A Novel Pro-Inflammatory Mechanosensing Pathway Orchestrated by the Disintegrin Metalloproteinase ADAM15 in Synovial Fibroblasts

Mechanotransduction is elicited in cells upon the perception of physical forces transmitted via the extracellular matrix in their surroundings and results in signaling events that impact cellular functions. This physiological process is a prerequisite for maintaining the integrity of diarthrodial joints, while excessive loading is a factor promoting the inflammatory mechanisms of joint destruction. Here, we describe a mechanotransduction pathway in synovial fibroblasts (SF) derived from the synovial membrane of inflamed joints. The functionality of this pathway is completely lost in the absence of the disintegrin metalloproteinase ADAM15 strongly upregulated in SF. The mechanosignaling events involve the Ca2+-dependent activation of c-Jun-N-terminal kinases, the subsequent downregulation of long noncoding RNA HOTAIR, and upregulation of the metabolic energy sensor sirtuin-1. This afferent loop of the pathway is facilitated by ADAM15 via promoting the cell membrane density of the constitutively cycling mechanosensitive transient receptor potential vanilloid 4 calcium channels. In addition, ADAM15 reinforces the Src-mediated activation of pannexin-1 channels required for the enhanced release of ATP, a mediator of purinergic inflammation, which is increasingly produced upon sirtuin-1 induction.


Introduction
Chronic inflammation in immune-mediated inflammatory joint diseases is perpetuated by immune cells and tissue-resident fibroblasts in the synovial membrane, which is a specialized connective tissue that lines the inner surface of the capsule of diarthrodial joints [1]. These synovial fibroblasts (SF) develop an aggressive phenotype characterized by an increased apoptosis resistance, a proteolytic attack on extracellular matrix (ECM) components, and infiltrative growth into cartilage and bone as well as the activation pro-inflammatory pathways [2]. Biomechanical loading is an important factor controlling site-specific localization of inflammation and tissue damage, to which activated SF considerably contribute to the inflammatory processes [3,4].
Synovial fibroblasts that are subjected to mechanical loading during the movement of joints perceive transmitted mechanical forces via their ECM receptors, e.g., integrins [5,6]. Focal adhesions, which contain the integrin receptors in their outer layer, provide anchorage to the ECM and transduce mechanical information [7]. Mechanosensitive ion All patients had given written informed consent and the study was approved by an ethic votum from the Ethics Committee of the Jena University Hospital. All patients met the American College of Rheumatology 1987 criteria for rheumatoid arthritis (RA) (Arnett FC, et al., the American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis (RA). Arthritis Rheum 1988; 31:315-24). The patients (aged between 45 and 71 years) had established RA of >3 years' duration, with an erosive disease course. The majority of patients (70%) were positive for rheumatoid factor and/or anti-cyclic citrullinated peptide (ACPA). The patients were treated with conventional disease-modifying antirheumatic drugs (DMARDs), such as leflunomide or methotrexate. In accordance with the recommendations of the German Society for Orthopedic Rheumatology, therapy with DMARDs was tapered off prior to joint surgery. Thus, at the time of obtainment of the synovial tissue, none of the patients were receiving antirheumatic treatment, with the exception of prednisolone and/or nonsteroidal anti-inflammatory drugs, at a dosage of ≤10 mg/day. Synovial fibroblasts were isolated and grown in DMEM (Gibco; Thermo Fisher, Schwerte, Germany, #41965-039) supplemented with 10% fetal bovine serum (FBS) (Gibco; #10500-064), 1% L-glutamine (Gibco; #15030-024), 100 units/mL penicillin and 100 µg/mL streptomycin (Gibco; #15140-122) at 37 • C in 5% CO 2 . SF were isolated and characterized, as described in detail earlier [31], and expressed the fibroblast-specific marker proteins cadherin-11 and FAP (fibroblast-activating protein) (Supplementary Figure S1). Fibroblasts between passages 3 and 6 were used for experiments. The chondrocyte cell line T/C28a4 permanently transfected with full-length ADAM15, a deletion mutant without the cytoplasmic domain, or an empty vector were cloned, generated and grown in DMEM/10% FCS, as described in detail previously [17].

