*2.5. Potential Immunomodulation by Cultured Human LNSC*

Next we analyzed the expression of genes potentially involved in LNSC-mediated T cell modulation, thereby we used IFNγ, which is produced by T cells upon their activation and differentiation, as a stimulus [41]. Genes analyzed in cultured LNSCs of passages 4–9 after 4, 24, and 48 h stimulation included interferon gamma receptor 1 (IFNGR1), co-stimulatory molecules cluster of differentiation 40 (CD40), CD80 and CD86 [27,42], immunosuppressive cytokines interleukin 10 (IL-10) and transforming growth factor beta 1 (TGFB1), which are both involved in Treg cell induction [43,44], and the negative T cell regulators CD274 (programmed cell death 1 ligand (PD-L1)), nitric oxide synthase 2 (NOS2) and indoleamine 2,3-dioxygenase 1 (IDO1) [25,26] (Figure 5A, healthy individuals *n* = 5, RA-risk individuals *n* = 5, and RA patients *n* = 5, passages 4–9). Under homeostatic conditions no differences between donor groups were observed. IFNGR1 was stably expressed, at similar levels in all donor groups and unaffected by stimulation (Figure S3), showing that all donor groups are equally equipped to respond to IFNγ. TGFB1 was strongly expressed under homeostatic conditions and not strongly affected by IFNγ (Figure 5A). In contrast, IL-10 was not detected by qPCR. For the co-stimulatory molecules, we found that CD40 was strongly expressed and slightly induced at 24 h, while expression of CD80 was very low and CD86 was undetectable (Figure 5A). On the other hand, the negative regulators CD274 (PD-L1), NOS2, and IDO1 strongly responded to IFNγ, with CD274 (PD-L1) and NOS2 expression peaking at 4 h and IDO1 at 24 h (Figure 5A). Furthermore, at protein level we could detect CD40 by flow cytometry on cultured human LNSCs and expression was upregulated after 72 h stimulation with IFNγ (Figure 5B) though no significant differences between donor groups were observed (passages 4–9) (Figure 5C). Finally, none of these findings reported here on mRNA or protein level correlated with any clinical parameter such as age, gender, or autoantibody titers. Overall, these data show that human LNSCs have the capacity to modulate the adaptive immunity.

**Figure 5.** Expression of immunomodulatory molecules in cultured human LNSCs. (**A**) Induction of transforming growth factor beta 1 (TGFB1) CD40, CD80, CD274 (PD-L1; programmed cell death 1 ligand), nitric oxide synthase 2 (NOS2), and indoleamine 2,3-dioxygenase 1 (IDO1) was assessed by qPCR in cultured LNSCs (passages 4–9) after stimulation with IFNγ at different time points (4, 24, and 48 h). Data are represented as fold induction (median with interquartile range) by comparing the mRNA levels in stimulated cells to corresponding unstimulated cells in 15 donors (*n* = 5 per donor group). The dotted line represents a fold induction of 1. (**B**) CD40 protein expression was measured by flow cytometry in CD45− stromal cells. Histogram depicts the increase in staining between isotype, unstimulated and stimulated for 72 h with IFNγ and is displayed for one representative donor (RA-risk individuals passage 8) out of 13 donors tested. (**C**) Induction of CD40 protein upon stimulation with IFNγ for 72 h is further presented as Mean Fluorescent Intensity (MFI) ratio (isotype/staining) in all donors measured (healthy individuals *n* = 5, RA-risk individuals *n* = 5, and RA patients *n* = 3, passages 4–9).

