RGMb impacts partial epithelial‐mesenchymal transition and BMP2‐Induced ID mRNA expression independent of PD‐L2 in nonsmall cell lung cancer cells

PD‐1/PD‐ligand‐axis immunotherapy‐mediated activation of T‐cells for cancer cell elimination is a promising treatment of nonsmall cell lung cancer (NSCLC). However, the effect of immunotherapy on intracellular signaling pathways in cancer cells still needs further delineation. Repulsive Guidance Molecule b (RGMb), a regulator of Bone Morphogenetic Proteins (BMPs) signaling, interacts with the PD‐ligand, PD‐L2, at cancer cell membranes. Accordingly, a clarification of the functions of RGMb and its relation to PD‐L2 might provide insight into NSCLC cell signaling responses to PD‐1/PD‐ligand‐axis immunotherapy. In this study, the functions of RGMb and PD‐L2 were examined using the two NSCLC cell lines HCC827 and A549. CRISPR/Cas9 was used to decrease the expression of RGMb and PD‐L2, while lentiviral vectors were used to increase their expression. Downstream effects were examined by RT‐qPCR and immunoassays. Ectopic expression of RGMb impacted BMP2‐induced expression of ID1 and ID2 messenger RNA (mRNA) independently of PD‐L2, while RGMb depletion by CRISPR/Cas9 did not affect the BMP2‐mediated induction of ID1, ID2, and ID3 mRNA. However, depletion of RGMb resulted in a partial epithelial‐mesenchymal transition (EMT) gene expression profile in HCC827 cells, which was not mimicked by PD‐L2 depletion. The results show that RGMb is a coregulator of BMP signaling and hence, ID mRNA expression and that RGMb can control the EMT balance in NSCLC cells. However, RGMb appears to exert these functions independently of PD‐L2, and accordingly, the PD‐1/PD‐ligand axis for immune surveillance in NSCLC cells.

Lung cancer is the second most common cancer type in the world and the leading cause of cancer death worldwide, responsible for 18% of all cancer deaths (Sung et al., 2021).Lung cancer is divided into two histological subtypes; small cell lung cancer and nonsmall cell lung cancer (NSCLC), which represent more than 80% of all lung cancer cases (Herbst et al., 2018).
Bone Morphogenetic Proteins (BMPs) constitute a subfamily of the Transforming Growth Factor β super-family of secreted signaling molecules (Corradini et al., 2009;Samad et al., 2005).BMPs are involved in a diverse array of cellular and systemic processes including cellular proliferation, differentiation, and migration (Corradini et al., 2009).
BMP2 and BMP4 form a functional subclass of BMPs hereafter abbreviated BMP2/4 (Goldman et al., 2009).BMP2/4 expression is upregulated in NSCLC and positively correlates with tumor stage and metastatic burden as well as shorter overall survival of NSCLC patients (Ju et al., 2019).In accordance, BMP2/4 has been shown to promote the proliferation, migration, and invasion of NSCLC cells (Chu et al., 2014;Fei et al., 2013;Huang et al., 2020).BMP2/4 binds preferentially to the high-affinity type I receptors BMPR1A (also known as ALK3) and BMPR1B (also known as ALK6), but can also signal through ACVRI (also known as ALK2) (Goldman et al., 2009).The subsequent association with low-affinity type II receptors BMPR2, ACVR2A, and ACVR2B results in a phosphorylation cascade involving the SMAD1/5/8 pathway or the p38 mitogen-activated protein kinase (MAPK) and extracellular signal-regulated protein kinase (ERK) pathway, leading to transcriptional activation of response genes (Corradini et al., 2009;Xia et al., 2011).
Well-characterized response genes are the Inhibitors of Differentiation (ID) genes ID1, ID2, and ID3, which encode a subfamily of the helix-loophelix transcription factors (Ling et al., 2014).
