Interleukin-22 is increased in multiple sclerosis patients and targets astrocytes

For the studies pertaining to the determination of IL-22 and IL-22BP levels in the blood, 141 MS patients and healthy control (HC) subjects were enrolled and divided into subgroups according to the disease type (Table 1). The diagnosis of MS followed the revised McDonald criteria [36]. The category of clinically active multiple sclerosis patients comprised relapsing remitting (RR)-MS or clinically isolated syndrome (CIS), who had a relapse that started less than 2 months prior to our assays. The category of clinically inactive multiple sclerosis patients included RR-MS and CIS patients who were in remission, as defined by an interval of more than 3 months after the last relapse. The category of progressive MS patient group contained patients with secondary progressive (SP)-MS or primary progressive (PP)-MS. Clinically inactive, SP- and PP-MS patients were not under any treatment within the 3 months prior to the blood draw. Among the 26 clinically active MS patients, four were on interferon-β, one on natalizumab, and one on fingolimod treatments. None of the MS patients had received corticosteroids in the previous 3 months. All enrolled patients and healthy control subjects signed an informed consent form, according to our institution review board.

For immunohistochemistry studies, brain biopsies from 11 non-MS patients were obtained from neurosurgical resections performed in the service of neurosurgery at the CHUV in Lausanne, hereafter referred to as the “Lausanne cohort” (Table 2). Tissues from five MS patients with their seven control counterparts obtained after postmortem autopsies were processed in Basel and named “Basel cohort” (Table 2). All human brain tissues were collected in accordance with local ethical committee from the University Hospitals of Lausanne and Basel and the UK MS Tissue Bank.

Human primary astrocytes (HA) derived from brain cerebral cortex were purchased from ScienCell Research Laboratory and were cultured according to the provider’s instructions. Briefly, HA were grown and cultured at 37 °C with 5 % CO₂, in astrocyte medium (AM), supplemented with 2 % fetal calf serum (FCS) and 1 % astrocyte nutritive supplement with 1 % penicillin/streptomycin (referred as “complete astrocyte medium”). For immunofluorescence, cells were fixed for 15 min with 4 % paraformaldehyde and stored in phosphate-buffered saline (PBS) at 4 °C. For flow cytometry, cells were resuspended with trypsin (BioConcept) and first labeled with LIVE/DEAD fixable violet dead cell stain (Life Technologies). Then, HA were stained with cytofix/cytoperm (BD Biosciences) with mouse anti-IL-22R1 (clone 305405, R&D Systems)/mouse IgG₁ isotype control (clone 11711, R&D Systems) or rabbit anti-IL-10R2 (sc-69580, Santa Cruz Biotechnology)/rabbit IgG isotype control (AB-105-C, R&D Systems) primary antibodies and followed by donkey anti-mouse IgG AF546 and goat anti-rabbit IgG AF488 (Invitrogen) secondary antibodies. Alternatively, cell suspension was directly used for staining as described in “Proliferation and cell death/apoptosis assays” section. Cells were processed with an LSRII flow cytometer (BD Biosciences) and were analyzed with FlowJo software (version 9.1.11, Treestar).

Peripheral blood mononuclear cells (PBMC) were freshly isolated by Ficoll (GE Healthcare Biosciences) as described previously [37]. PBMC subpopulations were sorted with anti-CD4, anti-CD8, anti-CD14, anti-CD19, and anti-CD56 MicroBeads (Miltenyi Biotec) with an autoMACS Pro Separator (Miltenyi Biotec) according to manufacturer’s instructions. The purity of sorted cells was checked by flow cytometry with the following antibodies: anti-CD4 APC-H7 (clone SK3, BD Biosciences), anti-CD4 ECD (clone SFCI12T4D11, Beckman Coulter), anti-CD8 FITC (clone RPA-T8, BD Biosciences), anti-CD11c APC (clone B-ly6, BD Biosciences), anti-CD14 PB (clone M5E2, BD Biosciences), anti-CD19 PE (clone HIB19, eBioscience), and anti-CD56 PE-Cy7 (clone NCAM16.2, BD Biosciences) on a LSRII flow cytometer (BD Biosciences). We did not pursue the proposed experiments if the purity of sorted cells was less than 90 %. Analyses were performed using FlowJo software (Treestar).

