T cell-tropic HIV efficiently infects alveolar macrophages through contact with infected CD4+ T cells

Alveolar macrophages (AMs) are critical for defense against airborne pathogens and AM dysfunction is thought to contribute to the increased burden of pulmonary infections observed in individuals living with HIV-1 (HIV). While HIV nucleic acids have been detected in AMs early in infection, circulating HIV during acute and chronic infection is usually CCR5 T cell-tropic (T-tropic) and enters macrophages inefficiently in vitro. The mechanism by which T-tropic viruses infect AMs remains unknown. We collected AMs by bronchoscopy performed in HIV-infected, antiretroviral therapy (ART)-naive and uninfected subjects. We found that viral constructs made with primary HIV envelope sequences isolated from both AMs and plasma were T-tropic and inefficiently infected macrophages. However, these isolates productively infected macrophages when co-cultured with HIV-infected CD4+ T cells. In addition, we provide evidence that T-tropic HIV is transmitted from infected CD4+ T cells to the AM cytosol. We conclude that AM-derived HIV isolates are T-tropic and can enter macrophages through contact with an infected CD4+ T cell, which results in productive infection of AMs. CD4+ T cell-dependent entry of HIV into AMs helps explain the presence of HIV in AMs despite inefficient cell-free infection, and may contribute to AM dysfunction in people living with HIV.


HIV can be detected in alveolar macrophages (AMs) from individuals with untreated HIV infection.
To determine levels of HIV nucleic acids in AMs from HIV-infected individuals, we performed bronchoscopies with bronchoalveolar lavage (BAL) in a cohort of HIV-infected ART naive and uninfected individuals living in Cape Town, South Africa (Table 1). All the participants in the cohort had evidence of immune sensitization to Mycobacterium tuberculosis but no indications of active disease. Cells from each bronchoscopy were adhered and non-adherent cells were removed; 99-100% of the remaining cells were AMs by microscopy (Supplementary Table 1). Using qPCR for HIV Gag, we detected Gag RNA in purified AMs from 4/11 (36.4%) participants and Gag DNA in AMs from 5/11 (45.5%) participants (Fig. 1A). While HIV RNA was not detected in all individuals, the BAL viral load was significantly higher in participants with detectable HIV RNA in the corresponding AM sample (Supplementary Fig. 1A). HIV RNA and DNA were both detected in samples from two participants, while two other participants had detectable HIV RNA only and three had detectable HIV DNA only. Overall, these findings were similar to prior reports using AMs from ART-naive participants in which HIV RNA and DNA was detectable in AMs from a median of 38.7% and 62.6% individuals, respectively, although the values varied widely across studies 18,20,21,40,[46][47][48][49][50] . These data demonstrate that a significant number of HIVinfected individuals harbor AMs with detectable HIV RNA and DNA. Table 1. Characteristics of participants in the Cape Town cohort. All participants were antiretroviral therapynaive and were sensitized to Mycobacterium tuberculosis as determined by QuantiFERON TB Gold test without clinical or radiographic evidence of active TB. BAL, bronchoalveolar lavage; ELF, epithelial lining fluid; AM, alveolar macrophage. Values are expressed as median with interquartile range. Statistical analysis was performed with Mann-Whitney test, except for smokers, where Fisher's exact test was performed. *p < 0.05 compared to healthy controls.

AM-and plasma-derived HIV is T-tropic.
In order to understand the entry properties of HIV derived from AMs, we isolated HIV env from lung and plasma samples. We sequenced HIV env RNA from AMs (27 isolates), BAL fluid (19 isolates), and plasma (21 isolates) from 3 HIV-infected ART-naive individuals (Supplementary Fig. 2). These sequences clustered by HIV-infected donor, but there was no compartmentalization when all AM-derived isolates were compared to all plasma-derived isolates. However, compartmentalization between AM and plasma derived sequences was observed in 2 of 3 donors ( Supplementary Fig. 2). To test the ability of  www.nature.com/scientificreports/ these AM-and plasma-derived Env proteins to allow infection of CD4+ T cells and MDMs, we cloned the env genes into an NL4-3 env deleted HIV backbone to generate replication-competent virus. We tested viruses containing Env from 8 AM and 9 plasma isolates selected to represent the sequence diversity found in the samples (arrows in Supplementary Fig. 2). Viruses containing all of the AM-and plasma-derived HIV Env isolates productively infected CD4+ T cells but did not infect MDMs ( Fig. 2 and Supplementary Fig. 3). These data indicate that all AM-and plasma-derived HIV primary isolates found in our study were T-tropic.

