Inhibitory CD200-receptor signaling is rewired by type I interferon

CD200 Receptor 1 (CD200R) is an established inhibitory immune receptor that inhibits TLR-induced cytokine production through Dok2 and RasGAP. RasGAP can be cleaved under certain conditions of mild cellular stress. We found that in the presence of cleaved RasGAP, CD200R loses its capacity to inhibit rpS6 phosphorylation. Furthermore, IFNα pre-stimulation of human mononuclear cells results in increased amounts of cleaved RasGAP. Coherently, upon pretreatment with increasing concentrations of IFNα, CD200R gradually shifts from an inhibitor to a potentiator of TLR7/8-induced IFNG mRNA production. In peripheral blood mononuclear cells from Systemic Lupus Erythematosus (SLE) patients, a prototypic type I IFN disease, we found an increased proportion of cleaved RasGAP compared to healthy controls. In line with this, in subsets of SLE patients the inhibitory function of CD200R is lost or converted to a potentiating signal for IFNG mRNA production. Thus, our data show that type I IFN rewires CD200R signaling and suggest that this cell-extrinsic regulation of signaling could contribute to perpetuation of inflammation in SLE.


Introduction
Immune inhibitory receptors function to prevent collateral damage by dampening excessive immune cell activation (Ravetch and Lanier, 2000). CD200 Receptor 1 (CD200R) is known as an inhibitory immune receptor that limits pro-inflammatory responses in myeloid cells . Indeed, CD200fc, a Fc-fusion protein of the best-known ligand for CD200R, has been shown to have immune-suppressive capabilities (Cox et al., 2012;Huang et al., 2018;Simelyte et al., 2008;Yin et al., 2016). Mice deficient in CD200R or in CD200 are healthy under homeostatic conditions. However, upon challenge, these mice generally show increased susceptibility to autoimmune diseases and immune pathology Rygiel et al., 2008;Simelyte et al., 2008;Snelgrove et al., 2008). We previously reported that CD200R inhibits TLR7 signaling, resulting in enhanced anti-viral responses but also enhanced immunopathology upon viral infection in mice . All these observations are consistent with an immune-suppressive role for the CD200R-CD200 axis. In line with this, absence of the CD200/CD200R inhibitory axis results in enhanced anti-tumor responses (Kretz-Rommel et al., 2007;McWhirter et al., 2006;Petermann et al., 2007;Rygiel et al., 2012). Coherently, tumor CD200 expression in multiple human cancers is associated with poor prognosis (Aref et al., 2018;Chen et al., 2017;El Din Fouad et al., 2018;Moreaux et al., 2006) and blocking the CD200/CD200R inhibitory axis has been proposed as checkpoint blockade therapy in human cancer (Mahadevan et al., 2019).
However, the CD200R-CD200 axis has also been reported as a co-stimulatory axis (Borriello et al., 1997;Wang et al., 2019). Furthermore, CD200R is required to 'license' TLR2mediated immune activation by Herpes Simplex Virus 1 in macrophages (Soberman et al., 2012), and the CD200R/CD200 axis in some cases was suggested to regulate protective antitumor responses (Erin et al., 2015;Erin et al., 2018;Liu et al., 2016). Taken together, although CD200R is generally considered to be an inhibitory receptor, there are data that suggest a more versatile and/or complex functionality of CD200R.
CD200R signaling is different from classical inhibitory receptors: CD200R does not signal through an immune tyrosine inhibitory motif (ITIM), but through an NPxY motif (Mihrshahi et al., 2009). CD200R inhibits mitogen activated protein kinases (MAPK) by activating Dok2 and RasGAP/p120 (Mihrshahi et al., 2009;Zhang et al., 2004). The N-terminus of RasGAP/p120 contains the SH2 and SH3 domains that are required to interact with Dok2, the C-terminus contains the GTPase activating protein (GAP) domain to suppress Ras-signaling . RasGAP/p120 is a target of sub-lethal caspase 3 activity and can be cleaved after amino acid position 455 . The resulting N-terminal fragment, or 'fragment N', has different signaling properties from the full length RasGAP and activates Akt signaling to promote cell survival . Currently it is unknown what the individual contributions of Dok2 and RasGAP are to CD200R signaling and if cleavage of RasGAP affects the outcome of CD200R ligation.

