Subpopulations of M-MDSCs from mice infected by an immunodeficiency-causing retrovirus and their differential suppression of T- vs B-cell responses

Abstract

Monocytic (CD11b⁺Ly6G^±/LoLy6C⁺) myeloid derived suppressor cells (M-MDSCs) expand following murine retroviral LP-BM5 infection and suppress ex vivo polyclonal T-cell and B-cell responses. M-MDSCs 3 weeks post LP-BM5 infection have decreased suppression of T-cell, but not B-cell, responses and alterations in the degree of iNOS/NO dependence of suppression. M-MDSCs from LP-BM5 infected mice were sorted into four quadrant populations (Ly6C/CD11b density): all quadrants suppressed B-cell responses, but only M-MDSCs expressing the highest levels of Ly6C and CD11b (Q2) significantly suppressed T-cell responses. Further subdivision of this Q2 population revealed the Ly6C^+/Hi M-MDSC subpopulation as the most suppressive, inhibiting T- and B-cell responses in a full, or partially, iNOS/NO-dependent manner, respectively. In contrast, the lower/moderate levels of suppression by the Ly6C^+/Lo and Ly6C^+/Mid M-MDSC Q2 subpopulations, whether versus T- and/or B-cells, displayed little/no iNOS dependency for suppression. These results highlight differential phenotypic and functional immunosuppressive M-MDSC subsets in a retroviral immunodeficiency model.

Introduction

Myeloid derived suppressor cells (MDSCs) are a heterogeneous cell population, which dampens T-cell mediated immune responses against several types of cancer, including melanoma, lung, mammary, and colon carcinomas (1–3). More recently, MDSCs have been identified as altering immune responses to viral infections, including cytomegalovirus (CMV), hepatitis C virus (HCV), human immunodeficiency virus (HIV), simian immunodeficiency virus (SIV), and murine LP-BM5 retrovirus (4–9). In murine models, MDSCs are classically defined as CD11b⁺GR-1⁺, with further subdivision into two subsets: CD11b⁺Ly6G^±/LoLy6C^+/Hi monocytic-MDSCs (M-MDSCs) and CD11b⁺Ly6G^+/HiLy6C^±/Lo granulocytic MDSCs (G-MDSCs) (aka polymorphonuclear (PMN)-MDSCs) (2, 10, 11). MDSCs display varying co-expression of other surface markers including, but not limited to, TLR-4, F4/80, FcγRIII/II (CD16/32), IL-4Rα (CD124), and/or CD115 (1, 2, 6, 10). Murine MDSCs may also express different combinations of chemokine receptors, especially CCR2, CXCR4, CXCR2, and/or CX₃CR1, which are important for egress of MDSCs out of the bone marrow and/or migration to tumor sites or sites of infection (12–21). MDSCs suppress T-cell responses by secretion of nitric oxide (NO), arginase-1 (Arg-1), reactive oxygen species (ROS), and other mechanisms. Arguably, M-MDSCs express higher levels of NO and lower levels of ROS, while G-MDSCs produce higher levels of ROS, along with lower levels of NO, with both subsets capable of producing some Arg-1 (10, 22).

In recent years, the earliest descriptions of MDSCs in retroviral infections were focused in the HIV, SIV, and LP-BM5 murine systems (5–8, 23). In recent studies, HIV-infected patients had increased MDSC frequency, which correlated with exacerbated HIV disease (increased viral load and decreased CD4 T cell counts) (7, 23, 24). MDSC frequency decreased after initiation of highly active antiretroviral therapy (HAART), consistent with an active role of MDSCs during HIV infection (7, 23). Ex vivo enrichment from the peripheral blood mononuclear cells (PBMCs) of HIV-infected patients revealed two MDSC phenotypes: granulocytic-like MDSCs (23); and monocytic-like MDSCs (7), both capable of suppressing in vitro T-cell responses. It remains unclear which mechanism(s) are used by these MDSCs to suppress immune responses ex vivo and which, if any, differential cell-surface or mechanistic MDSC phenotype(s) are contributing to in vivo retrovirus-induced immunodeficiency. Additional studies are warranted in systems, such as murine retroviral models, which allow well-controlled experimentation to examine MDSCs in retroviral systems.

LP-BM5 retrovirus-induced murine AIDS (MAIDS) causes profound and progressive immunodeficiency in mice, resulting in early activational events, such as splenomegaly, lymphadenopathy, and hypergammaglobulinemia (25–29), similar to human HIV/AIDS. LP-BM5-induced disease also features severe deficiencies of the responsiveness of T and B cells, an increased incidence of B-cell lymphomas, and higher susceptibility to opportunistic infections (26–29).

Recently, following LP-BM5 infection, our lab identified an expanded CD11b⁺ suppressive M-MDSC population, which was further characterized as Ly6G^±/LoLy6C⁺ (6). A proportion of these M-MDSCs co-express FcγRIII/II, F4/80, and/or TLR4 (6). M-MDSCs from LP-BM5 infected mice suppress T-cell responses in an iNOS-dependent, indoleamine 2,3-dioxygenase (IDO)-independent, and arginase-independent manner (6, 30). In addition, and distinctively, these M-MDSCs suppress B-cell responsiveness, in part due to an iNOS-dependent mechanism (6).

The goal of this study was to further define the cell-surface phenotype and function of M-MDSCs from LP-BM5 infected mice, utilizing sorted M-MDSC populations and subpopulations. We show differential suppression of T- and/or B-cell responses by distinct M-MDSC subsets, which in some cases suppress via different mechanisms. These studies provide a further understanding of these murine LP-BM5-derived M-MDSCs and may have broad implications, especially for understanding how MDSCs contribute to immunodeficiency and retroviral pathogenesis during SIV/HIV retroviral infections.

Materials and Methods

Mice

C57BL/6 (B6, w.t.) mice were purchased from the National Cancer Institute (NCI; Bethesda, MD) or Charles River (Wilmington, MA) and housed in the Center for Comparative Medicine and Research (CCMR) at the Geisel School of Medicine at Dartmouth. All animal experiments were done with the approval of the Institutional Animal Care and Use Committee of Dartmouth College, and in conjunction with the Dartmouth CCMR, an AALAC approved animal facility.

