Infection-Derived Monocytic MDSCs Require TGF-β to Suppress Filarial-Specific IFN-γ But Not IL-13 Release by Filarial-Specific CD4+ T Cells In Vitro

Lymphatic filariasis (LF) remains a major health problem with severe economic repercussions in endemic communities of Sub-saharan Africa, South-East Asia and South America. The rodent-specific nematode Litomosoides sigmodontis (Ls) is used to study the immunomodulatory potential of filariae and research has elucidated pathways involving regulatory T cells (Tregs), IL-10 producing cells and alternatively activated macrophages (AAMs) and that CD4+ T cells play a paramount role during infection. Myeloid-derived suppressor cells (MDSCs) have been identified and characterised in man in cancer and other pathologies. The hallmark of MDSC populations is the suppression of T and B cell responses using various mechanisms, which are mostly specific to the pathology or setting. However, until now, it remains unclear whether they play a role in filarial-specific responses. We report here that monocytic MDSCs (Mo-MDSCs, CD11b+Ly6C+Ly6G-) and polymorphonuclear MDSCs (PMN-MDSCs, CD11b+Ly6Cint/loLy6G+) expanded in the thoracic cavity (TC, the site of infection) and correlated positively with filarial life-stages in Ls-infected BALB/c mice. In vitro, only infection-derived Mo-MDSCs showed a suppressive nature by preventing IL-13 and IFN-γ secretion from filarial-specific CD4+ T cells upon co-culture with soluble worm extract. This suppression was not mediated by IL-10, IL-6 or TNF-α, and did not require cell-contact, nitric oxide (NO), IL-4/IL-5 signalling pathways or CCR2. Interestingly, neutralizing TGF-β significantly rescued IFN-γ but not IL-13 production by filarial-specific CD4+ T cells. In comparison to naive cells, PCR array data showed an overall down-regulation of inflammatory pathways in both infection-derived Mo-MDSCs and PMN-MDSCs. In conclusion, these primary data sets show activity and expansion of MDSCs during Ls infection adding this regulatory cell type to the complex milieu of host responses during chronic helminth infections.


INTRODUCTION
Approximately, 65 million people are infected with LF worldwide (1). To understand mechanisms underlying immunity to filarial infections in man, such as Wuchereria bancrofti which elicits lymphatic filariasis (LF) (2,3), researchers employ the Litomosoides sigmodontis (L. sigmodontis, Ls) rodent model since it reflects several aspects of LF infection. This includes partial patency since only a portion of immunocompetent BALB/c mice become patent (release microfilariae (MF) into the periphery) when naturally infected with Ls (4)(5)(6). The development of adult worms from larval stages in the host provokes different immune reactions and encompasses both innate (neutrophils, eosinophils and basophils) and adaptive responses (T and B cells). Indeed, both filarial-specific CD4 + T cells and Th2 responses have been found to be essential for Ls development and host immunity (6)(7)(8). Moreover, lack of Th2 associated cytokines, such as IL-4, IL-13 and IL-5, disables the effects of neutrophils and eosinophils on controlling worm burden and leads to elevated MF loads (9)(10)(11). Interestingly, studies have also shown that IFN-g supports neutrophil-mediated clearance of Ls adult worms and suggested that Th1 and Th2 immune responses act in a synergistic manner in order to control Ls infection (12,13). Furthermore, infection with Ls induces regulatory cell populations including Tregs, IL-10 producing cells and alternatively activated macrophages (AAMs) that can affect CD4 + T cells by making them hyporesponsive to the nematode (6,(14)(15)(16)(17). In this study, we investigated the yet unexplored role of another regulatory cell population: Myeloid-Derived Suppressor Cells (MDSCs). Two subsets of MDSCs, the monocytic MDSCs (Mo-MDSCs, CD11b + Ly6C + Ly6G -) and the polymorphonuclear MDSCs (PMN-MDSCs, CD11b + Ly6C int/lo Ly6G + ) have been identified in patients and murine models of disease. These include cancer, systemic inflammation and sepsis, autoimmune diseases as well as bacterial, viral and parasitic infections (18)(19)(20)(21)(22)(23)(24)(25)(26). Mo-MDSCs and PMN-MDSCs are reported to suppress T and B cell responses through receptors, cytokines, or soluble factors as reviewed (20,27). It has been shown that in infections caused by Heligmosomoides polygyrus bakeri, isolated CD11c -CD11b + Gr1 + cells, likely MDSCs, suppressed OVA-specific CD4 + T cell proliferation in a nitric oxide (NO)-dependent manner and nematode-specific IL-4 responses in vitro. Also, an adoptive transfer of these cells induced higher parasite burden indicating that these cells promoted chronic infection (28). This current work explored the role of Mo-and PMN-MDSCs over the course of Ls infection in BALB/c mice. Interestingly, both MDSC subsets were present in the thoracic cavity (TC), the site of infection and levels positively correlated with worm load. Since cells were virtually absent in naive mice, this indicates that MDSCs are part of the immune repertoire that is specifically drawn to the TC during infection. To test functional responses of infection-derived MDSCs subsets, we designed an in vitro MDSC:T cell co-culture assay and revealed that Mo-MDSCs but not PMN-MDSCs showed high suppressive abilities on the production of IL-13 and IFN-g but not IL-5 by Ls-specific CD4 + T cells. This suppression was independent of cell-contact, nitric oxide, IL-10, IL-6, TNF-a and the receptors CCR2 and IL-4Ra. Further analyses demonstrated that Mo-MDSCs used the TGF-b pathway to impair the production of IFN-g but not IL-13. Collectively, this study reveals that the expansion of MDSCs subsets during Ls infection is a specific host response and Mo-MDSCs have a defined role in controlling Ls-specific IL-13 and IFN-g production from CD4 + T cells.