CD200R signaling bifurcates and uses separate pathways to inhibit MAPK and PI3K/Akt
In order to study which signaling pathways are regulated by CD200R we generated U937 cells that ectopically express CD200R-GFP, or a truncated signaling dead control CD200RΔ-GFP (Fig. S1A). U937 cells have active Akt, Erk and rpS6 signaling pathways in culture. We assessed the outcome of CD200R ligation on these signaling pathways. Ligation of CD200R-GFP, but not CD200RΔ-GFP with an agonistic antibody (clone OX108) inhibited phosphorylation of mitogen-activated protein kinase (MAPK) Erk at Thr402/Tyr204 (p-Erk) and IL-8 secretion, similar to what was previously reported ( Fig. 1A and S1B) (Mihrshahi et al., 2009). Additionally, CD200R ligation inhibited phosphorylation of ribosomal protein S6 at Ser234/235 (p-rpS6) (Fig. 1B), which is a target of the Erk/MAPK pathway and of the Akt-mTORC1 pathways. Consistently, the inhibitory function of CD200R was not limited to MAPK. We found that CD200R also inhibited Akt at residues phosphorylated by mTORC2 (p-Akt473) and PI3K/PDK (p-Akt308) (Figs. 1C and 1D). Similar to antibody-mediated ligation, ligation of CD200R with recombinant human CD200 (rhCD200) inhibited p-Erk in U937 cells (Fig. S1C). We conclude that CD200R inhibits both the MAPK/Erk pathway and the Akt pathway and their common downstream target rpS6.
CD200R recruits and activates RasGAP through Dok2 (Mihrshahi et al., 2009). To determine the relative contribution of RasGAP and Dok2 in CD200R signaling, we generated U937-CD200R-GFP lines deficient for RasGAP or Dok2 using CRISPR/Cas9 (Fig. S1D). We used two different guides that yielded two different deficient U937 cell lines per target, and used these cells as separate polyclonal lines for experiments. In line with previous reports showing that Dok2 is essential for CD200R-mediated inhibition in U937 cells (Mihrshahi et al., 2009), Dok2 -/-U937-CD200R-GFP cells failed to inhibit both p-Akt and p-Erk, and IL-8 secretion (Figs. 1E-H). In contrast, without RasGAP, CD200R remained able to inhibit p-Akt, while CD200R still failed to inhibit IL-8 secretion and p-Erk (Figs. 1E-H). These data show that CD200R-signaling bifurcates at the level of Dok2 and RasGAP, and that in the absence of RasGAP, CD200R still inhibits Akt through a pathway that involves Dok2 (Fig. 1I).
CD200R signaling changes in the presence of cleaved RasGAP.
RasGAP is synthesized as a protein of approximately 120kDa, but can be cleaved by caspases during cell stress which disassociates the N-terminal Dok2-binding SH2-domain from the functional C-terminal GAP-domain . We ectopically expressed a N-terminal fragment of RasGAP that lacks the functional GAP-domain (RasGAP 1-455 ) in U937-CD200R cells (Fig. S2A) to test the functional consequence of cleaved RasGAP on CD200R function. Overexpression of RasGAP 1-455 did not affect the amount of p-Erk, p-Akt308, p-Akt473 or p-rpS6 without CD200R ligation (Fig. S2B). RasGAP 1-455 overexpression also did not significantly affect the ability of CD200R to inhibit p-Erk or p-Akt ( Fig. 2A and S2C). However, CD200R was not able to inhibit p-S6 in presence of RasGAP 1-455 overexpression ( Fig. 2A), showing that cleaved RasGAP affects the outcome of CD200R signaling.