LP-BM5 virus inoculation

LP-BM5 retrovirus was prepared as previously described (31, 32). Mice were given an intraperitoneal injection of 5×10⁴ plaque forming units (pfu) of LP-BM5 at 6-8 weeks of age. M-MDSCs were enriched from pooled splenocytes 3-6 weeks post infection (wpi).

Flow Cytometry

Surface staining was performed as previously described (33). Cells were stained with FITC-, PerCP-, PE-, APC-, Pe-Cy7, APC-Cy7-, or Brilliant Violet 421- conjugated antibodies and analyzed by a MACSQuant flow cytometer (Miltenyi Biotec; Auburn, CA) to detect the expression of murine TLR-4 (MTS510), IL-4Rα (CD124) (mIL4R-M1), F4/80 (BM8), FcγRIII/II (93), CD11b (M1/70), Ly6C (HK1.4), GR-1 (RB6-8C5), CXCR4 (L276F12), CXCR2 (SA045E1), CX₃CR1(SA011F11), or CCR2 (475301) (BioLegend; San Diego, CA; BD Biosciences; San Jose, CA; R&D Systems; Minneapolis, MN). Positive gates were selected based on isotype or fluorescence minus one (FMO) controls and data were analyzed using FlowJo software (Tree Star, Inc.). All representative flow cytometry dot plots and subsequent phenotypic analyses are of unsorted M-MDSCs.

M-MDSC Isolated and Cell Sorting

M-MDSCs were negatively selected for Ly6G, then positively selected for CD11b using paramagnetic beads and subsequent MACS column purification (Miltenyi Biotec) from pooled splenocytes (n=3-10) of infected mice, as previously described (6). Enriched cells, defined by isotype controls, were sorted using a FACS Aria (Miltenyi Biotec) into serum coated tubes. Unsorted (U) and sorted M-MDSCs were used as suppressor cells in suppression assays.

Suppression Assays

Responders cells (R) from naïve w.t. mice were cultured with suppressor cells (S; M-MDSCs) at a responder:suppressor (R:S) ratio of 3:1, unless otherwise noted, with supplemented medium and a final concentration of either 10ug/ml lipopolysaccharide (LPS) (B-cell), 0.75ug/ml concanavalin A (ConA) (T-cell), or 2.5ug/ml anti-CD3 and 1ug/ml CD28 (T-cell) stimulations, as previously described (6). Suppression assays were left untreated or treated with 100μM of the iNOS/NOS2 inhibitor L-NIL (Enzo Life Sciences) for the duration of the assay. Plates were pulsed with 1uCi [³H] thymidine in the last 6 hours of incubation, and assessed at 72 hours for thymidine incorporation. Percent suppression was calculated from the control response, as previously described (6).

NO Production

NO production was measured from cell supernatants of suppression assays (described above) after 65-72 hours of culture, using the Griess reagent for nitrite (Sigma-Aldrich; St. Louis, MO), as previously described (O'Connor, MA et al., in revision).

Statistical analysis

Statistical analyses between groups were tested using Student's t-test, and the Holm-Bonferroni method was used to correct for multiple testing. (ns) indicates no statistical difference.

Results

M-MDSCs derived earlier in the course of LP-BM5 infection exhibit decreased suppression of T-cell responses

Our lab has previously described retrovirus infection associated monocytic MDSCs (M-MDSCs) capable of suppressing polyclonal T-, and a novel finding, B-cell responsiveness, following LP-BM5 infection of B6 mice (6), but it remains unknown when these cells obtain their suppressive capabilities. To give insight into the functional development of these M-MDSCs we compared M-MDSCs from LP-BM5 infected mice 5 weeks post infection (wpi), the typical timepoint of M-MDSC assessment, versus an earlier 3wpi timepoint.

M-MDSCs were standardly isolated from LP-BM5 infected B6 mice, by Ly6G-depletion, followed by CD11b-enrichment (see Materials and Methods), from pooled splenocytes 3 or 5 wpi, and used in a suppression assay, as previously described (6). Briefly, M-MDSCs were co-cultured with naïve responder cells stimulated with polyclonal T- (anti-CD3/CD28) or B-cell (LPS) activators for 3 days. Suppression was calculated from changes in proliferation of responder cells in the absence or presence of M-MDSCs, as measured by [³H] thymidine incorporation. M-MDSCs from 3wpi mice displayed nearly equivalent suppression of B-cell responses (Fig 1A), as compared to M-MDSCs from 5wpi mice. Suppression of B-cell responses was significantly blocked (up to 40%) when cultures were treated with L-NIL, an iNOS-inhibitor (Fig 1A, B), as we previously described with another iNOS inhibitor (6), with no significant difference of L-NIL defined iNOS/NO dependency in cultures containing M-MDSCs from 3 versus 5wpi mice.

Figure 1. M-MDSCs derived earlier in the course of LP-BM5 infection exhibit decreased suppression of T-cell responses.

M-MDSCs were isolated from LP-BM5 infected mice 3 or 5wpi, co-cultured with naïve responder cells stimulated with LPS (B-Cell) (A) or anti-CD3/CD28 (T-cell) (C) in a standard suppression assay, as previously described (6). Histograms (A-E) represent the means of triplicate samples with standard deviations of a representative experiment, of a total of at least three independent experiments. The percent suppression of B-cell (A) or T-cell (C) responsiveness by M-MDSCs is depicted for both standard co-cultures and co-cultures with the addition of the iNOS inhibitor, L-NIL. The effect on suppression (blockade) by L-NIL addition on B-cell (B) versus T-cell (D) responsiveness is compared. (E) Supernates derived from parallel suppression assays in which responder cells were stimulated with anti-CD3/CD28, in the absence or presence of M-MDSCs, were assessed for nitric oxide (NO) using the Griess reagent for nitrite. *p<0.05, **p<0.01.