Cyclic Biaxial Tensile Strain
For the application of cyclic tensile strain, the Flexcell FX-3000 Tension System (Flexcell International Corp, Hillsborough, USA) was used, which is a computer-based system that uses a vacuum to mechanically strain cells adhering to flexible silicone membranes. A controlled vacuum is applied to a loading station, into which four 6-well culture plates are mounted. SF (3.5 × 10 5 cells/well) were grown in BioFlex ® culture plates coated with type I collagen (Flexcell; BF-3001C) for at least 48 h and were then subjected to continuous mechanical stimulation with an equibiaxial sinusoidal waveform at an elongation of 15% and a frequency of 1 Hz for various time points at 37 • C in 5% CO 2 . Unstimulated cultures were grown under the same conditions but without the straining protocol. Cells were harvested by scraping and used for Western blot and qPCR analysis, as well as NAD+, ROS and ATP assays.

ArrayStar LncRNA Array
SF (3.5 × 10 5 cells/well) grown in BioFlex/type I collagen plates for 48 h were mechanically strained for 3 h, and total RNA was isolated using the RNeasy kit from Qiagen (#74104). Then, 2 µg of DNase I-treated RNA was reverse-transcribed using the rtStar™ First-Strand cDNA Synthesis Kit from ArrayStar Inc, Rockville, MD, USA (#AS-FS-001). cDNA was amplified in 384-well PCR plates using the nrStar™ Human Functional LncRNA PCR Array from ArrayStar (#AS-NR-004-1), according to the manufacturer's instructions, in an ABI ViiA™ 7 cycler (Thermo Fisher Scientific). Normalization and subsequent data analysis were performed using software provided by ArrayStar Inc.

NAD+ Assays
NAD+ was measured using NAD/NADH-Glo detection assay (#G9071) from Promega, according to the manufacturer's instructions. Briefly, after cell harvesting by scraping and centrifugation in the microcentrifuge 5424 from Eppendorf (Hamburg, Germany), SF (17.5 × 10 3 cells/50 µL) were lysed with 50 µL of 0.2 M NaOH solution containing 1% dodecyl trimethylammonium bromide (DTAB) (Sigma-Aldrich; #D8639). Then, 50 µL of the lysate was transferred to a new tube and acidified with 25 µL of 0.4-M HCl solution. After incubation for 15 min at 60 • C, resulting in the degradation of NADH, samples were neutralized with neutralization buffer included in the assay (25 µL) and incubated with NAD/NADH-Glo detection reagent (100 µL) in white 96-well plates (Greiner Bio-One; Frickenhausen, Germany) for 30 min at room temperature. The luminescence was measured using the Mithras LB940 plate reader with a 10-second exposure time (Berthold Technologies, Bad Wildbad, Germany).

Reactive Oxygen Species (ROS) Assays
ROS were measured using the ROS-Glo H2O2 Assay (#G8820) from Promega. Briefly, after cyclic straining, cells were scraped together, and 5 × 10 3 cells (in 80 µL) were transferred to white 96-well plates (Greiner Bio-One) and incubated with 20 µL of a 25 µM H 2 O 2 substrate solution for 3 h at 37 • C. Then, 100 µL of ROS-Glo detection solution was added and incubated for 30 min, and luminescence was measured.

ATP Assays
ATP was measured using the ATP detection kit from Abcam; Cambridge, UK (ab113849). Briefly: after the stimulation, cells (0.35 × 10 6 ) were harvested in 1 mL of PBS, and 100 µL of either supernatant or cell suspension was incubated with a detergent solution (50 µL) in a 96-well white plate for 5 min on an orbital shaker. Substrate solution (50 µL) was added, incubated for 5 min in the dark, and luminescence was immediately measured. A standard curve served as a template for the calculation of ATP concentrations.

Preparation of Nuclear Fractions
After harvesting by scraping and centrifugation, cells were lysed in 20 mM HEPES, pH 7.2, 150 mM NaCl, 0.5% NP40, containing proteinase (Roche Diagnostics; 40 µL/mL lysis buffer) and phosphatase inhibitor cocktails (Roche Diagnostics; 10 µL/mL lysis buffer) for 20 min on ice with intermittent mixing, and centrifuged at 13,000 rpm for 30 sec at 4 • C. The pellet containing the nuclear fraction was washed twice with the lysis buffer and protein concentration was determined using the BCA protein assay from Pierce/Thermo Fisher.

