|Home | About | Journals | Submit | Contact Us | Français|
Unlike human immunodeficiency virus type 1 (HIV-1)-infected humans, African-origin, natural simian immunodeficiency virus (SIV) hosts, such as African green monkeys (AGMs), sustain nonpathogenic SIV infections and rarely vertically transmit SIV to their infants. Interestingly, chronically SIV-infected AGMs have anatomically compartmentalized SIV variants in plasma and milk, whereas humans and SIV-infected rhesus monkeys (RMs), Asian-origin nonnatural SIV hosts, do not exhibit this compartmentalization. Thus, it is possible that AGM SIV populations in milk have unique phenotypic features that contribute to the low postnatal transmission rates observed in this natural host species. In this study, we explored this possibility by characterizing the infectivity, tropism, and neutralization susceptibility of plasma and milk SIVsab env variants isolated from chronically SIVsab92018ivTF-infected AGMs. AGM plasma and milk SIVsab env pseudovirus variants exhibited similar infectivities, neutralization susceptibilities to autologous and heterologous plasma, and chemokine coreceptor usages for cell entry, suggesting similar abilities to initiate infection in a new host. We also assessed the cytokine milieu in SIV-infected AGM milk and compared it to that of SIV-infected RMs. MIP-1β, granulocyte colony-stimulating factor (G-CSF), interleukin-12/23 (IL-12/23), and IL-13 trended significantly higher in SIV-infected AGM milk than in that of RMs, while IL-18 and IL-6 trended significantly higher in SIV-infected RM milk than in that of AGMs. Taken together, our findings imply that nonviral maternal factors, such as the cytokine milieu, rather than unique characteristics of SIV populations in the milk contribute to the low postnatal transmission rates observed in AGMs.
IMPORTANCE Due to the ongoing global incidence of pediatric HIV-1 infections, including many that occur via breastfeeding, development of effective vaccine strategies capable of preventing vertical HIV transmission through breastfeeding remains an important goal. Unlike HIV-1-infected humans, African green monkeys (AGMs), the natural SIV host species, sustain nonpathogenic SIV infections, rarely transmit the virus postnatally to their infants, and exhibit anatomically compartmentalized SIV populations in milk and plasma. Identifying unique features of the anatomically compartmentalized milk SIV populations could enhance our understanding of how AGMs may have evolved to avoid transmission through breastfeeding. While this study identified limited phenotypic distinctions between AGM plasma and milk SIV populations, potential differences in milk cytokine profiles of natural and nonnatural SIV hosts were observed. These findings imply the potential importance of nonviral factors in natural SIV host species, such as innate SIV/HIV immune factors in milk, as a means of naturally preventing vertical transmission.
Now 4 decades into the human immunodeficiency virus type 1 (HIV-1) epidemic, more than 200,000 pediatric HIV-1 infections continue to occur annually despite considerable gains in antiretroviral (ARV)-based prevention strategies (1, 2). Nearly half of these ongoing pediatric transmission events occur postnatally via breastfeeding. The ARV-based approach to prevention of postnatal transmission falls short of preventing all infant infections due to challenges such as adherence, access, viral resistance, and acute maternal infection during lactation. Thus, preventing this route of vertical HIV-1 transmission through an efficient pediatric vaccine regimen remains essential for achieving an HIV-free generation.
Unlike HIV-1-infected humans, natural simian immunodeficiency virus (SIV) host species such as African green monkeys (AGMs) sustain nonpathogenic SIV infection and rarely transmit SIV vertically to infants (3,–7), likely as a result of >30,000 years of host-pathogen coevolution (8). This lack of pathogenic infection and vertical transmission is not due to complete control of viremia, as chronic viral replication populates both the systemic and breast milk compartments in natural hosts (9,–13). Several findings suggest that both maternal and infant immunologic features contribute to the very low postnatal transmission rates in natural SIV hosts. We previously demonstrated that lactating, SIV-infected AGMs exhibit less B cell dysfunction and more functional antibody responses in both plasma and milk than do rhesus monkeys (RMs), a nonnatural SIV host with vertical transmission rates similar to those for HIV-infected humans (12, 14). Moreover, the development of a neutralizing antibody response can arise early after acute infection in AGMs (15). This neutralizing response within the mammary compartment could contribute to the low vertical transmission rates in breastfeeding AGMs. Alternatively, studies have also demonstrated that infant AGMs (16) and sooty mangabeys (17), another natural SIV host species, exhibit low CCR5 coreceptor expression on CD4+ target cells, which could decrease the infant's risk of SIV acquisition.
In addition to characterizing maternal and infant immunologic features unique to natural SIV host species, genotypic analyses of the viral populations in plasma and milk demonstrated that, unlike HIV-1-infected humans (18) and SIV-infected RMs (19), SIV-infected AGMs exhibit anatomically compartmentalized SIV variant populations in the milk and plasma (9). While systemic SIV variants of natural hosts can be highly infectious and resistant to neutralization (20), breast milk SIV variants have not previously been characterized. Thus, the potential existence of a phenotypically distinct milk SIV population could contribute to the rarity of postnatal transmission in AGMs. In this study, we sought to characterize the infectivity, neutralization, and tropism phenotype of milk and plasma SIV sabaeus (SIVsab) variants from chronically infected AGMs in an attempt to elucidate unique features of these genotypically distinct viral populations that could contribute to genetic selection of SIVsab variants in the milk compartment or the low vertical transmission rates observed in AGMs. Identifying key features of nontransmitted viral populations to which an infant is chronically exposed may inform pediatric HIV-1 vaccine strategies to eliminate vertical transmission through breastfeeding.
