© 2019. Published by The Company of Biologists Ltd | Journal of Cell Science (2019) 132, jcs220665. doi:10.1242/jcs.220665

RESEARCH ARTICLE The cancer-associated meprin β variant G32R provides an additional activation site and promotes cancer cell invasion Henning Schäffler1, Wenjia Li4, Ole Helm2, Sandra Krüger3, Christine Böger3, Florian Peters1, Christoph Röcken3, Susanne Sebens2, Ralph Lucius4, Christoph Becker-Pauly1 and Philipp Arnold4,*

ABSTRACT Thus, a well-balanced ratio of shed and membrane-bound meprin β The extracellular metalloprotease meprin β is expressed as a is required for a specific cellular function. β homodimer and is primarily membrane bound. Meprin β can be Meprin displays a unique cleavage preference for negatively released from the cell surface by its known sheddases ADAM10 and charged amino acids around the scissile bond (Becker-Pauly et al., ADAM17. Activation of pro-meprin β at the cell surface prevents its 2011), and has been shown to play important roles for proper shedding, thereby stabilizing its proteolytic activity at the plasma collagen I maturation in the skin and for the mucus detachment in membrane. We show that a single amino acid exchange variant the small intestine (Wichert et al., 2017; Broder et al., 2013; Schutte β (G32R) of meprin β, identified in endometrium cancer, is more active et al., 2014). Increased levels of meprin have been found in fibrotic against a peptide substrate and the IL-6 receptor than wild-type meprin conditions of the skin and lung (Becker-Pauly et al., 2007; Biasin β β. We demonstrate that the change to an arginine residue at position 32 et al., 2014). The cellular localization of meprin has been shown to represents an additional activation site used by furin-like proteases in be apical in epithelial cells of the proximal tubules of the kidney and the Golgi, which consequently leads to reduced shedding by ADAM17. enterocytes of the small intestine (Wichert et al., 2017; Schutte We investigated this meprin β G32R variant to assess cell proliferation, et al., 2014; Butler et al., 1987; Beynon et al., 1981; Sterchi et al., β invasion through a collagen IV matrix and outgrowth from tumor 1982). It has also been reported that meprin relocates to the spheroids. We found that increased meprin β G32R activity at the cell basolateral side during cisplatin-induced acute kidney injury, where surface reduces cell proliferation, but increases cell invasion. it cleaves nidogen-1, a component of the basal membrane. This cleavage of nidogen-1 was not detected in meprin β-deficient KEY WORDS: Meprin, Protease, Shedding, ADAM17, Endometrium, animals (Herzog et al., 2015). Additionally, it has been shown in Cell invasion biochemical experiments, that meprin β cleaves the important basal membrane component collagen IV (Kruse et al., 2004). INTRODUCTION Previously, we found that the co-expression of the membrane- Meprin β is an extracellular, multi-domain protease that forms bound serine protease matriptase-2 together with meprin β increases disulfide-linked homodimers at the cell surface (Prox et al., 2015). the proteolytic activity at the cell surface and reduces the amount of The active site of the protease domain is shielded by an inhibitory shed meprin β (Jäckle et al., 2015; Wichert et al., 2017). Here, we pro-peptide and, to gain proteolytic activity, meprin β requires characterize two cancer-associated mutations (G32R and D47A) activation by soluble or membrane-bound serine proteases (Ohler that are located within the pro-peptide and that likely influence et al., 2010; Jäckle et al., 2015). Of note, activation of meprin β is a activation and activity of meprin β. Both variants were identified in fate-determining step of the protease. If activated on the cell surface, large-scale screenings of tumor samples and were deposited in the for example by matriptase-2, meprin β can no longer be publicly available repositories (e.g. BioMuta-HIVE) (Simonyan proteolytically released (shed) by its known sheddases ADAM10 and Mazumder, 2014). Previously, we found that the introduction of or ADAM17 (Jefferson et al., 2013; Wichert et al., 2017). When single amino acid changes can alter the cleavage preference of released from the cell surface in its inactive state by ADAM10 or meprin β, namely, that the D204A variant of meprin β processes the ADAM17, it will be activated by soluble serine proteases, amyloid precursor protein (APP) to a higher extent; however, it is including, for example, kallikrein 4 and kallikrein 5 (Ohler et al., restricted to the secretory pathway and cleaves APP there (Arnold 2010). This difference in localization substantially determines the et al., 2015). substrate accessibility, as substrates such as the IL-6 receptor can Here, we found that both the G32R and D47A variants are only be cleaved by membrane-bound meprin β (Arnold et al., 2017). transported to the cell surface and exhibit comparable expression with However, not all substrates are accessible for the membrane-bound wild-type meprin β. However, the G32R variant showed higher version of the protease including mucin 2 and pro-collagen I proteolytic activity at the cell surface due to the change to an arginine (Wichert et al., 2017; Broder et al., 2013; Schutte et al., 2014). residue at position 32, which provides an additional activation site, as revealed by N-terminal detection. Tumor cell invasion through a β 1Biochemical Institute, Otto-Hahn Platz 9, 24118 Kiel, Germany. 2Institute for collagen IV matrix was increased for cells expressing the meprin Experimental Cancer Research, Arnold-Heller-Str. 3, 24105 Kiel, Germany. 3Dept. G32R variant, as was outgrowth length from tumor spheroids of Pathology, Christian-Albrechts-University and University Hospital Schleswig- compared to that seen for a non-activate able meprin β variant. Holstein, Arnold-Heller-Str. 3/14, 24105 Kiel, Germany. 4Anatomical Institute, Otto-Hahn Platz 8, 24118 Kiel, Germany. RESULTS *Author for correspondence ([email protected]) Localization and cell surface expression of meprin β variants P.A., 0000-0003-3273-9865 G32R and D47A At first, we evaluated the cellular localization of the G32R and

Received 23 May 2018; Accepted 23 April 2019 D47A variants. Both of these amino acid positions are located in the Journal of Cell Science