Enrichment of Plasma Membrane by Cell Surface Biotinylation
Synovial fibroblasts (4 × 10 6 ) were grown to subconfluency in Petri dishes (10 cm diameter) for 48 h. Before biotinylation, the medium was aspirated, and the cells were washed twice with PBS. Cells were then incubated with the membrane-impermeable EZ-Link Sulfo-NHS-LC-LC-Biotin from Pierce (0.1 mg/mL in PBS, pH 8.0; #21338) for 15 min at room temperature. Cells were washed again twice with PBS and fixed with 1% paraformaldehyde (PFA) in PBS for 5 min at room temperature, then quenched with 2.5 M glycine for 5 min. After harvesting by scraping and microcentrifugation, cell pellets were lysed in 10 mM HEPES, pH 7.0, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, containing complete proteinase inhibitor cocktail (Roche Diagnostics; 40 µL/mL) for 20 min on ice and ultrasonicated. Cell debris was removed by centrifugation at 13,000 rpm for 5 min at 4 • C, and the supernatant was transferred to a new tube. Streptavidin-conjugated magnetic beads (10 µL beads/500 µg cell lysate; Thermo Fisher; #88816) and 1U benzonase nuclease (Merck; #70664) were added to the supernatant and incubated under rotation for 2h at room temperature. The magnetic beads were washed thrice with PBS/0.1% Triton X-100. In order to revert PFA-crosslinked protein complexes, the beads were boiled for 45 min at 80 • C in 20 mM Tris/HCl, pH 6.8, 0.5% SDS, 10% Glycerin v/v, 0,1% bromophenol blue w/v, and 1% β-mercaptoethanol.

Statistical Analysis
Statistical significance was determined using Student's t-test when comparing mean values (calculated from triplicate or quadruplicate measurements) from stimulated versus control conditions. The Wilcoxon signed-rank test was used for the comparison of one data set of measured mean values from different individual donors under stimulation, versus the matched data set from the donors under non-stimulated control conditions. P values are indicated as follows: * p < 0.05; ** p < 0.005; *** p < 0.0005.

Downregulation of lncRNA HOTAIR by Mechanical Strain Is Critically Dependent on ADAM15
SF from 4 different donors, pretreated with either specific siRNA for ADAM15 or non-silencing control siRNA were strained for 3 h. Subsequently, transcribed RNA was amplified using Arraystar lncRNA qPCR plates coated with primers for 372 diseaserelevant lncRNAs, and the overall top 20 up-/downregulated lncRNAs from all 4 donors were determined ( Figure 1A,C). Intersections of all differentially expressed lncRNAs (≥2fold up-regulated) revealed a total of 17 upregulated lncRNAs in synovial fibroblasts from 3 out of 4 of donors, e.g., EGOT, Novlnc76, and MACROD2, but not a single candidate was upregulated in all 4 donors ( Figure 1B), indicating some donor-dependent heterogeneity of mechanically upregulated lncRNAs. By contrast, the intersections of all lncRNAs downregulated by ≤2-fold revealed 2 lncRNAs, i.e., H-19 and HOTAIR, in all 4 donors with~4-fold downregulation ( Figure 1D), identifying HOTAIR as a key candidate regulated by mechanical force in the presence of ADAM15.

Strain-Induced SIRT1 Upregulation via ADAM15-Mediated Downregulation of HOTAIR
The validation of mechano-induced HOTAIR downregulation was performed by qPCR in SF from 7 different donors. GAPDH-normalized Ct values revealed that HOTAIR was only downregulated in ADAM15-expressing cells, whereas SF with silenced ADAM15 did not show any change in HOTAIR levels after cell straining for up to 9 h, shown in SF from one representative donor (Figure 2A  Next, we analyzed the mRNA and protein expression of SIRT1, a gene target of HOTAIR, under the abovementioned conditions. The quantification of SIRT1 showed increasingly higher mRNA and protein levels of up to 4-fold in ADAM15-expressing versus non-expressing SF ( Figure 2C,D), with increased SIRT1 expression in both nuclear and cytoplasmic fractions ( Figure A1). Moreover, HOTAIR silencing of SF unexposed to tensile strain resulted in a~3-fold increase in SIRT1 mRNA and protein levels ( Figure 2E,F), demonstrating that HOTAIR directly affects SIRT1 expression, which is in line with the notion that strain-induced SIRT1 upregulation is directly mediated by the ADAM15dependent downregulation of HOTAIR.

