HIV-Infected Macrophages Are Infected and Killed by the Interferon-Sensitive Rhabdovirus MG1

Human immunodeficiency virus type 1 (HIV-1) remains a treatable, but incurable, viral infection. The establishment of viral reservoirs containing latently infected cells remains the main obstacle in the search for a cure.

postinfection (dpi), distinct HSA-positive (HSA 1 ) and -negative (HSA 2 ) populations could be detected by flow cytometry (Fig. 1A). Similar to others (25), we found that HSA 1 MDM enriched by magnetic bead isolation continued to release HIV-1 p24 antigen into culture supernatants, while HSA 2 cells did not ( Fig. 1B and C). Thus, HSA 1 MDM were defined as HIV infected and HSA 2 MDM were defined as bystander populations.
Using this model, bulk MDM cultures consisting of both HSA 1 and HSA 2 cells were exposed to the green fluorescent protein (GFP)-expressing MG1 virus. After 48 h, the proportion of GFP-positive cells was assessed by flow cytometry (Fig. 2A, representative histograms). As evidence for the preferential infection of HIV-infected MDM by MG1, a significantly greater percentage of GFP-positive cells was seen in the HSA 1 MDM population than in the HSA 2 cells (Fig. 2B). Since this preferential infection could  Intact cells were analyzed (black gate), after which HSA 2 (blue gate) and HSA 1 (red gate) MDM were gated upon. The percentages of GFP 1 cells were then measured within HSA 1 and HSA 2 populations, as shown in representative histograms. Histogram peak counts (y axis) for HSA 2 and HSA 1 populations were normalized to that of the uninfected control for be explained, in part, by elevated surface expression of the MG1 receptor on HIVinfected cells, we measured the expression of the low-density lipoprotein receptor (LDLR) (26)(27)(28) and found that its expression was slightly higher on HSA 1 cells ( Fig. 2C and D). Differences in MG1 infection may therefore be partially receptor mediated, although the biological relevance in this small difference in surface expression remains undetermined.
MG1 preferentially kills HIV-infected monocyte-derived macrophages. Oncolytic rhabdoviruses, including MG1, have been found to trigger cell death pathways in cancerous cell lines and solid tumors (29)(30)(31). To investigate whether preferential infection by MG1 would lead to increased killing of HIV-infected cells, we chose to perform annexin V staining to measure surface levels of phosphatidylserine, an early indicator of apoptosis, on MDM following MG1 infection. First, we assessed whether or not MG1 infection would increase the frequency of annexin V 1 cells within HSA 1 and HSA 2 MDM populations as a whole at 24 h post-MG1 infection. Not only did the frequency of annexin V 1 cells increase in a multiplicity of infection (MOI)-dependent manner, but this frequency was higher within HSA 1 MDM than HSA 2 MDM ( Fig. 3A and B). In order to understand whether productive MG1 infection resulted in a difference in annexin V expression between HSA 2 and HSA 1 MDM, we next measured annexin V staining on the GFP 1 cells found within the HSA 1 and HSA 2 cell gates and found that GFP 1 /HSA 1 MDM contained a significantly higher percentage of annexin V 1 cells than GFP 1 /HSA 2 MDM (Fig. 3C).
As has been previously demonstrated (19,(32)(33)(34), the killing of HIV-infected cells was then confirmed by measuring integrated (proviral) HIV-1 DNA at 2 days post-MG1 infection. Consistent with the annexin V staining, an MOI-dependent decrease in proviral HIV-1 DNA was observed in HIV-infected MDM that had been exposed to infectious MG1 (Fig. 3D). The release of HIV-1 p24 antigen into cell supernatants was also inhibited by MG1 infection (Fig. 3E). Importantly, UV-inactivated MG1 had no effect on proviral HIV-1 DNA or p24 release, indicating that replication-competent oncolytic virus was required for the killing of HIV-infected MDM. Together, these data indicate that MG1 infection results in the preferential death of HIV-infected MDM.