In contrast, M-MDSCs from 3wpi mice had a significant impairment, but not complete abrogation, of the suppression T-cell responses, as compared to M-MDSCs from 5wpi mice (Fig 1C). Addition of the iNOS inhibitor completely blocked the suppression of T-cell responses by M-MDSCs from 5wpi mice, but surprisingly, did not block the lower level of suppression of T-cell responses by M-MDSCs from 3wpi mice (Fig 1C, D). These data indicated that 3wpi M-MDSCs suppressed T-cell responses in an iNOS-independent manner, in contrast to the iNOS-dependent manner of 5wpi M-MDSCs. As measured by Griess reagent, NO produced in suppression assays containing M-MDSCs from 3wpi mice, was significantly lower than cultures containing M-MDSCs from 5wpi mice (Fig 1E). Therefore, the amount of NO produced by M-MDSCs, in vitro, from 3wpi mice may not be sufficient for an iNOS-dependent mechanism of suppression of T-cell responses. These collective data indicated that suppression of T- versus B- cell responses, and/or suppression via an iNOS/NO dependent mechanism, by M-MDSCs may proceed at different kinetic rates post infection, suggesting that subpopulations within these heterogeneous cells may display differential suppression of T- and B-cell responses.

M-MDSC populations, as defined by flow cytometric quadrant analysis of CD11b vs Ly6C expression, differentially suppress B- and T-cell responsiveness

M-MDSCs from LP-BM5 infected mice display substantial heterogeneity with regard to Ly6C expression (6), and this heterogeneity may be associated with the differential suppression observed in Figure 1. M-MDSCs isolated from 5wpi mice were sorted (see Materials and Methods) into indicated quadrant populations, delineated by isotype controls, present at the following mean proportions: Ly6C^±/Lo CD11b⁺ (Quadrant 1, Q1: 8.2%±1.7%); Ly6C⁺CD11b⁺ (Quadrant 2, Q2: 61.0%±6.2%); Ly6C^±/LoCD11b^±/Lo (Quadrant 3, Q3: 18.2%±3.4%); and Ly6C⁺CD11b^±/Lo (Quadrant 4, Q4: 12.6%±2.4%) cells (Fig 2A). When sorted M-MDSCs were tested in suppression assays, as described above, at equivalent R:S ratios, cells from all four quadrants suppressed B-cell responsiveness (Fig 2B), with no significant difference in suppression between quadrant populations. In contrast, only sorted M-MDSCs in Q2 significantly inhibited T-cell responsiveness (Fig 2C), with the other M-MDSC quadrants (Q1, Q3, and Q4) contributing to little/no (depending on the overall activity levels of the individual experiments) suppression of the T-cell response. The possibility of FoxP3⁺ T regulatory (Tregs) cells contributing to the observed suppression of B- and/or T-cell responses was very unlikely due to their infrequency (<0.5%) in enriched M-MDSC preparations. Regarding suppression of T cells in particular, the prototypic target of Treg cells, two additional findings argue against this possibility: 1) the essentially complete iNOS/NO dependency of the mechanism of suppression by unsorted MDSCs and 2) the somewhat preferential distributions of the few contaminating CD4⁺FoxP3⁺ Tregs in quadrants which had little/no suppression of T-cell responses (Q1, Q3, and Q4). To provide insight into the total suppressive capacity of the M-MDSC quadrants, suppression was adjusted for the proportionality of cells (Fig 2A) and Q2 cells were clearly the major contributor of the suppression of both B-, and especially T-cell, responsiveness (Fig 2D).

Figure 2. M-MDSC populations, as defined by flow cytometric quadrant analysis of CD11b vs Ly6C expression, differentially suppress B- and T-cell responsiveness.

M-MDSCs were enriched from LP-BM5 infected mice, 5-6wpi. (A) A representative flow cytometry dot plot along with the quadrant gating scheme and frequencies of quadrant populations, is depicted: Ly6C^±/LoCD11b⁺ (Quadrant 1, Q1), Ly6C⁺CD11b⁺ (Quadrant 2, Q2); Ly6C^±/LoCD11b^±/Lo (Quadrant 3, Q3), and Ly6C⁺CD11b^±/Lo (Quadrant 4, Q4) cells. Histogram (A) depicts the means and standard deviations for seven independent experiments. The percent suppression of B-cell (LPS) (panel B) and T-cell (ConA) (panel C) responsiveness by sorted M-MDSC quadrant populations, versus unsorted (U) M-MDSCs, is given. Histograms (B-C) represent means of triplicate samples with standard deviations, and are representative of four independent experiments, with similar patterns of results. No significance differences in the suppression of B-cell responses were found between the quadrant subpopulations (B). In C, p<0.01 for: §versus Q1; ≠versus Q3; # versus Q4. (D) Overall contribution to suppression by each sorted M-MDSC quadrant population, as adjusted for the frequency of enriched M-MDSCs distributed into that quadrant (from A). Pie charts represent the means of four independent experiments.

For suppression of B-cell responses, Ly6C^+/Hi M-MDSCs constitute the single-most suppressive subpopulation, whereas suppressive function cannot be detected for the Ly6C^+/Mid subpopulation

As cells in Q2 contributed substantially to the suppression of B-cell responses and T-cell responses (Fig 2D), and because of the varied Ly6C density of Q2 (Fig 2A), we further subdivided these cells into three independent subpopulations, present at the following average proportions: Ly6C^+/Lo (6.4%±0.7%), Ly6C^+/Mid (30.6%±7.3%), and Ly6C^+/Hi (20.6%±1.6%) M-MDSCs (Fig 3). Having only one remaining sorting gate available, cells from the remaining three quadrants (Q1, Q3, and Q4), which display varying (lower) levels of both Ly6C and/or CD11b expression, and all which displayed suppression of B-cell responses (Fig 2B), were sorted into a single population: Q1,3,4 (39.0%±6.2%) (Fig 3).

Figure 3. M-MDSC Q2 subpopulation gating scheme.