Six female AGMs and 4 female RMs were hormonally induced into lactation, followed by intravenous inoculation with SIVsab92018ivTF or SIVmac251, respectively, as described previously (14, 21). Blood and milk were collected at acute and chronic time points up to 2 years postinfection. Peripheral blood mononuclear cells (PBMCs) and plasma were isolated by density gradient centrifugation using Ficoll-Plaque (GE Healthcare, Waukesha, WI). Milk of 5 uninfected, hormonally induced lactating AGMs and 6 additional hormonally induced lactating SIVsab9351BR-infected AGMs (22), as well as 2 naturally lactating and 10 hormonally induced lactating, uninfected RMs and 1 additional chronically SIVmac251-infected, naturally lactating RM, was employed to characterize the cytokine milieu in chronically infected and uninfected AGMs and RMs (21). Animals were maintained according to the Guide for the Care and Use of Laboratory Animals (23).
SIVsab plasma and milk env variants from 5 AGMs at 5 months after SIVsab92018ivTF infection were amplified and sequenced using single-genome amplification (SGA), as previously reported (9), and were found to be anatomically compartmentalized in 4 of 5 AGMs. For this study, isolates from the 4 AGMs with statistically significant genetic compartmentalization of milk and plasma virus variants were selected for plasma and milk env cloning and pseudovirus production. Successfully cloned and functional SIVsab env pseudoviruses included 13 plasma variants and 10 milk variants. These env genes were amplified directly from the initial SGA product using PCR and cloned into the pcDNA3.1D Topo mammalian expression vector (Invitrogen, Waltham, MA) using standard molecular biology techniques. Cloning was considered successful if the ligated env sequence did not include nonsynonymous mutations compared to the SGA sequencing results used to initially assess viral variant compartmentalization (9). In MEGA 6 (24), SIVsab variants were aligned by codon using MUSCLE (25), and maximum-likelihood trees were constructed using the general time-reversible substitution model with fixed gamma-distributed rate variation across sites.
SIVsab env variant clones were cotransfected in 293T cells using Jetprime transfection reagent (Polyplus Transfection, New York, NY) according to the manufacturer's instruction. Both SG3Δenv (26, 27) and NL4.3_lucΔenv backbones (supplied by David Montefiori, Duke University) were employed in the cotransfection to produce functional, non-replication-competent pseudoviruses, as previously described (28). After a 48-hour incubation at 37°C, culture supernatant was collected, sterile filtered, and stored at −80°C. SIVsab variant pseudovirus functionality was determined by assessing infectivity as the 50% tissue culture infectious dose (TCID50) in TZM-bl reporter cells. Other viruses used as controls throughout the study and prepared in the same manner included pseudoviruses Du156.12 env, Mn.3 env, 89.6 env, and SIVmac251.30 env, as well as the replication competent viruses SIVsab92018ivTF (HQ378594.1), SIVmac239 IMC (M33262), and pNL-LucR.T2A-SIVagm.sab92018.ps.
For certain experiments, replication competent reporter viruses encoding SIV Env and Renilla luciferase (LucR) were utilized. pNL-LucR.T2A-SIVagm.sab92018.ps was constructed by modifying the previously described pNL-LucR.T2A-Env.ecto HIV-1 reporter virus approach (29, 30) such that SIV env genes of interest could be expressed in an established robust, isogenic reporter virus backbone. It has been previously reported that SIV Env cytoplasmic tail truncations to 22 or 23 amino acids (aa) are selected for by replication of SIVmac in human cells, including PBMCs (31). We took advantage of this by inserting only the SIV env coding region encompassing the entire ectodomain, membrane-spanning domain, and the first 22 aa of the cytoplasmic tail into the NL-LucR.T2A proviral backbone; by introducing the premature stop codon at this position, the SIV sequence ends right before the SIV splice acceptor site for exon 2 of tat and rev. This region also contains the SIV Rev response element (RRE), which has been found to work sufficiently well in concert with HIV-1 Rev (32). In our pSP72mss HIV-1 env cloning intermediate, which contains a silent mutation BstBI restriction site in the membrane-spanning domain-coding region (nucleotides [nt] 8301 to 8306) (29), we mutated the HIV env start codon and 2 additional methionine codons up to the KpnI site at NL4-3 nucleotide 6343 (located just 3′ of the end of the vpu/env open reading frame [ORF] overlap) to create pSP72mssMet-minus. We PCR amplified the env region of interest from SIVagm.sab92018 TF (HQ378594.1), introducing a 5′ KpnI site and a 3′ BstBI site, removing an internal KpnI site, and introducing the premature env stop codon (“ps,” at the position corresponding to SIVsab92018ivTF Env aa 728). The respective final PCR fragments were ligated into pSP72mssMet-minus via KpnI/BstBI. This essentially replaced the HIV env sequence between the overlapping vpu ORF and the splice acceptor sites for exon 2 of tat/rev with that of SIVagm.sab92018.ps. The HIV/SIV chimeric region was ligated back into the NL-LucR.T2A-mssD proviral backbone via the EcoRI and BamHI sites in HIV NL4-3, to generate the SIV.env-IMC-LucR reporter proviral plasmid, pNL-LucR.T2A-SIVagm.sab92018.ps. Virus stocks were prepared by transfection of individual proviral DNA into 293T cells, and titers were determined on TZM-bl cells essentially as described above.
