Kinetics of Early Innate Immune Activation during HIV-1 Infection of Humanized Mice

Human immunodeficiency virus type 1 (HIV-1) infection is restricted to humans and some nonhuman primates (e.g., chimpanzee and gorilla). Alternative model systems based on simian immunodeficiency virus (SIV) infection of macaques are available but do not recapitulate all aspects of HIV-1 infection and disease. Humanized mice, which contain a human immune system, can be used to study HIV-1, but only limited information on early events and immune responses is available to date. Here, we describe very early immune responses to HIV-1 and demonstrate a suppression of cell-intrinsic innate immunity. Furthermore, we show that HIV-1 infection interacts differently with innate immune responses in blood and lymphoid organs.

model systems are available (e.g., simian immunodeficiency virus [SIV] [SIVmac]), these do not recapitulate all the complex host-pathogen interactions evolved by HIV-1, mostly because they present preadaptation strains (i.e., SIVcpz or SIVsm) or HIV-2 progenitors (5,6). Additionally, SIV and HIV-2 encode an additional accessory protein, Vpx, which alters disease progression by promoting replication in myeloid cells by blocking the action of the restriction factor SAM domain-and HD domain-containing protein 1 (SAMHD1) (7,8). Furthermore, in vitro models of HIV-1 persistence in T cells generally require preactivation of the cells, which results in their proliferation and altered phenotype. This is in stark contrast to infection in vivo, where any cellular activation is driven by endogenous processes (9). Even though more physiological model systems have been developed, including ex vivo tissue explant models (10,11) or cytokine-driven models of HIV-1 latency (12,13), they often rely on extensive manipulation of natural physiology, are often challenging to accurately control, and have not yet been demonstrated to be equivalent to naturally infected cells.
Humanized mice have been developed to bridge this gap, facilitating HIV-1 infection using human cells in vivo (14). These models largely rely on the ability of human hematopoietic stem cells (HSC) to utilize the murine bone marrow stem cell niche to reconstitute highly immunodeficient mice with all major human hematopoietic lineages (15). Even though advanced model systems incorporating implanted human fetal liver and thymus have been developed (16), the limited availability of human fetal tissue as well as ethical considerations make human HSC-engrafted humanized mice the most widely used and tractable model (15). Even though many humanized mouse models have been used to study HIV-1 infection, currently, no data are available on very early events in the path to HIV-1 persistence. In addition, early innate immune activation by HIV-1 through type I interferon (IFN) is still poorly understood, and even less is known in regard to the role of type III IFN in HIV-1 infection. Furthermore, it remains unclear to what extent HIV-1 infection activates (17)(18)(19) or interferes with (20)(21)(22) innate immune activation and IFN-based signaling. Even though many IFN-stimulated genes (ISGs) with antiretroviral activity have been described (3,23), it remains unclear to what extent these are present, induced, repressed, or evaded during HIV-1 infection in vivo.
Here, we demonstrate that ultrasensitive detection can be used to distinguish very early eclipse and burst phases preceding the onset of CD4 T cell loss in humanized mice. These phases give clear insight into the differential innate immune activation stages involved in establishing HIV-1 persistence as well as illustrating how HIV-1 subverts host efforts to restrict infection. This offers valuable insight into how HIV-1 interacts with the innate immune system early during infection and explains why common IFN-stimulated HIV-1 restriction factors may be limited in their capacity to control infection.

RESULTS
HIV-1 infection exhibits a multiphasic kinetic, stabilizing after 3 weeks. Despite many studies utilizing humanized mice to model HIV-1 infection (15), little information was available on very early virological and immunological events following infection. The generation of humanized mice by intrahepatic injection of HSC into sublethally irradiated NOD.Cg-Rag1 tm1Mom IL2rg tm1Wjl /SzJ (NRG) mice resulted in multilineage engraftment of human immune cells in all major organs within 3 months, generating HIS mice ( Fig. 1a and b). This included CD3 ϩ T cells, CD19 ϩ B cells, as well as lineagenegative cells (Fig. 1c). CD3 ϩ CD4 ϩ T helper cells, the main target for HIV-1 infection, as well as CD3 ϩ CD8 ϩ cytotoxic T cells were present at physiological frequencies, and despite their lower repopulation, both CD11b ϩ macrophages and CD14 ϩ monocytes were present in all major organs (Fig. 1d).