Type I IFN exposure induces RasGAP cleavage.
RasGAP can be cleaved by caspase 3 . In human healthy control peripheral blood mononuclear cells (PBMCs) we found that type I IFN can activate caspase 3 activity ( Fig. 2B), and thus potentially affects RasGAP cleavage. Indeed, 16h exposure to type I IFN (IFNα) increased the amount of cleaved RasGAP fragments in PBMC and isolated monocytes (Figs. 2C -2E). We found the previously reported RasGAP cleavage fragments (fragment N of ~56kDa and fragment N2 of ~40kDa ), but also products of other sizes suggesting the involvement of other proteases. As opposed to RasGAP, IFNα prestimulation of PBMC and monocytes did not noticeably affect Dok2 protein expression (Figs.
Type I IFN exposure converts CD200R from an inhibitory to immunostimulatory receptor.
In absence of RasGAP, CD200R selectively inhibited Akt (Fig. 1). PI3K/Akt can provide negative feedback on the production of type I cytokines (Fukao et al., 2002;Ruhland and Kima, 2009). To clarify the role of Akt and Erk in the regulation of TLR7/8 signaling, we used selective inhibitors of Akt (AKTVIII, 2μM) and Erk (U0126, 2.5μM) in primary human monocyte derived dendritic cells (moDCs) that endogenously express functional CD200R (Figs. S2C and S2D), stimulated with the TLR7/8 ligand R848. While inhibition of Erk resulted in reduced secretion of IL-8, inhibition of Akt potentiated secretion  in the majority of the donors (Fig. 3A). This demonstrates that selective inhibition of Akt, which occurs if only Dok2 and not RasGAP/p120 is functionally recruited to CD200R, could potentiate TLR7/8 mediated responses. Therefore, we assessed the function of CD200R in freshly replated moDCs versus 16h IFNα pre-exposed moDCs seeded on immobilized recombinant human CD200 (rhCD200). In freshly replated cells, CD200R ligation by CD200 inhibited TLR7/8 ligand R848-induced p-Akt473 and p-rpS6 (Fig. 3B, 3C and S2E). IFNα pre-exposure marginally affected the ability of CD200R to inhibit R848-induced p-Akt473 ( Fig. 3B). However, after pre-exposure to IFNα, CD200R lost inhibitory capacity for p-rpS6 in all donors and even potentiated R848-induced p-rpS6 in some donors (Fig. 3C). Overall, we conclude that IFNα driven inflammation favors production of cleaved RasGAP fragments which re-wires the CD200R signaling cascade.
We next assessed the inhibitory capacity of CD200R in PBMC from healthy controls (HC).
In PBMC cultures, we found that IFNG mRNA and IFNγ protein production was induced by R848 and inhibited by CD200R ligation with immobilized antibodies (Figs. 3D and 3E).
IFNG mRNA in these cultures was most likely produced by NK cells, but orchestrated by monocytes (Figs. S3A-S3C) because depletion of either NK cells or monocytes (Chan et al., 1991;Okamura et al., 1995), but not T cells, prevented production of IFNG mRNA.
We tested the effect of IFNα on CD200R function in human PBMC by using freshly isolated cells, or cells that were pre-stimulated for 16h culture in medium with increasing doses of IFNα. Freshly isolated or pre-treated HC PBMC were seeded on immobilized agonistic CD200R antibody followed by stimulation with R848. CD200R ligation inhibited R848induced IFNG mRNA expression in freshly isolated PBMC from most healthy controls.
However, pre-treatment of PBMC with IFNα abrogated or reversed the capacity of CD200R to inhibit IFNG mRNA in almost all healthy donors (Figs. 3F and S3D). With increasing dose of IFNα, the ratio of IFNG mRNA production by CD200R stimulated cells over isotype control antibody stimulated cells, changed from <1 (inhibition) to 0 (no effect of CD200R) to >1 (potentiation by CD200R). Of note, after overnight culture of PBMC, the inhibitory function of CD200R was also partially lost compared to freshly isolated cells (Fig. 3F). This coincided with increased mRNA expression of type I IFN-stimulated genes IFITM1 and MX1, suggesting a type I IFN-response induced by overnight culture (Fig. S3E).

Increased RasGAP cleavage in systemic lupus erythematosus PBMC
Type I IFN is a feature of many systemic auto immune diseases and its pathogenic role is well established for systemic lupus erythematosus (SLE) (Crow and Rönnblom, 2018).
Furthermore, caspase activity is upregulated in SLE PBMC (Krishnan et al., 2005). We therefore assessed the availability of Dok2 and RasGAP in PBMC of SLE patients. At the protein level, both Dok2 and RasGAP expression is slightly decreased in PBMC from SLE patients (Fig. 4A). However, similar to what we observe after IFNα pre-stimulation in healthy control PBMC, we found an increase in cleaved RasGAP fragments in PBMC from SLE patients (Fig. 4A). This results in a lower ratio of RasGAP/p120 over cleaved RasGAP (Fig. 4B). Together these data suggest that RasGAP/p120 cleavage is enhanced by IFNα in vivo, which is exemplified in SLE PBMC.