M-MDSCs were enriched from LP-BM5 infected mice. A representative flow cytometry dot plot, along with the Q2 subpopulation gating scheme, is depicted to define, based on Ly6C expression, Q2 subpopulations: Ly6C^+/Lo, Ly6C^+/Mid, and Ly6C^+/Hi M-MDSCs. Quadrants Q1, Q3, and Q4 were sorted into a common pool for testing as a single combined population (Q134). The frequencies of M-MDSC subpopulations are indicated by histogram means with standard deviations for eight independent experiments.

The suppression of polyclonal B-cell responses, utilizing the sorting scheme of M-MDSC Q2 subpopulations as depicted in Figure 3, was first examined. M-MDSCs from the pooled Q1,3,4 populations and the Ly6C^+/Lo Q2 subpopulation had low/intermediate, but significant, levels of suppression of B-cell responsiveness, compared to unsorted Ly6C^+/Mid M-MDSCs (Fig 4A). Unexpectedly, suppression of B-cell responsiveness by the Ly6C^+/Mid subpopulation was consistently undetectable (Fig 4A). In contrast, the Ly6C^+/Hi M-MDSC subset displayed potent inhibitory activity, significantly suppressing B-cell responses at higher levels compared to any other subpopulation (Fig 4A). Although the frequency of the Ly6C^+/Hi subpopulation was a minority, about 20% of all M-MDSCs (Fig 3) and about 35% of Q2 M-MDSCs (Interpretation from Fig 3), it was a major contributor to the suppression of B-cell responsiveness (Fig 4B).

Figure 4. For suppression of B-cell responses, Ly6C^+/Hi M-MDSCs constitute the single-most suppressive subpopulation, whereas suppressive function cannot be detected for the Ly6C^+/Mid subpopulation.

Unsorted (U) or sorted M-MDSC subpopulations (gating scheme in Fig 3) were co-cultured with naïve responder cells stimulated with LPS in a standard suppression assay. All histograms (A, C, D) represent the means of triplicate samples, with indicated standard deviations and are from a representative experiment, with similar patterns of results, of a total of four (A) and three (C, D) independent experiments. (A) Suppression of B-cell responses by unsorted and sorted M-MDSC subpopulations is given. p<0.01: §versus Q134 population; ≠ versus the Ly6C^+/Lo; or # versus Ly6C^+/Mid subpopulations. (B) Overall suppressive contribution of each Q2 subpopulation, adjusted for the frequency of M-MDSCs distributed into quadrant 2 (from Fig 3). The pie chart depicts the means of four independent experiments. (C-D) Suppression assays were standard or included the addition of the iNOS inhibitor, L-NIL. The amount of suppression (C) and L-NIL blockade (D) are given. No detectible suppression indicated (ND), and not applicable (NA). **p<0.01, (ns) no statistical difference.

To assess potential differential use of the iNOS/NO mechanism to suppress polyclonal B-cell responsiveness, M-MDSC subpopulations (except the Ly6C^+/Lo M-MDSCs due to limited cell yield) were treated with the iNOS inhibitor L-NIL in standard suppression assays (Fig 4C). As a control, as previously described (6), iNOS inhibition lead to significant, but partial blockade of the suppression of B-cell responses by unsorted M-MDSCs (Fig 4C, 4D). There was also significant partial blockade (up to one third) of suppression by Ly6C^+/Hi cells (Fig 4C, 4D). In sharp contrast, little/no iNOS-dependent suppression was consistently observed for the pooled Q1,3,4 M-MDSC population (Fig 4C, 4D). Ly6C^+/Mid cells were unable to suppress B-cell responses (Fig 4A), and therefore there was no “suppression” to be blocked. Nonetheless as an internal control, addition of L-NIL to a co-culture of responder cells and this M-MDSC subpopulation had no effect on responder cell proliferation (Fig 4C, 4D). Collectively, these data indicated that Q2 M-MDSC subpopulations had differential suppressive activities, and exhibited varying levels of iNOS-dependency for suppression of B-cell responses.

The Ly6C^+/Hi M-MDSC subpopulation contributes to the majority of the suppression of T-cell responsiveness

In direct comparison to unsorted M-MDSCs, which exhibited strong levels of suppressive activity (80% inhibition), the suppression of T-cell responses by these same M-MDSC subpopulations (Fig 3) was examined (Fig 5). The pooled Q1,3,4 M-MDSC populations had no/low levels of suppression of T-cell responses (Fig 5A), and corroborated the results observed from the individual quadrant analyses (Fig 2C). On an equivalent R:S ratio basis, the Ly6C^+/Lo and Ly6C^+/Mid M-MDSC subpopulations provided low/intermediate levels of suppression of T-cell responses (Fig 5A), with no significant difference in suppression between these two cell subsets over five independent experiments. However, given the substantially higher distribution of the Ly6C^+/Mid subpopulation (∼50% of Q2 M-MDSCs; calculation from Fig 3), the total suppressive activity of the Ly6C^+/Mid subpopulation was greater than that of the Ly6C^+/Lo subpopulation (Fig 5B). In contrast, the Ly6C^+/Hi cells were significantly more suppressive of T-cell responsiveness, compared to any other single M-MDSC subset, as demonstrated using multiple R:S ratios within a given experiment (Fig 5A, 5B). Therefore, even when again taking into account the minority frequency (∼35% of Q2 M-MDSCs; calculation from Fig 3) of this subpopulation, Ly6C^+/Hi cells were the single-most suppressive M-MDSC subpopulation of both T- (Fig 5) and B-cell (Fig 4) responses.

Figure 5. The Ly6C^+/Hi M-MDSC subpopulation contributes to the majority of the suppression of T-cell responsiveness.

