. 7); these regions have been identified by others to contain CTL epitopes (30). In addition, mutations C2 and C3 were also found in highly viremic RMs during the course of SIVmac239 infection (28).

Mutations discovered in Env could be attributed to Ab-related selection (A1 to A5) and were localized to gp120 C2, V3, and C4 as well as to the gp41 fusion peptide and N-terminal heptad repeat (NHR) (Fig. 6, bottom panel, and Fig. 7). Mutations at these locations have been described in neutralization escape viruses. Of note, mutation A4 is also located in the env/RRE overlap region. It is therefore possible that this mutation could affect the Rev-RRE interaction in the context of the virus life cycle.

In conclusion, our innovative adaptation strategy combined with deep sequencing has identified mutations generated during the different phases of our adaptation. We found intriguing “fitness” mutations as well as mutations expected to arise from selection pressures exerted by host cell-mediated and Ab-mediated pressures.

DISCUSSION

We constructed an R5 SHIV-A carrying env from the Kenyan HIV-1 KNH1144. SHIV-A underwent rapid serial passage through six RMs and then in an RM with CD8⁺ and CD20⁺ cells simultaneously ablated. Both of these approaches to adapt SHIV-A to macaques were designed to bypass adaptive host immunity. Infected blood was then transferred into two non-immune-depleted RMs, where progeny SHIV-A showed increased replicative capacity and caused AIDS. We reisolated SHIV-KNH1144p4, which was replication competent in RM PBMC, and categorized it as a tier 2 virus. Next-generation sequencing of early- and late-passage SHIV-A strains identified mutations that arose due to “fitness” viral optimization as well as mutations typical for adaptive host immunity in the late virus isolate.

SHIV infection of RMs is an important model to study Ab-based interventions. Preferential neutralization of HIV-1 strains from a given clade was described when tested against plasma samples collected from individuals infected with HIV-1 strains of the same clade (1). Earlier work showed that polyclonal antibodies from HIV-1 clade C-infected individuals could neutralize both HIV-1 clade B and C strains, whereas polyclonal Abs from HIV-1 clade B-infected individuals neutralized only HIV-1 clade B, not clade C (31). In this study, HIV-A strains were optimally neutralized by antibodies from HIV-A-infected persons. The fact that HIV-A reflects the majority of circulating strains in parts of Africa implies a need to develop SHIVs to evaluate candidate vaccines targeting HIV-A. The fully adapted SHIV-KNH1144p4 described here is the first R5 SHIV-A which was found to be pathogenic in RMs.

Not all HIV-1 envelopes fit well with the SIVmac239 building blocks, leading to problems with the replicative capacity of the new SHIVs tested in isolated RM PBMC. Since the parental infectious molecular SHIV-A clone was restricted in its ability to replicate in most RM PBMC cultures tested, the virus clearly needed to be adapted. Our SHIV-A adaptation strategy was designed to minimize the impact of adaptive host immunity. Our first approach involved virus passage at week 2, the expected time of peak viremia. At this time point, cell-mediated immune responses directed against Gag following intravenous SIV inoculation of RMs were measurable, whereas those against other virus targets appeared later (32). However, soluble factors released by CD8⁺ cells in culture systems showed viral suppression when PBMC samples collected as early as 1 week postinoculation were tested. HIV-1-specific IgG usually appears at about 3 to 4 weeks postinfection (33). Our serial SHIV-A passages at 2-week intervals are within the “window period” when host SHIV-specific Ab responses are expected to be absent and CD8⁺ cell-mediated cytotoxicity begins to emerge (32). We were disappointed to realize that the replication kinetics of SHIV-A after passage through the first six RMs were suboptimal and that the virus was suppressed in all six of these chronically infected animals.