CD200R function is abrogated or inverted in PBMC from SLE patients
The observed increased RasGAP cleavage could result in rewiring of the CD200R signaling in (part of) SLE patients. To test this, we used PBMC from clinically well-defined SLE patients and age-and sex-matched healthy controls (HC) in two different cohorts. Antibodymediated CD200R ligation inhibited TLR7/8-mediated IFNG mRNA expression in most HC PBMC in cohort 1 (Fig. 5A, Table S1). In contrast, in SLE PBMC with a detectable IFNG mRNA response, with the exception of one patient, CD200R was unable to inhibit IFNG expression, and even potentiated IFNG expression in response to TLR7/8 ligation in 5 out of 7 SLE patients (Fig. 5A, Table S1). In a separate replication second cohort, CD200R function was also lost or inverted in 7 out of 11 SLE patients (Fig. 5B, Table S1). We excluded that CD200R function was lost because of lower CD200R expression: instead in SLE patients CD200R expression was increased on non-classical inflammatory CD14 -CD16 + monocytes (Fig. S4A), and on CD56 + NK cells and CD19 + B cells (Fig. S4B). Medication used for SLE patients could potentially affect TLR7/8 responses and therefore CD200R function. Indeed, patients treated with hydroxychloroquine (HCQ), prednisone (Pred) or azathioprine (Aza) had decreased IFNG mRNA production after R848 stimulation (Fig.   S4C). However, none of these drugs had an influence of the function of CD200R in these patients (Fig. S4C). Coherently, HCQ pre-treatment of HC PBMC did not change the function of CD200R (Fig. S4D), confirming that the functional switch we observed for CD200R was not the result of medication. Interestingly, the reversal of CD200R function was most prominent in SLE patients with Lupus Nephritis (LN), a severe manifestation of SLE, suggesting a link between aberrant CD200R inhibition and disease severity (Fig. 5C).