1 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220665. doi:10.1242/jcs.220665 pro-peptide of the protease and might modulate its activity 17−/− cell line was used to prevent shedding mediated by (Fig. 1A). To assess the cell surface level, we transfected Cos-7 endogenous ADAM10 or ADAM17, which would change the cells with each variant and performed cell surface protein protease level at the cell surface. By using a peptide cleavage assay, biotinylation experiments. We observed no significant difference we found that the activity of the G32R variant at the cell surface was between the variants and wild-type meprin β; however, a trend significantly higher than the activity detected for D47A and wild- to a lower surface expression was found for the G32R variant type meprin β (Fig. 2A). In this assay, a meprin β-specific peptide (Fig. 1B,C). We confirmed this finding by fluorescence microscopy, substrate that is linked to a fluorogenic group (MCA) at the N- demonstrating that wild-type and the two meprin β variants were terminus and a quencher (DPN) at the C-terminus (see Materials and found mainly at the cell surface (Fig. 1D). Thus, the expression Methods) was added to the cells (Broder and Becker-Pauly, 2013). pattern and cellular localization of the two amino acid exchange Upon cleavage by the protease, an increase in fluorescence intensity variants did not differ from that of wild-type meprin β. was observed (Fig. S1A). Western blot analysis was used a control for equal expression of all meprin β variants (Fig. S1B). As The G32R meprin β variant shows increased activity previously we identified the serine protease matriptase-2 (MT-2) as As we hypothesized that the exchange of a glycine residue into an a membrane-bound activator of meprin β (Jäckle et al., 2015), we arginine residue or an aspartate residue into an alanine residue might were interested whether the two exchange variants can be further change meprin β activation and activity, we transfected HEK cells activated by MT-2. Therefore, the respective meprin β variant and that were double-negative for ADAM10 and ADAM17 (denoted MT-2 were co-expressed in ADAM10/17−/− HEK cells. The ADAM10/17−/−) cells with G32R, D47A and wild-type expression increase in proteolytic activity was again measured by performing plasmids for meprin β or a control plasmid (mock). The ADAM10/ the fluorogenic peptide cleavage assay (Fig. S1C) and western blot

Fig. 1. Expression pattern of G32R, D47A and wild-type meprin β. (A) Molecular model of meprin β on the cell surface with one monomer shown in dark gray and the other in light gray. The insert shows the pro-peptide (orange) and the three active site histidine residues (cyan) that coordinate the zinc ion (gold). The positions of the two amino acid exchange variants [glycine to arginine (G32R), and aspartate to alanine (D47A)] are shown in red. (B) Representative western blot of a surface biotinylation of Cos-7 cells, with the transferrin receptor 1 (transferrin-R 1) used as a loading control for the membrane fraction. Cells were transfected with the meprin β variants indicated and, after biotinylation, the biotin-bound proteins were pulled down and analyzed by western blotting. (C) Quantification (mean±s.d.) of four individual experiments as shown in B. Meprin β amounts in the lysate were divided by the amount in the surface fraction, and all three variants were detected at the surface to the same extent (n=4). ns, not significant (P>0.05, one-way ANOVA with Tukey post-hoc test). (D) Immunofluorescence images of Cos-7 cells expressing wild-type, D47A and G32R (red) meprin β, with the endoplasmic reticulum marker protein disulfide isomerase A6 (PDIA6) (green). Journal of Cell Science

2 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220665. doi:10.1242/jcs.220665

Fig. 2. Cell surface activity and shedding of the meprin β variants. (A) ADAM10/17-deficient HEK cells were transfected with a control plasmid (Mock) or one of the three meprin β variants [wild-type (WT), D47A or G32R]. The proteolytic activity was determined using a peptide cleavage assay and the meprin β inhibitor actinonin was added as a control (n=4). ***P<0001; ns, not significant (P>0.05) (one-way ANOVA with Tukey post-hoc test). Representative activity measurements and control western blots are presented in Fig. S1A,B. (B) As in A, but cells were co-transfected with the known meprin β activator matriptase-2 (MT-2). Note, that the activity is not significantly increased for the G32R variant, as the activity is already higher when compared to the other meprin β variants (n=4). **P<0.01; ***P<0.001; ns, not significant (P>0.05) (one-way ANOVA with Tukey post-hoc test). Representative activity measurements and control western blots are shown in Fig. S1C,D. (C) Representative western blot of the IL-6 receptor (IL-6R) and its soluble form (sIL-6R) generated by cells transfected with the indicated form of meprin β. Cell culture supernatant was collected after 24 h and used for precipitation of the sIL-6R. Note the increased signal for the sIL-6R upon co-expression with the G32R variant. (D) Quantification of four individual experiments as shown in C. The ratio of IL-6R to sIL-6R (relative shed ILR) were determined and analyzed (n=4). *P<0.05; **P<0.01; ns, not significant (P>0.05) (one-way ANOVA with Tukey post-hoc test). (E) Meprin β variants were co-expressed with ADAM17 in ADAM10/17-deficient HEK cells. To assess the shedding potential of ADAM17 against the different meprin β variants, cell culture supernatant and cell lysate was analyzed by western blotting, and the ratio of shed to membrane-bound meprin β was quantified (n=4). *P<0.05; ns, not significant (P>0.05) (one-way ANOVA with Tukey post-hoc test). A representative western blot is shown in Fig. S2A. (F) After activation of shed meprin β from experiments as in E with trypsin, activity was determined by performing a peptide cleavage assay (n=4). ***P<0.001; ns, not significant (P>0.05) (one-way ANOVA with Tukey post-hoc test). Representative activity measurements are shown in Fig. S2B. All graphs show mean±s.d. Activity for all experimental groups was normalized to the median value of the activity measured for wt meprin β. analysis of cell lysates was used a control for equal expression with the different meprin β variants and the known protein (Fig. S1D). For wild-type meprin β and the D47A variant, a substrate IL-6 receptor (IL-6R) (Arnold et al., 2017). The significant increase in activity was detected after co-transfection soluble proteolytically shed IL-6R (sIL-6R) was precipitated with MT-2. Again, we observed an increased activity of the G32R from the cell culture supernatant and analyzed by western variant in single-transfected cells, which was increased after co- blotting (Fig. 2C). The releases of sIL-6R into the supernatant transfection with MT-2. This increase was not significant; however, was significantly higher after co-transfection with G32R than the basal activity of the G32R variant is already elevated (Fig. 2B), for D47A or wild-type meprin β (Fig. 2D). Taken together, and thus the relative increase in activity is smaller than for D47A these results show that G32R variant of meprin β has a and wild-type meprin β. significantly increased proteolytic activity at the cell surface To analyze the proteolytic activity towards a membrane- without the need for additional expression of activating −/− bound substrate, we co-transfected ADAM10/17 HEK cells serine proteases. Journal of Cell Science