Impact of ADAM15 and SIRT1 on Histone Acetylation, ROS and NAD+
Tensile strain did not induce any change of histone acetylation in nuclear fractions of SF with downregulated ADAM15; however, acetylated histone was reduced by~3-fold in ADAM15-expressing SF after 6 and 9 h of strain ( Figure 3A). Correspondingly, ROS levels were significantly decreased by~2-fold and, in parallel, NAD+ levels increased by~2-fold in ADAM15-expressing SF ( Figure 3B,C).
As a confirmation that the ADAM15-elicited effects on ROS and NAD+ are mediated by SIRT1, the ROS and NAD+ assays of SIRT1-silenced SF revealed significantly increased ROS and decreased NAD+ levels, compared to SIRT1-expressing SF ( Figure 3D). Likewise, the inhibition of SIRT activity by the specific inhibitor selisistat resulted in significantly increased ROS and decreased NAD+ levels ( Figure 3E). Together, these data clearly show an impact of ADAM15 on SIRT1-mediated functions in mechanically strained SF.

Impact of JNK on ADAM15-Dependent Mechano-Signaling in HOTAIR/SIRT1 Regulation
Mechanical strain strongly enhanced phosphorylations of Src at Y416, its target phosphorylation site Y861 FAK, and JNK in ADAM15-expressing SF ( Figure 4A). In addition, co-incubation with the Src inhibitor dasatinib or JNK inhibitor SP600125 during 6 and 9 h of strain showed the substantial inhibition of Src/FAK and JNK phosphorylation by their respective inhibitors ( Figure 4B). qPCR analysis revealed that dasatinib does not affect the strain-induced regulation of HOTAIR or SIRT1; however, SP600125 completely abolished the strain-induced downregulation of HOTAIR ( Figure 4C), and concomitant upregulation of SIRT1 mRNA levels ( Figure 4D), thus revealing the critical role of JNK signaling in ADAM15-dependent HO-TAIR/SIRT regulation under mechanical strain.

Mechano-Induced Activation of TRPV4 and CAMK Upstream of JNK
Next, we investigated whether upstream calcium signaling effectors, such as CAMKs, the calcium channel TRPV4, and Ca 2+ -binding calmodulin (CaM) influence the detected, JNK-mediated HOTAIR/SIRT1 regulation. The selective inhibition of TRPV4 by GSK2193874 [32], CAMKK2 by STO-609 [33], CAMKII by KN-93 [34], or calmodulin by TFP [35] all blocked the mechano-induced downregulation of HOTAIR, and even caused its upregulation to various degrees ( Figure 5A). Correspondingly, SIRT1 mRNA and protein levels were significantly downregulated by all inhibitors (Figure 5B,C), indicating that HOTAIR/SIRT1 regulation is dependent on the activity of candidate effectors of mechano-induced calcium signaling. In addition, the specificity of these inhibitors on strain-induced c-jun/JNK phosphorylations revealed inhibition of >95% by STO-609 and KN-93, and~75% by GSK2193874 and TFP, and no inhibition of the other MAP kinases, ERK1/2 and p38 ( Figure 5D).
Correspondingly, the mechano-induced effects on NAD+ levels (upregulated 3-fold) and parallel measured ROS levels (downregulated 2-fold) were completely blocked by all 4 inhibitors ( Figure 5E,F), indicating that strain-induced SIRT1 upregulation involves the sequential activation of TRPV4 and CAMKs, finally leading to JNK-mediated HOTAIR downregulation.

Impact of ADAM15 and Calcium Signaling on Strain-Induced ATP Release
Next, we investigated SIRT1-associated effects on mechano-induced ATP production and release. When ADAM15 was expressed, mechanical strain significantly induced ATP release, by~7 fold from 26.4 nM to 195.6 nM (calculated median from 7 different donors), whereas only minor ATP release was detectable in ADAM15-silenced SF ( Figure 6A). In addition, mechanical strain did not influence the total ATP levels in ADAM15-expressing SF but reduced total ATP levels by~35% in ADAM15-silenced SF ( Figure 6B). Likewise, the inhibition of the TRPV4 channel, CaM, JNK or SIRT1 activity by their respective inhibitors completely blocked mechano-induced ATP release, and also inhibited total ATP levels by 40% ( Figure 6C,D), indicating the importance of ADAM15 and calcium signaling molecules in mechano-induced ATP release. In addition to known pro-angiogenic and pro-inflammatory effects, the released ATP may also operate as an autocrine stimulator of ADAM15 expression by SF in a positive feedback loop, showing upregulated signal intensities for the ADAM15 protein band upon 48 h of stimulation with ATP-γ-S ( Figure A2).