Sorted CD14⁺ cells were incubated for 6 days at 1*10e6 cells per ml in Roswell Park Memorial Institute (RPMI; Gibco, Life Technologies) supplemented with 10 % FCS (heat inactivated, PAA Laboratories), 50 ng/ml premium grade recombinant granulocyte macrophages colony-stimulating factor (GM-CSF) and 20 ng/ml premium grade recombinant IL-4 (Miltenyi Biotec) to obtain differentiated monocyte-derived DCs (moDCs). Cell differentiation quality was checked by flow cytometry with the following antibodies: anti-CD11c APC (clone B-ly6, BD Biosciences) and anti-CD14 PB (clone M5E2, BD Biosciences). Proper differentiation was considered as completed when at least 80 % of the harvested cells were CD11c⁺CD14⁻. For mRNA analysis, moDCs were lysed with RLT Plus buffer (Qiagen) and kept at −20 °C until further extraction.

Whole PBMC were left untreated or stimulated with 100 ng/ml staphylococcal enterotoxin B (SEB, Sigma) at 2*10e6 cells per ml for 18 h at 37 °C. Supernatants were harvested and stored at −80 °C until use. CD4⁺, CD8⁺, CD14⁺, CD19⁺, CD56⁺ sorted cells, and moDCs were either left untreated or stimulated at 2*10e6 cells per ml with 100 ng/ml SEB, 1 μg/ml resiquimod (R848) (InvivoGen)—a toll-like receptor 7 and 8 ligand, a potent activator of both monocytes and B cells—, 10 μ/ml CD3/28 beads (Gibco, Life Technologies) for 18 h or 100 ng/ml phorbol myristate acetate (PMA, Sigma) and 1 μg/ml ionomycin (iono, Sigma) for 6 h at 37 °C.

For mRNA analysis, cells were lysed with RLT Plus buffer (Qiagen) and kept at −20 °C until further extraction.

Measurement of IL-22 in the serum, cerebrospinal fluid (CSF), or supernatant of stimulated PBMC was performed with the Human IL-22 ELISA Ready-SET-Go (eBioscience) according to manufacturer’s instructions and read with Opsys MR (Dynex International) instrument.

IL-22BP was measured in the serum and CSF by a home-made kit. “Maxisorp Immunoplates” 96-well plates (Nunc) were coated with coating solution (15 mM Na₂CO₃, 34.8 mM NaHCO₃) mixed with goat anti-IL-22BP antibody (AF1087, R&D Systems) diluted 1:500 and incubated overnight at 4 °C. The next day, blocking was performed by filling 200 μl/well PBS containing 0.05 % Tween 20 and 1 % bovine serum albumin (BSA) (PBS/T-1 % BSA) (Sigma) with 2 h incubation at 37 °C. After three washes with wash buffer solution (BD Biosciences), 100-μl standard dilutions (1087-BP-025, R&D Systems) and samples diluted with 50 μl PBS/T-1 % BSA were added to each well and incubated overnight at 4 °C. The third day, plates were washed three times and 50 μl/well of rabbit anti-IL-22BP (sc-134974, Santa Cruz Biotechnology) diluted 1:200 in PBS/T-1 % BSA were added and 1 h incubation at 37 °C was performed. Following three washes, addition of 50 μl of mouse anti-rabbit biotinylated antibody (550346, BD Biosciences) diluted 1:3,000 in PBS/T-1 % BSA was performed and the plates were incubated 1 h at 37 °C. After another round of three washes, 50 μl/well of 1:200 streptavidin-HRP (DY998, R&D Systems) in PBS/T-1 % BSA was added and incubated at RT for 30 min. Finally, after six washing steps, ELISA was revealed and developed with 100 μl/well revelation buffer (DY999 R&D Systems) in a 20-min incubation period at RT, protected from light. Reaction was stopped with 50 μl/well 1 M H₂SO₄, and plates were read at 450 nm with Opsys MR (Dynex International) device. Detection limit was set at 0.5 ng/ml to fully guarantee specificity and reliability of positive samples, based on data of the recombinant IL-22BP standard curve (R&D Systems).

