MCAM/CD146 Signaling via PLCγ1 Leads to Activation of β1-Integrins in Memory T-Cells Resulting in Increased Brain Infiltration

Multiple sclerosis is a chronic auto-inflammatory disease of the central nervous system affecting patients worldwide. Neuroinflammation in multiple sclerosis is mainly driven by peripheral immune cells which invade the central nervous system and cause neurodegenerative inflammation. To enter the target tissue, immune cells have to overcome the endothelium and transmigrate into the tissue. Numerous molecules mediate this process and, as they determine the tissue invasiveness of immune cells, display great therapeutic potential. Melanoma cell adhesion molecule (MCAM) is a membrane-anchored glycoprotein expressed by a subset of T-cells and MCAM+ T-cells have been shown to contribute to neuroinflammation in multiple sclerosis. The role of the MCAM molecule for brain invasion, however, remained largely unknown. In order to investigate the role of the MCAM molecule on T-cells, we used different in vitro and in vivo assays, including ex vivo flow chambers, biochemistry and microscopy experiments of the mouse brain. We demonstrate that MCAM directly mediates adhesion and that the engagement of MCAM induces intracellular signaling leading to β1-integrin activation on human T-cells. Furthermore, we show that MCAM engagement triggers the phosphorylation of PLCγ1 which is required for integrin activation and thus amplification of the cellular adhesive potential. To confirm the physiological relevance of our findings in vivo, we demonstrate that MCAM plays an important role in T-cell recruitment into the mouse brain. In conclusion, our data demonstrate that MCAM expressed on T-cells acts as an adhesion molecule and a signaling receptor that may trigger β1-integrin activation via PLCγ1 upon engagement.


INTRODUCTION
Multiple sclerosis (MS) is a common autoimmune disease of the central nervous system (CNS) that is characterized by demyelination [reviewed in (1)]. One hallmark of MS is the wave-wise invasion of peripheral immune cells into the CNS forming so-called plaques which are characterized by high inflammatory activity, progressive destruction of the myelin sheath and neuronal death [reviewed in (1)]. Several previous studies have shown that, besides myeloid cells, blood-derived T-cells are the major drivers of local CNS inflammation in MS and among these, particularly T-cells of the Il-17 producing subpopulation (Th17/Tc17 subtype) (2)(3)(4)(5)(6). Tissue invasion of immune cells includes a highly regulated recruitment cascade that consists of tethering, adhesion, rolling and transmigration allowing the cells to leave the blood stream and to enter the target tissue. To overcome the blood brain barrier (BBB), T-cells largely rely on the interaction of the integrin VLA-4 (very late antigen-4; b1a4; CD49d/CD29) with endothelial VCAM-1 (vascular cell adhesion molecule 1) (7)(8)(9) and thus, Natalizumab, an antibody blocking VLA-4, is commonly used as a therapeutic approach (8). However, also VLA-4 independent mechanisms for T-cell recruitment into the brain must exist as patients experience relapses even during Natalizumab therapy (10).
MCAM expression has been linked to subsets of IL-17 producing T-cells (31,(34)(35)(36)(37)(38)(39) and is in fact a criterium for the definition of Th17 cells (37,40). We and others have previously shown that MCAM expression improves adherence of T-cells in an in vitro model of the BBB (41) and penetration of the blood cerebrospinal fluid barrier (BCSFB) in vitro and in vivo (23,31) and further, that MCAM expressing T-cells reside to active lesion sites in MS patients (41). Thus, MCAM expression might be an important mechanism of CNS invasiveness of T-cells. As the particular function of MCAM in T-cell migration remains elusive so far, the aim of this study was to characterize the contribution of MCAM-ligand interactions to T-cell invasion into the CNS using primary human and murine MCAM expressing effector memoryand central memory T-cells (T EM /T CM ) mechanistically by using different in vitro and in vivo approaches analyzing both adhesion and intracellular signaling.

Ethics Approval
All experiments including human material were approved by the local ethics committee (Ethik-Kommission der Ärztekammer Westfalen-Lippe und der Medizinischen Fakultät der Westfälischen-Wilhelms-Universität, registration number 2010-245-f-S) and performed according to the Declaration of Helsinki. All experiments involving mice were approved by the responsible animal protection authority (Landesamt für Natur-Umwelt-und Verbraucherschutz Nordrhein-Westfalen) and conducted according to the German Tierschutzgesetz.