Unsorted (U) or sorted M-MDSC subpopulations (gating scheme in Fig 3) were co-cultured with naïve responder cells stimulated with anti-CD3/CD28, at multiple responder:suppressor (R:S) ratios (indicated below each graph). The dashed lines separate the standard R:S 3:1 from M-MDSC titration to yield R:S 6:1 and 9:1 ratios. All histograms (A, C, D) represent the means of triplicate samples, with indicated standard deviations, and are from a representative of a total of five (A) or four (C, D) independent experiments. (A) Suppression of T-cell responses by unsorted and sorted M-MDSC subpopulations. p<0.01: §versus Q134 population; ≠ versus the Ly6C^+/Lo; or # versus Ly6C^+/Mid subpopulation. (B) Overall contribution to suppression by Q2 subpopulations, adjusted for proportionality of cells (from Fig 3). The pie chart represents the mean of five independent experiments. (C) Suppression assays were standard or included the addition of the iNOS inhibitor, L-NIL. The amount of suppression (C) and L-NIL blockade (D) are given. No detectible L-NIL blockade indicated (ND). **p<0.01, (ns) no statistical difference.

As previously described (6), essentially all the suppression of T-cell responses by unsorted M-MDSCs was blocked by the addition of L-NIL (Fig 5C, 5D). Similarly, the suppression of T-cell responses by the highly active Ly6C^+/Hi M-MDSC subpopulation was also almost completely blocked by L-NIL (Fig 5C, 5D) and indicated a very strong iNOS-dependence for suppression of T-cell responses. In contrast, the suppression of T-cell responses by the pooled Q1,3,4 and Ly6C^+/Mid M-MDSCs (Fig 5B) were essentially iNOS-independent (Fig 5C, 5D). The infrequency of FoxP3⁺ Tregs within all M-MDSC subpopulations studied, as described above, was inconsistent with any substantial contribution to suppression. Thus, although all three M-MDSC Q2 subpopulations could suppress T-cell responses, to varying degrees, the Ly6C^+/Hi M-MDSCs subpopulation was by far more suppressive and its inhibitory activity almost completely iNOS-mediated.

The highly suppressive Ly6C^+/Hi M-MDSC subpopulation displays increased F4/80 expression

M-MDSCs from LP-BM5 infected mice co-express established MDSC markers, including TLR-4, F4/80, FcγRIII/II (6), or IL-4Rα (unpublished data). Identification of an additional MDSC marker(s) to further phenotype the described flow-sorted M-MDSC subpopulations, could aid in understanding the functional underpinnings of their differential suppression of T versus B cell responsiveness. To fairly compare several MDSC surface markers over multiple experiments, we looked at co-expression of MDSC markers on the M-MDSCs subpopulations, relative to the expression levels on the unsorted M-MDSCs as a whole. Co-expression of TLR-4, FcγRIII/II, or IL-4Rα was similar between all Q2 M-MDSC subpopulations (Fig 6A), which displayed their defining differential level of Ly6C expression. In contrast, F4/80 co-expression was significantly higher on the Ly6C^+/Hi subpopulation, as compared to the Ly6C^+/Lo and Ly6C^+/Mid subpopulations, with no significant difference between these latter subpopulations (Fig 6A).

Figure 6. The highly suppressive Ly6C^+/Hi M-MDSC subpopulation displays increased F4/80 expression.

(A) The mean fluorescent intensity (MFI) of TLR4, FcγRIII/II, IL-4Rα, and F4/80 expression by M-MDSC subpopulations (gating scheme from Fig 3) were normalized to the expression by all enriched M-MDSCs. Histograms represent the relative MFI means of at least five independent experiments with standard deviations. p<0.01: ≠ versus Ly6C^+/Lo; #versus Ly6C+/ of F4/80 expression of M-MDSC subpopulations, from 3wpi and 5wpi mice, were normalized to all enriched 5wpi M-MDSCs. Histograms (B, C) represent the means of at least three independent experiments with standard deviations: p<0.01: ≠versus Ly6C^+/Lo; #versus Ly6C^+/Mid. No significant differences between 3wpi and 5wpi M-MDSC subpopulations were observed.

In Figure 1 we demonstrated decreased suppression of T-cell responses by M-MDSCs from 3wpi versus standard 5wpi mice, and wondered if M-MDSCs from these two timepoints were phenotypically distinct. First, the relative percentage of the Q2 M-MDSC subpopulations (Ly6C^+/Lo, Ly6C^+/Mid, and Ly6C^+/Hi) (Fig 6B), and the overall expression of Ly6C (data not shown), were similar between 3wpi and 5wpi M-MDSCs. In addition, the entire expression of TLR4 and FcγRIII/II of M-MDSCs and expression within M-MDSC subpopulations was also similar between M-MDSCs from 3 and 5wpi mice (data not shown). F4/80 expression of the entire M-MDSC population was similar between 3wpi and 5wpi M-MDSCs (data not shown), but again as seen in Figure 6A, increased Ly6C expression was strongly associated with increased F4/80 expression also on 3wpi M-MDSCs (Fig 6C).

Unfortunately, the functions of Ly6C and F4/80, especially in the context of MDSC development and function, remain poorly defined (34–37), therefore other phenotypic marker(s) associated with a functional role of MDSCs were needed to further define M-MDSC subpopulations from LP-BM5 infected mice

Chemokine receptor expression further delineates M-MDSCs from LP-BM5 infected mice

To gain insight into the function and/or development of the M-MDSCs, we examined the phenotypic expression of chemokine receptors (CCR2, CX₃CR1, CXCR2, and CXCR4) reported to be variably present on MDSC populations in other systems and involved in MDSC egress into circulation and/or recruitment to tumor sites and sites of infection (13, 16, 20, 21, 38–40). M-MDSCs from LP-BM5 infected mice expressed varying percentages of CCR2, CX₃CR1, and CXCR4, but essentially not CXCR2 (Fig 7A). CCR2, CX₃CR1, and CXCR4 (Fig 7B), but not CXCR2 (data not shown), were also observed on all three M-MDSC Quadrant 2 subpopulations, albeit at varying frequencies. The Ly6C^+/Hi M-MDSC subpopulation exhibited the highest percentage of positive cells and expression on a per cell basis (MFI) of CCR2 and CX₃CR1 (Fig 7B), compared to the other two M-MDSC Q2 subpopulations. The MFI expression of CXCR4 was also higher on the Ly6C^+/Hi M-MDSCs compared to the Ly6C^+/Lo and Ly6C^+/Mid M-MDSC subpopulations (Fig 7B). In Figure 6A there was no observed phenotypic difference between the Ly6C^+/Lo and Ly6C^+/Mid M-MDSC subpopulations, despite their differential suppression of B-cell responses (Fig 4A). Assessment of chemokine receptor expression by these subpopulations, however, revealed that the Ly6C^+/Mid M-MDSC subpopulation displayed increased CCR2 expression on a per cell (MFI) basis (Fig 7B), compared to the Ly6C^+/Lo M-MDSC subpopulation.