Ethics Statement
Wildtype (WT) and IL-4Ra/IL-5dKO BALB/c mice were kept under specific-pathogen-free (SPF) conditions at the Institute of Medical Microbiology, Immunology and Parasitology (IMMIP), University Hospital of Bonn (UKB), Bonn, Germany. Animals were bred in accordance with German animal protection laws and the EU guidelines 2010/63/E4. All in vivo experiments were reviewed and approved by the appropriate committee (Landesamt für Natur, Umwelt und Verbraucherschutz, NRW, Germany).

BMDC Preparation
Following erythrocyte-depletion of RBCs from leg bones (hip to ankle) of WT BALB/c mice, GM-CSF (20 ng/mL, Peprotech, New Jersey, USA) derived dendritic cells were generated over 7 days using the previously described protocol (6).

CD4 + T Cell Isolation
An enriched T cell fraction from pooled spleen and mediastinal lymph nodes (mLN) between 70-72 days Ls-infected BALB/c mice were obtained after depletion of antigen presenting cells (APCs) using MHC class II beads and MACS technology (Miltenyi Biotech, Bergisch Gladbach, Germany). Resulting cells were then stained with anti-CD4 (clone RM4-5) APCconjugated and anti-CD25 (clone PC61.5) PE-conjugated antibodies (Thermo Fisher Scientific). CD4 + CD25 neg and CD4 + CD25 hi T cells were then isolated using the FACS DIVA cell sorter (BD Biosciences, Heidelberg, Germany). We purposely removed the CD4 + CD25 hi T cells from the main CD4 + T cell population based on the findings from our earlier studies (6,31,32) since this now removed potential regulatory T cells (CD4 + CD25 hi Foxp3 + ) from the co-cultures allowing a more accurate assessment of the MDSC suppressive ability on effector CD4 + T cells fractions. Cells were isolated using BD FACS Aria III Cell Sorter (BD Biosciences, Heidelberg, Germany) at the Flow Cytometry Core Facility (FCCF), Bonn.

Cytokine Determination
Cytokine concentrations in the culture supernatants of restimulation assays were determined by ELISA according to the manufacturer's instructions (IFN-g and IL-13 Thermo Fisher Scientific; IL-5 BD Biosciences). ELISA plates were read and analysed at 450 and 570 nm using a Spectra Max 340pc384 photometer and SOFTmax Pro 3.0 software (Molecular Devices, Sunnyvale, California, USA).

Flow Cytometry Staining of MDSCs
Frequencies of MDSC subsets in Ls-infected BALB/c mice were determined using cells collected from the TC. After blocking (anti-CD16/CD32, clone 93) (Thermo Fisher Scientific), cells were stained with the antibodies used for MDSC isolation (section 3.6 MDSC isolation). Expression levels were determined using the FACS Canto flow cytometer (BD Biosciences) and analysis was performed using the gating strategy shown in Figure 1A, adapted from (33) with FlowJo VX software (Tree Star, Inc, Ashland, Oregon, USA).