Discussion
Our data show that CD200R signaling bifurcates at the level of Dok2, inhibiting the PI3K/Akt pathway through Dok2, and the Ras/Erk pathway through recruitment of both RasGAP to Dok2. Type I IFN exposure results in RasGAP cleavage which changes the outcome of CD200R signaling. Consequently, CD200R potentiates R848-induced IFNG mRNA production in PBMC from HC that are exposed to type I IFN, and in untreated PBMC of part of SLE patients. This functional switch is not binary: the functional outcome of CD200R ligation seems to progress along a sliding scale from inhibition to potentiation.
Previously, immune receptors with an immune tyrosine switch motif (ITSM) were shown to relay an activating or inhibitory signal by using different adaptor molecules (Wu and Veillette, 2016). For CD200R, we now propose a different mechanism, where extracellular cues can rewire the intracellular signaling of an immune receptor by inducing cleavage of their adaptor molecules. Hence, CD200R is a versatile receptor that, depending on cellextrinsic cues, can function anywhere between an inhibitory to potentiating immune receptor.
How would the immune system benefit from this versatility of CD200R? In homeostatic conditions, ligation of CD200R results in suppression of TLR responses in healthy donors ( Fig. S5A). This is in line with published work in CD200 or CD200R knock-out mice showing stronger inflammatory responses and immunopathology Karnam et al., 2012;Rygiel et al., 2009;Snelgrove et al., 2008), and that treatment with CD200-fc prior to disease onset dampens renal pathology in lupus prone mice (Yin et al., 2016). We reason that this inhibitory function is a useful threshold for TLR-mediated immune activation and keeps immune cells quiescent via CD200 expressed on, for example, endothelial cells, glomeruli of the kidney, and apoptotic cells (Li et al., 2012;Rosenblum et al., 2004;Wright et al., 2001).
However, we hypothesize that in the context of infection, immune cells would need to induce rather than inhibit type I immunity to clear intracellular pathogens (Fig. S5B). Our data show that in such an inflammatory milieu, CD200R can switch function to potentiate TLR-induced IFNG production, a hallmark type I immunity cytokine. This potentially enhances intracellular pathogen clearance, after which inflammation will cease. Thus, we hypothesize that the versatility of CD200R aids the immune system to quickly adapt to changing inflammatory environments. This beneficial feature of CD200R goes awry the moment there is type I IFN production in the absence of an (infectious) source of inflammation that can be cleared, as is the case for SLE (Fig. S5C). In a subset of SLE patients, CD200R indeed potentiated IFNG mRNA production.
We found that CD200R signaling impinges on Erk, Akt and their common downstream target rpS6. These targets can explain potentiation of R848-induced IFNG at several levels. Akt can provide negative feedback on TLR signaling (Fukao et al., 2002;Ruhland and Kima, 2009).
Indeed, inhibition of Akt potentiated inflammatory cytokine secretion in human moDC. Our data show that CD200R selectively inhibits Akt in absence of RasGAP, which would inhibit a negative feedback loop which leads to increased IFNG production. Additionally, in presence of cleaved RasGAP and after type I IFN treatment, CD200R loses its inhibitory capacity towards rpS6, a target of mTORC1. mTORC1 is involved in inflammasome activation (Moon et al., 2015), which via IL-18 production can drive IFNG mRNA production (Chan et al., 1991;Okamura et al., 1995). Rapamycin, an inhibitor of mTORC1, reduces lupus disease activity, measured among others by SLEDAI score (Fernandez et al., 2006;Lai et al., 2018). Besides direct inhibition of IFNG production by T cells, rapamycin could overrule the CD200R-effect on rpS6 as an additive layer for the effect of rapamycin treatment of SLE patients. Overall, the type I IFN-mediated rewiring of CD200R signaling to potentiate rpS6 signaling could contribute to the chronicity of SLE.
Interfering with the CD200-CD200R pathway is suggested to be a therapeutic opportunity in cancer and auto-immunity (Kretz-Rommel and Bowdish, 2008;Mahadevan et al., 2019;Simelyte et al., 2008). However, depending on the setting, the CD200-CD200R axis is reported to either have pro-and anti-tumor effects. Our data show that the outcome of CD200R-ligation depends on the inflammatory milieu. Thus, the outcome of therapeutic targeting of the CD200-CD200R axis will likely depend on the inflammatory status of the patient at the moment of treatment.
In conclusion, we demonstrate that the signal transduction machinery of the immuneinhibitory receptor CD200R is responsive to type I IFNs thereby allowing CD200R to progressively switch from inhibitory to stimulatory signaling depending on the environment.
As a general concept, the rewiring of inhibitory receptors by inflammatory cues has implications for therapeutic approaches targeting inhibitory receptors in patients. For moDC differentiation, monocytes were isolated from Buffy coats by ficoll gradient followed by percoll gradient and additional adhesion isolation. We are not informed about the sex and age of these donors. Monocytes were differentiated with GM-CSF (800U/ml) and IL-4 (500U/ml) in enriched RPMI (Gibco, 10% bovine serum, P/S, 2mM Glutamine) in a 5% CO 2 humidified 37°C incubator as previously described (Geijtenbeek et al., 2003), and used at day 6 or 7 after the start of differentiation.

Materials and Methods.
All cells were maintained at 37˚C with 5% CO2 in a humidified cell culture incubator.
CD200R-GFP was generated by adding GFP to the c-terminal domain of CD200R, preceded by a short flexible glycine-serine (GSGSGS) spacer. CD200R∆-GFP was truncated directly after the predicted transmembrane domain. To obtain CD200R-GFP and CD200R∆-GFP, U937 cells were transduced using standard retroviral transduction protocols and we used fluorescence-activated cell sorting to obtain equal expression levels.
U937 expressing full length CD200R-GFP were transduced with an in-house generated lentiviral construct containing both the gRNA and Cas9 (van de Weijer et al.). After transduction, cells were selected with puromycin and kept as poly-clonal lines. Successful deletion of the target gene was confirmed by intra-cellular antibody-staining followed by flow cytometry.
Where indicated, PBMC or moDC were pre-incubated for 16 hours with 1000U/ml IFNα 2a (Cell Sciences) or enriched RPMI (RPMI, 10% bovine serum, P/S, 2mM Glutamine) in a 5% CO 2 humidified 37°C incubator and washed with medium before proceeding with the experiment. Before seeding cells, the plates were washed three times with 100μl PBS. 100μl cell-suspension of moDC were seeded at 0.5E6 cells/ml and PBMC were seeded at 1E6/ml in enriched RPMI in a 96-well plate. After 2 hours of pre-incubation on the coated wells, we stimulated cells with 1 μg/ml of R848 (Invivogen). mRNA was isolated 4 hours after stimulation with R848, cell free supernatant was harvested 6 hours after stimulation with R848, and for flow cytometric analysis of phosphorylated proteins cells were fixed 30 minutes after R848 stimulation (see "PhosFlow Cytometry" for details).
For IL-8 secretion, cells were stimulated with 200ng/ml LPS (Salmonella Typhosa, Sigma) and cultured for 24 hours in a 5% CO 2 humidified 37°C incubator before harvesting cell-free supernatant for ELISA.

