Myeloid-Derived Suppressor Cells Confer Infectious Tolerance to Dampen Virus-Induced Tissue Immunoinflammation

Infection-induced inflammation could potentially disrupt corneal immunoprivilege, a mechanism that ensures visual acuity in most terrestrial vertebrates (1). This can potentially lead to an immunoblinding disease necessitating corneal transplantation to restore vision. Such procedures also have limited long-term benefits because of the persisting virus that, upon reactivation from latency, can cause the disease. HSV type 1 (HSV1) infection triggers the onset of a severe inflammation known as herpetic stromal keratitis (HSK), a leading etiology of infectious blindness worldwide (1, 2). In humans, HSK results when the virus reactivates from trigeminal ganglion and transports through a retrograde mechanism to the corneal stroma to elicit an immune reactivity (3, 4). However, a primary HSV1 ocular infection in mice induces a series of molecular and cellular events leading to HSK (2, 5). Innate immune cells are recruited soon after the infection and CD4⁺ T cells of Th1 or Th17 phenotype serve as the major orchestrator of chronic keratitis lesion (6–8). The infiltration of CD4⁺ T cells in corneal stroma becomes evident at 7 d postinfection (dpi) but peaks between 14 and 21 dpi (3, 9, 10). However, an early infiltration is also recorded in a study (11). CD4⁺ T cells further enhance the infiltration of innate immune cells that contributes to a protracted and potentially blinding inflammatory reaction. Several immunosuppressive mechanisms that include the activity of regulatory T cells (Tregs) and various anti-inflammatory molecules, a constellation of physiological soluble mediators in the ocular fluids, and the existing anatomical barriers help control excessive inflammation (2–4, 12, 13). Different modalities have been explored to mitigate HSK lesions. Examples include enhancing the function of Treg, blocking the infiltration of effector cells to corneal tissues and neutralizing proinflammatory cytokines (12–15). However, studies on the regulatory innate immune cells such as the myeloid-derived suppressor cells (MDSCs) in virus-induced inflammatory diseases including HSK are scanty (16, 17). Such studies would not only help us better understand the pathogenesis of virus-induced inflammation, but also provide new angles of attack to help manage immunopathological diseases. Furthermore, boosting the response of MDSCs could potentially serve as an alternative strategy to dampen HSK lesions. This is because the information on immunopathogenic versus protective epitopes is still lacking, and a promotion of innate immune regulatory mechanisms might make such information redundant (18, 19).

We investigated the role of MDSCs in HSK pathogenesis and tested the feasibility of using in vitro–differentiated MDSCs as a cell-based therapy to modulate the severity of HSV1 caused ocular lesions and the potential underlying cellular mechanisms. We demonstrated that CD11b⁺Gr1^lo-int cells were expanded in the lymphoid organs of HSV1 infected animals in early stages of infection. Isolated CD11b⁺Gr1^lo-int cells, but not CD11b⁺Gr1^hi cells, suppressed not only the proliferation of stimulated CD4⁺ T cells, but also their cytokine production. Similarly, in vitro–generated CD11b⁺Gr1^lo-int cells efficiently inhibited the proinflammatory functions of stimulated CD4⁺ T cells predominantly via contact dependent mechanisms. The animals receiving in vitro–generated MDSCs before or after ocular HSV1 infection controlled their HSK lesions and angiogenic response better as compared to controls. The donor cells were majorly recovered from the lymphoid organs of recipient animals, a suggestion for their major activity at the immune inductive sites. Furthermore, MDSCs promoted endogenous Foxp3⁺ Treg responses not only by stabilizing Foxp3 expression in precommitted Foxp3⁺CD4⁺ T cells, but also by a de novo conversion of suppressive Foxp3⁺CD4⁺ T cells from Foxp3⁻CD4⁺ T cells. Promoting MDSCs response, therefore, could serve as a multipronged strategy to confer infectious tolerance to the host and efficiently controls infection-induced protracted inflammation.

BALB/c, C57BL/6, Foxp3/GFP knock-in (KI), and OT II Rag^−/− × TCR transgenic mice were procured from The Jackson Laboratory. The animals were housed and bred in the individual ventilated cages in the small animal facility for experimentation of Indian Institute of Science Education and Research Mohali. Six- to eight-week-old female mice were used for experiments. Different strains of HSV1 were grown and titrated using Vero cells as described earlier (13). The virus was stored at −80°C until further use. Institutional Animal Ethics Committee approved all the protocols, and the experiments were performed strictly in accordance with the approved protocols.

Mice were anesthetized by i.p. injection with 2,2,2-Tribromoethanol (Avertin) prior to corneal infection. The corneas were scarified using a 32-gauge needle, and 2-μl drop containing the indicated doses of HSV1 strains were applied to the scarified corneas. The eyes were examined for the lesion and angiogenesis development at different dpi using a slit-lamp biomicroscope as described earlier (3). The scoring of lesions was performed as per the following scheme: 0, normal cornea; +1, mild corneal haze; +2, moderate corneal scarring and opacity; +3, severe corneal opacity but iris clearly visible; +4, opaque cornea with corneal ulceration; and +5, cornea rupture and necrotizing keratitis. The scoring or angiogenesis was performed as per the following events. A grade of 4 was given to one quadrant of eye lobe representing a centripetal growth of 1.5 mm toward the center. The scores of all four quadrants were summed to derive the angiogenesis index. Therefore, the range of angiogenic scores was from 0 to 16, in which 0 represents normal cornea and 16 represents maximum vascularization.