The biological material was lysed with RLT Plus buffer (Qiagen) and stored at −20 °C until RNA extraction. The RNA isolation was performed with the RNeasy Plus Mini kit (Qiagen). Up to 0.5-μg purified RNA (concentration measured with a NanoDrop 2000, Thermo Scientific) was taken for reverse transcription utilizing the Quantitect Reverse Transcription kit (Qiagen). Quantitative PCR was performed with the QuantiTect SYBR green PCR mix (Qiagen) and QuantiTect primer assays for 18S ribosomal RNA, IL-22BP set (Qiagen). MicroAmp Fast Optical 96-well reaction plate (Applied Biosystems, Life Technologies) was run with a StepOnePlus Real-Time PCR System (Applied Biosystems, Life Technologies). The relative expression of each sample was calculated with the “2e(−ΔCt)” equation where the Ct is defined as the cycle number at which the SYBR green fluorescence crosses the threshold (arbitrary set and fixed for all experiments performed) and the Δ is the difference between the Ct of the sample tested and the housekeeping gene Ct. Melting curve analysis was performed to ensure reaction specificity.

For immunohistochemistry staining, tissue sections were treated with 0.6 % hydrogen peroxide in 80 % methanol for 30 min to inactivate endogenous peroxidase and blocked with blocking buffer (1 % normal donkey serum, 0.1 % Triton, 0.05 % Tween) for 1 h. The tissue sections for myelin oligodendrocyte glycoprotein (MOG) were then additionally defatted in 100 % methanol for 8 min at −20C°. Sections were incubated with following primary antibodies overnight at 4 °C: mouse anti-MOG (clone Z12, kindly provided by R. Reynolds) to target myelin, mouse anti-IL-22R1 (clone 305405, R&D Systems)/mouse IgG₁ isotype control (clone 11711, R&D Systems), rabbit anti-glial fibrillary acidic protein (GFAP) (AB5804, Millipore; Z0334, DakoCytomation) to label astrocytes, and rabbit anti-Caveolin-1 (Cav-1) (N-20, Santa Cruz Biotechnology) for endothelia staining/rabbit IgG isotype control (AB-105-C, R&D Systems; 12–370, Millipore). Secondary biotinylated antibodies (Vector Laboratories) were applied for 1 h at room temperature, together with 4′,6-diamidino-2-phenylindole (DAPI, Invitrogen Life Technologies) counterstaining, followed by avidin-biotin complex reagent (Vector Labs) for 30 min. Color reaction was performed with 3-amino-9-ethylcarbazole. Cells were stained in hematoxylin for 5 min and rinsed afterwards under running tap water. Image acquisition was performed with a Zeiss Axiovision (Carl Zeiss Microscopy) microscope, and picture analysis was performed with Axiovision software (version V4.8.1.0, Carl Zeiss Microscopy).