Gene Expression Analysis
In order to evaluate the differences in the expression of 84 genes between isolated Mo-and PMN-MDSC populations (see MDSC Isolation) from pooled Ls-infected BALB/c mice (n=10), PCR Arrays using the RT2 SYBR Green Mastermix kit (Qiagen, Hilden, Germany) and the RT2 Profiler PCR Array Mouse Innate and Adaptive Immune Responses kit (Qiagen) were performed. These two subsets of MDSCs were also isolated from naive mice to serve as controls (n=30). 20 µl PCR components mix were added to each well of the RT2 Profiler PCR Array using Qiagility device (Qiagen), followed by cycling using a Rotor Gene Q (Qiagen). Data were analysed using the online RT2 Profiler PCR Array Data analysis 3.5 software at the  sabiociences.com website (Qiagen) and gene expression was normalized to Beta-actin. The threshold cycle (Ct) cut-off was set to 33 and the fold-change varied between 0.0001 and 15789. Fold-change values greater than one indicate a positive-or an up-regulation, and fold-change values less than one indicate a negative or down-regulation.

Statistical Analysis
Statistical analyses were performed using SPSS (IBM SPSS Statistics 25, Armonk, NY, USA) and graphs were generated using the software PRISM 5.02 (GraphPad Software, Inc., La Jolla, USA). After column statistics (D'Agostino & Pearson normality test, Shapiro-Wilk normality test and Kolmogorov Smirnov normality test), statistical differences between three or more groups were observed using ANOVA with Tukey posthoc test, for a parametric distribution. If data were non-parametric, a Kruskal-Wallis with Dunn's posthoc test was used. For correlation analysis, statistical significances were tested using the Spearman correlation test. P-values of 0.05 or less were considered significant.
Very few Mo-MDSCs or PMN-MDSCs were present in the TC of naive mice ( Figure 1B) whereas substantial numbers of these cells comprised the cellular milieu in Ls-infected mice 72 days p.i. ( Figure 1B). Moreover, frequencies and cell numbers of PMN-MDSCs were significantly higher than Mo-MDSCs ( Figure 1C). Correlation analysis of worm burden or MF counts with MDSC numbers also indicated that levels of both Mo-MDSCs and PMN-MDSCs positively correlated with worm burden ( Figure 1D) and MF in the TC ( Figure 1E). However, no correlation was displayed between MDSC levels in the TC and released MF numbers in blood ( Figure 1F). These data show that Mo-MDSCs and PMN-MDSCs infiltrate the TC of Ls-infected mice and positively correlate with parasite load at the site of infection.

Mo-MDSCs But Not PMN-MDSCs Suppress Ls-Specific CD4 + T Cell Responses
To test the suppressive properties of Mo-MDSCs and PMN-MDSCs that arose during Ls infection, an in vitro MDSC:T cell co-culture assay was designed (Figure 2A). Earlier studies have shown that CD4 + T cells from Ls-infected but not naïve BALB/c mice can be re-stimulated ex vivo with LsAg (a soluble extract prepared from adult Ls worms) (6) ( Figure S2A). Thus, CD4 + T cells (without the Treg compartment) from Ls-infected WT BALB/c mice (70-72 days p.i.) were co-cultured with GM-CSFderived BMDC from WT BALB/c naive mice and stimulated with LsAg in the presence or absence of Mo-or PMN-MDSCs from Ls-infected mice in a 1:1 ratio (MDSCs:T cells). This assay revealed that Mo-MDSCs were able to significantly suppress the production of IL-13 and IFN-g but not IL-5 by filarial-specific CD4 + T cells ( Figure 2B c.f. white and black bars). Surprisingly, PMN-MDSCs were not able to suppress the production of any of those measured cytokines ( Figure 2B c.f. white and grey bars). Of note, when compared to the main read-out parameter DC+CD4 + +LsAg, other controls had negligible levels of IL-13 and IFN-g ( Figure S2B) when stimulated with LsAg. Similarly, co-culture of Mo-MDSCs with GM-CSF-derived DC (without CD4 + T cells) had very low cytokine levels ( Figure S3) Moreover, the observed suppression shown by Mo-MDSCs was not caused by cell death (or whether antigen stimulation occurred before or after Mo-MDSCs were added to co-culture), since discrimination of dying (or adding Mo-MDSCs after antigen stimulation) still resulted in suppression of IL-13 and IFN-g ( Figures S1B, C). To determine whether the suppressive ability of Mo-MDSCs was connected to the infection state of the mouse (patent or latent), further assays were performed using Mo-MDSCs isolated from either MF+ or MF-mice. Figure S4 shows that the suppressive capacity of Mo-MDSCs on the production of IL-13 ( Figure S4A) or IFN-g ( Figure S4B) by Ls-specific CD4 + T cells is independent on the infection state of the mouse.