SIVsab env/SG3Δenv variant infectivity in TZM-bl cells was used to estimate viral infectivity. The TCID50 of viruses was assessed in 96-well cell culture plates using Tat-regulated firefly luciferase (LucF) reporter gene expression to quantify viral entry into various cell lines. Four replicate infectivity assays of each virus were performed, and the Reed-Muench method was used to calculate TCID50 as previously described (28). Viruses were serially diluted onto cells and allowed to infect for 48 h at 37°C prior to luminescence quantification (Promega, Madison, WI). The minimum detection limit for infection was defined as 2.5 the average signal of the cell-only controls. Prior to comparing plasma and milk SIVsab variant infectivities, the TCID50 was normalized to SIV Gag p24 concentration quantitated by enzyme-linked immunosorbent assay (ELISA) (PerkinElmer, Waltham, MA) as an estimate of total viral particles to account for variations intrinsic to the transfection protocol. To additionally assess viral infectivity in cells without CCR5 overexpression such as that seen in TZM-bl cells, SIVsab92018ivTF infectivity of M7-luc cells (supplied by David Montefiori, Duke University) was also measured (33) and compared to SIVsabivTF infectivity of TZM-bl cells. Variations of the human glioma cell line NP-2/CD4 with specific chemokine coreceptor expression of CCR5, CXCR4, or GPR15, or with no chemokine receptor (provided by Feng Gao) (34), as well as HOS cell-derived Ghost(3) cells expressing CD4, CCR5, and CXCR4 (obtained from the NIH AIDS Reagent Program) (35), were used to assess SIVsab/NL4.3_lucΔenv pseudovirus infectivity in the setting of these specific chemokine receptors. NP-2 and Ghost(3) cell line chemokine receptor expression was verified through the use of viral controls, including Du156.12/NL4.3_lucΔenv (CCR5 tropism) (36), Mn.3/NL4.3_lucΔenv (CXCR4 tropism), 89.6/NL4.3_lucΔenv (CCR5, CXCR4, and CCR3 tropisms) (37, 38), YU2 (CCR5 tropism), and YU2-6248wt (CCR5 and GPR15 tropism) (39). HIV Gag p24 quantification by ELISA was used to measure infectivity of YU2 and YU2-6248wt in the GPR15-expressing NP-2 cell line due to the lack of luciferase reporter gene expression for these viruses. Of note, infectivity comparisons between cell lines were not conducted due to inconsistent chemokine receptor expression levels between the cell lines.
Neutralization of SIVsab viruses by plasma and milk in TZM-bl and M7-luc cells was measured in 96-well cell culture plates using Tat-regulated LucF reporter gene expression to quantify an antibody-induced decrease in viral entry, as previously described (28). Briefly, heat-inactivated samples (56°C; 1 h) were serially diluted in duplicate and incubated with virus (to achieve ~50,000 relative light units [RLU] of LucF) for 1 h at 37°C to allow blocking. TZM-bl cells (1 × 104/well) were added and incubated for 48 h at 37°C prior to luminescence quantification with the Brightglow luciferase detection system (Promega, Madison, WI). Neutralization (reported as the median infectious dose [ID50]) was then calculated by comparisons to virus-only controls. Of note, unlike with TZM-bl cells, M7-luc cells required replication-competent virus to observe a detectable signal, and therefore, neutralization assays on M7-luc cells were performed only with the replication-competent virus SIVsab92018ivTF.
The ability of AGM plasma to neutralize SIVsab92018ivTF Env expressing SIV.env-IMC-LucR in healthy donor PBMCs was measured in 96-well cell culture plates using a previously described protocol with slight modifications (40). Briefly, human PBMCs from a healthy donor were activated with phytohemagglutinin P (PHA-P) (Sigma, St. Louis, MO) and interleukin-2 (IL-2) (ABL, Inc., Rockville, MD) for 24 to 72 h. Heat-inactivated samples (56°C; 1 h) were serially diluted in duplicate in a round-bottom 96-well plate and then incubated with the replication-competent, luciferase-expressing pNL-LucR.T2A-SIVagm.sab92018.ps. The virus input was adjusted to achieve >10× the RLU of LucR in virus control compared to uninfected cell control backgrounds. After 1 h of incubation at 37°C, activated PBMCs (3.5 × 105/well) were added, and plates were spinoculated at 25°C and 1,200 × g for 2 h. The plates were incubated at 37°C for 12 h, and then the cells were washed completely three times with fresh, IL-2-containing medium. After a total of 5 days of incubation at 37°C, the antibody-induced decrease in viral entry was quantified. The Viviren live cell substrate (Promega, Madison, WI) was used to detect Tat-regulated Renilla luciferase (LucR) reporter gene expression according to the manufacturer's instructions. Neutralization (reported as the median infectious dose [ID50]) was then calculated by comparisons to virus-only controls.
The chemokine receptor antagonist susceptibility of viruses was determined in 96-well cell culture plates by using Tat-regulated Luc reporter gene expression to quantify a chemokine receptor antagonist-induced decrease in viral entry, as previously described (41). TZM-bl cells (1 × 104/well) were incubated with 10 μM TAK-779 (CCR5 antagonist) or 1.3 μM AMD-3100 (CXCR4 antagonist) for 1 h at 37°C to allow chemokine receptor antagonism. Viruses were then added (to achieve ~50,000 RLU of LucF) and allowed to infect for 48 h at 37°C prior to luminescence quantification with the Brightglow luciferase detection system (Promega, Madison, WI). The degree of chemokine receptor antagonist susceptibility was defined as percent neutralization, which was calculated by comparison to virus-only controls. Additionally, HIV-1 pseudoviruses Du156.12/SG3 and Mn.3/SG3 functioned as positive controls for susceptibility to CCR5 and CXCR4 antagonism, respectively. The cutoff for detectable neutralization for each SIVsab variant was defined as 3 standard deviations above the value for the negative HIV-1 pseudovirus control for each antagonist (Du156.12/SG3 for AMD-3100 and Mn.3/SG3 for TAK-779).
Chemokine receptor antagonist titration assays with the CCR5 antagonist maraviroc (42) were also conducted. Antagonists were serially diluted and incubated with TZM-bl cells (1 × 104/well) for 1 h at 37°C. Viruses were added (to achieve ~50,000 RLU of LucF) and incubated for 48 h prior to luminescence quantification with the Brightglow luciferase detection system (Promega, Madison, WI). Chemokine receptor sensitivity was calculated as the half-maximal inhibitory concentration (IC50) through comparison to cell- and virus-only controls. AMD-3100 proved toxic to TZM-bl cells prior to achieving 50% neutralization.