Immunostainings were performed with the following primary antibodies: goat anti-IL-22 (AF782, R&D Systems)/goat IgG (AB-108-C, R&D Systems), mouse anti-IL-22R1 (clone 305405, R&D Systems)/mouse IgG₁ (clone 11711, R&D Systems), rabbit anti-IL-10R2 (sc-69580, Santa Cruz Biotechnology), rabbit anti-GFAP (AB5804, Millipore; Z0334, DakoCytomation), mouse anti-GFAP-Cy3 (for HA only, clone G-A-5, Sigma), rabbit anti-Caveolin-1 (N-20, Santa Cruz Biotechnology)/rabbit IgG (AB-105-C, R&D Systems; 12–370, Millipore), sheep anti-von Willebrand factor (VWF) (GTX74137, GeneTex) for vessel labeling, and chicken anti-microtubule-associated protein (MAP)-2 (ab5392, Abcam) to target neurons/chicken IgG (GTX35001, GeneTex) and with the following secondary antibodies: donkey anti-goat IgG AF488, donkey anti-mouse IgG AF546 and AF647, goat anti-rabbit IgG AF488, donkey anti-rabbit IgG AF647, goat anti-chicken IgG AF647, goat anti-sheep AF647 (Invitrogen) and, finally, donkey anti-rabbit IgG AF488 (Jackson ImmunoResearch) and donkey anti-rabbit IgG CSL467 (Santa Cruz Biotechnology). To reduce autofluorescence, tissue sections of the Basel cohort were incubated in cupric sulfate in ammonium buffer (10 mM CuSO₄, 50 mM CH₃COONH₃, pH 5.0) for 30 min before secondary antibody staining. Nuclei staining was done with DAPI (Invitrogen Life Technologies). Slices were mounted with Vectashield (Vector Laboratories). Image acquisition was done with a Zeiss LSM 710 Quasar (Carl Zeiss Microscopy) confocal, and picture analysis was performed with Zeiss ZEN 2009 (Carl Zeiss Microscopy), ImageJ (version 1.46r, National Institutes of Health), and Axiovision softwares (version V4.8.1.0, Carl Zeiss Microscopy). Images were taken, and post-acquisition processing (brightness and contrast) was done the same way for specific antibodies and their isotype controls.

For functional experiments, cells were cultured at low passage (three to six passages) in complete astrocyte medium, prior to transfer in 24-well plates (Costar) at 20,000 cells/well in RPMI only medium, 300 μl/well, on day −1. On day 0 and then every other day over a 9-day period, HA were treated with six different conditions: astrocyte medium without FCS as reference medium; RPMI only (referred as “untreated” or negative control) as standard minimal medium to provide only essential nutrient to the cells; IL-22 at 50 ng/ml (R&D Systems); tumor necrosis factor (TNF)α at 10 ng/ml (R&D Systems); TNFα and IL-22 co-treatment; and finally, 100 nM staurosporine (STS, Sigma) as positive control to induce apoptosis and cell death. TNFα was chosen as a pro-inflammatory cytokine considering its paramount role in MS [38].

To assess for cell proliferation, carboxyfluorescein succinimidyl ester (CFSE, Biolegend) staining was done at the beginning of the kinetic (day −1), such as already performed in the lab [39], whereas to assess for cell death and apoptosis, staining with 7-aminoactinomycin D (7-AAD, BD Biosciences) and Annexin V AF647 (Invitrogen Life Technologies) in Annexin-binding buffer (Invitrogen Life Technologies), respectively, were performed at day 1, 2, 3, 4, 5, 7, and 9, according to manufacturer’s instructions. Annexin V marker analysis was performed on 7-AAD⁻ pregated cells. Samples were run with an LSRII flow cytometer (BD Biosciences) and were analyzed with FlowJo software (version 9.1.11, Treestar).

First, using ELISA, we found that there was a strong trend (p = 0.07) for an increase of IL-22 protein in the serum of 63 MS patients as compared to 13 HC (Fig. 1a). Interestingly, the level of IL-22 in the serum of MS patients with active disease was higher than in the serum of inactive (p = 0.017) and progressive (p = 0.015) MS patients and, especially, of HC (p = 0.003) (Fig. 1b). IL-22 was not detectable in the CSF of patients with active MS (Fig. 1c). Of note, no lumbar puncture was performed in the other categories of study patients.