MG1-mediated killing of HIV-infected monocyte-derived macrophages is not mediated by soluble factors. Because cell death was observed in both GFP 1 /HSA 1 and GFP 2 /HSA 1 cells, we considered the possibility that MG1 infection was also causing the indirect, cytokine-mediated killing of HIV-infected cells. To address this, we performed supernatant transfer experiments, as depicted in Fig. 4A. Cells that received conditioned supernatants from MG1-infected MDM showed a small but significant drop in viability as measured by the MTT [3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2Htetrazolium bromide] assay (Fig. 4B). Both MG1 infection and conditioned supernatants also prevented the accumulation of HIV-1 p24 antigen in culture medium (Fig. 4C), but only MG1 infection resulted in a decrease in proviral HIV-1 DNA (Fig. 4D). Additionally, no effect on HIV-1 p24 release or proviral HIV-1 DNA was seen following exposure to either UV-inactivated MG1 or supernatants from MDM exposed to UV-inactivated MG1. Measurement of cytokines in filtered supernatants showed that concentrations of IFN-a2, tumor necrosis factor alpha (TNF-a), interleukin 4 (IL-4), and IL-6 were elevated at 48 h post-MG1 infection (Fig. 5). IFN-g was also measured but was undetectable in all samples analyzed. In summary, the eradication of HIV-infected MDM is not mediated by soluble factors and requires the presence of infectious MG1.
ISG induction is impaired within HIV-infected monocyte-derived macrophages. As MG1 is highly sensitive to IFN-I (17), we next wanted to determine if IFN-I signaling (C) Flow cytometry gating strategy and representative histograms depicting LDLR expression on HSA 2 (blue) and HSA 1 (red) MDM. Intact cells were gated (black), after which HSA 2 (blue) and HSA 1 (red) MDM were defined. The PE FMO control is shown as a filled, gray peak. Histogram peak counts (y axis) for HSA 2 and HSA 1 populations were normalized to that of the PE FMO control for visualization purposes. (D) LDLR expression on HSA 2 (blue) and HSA 1 (red) MDM (n = 7; P = 0.003 by paired, two-tailed t test). Data represent mean 6 SEM; n values represent separate biological replicates. influenced OV-mediated cytopathogenicity. We began by assessing MG1 infection and killing in IFN-a-pretreated MDM cultures and found that stimulation with IFN-a blocked both MG1 infection (Fig. 6A) and the reduction of proviral HIV-1 DNA in a dose-dependent manner (Fig. 6B). Interestingly, preferential infection and the corresponding reduction in proviral HIV-1 DNA were maintained after treatment with low doses of IFN-a, leading us to investigate whether or not differences in ISG expression existed between HIV-infected and bystander cells.
To do this, HIV-1-infected MDM cultures were stimulated with IFN-a, and levels of two representative ISG products, PKR and ISG15, were measured by flow cytometry (Fig. 7A). Basal expression of both proteins was higher in HSA 1 MDM than in both HSA 2 MDM and HIV-1 naive MDM ( Fig. 7B and C). Conversely, the relative IFNa-induced expression of PKR and ISG15 was lower in HSA 1 MDM ( Fig. 7D and E), indicating impaired IFN-a responsiveness in HIV-infected cells. This was not associated with differences in ISG15 or PKR mRNA levels in sorted HSA 1 and HSA 2 MDM ( Fig. 7F and G). Additionally, surface expression of the IFN-a/b receptor (subunits 1 and 2; IFNAR1/2) did not differ between HSA 1 and HSA 2 cells (Fig. 8), suggesting that differences in ISG induction in these cell populations was not due to receptor downregulation.
MG1 eliminates HIV-infected alveolar macrophages ex vivo. Finally, to begin to address the potential clinical implications of our findings, we investigated whether MG1-mediated killing of HIV-infected myeloid cells could be replicated in primary alveolar macrophages. As one of the only accessible sources of tissue-resident myeloid cells, AM are known to harbor replication-competent HIV-1 and thus act as a viral reservoir (2,(20)(21)(22)35). To assess whether HIV-infected AM were susceptible to MG1-mediated killing, cells were collected from PLWHIV (on suppressive ART for $3 years at the time of collection) via bronchoalveolar lavage, isolated by plate adherence, and infected with MG1.
Following MG1 infection of adherent AM, proviral HIV-1 DNA was measured by digital droplet PCR (ddPCR). As has been previously demonstrated using cells collected from the lungs of PLWHIV (35), the frequency of HIV-infected AM was highly variable. Despite a relatively small number of participants, a statistically significant decrease in proviral HIV-1 DNA was observed at 48 h post-MG1 infection ( Fig. 9), emphasizing the consistency of this effect. Additionally, MG1 infection did not appreciably reduce CD3 DNA copy number (as quantified by ddPCR) in comparison to MG1-uninfected AM, suggesting that HIV-uninfected AM were not killed by MG1. As primary, tissue-resident macrophages have been historically overlooked in HIV-1 cure research (reviewed in references 36), these data provide important insights into the future application of MG1 as an HIV-1 cure strategy.