The effects of type 1 interferons (IFNs) on SIV infection and/or replication in human PBMCs were determined in 96-well cell culture plates by using Gag p27 expression monitoring after IFN treatment. Human PBMCs from 2 healthy donors were CD8+ T cell depleted with magnetic beads (Miltenyi Biotec Inc., San Diego, CA) and activated with PHA-P (Sigma, St. Louis, MO) and IL-2 (ABL, Inc., Rockville, MD) for 24 to 72 h. Activated PBMCs (2 × 105/well) were incubated with serial dilutions of recombinant human IFN-αII (PBL, Piscataway, NJ) or IFN-β (Peprotech, Rocky Hill, NJ) at 37°C for 4 h. Replication-competent infectious molecular clones (IMCs) SIVsab92018ivTF and SIVmac239 IMC were then spinoculated on the cells at 25°C and 1,200 × g for 2 h, followed by 3 complete washes and the replacement of medium containing the appropriate concentrations of IFN. Assays were set at 37°C, and supernatant was collected at days 4 and 8 postinfection. After supernatant collection on day 4, the medium was partially replaced with the full appropriate IFN doses for each well. Virus production was quantified through p27 ELISA (Zeptometrix, Buffalo, NY) of day 0 (postwash) and day 8 supernatants. The IFN-induced, dose-dependent inhibition of viral production was calculated as IC50 through comparison to cell- and virus-only controls. IFN-α concentrations were converted to units of pg/ml using the manufacturer-supplied specific activity to report units consistent with the IFN-β.
Twenty-three cytokines/chemokines were measured using the nonhuman primate cytokine Milliplex kit (Millipore, Billerica, MA) according to the manufacturer's instructions. Groups of nonhuman primates included 5 uninfected AGMs, 6 chronically SIVsab9351BR-infected AGMs, 5 SIVmac251-infected RMs, and 12 uninfected RMs. The analytes measured were IL-1β, IL-1RA, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12/23 (p40), IL-13, IL-15, IL-17, IL-18, IFN-γ, granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), monocyte chemoattractant protein 1 (MCP-1), MIP-1α, MIP-1β, tumor necrosis factor alpha (TNF-α), transforming growth factor α (TGF-α), soluble CD40L (sCD40L), and vascular endothelial growth factor (VEGF). The analyte detection ranged from 0.64 to 10,000 pg/ml. Plasma or breast milk supernatant was incubated with antibody-coupled beads overnight, followed by incubation with biotinylated detection antibody and, finally, incubation with streptavidin-phycoerythrin (PE). Each sample, cytokine standards, and controls supplied by the manufacturer were run in duplicate on each plate. Multianalyte profiling was performed using a Luminex-100 system, and data were analyzed using BioPlex manager software (Bio-Rad, Hercules, CA). Analytes determined to be out of range by BioPlex were assigned values of half the difference between the lowest standard and zero. AGM IFN-α and RM IFN-α were measured with a rhesus ELISA kit from PBL (Piscataway, NJ). AGM IFN-β and RM IFN-β were measured with a rhesus ELISA kit from USCN (Wuhan, China). Plasma and milk samples were run in duplicate, and concentrations were calculated from the standard curves obtained with standards provided by the manufacturers.
A 384-well plate was coated with 2 μg/ml of anti-tenascin C (anti-TNC) rabbit polyclonal IgG (Santa Cruz Biotechnology, Dallas, TX) and then blocked using 7.5% bovine serum albumin (BSA) (Gibco, USA) at room temperature for 1 h. AGM and RM milk samples were delipidated via centrifugation at 21,000 × g and diluted in 7.5% BSA prior to room temperature incubation for 1 h on the blocked plates. Purified TNC (Millipore, Billerica, MA) was used as a standard ranging from 5 μg/ml to 5 ng/ml. All samples and standards were measured in triplicate. TNC binding was detected using 1 μg/ml of an anti-TNC mouse monoclonal antibody (T2H5; Fisher, USA) for 1 h at room temperature, followed by a 1:10,000 dilution of goat anti-mouse horseradish peroxidase (HRP)-conjugated antibody (Promega, Madison, WI) for 1 h at room temperature. A 5-min room temperature incubation with SureBlue Reserve tetramethylbenzidine (TMB) substrate (KPL, Gaithersburg, MD), followed by addition of TMB stop solution (KPL, Gaithersburg, MD), was used to detect the HRP, and the plate was read at 450 nm.
Plasma and milk SIVsab variant infectivity, CCR5 antagonist susceptibility, and chemokine receptor usage measures were compared using the Wilcoxon rank sum test. These results were also compared using mixed-effects models, which accounted for host clustering by assuming host-level random intercepts (results not shown). Neutralization sensitivity measures were dichotomized (nonneutralized [ID50 of <20] versus neutralized [ID50 of ≥20]) and assessed for the ability to predict compartment association in a mixed-effects logistic regression model with host random intercepts to account for host clustering. Plasma collection time at 1 or 2 years after infection was added as a covariate in the model. Fisher's exact tests were used to compare the proportions of plasma and milk SIVsab env variants that were neutralized by plasma or the CXCR4 antagonist AMD-3100. Neutralization measures in TZM-bl cells and M7-luc cells were compared using Spearman's rank test. For all analyses, differences between compartments were considered statistically significant at P value of <0.05, and all tests were two sided. These analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC). Cytokine levels in AGM and RM milk were compared in GraphPad Prism version 6 (GraphPad Software, La Jolla, CA) using Mann-Whitney tests for plate-based ELISA data and Wilcoxon rank sum tests with false discovery rate (FDR) corrections to account for multiple comparisons for multiplex ELISA data.