Repulsive Guidance Molecule b (RGMb) (also named DRAGON) can, in a context-dependent manner, either support or suppress the BMP2/4-signaling cascade (Healey et al., 2015;Kanomata et al., 2009;Siebold et al., 2017;Xia et al., 2010).RGMb is part of the RGM family of glycosylphosphatidylinositol (GPI)-anchored BMP coreceptors, which also include RGMa and RGMc (Corradini et al., 2009).RGMb is expressed in the nervous system and the immune system, and by epithelial cells including many types of cancer cells (Park et al., 2023;Siebold et al., 2017;Xia et al., 2011;Xiao et al., 2014).RGMb can be secreted to the intercellular space, attached to the cell surface through a C-terminal GPI anchor, or maintained intracellularly (Li, Ye, Kynaston, et al., 2012;Xiao et al., 2014).RGMb interacts with BMP2/4, BMP receptors, and the transmembrane protein Neogenin (NEO1) which is also involved in BMP-signaling (Samad et al., 2005).In addition, RGMb interacts with the cell membrane-localized immune checkpoint molecules CTLA4 and PD-L2, but not with PD-L1 (Sekiya & Takaki, 2019;Xiao et al., 2014).Both PD-L1 and PD-L2 participate in PD-1/PD-ligandaxis immune surveillance, and therapeutic antibodies targeting the PD-1 receptor on T-cells and PD-L1 on cancer cells show promising treatment results for cancer patients including patients with NSCLC (Sadeghirad et al., 2023).Downregulation of RGMb and PD-L2 expression was found to promote antitumor immunity in a mouse tumor model while blocking of the RGMb-PD-L2 interaction was found to overcome resistance to PD-1 and PD-L1 antibody-based cancer immunotherapy (Park et al., 2023).In addition, the interaction between RGMb and PD-L2 was found to promote respiratory tolerance in the lung (Xiao et al., 2014).RGMb expression is downregulated in advanced-stage NSCLC as well as in tumors with vascular invasion, and low RGMb tumor expression correlates with shorter overall survival of these patients (Li et al., 2016).Furthermore, RGMb inhibits adhesion, invasion, and metastasis of breast cancer, prostate cancer, and NSCLC cells, contrary to the functions of BMP2/4 (Li, Ye, Kynaston, et al., 2012;Li, Ye, Sanders, et al., 2012;Li et al., 2016).
Epithelial-mesenchymal transition (EMT) refers to the dedifferentiation and depolarization of epithelial cells to mesenchymal cells and plays a crucial role in the progression of NSCLC by increasing cancer cell stem-cell characteristics, motility, and therapeutic resistance (Legras et al., 2017;Nurwidya et al., 2012;Xiao & He, 2010).In EMT, the expression of numerous genes is up-or downregulated, for example, Vimentin (VIM), which is upregulated, and E-cadherin (CDH1), which is downregulated.EMT is promoted by the core transcription factors ZEB1, ZEB2, SNAI1, SNAI2, TWIST1, and TWIST2 (Singh et al., 2018) and regulated by several other transcription factors including the ID proteins, which have pleiotropic effects as either EMT-promoters or suppressors (Ling et al., 2014;Singh et al., 2018).In NSCLC cells, increased ID expression can promote EMT and is associated with dedifferentiated, proliferating, and invasive cell phenotypes as well as advanced tumor stage and poor prognosis (Ling et al., 2014).BMP2/4 signaling can promote EMT, demonstrated by BMP2/4-mediated upregulation of Vimentin, IDs, SNAI1, and SNAI2 and downregulation of E-cadherin (Bao et al., 2017;Fukuda et al., 2020;Mihajlovic et al., 2019;Park et al., 2015;Qiao et al., 2011).In line with this, RGMb also affects EMT, as decreased RGMb expression was found to result in upregulation of SNAI1 and TWIST1 in breast cancer cells (Li, Ye, Sanders, et al., 2012), while it led to downregulation of E-cadherin and upregulation of Vimentin in a squamous cell carcinoma of the head and neck cells (Zhang et al., 2020).
In this study, we examined the impact of RGMb on BMP2/4 signaling and EMT in NSCLC cells by utilizing CRISPR/Cas9-mediated RGMb depletion and lentivirus-based RGMb overexpression.In the same way, we examined the impact of PD-L2 on the function of RGMb.We found that RGMb inhibited the development of a partial EMT phenotype, and affected the BMP2-mediated induction of ID1 and ID2 messenger RNA (mRNA) expression and that these functions are not mimicked by PD-L2.
HEK293T cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% FCS and 1% penicillinstreptomycin.All cells were cultivated at 37°C and 5% CO 2 .Cell culturing was initiated using cells with the lowest possible growth passages and cell integrity was confirmed with fingerprinting of the given mutational status for HCC827 and A549 cells.For treatment with recombinant BMP2 (rBMP2), cells were cultivated in the presence of 100 ng/μL rBMP2 (R&D systems, 355-BM) dissolved in 4 mM HCl with 0.1% bovine serum albumin (BSA) (Sigma-Aldrich, A2153) for 1 or 24 h. 4 mM HCl with 0.1% BSA was used as a control.
The RT-qPCR data were analyzed using the X 0 -method (Thomsen et al., 2010).ACTB was used as a reference gene and primer efficiencies were determined in technical duplicates as previously described (Thomsen et al., 2010).Primers are listed in Supporting Information: Table S1.For retrospective mRNA expression analyses in HCC827 and A549 NSCLC cells, we used the depmap portal https://depmap.org.mRNA expression data for HCC827 cells possessing EMT were previously described (Vad-Nielsen et al., 2023).
These mRNA expression data are presented in Supporting Information: Table S2.