Flow Chamber Assays
Flow chamber assays were performed as described previously (41)(42)(43)(44)(45). Briefly, chambers were coated for 2 h at RT using recombinant human VCAM-1 (R&D; concentration as indicated in results section), recombinant human MCAM (Thermo; concentration as indicated in results section), and recombinant human laminin-411 (Biolamina, concentration as indicated in results section). Chambers were blocked in Casein (Blocker TM Casein; Thermo) for 1 h at RT. For the measurements, the cells were diluted to 1*10^6/ml in RPMI1640 (PAN Biotech) + 10% FCS (Sigma) + 1% PS (PAN Biotech), constant flow was established for 1 min and the flow was stopped for 30 s. Then constant flow was reestablished and then the adhering cells in the whole area of the capillary (0.02 x 0.2 mm) were counted under flow within 1 min. All flow chamber experiments were performed in a completely standardized way concerning the time of flow, the temperature (RT), the shear stress applied, the dilution of the cells and by one researcher blinded concerning the respective population and treatment. Regarding adherence to MCAM and laminin-411 all different conditions were assessed with cells from each donor to provide maximal comparability.

Phalloidin Staining of Human MCAM+/-Central Memory T-Cells
Glass cover slips were coated with human recombinant MCAM (Thermo; 10 µg/ml)/laminin-411 (Biolamina; 10 µg/ml) for 2h at RT. The cells were diluted to 1*10^6/ml in RPMI1640 (PAN Biotech) + 10% FCS (Sigma) + 1% PS (PAN Biotech) and incubated rolling on the coated cover slips on a shaker for 1h at RT at 100 rpm. Then, 8% PFA (Sigma) in PBS was carefully added (final concentration 4% PFA) to the cells to fix them in situ. After incubating at RT for 10 min, the cells were permeabilized in PBS (PAN Biotech) + 100 mM Glycine (Sigma) + 0.1% triton (AppliChem) for 10 min at RT. Cells were stained in Alexa Fluor647 Phalloidin (Thermo; 1:40 in PBS + 1% BSA) for 30 min at RT in the dark. Next, the cells were carefully washed in PBS and stained with DAPI (Sigma; 1µg/ml in PBS) for 10 min at RT in the dark. Cells were washed in PBS and mounted on object slides using fluorescence mounting medium (Dako).

VCAM-1 Binding Assays
Cells were diluted to 2*10^6/ml. 0,2 µg recombinant human VCAM-1 (R&D) per sample was labelled by adding anti human Fc-antibody (APC, Southern Biotech; 5:1) on ice for 10 min. Cells were stimulated by rolling over MCAM ligands as described above. Labelled VCAM-1 was added to the cells and incubated for 30 min at 37°C, cells were fixed in 4% PFA (Sigma) and VCAM binding was assessed on a BD Canto II flow cytometer.

Electroporation of Human T-Cells
Electroporation of human T-cells was performed using the Amaxa human T-Cell Nucleofector Kit (Lonza, VPA-1002) and the Nucleofactor 2b system (Lonza), according to the manufacturer's instructions using RPMI1640 (PAN Biotech) + 20% FCS (Sigma) + 1% PS (PAN Biotech). The cells were diluted to 1*10^6/ml. Centrifugation steps were performed at 400 x g for 5 min and the cells were incubated for 30 min at 37°C + 5% CO 2 after electroporation before plating. For the PLCg1 knock down predesigned silencer siRNA (nc and B 100 nM; ABC 300 nM; Human siRNA Oligo Duplex, locus ID 5335, origene) were used.

Western Blotting
Western blotting was performed using the BioRad system according to standard protocols. Cell lysates were diluted in 5x Lämmli buffer (1M Tris PH = 6.8; 10% SDS; 50% glycerol; 500 mM b-mercaptoethanol; 0.5% bromophenol blue; all Sigma), boiled for 10 min at 95°C and separated in 10-12% glycine gels, blotted on Amersham protran nitrocellulose membranes (pore size 0,45 µm; Sigma) and blocked in TBST + BSA (3%, PAN Biotech) for 1h at RT. Then the membranes were incubated with the following primary and secondary antibodies over night at 4°C or for 1h at RT, respectively: (rabbit anti SRCp416, 1:5,000; rabbit anti FAK1p397, 1:1,000; rabbit anti Plcg1p783, 1:1,000; rabbit anti p38, 1:2,000; anti rabbit-HRP, 1:2,000; all Cell Signaling). Membranes weredeveloped using the ECL kit (GE Healthcare) and light sensitive Amersham Hyperfilms (GE Healthcare). Quantification of protein expression was performed by ImageJ. The bands were labelled as region of interest, background subtraction and area under the curve quantification was performed using the plot lanes and label peak option for gels and normalized to the expression of p38 on the same membrane for individual blots.