Viral plasmids
CD200R was fused to eGFP by two-step sewing PCR. First CD200R or CD200RΔ, and eGFP were amplified with the following primers and appropriate sized fragments were isolated from agarose gels with the PCR clean-up Gel extraction (Macherey-Nagel). PCR fragments were mixed and a consecutive PCR was done with the CD200R-F and GFP-R primers before digestion with PmlI and NotI and ligation into pMX. We used sanger sequencing to validate the sequence in the plasmid matched the expected sequence. Relative expression = 2^(Ct(B2M)-Ct(target)).

Flow cytometry.
Cells were washed in flow cytometry buffer (PBS, 1% BSA, 0.01% sodium azide, from here on FACS buffer), and incubated with antibodies for 30 minutes on ice, washed twice and analyzed on a BD Fortessa or BD Canto II using DIVA 8 software. For multi-color experiments, compensation settings were generated using the compensation tool in DIVA 8.
For data analysis we used FlowJo V10 to gate for live cells, single cells and subsequent gating as indicated in supporting figures.

Intra cellular flow cytometry for phosphorylated proteins (PhosFlow)
For analyses of phosphorylated proteins with flow cytometry, cells were fixed after stimulations with a final concentration of 3% PFA in medium for 10 minutes at room temperature. Attached cells were detached by scraping with a plunger. Cells were harvested, transferred into a 96-well V-bottom plate and washed, and subsequently permeabilized with 100% ice old methanol for at least 5 minutes and a maximum of 2 weeks at -20˚C.
Before continuing with the staining protocol, the PFA/Methanol fixed cells were washed twice. Staining was performed in flow cytometry buffer and non-specific binding was prevented by incubating the cells with 5% normal goat or mouse serum for 45 minutes at room temperature. The primary antibodies were incubated shaking for 1 hour at room temperature or 16 hours at 4°C. Cells were washed three times in 150μl FACS buffer and incubated with a conjugated secondary antibody for 1 hour at room temperature while shaking. After washing three times, fluorescence was assessed on a BD Fortessa or BD Canto II using a high throughput sampler attached to the flow cytometers.
'live' single cells were gated and median fluorescent intensity was used for quantification.
For U937-CD200R and -CD200RΔ cells, cells were gated for similar expression of CD200R-GFP before assessing phosphorylation staining.
and categorized the effect of CD200R as follows: neutral: less than 15% inhibition or potentiation; inhibition: ≥ 15% inhibition; potentiation: ≥15% potentiation. We tested the effect of other thresholding ranging from 5% to 50% inhibition and potentiation, which did not affect the overall conclusion (not shown).The ratio to depict CD200R function was calculated as: target anti-CD200R /target isotype expression so that a value below 1 represents inhibition, and a value above 1 potentiation.
The inhibitory capacity of CD200R was calculated as follows: (% inhibition in the condition of interest)/( % inhibition in CD200R-GFP U937) x 100. A value of 100 indicates that the treatment did not affect CD200R function and a value of 0 indicates complete loss of function.

Statistics
In all figure legends we have indicated the statistical test used to determine significance, and the n of experiments, or the data of individual donors is depicted. Statistical analysis was performed in Prism 7.1 (GraphPad software) or in R 5.3.1. All tests in this study are twosided. All boxplots are plotted as median with 25 and 75 percentiles with handles range from minimum to maximum values. Bars are mean±SEM.