Cornea, spleen, lymph nodes (LNs), and bone marrows were collected from different groups of euthanized mice postinfection for immune analysis. The animals were perfused with 10–15 ml of cold sterile PBS before collecting lymphoid organs and the corneal tissues. Corneal tissues were harvested and processed by liberase (2.5 mg/ml) digestion for 45 min at 37°C in 5% CO₂ incubator. The prepared single-cell suspensions from lymphoid organs and corneal samples were counted for cell numbers, and 1 × 10⁶ cells were stained for flow cytometric analysis. All staining steps were conducted at 4°C and Fc block were performed prior to the respective Abs staining. The stocks of Abs (1 mg/ml) were diluted in a 1:100 ratio in FACS buffer (2% FBS in PBS), and 50 μl of the final volume were added to each sample. The samples were incubated on ice for 30 min, and subsequently, three washings were performed using FACS buffer. After washings, cells were resuspended in 200 μl of FACS buffer for acquisition using a BD Accuri Flow cytometer and were then analyzed using FlowJo software v10. For analysis, FACS plots were gated based on fluorescence minus one controls. Counting the numbers of cells in gated population in FACS plots in FlowJo software and extrapolating the data from total cell counts generated the represented cell counts for indicated markers (16).

All the Abs used were purchased from BD Biosciences, Tonbo Life Sciences, BioLegend, and eBiosciences. The Abs used were against CD16/32, CD45 (clone 30-F11), CD11b/FITC (clone M1/70), CD4/PE (clone GK1.5), Ly-6C/PerCP Cy5.5 (clone AL21), Ly-6G APC (clone 1A8), Gr1/APC (clone RB6-8C5), CD8/PerCP Cy5.5 (clone 53-6.7), Foxp3/APC (clone 3G3), PD1/PE (clone J43.1), CD80/PE (clone 16-10A1), F4/80/PE (clone T45-2342) and CD124/PE (clone mIL4R-M1), latency-associated peptide (LAP)/PerCP Cy5.5 (clone TW7-16B4), CD73/APC (clone TY11.8), and IL-17A FITC (clone 17B7). The analysis of inflammatory cells in corneal samples was performed in gated CD45⁺ cells. All Ab dilutions were prepared in FACS buffer. The reagents such as PMA, ionomycin calcium salt, and paraformaldehydes were purchased from Sigma-Aldrich. RPMI 1640, FBS, and penicillin/streptomycin were obtained from Life Technologies. H&E was from HiPrep, and OCT compound was obtained from Fisher HealthCare.

Mice were euthanized using CO₂ in euthanasia chambers and sprayed with 70% ethanol. The long bones were dislocated, cleaned up from muscles, and washed with 70% ethanol. Thereafter, multiple washing using sterile cold PBS were performed. The bones were then placed in a petri dish containing ice-cold complete RPMI 1640, and the bones were flushed using the same media for making a single-cell suspension (20). BM cells were incubated with 1 ml of RBC lysis buffer for 5 min at room temperature. The cells were then washed in PBS twice and resuspended in complete RPMI medium. The cells were counted using a hemocytometer, and 3 × 10⁶ cells were added to a 25-mm² flask in 5 ml complete RPMI 1640 supplemented with 20 ng/ml of each of the cytokines: IL-6, IL-4, and GM-CSF in humidified chambers of CO₂ incubator at 37°C (21). Some cells were also grown in RPMI 1640 without any cytokines and were used as controls. After 4 d, cells were washed, and some cells were stained for CD11b, Gr1, and Ly-6C for phenotypic characterization. Remaining cells were used for the downstream experiments.

CD11b⁺Gr1^hi and CD11b⁺Gr1^lo-int cells as shown in the gates of FACS plots from splenocytes of infected animals, in vitro–generated MDSCs; Foxp3/GFP^-ve CD4⁺ or Foxp3/GFP^+ve CD4⁺ T lymphocytes from pooled LN and spleen cells were sorted by BD FACSAria Fusion. The presort and postsort samples were analyzed for determining the purity of desired cell population.

CD4⁺ T cells were purified using mouse CD4⁺ T isolation kits from Invitrogen. The purified cells were resuspended in 1 ml of PBS. One milliliter of 10 μM CFSE solution was prepared and was added dropwise to the purified CD4⁺ T cells. The cells were gently mixed after every drop and were then incubated at room temperature for 10 min. After incubation, 2 ml of FBS was added, and the samples were centrifuged at 300 × g at room temperature for 5 min. The cell pellet was resuspended in complete RPMI 1640 and centrifuged for 5 min. The labeled cells were washed three times and further suspended in complete RPMI 1640.

Purified and labeled GFP⁻CD4⁺ T cells obtained from the lymphoid organs of C57BL6 (Foxp3/GFP KI) mice were cultured with sorted MDSCs that were either FACS sorted from the spleens samples of HSV1-infected mice at 15 dpi or generated in vitro from bone marrow precursor cells in different dilutions in a 96-well, flat-bottom plate precoated with anti-CD3 Abs (1 μg/ml). Soluble anti-CD28 (0.5 μg/ml) was added, and the cultures were incubated for 3 d. The cells were then harvested for flow cytometric analysis, and the supernatants were stored for cytokine measurement by ELISA. For measuring whether the suppressive activity of MDSCs against stimulated CD4⁺ T cell targets is contact dependent, we used transwell assays plates. The effector cell population was added in the lower chamber of the wells precoated with anti-CD3 and anti-CD28 Abs. The MDSCs were added in the upper chamber. For measuring the suppressive effects of MDSCs against Ag-specific target CD4⁺ T cells, two types of assays were performed. First, we collected and pooled LN and spleen cells from RAG1^−/− × TCR transgenic mice or (OT II mice) and labeled them with CFSE. These target cells were then stimulated either with the OVA protein (200 μg/ml) or an I-A^b peptide OVA_323–339 (50 μg/ml). These cells were then added with the graded dosages of MDSCs, and the dilution of CFSE in gated CD4⁺ T cells was measured 3 d later. Similarly, MDSCs were OVA_323–339 peptide pulsed and extensively washed prior to their incubation with CFSE-labeled OT II cells. In the second set of experiments, the cells from draining cervical LNs (CLNs) of HSV1-infected mice were labeled with CFSE. The labeled LN cells were either incubated with MDSCs or freshly isolated bone marrow cells. The UV-inactivated HSV1 was added to the cells to provide Ags for the stimulation of T cells. Three days later, the dilution levels of CFSE in stained CD4⁺ T cells were analyzed using a flow cytometer.