T-tropic HIV efficiently infects AMs through contact with infected CD4+ T cells. Previous work
has shown increased efficiency of M-tropic HIV entry into MDMs through uptake of or fusion with infected T cells 44,45 . However, whether this also occurs with T-tropic HIV, the primary virus during acute and chronic infection, and its relevance to primary AM infection remains unknown. To test T-tropic and M-tropic HIV entry into AMs via interaction with HIV-infected T cells, we infected PHA-activated CD4+ T cells with the CCR5-using T-tropic HIV strain JR-CSF. We then cultured HIV from these autologous infected CD4+ T cells with MDMs or AMs in the following conditions: (1) infected CD4+ T cells added directly to the macrophages ("cell-to-cell" or "CTC"); (2) infected CD4+ T cells separated from the macrophages by a transwell ("TW"); and (3) infected CD4+ T cell culture supernatant added to the macrophages ("SN") ( Fig. 3A). The cells or supernatant were added for 3 h before washing the macrophages. In MDMs and AMs, T-tropic HIV replication was significantly increased by contact with infected CD4+ T cells compared to the SN or TW conditions (Fig. 3B, C). Other reports have suggested that contact with T cells enhances M-tropic infection rates of MDMs 44,45 , but this has not been shown in AMs. To test this, we repeated the experiment with CD4+ T cells infected with the CCR5 and CXCR4-using M-tropic HIV strain 89.6. In MDMs, M-tropic infection was enhanced by contact with infected CD4+ T cells, similar to previous literature (Fig. 3D). However, in AMs, M-tropic HIV p24 levels were similar in the SN and CTC conditions by day 14 (Fig. 3E). In order to study whether productive HIV infection occurs in the T cell-macrophage co-cultures, we pre-treated the macrophages and T cells with the non-nucleoside reverse transcriptase inhibitor (NNRTI) efavirenz (EFV). We found that EFV pre-treatment inhibited T-tropic HIV replication by day 14 in the CTC condition in both MDMs and AMs ( Supplementary Fig. 4A, B). These data demonstrate that T-tropic HIV infects MDM and AM cultures more efficiently through contact with an infected

AM-derived HIV can productively infect MDMs more efficiently when there is contact with infected CD4+ T cells. Having established that infection of AMs and MDMs with a laboratory strain of
T-tropic HIV can be enhanced by contact with an infected CD4+ T cell, we tested whether infection with primary patient-derived AM-or plasma-derived HIV isolates could also be enhanced in macrophages through this same route. Using the experimental setup described above, we demonstrate that both AM-derived and plasma-derived primary isolates did not productively infect MDMs in the SN condition ( Fig. 4 and Supplementary Fig. 5), consistent with our observations of the T-tropic virus JR-CSF. On rare occasions, there was detectable p24 production in the cell-free virus (SN) condition. However, for all 3 AM-derived and 2 out of the 3 plasma-derived Envcontaining viruses, infection of MDMs was observed in the CTC condition ( Fig. 4 and Supplementary Fig. 5). These data indicate that primary HIV isolates made with Env derived from AMs infect MDMs inefficiently on their own, but can infect MDMs more efficiently via contact with an infected CD4+ T cell intermediate.

T cell-mediated infection of macrophages is mainly associated with CD4+ T cell contact.