Receptor-Dependent Pathways Are Not Crucial for Mo-MDSC Function
Research has indicated that MDSCs suppress T cell responses through the IL-4Ra. For example, cancer patients have been shown to have higher levels of MDSCs expressing IL-4Ra (36,37). Moreover, studies have reported that suppression through IL-4Ra is only attributed to the Mo-MDSC subset (38). Therefore, in order to decipher the mechanism(s) by which Mo-MDSCs suppress Ls-specific CD4 + T cell responses, we investigated the outcome of the suppressive ability of infectionderived Mo-MDSCs from IL-4Ra/IL-5 dKO BALB/c mice since earlier studies showed that these mice present high worm burden and MF load in the absence of these Th2 cytokines (9,11). Surprisingly, Mo-MDSCs derived from Ls-infected IL-4Ra/IL-5 dKO BALB/c mice were able to suppress IL-13 and IFN-g production by Ls-specific CD4 + T cell responses in a similar manner to those derived from WT BALB/c ( Figures 3A, B). This indicates that neither the IL-4Ra nor IL-5 was involved in the observed suppression. In regards to MDSC numbers in the TC, levels of Mo-MDSCs in either percentage or absolute cell number were significantly higher in Ls-infected IL-4Ra/IL-5 dKO BALB/ c ( Figures S5A, C) when compared to those in WT mice whereas populations of PMN-MDSCs remained comparable ( Figures  S5B, D). There was no difference in MDSC amounts in naïve IL-4Ra/IL-5 dKO BALB/c when compared WT mice (3.103% ± 0.345 vs 3.66% ± 1.149, respectively). CCR2 is a receptor present on the surface of Mo-MDSCs but not PMN-MDSCs (39) and reports have suggested that this receptor influences the suppressive activity of Mo-MDSCs (40)(41)(42). However, in assays blocking the CCR2 receptor, Mo-MDSCs were still capable of suppressing IL-13 and IFN-g production by Ls-specific CD4 + T cells, (Figures 3C, D and S6A, B for data including isotype controls).
Next, we assessed the requirement of cell-contact for Mo-MDSCs to suppress IL-13 and IFN-g production by filarialspecific CD4 + T cells using a transwell assay (Figures S7A, B). To avoid cell-contact, Mo-MDSCs were placed in an insert. Again, suppression of T cell cytokine production remained unaltered between MDSC co-cultures with or without the insert indicating that cell-contact was not required. These data tallied with the findings on the IL-4Ra receptor and emphasised that suppression was not receptor mediated.

MDSC-Related Cytokines and Nitric Oxide Pathways Are Not Essential for Mo-MDSC Suppression of LsAg-Specific CD4 + T Cell Responses
Evidence has been provided that MDSCs function through either soluble factors (43,44) or receptors (41,45) and since we found that suppression of CD4 + T cell responses by infection-derived Mo-MDSCs was not cell-contact dependent, we explored the requirement of cytokines using neutralizing antibodies. Increased levels of IL-10 and IL-6 levels have been associated with the augmentation of Mo-MDSC levels in patients or models of cancer and other diseases (46)(47)(48)(49). Additionally, studies have demonstrated that TNFR2 expression is required for the generation and function of Mo-MDSCs (45,50,51). Here, we found that the Mo-MDSC-mediated suppression of IL-13 ( Figures 4A-C) or IFN-g ( Figures 4D-F) by Ls-specific CD4 + T cells was sustained after neutralizing IL-10 ( Figures 4A, D), IL-6 ( Figures 4B, E) or TNF-a ( Figures 4C, F), indicating that these cytokines were not relevant for Mo-MDSC suppressive activity.
NO and ROS are factors associated with MDSC function and a study has reported that Mo-MDSCs limited autologous B cell proliferation and antibody production in a NO and PGE 2dependent manner (43,52) and recent findings showed that tumor-derived MDSCs regulated NO production in asthma pathogenesis (53). Interestingly, using a single application of L-NMMA (an inhibitor of nitric oxide synthase (NOS), an enzyme which catalyses NO production), we observed a trend of suppression of IFN-g production by Ls-specific CD4 + T cells in presence of Mo-MDSCs 24 hours after stimulation ( Figures 5A, B). Upon 72 hours stimulation, both IL-13 and IFN-g production were dampened by Mo-MDSCs and this was not affected by blocking NO pathways (Figures 5C, D). Moreover, this did not alter when multiple doses (at 0, 24 and (B) Resulting supernatants were screened for IL-13, IFN-g and IL-5 by ELISA. Graphs show mean ± SEM of combined co-culture assays using cells from 4-6 pooled mice from four independent infections. Statistical significances between the indicated groups were obtained after Kruskal-Wallis followed by Dunn's multiple comparison test.
48 hours) of the inhibitor were added (Figures 5E, F). This suggests that Mo-MDSCs do not use NO pathways to suppress filarial-specific IL-13 and IFN-g production by Ls-specific CD4 + T cells.