DISCUSSION
The development of an HIV-1 cure has focused largely on the identification, characterization, and eradication of latently infected CD4 1 T cells. Unfortunately, many of these strategies remain untested in other cellular reservoirs of HIV-1. We therefore sought to identify whether MG1, an established oncolytic rhabdovirus capable of tar- (Continued on next page) geting latently HIV-infected cell lines and primary CD4 1 T cells (19), could also infect and kill HIV-infected macrophages.
In agreement with our previous findings in CD4 1 T cells, we observed a higher frequency of MG1 infection in HSA 1 (HIV-infected) MDM than in HSA 2 (bystander) MDM. This coincided with the preferential killing of HIV-infected MDM, as measured by an increase in annexin V 1 cells and a reduction in proviral HIV-1 DNA. Next, we investigated whether MG1-mediated killing of HIV-infected MDM could be achieved via indirect means. UV-inactivated viral particles, for instance, have been found to induce the cytolytic killing of acute myeloid leukemia cells in vitro and in a murine model of leukemic blast crisis (37). OV-induced release of proinflammatory cytokines may also facilitate the indirect killing of tumor cells (38). In the context of HIV-1 infection, however, UV-inactivated MG1 did not kill HIV-infected MDM (Fig. 3D) or stimulate the release of proinflammatory cytokines such as TNF-a or IL-6 ( Fig. 5).
Similarly, treatment of HIV-infected cells with conditioned supernatants from MG1infected MDM did not eliminate HIV-infected cells but resulted in a significant reduction in HIV-1 p24 antigen release ( Fig. 4C and D). IFN-a2, TNF-a, IL-4, and IL-6 were elevated in supernatants from MG1-infected cells, which is consistent with previous literature describing cytokine release by cells infected with MG1 or with the closely related rhabdovirus, vesicular stomatitis virus (31,39,40). Cytokines including TNF-a, IL-6, and IFN-g have also been shown to block HIV replication in human macrophages (41)(42)(43)(44), suggesting that these soluble factors are responsible for the reduction in p24  Next, we investigated the role of IFN-I signaling in MG1 infection. The MG1 virus is highly sensitive to IFN-I signaling due to two genetically engineered amino acid substitution mutations within the G (Q242R) and M (L123W) proteins (17). It was therefore no surprise that IFN-a pretreatment protected both HSA 1 and HSA 2 MDM from MG1 infection in a dose-dependent manner (Fig. 6).
To investigate whether we could identify global IFN-I signaling differences between HSA 1 and HSA 2 MDM, we measured the basal and IFN-a-induced expression of two representative ISG. The observed baseline elevation of PKR and ISG15 in HSA 1 MDM was as expected, given that ISG induction has been shown to occur within the first 24 h of HIV-1 infection in MDM (46). Nonetheless, it was interesting to see that higher levels of ISG were not sufficient to block preferential MG1-mediated cytopathogenicity, as elevated ISG expression in malignant cells has been found to be protective against OV infection and killing (47). One possible explanation is the functional impairment of ISG in HIV-infected MDM. The interaction of PKR with a number of cellular and viral proteins in HIV-1-infected cells has already been shown to block this ISG from fulfilling its role as a key RNA sensing protein needed for the activation of cellular antiviral response pathways (48)(49)(50)(51). Additionally, given that pretreatment with low levels of IFN-a did not prevent MG1-mediated killing of HIV-infected cells (Fig. 6B), and that HIV-infected MDM are impaired in their ability to produce IFN-I (14), it is possible that additional defects in antiviral signaling that sensitize these cells to OV infection exist.