3 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220665. doi:10.1242/jcs.220665

Shedding of meprin β G32R mediated by ADAM17 is impaired activity for the G32R variant was seen in the peptide cleavage Activation of meprin β and its shedding from the cell surface are assay (Fig. 2F; Fig. S2B). closely connected (Wichert et al., 2017). We have previously The consensus activation site of meprin β cleaved by different found that only inactive pro-meprin β can be shed by ADAM10 tryptic serine proteases is located between arginine 61 and and ADAM17, but not active membrane-bound meprin β (Wichert asparagine 62 (Jäckle et al., 2015). Thus, a mutant that carries a et al., 2017). We hypothesized, that increased activation of serine at position 61 cannot be activated (Fig. 3A) (Jäckle et al., G32R at the cell surface influences meprin β shedding. Thus, 2015). We introduced an arginine residue at position 32, to create a we co-expressed ADAM17 together with the meprin β variants in G32R/R61S double mutant (Fig. 3A). The peptide cleavage assay ADAM10/17−/− HEK cells. Western blot analysis of cell culture revealed a significant increase in the activity of the double mutant supernatants revealed that the G32R variant was shed to a (G32R/R61S) compared to the R61S single mutation alone significantly lesser extend than D47A or wild-type meprin β (Fig. 3B). This suggests, that the arginine residue at position 32 (Fig. 2E; Fig. S2A). To assess the activity of shed meprin β in the can be cleaved by different serine proteases and that this cleavage cell supernatant, the protease was activated with trypsin, as only leads to a partial activation of meprin β. To further investigate the inactive pro-meprin β is released by ADAM proteases. As removal of the pro-peptide, we generated meprin β variants that indicated by our western blot analysis, a significantly decreased are additionally N-terminally Strep-tagged (Peters et al., 2019).

Fig. 3. An arginine residue at position 32 serves as an additional activation site in pro-meprin β. (A) Molecular model of membrane-bound meprin β. The insert shows the amino acid changes made to produce the G32R, R61S and G32R/R61S (G32R_R61S) mutants at position 32 and at the canonical activation site at position 61. Colors are as in Fig. 1A. (B) Cell surface activity measurements for ADAM10/17-deficient HEK cells transfected with a control plasmid (Mock) or one of the four meprin β variants [wild-type (WT), R61S, G32R or G32R/R61S]. Note that the insertion of an arginine at position 32 reconstitutes the loss of activity in the R61S mutant (n=4). *P<0.05; **P<0.01 (one-way ANOVA with Tukey post-hoc test). Representative activity measurements and western blots are shown in Fig. S2C,D. Activity was normalized to the median value of the activity measured for wt meprin β. (C) Cartoon of an N-Strep- and C-Flag-tagged meprin β variant. The Strep tag can only be detected for inactive meprin β, while the Flag tag can be detected for both, active and inactive meprin β. (D) Strep–Flag-tagged meprin β variants analyzed 36 h past transfection by western blotting to discriminate between active and inactive meprin β. Actin was used as a loading control. (E) Quantification of results shown in D reveals a significant reduction in N-terminal (Strep tag) to C-terminal (Flag tag) signal ratio (n=4). **P<0.01; ns, not significant (P>0.05) (one-way ANOVA with Tukey post-hoc test). (F) Representative western blot for experiments as in D, but after the addition of Pefabloc for 12 h at 24 h post transfection. (G) Quantification of results shown in F (n=4). *P<0.05; ***P<0.001; ns, not significant (P>0.05) (one-way ANOVA with Tukey post-hoc test). (H) Representative western blot for experiments as in D, but after the addition of proprotein convertase inhibitor for 12 h 24 h post transfection. (I) Quantification of results shown in H (n=4). ns, not significant (P>0.05) (one-way ANOVA with Tukey post-hoc test). (J) Summarizing cartoon illustrating the activation of G32R meprin β in the Golgi and of WT meprin β at the cell surface. All graphs show mean±s.d. Journal of Cell Science

4 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220665. doi:10.1242/jcs.220665

Thus, the activated and non-activated meprin β can be discriminated is pre-activated in the Golgi and delivered to the cell surface as an (Fig. 3C). To further elucidate the activating mechanism, different (at least partially) active enzyme. serine protease inhibitors were used. To inhibit cell surface activation of meprin β, Pefabloc, a non-membrane transmissible inhibitor was Cell proliferation and migration added to the cells. To inhibit intracellular furin-like serine proteases We next elucidated whether the shift in proteolytic activity of the in the Golgi network, a proprotein convertase inhibitor (PCI) was meprin β variants has functional consequences, and evaluated cell applied. Untreated conditions revealed no significant difference proliferation and cell invasion, which are hallmarks of cancer between WT and R61S variants (Fig. 3D,E), which might be due to development. Increased cell proliferation and decreased apoptosis the fast turnover of active meprin β at the cell surface as revealed by promote tumor growth. Cell migration/invasion is needed for the co-expression experiments with MT-2 (Jäckle et al., 2015). formation of metastases released from the primary tumor. Changes in However, both variants carrying an arginine residue at position 32 the proliferative and invasive potential of HeLa cells upon transfection (G32R and G32R/R61S) have a significantly lower N-terminal than with different meprin β variants were measured by using an C-terminal signal (Fig. 3D,E). The addition of Pefabloc did not xCelligence real-time cell analyzer (ACEA Biotechnologies). In lead to a marked increase in N-terminal Strep-tag signal for G32R these assays, increased proliferation or invasion leads to increased and G32R/R61S, indicating only a minor cleavage at position R32 electrical impedance values in the respective experimental setup, at the cell surface (Fig. 3F,G). Inhibition of furin-like proprotein expressed in an arbitrary unit, the cell index (CI). We observed convertases by using PCI increased the level of detectable decreased cell proliferation upon expression of all meprin β variants in N-terminus (Strep-tag) to wild-type and R61S levels for G32R and HeLa cells. The reduction seen for the G32R variant was significantly G32R/R61S variants, respectively (Fig. 3H,I). Thus, we suggest that higher when compared to cells that did not express meprin β (Fig. 4A). the protein can be cleaved at R32 in the Golgi, unlike R61, which is Another hallmark of cancer cells is their ability to invade through the only accessible for cell surface serine proteases (Fig. 3J). basal lamina. Here, we mimicked this basal laminar with a layer of Taken together, we found that the ratio between membrane- collagen IV (Col IV), one major component of it. We transfected HeLa tethered and shed meprin β was altered upon the introduction of an cells and allowed them to invade through a collagen IV matrix. We arginine residue at position 32. This correlated with an increased observed that meprin β-expressing cells showed a significantly higher proteolytic activity at the cell surface and decreased proteolytic invasive potential than control cells. Interestingly, we found an even activity in the supernatant. Additionally, we found that G32R higher invasion rate for cells expressing the G32R variant (Fig. 4B).