We thus decided on a more radical approach: to remove CD8⁺ cells (T cells and NK cells) as well as B cells (CD20⁺ cells) in order to overcome viral suppression by soluble factors generated and CTL activity by the former and to preclude Ab generation by the latter—with the sole aim to maximize virus replication in the next animal. This double-immunodepletion strategy showed the desired result: sustained viral loads of >10⁷ copies/ml. Reverse transcriptase makes a certain number of errors in each round of replication, and these errors are random and subjected to “survival-of-the-fittest” selection. Therefore, the more replication cycles there are, the more mutations will arise, with higher chances of ever-fitter mutants being generated. We reasoned that such high viremia levels, reflected by the substantially greater AUC, would accelerate the selection of mutated progeny virions better able to replicate in the host environment of Macaca mulatta. To test whether this was indeed the case, we examined viral replication kinetics and pathogenicity in two additional animals that were not subjected to immune ablation prior to virus inoculation. The fact that both of these RMs demonstrated not only high vRNA peaks but also sustained steady-state viremia confirmed the markedly improved adaptation of virus found at the time the doubly immunodepleted RM underwent necropsy. This virus had depleted the absolute numbers of CD4⁺ T cells, which led to the development of opportunistic infections. Thus, we have proof-of-concept that SHIV-A had become pathogenic. Limitations of our studies allowed us to enroll only two animals for this experimental phase. The final SHIV-A stock (strain SHIV-KNH1144p4) was isolated at week 36 from animal RZY-11, when AIDS had developed. Virus isolated at this late time point had been exposed to host adaptive immunity; the sustained vRNA loads implied that this virus is no longer easily suppressed by host factors. Ablation of host cellular as well as humoral immunity may be an efficient strategy for adapting other pathogens to a new host species.

We hypothesized that SHIV-A adaptation with minimal selection pressure exerted by host adaptive immunity would yield a distinct signature of mutations, termed “fitness” mutations, in the viral genome that would markedly differ from mutations arising due to CTL and Ab pressures. Such “fitness” mutations could result in better conformity of the HIV-A Env molecules with SIV-derived virion building blocks or host proteins. Likewise, “fitness” mutations in noncoding regions could also accelerate virus replication and result in the outgrowth of virus variants in the absence of adaptive T-cell and antibody-mediated host immunity. To decipher such “fitness” mutations, we performed next-generation sequencing of the early isolate, SHIV-KNH1144p1, and the late SHIV-KNH1144p4. “Fitness” mutations were indeed identified, since genomic RNAs isolated from the early SHIV-KNH1144p1 isolate contained four nucleotide substitutions, which were present in >80% of the viral sequences analyzed. The fact that these four mutations became fixed in 100% of the progeny virus later on is an indication that these newly emerged virus quasispecies were able to totally outcompete the parental virus clone—a true situation of survival of the fittest at the level of the entire animal. These initial four “fitness” mutations were found in different parts of the entire viral genome and consisted of mutations at (i) the PBS, (ii) the C1 region of gp120, (iii) the second Rev exon, and (iv) the endocytic signal region of gp41 (26, 29). The rapid serial passage strategy clearly had prevented the selection of virus with significantly divergent envelopes compared to the parental construct, an indication of the lack of Ab-mediated selection pressure during the early serial adaptation strategy.

Finding a mutation in the middle of the PBS came as a surprise, given that SIVmac239, from which the SHIV backbone had been derived, is virulent in RMs. However, Alexander et al. (28) found that after prolonged replication in RMs, the same PBS T-to-C mutation was selected in vivo, indicating that even the cloned SIVmac239 was not optimally adapted to RM species. In this study, the T-to-C changes were consistently observed in the SIV sequences amplified from all four SIVmac239- or SHIVnef-infected macaques and thus were unlikely due to PCR errors (28).

SIVmac239 preferentially uses cellular tRNA(Lys,3) to initiate reverse transcription, which requires the interaction between the 18-nucleotide PBS and tRNA(Lys,3). The T-to-C mutation at position eight of the PBS (M1* mutation [Fig. 6]) renders the mutated viral PBS exactly complementary to the tRNA(Lys,3). When engineered into the parental infectious RT-SHIV molecular clone, the virus carrying this T-to-C mutation in the PBS (RT-SHIV-TC) replicated much faster than the parental virus in both RM and human cells (34). The fact that both wild-type SIVmac239 and our parental SHIV-A construct underwent identical mutations in RMs represents convergent molecular evolution, even in the context of different viral backbones, and implies that the same selection pressure was operating. The potential mechanisms involved in the selection of the SIVmac239 PBS mutation were discussed in two recent publications, although opinions differed (35, 36).

Two other mutations, C2* and C3*, that were found in the late SHIV-KNH1144p4, have also been observed in SIVmac239-infected RMs after prolonged viremia; both resulted in amino acid changes in Pol (28). These mutations are also examples of convergent evolution of the SIVmac239 genome in different RMs, indicating that SIVmac239 adaptation to the rhesus species is not optimal. Mutations that result in Pol amino acid changes could also have arisen due to selection pressure exerted by host CTL pressure. We have observed other CTL-related mutations, especially in Gag and Nef, which are well-known CTL targets (30, 37).