Electroporation of Murine T-Cells
Electroporation of murine T-cells was performed using the Amaxa murine T-Cell Nucleofector Kit (Lonza) and the Nucleofactor 2b system (Lonza), according to the manufacturer's instructions. Centrifugation steps were performed at 400 x g for 5 min and the cells were incubated for 30 min at 37°C + 5% CO 2 after electroporation before plating. To knock down PLCg1 in murine cells predesigned silencer siRNA (nc and A 300 nM, ABC 300 nM; Mouse siRNA Oligo Duplex, locus ID 18803, origene) were used.

Adoptive Transfer of Murine T-Cells
Murine CD4+ T-cells were isolated from spleens of 6 2D2 transgenic mice (male + female, 8-12 weeks of age), differentiated and transfected as described above. 24 h post siRNA transfection, labelling with cell tracker green (CMFDA; Thermo) was performed according to the manufacturer's instructions. The cells were washed and 3*10^6 T-cells in 100 µl NaCl were transplanted per recipient mouse by i.v. injection.

Viability Test of Murine and Human T-Cells
For testing cell viability murine and human T-cells, a combination of the Zombie green fixable viability kit (Biolegend) and 7-AAD (Biolegend) was applied strictly according to the manufacturer's instructions. Briefly, cells were washed in PBS (PAN Biotech) and resuspended in 100 µl PBS (PAN Biotech) + Zombie stain (1:500) per 1*10^6 cells, incubated for 30 min at RT in the dark and washed in PBS (PAN Biotech) + 10% FCS (Sigma). Then 7-AAD staining was performed 1:20 in PBS (PAN Biotech), the cells were fixed in 4% PFA (Sigma) and the cell viability was assessed on a BD Canto II flow cytometer.

Statistics
The indicated numbers of performed experiments (n) reflect the number of individual donors. Statistical analysis was performed using the Graphpad PRISM 6.01 software. For all experiments the distribution of values was assessed and accordingly, group comparisons were performed applying the appropriate T-test (normal or Mann-Whitney). Multiple comparisons were performed applying the two-way ANOVA including correction for multiple testing (Sidak). Statistical significance was assumed for p < 0.05 (*), p < 0.01 (**), p < 0.001 (***).  Figure S1 a) (46). MCAM+ T EM and T CM thereby exhibited equal levels of MCAM expression (Figure S1 b) and expression of IL-17a was confirmed to be associated with the MCAM+ subpopulations of T CM and T EM (Figure S1 c).

Co-Coating of Flow Chambers
First, to analyze whether and how MCAM expression on different T-cell subsets affects the adhesive properties of human central-and effector memory T-cells (T CM , T EM ), we performed flow chamber assays in vitro using primary human MCAM+ and MCAM-T EM and T CM separately ( Figure 1A). However, as our assays analyzing MCAM mediated adhesion revealed no differences between T EM and T CM we focus on T CM in the main figures and the data concerning T EM is presented in the supplemental material. To provide a mechanistic analysis of  Figure 1B), thereby confirming the feasibility of the assay (7)(8)(9).
Both MCAM and laminin-411 are shown to be expressed on the endothelium or in the extracellular matrix (ECM) of the BBB as well as the choroid plexus (23,31), which represent the main entry site of peripheral immune cells to the CNS (47,48). As the physiological expression levels of endothelial MCAM have been reported to be variable (49,50) we analyzed the adhesive properties of T-cellular MCAM by performing titration experiments using flow chambers coated with increasing amounts of MCAM (homophilic interaction) or laminin-411 (heterophilic interaction) using cells from the same donor to provide maximal comparability. We found that only MCAM expressing memory Tcells adhered to MCAM in our flow chambers ( Figure 1C, Figure  S2 a). In contrast, both MCAM+ and MCAM-T CM adhered to laminin-411 ( Figure 1D, Figure S2 b), probably due to the expression of laminin binding integrin a6 ( Figure S1 d/e). The interaction of MCAM with both ligands was concentration dependent and specific, as the adhesion could be abolished by a blocking MCAM antibody ( Figures 1C, D, Figure S2 a/b). In addition, we hypothesized that MCAM engagement induces an intracellular signaling cascade that leads to b1-integrin activation and thereby strengthens adhesion. To test this hypothesis, we used co-coated flow chambers with either MCAM or laminin-411 in combination with increasing concentrations of VCAM-1. We demonstrated that the adherence of MCAM+ T-cells is significantly increased in capillaries coated with MCAM/laminin-411 in addition to low concentrations of VCAM-1 (Figures 1E, F; Figure S2 c/d). This phenomenon exceeded the simple additive effect, thus indicating that MCAM binding indeed induces a signaling cascade supporting VLA-4 activation and VCAM-1 binding. This assumption was further affirmed by the fact that increased adherence to VCAM-1 in presence of laminin-411 was only induced in MCAM+ T CM , even though MCAM-T CM also bind laminin-411 ( Figure 1F). Morphological analyses of MCAM+ T CM after adherence to MCAM/laminin-411 additionally point towards an activated phenotype upon interaction with the substrate (Figures 1G, H).