We previously reported anatomic compartmentalization of SIVsab env variants in the milk and plasma of four of five chronically SIV-infected AGMs (9). For this study, we focused on cloning diverse milk and plasma SIVsab env variants from the four AGMs that demonstrated anatomic compartmentalization, as our goal was to phenotypically characterize the genetically distinct milk and plasma SIV populations. A maximum-likelihood tree of all SIVsab env variants isolated from 5 months postinfection (Fig. 1A) also demonstrated the expected viral variant segregation between animals, in addition to the reported anatomic compartmentalization of milk and plasma SIVsab env variants. Of the 4 AGMs in this study, SIVsab env variants from AGM 89 were the most similar to the SIVsab92018ivTF challenge virus and also demonstrated large groups of identical viruses in milk and plasma. In contrast, SIVsab env variant populations isolated from AGMs 91 and 94 consisted of highly diverse variants. AGM 92 SIVsab env variants were more distinct from SIVsab92018ivTF than those isolated from AGM 89 but similarly demonstrated several clusters of identical variants. Thus, we analyzed phenotypic data from this set of cloned SIVsab env variants (Fig. 1B) using Wilcoxon rank sum tests, but we confirmed these results with mixed-effects models to account for these differences in SIVsab env evolution between monkeys.
We first assessed the infectivities of plasma and milk SIVsab variants to determine the contribution of infectivity to both the anatomic compartmentalization of milk and plasma SIV variants in AGMs and the low postnatal transmission rate observed in this natural SIV host species. The infectivities of plasma and milk SIVsab variants (quantitated as TCID50/ng of Gag p24 in TZM-bl reporter cells) were not significantly different between compartments (Wilcoxon test; P = 0.10) (Fig. 2A). However, TZM-bl cells are engineered to overexpress CCR5, whereas natural SIV host species have been shown to express low levels of CCR5, particularly during infancy (16, 17). Thus, we sought to determine whether infectivity in the TZM-bl reporter cell line reliably predicted infectivity of SIVsab variants in a cell line lacking CCR5 overexpression, M7-luc cells (33). Unlike the TZM-bl reporter cell line, which is suitable to assess pseudovirus infectivity, the M7-luc cell line requires replication-competent viruses to observe detectable infection. Therefore, the replication-competent challenge virus SIVsab92018ivTF was employed to compare infectivity and neutralization sensitivity between the TZM-bl and M7-luc cell lines. The neutralization sensitivity of SIVsab92018ivTF to plasma of chronically SIV-infected RMs and AGMs and milk of chronically SIV-infected AGMs in TZM-bl cells strongly correlated with that in M7-luc cells (Spearman's rank test; r = 0.95, P < 0.0001) (Fig. 2B). To further strengthen our analysis, we extended it to include primary cells. The neutralization sensitivity of the replication-competent, luciferase-expressing pNL-LucR.T2A-SIVagm.sab92018.ps was assessed against plasma of chronically infected AGMs in both healthy human PBMCs and TZM-bl cells and was shown to correlate strongly, albeit nonsignificantly (Spearman's rank test; r = 0.83, P = 0.06), between the two assay formats. Furthermore, the infectivities of SIVsab92018ivTF were comparable in the TZM-bl and M7-luc cell lines (TZM-bl TCID50/ml = 69,877; M7-luc TCID50/ml = 72,407) (Fig. 2C). Based on these comparisons, the measured infectivity of SIVsab plasma and milk pseudovirus variants in TZM-bl cells likely represents that measured in a cell line that does not overexpress CCR5, such as the M7-luc cell line.
To determine if neutralization sensitivity differs between plasma and milk SIVsab variants and accounts for the anatomic compartmentalization, we assessed SIVsab env variant neutralization sensitivity to autologous and heterologous plasma of SIV-infected AGMs and RMs. No difference in the kinetics or magnitude of neutralization sensitivity to autologous plasma was observed between plasma and milk SIVsab env variants across multiple time points (preinfection and 4 months, 1 year, and 2 years postinfection) (Fig. 3). While plasma neutralization of the challenge virus strain was detected by 6 weeks in SIV-infected AGMs (15), neutralization of the SIVsab env variants isolated at 5 months postinfection was first detected in plasma from 1 year postinfection, and plasma neutralization sensitivity was similar for most variants at 2 years postinfection (Fig. 3A). The magnitude of autologous neutralization titers against the phylogenetically compartmentalized variants did not predict SIVsab variant association with the plasma or milk compartments (dichotomized mixed-effects model; P = 0.21, odds ratio [OR] = 0.28 [0.04 to 2.12]). Additionally, autologous plasma from 1 and 2 years postinfection neutralized similar proportions of plasma (10/13) and milk (6/9) SIVsab variants (Fisher's exact test; P = 0.65) (Fig. 3B). The neutralization sensitivity of SIVsab plasma and milk env variants to heterologous plasma of chronically SIV-infected AGMs was also measured. One-year-postinfection plasma from AGM 89 was unable to neutralize SIVsab env variants from any other animal. Accordingly, SIVsab env variants isolated from AGM 89 were neutralized by most heterologous 1-year-postinfection plasma of other AGMs. Plasma from AGMs 91 and 94 collected at 1 year postinfection demonstrated strong neutralization against the majority of heterologous SIVsab env variants, whereas plasma from AGMs 92 and 93 collected at 1 year postinfection neutralized few heterologous SIVsab env variants, demonstrating variability in the host antibody and viral envelope evolution between AGMs. No detectable heterologous neutralization of SIVsab variants was mediated by plasma of SIVmac251-infected RMs at 2 years postinfection (ID50 of <20 for all variants).
Chemokine receptor usage by the SIVsab plasma and milk env variants isolated from SIV-infected AGMs was probed to elucidate potential mechanisms resulting in the anatomic compartmentalization of milk and plasma variants in this species, particularly in the setting of low CD4+ T cell number in the milk compartment (9). As most HIV and SIV variants predominantly utilize CCR5 or CXCR4 for viral entry, tropism for these coreceptors was initially explored for SIVsab variants using a chemokine receptor antagonist-mediated decrease in viral entry into TZM-bl cells. CCR5-tropic (Du.156.12) and CXCR4-tropic (Mn.3) viruses were employed as controls in these assays. CCR5 antagonism using TAK-779 resulted in inhibition of all SIVsab plasma and milk env variants (Fig. 4A). Surprisingly, CXCR4 antagonism using AMD-3100 also resulted in inhibition of a subset of plasma SIVsab env variants (4/13 plasma SIVsab variants) (Fig. 4B), as well as the challenge virus, SIVsab92018ivTF, above a conservative cutoff defined as 3 standard deviations above the value for the CCR5-tropic virus control. CCR5 antagonism with increasing concentrations of the CCR5 antagonist maraviroc further confirmed the similarities between SIVsab plasma and milk variant CCR5 antagonist susceptibility, reported as IC50 (Wilcoxon test; P = 0.64) (Fig. 4C).