The mutation M3 that affects the endocytic signal region within the cytoplasmic domain of the HIV-1 gp41 transmembrane protein is particularly important for incorporation of the viral envelope into the virion (26, 29). During the maturation of the particle, viral Gag interacts with this gp41 region; the latter is also important for rapid endocytosis of virions (38). M3 may have been selected and fixed in progeny SHIV-A due to a better three-dimensional fit between SIVmac239 Gag elements and the newly inserted clade A HIV-1 Env in parental SHIV-A.

As expected, most mutations located in SHIV-KNH1144p4 Env occurred in regions previously shown to be related to virus escape from neutralizing Ab responses, including the V3 loop or CD4bs (39–41).

We have attributed mutations C4 and C5 found in Nef of the late virus, which caused amino acid changes, to CTL pressure based upon earlier work by others, according to whom SIVmac239 Nef is a CTL target (42, 43). Interestingly, it is also possible that these mutations were selected because the amino acid substitutions may interfere with the Nef-tetherin interaction. The parental SIVmac239 Nef antagonizes host tetherin, a protein that is implicated in hindering release of infectious lentiviruses from target cells (37, 44, 45). SIV Nef proteins were found to downregulate and antagonize rhesus tetherin through direct physical interaction (44). Therefore, mutations C4 and C5, which interfere with the Nef-tetherin counteraction, may contribute to increased viral fitness via this mechanism.

In summary, we have constructed an R5 SHIV-A carrying env of the Kenyan HIV-1 KNH1144 and have adapted the parental construct to RMs according to a novel double-immunodepletion strategy that resulted in sustained, very high levels of viremia. This in turn led to optimization of viral fitness in vivo as shown by improved viral loads during chronic viremia and disease progression. The adapted virus, the biological isolate SHIV-KNH1144p4, was pathogenic in RMs. This SHIV-A/RM model will provide a unique opportunity to evaluate the efficacy of anti-HIV-1 clade A candidate vaccines. Next-generation sequencing of early- and late-passage SHIV-A strains identified mutations that arose due to “fitness” viral optimization as well as mutations typical for adaptive host immunity and provide important insights for the viral fitness. Our viral adaptation strategy could be applied to adapt other viral pathogens to new host species.

MATERIALS AND METHODS

Construction of the infectious molecular clone SHIV-KNH1144.

HIV-1 KNH1144 (GenBank accession no. AF457066) was isolated from Nairobi, Kenya (gift of Francine McCutchan); the gp160 gene was generated by nested PCR using DNA purified from PBMC of blood donated at the Kenyatta National Hospital and found to be HIV-1 positive (2). While no clinical information is available regarding the donor of the HIV-1-positive blood, the fact that this person donated blood is an indication that he/she probably was asymptomatic at the time of the donation. HIV-1 KNH1144 env was cloned into the SHIV-1157ipd3N4 backbone (15), which resulted in the infectious molecular clone SHIV-KNH1144 (Fig. 1); virus was generated by transfecting proviral DNA in 293T/17 cells.

Construction of the reporter virus NL-LucR.KNH1144p4.

The NL4-3 env region of pNL-LucR.T2A (gift of Christina Ochsenbauer, University of Alabama, Birmingham) was first cloned into pBR322 using the EcoRI/BamHI sites to generate vector pBR322-NL4-3. Next, env of SHIV-KNH1144p4 was excised using KpnI/BamHI sites and cloned into the pBR322-NL4-3 backbone to generate the shuttle vector pBR322-KNH1144p4. Finally, using the EcoRI/BamHI sites, env of pBR322-KNH1444p4 was inserted into the backbone of pNL-LucR.T2A to obtain the final construct, pNL-LucR.KNH1144p4. Infectious NL-LucR.KNH1144p4 was generated by transfecting pNL-LucR.KNH1144p4 proviral DNA in 293T/17 cells and tested for infectivity in TZM-bl cells.

Animal inoculations and virus passages.