Infection of these mice with the CCR5-tropic HIV-1 recombinant virus NL4.3 (BaL Env) resulted in stable viremia, as determined by ultrasensitive reverse transcriptase quantitative PCR (qRT-PCR) (Fig. 2a). Notably, this method allowed us to observe three phases during infection: an eclipse phase up to day 4 followed by a burst phase of viral replication before expansion of HIV-1 RNA copies in the plasma of mice 4 to 9 days before stable peak viremia was reached (Fig. 2a). Nonhumanized control cohorts were utilized to demonstrate rapid virological decay in the absence of human cells (Fig. 2b). Strikingly, comparing the initial decay of HIV-1 RNA in the sera of HIV-1-infected humanized and nonhumanized mice within the first 3 days of the eclipse phase, the serum HIV-1 RNA half-life was significantly prolonged in mice containing human immune cells (Fig. 2c). At day 22 postinfection, de novo-produced infectious HIV-1 particles were detectable in the serum of infected mice, as determined by titration on GHOST cells (Fig. 2d). p24-expressing HIV-1-infected CD3 ϩ CD4 ϩ T cells were detectable with a peak at 7 days postinfection before contracting to stable levels, detected in both peripheral blood and splenic tissue of these mice ( Fig. 2e to g). This shows that HIV-1 infection in humanized mice can be assessed at very early time points postinfection and that HIV-1 viremia initially contracts prior to the establishment of stable serum viremia. Among the hallmarks of HIV-1 infection is the resulting longitudinal loss of CD4 T cells (24). Critically, and similar to HIV-1 infection in humans, in this model, HIV-1 infection resulted in an overall reduction of total CD4 T cell counts in peripheral blood and spleen (Fig. 2h to j). This progressively worsened and, in some respects, therefore recapitulated progression to AIDS in human infection ( Fig. 2h to j). As expected, the infection perturbs the total human leukocyte engraftment compared to uninfected mice, leaving the murine CD45-expressing leukocyte population unchanged (Fig. 2k). This demonstrates that the trajectories of HIV-1 infection in humanized mice bear similarities to those observed in humans.
HIV-1 distinctly suppresses early innate immune activation. Reports of early HIV-1 infection in patients suggest a strong peripheral type I IFN response to infection (25). However, little information was available on the kinetics and the correlation of peripheral circulation with lymphoid responses. Thus, it remained unclear whether HIV-1 is able to suppress early host responses. Since HIV-1 infection is associated with immune hyperactivation (26), we aimed to measure the extent of innate immune activation throughout the establishment of HIV-1 persistence in humanized mice. To this end, we evaluated the induction of type I and III IFN at early, intermediate, and late time points. Interestingly, at only 24 h postinfection, HIV-1 specifically downregulated type I IFN (IFN-␣ and IFN-␤) expression in peripheral blood lymphocytes ( Fig. 3a and b), while neither of the type III IFNs was expressed ( Fig. 3c and d). During the viral eclipse phase up to day 7, HIV-1 infection was associated with marked elevations of both type I and type III IFN in peripheral blood lymphocytes ( Fig. 3a to d). Once HIV-1 infection was fully established at day 35 postinfection, type I and III IFN returned to baseline levels in the peripheral circulation ( Fig. 3a to d).
To evaluate what mechanism drives the observed peripheral and splenic production of type I and III IFN at the peak of innate immune activation, prior to the HIV-1associated depletion of CD4 T cells or macrophages ( Fig. 4a and d), we evaluated the presence of pattern recognition receptors (PRR) able to sense viral RNA on peripheral as well as splenic CD4 T cells ( Fig. 4b and c) and macrophages ( Fig. 4e and f) 10 days following infection with HIV-1. At this point, HIV-1 infection resulted in a trend toward suppression of Toll-like receptor 3 (TLR3) protein expression on splenic but not peripheral CD4 T cells, whereas peripheral TLR7 and TLR9 protein expression was elevated after HIV-1 infection of both CD4 T cells and macrophages (Fig. 4b, c, e, and f).
To determine whether this differential TLR profile may be involved in the recognition of HIV-1 infection, we performed phosflow pathway analysis (39) of nuclear factor B (NF-B), TANK-binding kinase 1 (TBK1), and interferon regulatory factor 3 (IRF3) in peripheral and splenic CD4 T cells and macrophages 10 days following infection of humanized mice with HIV-1 (Fig. 4g to m). While splenic CD4 T cells respond to HIV-1 infection by inducing phosphorylation of the NF-B subunit p65 at serine 536, this was not observed in peripheral CD4 T cells or macrophages ( Fig. 4h and k). Even though many PRR pathways involve the phosphorylation of TBK1, only macrophages, but not CD4 T cells, exhibited elevated mean fluorescence intensities of phosphorylated TBK1 (Ser172) ( Fig. 4i and l). Peripheral but not splenic CD4 T cells additionally showed a trend toward elevated levels of phosphorylated IRF3 (Ser396) (Fig. 4j and m).