PANX1 Activity Is Controlled by ADAM15
Next, we investigated whether mechano-induced ATP release involves an ADAM15dependent activation of the ATP export channel PANX1. SF exhibited markedly enhanced, persistent phosphorylation of PANX1 and Src for up to 9 h strain, compared to ADAM15silenced SF ( Figure 7A). As Src has been shown to activate PANX1 by phosphorylating Y198 [36], SF were strained in the presence of the tyrosine kinase inhibitor dasatinib, which resulted in the complete inhibition of phosphorylations of Src at Y416 and PANX1 at Y198 ( Figure 7B). In addition, the inhibition of Src and PANX1 by dasatinib and carbenoxolone [37], respectively, significantly inhibited strain-induced ATP release to the basal level of unstimulated cells, without altering total ATP levels ( Figure 7C,D). To confirm the direct impact of ADAM15 on PANX1-triggered ATP release, both HOTAIR and ADAM15/HOTAIR were silenced by siRNAs. The single knockdown of HOTAIR in ADAM15-expressing strained SF resulted in a significantly increased ATP release. This is likely due to SIRT1-upregulation as a consequence of complete HOTAIRsuppression, as controlled by qPCR (data not shown), which clearly exceeds the ADAM15mediated regulatory effect imposed by mechanical force alone. However, on the one hand, a double knockdown of ADAM15/HOTAIR resulted in a significant reduction of ATP release to the low levels measured under the conditions of single ADAM15 knockdown ( Figure 7E), and, on the other hand, revealed the highest total ATP levels induced by mechanical strain ( Figure 7F). Together, our data clearly show both a strain-induced increase in ATP-production via ADAM15/HOTAIR-mediated SIRT1 upregulation, as well as an independent activating impact on the ATP release channel PANX1 by ADAM15, which in the case of its compromised expression leads to an impaired ATP release.

Binding of ADAM15 to TRPV4 Is Critical for Its Membrane Localization
Since ADAM15 and TRPV4 are both membrane-integrated molecules, our further studies investigated the hypothesis of their direct interaction. Co-immunoprecipitations (IP) using either ADAM15-or TRPV4-specific antibodies reveal the binding of both proteins in ADAM15-expressing SF ( Figure 8A). IPs from T/C28a4 cell lines transfected with full-length ADAM15 (814 amino acids,~100 kDa) or a deletion mutant lacking the cytoplasmic domain (100 amino acids,~10 kDa) show that the co-precipitation of TRPV4 with ADAM15 depends on the presence of its cytoplasmic domain ( Figure 8B). Accordingly, im-munofluorescence stainings of SF demonstrate the colocalization of ADAM15 and TRPV4 in the foci at the cell membrane ( Figure 8C). Moreover, TRPV4 detection was confined to enriched cell membrane preparations of ADAM15-expressing SF, while remaining at the detectability threshold in membranes from ADAM15-silenced cells ( Figure 8D), indicating that interaction between ADAM15 and TRPV4 is crucial for the cell membrane retention of the calcium channel. A graphic summary of our results is shown in Figure 8E. in ADAM15-expressing cell membranes only. Co = non-biotinylated cell lysates, purified on streptavidin magnetic beads, served as the background control (N). Representative results out of three independent experiments are shown. (E) Diagram of summarized results: cyclic strain results in ADAM15-mediated activation of JNK, the downregulation of HOTAIR and subsequent upregulation of SIRT1, leading to decreased ROS, increased NAD+ levels and ATP release. In parallel, ADAM15-mediated Src activation results in the phosphorylation of PANX1, thereby activating PANX1-mediated ATP release. The interaction with ADAM15 inhibits the constitutive cycling of the mechanosensitive calcium channel TRPV4 to the endosome (dashed double arrow). The knockdown of ADAM15 not only blocks mechanical force-induced JNK-and HOTAIR-dependent upregulation of SIRT1 completely but also eliminates all respective downstream effects on NAD+, ROS and ATP, including its release as a purinergic mediator of inflammation. ECM-extracellular matrix.