In vitro–generated MDSCs were FACS sorted, and indicated numbers were transferred in either C57BL/6 or Foxp3/GFP KI mice before or after ocular HSV1 infection. The progression of disease in control and MDSCs recipient mice was assessed over time. At the end of the experiments, animals were sacrificed. Different cell populations isolated from the lymphoid organs and at corneal tissues were analyzed for different surface markers and cytokine production.

In vitro–generated MDSCs were FACS sorted and labeled with CFSE or CellTrace Far Red dye. A total of 3 × 10⁶ cells were i.v. injected before or after HSV1 infection in mice. Next day, posttransfer, different lymphoid organs such as the bone marrows, spleens, CLNs, iliac LNs, brachial LNs, and mediastinal LNs as well as corneas were collected from sacrificed mice. Single-cell suspensions were prepared to measure for the presence of CFSE-positive cells.

Single-cell suspension prepared from isolated spleens and LNs of HSV1 animals from different groups were analyzed for cytokine production. The cells were either added with UV-inactivated HSV1 or the PMA (10 ng/ml)/ionomycin (1 ng/ml). HSV1-pulsed cells were incubated overnight in humidified CO₂ incubator at 37°C, and in last 4 h, brefeldin A was added to induce an accumulation of cytokines. In PMA/ionomycin-stimulated cells, brefeldin A was added for the duration of 5 h. Cells were then stained for surface markers, such as CD4, using fluorochrome-labeled Abs. After surface staining, the cells were fixed and permeablized with BD FixPerm Kit and subsequently stained for the anti-cytokine Abs. The samples were acquired and analyzed by BD Accuri C6.

Sorted Foxp3⁻ or Foxp3⁺CD4⁺ T cells from Foxp3/GFP KI mice were stimulated with plate-bound anti-CD3 and soluble CD28 in the presence or absence of sorted MDSCs in graded doses for 3 d in humidified CO₂ incubator at 37°C. CD4⁺ T cells were analyzed for Foxp3/GFP expression after incubation period was over. In some experiments, the coculture of MDSCs and CD4⁺ T cells were added with anti–TGF-βR1 Ab (Sigma-Aldrich). To measure the suppressive activity of converted Foxp3/GFP^+ve cells, graded doses of Foxp3⁺ cells were added to CellTrace Far Red (Thermo Fisher Scientific)-labeled, OVA-stimulated OT II cells. The activation profile of the responder cells was measured 24 h later by flow cytometry.

Animals were sacrificed at different dpi, and corneal samples were fixed in 4% parafromaldehyde prepared in 10 mM PBS at 4°C. The samples were then dehydrated in a sucrose gradient from 5 to 20% in 10 mM PBS at room temperature and embedded in tissue blocks in the OCT compound. The blocks were frozen at −80°C. Six-micrometer tissue sections cut using a Leica cryotome, and the slides were stored at −20°C until further use. H&E staining was performed. The stained sections were dried at room temperature, and coverslip were mounted with medium. After drying, the sections were photographed on Leica DMi8 microscope, and the images were analyzed for size and scale bar using ImageJ software.

The representative data were shown from at least three independent experiments. The data obtained for different groups were analyzed either by ANOVA or Student t test using Graph Pad Prism software v5.03, as indicated in the figure legends. The levels of significance between different groups was determined as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and NS.

We monitored the progression of ocular disease in HSV1-infected mice. Additionally, the kinetics of CD11b⁺Gr1⁺ cells were also measured in the cornea and lymphoid tissues of infected mice at different dpi. A subset of CD11b⁺Gr1⁺ cells possess immunosuppressive activity and, therefore, are named as MDSCs (22–24). Infected animals progressively developed HSK lesions and angiogenic response (Fig. 1A–C). More than 80% of animals developed a lesion score of ≥2 by day 13. Majority of infected eyes demonstrated a lesion score of 5 and angiogenic score of 15 by 18 dpi (Fig. 1A–C). Histological analysis of H&E-stained corneal sections also revealed a progression in inflammation. Thus, the thickness of corneal stroma as well as the cellular infiltrates increased as the disease progressed from 2 to 18 dpi (Fig. 1D). We sacrificed mice on different dpi to isolate cornea and lymphoid organs for cellular analysis. The inflammatory leukocytes (CD45⁺ cells) infiltrated corneal tissues in high frequencies postinfection (Fig. 1E–J). As early as 2 dpi, ∼10% of total live cells recovered were leukocytes, and their frequencies and numbers per cornea increased gradually by 18 dpi (Fig. 1E–G). Among the gated CD45⁺ cells, two populations (CD11b⁺Gr1^hi and CD11b⁺Gr1^lo-int) were evident in corneal samples. The ratios were skewed toward CD11b⁺Gr1^lo-int cells early during the ocular infection at 6 dpi in comparison with those at subsequent times (i.e., 13–30 dpi). Thus, when the corneal inflammation was barely evident (i.e., a lesion scores of ≤1 at 6 dpi), the ratios were skewed toward CD11b⁺Gr1^lo-int cells both in frequencies and total numbers. However, when the lesion scores progressed to ≥3 between 13 and 30 dpi, the frequencies, as well as numbers, of CD11b⁺Gr1^lo-int cells remained similar, but those of CD11b⁺Gr1^hi cells increased greatly, suggesting for a dichotomy in the functionality of two populations (Fig. 1A, 1H–J). CD11b⁺Gr1^hi cells were also considered as neutrophils and were more frequent immediately after ocular HSV1 infection, and such cells were shown to exhibit antiviral activity (25, 26). A biphasic infiltration of neutrophils is clearly evident with the first wave infiltrating in corneal tissues at 2 dpi and the second wave at 6–7 dpi (3). The frequencies and the numbers of neutrophils (CD11b⁺Gr1^hi) as well as CD4⁺ T cells increased steadily in the corneal tissues up to 18 dpi (Fig. 1K–M). Similar results were obtained for infected C57BL/6 mice (Supplemental Fig. 1). CD11b⁺Gr1^lo-int cells also expanded in the spleens early after HSV1 infection, suggesting for an anti-inflammatory response soon after HSV1 infection, but the cues responsible for their recruitment are yet to be identified (Fig. 1N–P). The inverse correlation between the disease progression and CD11b⁺Gr1^lo-int cells recovered from corneal as well as splenic tissues indicated for their potential regulatory roles in the causation keratitis.