Fig. 4. Cell proliferation and migration. (A) A cell proliferation assay (see Materials and Methods) revealed that there was significantly reduced proliferation of HeLa cells transfected with the G32R meprin β variant when compared to mock-transfected cells, or cells transfected with wild-type (WT) or D47A meprin β. CI, cell index (presented relative to the results for Mock). (B) A cell invasion assay through a collagen IV matrix revealed a significant increase for cells transfected with meprin β. This increase is significantly higher when cells express the G32R variant. (C) An example spheroid from mock-transfected HeLa cells is shown at 48 h past evasion start. (D) A spheroid with the evasion area marked. In green, a manual outline is shown, which enabled semi-automated finding of cells. (E) Spheroid marked with two circles to assess the maximum outgrowth distance. Scale bars: 250 µm. (F) Statistical analysis of results of the spheroid assay for HeLa cells transfected as indicated. No significant differences were found for the same time point between the groups (n=5) (two-way ANOVA with Tukey’s post-hoc test). (G) Statistical analysis of the outgrowth distance over time for the spheroid assay (n=5). *P<0.05 (two- way ANOVA with Tukey’s post-hoc test). All graphs show mean±s.e.m. Journal of Cell Science

5 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220665. doi:10.1242/jcs.220665

To assess cellular properties in a more tumor-like environment, DISCUSSION we used a spheroid evasion assay. Therefore, we generated tumor In this study, we investigated the molecular and cellular consequences spheroids from HeLa cells (Fig. 4C) and evaluated the outgrowth of the cancer-associated meprin β variants G32R and D47A, both area (Fig. 4D) and the outgrowth distance (Fig. 4E) after 24 h, 48 h identified in endometrium cancers. We found that the G32R single and 72 h. The cellular outgrowth did change over time, but was not amino acid exchange variant had higher cell surface activity significantly altered among cells expressing the different meprin β compared to wild-type meprin β through a premature activation in variants at a given time point. Western blotting was performed to the Golgi network and thus changes the balance between the amount confirm similar levels of expression of meprin β variants (Fig. S2D). of soluble and membrane tethered protease because of impaired When comparing the outgrowth distance, we found it to be shedding by ADAM17. As a consequence, cleavage of a peptide significantly longer for cells expressing the G32R compared to the substrate and the IL-6R by the meprin β variant G32R was increased. inactive R61S variant at 48 h (Fig. 4G). Previously, we had shown that matriptase-2, a membrane-bound serine protease, activates meprin β at the cell surface and thereby Meprin β is expressed in murine and human endometrium impairs its shedding by ADAM proteases (Jäckle et al., 2015; As both meprin β variants are found in endometrial tumors Wichert et al., 2017). Here, we could show that the G32R variant has (COSMIC database IDs 987719 and 987721), we investigated a similar effect, but relies on the endogenously expressed proprotein general expression of this protease in the endometrium. Therefore, convertases in the Golgi. This renders the G32R mutation a valuable we analyzed tissues from wild-type and meprin β-deficient tool to simulate increased meprin β activity at the cell surface. In a [Mep1b knockout (ko)] animals by immunohistochemistry. previous study, we found that a premature activation of meprin β in We found that meprin β is expressed at the apical side of cells in the endoplasmic reticulum impairs its transport to the cell surface the mucosal cell layer (Fig. 5A). In unaffected control areas of (Arnold et al., 2015). The present data suggests now that active human endometrium samples, we could confirm the expression of meprin β can be transported from the Golgi to the cell surface. meprin β and, as in the murine sample, found it at the apical side In this study, we show that the metalloprotease meprin β is of endometrial gland cells (Fig. 5B). To evaluate the expression present at the apical side of epithelial cells of the murine and human in endometrial tumor samples, these were stained for meprin β,too. endometrium. This is in line with the expression at the apical side of We observed that the distribution of meprin β was changed enterocytes in the small intestine (Wichert et al., 2017; Schutte et al., from apical to also showing basolateral localization within the 2014) and of proximal tubule cells in the kidney (Butler et al., cells (Fig. 5B). As the cellular polarity is lost in tumor tissue 1987). The exact function of meprin β in healthy endometrium is (Khursheed and Bashyam, 2014), a redistribution of meprin β currently unknown. However, as meprin β is important for proper would be expected. Some cells within the tumor seem to express mucus detachment in the small intestine (Wichert et al., 2017; meprin β at higher levels than others and additional cells positive Schutte et al., 2014), it might have a similar function in the for meprin β were detected outside the tumor (Fig. 5B). Taken endometrium. We found that the subcellular localization of meprin together, we found meprin β expression at the apical side of cells in β changes in endometrial tumors from the apical to also being control endometrial tissue, which was changed in samples from present at the basolateral side, which might be induced by a loss of tumor patients. cell polarity in these tumor cells (Khursheed and Bashyam, 2014).