TGF-b Is Critical for Filarial-Specific
Mo-MDSC Suppression of IFN-g TGF-b is a cytokine which is usually associated with regulatory networks (54). Therefore, MDSC suppression assays were performed whilst blocking TGF-b. Interestingly, the presence of anti-TGF-b neutralizing antibody in the culture did not impair the ability of Mo-MDSCs to suppress IL-13 production ( Figures 6A and S6C) but the suppression of IFN-g production by filarial-specific CD4 + T cells was significantly abrogated (Figures 6B and S6D). When compared to the main read-out parameter the secretion of TGF-b is diminished in the presence of Mo-MDSCs ( Figure S6E, c.f. white and bars). Collectively, these data show that TGF-b is essential for the suppression of IFN-g production by filarial-specific CD4 + T cells. Figure 7 displays schematically the role of TGF-b in the suppression of IFN-g production by filarial-specific CD4 + T cells.

Inflammatory Pathways Are Shut-Down in Infection-Derived MDSC Subsets
Since gene profiling can inform on MDSC features and has been associated with MDSC function (55, 56), we assessed the regulation of innate and adaptive genes comparing isolated Mo-MDSCs and PMN-MDSCs from the TC of infected WT BALB/c mice on day 72 p.i. to those of naive mice which served as controls. Interestingly, data from the PCR array (84 genes) showed that PMN-MDSCs and especially Mo-MDSCs isolated from infected mice displayed an up-regulation of FOXP3 (Forkhead box P3), a transcription factor for regulatory T cells (Tregs) and NF-kB1 (Nuclear factor of kappa light polypeptide gene enhancer in B-cells 1, p105) (Tables 1, 2). However, a down-regulation of genes including C3 (Complement component 3), CCR6 (Chemokine  Each graph shows mean ± SEM of combined co-culture assays using cells from 5-10 pooled mice from two independent infections. Statistical significances between the indicated groups were obtained after Kruskal-Wallis followed by Dunn's multiple comparison test. (C-C motif) receptor 6) and IL-18 (Interleukin 18) was observed in the two subsets ( Tables 3, 4). Mo-MDSCs alone showed negative fold-changes in the expression of the genes IL-1a and NOD1 (Nucleotide-binding oligomerization domain containing 1) ( Table 3) but displayed up-regulated expression of IFN-g and IL-13 when compared to those from naive controls ( Table 1). These results give a primary indication that gene expression within the two MDSC subsets present high regulatory and rather low inflammatory properties.