After measuring the kinetics of myeloid cells in corneal tissues and lymphoid organs of HSV1-infected mice, we phenotypically characterized CD11b⁺Gr1⁺ cells by analyzing the expression of a panel of molecules such as CD80, F4/80, CD124 (IL-4R), and PD1. Many of the molecules have been associated with the cellular functions (27). After HSV1 infection, a greater proportion of CD11b⁺Gr1^lo-int cells upregulated PD1 (up to 4-fold) and CD80 (up to 2-fold) (Fig. 2A, 2B). A small, but significantly increased, proportion of IL-4R⁺ cells were present in HSV1-infected mice. Similarly, the expression of some molecules such as PD1 and CD80 on a per-cell basis, as measured by mean fluorescence intensity (MFI) values, was also greater on the cells isolated from infected mice as compared with controls (Fig. 2C). Essentially, similar results were obtained for CD11b⁺Gr1^hi cells as well (Supplemental Fig. 2). Therefore, we further assessed the functionality of such cells using in vitro suppression assays. Both the populations were included in the analysis because the uniquely expressed molecules to mark MDSCs have not been well characterized (17, 22). Moreover, MDSCs are divided into monocytic and granulocytic lineage, and both are phenotypically characterized as CD11b⁺Gr1⁺ in mice (22). Both of these subsets, (CD11b⁺Gr1^hi and CD11b⁺Gr1^int cells) isolated from tumor-bearing mice, were shown to possess suppressive activity (21).

CD11b⁺Gr1^lo-int as well as CD11b⁺Gr1^hi cells collected from splenic tissues of HSV1-infected mice at 15 dpi were FACS purified. The purity of sorted cells was >90% (Fig. 2D). The purified CD11b⁺Gr1^hi or CD11b⁺Gr1^lo-int cells were added in graded numbers to coreceptor-stimulated, CFSE-labeled CD4⁺ T cells. In the absence of CD11b⁺Gr1⁺ cells, CD4⁺ T cells extensively diluted their CFSE content. More than 80% of CD4⁺ T cells divided, and a large majority of cells divided more than five times (Fig. 2E, 2F). CD11b⁺Gr1^lo-int cells inhibited the proliferation of CD4⁺ T cells in a dosage-dependent manner (Fig. 2E, lower panel, and Fig. 2G). At 1:2 ratio of CD4⁺ T cells and MDSCs, only 45% of cells divided. Among the divided cells, a minimal fraction of cells divided more than four times (Fig. 2E, lower panel). The culture supernatants collected from these wells were measured for cytokine accumulation. An addition of CD11b⁺Gr1^lo-int cells inhibited the production of both IL-2 and IFN-γ from stimulated CD4⁺ T cells (Fig. 2H, 2I). When CD11b⁺Gr1^hi cells were added to the stimulated CD4⁺ T cells, the extent of CFSE dilution, the percentage of divided cells, as well as the cytokine production remained unchanged (Fig. 2E, upper panel, Fig. 2F, and Supplemental Fig. 2D, 2E).

Our results, therefore, demonstrate that CD11b⁺Gr1^lo-int MDSCs expands during early stages of HSV1 infection and these cells are also recruited to corneal tissues.

We demonstrated that CD11b⁺Gr1^lo-int cells expanded in HSV1-infected mice were immunosuppressive but despite their expansion, HSV1-infected animals developed stromal keratitis lesions. Therefore, we reasoned that an inefficient induction of MDSCs could contribute to the development of ocular inflammatory lesions. Therefore, repleting infected animals with in vitro–generated MDSCs would help control HSK. MDSCs were generated in vitro from bone marrow precursor cells as described in 2Materials and Methods section (21). We obtained between 50 and 75% of live cells with a phenotype of CD11b⁺Gr1^int in different experiments (Fig. 3A). A significant proportion of these cells also expressed molecules such as PD1 (>34%), CD80 (>65%), and F4/80 (>94%) (Fig. 3B, 3C). We then FACS sorted CD11b⁺Gr1^int cells to measure their suppressive potential against coreceptor stimulated or an Ag driven proliferation of CFSE-labeled CD4⁺ T cells. The stimulated CD4⁺ T cells divided extensively (∼80% cells diluted CFSE), but an addition of CD11b⁺Gr1^int cells in graded numbers greatly inhibited their proliferation (Fig. 3D, 3E). At a 1:16 and 1:32 ratio, ∼40 and ∼60% cell diluted their CFSE content (data not shown). We also measured the cytokines produced by stimulated cells in the culture supernatants. The levels of both IL-2 and IFN-γ were reduced when MDSCs were added to stimulated CD4⁺ T cells (Fig. 3F, 3G). To explore the mechanism of suppression by MDSCs, we measured TGF-β level in the culture supernatants, as it was reported earlier that TGF-β could be involved (28). Similar levels of TGF-β were present in the culture supernatant of both the precursor bone marrow cells as well as differentiated MDSCs (Fig. 3H). When MDSCs were added to the coreceptor-stimulated CD4⁺ T cells, no significant differences in the levels of TGF-β were found, which might suggest for a minimal if any role of soluble TGF-β in causing the suppression of effector CD4⁺ T cells (Fig. 3I).