Fig. 5. Expression of meprin β in the endometrium. (A) Sections of endometrium samples from wild-type (Mep1b wt) and meprin β-knockout (Mep1b ko) mice stained with H&E (left) or with antibody against meprin β (right). Note, in Mep1b wt animals a clear signal is detected for meprin β at the apical side of cells (arrows). This signal is missing in Mep1b ko animals. (B) Sections from human endometrium control and tumor sample tissue stained for meprin β. The control section shows a clear signal at the apical side (arrows). In tumor samples, this strict apical expression is lost and more randomly distributed signals can also be detected at the basolateral side (arrows). Journal of Cell Science

6 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220665. doi:10.1242/jcs.220665

The partial redistribution of meprin β from the apical to the deficient for ADAM10/17 (Riethmueller et al., 2017) or wild-type were basolateral side enables direct contact with well-described substrates seeded in Dulbecco’s modified Eagle’s medium (DMEM) plus GlutaMax in the basal lamina, such as collagen IV, laminin and nidogen, and (Thermo Fisher Scientific) supplemented with 10% fetal calf serum (FCS) makes them accessible for proteolytic cleavage (Kruse et al., 2004). and 1% penicillin and streptomycin (P/S) (Thermo Fisher Scientific). Cells An increased cleavage of nidogen-1 at the basal lamina has been were grown to 70% confluence and were then transfected with the respective DNA plasmids by using polyethylenimine (PEI). After 24 h, the cell culture described for meprin β in a mouse model of cisplatin-induced acute β medium was changed to DMEM plus GlutaMax without further supplements kidney injury. In this model, the cell polarity is lost and meprin is to enhance the proteolytic potential of meprin β. After an additional 24 h, the also distributed to the basolateral side (Herzog et al., 2015). Here, we cell supernatant and cells were harvested. Cells were washed three times with demonstrate that this accessibility enhances the invasive potential of PBS, lysed with lysis buffer (PBS pH 7.4, 1% Triton X-100, supplemented meprin β-expressing HeLa cells through a collagen IV matrix. This is with Complete EDTA free (Roche)], and protein content was determined by in line with recent studies, demonstrating that meprin β increases using a BCA kit (Thermo Fisher Scientific). Samples were mixed with transmigration of Lewis lung carcinoma cells through an endothelial loading buffer and heated to 98°C for 10 min. Detection of soluble IL-6R was cell layer in vitro (Bedau et al., 2017a,b) and that fewer immune cells performed as described previously (Arnold et al., 2017). For experiments enter a peripheral site of inflammation in mice deficient for meprin β using inhibitors, cells were transfected as described above and medium was (Bedau et al., 2017a,b). This indicates a role for meprin β in the changed to inhibitor-containing medium supplemented with 10% FCS and 1% P/S for 12 h prior to cell harvesting. Pefabloc (Merck, Darmstadt, promotion of trans-endothelial migration for different cell types. Our β Germany) was added at 0.5 mg/ml and proprotein convertase inhibitor tumor spheroid assays reveal only minor effects for meprin on (Merck, Darmstadt, Germany) at 22.5 µM. outgrowth rate and distance. While meprin β might play an important role for the penetration of the basal lamina, its role during cell Cell surface protein biotinylation migration seems less pronounced and other proteases might take over Cos-7 cells were transiently transfected with the respective meprin β variants such as soluble and membrane type 1 matrix metalloproteinases and protein expression was allowed to procedure for 48 h. Cells were cooled to (MT1-MMP; also known as MMP14) (Thakur and Bedogni, 2016). 4°C and washed with PBS-CM (PBS supplemented with 0.1 mM CaCl2 and Cell proliferation and survival can be influenced by extracellular 1mMMgCl2), and incubated with biotin solution (1 mg/ml NHS-SS-Biotin proteolytic action. This was shown for the extracellular from Thermo Fisher Scientific in PBS-CM) at 4°C for 30 min, prior to metalloprotease ADAM17. It cleaves epidermal growth factor quenching with quenching buffer (50 mM Tris-HCl pH 8.0 in PBS-CM). Then cells were lysed and a BCA assay was performed as described above, and receptor ligands from the cell surface and thereby induces cell β lysate controls were taken for western blot analysis. From the remaining cell proliferation. Here, we found that increased activity of meprin on the lysate, ∼600 µg protein was taken and incubated with 75 µl of Streptavidin– cell surface reduced cell proliferation. Another pathway influenced by agarose resin (Thermo Fisher Scientific), washed three times with lysis buffer, ADAM17, emerging in different tumor entities, is interleukin 6 (IL-6) mixed with loading buffer and heated for denaturation. Subsequent western and IL-6R-induced trans-signaling (Bergmann et al., 2017; Schmidt blot analysis of the biotinylated and non-biotinylated samples revealed the et al., 2018). Here, ADAM17 releases the IL-6R from the cell surface presence of meprin β variants at the cell surface. As a control, transferrin 1 of one cell, and then this soluble receptor forms a signaling complex receptor was detected. Antibody details are given in Table S2. together with its cytokine and gp130 (also known as IL6ST) on Peptide cleavage assay another cell to induce cell survival via STAT3 phosphorylation β (Schaper and Rose-John, 2015). Interestingly, the G32R variant of To assess the activity of meprin on the cell surface, a peptide cleavage β assay was used as described previously (Arnold et al., 2015). In brief, a meprin sheds more IL-6R from the cell and we have also previously β β substrate peptide, specifically cleaved by meprin and coupled to a shown that wild-type meprin generates biologically active IL-6R, fluorophore (7-methoxycoumarin-4-acetic acid N-succinimidyl ester; which is capable of inducing trans-signaling (Arnold et al., 2017). MCA) on one side and a quencher (2,4-dinitrophenyl; DPN) on the other Taken together, we identify a single amino acid exchange variant side, was added to the cell supernatant. Upon cleavage, an increase in the of meprin β (G32R) that changes its proteolytic activity at the cell fluorescence signal is detected at 320 nm. The increase of this signal was surface and thereby diminishes its ADAM17-mediated shedding. detected in a 96-well plate reader (Infinit F200 Pro, Tecan, Crailsheim, We observed that cellular properties, such as proliferation and Germany) for the time indicated in the experiments. As a control, actinonin trans-migration, were directly linked to the amount of active meprin (Sigma Aldrich, Darmstadt, Germany), a meprin β inhibitor (10 μM), was β on the cell surface. Additionally, we could show that the added to the cell supernatant. expression at the apical side of endometrial mucosal cells is lost in β β Activation of meprin by trypsin tumor tissue and found cells positive for meprin in tumors. The β β To activate soluble pro-meprin that had been released to the supernatant exact role of meprin in the tumor environment has to be further through the action of ADAM17, pancreatic trypsin was used as described investigated in upcoming studies. previously (Wichert et al., 2017).