To further characterize the cellular location of HIV in infected macrophages, we imaged T cell-macrophage cocultures using immunofluorescent confocal microscopy. As before, autologous CD4+ T cells were infected with T-tropic HIV, and T cell supernatant (SN) or HIV infected CD4+ T cells (CTC) were added to AMs. The samples were stained with anti-CD3 (T cells), anti-CD68 (macrophages) and anti-HIV Gag, and imaged after 14 days (Fig. 5A). Staining was quantified to localize HIV Gag. As expected, we detected Gag staining within AMs primarily in the CTC but not in the SN condition ( Fig. 5A and Supplementary Fig. 6A). The CD3 staining was limited to a discrete intracellular area, suggestive of an endocytosed CD3+ T cell. Gag+ CD3+ T cells that were    www.nature.com/scientificreports/ other T cells were in close contact with the AM, suggesting that T cell internalization was not required for HIV Gag staining in the AM cytosol. HIV Gag was rarely observed only within the cytosol of T cells internalized by AMs without HIV Gag staining also in the macrophage cytosol ("CD68+ CD3+ Gag+ in T cell only"; Fig. 5D and Supplementary Fig. 6B). Both the SN and CTC conditions had a similar percentage of Gag+ AMs with no T cell staining (Fig. 5D, Supplementary Fig. 6C), which may be the result of archiving of HIV in endosomal compartments 51 . The percentage of Gag+ AMs with no T cell staining was the same on day 0 after 3 h of incubation and on day 14, supporting the conclusion that this staining in cell-free infection is due to phagocytosis or archiving (data not shown). These findings show that T-tropic HIV can be found in the AM cytosol after coculture with infected T cells, suggesting that HIV is transmitted from an infected CD4+ T cell to the AM cytosol, and results in productive infection of AMs.

Discussion
While HIV nucleic acid has been detected in AMs during early stages of HIV infection, circulating HIV at this stage is primarily T-tropic and enters macrophages inefficiently in vitro. The tropism of HIV from AMs and the mechanism of HIV entry into AMs are both unknown, and may have a significant impact on the burden of pulmonary disease observed in those living with HIV. We studied AMs obtained by bronchoscopy in a cohort of ART-naive HIV-infected and uninfected individuals in Cape Town, South Africa and demonstrate that HIV RNA and DNA can be detected within AMs. We show that AM-derived HIV isolates are T-tropic and efficiently infect T cells but not macrophages; however, these isolates can productively infect macrophages through contact with infected CD4+ T cells. Our findings indicate that CD4+ T cell-dependent infection of AMs is a route of infection that may explain the presence of T-tropic HIV in AMs despite inefficient cell-free infection. This route of AM infection may be an important contributor to the burden of pulmonary disease observed in those living with HIV infection.
A few characteristics of the study population are notable, including sensitization to Mycobacterium tuberculosis and frequency of detection of HIV nucleic acids. The volunteers in the study were all sensitized to Mycobacterium tuberculosis, which could potentially increase the ability of AMs to be infected by HIV, including increased replication of HIV at sites of TB infection [52][53][54][55] . This reflects the high rates of latent tuberculosis infection seen in this population. However, the detection frequency of HIV Gag RNA and DNA were consistent with other studies, including those performed in non-TB sensitized populations. We observed HIV Gag RNA in AMs from 36.4% of participants and Gag DNA in AMs from 45.5% of participants, which is consistent with studies using qPCR to detect HIV nucleic acids in AMs 18,20,21,40,[46][47][48][49][50] .