#### *2.6. Ex Vivo Human LNSCs Express HLA-DR and Low Level of Co-Stimulatory and Co-Inhibitory Molecules*

Finally, we acquired larger pieces of human LNs from kidney transplant recipients which enabled the directly ex vivo analysis of HLA-DR, co-stimulatory, and co-inhibitory molecules on human LNSCs in comparison to DCs (live CD45+CD11c+) as classical antigen presenting cells. In line with murine LNs [31,45], the human LN stromal compartment consists of four distinct subsets when gated on live CD45− cells and additional staining for CD31 and PDPN, thereby considering the PDPN+ population as one population containing FRCs, MRCs, as well as FDCs (Figure 6A,B). However, we observed more BECs and DNs in human LNs compared to mice [31,45]. HLA-DR was expressed on all subsets of human LNSCs but at different levels and considerably lower than DCs (Figure 6C,D). The MFI of HLA-DR in FRCs, LECs, and BECs was higher than DNs which is comparable to MHC-II expression described for murine LNSCs (Figure 6C,D) [31]. Similar to cultured human LNSCs (Figure 5A), co-stimulatory ligand CD80 was detected on freshly digested LNs at protein level in which the expression was relatively higher on LECs compared to other subsets. Furthermore, we could detect low levels of CD86 surface expression on different subsets of LNSCs while we were unable to detect the CD86 at mRNA level in cultured stromal cells (Figure 6C,D). CD40 expression was present on all the subsets and at lower levels than DCs (Figure 6C,D). The expression level of PD-L1+ on endothelial cells (LECs and BECs) was equivalent to DCs but lower on FRCs and DNs. These data noticeably show that ex vivo LNSCs possess HLA-DR, co-stimulatory and co-inhibitory molecules.

**Figure 6.** Expression of HLA-DR and co-stimulatory molecules on freshly isolated human LNSCs. (**A**) Human LNs obtained from kidney transplantation recipients were enzymatically digested, and stained for LNSC subsets based on the expression of CD31 and Podoplanin (PDPN) when gated on live CD45− cells. (**B**) Scatter plot represents the frequencies of different subsets of human LNSCs (hLNSCs) directly after digestion. (**C**) Stromal cells were gated as live CD45− cells and expression of co-stimulatory and inhibitory molecules on different subclasses of LNSCs and CD45+CD11c+ dendritic cells (DCs) were assessed using flow cytometry. Histograms show the expression of indicated molecules on different stromal cell subsets and DCs. The fluorescence minus one (FMO) was used as negative control. (**D**) Scatter plots represent mean fluorescence intensity of HLA-DR, CD80, CD86, CD40, and PD-L1 of each subset of LNSCs when gated on live CD45− cells in comparison to DCs. Data are represented as mean with SEM (*n* = 3–10 kidney recipients).

#### **3. Discussion**

Herein we demonstrate for the first time that PADI2 and PADI4 enzymes as well as citrullinated proteins targeted by ACPAs are present in human LN tissue as well as in cultured human LNSCs during health and different phases of RA. Furthermore, we show that human LNSCs express certain PTAs as well as transcription factors AIRE and DEAF1 together with molecules involved in antigen presentation and immunomodulation. Of interest, our data points towards an altered LN microenvironment in RA patients compared to healthy individuals as seen by variable expression levels of some PTAs.

Citrullination occurs during life under homeostatic conditions but increases in many tissues during inflammation [46]. With PADIs and citrullinated proteins observed in healthy individuals, ACPA-negative and ACPA-positive LN tissue, and cultured LNSCs, our data clearly reveals that citrullination is on-going in human LNs and that this process occurs in both healthy individuals and RA patients' LNSCs. However, a deeper conclusion cannot be drawn since the antibodies used are reactive to several citrullinated targets [47] and we only investigated a small number of donors in this explorative study.

In the context of peripheral tolerance by LNSCs we confirm expression of DEAF1 and AIRE, with AIRE protein being localized primarily nuclear [11] with some cytosolic AIRE as reported in other mammalian cells [48]. Furthermore, we observed a differential expression pattern of disease-related PTAs in LNSCs of RA-risk individuals and RA patients. Of interest, these differences were especially pronounced in LNSCs derived from ACPA negative individuals, which form potentially a genetically distinct patient group [49]. Expression of RRAD was found to be significantly lower in LNSCs of RA patients compared to healthy individuals. At first glance there is no link to autoimmunity since RRAD overexpression is associated with type II diabetes [37,50], which is caused by acquired insulin resistance. However, RA is an important risk factor [51] for this type of diabetes, possibly due to RA-driven inflammation [52]. Since it has been shown that murine LNSCs can control the formation of autoreactive T cells by presenting specific PTA in the context of MHC molecules [27,32,53], it is possible to speculate that LNSCs obtained from RA patients are not capable of suppressing RRAD-specific T cells due to lower expression of this PTA. Although the limitation of our study is the low number of PTAs we analyzed. Investigating the expression of various PTAs by LNSCs from RA patients will be of interest in future studies to confirm this observation. Moreover, further experiments showing antigen presentation and tolerance induction by human LNSCs are necessary, though highly challenging to conduct. So far, we lack the knowledge on well-defined PTA presented by human LNSCs and the availability of corresponding autoreactive human T cells. However, recently a multi-tetramer assay has been developed to detect citrulline-specific T cells in both healthy individuals and RA patients [54]. These tetramers might allow us to study tolerance capacity of LNSCs in controlling citrulline-specific T cells using our in vitro model containing expanded human LNSCs.