We further investigated whether the inhibitory effects of MDSCs was mediated by their secreted products or occurred in a contact dependent fashion using transwell assays. CFSE-labeled CD4⁺ T cells were stimulated through their coreceptors with in vitro–generated, FACS-sorted MDSCs either to establish their direct contact with CD4⁺ T cells or in the upper chamber of the transwell plates. In the absence of MDSCs, more than 55% of CD4⁺ T cells divided (Fig. 3J, 3K). Very few CD4⁺ T cells divided when in direct contact with the MDSCs (∼10% CD4⁺ T cells diluted CFSE) and only ∼30% of CD4⁺ T cells diluted CFSE when there was no direct contact. Furthermore, the frequencies of CD4⁺ T cells within each division were lower for the cells cultured in a transwell format as compared with those maintaining direct contact (Fig. 3I). These results suggest that the contact between MDSCs and CD4⁺ T cells is crucial for their optimal suppressive effects of MDSCs.

We also measured the suppressive activity of CD11b⁺Gr1^int cells using TCR transgenic CD4⁺ T cells that specifically respond to I-A^b–restricted peptide (OVA_323–339) of chicken OVA (Fig. 4A–D). The pooled cells from the lymphoid organs of OT II transgenic mice were labeled with CFSE and incubated with OVA protein, which were then cocultured in the presence of graded doses of in vitro–generated MDSCs. At 1:2 dilutions of MDSCs and OT II cells, <10% of CD4⁺ T cells divided (Fig. 4B). More of the CD4⁺ T cells divided when the frequencies of MDSCs were decreased in a graded fashion. MDSC to CD4⁺ T cell ratios of 1:4, 1:8, and 1:16 led to ∼20, ∼56, and ∼72% of divided CD4⁺ T cells, respectively (Fig. 4A, 4B). As the protein was added to the mixture of MDSCs and pooled OT II cells, there remains a possibility that MDSCs themselves might have participated in the uptake, processing and presentation of OVA protein to CD4⁺ T cells to induce their stimulation or tolerance. We therefore evaluated the Ag processing and presentation as well as peptide-displaying abilities of MDSCs that might result in the proliferation of OT II cells. MDSCs were either pulsed with the OVA protein for 2 h or its derivative OVA_323–339 peptide for 45 min. Pulsed MDSCs were added to OT II cells. In such experiments, MDSCs failed to induce the proliferation of OT II cells, suggesting for their inability to present Ags to induce an activation of T cells (Fig. 4C, 4D). Similar results were also obtained in earlier studies in which Ag-specific proliferation of OVA-reactive OT I cells was investigated (29).

The suppressive activity of MDSCs was also measured against HSV1-reactive CD4⁺ T cell targets. LN cells were pulsed with inactivated HSV1 in the presence or absence of graded doses of in vitro–generated CD11b⁺Gr1^int cells. Approximately 30% of gated CD4⁺ T cells diluted CFSE after 3 d of stimulation and CD11b⁺Gr1^int cells inhibited their proliferation in a dosage-dependent manner (Fig. 4E). Less than 4% of CD4⁺ T cells divided at 1:1 ratio of CLN cells and in vitro–generated MDSCs (Fig. 4E, 4D). These results suggested that MDSCs suppressed HSV1-reactive CD4⁺ T cells.

Taken together, our results demonstrate that in vitro–generated MDSCs inhibited CD4⁺ T cells in a contact-dependent manner when the latter population is stimulated either through coreceptors or that likely responded through TCRs.

Having established the suppressive activity of in vitro–generated MDSCs against stimulated CD4⁺ T cells, we investigated whether repletion of MDSCs during HSV1 infection modulates ocular inflammation. We transferred 1 × 10⁶ MDSCs in C57BL/6 mice and monitored the disease progression in infected mice up to 15 dpi (Fig. 5A). MDSC-recipient animals controlled their HSK lesions and angiogenic response better as compared with control mice (Fig. 5B–D). We also analyzed the distribution and phenotype of immune cells recovered from the corneal tissues and lymphoid organs of MDSC-recipient and control animals. Inflammatory leukocytes (CD45⁺ cells) and CD4⁺ T cells infiltrated to a lower extent in the corneal tissues of MDSCs recipient mice as compared with controls (Fig. 5E–G). The phenotypic analysis of draining LN cells revealed a significantly greater proportion (33 versus 40%) and numbers (1.5- to 2-fold more) of CD62L^hiCD4⁺ T cells in the draining CLNs of MDSC recipients as compared with control animals. Therefore, MDSCs might help sequester CD4⁺ T cells in the draining LNs and inhibit their migration to corneal tissues (Fig. 5H–J). We also measured anti-inflammatory Foxp3⁺ Treg and proinflammatory IL-17–producing Th17 cells in the lymphoid organs of these animals (30–32). Although the frequencies of Foxp3⁺CD4⁺ T cells were not significantly different in the MDSC-recipient mice as compared with control mice, their total numbers were more in MDSC-recipient animals (Fig. 5K, 5M). We also observed a reduction in the frequencies as well as the number of Th17 cells in the MDSC-recipient animals as compared with control, but the overall results were NS (Fig. 5N, 5O). These results suggested that MDSCs promoted the endogenous regulatory mechanisms and that a combined activity of transferred MDSCs and endogenous Foxp3⁺ Tregs helped control the disease progression, probably by interfering with the migration of effector cells to inflammatory sites. To track the distribution of transferred MDSC, we labeled these cells with CFSE and injected 3 × 10⁶ cells i.v. The distribution was measured 24 h later in different lymphoid organs. We observed detectable frequencies of transferred MDSCs in most of the lymphoid organs such as spleens, CLNs, mesenteric LN, brachial LNs, and iliac LNs as well as in the bone marrow of recipient animals. The transferred cells were recovered from most of the lymphoid organs, but their frequencies were 3-fold higher in spleens (Fig. 5P, 5Q).