Signaling via the MCAM Molecule Triggers SRC, FAK1 and PLCg1 Activation in MCAM + Memory T-Cells
To investigate which signaling pathway is induced by MCAM engagement, we stimulated MCAM+ and MCAM-primary human memory T-cells with either MCAM or laminin-411 under shear conditions in vitro (Figure 2A). Induction of phosphorylation of the intracellular kinases SRC, FAK1 and PLCg1 upon interaction with MCAM-specific ligands was only detectable in MCAM expressing T-cells, but not in MCAM-Tcells ( Figures 2B-D, Figures S3 a-f), whereas PYK2, PLCg2 or WASP did not get activated upon stimulation with MCAM-ligands ( Figures S3 g-j). Importantly, increased phosphorylation of SRC, FAK1 and PLCg1 was abolished in MCAM expressing T-cells pretreated with a blocking MCAM antibody before stimulation ( Figures 2B-D, Figures S3 a-f).

Pharmacological Inhibition and Specific Knockdown of PLCg1 Significantly Decreases VCAM-1 Binding in Primary Human MCAM+ Memory T-Cells by Inhibiting b1-Integrin Activation
To further substantiate the finding that MCAM signaling induces SRC, FAK1 and PLCg1 activity, we blocked the activation of the different kinases by commercially available inhibitors. In flow chamber experiments using capillaries co-coated with VCAM-1 in combination with either MCAM or laminin-411, application of the PLCg inhibitor U73122 significantly reduced adhesion compared to its control U73433 ( Figures 3A, B, Figure S4 a).

Notably, PLCg inhibition only lead to decreased adherence in MCAM expressing memory T-cells, but not in MCAM-T-cells.
These findings were further confirmed in a VCAM-1 binding assay ( Figure 3C and Figure S4 b). Interestingly, in the VCAM-1 binding assays PLCg inhibition lead to reduced VCAM-1 binding also in MCAM-T-cells suggesting that PLCg plays a central role in VLA-4 activation.
As U73122 is not specific for PLCg1 but also inhibits PLCg2, we next sought to specify our findings by performing a specific PLCg1 knock down in primary human memory T-cells using different siRNA oligonucleotides (Figures 3D, E; nc: negative control, B: siRNA B, ABC: mixture of siRNA A, B, and C). Cell viability was assessed by a flow cytometry-based approach and shown to be comparable in all treatment groups before and after the PLCg1 knock down (Figure S4 d). VCAM-1 binding assays of oligonucleotide treated MCAM+/-memory T-cells were performed 48h after inducing the PLCg1 knock down and confirmed that VCAM-1 binding was increased in MCAM+ Tcells upon in vitro stimulation with MCAM and laminin-411 (negative control). Importantly, this effect was absent in PLCg1 knock down cells (oligonucleotides B, ABC) and in MCAM-Tcells ( Figure 3F, Figure S4 e).
Next, we sought to go one step further and test whether we can additionally link MCAM signaling via PLCg1 to b1-integrin activation. To that end, we used a flow cytometry-based assay to investigate b1-integrin activation (using an antibody that specifically detects the full active conformation) after in vitro stimulation of the cells with either MCAM or laminin-411. Here, we demonstrated that the stimulation with either MCAM or laminin-411 significantly increased b1-integrin activation in MCAM+ memory T-cells, but not in MCAM-cells (U73433; Figure 3G and Figure S4 f). Pharmacological inhibition of PLCg also inhibited MCAM-triggered b1-integrin activation on MCAM+ memory T cells (U73122; Figure 3G and Figure S4 f/g).