Figure 7. Chemokine receptor expression further delineates M-MDSCs from LP-BM5 infected mice.

(A, B) Representative flow cytometry histograms and average frequencies, from four independent experiments, of CCR2, CX₃CR1, CXCR2, and CXCR4 on enriched M-MDSCs (A) and M-MDSC subpopulations (gating scheme from Fig 3) (B). (B) The mean fluorescent intensities (MFI) of CCR2, CX₃CR1, and CXCR4 expression by M-MDSC subpopulations were normalized to the expression by all enriched M-MDSCs. (C,D) The average frequency of expression of (C) none or one only, or (D) two or three chemokine receptors by M-MDSC subpopulations. Histograms (A-D) represent the means of four independent experiments with standard deviations. Fluorescence minus one (FMO). p<0.05: ≠versus Ly6C^+/Lo; #versus Ly6C^+/Mid; §versus Ly6C^+/Hi, using Student's t-test, and the Holm-Bonferroni method to correct for multiple testing.

To address the overlap of expression of chemokine receptors within each M-MDSC subpopulation, we first focused on M-MDSCs expressing only a single chemokine receptor (Fig 7C). The Ly6C^+/Lo and Ly6C^+/Mid M-MDSC subpopulations had larger proportions of cells expressing no chemokine receptors or just the CXCR4 chemokine receptor compared to the Ly6C^+/Hi M-MDSC subpopulation (Fig 7C). In contrast, the very high level of each individual chemokine receptor frequency in the Ly6C^+/Hi M-MDSC subpopulation, observed in Figure 7B, suggested extensive chemokine receptor co-expression. Indeed, the Ly6C^+/Hi M-MDSC subpopulation expressed the highest proportion of all three chemokine receptors, CCR2⁺CX₃CR1⁺CXCR4⁺, and the lowest frequency of CCR2⁺CXCR4⁺ dual-expressing cells, all compared to the other two M-MDSC subpopulations (Fig 7D). Collectively these results provide further phenotypic distinction between these three M-MDSC subpopulations—perhaps of relevance to the ability of these subpopulations to suppress T-cell versus B-cell targets with differential efficiencies.

Discussion

In these studies, we report differential suppression of T- and B-cell responsiveness by quadrant analysis (Q) populations and Q2 subpopulations of M-MDSCs from LP-BM5 retrovirus infected mice, as delineated by the densities of CD11b and/or Ly6C expression. First, M-MDSCs from all four quadrant populations suppressed B-cell responsiveness, but only cells from Quadrant 2 significantly suppressed T-cell responses (Fig 2). Upon analysis of this Q2 population, a highly suppressive Ly6C^+/Hi M-MDSC subpopulation was identified. The Ly6C^+/Hi M-MDSC subpopulation was significantly more suppressive of both T- and B-cell responsiveness, as compared to any other single Q2 subpopulation (Fig 4 and 5), and had the highest co-expression of cell surface F4/80 (Fig 6A), CCR2, and CX₃CR1 (Fig 7B). Second, suppression of B-cell responses by the Ly6C^+/Mid M-MDSC subpopulation was undetectable (Fig 4), although these cells consistently and substantially suppressed T-cell responses (e.g. ∼34%, Fig 5B) and comprised ∼30% of all enriched M-MDSCs (Fig 3). The suppression of T-cell responses by the Ly6C^+/Mid M-MDSC subpopulation was essentially independent of iNOS/NO (Fig 5), which served to further distinguish this subpopulation. Other observations made here were also in keeping with a dichotomy, under certain circumstances, of MDSC suppression of T-versus B-cell responses. For example we identified several M-MDSC populations with little/no suppression of T-cell responses: Q1, Q3, and Q4 (Fig 2C) and pooled Q1,3,4 (Fig 5A) populations, whereas these same M-MDSCs suppressed B-cell responsiveness. Additionally, using M-MDSCs derived from different kinetic stages of LP-BM5 retrovirus infection, it was determined that the ability to suppress B-cell (versus T-cell) responses developed earlier post infection (Fig 1C, 1E), despite no observed phenotypic differences among prominent MDSC-associated markers, including expression of F4/80 (Fig 6B, 6C).

Our results thus identified a Ly6C^+/Hi M-MDSC subpopulation, robustly suppressive of both T- and B-cell responsiveness, and displaying high levels of both Ly6C and F4/80 (Fig 6A). Unfortunately, the functions of Ly6C and F4/80 remain poorly defined, and to our knowledge, there is no agreed-upon ligand for F4/80 (34–37). It is possible that the differential expression of Ly6C by M-MDSC populations and subpopulations from LP-BM5 infected mice may allow further subdivision on the basis of expression of Ly6C1 and/or Ly6C2, which currently are indistinguishable using available antibodies (35). Macrophages from F4/80 knockout mice display normal macrophage development and no overt impairment of macrophage functions (36, 41). However, F4/80 may be important in the development and/or function of M-MDSCs, particularly the Q2 Ly6C^+/Hi subpopulation, during LP-BM5 retrovirus infection, but presently this possibility remains unclear.