We performed additional experiments to assess the therapeutic value of MDSCs in modulating HSK lesions by transferring MDSCs on day 4, a time that initiate the activation and recruitment of effector cells to infected corneal tissues (Fig. 6). A transfer of 1 × 10⁶ MDSCs in the clinical stage of HSV1 infection at 4 dpi significantly reduced the lesion scores as well as angiogenic response in recipient mice as compared with controls (Fig. 6B–D). The average lesion scores and angiogenesis scores in controls and MDSC recipients, as compared with the control group, were 3 versus 4.2 and 6 versus 12.5, respectively (Fig. 6C, 6D). Up to a 3-fold reduction in the frequencies of inflammatory leukocytes (CD45⁺ cells) and CD4⁺ T cells was observed in the corneal tissues of MDSC-supplemented mice in comparison with controls (Fig. 6E–G). We also observed a greater proportion of CD62L^hi in CD4⁺ T cells present in the draining CLNs of MDSC-recipient mice as compared with controls (Fig. 6H, 6I). The proportion of Foxp3⁺CD4⁺ T cells in the draining CLNs and those infiltrating to corneal sites were more in MDSC-recipient mice as compared with controls (Fig. 6J–L). A reduction in the proinflammatory Th17 cells in the CLNs of MDSC-recipient mice was also evident (Fig. 6M, 6N). Therefore, our results show that in vitro–generated MDSCs also reduce the severity of HSK when injected therapeutically at the beginning of clinical stage of HSK.

We then aimed to measure whether transferred MDSCs could be detected in corneal tissues or their activity was only concentrated in the immune inductive sites. We transferred 3 × 10⁶ CFSE-labeled MDSCs at a time when the clinical disease was clearly evident at 10 dpi. A day after the transfer, CFSE-positive cells were measured in inflamed cornea and the lymphoid tissues. Although we observed detectable frequencies of transferred MDSCs in most of the lymphoid organs such as spleens, draining LNs (CLNs), and nondraining LNs such as mesenteric LN, brachial LNs, and iliac LNs as well as in the bone marrow of recipient animals, such cells were barely detected in the inflammed corneal tissues, suggesting for their predominant role in the lymphoid organs (Fig. 6O, 6P, data not shown). Taken together, our results demonstrate that an administration of in vitro–generated MDSCs before or after ocular HSV1 infection dampens corneal inflammation, and the donor MDSCs predominantly homed to the lymphoid organs.

MDSCs recipients expanded their endogenous Foxp3⁺ Tregs. We therefore explored the possible cellular mechanisms by performing two types of in vitro experiments aimed at measuring the influence of MDSCs on already committed Foxp3⁺ Tregs as well as to assess their ability to cause a de novo conversion of conventional CD4⁺ T cells into Foxp3⁺ Tregs (Fig. 7). The inability of Foxp3⁺CD4⁺ T cells to retain Foxp3 expression is known as plasticity and that could potentially serve as a deterrent for cell-based anti-inflammatory therapies (33). FACS-sorted GFP⁺CD4⁺ T cells from Foxp3/GFP KI mice were stimulated using anti-CD3 and anti-CD28 Abs (Fig. 7A, upper panel). A large majority of stimulated GFP⁺CD4⁺ T cells lost their GFP expression within 3 d, but most CD4⁺ T cells retained Foxp3 expression when MDSCs were added to the cultures (Fig. 7A, upper panel, and Fig. 7B). The MFI values for GFP expression were also significantly enhanced with an addition of MDSCs (Fig. 7C). These results suggested that MDSCs helped precommitted Foxp3⁺CD4⁺ T cells retain Foxp3 and stabilized their phenotype. We also measured the ability of MDSCs to cause a de novo conversion of conventional Foxp3⁻CD4⁺ T cells into Foxp3⁺CD4⁺ T cells. FACS-sorted Foxp3/GFP⁻CD4⁺ T cells were stimulated via their coreceptor in the presence or absence of MDSCs. There was no contamination of Foxp3/GFP⁺ cells in the sorted population (Fig. 7A, lower panel). As the proportions of MDSCs increased, the induction of Foxp3 also increased in Foxp3/GFP⁻CD4⁺ T cells (Fig. 7A, lower panel, Fig. 7D). As many as 70% of GFP⁻ cells became GFP⁺ at 1:1 ratio of MDSCs and Foxp3/GFP⁻CD4⁺ T cells (Fig. 7D). In different experiments, we obtained up to a 90% conversion of Foxp3⁺ cells from Foxp3⁻CD4⁺ T cells (Fig. 8). We also recorded the expression levels of Foxp3/GFP in converted cells by measuring its MFI values and observed a dosage-dependent increase in Foxp3 expression (Fig. 7E). The suppressive activity of induced Tregs was measured against cognate peptide–stimulated OT II cells. Splenocytes from OT II TCR transgenic mice were labeled with CellTrace dye so as to identify these cells from converted Foxp3/GFP⁺ cells. Thereafter, labeled cells were pulsed with OVA_323–339 peptide and different ratios of MDSCs converted Foxp3⁺GFP⁺ T cells were added (Supplemental Fig. 3). Two days later, the expression of CD69 as an indicator of activation was measured in CellTrace–positive CD4⁺ T cells. We observed a dosage-dependent inhibition of effector CD4⁺ T cell targets by MDSC-converted Tregs (Fig. 7F, 7G).