MATERIALS AND METHODS Histology and immunofluorescence Generation of the meprin β G32R and D47A variants Murine uterus was removed from wild-type or meprin β-deficient animals The two meprin β variants, G32R and D47A, were generated on the basis of (Yura et al., 2009) after devitalization of the mouse in accordance with the a wild-type meprin β pcDNA4/TO-construct that carries a C-terminal FLAG local animal protection law (animal protection law of Schleswig-Holstein, tag. Individual nucleotides were exchanged according to the exchange internal animal facility approval number 750). The uterus from 12-week-old variants reported (BioMuta database) using appropriate primers (Table S1) female mice was fixed in 4% paraformaldehyde (PFA) overnight and was and the QuikChange II XL site-directed mutagenesis kit (Agilent then post-fixed in 1% PFA. Samples were embedded in paraffin and Technologies, CA). The sequence of the obtained variants was confirmed sectioned into 7 µm slices. H&E staining was performed following routine by sequencing of the complete meprin β cDNA inserted in the expression protocols (Arnold et al., 2016). plasmid (GATC Biotech, Konstanz, Germany). We retrieved specimens of non-neoplastic endometrium and endometrial cancer from the archive of the Dep. of Pathology, University Medical Center Cell culture and transient transfection Schleswig-Holstein; approval was given by the Ethics Committee of the Cell culture experiments were performed as described previously (Arnold University Medical Center Schleswig-Holstein, Schwanenweg 20, 24105 et al., 2015). Briefly, Cos-7 and human embryonic kidney (HEK) cells Kiel. Working with human samples was approved by local authorities Journal of Cell Science

7 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220665. doi:10.1242/jcs.220665