| Lentivirus-based cDNA expression
For RGMB cDNA expression, pLV-RGMB-GFPSpark was used (SinoBiological, HG17194-ACGLN), and for PD-L2 expression, PDCD1LG2 cDNA, encoding human PD-L2, was cloned into the pCCL-CMV-intron+PGK-GFP vector.The empty pCCL-CMV-intron+PGK-GFP vector was used for control.Lentiviral production was performed in HEK239T cells, as previously described (Dietz et al., 2022).Virus titers were calculated based on the number of GFP-positive HCC827 cells after transduction.A multiplicity of infection of five was used for the generation of PD-L2 and RGMb overexpressing HCC827 cells with no subsequent selection.
Flow cytometry analysis was performed within 24 h on a NovoCyte Quanteon 4025 (Agilent) and data were analyzed using FlowJo (BD Biosciences).The gating strategy is shown in Supporting Information:
Imaging was performed on a Leica/LEITZ DMRB fluorescence microscope with a monochrome digital camera (Leica).

| Statistics
Expression data are shown with means and standard deviations (SDs).
Fold changes (FC) in gene expression were considered relevant if the data fulfilled the following criteria: FC > 2 or FC < 0.5 and p < .05using an unpaired, two-tailed, two-sample t-test of technical triplicates.1b and Supporting Information:

3
Table S2) (Li, Ye, Kynaston, et al., 2012;Xiao et al., 2014).The addition of the C-terminal tag on the ectopic RGMb potentially decreases the rate of GPI processing (Caras, 1991;Guizzunti & Zurzolo, 2014), but we note that the flow-cytometry analyses revealed the presence of membrane-bound RGMb (Figure 1b).A tendency toward increased mRNA expression levels of ID1 and ID2, but not ID3, in the presence of ectopic RGMb, was observed, while ectopic PD-L2 did not result in a similar effect (Supporting Information: Figure S1D).To further study the effects of ectopic induction of ID1 and ID2 mRNA was lower in the HCC827eRGMb cells compared to the HCC827 empty vector control cells, and the effect was in particular evident for ID2 (Figure 1c and Supporting Information: Figure S1E).This is indicating that RGMb can influence BMP2-mediated induction of ID1 and ID2 mRNA expression in NSCLC cells.No effect was observed for ectopic RGMb on ID3 mRNA induction, indicating a difference in the regulation of ID mRNA expression in response to BMP2-stimulation (Figure 1c and Supporting Information: Figure S1E).A similar effect on ID1, ID2, and ID3 mRNA induction was not observed following ectopic PD-L2 expression (Figure 1c and Supporting Information: Figure S1E).These results indicate that RGMb can impact active BMP2/4 signaling in NSCLC cells and, importantly, that this function is not mimicked by PD-L2.

| CRISPR/Cas9-mediated RGMb depletion in NSCLC cells
To further study the function of RGMb in HCC827 cells, we used CRISPR/Cas9 to create RGMb-depleted HCC827 cells (HCC827d RGMB cells).For this, HCC827Cas9 cells were transduced with shown).The indels did not result in severe nonsense-mediated mRNA decay, as the RGMB mRNA expression levels did not differ significantly between the HCC827 sgRNA control and the HCC827dRGMB cells (Figure 2b).This could be due to the proximity of the indels to the start codon, as mRNA with premature termination codons proximal to the start codon can be resistant to nonsensemediated mRNA decay (Pereira et al., 2015;Popp & Maquat, 2016).
Western blot analyses did not reveal a reduction in the RGMb level in HCC827dRGMB cells concordant in quantity with the genetic data (Supporting Information: Figure S3A).We next performed immunofluorescence assays.This showed that decreased RGMb expression was observed in HCC827dRGMB cells compared to the HCC827 sgRNA control cells, as a less intense RGMb staining was observed for the HCC827dRGMB cells in analyses of multiple cell slide sections (Figures 2c,d).This is in line with the high KO percentage from the genetic data (Figure 2a).

| BMP2 induces ID mRNA expression independently of endogenous RGMb
To examine the effect of RGMb on basal BMP2/4 signaling in NSCLC cells, we assessed the mRNA expression of BMP2/4 signaling response genes ID1, ID2, and ID3 in HCC827dRGMB cells relative to HCC827 sgRNA control cells.Depletion of endogenous RGMb did not impact BMP2/4-mediated ID expression under the given experimental conditions, as we found the expression of ID mRNA to be similar between HCC827dRGMB cells and HCC827 control cells (Figure 2e).Acknowledging that basal BMP2/4 expression is low (Supporting Information: Table S2), we addressed the mRNA expression of ID1, ID2, and ID3 in HCC827 sgRNA control and HCC827dRGMB cells after stimulation with rBMP2.ID1, ID2, and ID3 mRNA expression levels were elevated by rBMP2 stimulation (1 h and 24 h) in both HCC827 sgRNA control and HCC827dRGMB cells (Figure 2f and Supporting Information: Figure S3B).Altogether, this indicates that the endogenous RGMb in HCC827 cells does not confer regulation of BMP2-mediated ID mRNA induction.