We next assessed SIVsab variant infectivity in cell lines with distinct chemokine receptor expression of CCR5 and CXCR4, as well as GPR15, as this chemokine receptor has been reported to be utilized as a coreceptor by some SIV strains (20, 43,–46). Chemokine receptor expression was confirmed in the cell lines with virus controls of the CCR5-tropic Du.156.12, the CXCR4-tropic Mn.3, the CCR5- and CXCR4-tropic 89.6, the CCR5-tropic YU2, and the CCR5- and GPR15-tropic YU2-6248wt. In general, the parental cell line (CD4 only) was unable to be infected by SIVsab env variants or control viruses (Fig. 5A). As expected, all SIVsab env variants were able to infect cell lines expressing CCR5 and CCR5/CXCR4, yet all env variants were also able to infect cells expressing only CXCR4. The majority were able to infect cells expressing only GPR15 as well (22/23 env variants). The infectivity of the SIVsab plasma and milk variants normalized to SIV Gag p24 (TCID50/ng Gag p24) was not significantly different in any of these infected cell types (Fig. 5B to toE).E). Virus infectivities across cell lines were not compared due to potential variations in chemokine receptor expression.
Given the lack of phenotypic differences between SIVsab plasma and milk env variants, we next explored the breast milk immunologic milieu in lactating natural and nonnatural SIV hosts as a potential contributor to the observed anatomical compartmentalization and low vertical transmission rates in AGMs. HIV-1 transmitted/founder (T/F) variants have been characterized by their ability to resist inhibition by type 1 IFN responses (47), but the impact of type 1 IFNs on SIV transmission or replication efficiency has not been well characterized. Unlike humans and nonnatural SIV host species, natural SIV host species have been shown to largely contain their type 1 IFN responses following SIV infection (48), which may alter the impact of type 1 IFN on the early replication of SIV T/F variants. As SIVmac infection of RMs does not result in milk/plasma virus variant compartmentalization (19) and can result in postnatal transmission (49), this nonnatural SIV host species provides a SIV infection model to contrast with AGMs. Thus, we assessed the impact of IFN-α and IFN-β on SIVmac239 IMC compared to the SIVsab92018ivTF strain. IFN-α and IFN-β were both able to inhibit SIVsab92018ivTF and SIVmac239 IMC to similar degrees in human PBMCs from two uninfected donors (for IFN-α, SIVsab92018ivTF mean IC50 = 85.4 pg/ml and SIVmac239 IMC mean IC50 = 16.8 pg/ml; for IFN-β, SIVsab92018ivTF mean IC50 <3.7 pg/ml and SIVmac239 IMC mean IC50 = 7.0 pg/ml) (Fig. 6). Therefore, the susceptibilities of these SIVsab92018ivTF and SIVmac239 challenge strains to type 1 IFNs do not differentiate them to explain the resulting compartmentalization of viruses in milk of AGMs, but not RMs (19). As this approach to quantifying an IFN-induced decrease of viral replication required replication-competent IMCs, we were unable to characterize all SIVsab env variants based on type 1 IFN resistance.
Given the observed antiviral effect of type 1 IFNs on SIV infection and replication, concentrations of type 1 IFNs, other antiviral and inflammatory cytokines, and the innate antiviral protein tenascin C (TNC) were measured in AGM milk at 1 year postinfection and compared to those in of SIV-infected RMs (Table 1). IFN-α levels were undetectable in AGM and RM milk using our method, and IFN-β levels in milk were similar in these species. Interestingly, IL-6 and IL-18 levels were significantly lower in AGM milk than in that of RMs, whereas MIP-1-β, G-CSF, IL-12/23, and IL-13 levels were higher in AGM milk than in that of RMs. After correcting for multiple comparisons, none of these comparisons remained significant, yet all those listed above reached the minimum attainable FDR-corrected P value with this number of analytes tested (P = 0.1645). Surprisingly, TNC, which has been shown to possess some HIV-neutralizing activity (50), was found at appreciably higher concentrations in chronically infected RM milk (median [TNC] = 39,728 pg/ml) than in AGM milk (median [TNC] <5 pg/ml) (Mann-Whitney test, P = 0.01).
Finally, we quantified the cytokine milieu in milk from uninfected AGMs (n = 5) and compared it to that in uninfected RMs (n = 12) and observed similar differences in the cytokine milieu as in infected animals (Table 2). In milk from uninfected animals, G-CSF, IL-2, IL-13, IL-15, and MCP-1 concentrations were significantly higher in AGMs than in RMs, whereas IL-6 and IL-18 levels were higher in RMs than in AGMs. TNC concentrations were similar in uninfected RM milk (median [TNC] = 16,000pg/ml) and in uninfected AGM milk (median [TNC] = 4,100pg/ml) (Mann-Whitney test, P = 0.11). Interestingly, for both species, the cytokine milieu in the milk at 1 year postinfection was similar to that observed in uninfected animals (statistical comparison not shown). Regardless, the potentially distinct milieu of innate immune factors both preinfection and during chronic infection in the breast milk of AGMs versus RMs remains a possible contributor to the anatomic compartmentalization and low postnatal transmission risk in AGMs.
Identifying factors responsible for the rare postnatal vertical SIV transmission in natural SIV hosts remains an important pursuit due to the potential for translation to strategies that will prevent breast milk HIV-1 transmission in humans, which is responsible for a high proportion of ongoing pediatric HIV-1 infections. Our previous work identified anatomic compartmentalization of SIV populations in the milk and peripheral blood in AGMs. Conversely, HIV-infected humans and SIV-infected RMs do not demonstrate anatomically distinct viral populations in breast milk. Therefore, we sought to explore compartment- and virus-specific factors that could potentially contribute to the compartmentalization of viral variants in the mammary glands of AGMs and/or the low postnatal transmission rate in this species.