(D 417/18). Diagnosis was based on histological assessment by a certified Acknowledgements consultant pathologist. Immunohistochemical meprin β staining of human We acknowledge Inez Götting, Bettina Facompréand Katrin Neblung-Masuhr for tissue specimens was carried out with the Bondmax automated slide staining excellent technical help. system (Leica Biosystems), using the Polymer Refine Detection Kit (Leica Biosystems), the Epitope Retrieval Solution 1 for 30 min and anti-meprin β Competing interests The authors declare no competing or financial interests. antibody (Table S2, dilution 1:1000). Immunostaining was visualized with DAB and counterstained with hemalaun. Author contributions Immunofluorescence (IF) staining was performed as described previously Conceptualization: P.A., C.B.-P.; Methodology: W.L., O.H., S.S.; S.K., C.B. (Arnold et al., 2015). Briefly, Cos-7 cells were seeded on glass slides and Investigation: H.S., W.L., S.K., C.B., P.A.; Resources: F.P., C.R.; R.L., C.B.-P. left overnight. Cells were then washed in PBS three times and fixed in 4% Data curation: H.S., W.L., P.A.; Writing - original draft: P.A.; Writing - review & PFA for 30 min. After permeabilization with saponin, cells were incubated editing: all authors; Visualization: H.S., W.L.; P.A.; Supervision: C.B.-P., P.A.; with the primary antibody (see Table S2) for 1 h, washed and then incubated Project administration: P.A.; Funding acquisition: C.B.-P., P.A. with the respective secondary antibody coupled to Alexa Fluor 488 or Alexa Fluor 594. Images were acquired on an Olympus FV 1000 confocal laser Funding scanning microscope (Olympus, Hamburg, Germany). HS was supported by a stipend from the medical faculty of the Christian Albrechts University Kiel. This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – project number 125440785 [SFB 877 Cell proliferation and invasion Projects A9, A15 (to C.B.-P.) and A13 (to P.A.)]. Transfected HeLa cells were analyzed for their proliferative and invasive potential using the xCELLigence DP real-time cell analyzer (ACEA, Supplementary information Biosciences) according to manufacturer’s instructions. Briefly, for Supplementary information available online at proliferation assays, cells were harvested, resuspended in DMEM/ http://jcs.biologists.org/lookup/doi/10.1242/jcs.220665.supplemental GlutaMax with 10% FCS and seeded at 104 cells per well in an E-plate in triplicates. Proliferation was tracked by measuring the impedance on the References bottom well electrode, representing the area covered by cells, as relative CI Arnold, P., Schmidt, F., Prox, J., Zunke, F., Pietrzik, C., Lucius, R. and Becker- units every 30 min for 72 h. Data are presented as the area under the curve of Pauly, C. (2015). Calcium negatively regulates meprin beta activity and the resulting graph, normalized to control-transfected HeLa cells. For attenuates substrate cleavage. FASEB J. 29, 3549-3557. doi:10.1096/fj.15- 272310 invasion assays, transfected HeLa cells were harvested, resuspended in Arnold, P., Rickert, U., Helmers, A. K., Spreu, J., Schneppenheim, J. and 4 DMEM/GlutaMax with 1% FCS and seeded at 5×10 cells per well in a Lucius, R. (2016). Trefoil factor 3 shows anti-inflammatory effects on activated CIM-plate in triplicates. Each well was previously coated with 12 µg of microglia. Cell Tissue Res. 365, 3-11. doi:10.1007/s00441-016-2370-5 human placenta-derived collagen IV (Sigma Aldrich) and allowed to dry, Arnold, P., Boll, I., Rothaug, M., Schumacher, N., Schmidt, F., Wichert, R., providing a matrix that cells had to invade through, to be measured on the Schneppenheim, J., Lokau, J., Pickhinke, U., Koudelka, T. et al. (2017). Meprin metalloproteases generate biologically active soluble interleukin-6 flip side of an 8 µm pore membrane. CI measurements were performed at receptor to induce trans-signaling. Sci. Rep. 7, 44053. doi:10.1038/srep44053 30 min intervals for 24 h. Data are presented as end point CI values, Arolas, J. L., Broder, C., Jefferson, T., Guevara, T., Sterchi, E. E., Bode, W., normalized to those of mock transfected HeLa cells. Stocker, W., Becker-Pauly, C. and Gomis-Ruth, F. X. (2012). Structural basis for the sheddase function of human meprin beta metalloproteinase at the plasma Production and analysis of tumor spheroids membrane. Proc. Natl. Acad. Sci. USA 109, 16131-16136. doi:10.1073/pnas. Tumor spheroids were generated from 10,000 HeLa cells following a 1211076109 Becker-Pauly, C., Howel, M., Walker, T., Vlad, A., Aufenvenne, K., Oji, V., Lottaz, previously described protocol (Vinci et al., 2015). HeLa cells were D., Sterchi, E. E., Debela, M., Magdolen, V. et al. (2007). The alpha and beta transfected 24 h before spheroid formation was started. Images of tumor subunits of the metalloprotease meprin are expressed in separate layers of human spheroids were acquired (5× magnification, Leica) at 24 h, 48 h and 72 h epidermis, revealing different functions in keratinocyte proliferation and after matrix (Geltrex® Matrix, Gibco, Life Technologies, catalog no. differentiation. J. Invest. Dermatol. 127, 1115-1125. doi:10.1038/sj.jid.5700675 A1413202) was supplemented. Becker-Pauly, C., Barré, O., Schilling, O., Auf dem Keller, U., Ohler, A., Broder, ̈ ̈ To analyze the maximum evasion length, spheroids were marked with C., Schutte, A., Kappelhoff, R., Stocker, W. and Overall, C. M. (2011). Proteomic analyses reveal an acidic prime side specificity for the astacin circles and the distance from the center of the spheroid circle to the metalloprotease family reflected by physiological substrates. Mol. Cell. furthest cell that evaded out was measured. Afterwards, the radius of the Proteomics 10, M111 009233. doi:10.1074/mcp.M111.009233 spheroid was subtracted. Data were analyzed using ImageJ and GraphPad Bedau, T., Peters, F., Prox, J., Arnold, P., Schmidt, F., Finkernagel, M., Prism 6. Kollmann, S., Wichert, R., Otte, A., Ohler, A. et al. (2017a). Ectodomain To analyze the evasion area, cells that evaded from the spheroid were shedding of CD99 within highly conserved regions is mediated by the metalloprotease meprin beta and promotes transendothelial cell migration. selected automatically in Image-Pro 10 (Media Cybernetics) and noise FASEB J. 31, 1226-1237. doi:10.1096/fj.201601113R signals were deleted manually. The evasion area was calculated by the sum Bedau, T., Schumacher, N., Peters, F., Prox, J., Arnold, P., Koudelka, T., Helm, of pixels of the selected area. Statistical analysis was performed using O., Schmidt, F., Rabe, B., Jentzsch, M. et al. (2017b). Cancer-associated GraphPad Prism 6. mutations in the canonical cleavage site do not influence CD99 shedding by the metalloprotease meprin beta but alter cell migration in vitro. Oncotarget 8, 54873-54888. doi:10.18632/oncotarget.18966 Molecular imaging Bergmann, J., Müller, M., Baumann, N., Reichert, M., Heneweer, C., Bolik, J., Molecular graphics were produced using UCSF Chimera (Pettersen et al., Lücke, K., Gruber, S., Carambia, A., Boretius, S. et al. (2017). IL-6 trans- 2004). The basis for imaging was the crystal structure of pro-meprin β (PDB signaling is essential for the development of hepatocellular carcinoma in mice. ID 4GWM) and a molecular model of the membrane-bound form, Hepatology 65, 89-103. doi:10.1002/hep.28874 containing the EGF-like domains, the transmembrane helix and the C- Beynon, R. J., Shannon, J. D. and Bond, J. S. (1981). Purification and terminal tail (Arolas et al., 2012). Single amino acid exchanges were characterization of a metallo-endoproteinase from mouse kidney. Biochem. J. swapaa 199, 591-598. doi:10.1042/bj1990591 produced with the UCSF Chimera command line tool (https://www. Biasin, V., Marsh, L. M., Egemnazarov, B., Wilhelm, J., Ghanim, B., Klepetko, cgl.ucsf.edu/chimera/). W., Wygrecka, M., Olschewski, H., Eferl, R., Olschewski, A. et al. (2014). Meprin beta, a novel mediator of vascular remodelling underlying pulmonary Statistical analysis hypertension. J. Pathol. 233, 7-17. doi:10.1002/path.4303 All statistical analysis was performed using GraphPad Prism 6, and the Broder, C. and Becker-Pauly, C. (2013). The metalloproteases meprin alpha and test indicated in the figure legend was used for the individual meprin beta: unique enzymes in inflammation, neurodegeneration, cancer and fibrosis. Biochem. J. 450, 253-264. doi:10.1042/BJ20121751 experiments. The density of western blot signals was detected using Broder, C., Arnold, P., Vadon-Le Goff, S., Konerding, M. A., Bahr, K., Muller, S., ImageJ. For soluble protein signals, the corresponding lysate form was Overall, C. M., Bond, J. S., Koudelka, T., Tholey, A. et al. (2013). used for quantification. Metalloproteases meprin alpha and meprin beta are C- and N-procollagen Journal of Cell Science