Next, we found that the supernatant of SEB-stimulated PBMC of 74 MS patients secreted a higher amount of IL-22 than of 32 HC (p = 0.0436, Fig. 1d), a finding which was ascribable to the active category of MS patients (active versus HC: p = 0.0048, active versus inactive: p = 0.0216, Fig. 1e). Then, we investigated which leukocyte subtypes secreted IL-22. We found that CD4⁺ T cells accounted for most of the production of IL-22; nevertheless, and as already reported, monocytes, B cells, CD8⁺ T cells, and natural killer (NK) cells were also able to produce and secrete significant amount of IL-22 (Additional file 1: Figure S1) [40]. Of note, unstimulated PBMC released a low, but not null, level of IL-22, consistent with previous reports [41]. Therefore, we tested several polyclonal stimulations (SEB, R848, PMA/ionomycin, and CD3/CD28 beads), and all showed similar efficacy, except for R848 which was induced much less IL-22 secretion from CD4⁺ T cells than other stimulants (Additional file 1: Figure S1).

To further examine the putative implication of IL-22 in MS, we looked at its soluble antagonist, i.e., IL-22BP. Indeed, IL-22BP gene polymorphism has been recently associated with MS [32]. Looking first at the protein level, we did not find a difference in terms of IL-22BP protein in the sera of 63 MS patients versus 13 healthy controls (HC); however, there was a trend (p = 0.14) towards an increased secretion of IL-22BP in MS patients as compared to HC (Fig. 1f). Those 76 study subjects and patients were the very same who were tested for the content of IL-22 in the serum (see above). Some MS patients harbored high levels of soluble IL-22BP, reaching levels of 10 ng/ml and more (Fig. 1f–g); however, there was no difference between the categories of MS patients (Fig. 1g). Interestingly, IL-22BP was detected in the CSF of 13/15 active MS patients who had a lumbar puncture at the same time as this assay (Fig. 1h).

Then, we found that among different sorted subpopulations of blood immune cells, CD14⁺ monocytes and, especially, in vitro differentiated moDCs contained the highest levels of mRNA coding for IL-22BP (Additional file 2: Figure S2), confirming previous literature data [31, 35, 42, 43]. We found an increased expression of IL-22BP mRNA in the monocytes of 51 patients as compared to 30 HC (p = 0.016; Fig. 1i) and in the moDCs of a subset of 15 MS patients as compared to 10 HC (p = 0.016; Fig. 1k). Interestingly, although there was no difference between the categories as a whole (p = 0.108 with the Kruskal-Wallis test), the higher expression of IL-22BP mRNA in the monocytes of MS patients seemed to be mostly accountable to the category of active MS (active MS patients versus HC: p = 0.037, Fig. 1j).

The fact that IL-22 and its soluble antagonist receptor, IL-22BP, are differentially expressed in MS patients as compared to HC suggests that this cytokine may be involved in the immunopathogenesis of MS. For this reason, we looked for putative target cells of IL-22 in the CNS. This cytokine is recognized through the IL-22 heterodimeric receptor composed of IL-10R2 and IL-22R1 subunits [44]. Of note, IL-10R2 can also bind IL-10, IL-26, IL-28(α/β), and IL-29, whereas IL-22R1 binds also IL-20 and IL-24 [30, 45], but only IL-22 is recognized by these two subunits together [45]. IL-10R2 has been previously shown to be expressed ubiquitously among tissues from hematopoietic and nonhematopoietic origins, whereas IL-22R1 expression was absent on hematopoietic cells [14, 41].