Knockdown of Plcg1 Results in Impaired CNS Infiltration of MCAM+ T-Cells In Vivo
To analyze whether our finding that MCAM-triggered signaling leads to b1-integrin activation via PLCg1 also affects T-cell behavior in vivo, we performed adoptive cell transfer experiments. For this, we first isolated CD4+ T-cells from spleens of 2D2 transgenic mice and cultured them under conditions that were reported to expand, activate and enrich MCAM+/Th17-cells (23,31). On average 75% of CD4+ cells were shown to be MCAM+ IL17a+ after differentiation ( Figures  4A, B). Next, Plcg1 knock down was performed using a set of different siRNAs. The knock down efficiency was variable between the different siRNAs as shown in Figures 4C, D (nc: negative control, A: siRNA A, ABC: mixture of siRNA A, B, and C) but cell viability was comparable between the differently treated cells ( Figure S5). Twenty-four hours after inducing the Plcg1 knock down the cells were labelled with cell tracker green, and re-transplanted into wildtype mice by i.v. injection. Fortyeight hours after injecting the cells, we analyzed the number of transplanted cells in the cortex and the plexus choroideus (4 th ventricle; Figures 4E-H) and found that knocking down Plcg1 in MCAM+ Th17-cells significantly decreased the number of recruited cells in both analyzed compartments ( Figures 4G, H) as illustrated in representative images [ Figure 4E (choroid plexus), F (cortex)]. Importantly, the cell numbers in CNS tissue in our experiments were comparable to previous studies (23). These findings show that Plcg1 signaling in MCAM+ Tcells plays an important role for the recruitment of CD4+ T-cells into the brain.