Initially, we compared phenotypic characteristics of the anatomically compartmentalized SIVsab env variants isolated from milk and plasma. We hypothesized that milk SIVsab env pseudovirus variants could have a generally lower infectivity or higher neutralization susceptibility than plasma SIVsab variants, resulting from compartment-specific antibody responses or other factors and leading to the low transmission rate through breastfeeding. As plasma and milk SIVsab env variants demonstrated similar infectivity in TZM-bl cells, this characteristic likely did not contribute to the anatomic compartmentalization or the low transmission rates observed in AGMs. This cell line allows the assessment of env pseudovirus infectivity but superficially expresses high levels of CCR5. Yet, infant AGMs are known to express low levels of this chemokine coreceptor (16). Thus, we then compared the infectivity of the challenge virus in TZM-bl cells to that measured in M7-luc cells, which express a more physiologic level of CCR5. The similar SIV infectivity measures in both cell lines support the validity of our assessment of SIVsab env pseudovirus variant infectivity in the TZM-bl cell line. We next assessed the neutralization sensitivity of milk and plasma SIVsab variants in AGMs. At 1 and 2 years after SIV infection, autologous plasma demonstrated similar neutralization potency against SIVsab plasma and milk variants. Interestingly, there was variability in the neutralization potency of heterologous plasma by different individual AGMs. One-year-postinfection plasma from AGM 89 was unable to neutralize heterologous SIVsab env variants, and AGM 89 SIVsab env variants were easily neutralized by most heterologous AGM plasma samples. Contrarily, plasma from AGMs 91 and 94 neutralized a large proportion of heterologous SIVsab env variants. This discrepancy may be partially due to disparate SIVsab env variant and antibody evolution rates in these animals. AGM 89 SIVsab env variants only minimally diverged genotypically from SIVsab92018ivTF, and the population consisted largely of multiple clonally amplified variants, whereas AGM 91 and AGM 94 SIVsab env variant populations appear to be farther removed from the wild-type challenge and demonstrated greater diversity. Regardless, the observed differences in heterologous neutralization susceptibility of SIVsab env variants remained independent of the SIVsab env variant anatomic compartment of origin. Taken together, these results failed to implicate either viral infectivity or neutralization sensitivity as a viral characteristic responsible for the selection of SIVsab variants into the mammary gland compartment or the low postnatal transmission rate observed in AGMs.
We previously demonstrated that the viral load in milk of AGMs is comparable to that in plasma despite an extremely low number of target CD4+ T cell population in milk (9, 12). Therefore, use of an alternative target cell population could contribute to virus replication and compartmentalization in the milk compartment. While none of the SIVsab env variants appeared to be CD4 independent, target cell chemokine receptor requirements could vary depending on the SIVsab variants chemokine coreceptor tropism. SIV strains have historically been reported to employ alternative coreceptors more commonly than HIV-1 strains (43,–45, 51, 52), which predominantly employ either CCR5 or CXCR4. The plasma and milk SIVsab variant chemokine receptor promiscuity was probed as a potential mechanism for selection of viruses that seed the milk compartment, which may in turn affect the ability of milk viruses to initiate infection in the infant. Initially, we antagonized CCR5 and CXCR4 in TZM-bl cells to approximate the dependence of SIVsab variants on each respective chemokine receptor. Notably, while all SIVsab variants were susceptible to CCR5 antagonism, several SIVsab plasma variants and the SIVsab92018ivTF challenge virus were also susceptible to CXCR4 antagonism, demonstrating some degree of chemokine receptor promiscuity. Cell lines expressing chemokine receptors CCR5, CXCR4, and GPR15 then were used to confirm SIVsab chemokine receptor promiscuity. This method demonstrated that certain plasma and milk SIVsab variants were capable of infecting cells expressing CCR5, CXCR4, or GPR15. Moreover, plasma and milk SIVsab variants demonstrated similar infectivity in these cell lines, suggesting that a disparate usage of select chemokine receptors by SIVsab variants is unlikely to contribute to the observed SIVsab variant compartmentalization. Several reports have demonstrated that low CCR5 expression on T cells in the infant gut of natural SIV hosts likely impacts vertical transmission rates (16, 17). Thus, the finding that these chronic SIVsab milk variants demonstrate appreciable tropism toward non-CCR5-expressing target cells suggests that other factors also contribute to the low postnatal transmission rates observed in natural SIV hosts. Additional work characterizing the full extent of SIVsab variant chemokine receptor promiscuity and infection efficiency and a more complete characterization of chemokine receptor expression levels on SIV-permissive cells in infant natural SIV hosts could further elucidate the extent to which low infant target cell availability contributes to the observed lack of vertical transmission in these species. Importantly, our characterization of SIVsab variants was limited, as we cloned only the env gene and not the entire replication-competent SIVsab genome. Therefore, our findings may fail to identify unique and impactful features of the anatomically compartmentalized SIVsab variants found in AGM milk in other regulatory or structural genes. Additionally, we were unable to characterize certain features of the SIVsab milk and plasma variants, such as susceptibility to type I interferons and other intracellular cytokines.
