The CCL2 Chemokine Promotes Early Seeding of the Latent HIV Reservoir

HIV infects long-lived CD4 memory T cells establishing a latent viral reservoir that necessitates lifelong anti-retroviral therapy (ART). How this reservoir is formed so swiftly remains unknown. We now show the innate inflammatory response to HIV infection results in CCL2 chemokine release, which can drive recruitment of cells expressing the CCR2 receptor including a subset of central memory CD4 T cells. Supporting a role for the CCL2/CCR2 axis in rapid reservoir formation, we find 1) treatment of humanized mice with anti-CCL2 antibodies during HIV infection decreases reservoir seeding and 2) CCR2/5+ cells from the blood of HIV-infected individuals on long term ART contain significantly more provirus than CCR2/5-negative memory or naïve cells. Together, these studies support a model where the host’s innate inflammatory CCL2 response to HIV infection recruits CCR2/5+ central memory CD4 T cells to zones of virus-associated inflammation likely contributing to rapid formation of the latent HIV reservoir. GRAPHICAL ABSTRACT Why is the latent HIV reservoir established so early following infection? An innate immune response occurs during acute infection that establishes a “zone of inflammation” (step 1). The CCL2 chemokine is produced in part through IFI16 sensing of HIV DNA in abortively infected cells. CCL2 promotes rapid recruitment of CCR2/5+ memory CD4 T cells (step 2). Many of these cells become productively infected (step 3) and a fraction become latently infected (step 4). Thus, HIV hijacks the host inflammatory response to rapidly establish the latent reservoir. In support of this model, we find HIV reservoir reduction in humanized mice treated with anti-CCL2 antibodies during early infection. Further, we find that CCR2/5+ CD4 T cells harbor a substantial fraction of detectable proviruses in the blood of HIV-infected individuals on long-term suppressive ART.

this model, we find HIV reservoir reduction in humanized mice treated with anti-CCL2 antibodies

25
The latent HIV reservoir is comprised of infectious proviruses predominantly within memory CD4 26 T cells. These cells are not effectively depleted by antiretroviral therapy (ART), making lifelong 27 treatment a requirement to prevent disease progression. Targeted elimination or control of the 28 latent HIV reservoir is critical for achieving an HIV cure. In rhesus macaques, a stable SIV 29 reservoir is established within 3 days or less following infection 1 . Likewise, the HIV reservoir is 30 rapidly seeded in humans. Administration of ART immediately after detection of HIV RNA in 31 plasma prior to detection of p24 gag or seroconversion (Fiebig stage I) in closely monitored, high 32 risk individuals, fails to prevent viral rebound during a subsequent analytical treatment interruption A potential mechanism for the recruitment of memory cells to the site of infection is via chemoattraction. Acute infection by HIV drives the production of multiple cytokines. Prominent among these is the C-C motif ligand 2 chemokine (CCL2/MCP-1) 12-14 . CCL2 is one of the first rapid, synchronous infection of CD4 T cells purified from human tonsils. When measured 18 hours decreased CCL2 release following HIV infection compared to non-targeted controls ( Figure 1D).
Knocking out cGAS trended to decrease CCL2 release but did not reach statistical significance.

96
Together, these results indicated that the innate DNA sensing pathway involving IFI16 and STING

164
To test whether CCR2/5+ cells were able to support latent infection, the cells were sorted as 165 above, then infected with a luciferase reporter virus in the presence of saquinavir (a viral protease 166 inhibitor) to prevent viral spread. The cells were then allowed to rest in the presence of ART to 167 establish latency, as previously described 43 . After five days in culture, the cells were stimulated raltegravir, to ensure measurement of post-integration latency. Luciferase expression was measured as a readout of viral reactivation. CCR2/5+ cells underwent similar HIV reactivation tonsil CCR2/5+ cells ex vivo.

175
We also measured the survival of CCR2/5+ cells following infection with both X4 and R5-tropic 176 virus. We found that the fraction that died following infection was quite similar to that of total tonsil

184
This model utilized R5-tropic virus allowing the establishment of a latent HIV reservoir in human 185 lymphoid tissue that persists despite ART 44 . Mice were treated with a blocking anti-CCL2 antibody 186 (n=35) or an isotype control antibody (n=34) 24 hours prior to and during repeated intrarectal HIV 187 inoculations that occurred daily for 5 days (see Figure 5A). The intrarectal infection protocol 188 employed was previously shown to produce ~50% infection rates 45 and was selected to

192
Mice were maintained on ART chow until week 12, and then switched to normal chow. Following the control antibody group exhibited detectable plasma viremia while no animals in the CCL2 total, 16 animals (47%) from the isotype control arm had detectable HIV, compared to 5 (14%) 198 from the CCL2-antibody treated group (Fisher's exact p= 0.0041). Together, these findings 199 suggested that treatment with the anti-CCL2-blocking antibody, but not an isotype-control

213
To quantify proviral levels, genomic DNA from sorted cells was isolated and subjected to ALU-

214
Gag droplet digital qPCR to detect integrated HIV provirus. This technique is based on a nested is used to detect HIV. Using this assay, we observed that CCR2/5+ cells contained significantly 217 higher levels of integrated HIV than CCR2/5-memory cells or naïve cells (Figure 6 B-D).