Our data reveal that both ex vivo and in vitro, human LNSCs might have the potential to present PTAs directly as they strongly express HLA-DR especially after stimulation with IFNγ. Of interest, our study shows that human FRCs, and also DNs have equal capacity to induce HLA-DR, pointing towards potentially similar roles of these two subsets in humans [21]. However, studies in mice diverge on whether LNSCs present directly to CD4+ T cells. Two studies promoted direct presentation [31,55] and shuttling MHC class II-peptide complexes from DCs [29], while another group showed that LNSCs shuttle antigens to DCs and do not induce tolerance themselves [30]. Similarly, in humans, LECs were able to process antigens, but failed to induce allogeneic CD4+ T cell proliferation [26]. Since LNSCs needed several weeks to grow to confluence and did not contain CD45 positive hematopoietic cells such as DCs, we can conclude that human LNSCs express MHC class II themselves. Accordingly, our ex vivo analyses reveal that human LNSCs express low level HLA-DR in comparison to DCs. Furthermore, the antibody L243 used to stain MHC class II molecules detects a conformational epitope on HLA-DRαβ which depends on the peptide-loading and consequent correct folding of the αβ heterodimer [56], indicating the presence of functional MHC class II molecules containing peptides on the LNSC cell membrane. Nonetheless, this culture technique for human LNSCs provides a robust

basis for applying modern techniques in future research to identify MHC class loaded peptides on LNSCs by mass spectrometry [57] or reverse immunology using bioinformatics [58].

Next to possible antigen presentation by LNSCs, we demonstrated in this study that human LNSCs possess an arsenal of immunomodulatory molecules both in vitro on cultured human LNSCs and ex vivo on freshly isolated human LNSCs. As previously reported, PD-L1 expression is mostly restricted to endothelial cells (LECs and BECs) on freshly isolated human LN [59]. Similarly, our data also showed that PD-L1 protein expression was higher on endothelial cells in comparison to FRCs and DNs. Interestingly, at mRNA level we could detect low level of PD-L1 and HLA-DR in cultured stromal cells and the expression was lower in RA-risk individuals and RA patients compared with healthy individuals, but future studies are needed to confirm this. Moreover, our data reveal very low levels of co-stimulatory molecules CD80, CD86, and CD40 expression on freshly isolated human LNSCs. This is interesting since tolerogenic DCs possess a low level of costimulatory molecules and therefore can control autoreactive T cells. From these observations we can conclude that human LNSCs have the machinery to interact and influence lymphocytes and therefore to potentially regulate tolerance and adaptive immunity.

Overall, our explorative study shows for the first time citrullination in human LNSCs targeted by autoantibodies isolated from RA patients. Further challenging mechanistic studies are required to investigate whether LNSCs from RA (-risk) patients have an altered tolerogenic effect on autoreactive T cells. To study LNSCs immunoregulatory function, in vitro expansion of LNSCs is required, which is a limitation of this study. The difficulty of obtaining LN biopsies from a large number of individuals and the slow growth rate of human LNSCs limits the number of donors analyzed in this study. In addition, the variation between donors is high. However, by revealing that human LNSCs exhibit the tools to induce tolerance as observed in mice, they become an attractive new therapeutic target to exploit in tolerance maintenance and induction.