DISCUSSION
MS is a devastating autoinflammatory disease of the CNS, characterized by focal highly inflammatory plaques in the CNS. The recruitment of peripheral Th17 cells has been shown to contribute markedly to initiation of local inflammation and tissue destruction (2)(3)(4)(5)(6). Invasion of Th17 cells into the CNS is highly dependent on the interactions of VLA-4 expressed on T-cells with endothelial VCAM-1. Therefore, VLA-4 blocking (e.g. by Natalizumab) interferes with Th17 cell recruitment to the CNS resulting in significantly decreased local inflammation and progression of the disease (8). However, patients may experience relapses during Natalizumab therapy (10), indicating a VLA-4 independent recruitment mechanism of Th17 cell into the brain. We and others have previously proposed that MCAM, which represents a receptor of so far unknown function on T-cells, substantially contributes to brain inflammation in MS by 411 resulted in a significantly higher T-cell adherence. Our flow chamber assays were performed in a standardized way regarding temperature, applied shear stress, cell dilution as well as quantified area and time of flow rather than representing the physiologic conditions of the cerebral microcirculation. The limited abundance of MCAM+ T-cells in the blood of healthy donors remains a technical issue, however, these data together with our biochemistry data and reporter antibody experiments demonstrate mechanistically that MCAM engagement induces the activation of intracellular signaling pathways that lead to b1-integrin activation. Interestingly, we only observed this for low concentrations of VCAM-1, indicating that the MCAM-mediated increase of cellular adhesion to VCAM-1 is of particular importance if VCAM-1 is expressed at low levels, e.g. in non-inflamed tissue or under immune suppression. MCAM-mediated intracellular signaling via the SRC kinase FYN and the tyrosine kinases PLCg, PYK2 and FAK1 have been described before in endothelial cell lines (32,33), but not in leukocytes. We stimulated MCAM+ and MCAM-memory Tcells with MCAM or laminin-411 under shear conditions and demonstrated that MCAM engagement induces the phosphorylation of SRC, FAK1 and PLCg1. Importantly, induction of SRC, FAK1, and PLCg1 phosphorylation was absent in MCAM-cells or if MCAM+ cells were pre-treated with a blocking anti-MCAM antibody. Interestingly, SRC and FAK1 are known to be involved in integrin outside-in-signaling, whereas PLCg is involved in insight-out as well as outside-in signaling of b1integrins (53)(54)(55)(56). By using inhibitor pre-treated and knock-down cells in flow chamber and reporter antibody experiments, we demonstrated that PLCg1 is involved in MCAM-mediated b1integrin activation. Inhibition of SRC or FAK1 was not sufficient to counteract MCAM-mediated adherence, indicating that other kinases upstream of PLCg1 play an additional role. However, our blocking and knock-down experiments show that PLCg1 activity is crucial for MCAM-mediated integrin b1 activation. Thus, specific inhibition of PLCg1 in patients experiencing relapses during Natalizumab therapy might be a new treatment option to reduce T-cell recruitment into the central nervous system and therefore reduce relapse episodes.
Here, we provide the first functional proof of a mechanistic link between MCAM-mediated intracellular signaling and integrin activation. Notably, besides pairing with the integrin subunit a4 (forming integrin a4b1/VLA-4) integrin subunit b1 also forms heterodimers with a1, -2, -3, -4, -5, -6, -7, -8, -9 -10, -11, and -v which then bind to different ECM molecules and signal transmitting factors such as different species of collagen, fibronectin, vitronectin, laminin, as well as thrombospondin and osteopontin, and membrane receptors such as MadCAM-1 and VCAM-1 [reviewed in (57)]. We show that MCAM+ T-cells also express integrin a5 and a6, thus, broad activation of b1-integrin by MCAM mediated PLCg1 signaling might be a mechanism of increased cell adherence to the ECM in various tissues. However, we did not further investigate this. As MCAM and b1-integrins are also involved in cancer metastasis (11,58,59), our data might provide a mechanistic link between these molecules and disease progression also in that context. Our adoptive transfer experiments implementing specific knock down of plcg1 in MCAM expressing myelin oligodendrocyte glycoprotein-specific T-cells (2D2 donor mice) further underline the relevance of MCAM-mediated signaling via plcg1 in the context of T-cell recruitment into the brain in vivo, however, whether additional MCAM dependent signaling pathways are involved in VLA-4 independent T-cell recruitment to the CNS and which impact MCAM mediated T-cell recruitment via plcg1 activation has in the context of disease (e.g. EAE) remains to be determined. Notably, our data are line with previous studies describing impaired CNS infiltration and delay of disease onset/severity in different mouse models of MS with endothelial MCAM knock out or anti-MCAM antibody treated mice (23,60). We have previously shown that MCAM blockade impairs the recruitment of MCAM+ T-cells to the CNS via the choroid plexus and delays the disease onset in EAE induced mice lacking integrin a4 expression in CD4+ T-cells and ameliorates the disease beyond the blocking of VLA-4 in a spontaneous model of MS (2D2 mice) (23).
Ours and previously published studies (23, 31, 41, 60) thus provide reliable evidence for the relevance of MCAM dependent CNS invasion of T-cells, indicating MCAM blockade to be a suitable treatment approach for MS patients. Clinical trials implementing anti-MCAM antibodies in psoriasis patients have reported limited success, however, leukocyte recruitment to the brain in MS patients might still be efficiently constrained by anti-MCAM antibody treatment.

CONCLUSION
Our study using primary human T-cells in in vitro assays and a murine in vivo model of T-cell recruitment to the CNS provides first evidence for MCAM-mediated ß1-integrin activation via PLCg1 engagement and subsequent amplification of CNS invasiveness.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
All experiments including human material were reviewed and approved by the local ethics committee (Ethik-Kommission der Ärztekammer Westfalen-Lippe und der Medizinischen Fakultät der Westfälischen-Wilhelms-Universität) and performed according to the Declaration of Helsinki. The patients/participants provided their written informed consent to participate in this study. All experiments involving mice were reviewed and approved by the responsible animal protection authority (Landesamt für Natur-Umwelt-und Verbraucherschutz Nordrhein-Westfalen) and conducted according to the German Animal Protection Law.

AUTHOR CONTRIBUTIONS
LZ, SH, PK, and KK performed the experiments. LZ analyzed the data and wrote the manuscript. AZ revised the manuscript. TS-H, HW, NS, and AZ conceived and designed the study. All authors contributed to the article and approved the submitted version.

FUNDING
This study was supported by the German Research Foundation (DFG) Grant CRC128 Project B1 (NS, AZ).