218
Additionally, by comparing the levels of HIV DNA in the sorted populations to that in the total population of sorted CD4 T cells, we were able to quantify the percentages of total HIV provirus CCR2/5+ population, while in others this percentage was lower. The median percentage of HIV 223 provirus within the CCR2/5+ cells was 37% ( Fig 6D). Importantly, the CCR2/5+ cells had the    264 phosphorylation and Type I IFN production 27 ; and further, that STING stimulates CCL2 production 265 by activating STAT6 28 . When sgRNAs targeting IFI16 and STING were deployed with Cas9 RNPs to knockout these genes in lymphoid CD4 T cells, a partial decline in CCL2 production following a primary activator of STING 29 , only trended to decreasing CCL2 production not reaching 269 statistical significance.

271
Of note, the IFI16 and STING knockouts only partially ablate production of CCL2 in the HIV-272 infected CD4 T cells. This likely is due to the fact that CCL2 production during HIV infection 273 occurs through multiple mechanisms. As noted above, HIV gp120 52-54 , Tat 55-57 , Nef 58 , and p17

CCR2/5+ cells are targets of HIV infection
We find that sorted CCR2/5+ cells from unstimulated tonsils are permissive to HIV infection 290 ( Figure 4). However, it is important to sort these cells prior to infection before R5-tropic viruses 291 induce a downregulation of surface CCR2 changing the cell's phenotype ( Figure 4D). We suspect 292 that this loss of CCR2 expression is due to viral envelope interactions with CCR2. Non-infected 293 bystanders (GFP-) as well as infected cells (GFP+) exhibit downregulation of CCR2 following 294 exposure to R5-tropic, but not X4-tropic NL4-3. In addition to the homology shared by CCR2 and 295 CCR5, previous studies have described CCR2 binding by HIV Env, and multiple HIV entry      Table S2.

403
Concentrated white blood cell preparations from healthy volunteers were obtained from Vitalant.

Virus strains
Fresh human tonsil tissue or splenic tissue was processed and cultured as previously described     Resources Table), diluted in blocking buffer (1:1,000) and incubated with rocking overnight at 501 4°C. The following day, membranes were washed in PBS-T and incubated with HRP-conjugated 502 secondary antibody diluted 1:10,000 in blocking buffer for 2 hours at room temperature with 503 continuous rocking. After secondary incubation, the membranes were washed and TMB substrate 504 was added. The reaction was allowed to proceed for 5 minutes, and light was measured by 505 exposure to film.

518
Fluorescence-activated cell sorting was performed on unfixed live cells, following isolation and 519 surface staining as described above. Single cells from a lymphocyte scatter gate were gated as 520 "live" (zombie low/negative, see Key Resources Table), and subsequently as CD3+/4+ T cells.

527
Mass cytometric analyses was performed as previously described 68 . Cells were isolated from 528 fresh PBMCs or HLAC as described above and stained with a panel of lanthanide metal-529 conjugated antibodies (See Table S1). Antibody staining was performed in a volume of 100 µL

571
Depending on the available sample volume, the detection limit was 375 cp/mL during follow-up                         Cas9RNPs and two independent guide RNAs (gRNAs) were used to knockout IFI16, STING and cGAS protein expression in primary tonsil CD4 T cells (NT: non-targeting gRNA control). (D) CCL2 protein production was significantly diminished (but not eliminated) in HIV-infected cultures when IFI16 or STING were knocked out. Supernatants were harvested 24 hours after infection for MSD analysis of CCL2 production. Fold increase versus uninfected control is shown for each knockout culture. Experiments were performed with 4-6 independent tonsils. Error bars denote SEM; ANOVA with Tukey's multiple comparisons, * p < 0.05, ** p < 0.01, *** p < 0.001.   To assess the ability of tonsil CCR2/5+ cells to support latent infection, a previously described primary CD4 T cell latency model was employed 43 . Cells sorted as in panel B were spinoculated with NL4-3-luciferase reporter virus and cultured in the presence of ART for 5 days. Cells were then reactivated with anti-CD3/28 beads for 24 hours. Virus production after reactivation was measured by quantitating luciferase activity. Data are presented as fold increases in stimulated cells relative to unstimulated cells. Experiments involved the analysis of 4-8 independent tonsil preparations. Error bars denote SEM; ANOVA with Tukey's multiple comparisons, * p < 0.05, ** p < 0.01, *** p < 0.001.  (Bottom) Cells were gated for CCR2/5 expression as in Figure 2. Shown here are a panel of markers which were different or similar between lymphoid tissue (blue) and blood-derived (red) CCR2/5+ cells.