Our results, therefore, demonstrate that MDSCs enhance endogenous Treg mechanisms not only by stabilizing their phenotype, but also by causing a de novo generation of suppressive Foxp3⁺CD4⁺ T cells from conventional Foxp3⁻ T cells.

We explored whether TGF-β signaling is involved in the MDSC-mediated conversion of Foxp3⁺ Tregs. Numerous studies have shown the role of TGF-β in Foxp3 induction in stimulated non-Foxp3⁺ T cells (34–36). We also observed a basal level of TGF-β in the culture supernatant of MDSCs and CD4⁺ T cells (Fig. 3I). We added a neutralizing anti–TGF-βR Ab in the cocultures of coreceptor-stimulated Foxp3/GFP⁻CD4⁺ T cells. In these experiments, ∼90% of CD4⁺ T cells became Foxp3⁺ (Fig. 8A, 8B). With the addition of anti–TGF-βR Abs, we observed no reduction in the proportions of induced Foxp3⁺CD4⁺ T cells, suggesting that the MDSC-induced conversion of Tregs is independent of TGF-β signaling (Fig. 8A, 8B). We also measured the expression of CD73, an ecto-5′-nucleotidase that, in conjunction with CD39, marks Tregs and is also involved in the suppressive activity by generating adenosine in the pericellular space (37, 38). Although a high proportion of activated CD4⁺ T cells expressed CD73, the molecule was downregulated on converted CD4⁺Foxp3⁺ Tregs, and the effect was independent of TGF-β signaling (Fig. 8A, 8C). We then measured the influence of TGF-β signaling in the MDSC-mediated stabilization of precommitted CD4⁺Foxp3⁺ Tregs. As observed in earlier experiments, the coincubation of MDSCs with Foxp3⁺ Tregs helped them retain Foxp3 expression independent of TGF-β signaling (Fig. 8D, 8E). The analysis of CD73 expression on precommitted Tregs, however, yielded surprising results. Foxp3⁺CD4⁺ T cells coincubated with MDSCs not only retained their Foxp3 expression, but also upregulated surface CD73, an effect that occurred largely independent of TGF-β signaling (Fig. 8D, 8F). These results, therefore, indicated for a dichotomy in the effect of MDSCs on precommitted and de novo–generated Tregs. We also measured the expression of LAP on MDSC-induced Foxp3⁺ Tregs. The expression of LAP by Tregs is also involved in their suppressive activity as well as converting non-Tregs into Tregs (39, 40). The induced Tregs expressed high levels of surface LAP as compared with stimulated CD4⁺ T cells (Fig. 8G). When the anti–TGF-βR Ab was included in the cocultures of MDSCs and stimulated CD4⁺ T cells, the expression of LAP on CD4⁺ T cells was significantly reduced (Fig. 8H). However, the level of Foxp3 expression on such cells was not influenced by the addition of anti–TGF-βR Ab (Fig. 8B). Therefore, our results demonstrate that MDSC-mediated de novo generation of Tregs may not require TGF-β, but the induced Tregs might require TGF-β signaling for their optimal function. Thus, the levels of LAP expressed by Tregs are shown to regulate their activity (39).

MDSCs play critical role in promoting the progression of various tumors and reducing the severity of some autoimmune diseases (22). These cells are also recruited in the course of some infections, but their contribution in the pathogenesis of viral infections remains less well explored. In this study, we investigated the role of MDSCs during ocular HSV1 infection and demonstrated an early recruitment of CD11b⁺Gr1^lo-int cells that suppressed effector CD4⁺ T cells in a predominantly contact-dependent manner. Repletion of HSV1-infected mice with in vitro–generated MDSCs controlled the severity of HSK lesions, and the donor cells preferentially homed to lymphoid organs. MDSCs promoted endogenous Foxp3⁺ Tregs at least by two mechanisms (i.e., by stabilizing Foxp3 expression in already-differentiated Foxp3⁺ T cells and also by converting a large number of conventional CD4⁺ T cells into Foxp3⁺CD4⁺ T cells). MDSC-enhanced Treg responses were likely independent of TGF-β signaling. Therefore, therapeutic value of MDSCs could be harnessed as a multipronged strategy that not only directly suppresses inflammatory response but also promotes endogenous regulatory mechanisms to dampen virus-induced inflammatory reactions.