DISCUSSION
In this study, we show that MDSCs are induced and expanded at the site of infection following exposure to L. sigmodontis ( Figure 1) and moreover that the monocytic subset displays distinct suppressive activities on filarial-specific CD4 + T cell responses ( Figure 2). Furthermore, the suppression was tailored since not all cytokine responses were suppressed e.g. IL-5 ( Figure 2B), adding credence to the overall findings.
In cancer, MDSC frequencies have been associated with both the state and progression of pathology (46,57). Using the laboratory BALB/c mouse strain, which allows the complete Ls life-cycle, we found that in the absence of infection, very few Mo-MDSCs and PMN-MDSCs were within the TC fluid whereas large numbers of these cells were identified after infection ( Figure 1B). This supports earlier studies that revealed that levels of MDSC populations increase under inflammatory conditions (20,58). Interestingly, there was a higher number of PMN-MDSCs when compared to Mo-MDSCs ( Figure 1C) which reflects observations in tumour models (19) but different compositions were found in Trypanosoma congolense-infected mice (59). In our model, MDSC infiltration positively correlated with adult worm burden ( Figure 1D). Therefore, we suggest that while adult worms develop, the host offers a favourable milieu for MDSC infiltration or vice versa, promoting chronic infection. This is in line with recent findings that demonstrated that after adoptive transfer of MDSCs, there was a persistent chronic phase of infection with the nematode Heligmosomoides polygyrus bakeri (28).  Upon re-stimulation of CD4 + T cells from infected mice with LsAg, infection-derived Mo-MDSCs but not PMN-MDSCs were able to impair the release of IFN-g, and IL-13 but not IL-5 ( Figure 2B). This suppression was revealed not to be dependent on patency since Mo-MDSCs derived from both MF-and MF+ Ls-infected mice groups were similarly capable of suppression ( Figure S4). Studies in cancer patients have demonstrated the suppressive function of MDSCs on CD4 + T cell responses even though the effect of Mo-MDSCs was not specifically ascertained (60). In contrast, data from primary glioblastoma patients have indicated that Mo-MDSCs were not able to suppress T cell responses, confining the suppressive function in that scenario to PMN-MDSCs. Indeed, these authors have observed diverse MDSC subsets including monocytic, granulocytic, neutrophilic, eosinophilic and immature MDSCs (61). In regards to murine settings, Zhu and colleagues found that in a murine model of hepatitis, SSC hig h CD11b hig h Ly-6C hi gh Ly-6G low MDSCs, corresponding to Mo-MDSCs, were the only subset capable of suppressing CD4 + T cell responses while the other MDSC-like compartment was not capable of doing so (62). Elsewhere, studies in C57BL/6 mice have noted the inhibition of not only T cell responses but also of B cell responses by Mo-MDSCs in murine LP-BM5 retroviral infections (63,64). PMN-MDSC numbers in the TC were higher when compared to Mo-MDSCs ( Figure 1C) but only Mo-MDSCs displayed a suppressive function on LsAg specific CD4 + T cell responses ( Figure 2B), suggesting that the Mo-MDSC-mediated suppressive function was not associated with elevated Mo- by ELISA. Each graph shows mean ± SEM of combined co-culture assays using cells from 5-10 pooled mice from three independent infections. Statistical significances between the indicated groups were obtained after Kruskal-Wallis followed by Dunn's multiple comparison test. In earlier studies we showed that Ls-infected IL-4Ra/IL-5 dKO BALB/c mice had higher worm burden and full patency when compared to the WT group (9). Since studies have suggested that the presence of the IL-4Ra is associated with the suppressive activity of Mo-MDSCs in melanoma and colon carcinomas (38), we investigated the role of this receptor too.
However, our results clearly show that the lack of IL-4/IL-5 signalling does not impair the suppression of CD4 + T cell responses by Mo-MDSCs ( Figures 3A, B) but does seem to facilitate Mo-MDSC infiltration into the TC of Ls-infected IL-4Ra/IL-5 dKO BALB/c mice ( Figure S5). These findings support previous studies which showed that in tumour-harbouring IL-4Ra KO mice, the IL-4Ra receptor was not required for MDSC function (65). Furthermore, our infection-derived Mo-MDSCs kept their suppressive ability even after CCR2 blockade ( Figures 3C, D). In contrast to that and using MDSCs generated from CCR2 -/-C57BL/6 mice, Qin and colleagues showed that these cells failed to migrate to islet allografts and to suppress CD8 + T cells under in vivo conditions which resulted in enhanced Tregs activity. However, they were able to exert in vitro inhibition (66). Further studies are required to ascertain whether blocking CCR2 truly affects the suppressive nature of filarial-specific Mo-MDSCs.