Here, to explore whether human brain CNS cells express the IL-22 receptor and whether IL-22 is present in the brain, we analyzed brain tissue sections from five MS cases and 18 non-MS controls. Using immunohistochemistry peroxidase staining, we aimed at detecting IL-22 and IL-22R1 on adjacent tissue sections with pictures taken at the exact same area for all samples (Fig. 2). We were able to detect IL-22 and IL-22R1 in the GM and the WM of both control subjects (Fig. 2a) and MS patients, either in areas deprived of lesions (Fig. 2b, c) or at the vicinity of plaques (Fig. 2d). Isotype controls showed that there was no unspecific staining neither of the primary nor of the secondary antibodies used in Fig. 2 (Additional file 3: Figure S3). Of note, MS plaques were identified by MOG and HE staining (example is shown in Fig. 2d). The expression of the cytokine and its receptor was clearly stronger in MS than control tissue (Fig. 2b–d compared to Fig. 2a). Interestingly, the morphology of the IL-22- and IL-22R1-positive cells clearly looked alike GFAP-positive astrocytes (Fig. 2c, arrow, first to third row), suggesting that the receptor and its cytokine were present on the same cells, for example see the two upper panels of Fig. 2c. The IL-22- and IL-22R1 expressing cells were often found in the vicinity of blood vessels. Since IL-22R1 was identified to be expressed by endothelial cells in MS [22], we stained for Caveolin-1. However, we found that the staining of IL-22 and its receptor reflected more the staining of GFAP than the one of Caveolin-1 (Fig. 2c, d in particular). Caveolin-1 delineates very nicely in blood vessels, whereas IL-22 and IL-22R1 are localized on fine processes around blood vessels and on cell bodies of stellate cells in proximity of blood vessels. Moreover, in MS lesions where strong astrogliosis occurs, IL-22 and IL-22R1 expressions was also enhanced (Fig. 2d).

Next, using immunofluorescent confocal microscopy, we wanted to determine whether IL-22 and IL-22 receptor indeed colocalize on the same cells. We also wanted to ascertain their expression on astrocytes as indicated by light microscopy (Fig. 2). We found that IL-22 colocalized with its receptor in the brain tissues of control subjects (Fig. 3a–d) as well as in the brain tissues of MS patients (Fig. 4). In the brain of control subjects, the IL-22/IL-22R1 colocalization seemed to be slightly more expressed in the GM than in the WM (Fig. 3a, b), whereas in MS, it was clearly stronger in the plaques, either in the WM (Fig. 4a, b) or the subpial GM (Fig. 4c, d), than in the normal appearing white matter (NAWM) (Fig. 4a, b). We were also able to confirm that the IL-22/IL-22R1 couple was expressed on astrocytes in the brain of control subjects (Fig. 3a, c) as well as of MS patients (Fig. 4a, c). By contrast, co-staining of the IL-22/IL-22R1 with Cav-1 was rarely detected (less than 1 % of Cav-1⁺ structures were also positive for IL-22/IL-22R1) (Fig. 3b). By analyzing adjacent slices of pictures taken at the very same location, we noticed strong IL-22R1 expression in white matter plaques (Fig. 4a, b) as well as in subpial lesions in gray matter (Fig. 4c, d).

To ascertain the specificity of the antibodies used for immunofluorescent confocal microscopy and to rule out autofluorescence, we performed isotype stainings, using the exact same protocol, acquisition parameters, and post-processing analysis methodology as for specific antibodies (Additional file 4: Figure S4 and Additional file 5: Figure S5). Of note, the pictures of the isotype controls were taken at the very same location in the tissue slices as the pictures of the specific antibodies, except for Additional file 4: Figure S4c, which was not immediately adjacent to Fig. 3c, d. Otherwise, tissue pictured in Additional file 4: Figure S4a, b was immediately adjacent to the one represented in Fig. 3a, b; the same was true for Additional file 5: Figure S5a with Fig. 4a, b and Additional file 5: Figure S5b with Fig. 4c, d.