MDSCs are categorized as monocytic and granulocytic cells based on their relative expression levels of Gr1 (21). Both CD11b⁺Gr1^hi and CD11b⁺ Gr1^int cells were shown to be suppressive during the progression of cancers (22). MDSCs were also expanded during infectious diseases, such as trypanosomiasis, leishmaniasis, and candidiasis (41–43). A few studies focused to decipher their roles during the pathogenesis of viral infections (17, 44, 45). Accordingly, a chronic infection caused by lymphocytic choriomeningitis virus clone 13 expanded a population of myeloid cells with potent suppressive activities that compromised antiviral immune response (44). Interestingly, the Armstrong strain of lymphocytic choriomeningitis virus that causes an acute infection failed to signal efficient MDSCs response. Therefore, MDSCs response could potentially decide the outcome of infection (44). Some reports investigating the role of MDSCs during HSV1 infection showed less-efficient responses of MDSC-like cells (25, 45, 46). However, polymorphomononuclear cells were efficiently expanded, and many such cells infiltrated corneal tissues that not only helped control virus growth, but also contributed to the disease progression (25, 45, 46). We aimed to identify the suppressor cell population based on their relative expression of myeloid and granulocytic markers. We also explored whether the disease outcome could be modulated by the therapeutic administration of MDSCs. We observed the suppressive activity in CD11b⁺Gr1^lo-int cells in comparison with CD11b⁺Gr1^hi cells against effector CD4⁺ T cells, the orchestrator of HSK lesion (Fig. 3). A lack of suppressive activity by CD11b⁺Gr1^hi cells could be attributed either to their short life span or inability to produce anti-inflammatory molecules. Earlier studies showed their proinflammatory properties (5, 25, 46, 47). A recruitment of CD11b⁺Gr1^lo-int cells upon HSV1 infection to corneal tissues inversely correlated with the disease progression, which suggested that the repletion of MDSCs could help reduce the disease severity. Therefore, we focused our further studies to explore the therapeutic value of in vitro–generated MDSCs from bone marrow precursor cells. HSV1 infection induces a potent inflammatory response not conducive for the efficient immunoregulation at least during the acute phase of response. Thus, cytokines such as IL-6, IL-1β, and TNF-α could dampen Treg function either by making them more plastic or altering their transcriptional program to facilitate migration to less-inflamed locations (33, 48). Therefore, cells of innate immune origin with suppressive functions could provide lasting regulatory effects. These cells are likely to be refractory to modulation by an inflammatory microenvironment (22). We observed that MDSCs controlled HSK lesions during the clinical stage of HSV1 infection. Moreover, therapeutic application of MDSCs could obviate the need of identifying epitopes for which a tolerance is induced. Our results also showed that MDSCs were not efficient in Ag presentation (Fig. 4).

Animals transferred with in vitro–generated MDSCs expanded Foxp3⁺CD4⁺ T cells. Therefore, MDSCs not only provided a quick anti-inflammatory response but also maintained its sustenance by promoting endogenous Tregs. The observations that MDSCs stabilized the expression of Foxp3 in already-committed Tregs could be clinically relevant. Many strategies to promote the generation of Tregs fail to stabilize their phenotype, and this property could be potentially damaging (49). A therapy with MDSCs might not suffer from such problems. Moreover, the ability of MDSCs to convert Foxp3⁻ T cells into Foxp3⁺ cells by their de novo differentiation further advances the concept of infectious tolerance originally proposed by Gershon and Kondo (50) in 1970s. Thus, certain Tregs conferred to the host an infectious tolerance by imparting the target cells with a regulatory phenotype, thereby prolonging the anti-inflammatory response (51).

How MDSCs perform their function has been explored by several investigators (21, 22, 52). The suppressive activity of MDSCs could either be mediated by a contact-dependent manner or by their secreted products such as arginase 1, inducible NO synthase, and reactive oxygen species (22, 52). Interfering with the constellation of costimulatory molecules or engaging inhibitory receptors or the membrane bound anti-inflammatory molecules such as LAP could also contribute to a contact-dependent inhibition (52). We observed an upregulation of several molecules by MDSCs. These included CD80 and PD1, which could contribute to their suppressive functions (Figs. 2F, 3A). The role of TGF-β in the suppressive activity has been proposed. TGF-β not only cause an induction of Tregs, but also their stabilization (39). However, our preliminary experiments involving inhibition of TGF-β signaling did not support its role in MDSC-induced conversion of Tregs or their stabilization (Figs. 3H, 3I, 8). This could rather suggest for a potent immunoregulation caused by MDSCs. Thus, the presence of TGF-β along with proinflammatory cytokines such as IL-6 could generate proinflammatory Th17 cells that limit the extent of immunosuppression (28, 53). MDSC-mediated effects on precommitted Tregs and the de novo generations of Foxp3⁺ T cells were seemingly caused by different mechanisms. Accordingly, during the conversion process of non-Tregs into Foxp3⁺ Tregs, the latter population downregulated CD73, an ecto-5′-nucleotidase, whereas its expression was upregulated by precommitted Tregs cocultured with MDSCs (Fig. 8). CD73, along with CD39, controls the suppressive activity of Tregs by modulating nucleotide metabolism (37, 38). The dichotomy in the effects of MDSCs on induction of Tregs and their phenotypic stabilization of already-differentiated Tregs could help diversify the functional properties to achieve persistent suppression. The underlying molecular details could provide greater insights into the cellular interactions and the tolerogenic responses.

We recovered adoptively transferred MDSCs from most lymphoid organs suggesting their predominant effects at immune inductive sites either by interfering with the activation of effector cells or by sequestering disease-causing CD4⁺ T cells to secondary lymphoid organs. The latter effects could be caused by the modulation of homing molecules of these cells. Our observations that CD4⁺ T cells from MDSC recipients expressed enhanced levels of CD62L and produced fewer proinflammatory molecules support this hypothesis (Figs. 5, 6). MDSCs not only suppress the cells of the adaptive immune system, but also those of innate arm, and the latter are regulated by inhibition of cytokine production (52). Clearly, future studies focusing on their cellular and molecular mechanisms are likely to provide insights into MDSCs mediated immunosuppression.

The authors have no financial conflicts of interest.