Next, we questioned if the lack of RGMb-mediated regulation of ID mRNA expression was specific to HCC827 cells.To address this, we performed CRISPR/Cas9-mediated RGMb depletion in the A549 NSCLC cell line to generate A549dRGMB cells (Supporting Information: Figures S4A-D).As for the HCC827 cells, no significant changes in mRNA levels of ID1, ID2, and ID3 were observed in A549dRGMb cells compared to A549 sgRNA control both without rBMP2 stimulation and after 1 and 24 h of rBMP2 stimulation (Supporting Information: Figure S4E and S4F).Collectively, the results show that stimulation with rBMP2 can induce expression of ID1, ID2, and ID3 mRNA in NSCLC cell lines and that the endogenous RGMb does not regulate the BMP2/4-mediated ID mRNA induction.

| CRISPR/Cas9-mediated RGMb depletion in HCC827 cells induces a partial EMT mRNA expression profile
An experimentally-mediated decrease in RGMb expression promotes EMT characteristics in cell lines of breast cancer and squamous cell carcinoma of the head and neck (Li, Ye, Sanders, et al., 2012;Zhang et al., 2020).Therefore, we examined if the CRISPR/Cas9-mediated depletion of RGMb in HCC827dRGMB cells promoted a shift in gene expression reflecting EMT characteristics.HCC827 cells normally possess an epithelial morphology (Karacosta et al., 2019;Li et al., 2011).We found that mRNA expression was upregulated for the mesenchymal markers VIM and ZEB1 in the HCC827dRGMB cells compared to the HCC827 sgRNA control cells (Supporting Information: Figure S5A).To verify this result we used another RGMB sgRNA, sgRNA-6, to generate CRISPR/Cas9-mediated RGMb depletion (Supporting Information: Figure S5B).This resulted in similar results, showing upregulation of VIM mRNA expression, as well as a tendency towards upregulation of ZEB1 mRNA expression (Supporting Information: Figure S5C).To further explore the development of the EMT characteristics, we made an independent biological replicate of RGMb depletion using sgRNA-3.Again, VIM and ZEB1 mRNA upregulation was observed in the HCC827dRGMB cells, as well as upregulation of ZEB2 and TWIST1 mRNA levels, whereas mRNA expression levels were unchanged for SNAI1, SNAI2, and TWIST2 (Figure 3a).Furthermore, the mRNA expression of the three epithelial markers CDH1, GRHL2, and ESRP1 was downregulated (Figure 3a).In addition, the cell shape morphology of the HCC2827dRGMB cells was more elongated than the HCC827 sgRNA control cells and presented with less condensed cell-cell contacts (Figure 2c).
As A549 cells have a more mesenchymal gene expression phenotype (Supporting Information: Table S2), we expected these cells to be less prone to a shift in EMT marker expression mediated by RGMb depletion.As expected, no changes in mRNA expression of EMT markers were observed between A549dRGMB and A549 sgRNA control cells (Supporting Information: Figure S5D).
We observed a tendency towards decreased mRNA expression of PDCD1LG2 and CD274 (encoding PDL1) in HCC827dRGMB cells compared to HCC827 sgRNA control cells (Figure 3b).The decreased mRNA expression of PDCD1LG2 could reflect the presence of partial EMT, as mRNA expression of the CD274 and PDCD1LG2 genes are highly correlated and is downregulated as a result of EMT in NSCLC cells (Larsen et al., 2019;Lou et al., 2016;Tieche et al., 2019).In addition, RGMb depletion did not have any influence on PDCD1LG2 mRNA expression in A549 cells (Supporting Information: Figure S5E).
IFNγ is a well-described inducer of PDCD1LG2 and CD274 mRNA expression in NSCLC cell lines including HCC827 cells (Gao et al., 2018).However, we observed a poorer IFNγ-mediated mRNA induction in the HCC827dRGMB cells (Figure 3c).This is in line with the described inhibitory effect of EMT on IFNγ signaling in lung cancer cells (Tseng et al., 2022).For two other IFNγ-response genes, BMP2/4 stimulation was previously shown to upregulate core EMT transcription factors SNAI1 and SNAI2 in NSCLC cells (Bao et al., 2017;Mihajlović et al., 2019).Therefore, we assessed the mRNA expression of various EMT-markers in HCC827 sgRNA control and HCC827dRGMB cells after stimulation with rBMP2.After 24 h of rBMP2 stimulation SNAI1 mRNA expression was upregulated 2.5-fold in HCC827 sgRNA control cells and the same tendency was observed for the HCC827dRGMB cells albeit to a lesser extent (Figure 3d).After 1 h of rBMP2 stimulation, a similar pattern of SNAI1 mRNA induction was observed (Supporting Information: Figure S5F).