As evidence of altered IFN-I signaling in HIV-infected MDM, we also showed that the IFN-a-mediated induction of ISG15 and PKR was impaired in HSA 1 MDM. This is consistent with findings previously reported by Ranganath    Finally, we demonstrated that MG1 could eliminate HIV-infected macrophages ex vivo, using AM from ART-treated PLWHIV. AM form a unique source of accessible, tissue-resident myeloid cells that form a known HIV-1 reservoir (60). The ability to eradicate these cells using an IFN-I-sensitive OV therefore has important implications for the development of an HIV-1 cure. Low cell numbers, and the relative fragility of cells obtained by bronchoalveolar lavage, unfortunately prevented further investigation of markers of MG1 infection, cell death, and IFN-I signaling, as was possible with in vitro HIV-infected MDM. This, combined with the relatively small number of donors recruited for bronchoalveolar lavage and AM collection, represents a current limitation of this study objective. Still, the decision to assess MG1-mediated elimination of HIVinfected cells using primary AM from PLWHIV represents an important step forward in the field of HIV-1 cure research. Primary, tissue-resident macrophages have consistently been overlooked as a source of replication-competent HIV-1 and, consequently, are a necessary target for novel HIV-1 cure strategies. Future work assessing the efficacy of latency reversal agents, allogeneic stem cell transplant, or other gene therapy approaches on HIV-1 reservoir size must therefore consider the myeloid HIV-1 reservoir as part of the experimental design (36).
In summary, these results demonstrate the preferential infection and killing of HIVinfected MDM by the IFN-I-sensitive oncolytic virus MG1. In doing so, we have fulfilled a current objective in HIV-1 cure research, which is to consider and account for myeloid HIV-1 reservoirs when testing new therapeutic strategies that have already been investigated in CD4 1 T cells (36). Going forward, the ability of MG1 to reduce the size of the HIV-1 reservoir will need to be further assessed using a relevant in vivo model of HIV-1 infection. Should ongoing phase I/II clinical trials of MG1 in cancer (ClinicalTrials registration numbers NCT02285816, NCT02879760, and NCT03618953) identify this oncolytic virus as safe, a proof-of-concept clinical trial to investigate the safety and impact on HIV reservoir size in PLWHIV may be feasible. Ethics statement. Experiments requiring healthy volunteers were approved by the Ottawa Health Science Network Research Ethics Board (protocol no. 2005388-01H), and all participants provided written informed consent. The collection and use of alveolar macrophages from PLWHIV were approved by the Institutional Review Boards of the MUHC (no. 15-031) and the Université du Québec à Montréal (no. 602). All study participants provided written informed consent.
Monocytes were separated from healthy donor peripheral blood mononuclear cells (PBMC) by plate adherence. Following isolation by density gradient centrifugation, PBMC were resuspended at 6.25 Â 10 6 /ml in warm, serum-free RMPI 1640 with PenStrep. A total of 1.25 Â 10 8 PBMC were then plated in 150-cm 2 polystyrene tissue culture dishes (Sarstedt, Nümbrecht, Germany) and left to adhere for 2 h at 37°C. Plates were washed 3 times with endotoxin-free PBS (pH 7.4; Gibco) to remove nonadherent lymphocytes, and 20 ml of warmed RPMI 1640, supplemented with PenStrep and 10% heat-inactivated human AB serum (Mf medium) and M-CSF (25 U/ml), was added to the plate. Adherent cells were incubated at 37°C with 5% CO 2 for 7 days. At 3 days postplating, cells were washed twice with warmed endotoxin-free PBS, and 20 ml of Mf medium was added to the plate. On day 8, adherent MDM were washed twice with endotoxin-free PBS, detached using Accutase (Millipore-Sigma) and gentle scraping with a Sarstedt cell scraper, and counted by trypan blue exclusion. MDM were then pelleted by centrifugation (300 Â g for 10 min), resuspended at 2.5 Â 10 5 cells/ml in Mf medium, and plated in either 6-well (5 Â 10 5 cells/well) or 12-well (2.5 Â 10 5 cells/well) plates for further experiments.
AM were isolated from bronchoalveolar lavage fluid by plate adherence, as described previously (35,63,64). Briefly, participants recruited at the McGill University Health Centre (MUHC, Montreal, Canada) were cART-treated PLWHIV, with suppressed plasma viral load for $3 years and without respiratory symptoms or active illness. A total of 50 to 100 ml of lavage fluid was collected during bronchoscopy. Cells were pelleted and washed at 180 Â g for 10 min and then counted by trypan blue exclusion. Cells were then resuspended in serum-free RPMI 1640 at 5 Â 10 5 cells/ml and plated in 24-well plates at 2.5 Â 10 5 cells/well for 2 h at 37°C. Nonadherent cells were removed by rinsing the wells with endotoxinfree PBS, and adherent cell were covered with 500 ml of Mf medium. Prior to collection, adherent AM were rinsed an additional 3 times using endotoxin-free PBS, in order to remove cellular debris and nonadherent lymphocytic cells. AM were detached at 37°C for 30 min using CellStripper dissociation reagent (Corning, Fisher Scientific), followed by gentle pipetting.