The Indian-origin RMs enrolled in this study were prescreened and found to be free of SRV/D, STLV-1, and SIV and to be Mamu-B*008 and Mamu-B*017 negative. The RMs were maintained in accordance with the Guidelines for the Care and Use of Laboratory Animals and the institutional guidelines at the Yerkes National Primate Research Center (YNPRC) at Emory University (Atlanta, GA). All experiments were approved by the Institutional Animal Care and Use Committees of the YNPRC and the Dana-Farber Cancer Institute.

The first RM recipient (Fig. 2) was inoculated i.v. with 7 ml of cell-free virus, which was generated by first transfecting 293T/17 cells with SHIV-KNH1144 proviral DNA, collecting and filtering supernatant of the transfected cells, and then infecting cultured naive RM PBMC. Plasma vRNA loads were measured weekly; if there were >1 × 10⁴ copies/ml at week 1, 10 ml of blood was transferred to the next RM at week 2 postinoculation. The third RM (RDd-11) was inoculated with 10 ml virus supernatant generated by coculture with PBMC from the second RM (RFn-9). SHIV-KNH1144p1 was isolated from the last RM (RHf-12) of the first series of passages; a stock of SHIV-KNH1144p1 was generated by coculturing PBMC from RHf-12 with naive RM PBMC. Next, SHIV-KNH1144p1 was inoculated into RM RTp-9 that was depleted of CD8⁺ cells by the anti-CD8 MAb cM-T807 (50 mg/kg i.v. on days −2 and 20). RM RTp-9 was simultaneously depleted of B cells with the anti-CD20 MAb Rituxan (20 mg/kg i.v. on days −4, 0, and 4). RTp-9 had to be euthanized at week 8 due to AIDS-related opportunistic infections. Ten milliliters of its blood was inoculated into two RMs (RZs-7 and RZy-11). SHIV-KNH1144p4 was isolated from RZy-11 at week 36 by coculturing with PBMC from a naive RM.

Histopathology.

For histopathologic examination, various tissue samples were first fixed in 10% neutral buffered formalin and then processed routinely. After being embedded in paraffin, the samples were sectioned at 5 μm, and then stained with hematoxylin and eosin (H&E).

Virus production and viral replication assays.

Virus was produced by coculturing PBMC from a naive RM donor with PBMC from the infected RMs. PBMC were isolated by Ficoll gradient and stimulated for 3 days with medium supplemented with 5 μg/ml concanavalin A. After washing, the cells were maintained in medium with 20 U/ml interleukin-2 (IL-2) and 10 ng/ml TNF-α. SIV p27 levels were monitored every other day. Virus was harvested and filtered through a 0.45-μm filter. Reisolated SHIV-KNH1144p4 was harvested on days 7, 9, and 11, with a virus titer of 7.1 × 10⁹ copies/ml, 1.8 × 10⁹ copies/ml, and 1.9 × 10⁹ copies/ml, respectively. Parental SHIV-KNH1144 and adapted SHIV-KNH1144p4 were tested for growth in PBMC of 10 naive RM donors.

Coreceptor usage assay.

SHIV-KNH1144 and SHIV-KNH1144p4 were tested for usage of various coreceptors (CCR1, CCR2, CCR3, CCR5, CXCR4, GPR15/BOB, and CXCR6/BONZO/STRL33) expressed on either engineered U87.CD4 or Ghost cells (NIH AIDS Reagent Program). Cells were seeded at 1 × 10⁵ cells/well in a 12-well tissue culture plate (Greiner Bio-one) and 24 h later were infected with the test or control viruses with 1 ng/ml p24 or 10 ng/ml p27. The following day, cells were washed and incubated for 7 to 10 days. Supernatants were collected at various time points and tested for p24/p27 levels using commercially available kits (p24/p27 antigen capture assay; ABL). SHIV-KNH1144p1 and SHIV-KNH1144p4 were also tested for CXCR4 tropism using CEMx174-GFP cells; 5 × 10⁵ cells/well in a 12-well tissue culture plate were exposed to equal amounts of test and control viruses and incubated for 6 days. Cells were checked for GFP expression under a fluorescence microscope.

Neutralization assays.

The neutralization sensitivity for SHIV-KNH1144, SHIV-KNH1144p1, and SHIV-KNH1144p4 was tested with the TZM-bl cell-based assay (46). Briefly, antibody/serum was serially diluted in a 96-well plate, followed by incubation with virus for 1 h. The virus-antibody mixture was then incubated with 10⁴ freshly trypsinized TZM-bl cells per well containing 15 μg/ml DEAE-dextran for 48 h before lysis with Bright-Glo reagent. Luminescence was read immediately on a luminometer.