We note that decreased SNAI1 and SNAI2 mRNA expression following acquired EMT in HCC827 cells was previously described (Jakobsen et al., 2017).mRNA expression of PDCD1LG2 and CD274 was not altered by rBMP2 stimulation of HCC827 sgRNA control and HCC827dRGMB cells (Figure 3d).

| PD-L2 depletion does not mimic the RGMb depletion-mediated partial EMT phenotype
We next examined whether the inhibitory effect of RGMb on the acquisition of the partial EMT phenotype of HCC827 cells was dependent on the coexpression of PD-L2.We performed CRISPR/ Cas9-mediated PDCD1LG2 depletion to generate HCC827dPD CD1LG2 cells.The HCC827dPDCD1LG2 cells had PDCD1LG2 indel and KO percentages of 97% and 82%, respectively (Figure 4a).

A decreased cell surface expression of PD-L2 in HCC827dP
DCD1LG2 cells was identified by flow cytometry (Figure 4b) and a concordant decrease in PDCD1LG2 mRNA expression was observed (Figure 4c).We analyzed the mRNA expression of a selected set of EMT marker genes, as well as the BMP2/4 response genes ID1, ID2, and ID3, and no significant changes were observed for HCC827dPD CD1LG2 cells compared to HCC827 sgRNA control cells (Figure 4c).
Lastly, the expression of RGMB was unchanged in the HCC827dPDC D1LG2 cells (Figure 4c).Thus, no indication of a direct auto-regulatory loop between PD-L2 and RGMb expression was found.In conclusion, these results show that PD-L2 depletion in HCC827 cells does not mimic the partial EMT mRNA expression profile driven by RGMb depletion.

| DISCUSSION
In NSCLC, BMP signaling promotes proliferation, migration, and invasion of cancer cells and is associated with tumor stage, metastatic burden, and shorter overall survival (Chu et al., 2014;Fei et al., 2013;Huang et al., 2020, Ju et al., 2019, Langenfeld et al., 2005).There is little to no activity of the BMP signaling cascade in the epithelial cells of normal lung tissue with reactivation preferentially occurring in inflamed, damaged, or transformed bronchial epithelial cells (Langenfeld et al., 2013).This suggests that there is a therapeutic window to target BMP signaling pathways in NSCLC (Langenfeld et al., 2013).In BMP signaling, the number of BMP ligands exceeds the number of receptors.Thus, activation of the specific downstream SMAD and MAPK/ERK pathways requires additional membraneassociated regulatory factors, including the family of RGM proteins.
RGMs bind both BMPs and BMP receptors and thereby facilitate a shift in type II receptor usage from BMPR2 to ACVR2A.In addition, RGMs may, in a ternary complex with BMPs and NEO1, signal through the RHO kinase pathway and trigger cytoskeletal rearrangements (Corradini et al., 2009).RGMb is involved in BMP2/4 signaling, in particular, and has been shown to both promote and inhibit BMP2/ 4 signaling (Tian & Liu, 2013).Soluble RGMb, formed by shedding, can inhibit BMP2/4 signaling by titrating the amounts of BMP2/4 available for BMP receptor interaction at the cell membrane, as well as through competitive binding to the same site of BMP2/4 as type I receptors (Mueller, 2015).In addition, RGMb is also proposed to inhibit BMP2/4 signaling through interaction with a yet undescribed membrane-associated cosuppressor protein (Kanomata et al., 2009;Tian & Liu, 2013).By contrast, membrane-associated RGMb is also shown to be an activator of BMP2/4 signaling through a mechanism proposed to involve the endosome-dependent release of BMP2/4 from RGMb, enabling the generation of BMP2/4, type I receptor, and type II receptor ternary complexes (Mueller, 2015).Thus, the role of RGMb as genes ID1, ID2, and ID3 at basal BMP2/4 concentrations nor after stimulation with a high dose of rBMP2 in the NSCLC cell lines HCC827 and A549.This may be due to a relatively low expression level of RGMB mRNA in the NSCLC cell lines (Supporting Information: Table S2).Successful RGMb-depletion by CRISPR/Cas9 was validated by genetic analyses revealing the presence of a high proportion of reading frame-disrupting indels in the 5'-end of RGMB.However, Western blot analyses of RGMb did not show reduced protein levels.