To better understand the phenotype of HIV derived from AMs, we generated replication-competent strains of HIV using env isolated from individuals with untreated HIV infection. We found that 8 separate AM-derived viruses and 9 plasma-derived viruses, selected to maximize sequence diversity, all had a T-tropic phenotype. This suggests that T-tropic HIV is infecting AMs in vivo despite inefficient infection of macrophages by these isolates in vitro. We used PCR-based methods to amplify virus, which could result in HIV recombination within the samples. However, the consistency of the entry phenotype across separate AM-derived Env constructs suggests that the tropism of macrophage-derived Env is more broadly applicable. In addition, the adherence method of isolating AMs from BAL, while resulting in 99-100% macrophage purity by RapidDiff stain, does not allow us to rule out a contribution of residual infected T cells to the HIV we detected and isolated. This method has been used in recent prior literature measuring HIV in AMs 22,56 and led to minimal T cell inclusion in our co-culture studies, but could have resulted in leftover HIV-infected CD4+ T cells. A third caveat is that cell-free virus occasionally replicated over time in the cell-free conditions, although more rarely than with cell-to-cell contact. This phenomenon suggests that cell-free infection of MDMs by this T-tropic virus is possible but is less efficient than cell-to-cell infection. M-tropic HIV strains have 30-fold more efficient entry into macrophages than T-tropic strains do 23 , which is consistent with the idea that T-tropic strains occasionally enter MDMs.
Our data suggests that the p24 increase in AM-T cell co-cultures is due to productive infection of AMs. We cannot rule out new infection of CD4+ T cells that remained in the co-cultures after three rounds of washing. Treatment of these co-cultures with the reverse transcriptase inhibitor efavirenz inhibited viral replication. This shows that new viral production originated from cells infected after co-culture of CD4+ T cells and macrophages, and did not come from viral release from residual infected cells. It is possible that the newly infected cells are CD4+ T cells. However, in our microscopy staining, CD4+ T cells were at low prevalence (Fig. 5B), and most of the HIV Gag staining was localized to macrophages (Fig. 5D). Our work indicates that cell-to-cell infection of macrophages may be an important mechanism of T-tropic viral entry into AMs in the acute and chronic stages of infection.
While dendritic cells, alveolar epithelial cells, CD8+ T cells, NK cells and neutrophils may impact HIV infection dynamics in the alveolar space, either directly by acting as targets for infection or indirectly through cytokine secretion 57 , the main cells infected with HIV in the lung in vivo are CD4+ T cells and alveolar macrophages 58,59 .
A number of possibilities may explain the mechanism of the observed enhancement. The formation of an immunological and/or virological synapse, which involves the clustering of immune receptors and HIV virions at the interface between an AM and CD4+ T cell, may help overcome the low density of HIV entry receptors on AMs 44,60,61 . Fusion of the infected CD4+ T cell with the macrophage is another proposed mechanism of HIV infection of macrophages 45 , although we did not observe multinucleated giant cells or macrophages with surface CD3 expression, suggesting that this may not be occurring in our system with high frequency. It is also possible that autologous CD4+ T cells may activate AMs and increase HIV transcription 62 , as has been shown in dendritic cells 63 . In the CTC condition, we observed AMs with cytosolic Gag staining with T cells closely apposed to the membrane or internalized. This implies that AM phagocytosis of T cells occurs but may not be required  Fig. 3A). However, HIV-infected macrophages are known to persist for long periods because they are more resistant to the cytopathic effects of HIV 35 , the cytotoxic activity of CD8+ T cells 66 , and apoptosis [67][68][69] . Infected AMs may persist during ART treatment and may contribute to the dysfunction 22,69 either directly or through indirect effects of HIV components or other factors on bystander AMs 70 . Persistent AM dysfunction may contribute to the increased rate of lung inflammation and infectious disease seen in people with HIV on ART [71][72][73][74] . An understanding of the mechanism behind AM infection may help efforts to address persistent sources of inflammation in the lung in people with HIV on ART.