8 RESEARCH ARTICLE Journal of Cell Science (2019) 132, jcs220665. doi:10.1242/jcs.220665

proteinases important for collagen assembly and tensile strength. Proc. Natl. Riethmueller, S., Somasundaram, P., Ehlers, J. C., Hung, C. W., Flynn, Acad. Sci. USA 110, 14219-14224. doi:10.1073/pnas.1305464110 C. M., Lokau, J., Agthe, M., Dusterhoft, S., Zhu, Y., Grotzinger, J. et al. Butler, P. E., McKay, M. J. and Bond, J. S. (1987). Characterization of meprin, a (2017). Proteolytic origin of the soluble human IL-6R in vivo and a decisive membrane-bound metalloendopeptidase from mouse kidney. Biochem. J. 241, role of N-glycosylation. PLoS Biol. 15, e2000080. doi:10.1371/journal.pbio. 229-235. doi:10.1042/bj2410229 2000080 Herzog, C., Marisiddaiah, R., Haun, R. S. and Kaushal, G. P. (2015). Basement Schaper, F. and Rose-John, S. (2015). Interleukin-6: biology, signaling and membrane protein nidogen-1 is a target of meprin beta in cisplatin nephrotoxicity. strategies of blockade. Cytokine Growth Factor. Rev. 26, 475-487. doi:10.1016/j. Toxicol. Lett. 236, 110-116. doi:10.1016/j.toxlet.2015.05.005 cytogfr.2015.07.004 Jäckle, F., Schmidt, F., Wichert, R., Arnold, P., Prox, J., Mangold, M., Ohler, A., Pietrzik, Schmidt, S., Schumacher, N., Schwarz, J., Tangermann, S., Kenner, L., C. U., Koudelka, T., Tholey, A. et al. (2015). Metalloprotease meprin beta is activated by Schlederer, M., Sibilia, M., Linder, M., Altendorf-Hofmann, A., Knosel, T. et al. transmembrane serine protease matriptase-2 at the cell surface thereby enhancing APP (2018). ADAM17 is required for EGF-R-induced intestinal tumors via IL-6 trans- shedding. Biochem. J. 470, 91-103. doi:10.1042/BJ20141417 signaling. J. Exp. Med. 215, 1205-1225. doi:10.1084/jem.20171696 Jefferson, T., Auf dem Keller, U., Bellac, C., Metz, V. V., Broder, C., Hedrich, J., Schutte, A., Ermund, A., Becker-Pauly, C., Johansson, M. EV., Rodriguez- Ohler, A., Maier, W., Magdolen, V., Sterchi, E. et al. (2013). The substrate Pineiro, A. M., Backhed, F., Muller, S., Lottaz, D., Bond, J. S. and Hansson, degradome of meprin metalloproteases reveals an unexpected proteolytic link G. C. (2014). Microbial-induced meprin beta cleavage in MUC2 mucin and a between meprin beta and ADAM10. Cell. Mol. Life Sci. 70, 309-333. doi:10.1007/ functional CFTR channel are required to release anchored small intestinal s00018-012-1106-2 mucus. Proc. Natl. Acad. Sci. USA 111, 12396-12401. doi:10.1073/pnas. Khursheed, M. and Bashyam, M. D. (2014). Apico-basal polarity complex and 1407597111 cancer. J. Biosci. 39, 145-155. doi:10.1007/s12038-013-9410-z Simonyan, V. and Mazumder, R. (2014). High-performance integrated virtual Kruse, M. N., Becker, C., Lottaz, D., Köhler, D., Yiallouros, I., Krell, H. W., environment (HIVE) tools and applications for big data analysis. Genes (Basel) 5, Sterchi, E. E. and Stöcker, W. (2004). Human meprin alpha and beta homo- 957-981. doi:10.3390/genes5040957 oligomers: cleavage of basement membrane proteins and sensitivity to Sterchi, E. E., Green, J. R. and Lentze, M. J. (1982). Non-pancreatic hydrolysis of metalloprotease inhibitors. Biochem. J. 378, 383-389. doi:10.1042/bj20031163 N-benzoyl-l-tyrosyl-p-aminobenzoic acid (PABA-peptide) in the human small Ohler, A., Debela, M., Wagner, S., Magdolen, V. and Becker-Pauly, C. (2010). intestine. Clin. Sci. (Lond.) 62, 557-560. doi:10.1042/cs0620557 Analyzing the protease web in skin: meprin metalloproteases are activated Thakur, V. and Bedogni, B. (2016). The membrane tethered matrix specifically by KLK4, 5 and 8 vice versa leading to processing of proKLK7 thereby metalloproteinase MT1-MMP at the forefront of melanoma cell invasion and triggering its activation. Biol. Chem. 391, 455-460. doi:10.1515/bc.2010.023 metastasis. Pharmacol. Res. 111, 17-22. doi:10.1016/j.phrs.2016.05.019 Peters, F., Scharfenberg, F., Colmorgen, C., Armbrust, F., Wichert, R., Arnold, Vinci, M., Box, C. and Eccles, S. A. (2015). Three-dimensional (3D) tumor spheroid P., Potempa, B., Potempa, J., Pietrzik, C. U., Hasler, R., (2019). Tethering invasion assay. J. Vis. Exp. 99, e52686. doi:10.3791/52686. soluble meprin alpha in an enzyme complex to the cell surface affects IBD- Wichert, R., Ermund, A., Schmidt, S., Schweinlin, M., Ksiazek, M., Arnold, P., associated genes. FASEB J. fj201802391R. doi:10.1096/fj.201802391R Knittler, K., Wilkens, F., Potempa, B., Rabe, B. et al. (2017). Mucus detachment Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., by host metalloprotease meprin beta requires shedding of its inactive pro-form, Meng, E. C. and Ferrin, T. E. (2004). UCSF Chimera–a visualization system for which is abrogated by the pathogenic protease RgpB. Cell Rep. 21, 2090-2103. exploratory research and analysis. J. Comput. Chem. 25, 1605-1612. doi:10. doi:10.1016/j.celrep.2017.10.087 1002/jcc.20084 Yura, R. E., Bradley, S. G., Ramesh, G., Reeves, W. B. and Bond, J. S. (2009). Prox, J., Arnold, P. and Becker-Pauly, C. (2015). Meprin alpha and meprin beta: Meprin A metalloproteases enhance renal damage and bladder inflammation after procollagen proteinases in health and disease. Matrix Biol. 44-46, 7-13. doi:10. LPS challenge. Am. J. Physiol. Renal. Physiol. 296, F135-F144. doi:10.1152/ 1016/j.matbio.2015.01.010 ajprenal.90524.2008 Journal of Cell Science

9