Discussion
In the present study, we identified a strict ADAM15-dependent effect on the biomechanical regulation of the energy sensor SIRT1 and the underlying mechanisms via lncRNA HOTAIR and Ca 2+ -signaling effectors, culminating in the release of ATP in synovial fibroblasts. These cells are key players in tissue invasion and destruction in inflammatory joint diseases through the proteolytic degradation of extracellular matrix proteins in cartilage and bone, eventually leading to the destruction of articular joints [3,38]. Whereas pressure forces in the joint are predominantly perceived via osteocytes in bone and chondrocytes in cartilage [39], the synovial tissue is also exposed to mechanical strain by traction forces in the joint capsule and synovial fluid shear stress during joint movement. To mimic the biomechanical forces in the synovial membrane, we selected an in vitro model using the Flexcell ® device for the application of controlled cyclic tensile strain on SF, to study the impact of ADAM15 on induced mechanotransduction.
We identified that ADAM15 is crucial for the mechano-induced downregulation of lncRNA HOTAIR, which then results in an up-regulated expression of SIRT1. Accordingly, HOTAIR has also been described as a mechanosensitive lncRNA, whose repression in stretched aortic valve cells led to an increased expression of calcification genes [40]. The reciprocal upregulation of SIRT1 upon HOTAIR downregulation has been described in studies on the impact of the hepatitis C virus core protein on hepatocyte metabolism [41]. Studies on hepatic insulin resistance also reported an inhibitory effect of upregulated HOTAIR on SIRT1 expression [42]. Thus, the regulatory impact of HOTAIR on SIRT1 seems to be complex and likely context-and/or cell-type-dependent, as illustrated by its potential for SIRT1 upregulation via sponging miRNA34a, which is associated with cardio-protective effects in a murine cardiomyopathy model [43].
In RA synovium, SIRT1 was shown to be upregulated and associated with proinflammatory cytokine production and apoptosis resistance [26]. In this respect, our investigations provide additional new mechanistic insights into mechano-induced SIRT1 upregulation in SF regarding the crucial dependency of ADAM15. Moreover, as ADAM15 is known for various anti-apoptotic effects on synovial fibroblasts [16,18], its newly elucidated impact on mechanically regulated SIRT1 may complement the already revealed spectrum of mechanisms with a modulatory effect on deacetylase activity. The tumor suppressor p53 is a well-studied SIRT1 target [44], whose inactivation by deacetylation causes increased apoptosis resistance to oxidative and genotoxic stress [45,46], thereby likely promoting the aggressive growth of inflamed synovial tissue. Accordingly, the increased invasiveness and cellularity of SF in cartilage and bone erosions has been demonstrated as a consequence of p53 inhibition [47].
A central focus of our investigation was the elucidation of the upstream mechanotransduction pathway; in particular, the molecular interactions with ADAM15. In addition to known ADAM15-mediated Src signaling [16,18,19], the activation of c-jun/JNK, which had already been implicated in the mechanosensing of fibroblasts from other tissues [48][49][50], turned out to be the critical MAPK pathway in the regulation of HOTAIR/SIRT1. Moreover, the described mechanotransduction pathways leading to JNK activation also involve Ca 2+dependent mechanisms [50,51]. Thus, studies on mechanosignaling in endothelial cells via direct force application through α1-integrins uncovered stress-induced displacements in the focal adhesion assembly, associated with instantaneous, localized Ca 2+ influx through TRPV4 channels in the plasma membrane [52,53]. Accordingly, our studies provide unequivocal evidence for the involvement of mechanosensitive TRPV4 channels, linked to the subsequent activation of CAMKs and, finally, to c-Jun/JNK induced in ADAM15expressing SF by cyclic tensile strain. Thus, the triggering of the β1-integrins via tensile forces in the collagen matrix could be localized to focal adhesions in the cell membrane of endothelial cells [53], a site at which ADAM15 expression, co-localizing with FAK, has been demonstrated [16]. Moreover, the involvement of ADAM15 in Ca 2+ -dependent CaM signaling upon Fas receptor stimulation has been shown in SF [18].
Thus, the revealed colocalization in the cell membrane indicates a potential functional link between TRPV4 and ADAM15 in tensile force perception by SF. Accordingly, our co-immunoprecipitation studies provide conclusive experimental evidence for a direct interaction of the two proteins in critical dependency on the cytoplasmic domain of ADAM15, which is a key factor in promoting TRPV4 enrichment in the cell membrane. The expression of ion channels at the cell surface is essential for their activity and downstream cellular functions, as TRPV4 trafficking to the plasma membrane and its internalization by endocytosis is complex and tightly controlled, involving, e.g., the TRPV4-interacting protein PACSIN 3 [54] or PI3K, PKC, and RhoA signaling pathways [55]. Whereas our studies provide unequivocal evidence for ADAM15-dependent TRPV4 membrane localization, resulting in an upregulated mechano-induced activation of CAMK signaling, elucidation of the precise mechanisms of its impact on TRPV4 membrane-targeting is beyond the scope of the present study and presents an area for future investigation.