Production The GFP-expressing recombinant OV MG1 (obtained from John Bell and David Stojdl) was propagated and titered on Vero cells, as described previously (17,65). UV inactivation of MG1 stocks was performed as described previously (19).
In vitro HIV-1 infection and enrichment of HSA + MDM. MDM were infected with HIV NL4.3 BAL-IRES-HSA for 6 days at 37°C with 5% CO 2 . HIV-1 infection was confirmed by quantitative PCR (qPCR) (66), p24 ELISA, and surface expression of virus-encoded murine heat-stable antigen (HSA) by flow cytometry (24). HSA-expressing MDM were isolated by positive selection using Miltenyi LS columns in combination with the MidiMACS magnet (Miltenyi Biotech). The HSA sorting protocol was optimized by the laboratory of Michel Tremblay, as described previously (24,25). Purity of the HSA 1 and HSA 2 fractions was assessed by flow cytometry.
In vitro MG1 infection and supernatant transfer experiments. MDM and AM were infected with MG1 in RPMI 1640, supplemented with PenStrep and 10% heat-inactivated human AB serum (Mf medium), with 10 mM maraviroc. Cells pellets were collected at 48 hpi and stored at 280°C for quantification of HIV-1 DNA. MG1 infection, frequency of HSA-expressing cells, and cell viability were assessed in MDM cultures by flow cytometry and MTT assay.
For supernatant transfer experiments, MDM from healthy donors were infected with either MG1 or HIV NL4.3 BAL-IRES-HSA. At 48 hpi, supernatants from MG1-infected cells were collected, filtered with Amicon Ultra centrifugal filter units (molecular weight cutoff [MWCO], 100 kDa; Millipore Sigma) and stored at 280°C. Removal of infectious MG1 was confirmed on Vero cells by flow cytometry. At 6 dpi, HIV-infected MDM cultures were infected with MG1 at an MOI of 10, left uninfected, or treated with UVinactivated MG1. In duplicate, HIV-infected MDM were treated with filtered supernatants, at a 1:1 ratio with fresh Mf medium. At 48 h post-MG1 infection, cell viability was assessed by MTT assay and proviral HIV-1 DNA was measured by qPCR. Cell-free supernatants were collected every 2 days for quantification of HIV-1 p24 antigen by ELISA. MDM were pelleted at 6 dpi and stored at 280°C for quantification of proviral DNA.
IFN-a stimulation of MDM. MDM were stimulated with IFN-a at 6 days post-HIV-1 infection. IFN-a was diluted in Mf medium and added directly to the wells, after which MDM were incubated at 37°C for 16 h for mRNA isolation or 24 h for flow cytometry analysis.
For HSA staining, cells were resuspended in cold PBS plus 10% human AB serum and 20% NGS and incubated at 4°C for 20 min. The anti-HSA antibody (clone M1/69; BioLegend) or HSA isotype control antibody (rat IgG2b κ isotype clone RTK4530; BioLegend) was then added at 3 mg/ml, and cells were incubated at 4°C for 15 min. Cells were then washed twice and prepared for either further staining or for sorting. T.S.S. validated the methodologies, designed and performed the experiments, analyzed the data, and drafted the final manuscript. N.R. assisted in optimizing the MG1 infection experiments, as well as the MTT assay and HIV-1 p24 ELISA utilized in the study. S.C.B.S. assisted in the collection of healthy donor blood, the measurement of supernatant cytokines by Luminex (Fig. 5), and the overall editing of the final manuscript. S.S., O.M., C.T.C., and M.-A.J. assisted in patient recruitment, coordination of patients undergoing bronchoalveolar lavage, and collection of alveolar macrophages. S.C.C. and J.B.A. participated in the overall study design and assisted with data analysis and interpretation, as well as the drafting and editing of the manuscript. All authors read and approved the final manuscript prior to submission.
We have no competing interests to declare.