CCR2 is mainly involved in the migration of cells, which occurs during the course of infection. There remains a debate as to whether the anti-CCR2 neutralizing Ab used in this study, was suitable to block the activity of the receptor in vitro although a study reported that anti-CCR2 treatment had minimal influence on tumour regression (67). Perhaps Mo-MDSCs require CCR2 to influence only CD8 + T cell but not CD4 + T cell responses. In addition, the fact that these authors used generated MDSCs could explain the conflicting results observed here. Furthermore, data from Schmid et al. indicated that the CCR2 receptor was impaired in Mo-MDSCs derived from Leishmania major infected BALB/c mice when compared to those from C57BL/6 mice (68), suggesting that the presence and thus the activity of this receptor in mice may be strain-dependent. Therefore, more experiments using CCR2 -/-BALB/c mice and especially, using To gain further insight into the Mo-MDSC profiles, we additionally performed a PCR-array on isolated subsets and elucidated several genes that were up-and down-regulated when compared to naïve MDSC populations. These included the gene responsible for IL-1a expression which was shut down in Mo-MDSCs during Ls infection (Table 3). Interestingly, MDSCs have been shown to phagocytose Mycobacterium tuberculosis and produce high levels of IL-1a (70). Additionally, IL-18, a cytokine released upon activation of the inflammasomes, was found to be down-regulated in both Mo-MDSC and PMN-MDSC subsets isolated from Ls-infected mice, when compared to those of naive mice (Tables 3, 4). The expression of NLRP3 (NOD-like receptor family, pyrin domain containing 3) inflammasome was also downregulated in PMN-MDSCs (Tables 4). Interestingly, studies in patients with cryopyrin-associated periodic syndromes under anti-IL-1 therapy have displayed elevated MDSC numbers with suppressive activity (71). Recent data have indicated a positive correlation between the NLRP3 inflammasome expression and the MDSC markers CD11b and CD33 in a human head and neck squamous cell carcinoma (HNSCC) tissue microarray (72). Whether the inflammasome-signalling pathway plays a role in the suppressive capacity of filarial-induced Mo-MDSCs requires further studies. Interestingly, preliminary in vivo data from Ls-infected ASC -/-BALB/c mice showed no differences in Mo-MDSC numbers on day 72 post-infection (data not shown).
With regards to gene expression profilling, the data-sets here revealed that there was a down-regulation of the complement component C3 gene in both Ls-infected Mo-MDSC and PMN-MDSC subsets. Consistent with these results, an earlier report indicated that a complement component C3 deficiency in a diabetes model, was associated with favourable conditions for MDSC induction and associated disease outcome (73). In contrast, hepatic stellate cells derived from mice lacking complement component C3, have been shown to display an impaired MDSC induction and in doing so, exhibited no protection on co-transplanted islet allografts (74). This indicates that the regulation of the complement component C3 by MDSCs is linked to inflammation scenarios. We found that the FOXP3 gene is up-regulated in Mo-MDSCs and PMN-MDSCs during Ls infection, suggesting that both subsets might share the expression of the transcription factor Foxp3 together with Tregs, which could support their role as regulator of immune responses. Further studies should confirm these data and investigate the role of FOXP3 in MDSCs and how this marker could possibly influence MDSC function.
In summary, our study shows that MDSC subsets accumulate during chronic Ls infection at the site of infection. Moreover, Tnf receptor-associated factor 6 0 0.000131 infection-derived Mo-MDSCs exhibit a suppressive function on the production of IL-13 and IFN-g by Ls-specific CD4 + T cell responses. Strikingly, IFN-g production was rescued after blocking TGF-b and reveals that Mo-MDSCs are a relevant cell-type involved in filarial immunomodulation. Future studies should incorporate in vivo experiments which block Mo-MDSC development during infection or analyze the outcome of infection in the absence of TGF-b. Both scenarios hold difficulties since the long-term administration of neutralizing antibodies could not only alter the outcome of infection per se but also the development of host:helminth interactions which could lead to skewed helminth-specific immune responses. Thus, to evaluate the functional role of these subsets in vivo, preclinical models need to be established which allow tracking and depletion on-demand capabilities such as the DEREG model (75,76).

DATA AVAILABILITY STATEMENT
We have provided an additional file in which all raw data related to gene expression findings (PCR array results) are made available.

ETHICS STATEMENT
The animal study was reviewed and approved by Landesamt für Natur, Umwelt und Verbraucherschutz, NRW, Germany.

AUTHOR CONTRIBUTIONS
RSET and LEL designed the research project. RSET, MR, AW, KA and LEL planned and executed experiments. RSET and LEL analysed the data which was then discussed with MR and KA. MM provided the CCR2 blocking antibody. RSET and LEL wrote the manuscript which was then amended and critically assessed by all authors including AH.

FUNDING
RSET received a scholarship awarded by the German Academic Exchange Committee (DAAD) and the work received support from intramural funding by the University Hospital of Bonn (BONFOR