Taken together, these results indicate that the HIV-1-induced induction of TLR7 and TLR9 expression in PBMC may be at least partially responsible for peripheral interferon responses.
HIV-1-associated early immune activation is driven by both T cells and macrophages in peripheral but not lymphoid tissue. Since HIV-1 infection resulted in a very compartmentalized, early IFN response in PBMC but not splenic tissue, we assessed the distribution of the respective receptors for type I and III IFN 10 days following HIV-1 infection by flow cytometry (Fig. 5a to d). As expected, the majority of macrophages in either PBMC or spleen expressed either IFNAR1, IL28RA, or IFNAR1 and IL28RA ( Fig. 5b and d), whereas only a minority of CD4 T cells expressed predominantly IFNAR1 but not IL28RA (Fig. 5a and c). To assess which cell populations responded to the production of type I and III IFN, we next examined the level of STAT1 phosphorylation at tyrosine 701, which is indicative of activated JAK/STAT signaling downstream of type I and III IFN receptors ( Fig. 5e and f). This showed that macrophages in PBMC but not in spleen exhibited activated JAK/STAT signaling in response to HIV-1 infection, indicative of the observed production levels of type I and III IFN ( Fig. 5e and f). This also directly correlated with the production of ISGs, as exemplified by tetherin/BST2, which was preferentially induced by HIV-1 infection in CD4 T cells and macrophages in peripheral circulation but was not induced in splenic CD4 T cells or macrophages.
This compartmentalized IFN response additionally translated to the selective induction of well-known ISGs and HIV-1 restriction factors, including myxovirus resistance protein 2 (MX2), IFITM1, and IFI44, in PBMC at the peak of the initial HIV-1 replication burst (Fig. 6a to e). Surprisingly, however, both at the eclipse phase of infection as well as during late-stage chronic infection, HIV-1 infection suppressed baseline levels of IFI44 and showed similar trends with SAMHD1 and MX2 expression, indicating that HIV-1 actively shapes its environment.
Taken together, these results demonstrate that during early establishment of HIV-1 persistence in this model, innate immune activation in different compartments is highly dynamic, exhibiting signs of HIV-1-associated innate immune evasion and repression of expression of key HIV-1 restriction factors.
Exogenous type I IFN can override HIV-1-induced repression of restriction factors only in peripheral circulation. Based on recent experimental studies of SIV infection of macaques, it is possible that type I IFN accelerates HIV-1 disease progression (27,28). However, details on the responsiveness of HIV-1-infected cells to IFN and the subsequent expression of IFN-induced restriction factors are still lacking. To evaluate whether chronic HIV-1 infection durably suppresses the induction of IFN responses and HIV-1 restriction factors, we treated HIV-1-infected humanized mice with 1,000 IU/ g/day exogenous recombinant IFN-␣ (rIFN-␣) for five consecutive days. This treatment resulted in a significant reduction of circulating serum HIV-1 RNA levels ( Fig. 7a and b), indicating that the elicited inflammatory responses partially inhibited HIV-1 replication. Notably, administration of rIFN-␣ in HIV-1-infected humanized mice exacerbated the observed HIV-1-associated CD4 T cell depletion, indicating that ISGs and proinflammatory cytokines induced by type I IFN contribute to HIV-1-associated pathogenesis (Fig. 7d).
As expected, neither IFN-␣, IFN-␤, nor IFN-1 mRNA expression was elevated by rIFN-␣ (data not shown), whereas IFN-3, which is itself IFN stimulated, was induced by 3 log 10 units at the end of treatment ( Fig. 7e and j). rIFN-␣ treatment resulted in statistically significant increases of both SAMHD1 and MX2 in PBMC of HIV-1-infected humanized mice, whereas IFITM1 and IFI44 remained unchanged (Fig. 7f to i). Notably, despite a Ͼ2-log 10 induction of IFN-3 in response to rIFN-␣ treatment in splenocytes of HIV-1-infected humanized mice, HIV-1 restriction factors, including SAMHD1, MX2, IFITM1, and IFI44, remained unchanged or were downregulated by rIFN-␣ (Fig. 7k to n). This indicates that HIV-1 infection does not result in suppression of IFN responsiveness in peripheral circulation but induces a compartmentalized interference with the induction of HIV-1 restriction factors in lymphoid tissue.