Simultaneously with the ADAM15-mediated activation of the mechanosensitive TRPV4, a newly uncovered function in mechanotransduction is its modulation of mechanoinduced ATP release through activation of the PANX1 channel by Src. The effector loop of ADAM15-dependent mechanosignaling pathways culminates in the release of ATP as a purinergic mediator, capable of activating a broad spectrum of inflammatory responses (reviewed in [56]). The close proximity of SF to other cells in the synovial tissue, e.g., monocytes/macrophages, dendritic cells, mast cells, and endothelial cells, promotes the pro-inflammatory potential of the released ATP, which is limited by ectonucleotidase activity-dependent metabolization in the extracellular space [56]. However, the effects of ATP are not confined to the stimulation of purinergic receptors involved in inflammasome activation [29] or K ATP channels to induce angiogenesis [57], but instead include the potential for activation of the mannan-binding lectin (MBL) pathway of complement activation by the direct binding of ATP to MBL [58]. The latter aspect is noteworthy as, more recently, mechano-induced complement activation has been described as a mechanism promoting disease chronicity in the experimental mouse model of collagen II antibody-induced arthritis [59].
Moreover, we have shown that ATP-γ-S can upregulate ADAM15 in synovial fibroblasts, thus potentially acting as an autocrine stimulator of ADAM15 expression upon strain-induced ATP release. ADAM15 has also been shown to be upregulated by shear stress via the transcription factor KLF2, thereby promoting the survival of endothelial cells [60]. It is tempting to speculate that the upregulation of ADAM15, triggered by ATP, is a general mechanism that might also occur in other cell types apart from fibroblasts since arterial shear stress could be demonstrated to induce ATP release via the PANX1 channels in human platelets [61]. The positive feedback regulation of ADAM15 expression by ATP is supplemented by the potential of ATP to induce the release of IL-1β [62], a known stimulator of ADAM15 expression [63], via inflammasome activation in neighboring cells. In addition to the release of ATP as a purinergic pro-inflammatory mediator, we also demonstrated an upregulation of the chemokine CCL2 as an earlier-described crucial mediator of mechanoinflammation [3], in mechanically strained SF in strict dependency on ADAM15-regulated SIRT1 (results not shown).
Our elucidation of the critical impact of ADAM15 on the orchestration of mechanoinflammation in SF suggests its potential as a target for therapeutic intervention, which is supported by data on the amelioration of murine collagen-induced arthritis through treatment with ADAM15-specific siRNA [64]. Our investigations reveal the underlying mechanosignaling orchestrated by ADAM15, which exerts cell-adhesive properties to collagen type II [17] and directly interacts with collagen-binding integrins via its disintegrin domain [15], while using cytoplasmic structures to promote the focal cell membrane density of the mechanosensitive TRPV4 channel and the Src-mediated activation of the ATP release channel, PANX1. In this respect, ADAM15 provides a crucial scaffold for mechanosignaling events in synovial fibroblasts. Moreover, its highly upregulated expression in inflammatory diseases, such as rheumatoid arthritis and osteoarthritis, as well as in various cancers [14,63,65], may suggest that this ADAM15-mediated mechanosignaling also occurs in cell types different from (synovial) fibroblasts, contributing to infiltrative growth and the perpetuation of tissue inflammation.
In conclusion, our studies have elucidated a novel and crucial ADAM15-dependent mechano-inflammatory pathway in synovial fibroblasts, which, due to its positive feedback regulation and well-established connection to mechanosensing focal adhesions, may substantially contribute to fueling inflammatory processes.

Institutional Review Board Statement:
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of the university hospital, Jena, Germany (3951-12/13).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest. Appendix A Figure A1. SIRT1 distribution in nucleus and cytoplasm. Immunoblots of whole-cell lysates (wcl), lysates from isolated nuclei (10 µg) and cytoplasm (20 µg) from RASFs, showing increased SIRT1 levels in both nuclear and cytoplasmic fractions after straining for 0-9 h.