A variety of factors produced by tumor cells and/or activated T-cells, including, but not limited to GM-CSF, IFNγ, IL-10, IL-6, IL-12, and IL-13, are implicated in the activation and/or expansion of MDSCs for a variety of tumor models (10). Relative to the LP-BM5 retroviral infection system, changes in the cytokine profile in the first week of infection, reported to be characterized by Th1 (IFNγ) and Th2 cytokines (IL-15, IL-4, IL-10) (42), versus the predominately Th2 cytokines (IL-4, IL-10, IL-6) observed at later stages of infection (42), may influence M-MDSC accumulation and/or function. Related to the results here, it is possible that a lack of “full” MDSC activation may explain why some M-MDSC populations/subpopulations have a relatively decreased dependence on iNOS/NO for suppression (Fig 1D, 4D, and 5D), and consequently decreased suppression of T-cell responses (Fig 1C, 2C). Similarly, other local combinations of cytokines and other micro-environmental factors may explain the alternative deficiency in suppressing B-cell targets (Fig 4A) versus sufficiency in suppressing T-cell targets (Fig 5A) by the same Ly6C^+/Mid M-MDSC subpopulation. Which retroviral and/or infection-induced host factors specifically drive M-MDSC expansion and/or activation in the LP-BM5 retroviral model remain unknown.

Several chemokine receptors are found on immature myeloid subsets, including MDSC subpopulations (13, 14, 20, 38, 40, 43–46), and are important for migration of cells out of the bone marrow and/or into sites of tumor or infection (14, 17–19, 47, 48). A very recent study of murine hepatocellular carcinoma compared the chemokine receptor gene expression profiles of splenic M-MDSC and G-MDSC (aka PMN-MDSCs) subsets (21). These authors observed roughly equivalent CXCR4 expression by both M-MDSCs and G-MDSCs, but upregulated expression of CCR2 and CX₃CR1 on M-MDSCs, versus increased CXCR2 expression by G-MDSCs (21). Our data further support these findings, as splenic M-MDSCs from LP-BM5 infected mice expressed CCR2, CX₃CR1, and CXCR4, but not CXCR2 (Fig 7A), albeit to varying degrees on the different M-MDSC Q2 subpopulations. It was specifically of interest to observe that the highest chemokine receptor expression of CCR2 and CX₃CR1 (Fig 7B) and highest frequencies of CCR2⁺CX₃CR1⁺CXCR4⁺cells (Fig 7D) was observed for the highly suppressive Ly6C^+/Hi M-MDSC subpopulation, compared to all other M-MDSCs subpopulations. Phenotypic expression of chemokine receptors further distinguished the M-MDSC subpopulations (Fig 7B-D), and these differences in chemokine receptor expression may help to elucidate the M-MDSC type(s) needed to suppress specific cellular targets (e.g. T-cell versus B-cell). Further studies are needed to determine whether these chemokine receptors play a role in M-MDSC suppressive activity (including cellular targets) per se and/or M-MDSC recruitment to the spleen in LP-BM5 infected mice.

In addition, our lab and others have published extensively on the cellular and molecular requirements for LP-BM5 induced pathogenesis, in the overall context that LP-BM5 co-opts immune cells and normal immune interactions as requisites for pathogenesis, including the profound and broad immunodeficiency. CD4⁺ T cells and B-cells (26, 29, 49, 50), interactions between CD40 on B-cells and CD40L on T-cells (31, 51–53), and recruitment of TNF receptor-associated factor (TRAF) to the TRAF6 binding site on the CD40 cytoplasmic tail (54) are required for LP-BM5 pathogenesis. These results and others (55, 56) have provided evidence for a central role of pathogenic CD4⁺ T-cells in driving disease. M-MDSCs may expand and/or become activated as a consequence of these interactions and then reciprocally down modulate T cells and/or B cells—either those T/B cells necessary for LP-BM5 disease pathogenesis or as the target cells of the effector mechanisms of the ultimate severe immunodeficiency of advanced MAIDS.

MDSCs from HIV-infected patients and SIV-infected macaques suppress T-cell responses, but, to our knowledge, suppression of B-cell responses by these cells remains under reported (5, 7, 8, 23). How HIV-infection-associated MDSCs suppress T-cell responses remains unclear: for example, one study reports an arginase-dependent, iNOS/ROS-independent mechanism (7), while another report implicates an arginase-independent, iNOS/ROS-dependent mechanism (5). In the context of LP-BM5 infection, we have previously shown, both by using multiple specific inhibitors and iNOS^-/- mice, that enriched but unsorted M-MDSCs suppress T-and B-cell responses in an arginase-independent, iNOS-dependent manner, albeit with varying levels of iNOS/NO dependency (6). Here, the Ly6C^+/Hi M-MDSC Q2 subpopulation also exhibited complete iNOS-dependent suppression of T- (Fig 5C, D), and partial iNOS-dependent suppression of B-cell (Fig 4C, D), responses. In contrast, other M-MDSC subsets suppressed only T-cells (Ly6C^+/Mid subpopulation) or B- and T- cells (Q1,3,4 population) with little dependence on iNOS/NO (Fig 4C, D and 5C, D).