DISCUSSION
Chronic HIV-1 infection is associated with immune hyperactivation. Even though the kinetic of natural HIV-1 infection has been extensively studied in the past, limited information is available on very early events following infection. Even though surrogate models for HIV-1 infection have been used in the past to dissect this, vast genetic   serum RNA loads to those of mice without humanization. This far precedes the significant loss of CD4 T cells, which becomes notable only after 1 month of infection.
Strikingly, HIV-1 is able to suppress baseline peripheral IFN and ISG levels very early following infection, before HIV-1 RNA levels become detectable in peripheral circula-  tion. This early suppression may contribute to the spread of HIV-1-infected cells, since key HIV-1 restriction factors, such as SAMHD1 and MX2, are present in host cells at lower levels. Only 1 week following infection, when serum HIV-1 RNA is already detectable, there is a significant peripheral type I and III IFN signature. However, the lymphatic system, as evidenced by measurements in the spleen, does not experience type I IFN production and instead exhibits a type III IFN signature. This absence of splenic type I IFN responses coincides with a trend toward reduced expression of the RNA sensor TLR3 on splenic CD4 T cells. Additionally, in contrast to peripheral CD4 T cells, HIV-1 infection does not induce the expression of TLR7 in splenic CD4 T cells or macrophages. This suggests a skewing of HIV-1-induced signaling toward proinflammatory NF-B, rather than IFN-inducing IRF3 signaling in T cells and macrophages in the spleen. This could contribute to the pyroptosis-induced CD4 T cell loss characteristic during HIV-1 infection (29). In contrast, peripheral CD4 T cells show a trend toward increased phosphorylation and activation of IRF3. This may, at least in part, explain the observed differential and compartmentalized type I and III IFN production. This is furthermore underpinned by the absence of an ISG response in the spleen, whereas multiple ISG and HIV-1 restriction factors are upregulated on the mRNA and protein levels in the peripheral circulation. This absence of early IFN responses explains why HIV-1 exhibits an early reservoir expansion in the near absence of restriction by host factors. Notably, even after the early peripheral induction of IFN and ISG responses, persistence of HIV-1 is associated with reduced expression of select ISGs and restriction factors in peripheral blood, which further underpins the ability of HIV-1 to evade host restriction. This suppression of key HIV-1 restriction factors may render the host more susceptible to systemic dissemination of HIV-1 infection.
Despite this evasion mechanism, PBMC retain their responsiveness to type I IFN, as exogenous treatment with recombinant IFN-␣ results in the induction of SAMHD1, IL28b, and MX2. However, splenic cells of HIV-1-infected humanized mice, upon treatment with recombinant IFN-␣, display an inverse response to type I IFN, resulting in repressed restriction factor expression. In particular, SAMHD1 is significantly downregulated after IFN treatment, suggesting that HIV-1 may have evolved means to facilitate its dissemination in lymphatic tissue despite immune activation. This cannot be explained by a potential hurdle for rIFN-␣ to penetrate to the spleen upon exogenous administration, since IL28b, which itself is IFN stimulated, is readily detectable in the spleen. Surprisingly, another study using BLT mice, which exhibit a more functional adaptive immune system with the T cells' ability to mount protective antigen-specific immune responses, demonstrated that CD4 T cell loss is delayed and that MX1 levels are chronically elevated (30). We observe very similar elevations of MX1; however, MX1, in contrast to MX2, has not been shown to restrict HIV-1 infection, and a comparative analysis in humanized mice has not been previously performed (31). Thus, the different kinetics of CD4 T cell loss may be due to antigen-specific T cell responses against HIV-1 infection releasing IFN-␥. Additionally, another study demonstrated that blocking type I IFN signaling results in T cell recovery during HIV-1 infection, supporting the notion that type I IFN might be associated with disease progression (32).