Deep sequencing and analysis of SHIV-KNH1144p1 and SHIV-KNH1144p4 isolates.

Cell culture supernatant containing virus was centrifuged at 28,000 rpm (SW28.1 rotor; Beckman) for 2 h at 4°C on a 20% sucrose cushion; the pellet was suspended in 1% of the original volume in phosphate-buffered saline (pH 7.2). Genomic vRNA was extracted using the QIAamp viral RNA minikit (Qiagen) as recommended by the manufacturer; an on-column RNase-free DNase (Qiagen) treatment step was included, and the carrier RNA provided with the kit was replaced with linear polyacrylamide (5 μg/ml; Life Technologies). While we would have preferred to isolate vRNA also from the corresponding plasma samples from animal RHf-12 (for SHIV-KNH1144p1) and animal RZY-11 (for SHIV-KNH1144p4), such samples were lost due to our major move across the United States. Therefore, only biological isolates expanded briefly in RM PBMC cultures were available for deep sequencing for these two SHIV-A strains.

Afterwards, ∼500 ng vRNA was used for sequencing library preparation by following the Illumina TruSeq RNA sample preparation guide with modifications. The first step in the workflow involved total RNA fragmentation directly. Total RNA was fragmented into small pieces using divalent cations under elevated temperature. The cleaved RNA fragments were copied into first-strand cDNA using reverse transcriptase and random primers. This was followed by second-strand cDNA synthesis using DNA polymerase I and RNase H. These cDNA fragments then went through an end repair process, the addition of a single “A” base, and then ligation of the adapters. The products were then purified and enriched with PCR to create the final sequencing library. The libraries were subjected to a quantification process and pooled for cBot amplification and a subsequent sequencing run with the Illumina HiSeq 2000 platform. After the sequencing run, demultiplexing with CASAVA (v1.8.2; Illumina Inc.) was employed to generate the fastq file for each sample. All sequencing reads were aligned with the reference genome of parental SHIV-KNH1144 using the Burrows-Wheeler aligner (BWA) (47). The genome coverage for each sample was determined by using BEDTools (47, 48). The program VarScan (49, 50) was used to identify single-nucleotide variants present at 1%.

RNA isolation and deep sequencing of SHIV-KNH1144p2.

Plasma samples from RM RTp-9 were used to isolate SHIV-KNH1144p2 genomic RNA. Plasma was thawed, transferred to ultracentrifuge tubes on top of a 2-ml 20% sucrose cushion, and centrifuged (SW28 rotor; Beckman Coulter) (4 h at 24,000 rpm at 4°C). The supernatant was removed and the sample resuspended in 1 ml phosphate-buffered saline before vRNA was purified using the QIAamp viral RNA minikit (Qiagen). Contaminating DNA was digested with the RNase-free DNase set (Qiagen). RNA samples were prepared for sequencing as previously described (51). Briefly, DNA, rRNA, and mRNA were removed from harvested nucleic acids by using a Turbo DNA-free kit (Life Technologies), a Ribo-Zero magnetic gold kit (Epicenter Biotechnologies), and RNA purification beads (Illumina), respectively. The remaining material was cleaned and concentrated by using an RNA Clean and Concentrator-5 kit (Zymo). Sequencing libraries were generated by using the Illumina TruSeq total RNA sample preparation kit, according to the manufacturer’s instructions. The library was then sequenced by applying sequencing by synthesis (Illumina) using the 150-bp paired-end format on an Illumina MiSeq system. Initial data analysis occurred by using the Illumina pipeline to generate a FASTQ file containing all the reads. This was then mapped to the parental virus sequence by using Lasergene Seq-Man NGen. Quality trimming to mers was performed on reads with a minimum similarity of 93%. Single nucleotide polymorphisms (SNPs) were quantified by determining the number of reads at each position and comparing the variability to the reference sequence. This permits the expression of SNP abundance as a percentage of total reads.

Data availability.

The sequences determined in this study are available in GenBank under the following accession numbers: for SHIV-KNH1144p1, MK573511; for SHIV-KNH1144p2, MK573512; and for SHIV-KNH1144p4, MK573513.