ART reverses some but not all of the functional phagocytic defects in alveolar macrophages, which may be due to a differential impact of ART on infected and bystander AMs 75 . Direct infection of macrophages can activate immune pathways including type I interferons that affect bystander function 76,77 and chemokines which recruit other immune cells 66,78 . A number of papers have explored mechanisms of macrophage functional impairment in people with HIV on ART that are not dependent on direct infection, which include gp120-induced inhibition of apoptosis 69 , nef-induced inhibition of phagocytosis 8,79 and downregulation of CD36 80 , as well as posttranslational modification of Fc receptors 81 and changes in reactive oxygen species generation 82 . Notably, HIV proteins including gp120 and nef persist in bronchoalveolar lavage fluid in people with HIV on ART 69,83 . Finally, data from ART-naïve macaques shows that only about 1 in 20,000-100,000 AMs is productively infected with SIV 84 , so HIV-infected AMs likely only make up a small minority of AMs in the lung in ART-naïve or treated individuals. The mechanisms of HIV-mediated AM impairment, even in the absence of ART, are most likely attributable to an indirect effect of either HIV or soluble factors produced by infected AMs on uninfected AMs.
The finding that HIV isolated from AMs is T-tropic has important implications for our understanding of the HIV reservoir. The vast majority of HIV strains, including transmitted/founder strains, are T-tropic 25,33 . If T-tropic HIV is able to infect AMs, the majority of people with HIV have the potential to have infected AMs, as well as infected macrophages in other tissues. This has implications for macrophages as a reservoir for HIV. It also suggests that tissue macrophages may need to be targeted in cure strategies.
The potential for AMs to be productively infected by contact with an infected CD4+ T cell intermediate has several important implications for our understanding of HIV pathogenesis. Productive HIV infection has been shown to have many effects on macrophage function, including the generation of defects in phagocytosis, proteolysis, and cytokine production 9 . Additionally, infected macrophages may efficiently transmit HIV to other CD4+ T cells and macrophages, as has been shown for M-tropic HIV, through efferocytosis of infected macrophages or cell-to-cell contact 61,85 . HIV-infected macrophages are resistant to cell death and in non-human primates can harbor replication-competent SIV after ART 65 , leading to a cellular reservoir that may play a role in the failure of current attempts at HIV cure. Finally, HIV infection in macrophages via contact with T cells may play a role in the chronic immune activation that is seen with HIV infection. Multiple studies have associated macrophage-derived products with increased morbidity and mortality during HIV infection, including those on ART. These include IL-6, soluble CD14 (sCD14), soluble tumor necrosis factor receptor 1 (sTNFR1), sTNFR2, and indoleamine 2,3-dioxygenase (IDO) activity [86][87][88][89] . Therefore, AMs infected with T-tropic HIV via T cell contact may lead to multiple immune defects which may contribute to an increased incidence of lung disease and promotion of HIV disease progression in people living with HIV.

Methods
Human participants. Bronchoscopies were performed in a cohort of participants with HIV infection and uninfected control individuals residing in Cape Town, South Africa, as previously described in 90 . All participants in the cohort were sensitized to Mycobacterium tuberculosis as defined by positive results of an interferon γ release assay (IGRA; Quantiferon, Cellestis), without active TB as defined by absence of signs or symptoms of active TB, lack of clinical findings by chest X-ray, and a negative BAL TB culture. All volunteers who had previously been diagnosed with or treated for TB were excluded. The HIV-infected participants in the Cape Town cohort were asymptomatic and not using antiretroviral therapy. All HIV + IGRA + volunteers were eligible for INH prophylaxis according to South African National guidelines, and were offered isoniazid prophylactic treatment for 6 months, which was the indication at the time. All samples were taken prior to INH treatment.
For in vitro experiments, BAL was also collected from a cohort of HIV-uninfected individuals, recruited from outpatient clinics at local Boston hospitals, following institutional review board approval (IRB protocol # 2013P000063) and written informed consent. For HIV infection experiments involving monocyte derived macrophages (MDMs), cells were isolated from buffy coats of anonymous healthy blood donors obtained from the Massachusetts General Hospital (MGH) blood donor center (Boston, MA) under protocol # 2005P001218.