The major isoform of RGMb is 437 amino acids, is proteolytically processed in the N-and C-terminals has a potential internal protease cleavage site, and is post-translationally modified through two N-linked glycosylation events (Xiao et al., 2014).In line with this, published Western blot analyses show the presence of RBMb bands in sizes ranging from approximately 25 kDa to 55 kDa (Li et al., 2016;Samad et al., 2005;Sekiya & Takaki, 2019;Shi et al., 2015;Xiao et al., 2014), with an approximately 37 kDa band being recognized as internally cleaved RGMb (Xiao et al., 2014).Here, we detected a minor intensity decrease in a band of approximately 37 kDa following RGMb-depletion.We note that in situ detection of RGMb could detect successful CRISPR/Cas9-mediated depletion.Previously, we have observed that depletion of PD-L2 in HCC827 cells, which like RGMb is also expressed at low endogenous levels, was undetectable by Western blot analysis, but detectable by in situbased techniques, whereas depletion of PD-L1, which is expressed at a higher endogenous level, was detectable by both methodologies (data not shown).Depletion of RGMb was functionally substantiated in HCC827 cells by the resulting acquisition of a partial EMT phenotype.We found no direct evidence of a connection between the functions of RGMb related to BMP2/4-mediated ID expression and the acquirement of a partial EMT phenotype.It is clear that BMP2/4 signaling stimulates mRNA expression of ID1, ID2, and ID3, but whether ID upregulation is EMT-promoting or EMT-inhibiting seems to depend on the cellular context (Langenfeld et al., 2013).
Furthermore, we note that partial EMT was acquired in  HCC827dRGMB cells despite the similar mRNA expression levels of ID1, ID2, and ID3 in HCC827dRGMB and HCC827 control cells.In contrast, complete EMT in HCC827 cells is associated with increased mRNA expression levels for ID1, ID2, and ID3 (Supporting Information: Table S2).
We also examined the effect of ectopic expression of RGMb in HCC827 cells and found that the induction of mRNA expression of the ID proteins have at least some nonoverlapping functions (Ling et al., 2014).The latter is illustrated in NSCLC cell lines where ID3 induces apoptosis (Chen et al., 2016;Li, Zhu, Yu, et al., 2012), while ID1 promotes proliferation (Cheng et al., 2011).One limitation of our RGMb expression studies is that the addition of the C-terminal tag to RGMb could interfere with the efficiency of GPI-anchor-mediated cell surface localization.Unanchored RGMb could potentially have an inhibitory effect on BMP2/4 signaling mimicking that of soluble RGMb formed by shedding, which could also explain the observed decrease in rBMP2-mediated ID1 and ID2 mRNA induction.However, flow cytometry analysis validated the presence of at least some RGMb on the cell membrane of the HCC827eRGMB cells, which could not be detected on the HCC827 empty vector control cells.In line with this, it has been shown that C-terminal protein extensions does not block the GPIattachment process, but can induce a decrease in the rate of processing (Caras, 1991;Guizzunti & Zurzolo, 2014).Changing the ratio of GPI-anchored relative to soluble RGMb, potentially in a cell-type dependent context, could be used to determine the functions of RGMb related to BMP2/4-signaling and IDexpression.Although the exact modes of action of RGMb have yet to be defined, our NSCLC cell results support previous studies that RGMb impacts BMP2/4-mediated regulation of ID expression in cancer cells (Kanomata et al., 2009;Mueller, 2015;Tian & Liu, 2013).
The trans-membrane protein PD-L2 is a binding partner for RGMb and the two proteins can be coexpressed in NSCLC cells (Larsen et al., 2019;Xiao et al., 2014).Accordingly, PD-L2 is a potential candidate for being the yet undescribed membraneassociated cosuppressor protein of RGMb for inhibition of BMP2/4 signaling (Kanomata et al., 2009;Tian & Liu, 2013).Targeted-and immuno-therapies for NSCLC patients notoriously encounter problems with intrinsic and acquired resistance, which both in vitro and in vivo can be assigned as a result of EMT (Ashrafizadeh et al., 2020;Jakobsen et al., 2016;Zhou et al., 2020).Moreover, successful PD-1/PD-ligand-axis immunotherapy of NSCLC patients depends on the tumor EMT status, partially due to the lower levels of PD-L1 and PD-L2 expression resulting from EMT (Taki et al., 2021).Here, we characterized the functions of RGMb in NSCLC cells with a focus on the potential functional involvement of PD-L2.The identified