How this apparent conundrum can be explained is unclear, but there are two broad possibilities. First, it may well be that the MDSC populations and subpopulations represent relatively stable phenotypic and functional subsets that either inherently, or after activation, display differential suppressive mechanisms, depending on the susceptibility of the T- versus B-cell targets to the various M-MDSC inhibitory mechanism. Alternatively, or in addition to, there may be important cross-regulation of immunosuppressive mechanisms by certain populations/subpopulations of M-MDSCs. Cross-talk between tumor-associated MDSCs with macrophages or dendritic cells (57), and/or tumor tissue (58), can enhance immunosuppression. Therefore, it can be hypothesized that cross-talk between MDSC subsets may also exist to enhance the overall level of inhibitory activity, and the diversity of molecular mechanisms employed, to achieve more global immunosuppression in tumor and retroviral models. For example, MDSCs can release chemokines which enhance tumor growth by recruitment of Tregs to the tumor site (59, 60), but may also influence the recruitment of other MDSC subsets. Nitric oxide can be a direct regulator of gene expression (61) and cellular activation (62), therefore one can speculate that NO production by one MDSC subpopulation may influence the gene expression and/or post-transcriptional activation of another MDSC subset(s). This cross-regulation may influence the overall suppressive capacity of the latter MDSC subpopulation, as well as its mechanism(s) of suppression, similar to autocrine—paracrine iNOS polarization in macrophages (63). As an example, M-MDSCs when tested individually in isolation as in our sorting experiments, such as the Q1,3,4 populations or Ly6C^+/Mid subpopulation, may lack exposure to exogenous factors from other M-MDSC subsets to drive iNOS-mediated suppression—factors that they are normally exposed to in the unsorted ex vivo M-MDSC preparation (and in vivo). This lack of normal M-MDSC cross-talk may explain why suppression by certain isolated M-MDSC subsets appear to be iNOS-independent (Fig 4 and 5). Analogous or alternative kinds of cross-talk, but between M-MDSCs and polyclonally activated responder cells, such as IFNγ production by T-cells or IL-6 by B-cells, may also contribute to the differential iNOS-dependence in the M-MDSC suppression of T- and B-cells. For example, in some tumor systems, IL-6 (64) and IFNγ (38) can influence MDSC expression of arginase and iNOS.

The current understanding of suppression of B-cell responses by MDSCs is very limited to only a few recent reports (6, 65), but the threshold for susceptibility to iNOS-mediated (and iNOS-independent-mediated) suppression of T-cells versus B-cells may vary. The ability of a given M-MDSC population or subpopulation to produce sufficient NO to achieve these thresholds may be regulated by arginine (L-Arg) levels available to the iNOS and/or arginase metabolic pathways within the M-MDSC. Several cross-inhibitory interactions have been reported to exist between these two pathways (63), and they can also synergize to generate reactive nitrogen and oxygen species (ROS), such as peroxynitrites and H₂O₂ (66). Low levels of arginine within the cell, due to increased iNOS activity and/or decreased L-Arg uptake into the cell, can substantially influence these alternative metabolic pathways. In addition, only the iNOS pathway can be “rescued,” during L-Arg unavailability, by cytosolic enzymes via conversion of citrulline into arginine; in contrast, the arginase pathway lacks such a “rescue” mechanism (63). Additionally, low levels of arginine can promote the generation of ROS (66). The single-most dominant mechanism of suppression of B-cell responses by unsorted M-MDSCs from LP-BM5 infected mice is iNOS/NO; but production of ROS, normally a minor contributor to suppression also occurs and may become a significant compensatory mechanism in the absence of iNOS/NO (Rastad JL and Green WR. Manuscript in preparation). Therefore, arginase, ROS, and/or other alternate mechanisms may also contribute to suppression of T- versus B-cell responses by M-MDSC subpopulations, especially when iNOS/NO is limited.

One of most unexpected results from these studies was the consistent lack of detectable suppression of B-cell responses by the Ly6C^+/Mid subpopulation. This finding may complement two intriguing and very recent observations from our lab: 1) enriched M-MDSCs from LP-BM5 infected mice, whose CD4 T-cell compartment has been depleted of FoxP3⁺ T regulatory cells (nTregs) prior to their transfer intro TCRα^-/- recipients, in an adoptive transfer model of LP-BM5 induced disease, exhibit a significant increase in the proportion of the Ly6C^+/Mid Q2 M-MDSC subpopulation; and 2) consistent with the findings here, M-MDSCs isolated from these nTreg-depleted mice display increased suppression of T-cell, but not B-cell, responsiveness (O'Connor, MA et al., in revision). Several possibilities may explain why the Ly6C^+/Mid subpopulation is unable to suppress B-cell responses: 1) in isolation, these Ly6C^+/Mid cells may lack the required cross-regulation provided by other cells types, including other M-MDSCs, to become active suppressors of B-cell responses; 2) an alteration in autocrine and/or paracrine cytokine signaling could alter iNOS expression and/or other activated functions within the Ly6C^+/Mid cells, resulting in decreased NO production, and thus potentially increasing alternative mechanism(s) of suppression; 3) the non-iNOS/NO mechanism(s) available to the Ly6C^+/Mid subpopulation may only efficiently target a pathway or receptor found in/on T-cells, similar to CD3(x003B6) down regulation on T-cells following extracellular arginase-induced L-Arg depletion (67); and/or 4) the Ly6C^+/Mid subpopulation, as a newly defined M-MDSC subset, may represent a stable population with a distinct immunosuppressive mechanism(s) and target cell specificity. Future analysis of how this Ly6C^+/Mid M-MDSC subpopulation differs—functionally, phenotypically, and/or epigenetically—from the other Q2 subpopulations, and its potential lineage relationships to these subsets, may be key to understanding why these Ly6C^+/Mid M-MDSCs display differential suppression of T- versus B-cell responses.

In conclusion, our studies contribute to a more incisive understanding of retrovirus-induced M-MDSCs, and overall of MDSCs in general, by highlighting distinct M-MDSC populations and subpopulations with partially overlapping, but also distinguishing, mechanisms of suppression. Future studies, identifying the genetic and metabolic processes driving M-MDSC effector function, may help define the differential involvement of iNOS/NO dependent and independent mechanisms of suppression. Further characterization of T- versus B-cell target specificity by M-MDSC, as well as G-MDSC, subsets and their suppressive mechanism(s) utilized are needed to accurately identify and selectively target MDSCs, which contribute to the local and/or global immunosuppression that either promotes pathogenesis directly, and/or interferes with the development and maintenance of protective immunity in viral diseases, tumor systems, and autoimmunity.

Highlights.

M-MDSC subpopulations from LP-BM5 infected mice suppress T- and B-cell responses differentially

The Ly6C^+/Hi M-MDSC subpopulation suppresses T- and B-cells and is dependent on iNOS/NO

Suppression of B-cell responses by M-MDSCs, develops earlier than suppression of T-cell responses

The Ly6C^+/Mid M-MDSC subpopulation suppresses T-cell, but not B-cell, responsiveness