Bronchoalveolar lavage (BAL). Bronchoscopies were performed under conscious sedation via standard
technique 15,90 . For each BAL in the Cape Town cohort, samples were collected from the right middle lobe by washing with 160 ml of normal saline. After adherence, approximately 10,000 cells were centrifuged onto a coated microscope slide using a Cytospin Centrifuge (Shandon 3.0). The slide was then allowed to dry, dipped repeatedly in fixative and differentially stained by dipping the slide consecutively in two contrasting dyes (Rapid-Diff Kit, Clinical Sciences Diagnostics). The differential stain allows characterization of lymphocytes, macrophages and neutrophils, based on morphology. The slide was then viewed under immersion oil magnification using a light microscope and a total of 200 cells were counted. The BAL samples consisted of a median of 96% www.nature.com/scientificreports/ and 93% macrophages prior to adherence, in the HIV+ and HIV-BAL samples, respectively. This was enriched further by adherence for 20 min in order to eliminate the non-adherent cell fraction. For the Boston cohort, samples were collected from the lingula and the right middle lobe by washing 120 ml of normal saline in each segment. Alveolar macrophages (AMs) from bronchoscopy participants were isolated by 20-60 min of adherence. AMs for in vitro infection were washed 3 times with PBS and cultured in RPMI with 10% (v/v) FCS. BAL CD4+ T cells were obtained at bronchoscopy of ART-naive uninfected participants using a Human CD4+ T cell Enrichment kit (StemCell EasySep) on non-adherent BAL cells.
Cell isolation from blood. Peripheral blood from the bronchoscopy study participants recruited in Boston was obtained at least 7 days before the bronchoscopy and PBMCs were processed and cryopreserved. Ficoll gradients were used to isolate peripheral blood mononuclear cells (PBMCs). Monocytes and CD4+ T cells were then isolated by CD14+ positive selection (Miltenyi) and CD4+ T cell negative selection (StemCell EasySep), respectively. MDMs were obtained by maturing monocytes in RPMI with 10% GemCell US Origin Human Serum AB (GemBio) for 7 days.

Measurement of HIV nucleic acids.
For measurement of HIV RNA and DNA, AMs were lysed in RLT Plus lysis buffer (QIAGEN) with 1% β-mercaptoethanol (Sigma). The lysate was run through QIAShredder columns (QIAGEN), and isolated with AllPrep Micro DNA/RNA kits (QIAGEN). HIV Gag primers were used to obtain the number of copies of HIV RNA and DNA, and CCR5 primers were used to determine the number of cells per sample using TaqMan qRT-PCR (Applied Biosystems) 91 . Standard curves were prepared using preamplification of a purified restriction digested HxB2 plasmid for HIV gag, kindly provided by Todd Allen and Karen Power, and by preamplification of a CCR5 plasmid 91 .

HIV infection of CD4+ T cells and macrophages.
For cell-free infection experiments, 2 × 10 4 infectious units of HIV, or for mock infection, media, were added to 10 5 cells (MDMs, T cells or AMs) in 100 μl total volume for 12 h at 37 °C, then washed three times with PBS and replaced with 200 μl media. Supernatant was collected for p24 measurement at days 1 (1 h after changing media), 2, 7, and 14. CD4 + T cell cultures had low viability at day 14, consistent with prior literature 100 , so samples were not collected at this point, and day 7 was the final timepoint. HIV p24 levels were measured by p24 ELISA (PerkinElmer).