The
RGMB sgRNA-3 5'-GCGGCGCTGCTCAACCTCGG-3' has an MIT specificity score of 93, Doench efficiency score of 94%, and Moreno-Mateos (in vitro) score of 98%.There were no potential off-target sites with less than three mismatches.The RGMB sgRNA-6 used for verification of EMT-results 5'-AGCCTCGGGCTGCTCCACGC-3' has an MIT specificity score of 80, Doench efficiency score of 78%, and Moreno-Mateos (in vitro) score of 79%.The sgRNA-6 had one potential off-target with two mismatches in an intron for the MPP3gene.The PDCD1LG2 (PD-L2) sgRNA 5'-CAAGTCCAAGTGAGG GACGA-3' has an MIT specificity score of 82, Doench efficiency score of 91%, and Moreno-Mateos (in vitro) score of 71%.The sgRNA had two potential off-targets with two mismatches in intronic or intragenic regions.sgRNA oligonucleotides for RGMB sgRNA-3 5'-CACCGGCGGCGCTGCTCAACCTCGG-3' and 5'-AAACCCGAGG TTGAGCAGCGCCGCC-3'; RGMB sgRNA−6 5'-CACCGAGCCTCG GGCTGCTCCACGC-3' and 5'-AAACGCGTGGAGCAGCCCGAGGC TC-3'; and PDCD1LG2 5'-CACCGCAAGTCCAAGTGAGGGACGA-3' and 5'-AAACTCGTCCCTCACTTGGACTTGC-3' were annealed and inserted into plentiGuide-Hygro (Addgene, 139462) for RGMB and plentiGuide-Puro (Addgene, 52963) for PDCD1LG2.An inserted scrambled sgRNA sequence 5'-ACGGAGGCTAAGCGTCGCAA-3' followed by 14 days of hygromycin or puromycin resistance selection.For the determination of CRISPR/Cas9 efficiency, DNA extraction was performed using the MasterPure DNA purification kit (Illumina, MCD85201).PCR was performed in a total amount of 25 μL containing 50 ng DNA, 12.5 μL Taq DNA polymerase 2x Master Mix RED (Ampliqon, A180306), 5 μL betaine (5 M), 1.25 μL DMSO, 1 μL forward primer (10 μM), 1 μL reverse primer (10 μM) and nucleasefree water using the following program: 5 min at 95°C, 35 cycles of PCR (30 s at 95°C, 30 s at 57°C and 30 s at 72°C) and 5 min at 72°C.The following primers were used: RGMB 5'-GCTCACGG TCTTTGTGTCTTC-3' and 5'-CTTCGAGACCTGGGGCTATT-3' or PDCD1LG2 5'-AAAAGGATTCGCCACGGGA-3' and 5'-GATCCTCTGT TGGAGAGATGC-3'.The amplified DNA fragment was purified from the PCR product using the NucleoSpin ® Gel and PCR Clean-up kit (Macherey-Nagel, 740609), Sanger sequencing was performed by Ectopic expression of RGMb impacts BMP2-mediated ID1 and ID2 mRNA induction.(a) mRNA expression of RGMB and PDCD1LG2 mRNA in transduced HCC827 cells assessed by RT-qPCR.mRNA levels are normalized to the mRNA levels of the HCC827 empty vector control cells.The plot is based on one biological sample analyzed in triplicates.(b) Flow cytometry density plots showing the cell surface expression of PD-L2 and RGMb in transduced HCC827 cells.The PD-L2+RGMb plot for stained cells was used to set experimental gates.(c) mRNA expression of ID1, ID2 and ID3 in transduced HCC827 cells after 24 h rBMP2 stimulation assessed by RT-qPCR.mRNA levels are normalized to the mRNA levels of HCl-treated cells of the same genotype.mRNA expression results are based on one biological sample analyzed in triplicates.An independent biological replicate is shown in Supporting Information: FigureS1.In (a) are statistical analyses for individual genes performed relative to control cells and in (c) between HCl and BMP2 treated samples, as well as between rBMP2 treated samples with control and eRGMb lentivirus transduction.Fold change (FC) > 2 or FC < 0.5 and p < .05are marked with #. lentiviral particles carrying a sgRNA, sgRNA-3, targeting the RGMb open reading frame in the first coding exon of mRNA isoform 2 and the second coding exon of mRNA isoform 1(Rotwein, 2019).HCC827Cas9 cells were transduced with lentiviral particles carrying a scrambled sgRNA to generate HCC827 sgRNA control cells.Sanger sequencing revealed an indel percentage of 91% and a percentage of the open reading frame-disrupting indels, hereafter termed knockout (KO) percentage, of 85% (Figure2a).No indication of a potential natural selection for intact RGMb was observed, as the indel and KO percentages did not change after prolonged cell culture (data not and ISG56, the effect of RGMb depletion was less pronounced (Figure3c).Altogether, the results indicate that RGMb possesses an indirect regulatory function on PD-L2 expression mediated by the inhibition of RGMb on the development of partial EMT.