Additional staining performed on biopsies from control patient L-C5 confirmed that there was a very strong colocalization of IL-22R1 with GFAP (Additional file 6: Figure S6a) but not with Cav-1 (Additional file 6: Figure S6b, c) or with MAP-2 (Additional file 6: Figure S6c). Staining with anti-VWF, an alternative blood vessel marker, led to the same results as with anti-Cav-1, i.e., no colocalization with IL-22R1 (Additional file 7: Figure S7). Interestingly, IL-22R1 expression was particularly high in the close vicinity of blood vessels, lining them up. However, we could show that this proximity was not due to endothelial expression of IL-22R1 (Additional file 6: Figure S6b) but was rather attributable to expression by astrocytic feet. Of note, to ascertain that the detection of IL-22R1 used so far indeed indicated the expression of the whole IL-22 receptor, we assessed the expression of the other subunit of the IL-22 receptor, i.e., IL-10R2, and found that, indeed, IL10R2 colocalized with IL-22R1, clearly establishing that the heterodimeric IL-22 receptor complex is fully expressed in the CNS (Additional file 6: Figure S6d). Altogether, this set of experiments establishes that the IL-22 receptor is expressed in the human brain, mostly on astrocytes, and that IL-22 colocalize with its receptor. We also demonstrate that even if the IL-22/IL-22R1 couple is present in the brain of control patients with other neurological disease than MS, it nevertheless predominates in the brain of MS patients, in particular in the plaques.

Having shown that the IL-22 receptor was expressed in the human brain, predominantly by astrocytes, we decided to investigate the role of IL-22 on astrocytes. To this end, we selected human primary astrocytes. The purity of those primary astrocytes, as assessed by GFAP staining on flow cytometry, was close to 80 % (Additional file 8: Figure S8). Such as shown by flow cytometry (Fig. 5a) and by immunofluorescence (Fig. 5b–d), these astrocytes expressed both subunits of the IL-22 receptor, confirming that IL-22 could bind to these cells.

In order to determine whether IL-22 has a protective or deleterious effect on astrocytes, we studied its impact in terms of cell death, such as measured by 7-AAD, a marker of cell viability.

We found that IL-22 protected human astrocyte from cell death. Indeed, nutriment-deprived astrocytes that had been treated with IL-22 (plain blue line) exhibited a significant higher survival than astrocytes cultured in the same conditions but without IL-22 (dashed blue line) (Fig. 6a, c). Even if the improved survival of astrocytes attributable to IL-22 was relatively mild (increase of survival 2.4 % on day 2, 7 % on day 3, and 4.2 % on day 5), it was significant and occurred during a time of the culture (day 2, 3, and 5), which corresponded to a nadir in terms of astrocyte survival (Fig. 6c, blue lines).

Then, we assessed whether the protective function of IL-22 on astrocytes was maintained in a still more hostile environment, i.e., under inflammatory conditions. Therefore, we compared TNFα-treated astrocytes with or without adding IL-22. We found that whereas there was a steady and constant cell death of the TNFα-treated astrocytes population which had not received IL-22 (dashed red line) (Fig. 6b, c), there was a much better survival of astrocytes co-treated with TNFα and IL-22 (plain red line) from day 7 (Δ = 8.9 %) to day 9 (Δ = 20.6 %) (Fig. 6b, c). Regarding TNFα- versus IL-22- and TNFα-treated human astrocyte conditions, TNFα-treated cells showed steady shrinkage of surviving cell number till the end of the kinetic (Fig. 6c). Addition of IL-22 to the TNFα-treated cells drastically ameliorated cell survival from day 7 on, as reflected by the decreased frequency of 7-AAD-positive coupled with a rise of 7-AAD-negative cell frequency, i.e., more alive astrocytes.

In an attempt to determine if an anti-apoptotic mechanism may account for the protective effect of IL-22, we assessed whether, among 7-AAD-negative living cells, there were cells in an early stage of apoptosis, such as revealed by Annexin V staining. Gating on live cells only (7-AAD-negative), we found that the mean fluorescence intensity (MFI) of Annexin V was significantly less intense in IL-22-treated cells than their untreated counterparts on the first day of culture (27.1 % MFI difference), showing that the former was less apoptotic than the latter (Fig. 6d). Furthermore, we found that IL-22 treatment significantly attenuated the pro-apoptotic effect of TNFα on day 9 of the assay (22.8 % MFI decrease, Fig. 6d).

Finally, some authors have shown that IL-22 promotes proliferation of epithelial cells [19]. Thus, we examined whether the pro-survival effect of IL-22 on astrocytes could be ascribed to this property of IL-22. However, we did not find such an effect of IL-22 (data not shown).