Co-culture of HIV infected CD4+ T cells and macrophages. For co-culture experiments with HIV-
infected T cells, CD4+ T cells from matched donors were activated for 3 days with 2 ng/ml phytohemagglutinin (PHA-P) (ThermoFisher) and 10 ng/ml IL-2 (kindly provided by Alicja Trocha and the Ragon Institute Protein Core). 2 × 10 4 infectious units of HIV were added to 10 5 cells in each well and spin-infected by centrifugation at 800 ×g for 90 min at 4 °C; the cells were cultured for 12 h, washed and the media was replaced 15 . On day 4 after infection, autologous HIV-or mock-infected T cells were spun at 500 ×g for 3 min, the supernatant was collected, and the T cells were resuspended in fresh media. MDMs or AMs were cultured with one of three preparations for a 1:1 ratio of T cells to macrophages: supernatant collected from the CD4+ T cells (supernatant or "SN"), washed CD4+ T cells (cell-to-cell or "CTC"), or washed CD4+ T cells in the upper chamber of a 0.4 μm transwell (Transwell Costar) (transwell or "TW"). After incubation for 3 h at 37 °C, the cells were washed three times for 60 s with ice cold PBS with 10 mM EDTA, and media was replaced. Supernatant was collected for p24 measurement at days 0 (1 h after changing media), 2, 7 and 14 from the same wells with collection and replacement of 25% of the media volume. p24 levels were measured by ELISA (PerkinElmer). For experiments with AMs, autologous PBMCs were thawed 7 days before the bronchoscopy, and CD4+ T cells were isolated and activated for three days as above. In Fig. 4 and Supplementary Fig. 5, samples were only included in the final dataset if there was detectable HIV infection in the CD4+ T cells before they were added to the macrophages.
HIV primary isolate sequencing and production of primary isolates. Viral  www.nature.com/scientificreports/ vRNA were generated by reverse transcription using the SuperScript III protocol (Invitrogen). Nested PCR was performed for the env gene as described in 101 , and amplicons were cloned into a pHDM vector using the In-Fusion HD Cloning kit (Clontech) and sequenced on an Illumina MiSeq. Based upon analysis of env sequence diversity, 22 isolates were selected to represent the diversity of sequences from the 3 donors, with at least one sequence from each clade (arrows in Supplementary Fig. 2). pNL4-3 ΔEnv GFP was obtained through the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH: HIV NL4-3 Δenv EGFP Reporter Vector from Drs. Haili Zhang, Yan Zhou, and Robert Siliciano (cat# 11100) 102 . The amplicons were PCR amplified from the pHDM vectors and cloned into the pNL4-3 vector. HIV stocks were produced, frozen and titered on GHOST cells as above. Experiments in Fig. 2 and Supplementary Fig. 3 used 2 AM-derived and 2 plasma-derived viral isolates from donor 1, 3 AM derived and 2 plasma-derived viral isolates from donor 2, and 3 AM derived and 5 plasma-derived viral isolates from donor 3, which are indicated by arrows with and without a black border in Supplementary Fig. 2. Experiments in Figs. 4 and 5 used one viral isolate from each viral donor and compartment, which are indicated by arrows without a black border.
Microscopy. Poly-L-lysine coated coverslips were used to culture MDMs and AMs with infected CD4+ T cells or supernatant from infected CD4+ T cells under SN or CTC conditions as described above. Cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% saponin, blocked with 1% BSA, and stained with antibodies against HIV Gag (conjugated to FITC, Beckman Coulter), CD3 (rabbit polyclonal, Dako), and CD68 (KP1, BioCare Medical), then Alexa Fluor 568 Goat anti-Rabbit and Alexa Fluor 647 Goat anti-Mouse (Invitrogen), then DAPI and mounted with ProLong Diamond mounting media (Thermo Fisher). Slides were scanned with a TissueFAXS SL Q spinning disc Confocal microscope (TissueGnostics USA) using a Zeiss Plan-apochromatic 63 × 1.4NA oil immersion objective. Image analysis involved counting 100 cells from each slide on a single Z-stack image and classifying each cell as CD68± , CD3± , and Gag±, and if Gag+ , whether it was present in CD68+ and/or CD3+ areas of the cell.
Quantification and statistical analysis. Statistical details of experiments can be found in each figure legend. Nonparametric tests were used to compare medians between groups unless noted otherwise. The Mann-Whitney test was used for 2 groups and the Kruskal-Wallis test followed by Dunn's multiple comparison posttest was used for > 2 groups. Wilcoxon signed rank was used to compare continuous data between two time points. Spearman's correlation coefficients were used to examine associations between variables. Differences were considered significant at p < 0.05. For figures marked "fold increase", the value at the final timepoint was compared to the initial timepoint; rare values below 1 were normalized to the value of 1. Prism 8 (Graphpad) was used for all analyses 15 .