RGMb depletion in HCC827dRGMB cells.(a) RGMB indel percentage and the open reading framedisrupting indel (knockout (KO)) percentage.(b) RGMB mRNA expression in HCC827 sgRNA control and HCC827dRGMB cells.mRNA levels are normalized to the mRNA level of the HCC827 sgRNA control cells (c) RGMb expression levels in HCC827dRGMB and HCC827 sgRNA control cells were assessed by indirect immunofluorescence.Two representative images are shown for each cell type with DAPI staining of cell nuclei (blue) and anti-RGMb antibody staining (green).(d) Staining of HCC827 control sgRNA cells without primary anti-RGMb antibody.DAPI staining (blue) and secondary anti-rabbit antibody staining (green) are shown.(e, f) Analysis of the impact of RGMb depletion on rBMP2mediated ID1, ID2, and ID3 mRNA expression.(e) Basal ID1, ID2, and ID3 mRNA expression in HCC827dRGMB and HCC827 sgRNA control cells.mRNA levels are normalized to the mRNA level of the HCC827 sgRNA control cells.(f) ID1, ID2, and ID3 mRNA expression in HCC827dRGMB and HCC827 sgRNA control cells after 24 h rBMP2 stimulation.mRNA expression levels of the rBMP2-stimulated cells are shown relative to the HCl-treated cells of the same genotype.(b, e, f) mRNA expression assessed by RT-qPCR.The plots are based on one biological sample analyzed in triplicates and results are shown as mean values with error bars representing standard deviations.Statistical analyses were performed for individual genes between control and dRGMb samples, as well as between HCl and BMP2 treated samples in panel (f).Genes with FC > 2 or FC < 0.5 and p < .05are marked with #. activator of BMP2/4 signaling is most likely a reflection of the complex coordination of the stoichiometric amounts of the various components in the signalosome and precise compartmentalization.In alignment with such complex regulation, we observed that the regulatory role of RGMb in NSCLC cells depended on the amount of RGMb, as well as the concentration of BMP2/4.We found that endogenous RGMb was not required for regulation of the mRNA expression levels of the BMP-response F I G U R E 3 (See caption on next page).

F
I G U R E 3 RGMb depletion generates partial EMT mRNA expression changes in HCC827 cells.(a, b) mRNA expression levels of EMT markers (a) and PDCD1LG2 (encoding PD-L2) and CD274 (encoding PD-L1) (b) in HCC827 sgRNA control and HCC827dRGMB cells.mRNA levels are normalized to the mRNA level present in the HCC827 sgRNA control cells.(c) IFNγ-mediated induction of PDCD1LG2, CD274, RGMB, and IFNγ response genes IRF1 and ISG56 mRNA expression after 48 h of IFNγ stimulation.mRNA expression levels of the IFNγ-stimulated cells are shown relative to the control-treated cells of the same genotype.(d) mRNA expression of EMT markers in HCC827dRGMB and HCC827 sgRNA control cells after 24 h rBMP2 stimulation.mRNA expression levels of the rBMP2-stimulated cells are shown relative to the HCl-treated cells of the same genotype.mRNA expression results from RT-qPCR are shown as mean values with error bars representing standard deviations.The plots are based on one biological sample analyzed in triplicates.In (a) and (b) are statistical analyses performed for individual genes between control and dRGMb samples, in (c) between samples without and with IFNγ stimulation, and in (d) between control and dRGMb samples, as well as between the HCl and BMP2 treated samples.Genes with FC > 2 or FC < 0.5 and p < .05are marked with #.

ID1
and ID2, but not ID3, was reduced upon stimulation with high doses of rBMP2.This indicates that RGMb is a coinhibitor of specifically ID1 and ID2 mRNA expression induction conferred by rBMP2 stimulation.However, in the absence of rBMP2, tendencies of elevated ID1 and ID2, but not ID3, mRNA expression levels were observed as a result of ectopic expression of RGMb.This elevation in ID1 and ID2 mRNA expression will, in itself, reduce the effect of rBMP2 stimulation, if there is a limitation to the possible induction of ID mRNA expression levels by rBMP2.Thus, the results showing the inhibitory function of RGMb could be a reflection of RGMb coactivation of ID1 and ID2 mRNA expression in the HCC827 cells beforehand rBMP2 stimulation.The dissimilar rBMP2-mediated effects we observed in HCC827eRGMb cells regarding induction of ID1 and ID2 relative to ID3 mRNA expression levels are consistent with the fact that expression of the ID genes can be regulated by different mechanisms and that functions of RGMb related to EMT, as well as BMP2-mediated ID1 and ID2 mRNA induction, were not mimicked by either PD-L2 depletion or overexpression.Concludingly, our results do not indicate that PD-L2 is the membrane-bound cosuppressor of BMP2/4signaling acting together with RGMb in NSCLC cells.5 | CONCLUSIONThe present study shows that RGMb can be involved in regulating BMP2/4 signaling and thereby ID mRNA expression in NSCLC cells.Moreover, we present evidence that RGMb is involved in the regulation of the EMT balance of NSCLC cells.EMT has pivotal roles in the progression of metastatic disease, intrinsic and acquired treatment resistance, and PD-1/PD-ligand-axis immunotherapy response, and therefore, this observation could have important future diagnostic and treatment implications for NSCLC.Decreasing and increasing PD-L2 expression did not mimic the identified RGMb functions, indicating that RGMb can act independently of PD-L2 in the execution of these regulatory functions in NSCLC cells.