Taken together, our data show that IFN and ISG responses during acute HIV-1 infection in humanized mice are highly dynamic, exhibiting evidence for early viral suppression and peak recognition of HIV-1 by the innate immune system during the initial burst of HIV-1 replication. The trend observed indicating reduced TLR3 expression on splenic CD4 T cells upon HIV-1 infection could additionally contribute to the lack of HIV-1 RNA recognition. Since most of the initially induced HIV-1 restriction factors are returning to baseline or below-baseline expression levels during chronic HIV-1 infection, this could explain why HIV-1 is largely uncontrolled at later stages during infection. The induction of other ISGs without direct antiretroviral action could potentially result from other activation pathways and does not necessarily reflect the responses of ISGs in general. Indeed, CXCL10, which is commonly used as a marker for IFN stimulation and was shown to be elevated in untreated HIV-1-infected patients (33), was recently shown to be induced via TLR7/9 activation (34) rather than through JAK/STAT signaling. Most strikingly, the observed very early suppression of IFN and ISG responses in vivo has not been observed thus far. This early interference of pathogen sensing and IFN activation by HIV-1 may function as a viral strategy to establish an early reservoir and to eliminate restriction factor roadblocks preventing the establishment of HIV-1 persistence. The mechanism by which HIV-1 infection results in the transcriptional downregulation or destabilization of IFN mRNA remains elusive. However, many host proteins as well as viral accessory proteins (e.g., Vpr) are directly packaged within the HIV-1 virion and could contribute to this early downmodulation of the IFN system. This suggests that careful future analysis of single-cell transcriptional responses in vitro and in vivo is required to delineate which restriction factors are present and functional in HIV-1-infected as well as -uninfected bystander cells.

MATERIALS AND METHODS
Generation of humanized mice. NOD.Cg-Rag1 tm1Mom IL2rg tm1Wjl /SzJ (NRG) mice were obtained from The Jackson Laboratory and housed and bred at the Imperial College London CBS animal facility. Human CD34 ϩ hematopoietic stem cells (HSC) were isolated from human fetal livers as previously described (35). Less-than-4-day-old NRG mice were sublethally irradiated with 100 cGy and 4 h later were intrahepatically injected with 1 ϫ 10 5 purified CD34 ϩ HSC. Each experimental cohort was designed with groups comprising equal engraftment levels and numbers of animals of each gender and more than one HSC donor (Table 1).
HIV-1 infection of humanized mice and interferon treatment. Wild-type NL4.3 BaL was generated via transfection in 293T cells and then purified on a 20% sucrose cushion prior to injection, as previously described (36). Humanized mouse cohorts were generated based on equal numbers from different HSC donors, engraftment levels, age, and sex prior to infection. Humanized mice were intravenously injected with 2.1 IU of reverse transcriptase as measured by SG-PERT, as previously described (37), and mockinfected mice were subsequently injected with an equal volume of the vehicle. For the IFN-␣-treated cohorts, HIV-1-infected humanized mice, which displayed stable viremia, were intraperitoneally injected with 1,000 IU/g of body weight IFN-␣2 (Invitrogen, Paisley, UK) daily for 5 days, and blood and tissues were then immediately harvested.
Isolation of leukocytes from murine tissues. Murine peripheral blood was isolated longitudinally using tail vein bleeding. Peripheral blood mononuclear cells were isolated from whole blood using murine red blood cell lysis buffer (Alfa Aesar, MA, USA) according to the manufacturer's protocol. Fresh tissue isolated from humanized mice was washed in phosphate-buffered saline (PBS) and digested using 0.1% (wt/vol) collagenase digestion buffer for 30 min at 37°C. The digested tissue was then homogenized through a 70-m cell strainer and loaded onto a Ficoll gradient according to the manufacturer's protocol, and the leukocyte "buffy" layer was washed in PBS and stained for flow cytometry.
Isolation of RNA from peripheral blood mononuclear cells and splenic tissue. Fresh splenic tissue was harvested from humanized mice, directly stored in RNAlater stabilization solution (Life Technologies, Carlsbad, CA, USA), and frozen at Ϫ80°C. Upon RNA extraction, the samples were thawed and resuspended in the appropriate amount of RLT buffer containing ␤-mercaptoethanol according to the weight of the tissue or cell number. The tissue was then processed using the TissueLyser LT instrument (Qiagen, Manchester, UK) for 5 min at maximum speed. The RNA was processed using QIAshredder columns (Qiagen), followed by direct RNA isolation using the RNeasy minikit (Qiagen), according to the manufacturer's protocol.
Flow cytometry analysis. To evaluate engraftment, 50 l of whole blood was collected from a superficial tail vein 12 weeks following intrahepatic injection, and PBMC were extracted as described above. The isolated cells were then stained using anti-mouse CD45-BV605 (BD Bioscience [BD], NJ, USA) and anti-human CD45-allophycocyanin (APC) (BD), and engraftment was measured by the number of human CD45 ϩ cells in the total leukocyte population.