Given the lack of obvious phenotypic differences between milk and plasma SIVsab variants, we next assessed maternal innate factors in breast milk as potential contributors to the compartmentalization and low postnatal transmission rate in AGMs. The cytokine milieu after HIV-1 infection has been shown to play an important role in disease pathogenesis (53,–55), and recently, the potent antiviral effects of type 1 interferons on HIV-1 transmitter/founders and chronic variants have been well characterized (47). Yet, SIV infection of AGMs is characterized by controlling the systemic, chronic cytokine inflammation typical in HIV-infected humans, including the lack of sustained type 1 interferon levels in plasma (48). We identified that both SIVsab92018ivTF and SIVmac239 IMC infection productivities were susceptible to low levels of type 1 IFN. In fact, even though we were unable to detect IFN-α in AGM and RM milk, the IC50 for IFN-α-induced inhibition of SIVsab92018ivTF and SIVmac239 IMC infections was below or just above the limit of detection for our type 1 IFN quantification method. While no differences in type 1 IFN levels were detected in milk of either SIV-infected AGMs or RMs, several other cytokines tested demonstrated potentially differing levels in milk from the two species. The number of infected AGMs and RMs included in this study coupled with the large cytokine panel did not allow for significant comparisons between these groups that withstood multiple-comparison testing at an alpha level of 5%. Additionally, our characterization of the milk cytokine milieu in uninfected AGMs and RMs reveals differences between these species that are similar to those found in the infected animals. These early analyses suggest a potential importance of the cytokine milieu, and several cytokines in particular may contribute to the low postnatal transmission rate in AGMs. IL-6 has been reported to stimulate (56) and inhibit (57) viral production in monocytic cells, and increased IL-6 plasma levels have been correlated with ongoing viral production in HIV-infected humans (58). Therefore, the low levels of IL-6 found in AGM milk compared to RM milk may suggest that the absence of proviral IL-6 effects could contribute to the lower vertical transmission rates found in AGMs. Yet, this possibility requires further exploration given that milk SIV loads are similar in AGMs and RMs (12). Unexpectedly, we found higher levels of IL-13, which has been shown to stimulate HIV production in chronically dosed monocytes (59), in the milk of AGMs than in that of RMs. Interestingly, in milk from infected AGMs we observed potentially elevated levels of MIP-1β compared to those in milk from infected RMs. While the exact mechanisms of action remain uncertain (60), MIP-1β has been shown to inhibit HIV and SIV in vitro (61), and high levels of MIP-1β have been associated with asymptomatic HIV infection (62). It is possible that MIP-1β may contribute to the very low vertical transmission rates in AGMs by inhibiting the initial infection in AGM infants during breastfeeding.
Interestingly, even though AGMs and other natural SIV host species exhibit low vertical transmission rates through breastfeeding (4, 6), the innate antiviral protein TNC was found at appreciably higher levels in the milk of RMs than in that of AGMs. This suggests that TNC does not largely contribute to protection from vertical transmission in this natural SIV host model species. Of note, cytokine characterization was not species specific, which introduces potential variable sensitivity for species-specific cytokine quantification in AGMs. Future work both confirming cytokine levels in SIV-infected AGM and RM milk and functionally characterizing the anti-SIV impacts of cytokines found at distinct concentrations could yet elucidate innate contributors to the very low vertical transmission rates found in natural hosts. Additionally, this study mainly employed a chemically induced lactation model for both the RMs and AGMs (with the exception of 1 SIV-infected RM and 2 uninfected RMs), and results from a natural lactation model may vary. Furthermore, it remains unclear whether chemically induced lactating AGMs actually maintain the very low vertical transmission rates associated with natural SIV host species, which is an important factor influencing the validity of this model. Yet, in the context of RMs, artificial induction of lactation has been shown to yield milk with a similar cellular composition and heightened antibody levels compared to those in naturally lactating RMs, which are conditions unlikely to increase vertical transmission rates (21).
In conclusion, we did not identify viral factors contributing to the observed anatomic compartmentalization of SIVsab variants in milk of AGMs. It is possible that this compartmentalization may arise through stochastic seeding of SIVsab variants into the milk followed by clonal expansion of these variants through local replication. The low CD4+ T cell availability in the milk of AGMs coupled with viral loads comparable to those in the plasma indicates that local SIV replication in the milk may contribute to this clonal expansion. However, additional characterization of CD4+ cell populations in AGM milk as well as of the true extent of chemokine receptor promiscuity for SIVsab milk and plasma variants could elucidate selection pressures leading to compartmentalization. Furthermore, our identification of chemokine receptor promiscuity of chronic SIVsab variants, including a T/F variant, supports the possibility of distinct target cell populations mediating transmission in infant and adult natural SIV hosts (46, 63,–65). Additionally, it is possible, and has been suggested (66,–68), that cell-associated virus is the source of transmitted virus in postnatal transmission. The lack of differences in cell-free virus phenotype to explain the compartmentalization in AGMs and the established low target cell number could support the importance of cell-associated virus in postnatal HIV/SIV transmission. The possibility of cell-associated virus transmission has major implications for the ability of antiretroviral therapy (ART) to eliminate postnatal HIV transmission, as cell-associated DNA levels are reported not to change with maternal ART (69, 70). While this study ultimately found minimal differences between SIVsab env virus populations in the milk and plasma of AGMs that could contribute to the anatomic compartmentalization and/or the low vertical transmission rates in this species, identifying the key maternal, infant, or viral factors contributing to this unique natural protection in AGMs should remain a priority to inform non-ART-based strategies to eliminate pediatric HIV-1 transmission.
We thank Beatrice Hahn for generously providing the SIVsab92018ivTF infectious molecular clone used to infect the AGMs, David Montefiori for providing the M7-luc cell line, Fen Gao for providing technical expertise and the NP2/CD4 cell lines, and Vineet KewalRamani, Dan Littman, and the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH, for providing the Ghost(3) CXCR4+ CCR5+ cell line. TZM-bl cells and SG3Δenv were provided by John Kappes and Xiaoyun Wu through the NIH AIDS Reagent Program.
This work was also partially funded by the Center for HIV/AIDS Vaccine Immunology (CHAVI; AI0678501). Work by C.O. was supported through resources of the University of Alabama at Birmingham Center for AIDS Research (CFAR) Virology and Sequencing Cores (P30 AI27767) and a subcontract of the Comprehensive Antibody Vaccine Immune Monitoring Consortium (CA-VIMC) (grant 1032144), which is part of the Collaboration for AIDS Vaccine Discovery (CAVD)/CAVIMC, funded by the Bill & Melinda Gates Foundation.