Identification of Adamts4 as a novel adult cardiac injury biomarker with therapeutic implications in patients with cardiac injuries

Pathological cardiac remodeling as an aftermath of a severe cardiac injury can lead to ventricular dysfunction and subsequent heart failure. Adamts4, a metalloproteinase, and disintegrin with thrombospondin-like motif, involved in the turnover of certain extracellular matrix molecules and pathogenesis of osteoarthritis, also plays a role in cardiac remodeling although little is presently known about its expression and function in the heart. Here, we have investigated the dynamic expression pattern of Adamts4 during cardiogenesis and also in the adult heart. To our surprise, adult cardiac injury reactivated Adamts4 expression concomitant with fibrosis induction. To better understand the mechanism, cultured H9c2 cardiomyocyte cells were subjected to ROS injury and Hypoxia. Moreover, through combinatorial treatment with SB431542 (an inhibitor of Tgf-β1), and Adamts4 siRNA mediated gene knockdown, we were able to decipher a regulatory hierarchy to the signal cascade being at the heart of Tgf-β regulation. Besides the hallmark expression of Adamts4 and Tgf-β1, expression of other fibrosis-related markers like Collagen-III, alpha-SMA and Periostin were also assessed. Finally, increased levels of Adamts4 and alpha-SMA proteins in cardiac patients also resonated well with our animal and cell culture studies. Overall, in this study, we highlight, Adamts4 as a novel biomarker of adult cardiac injury.

However, there is not much literature about Adamts4 expression and function in the developing and adult heart at the basal level and also in post cardiac injury. Only recently studies have shown that Adamts4 along with Adamts1 levels remain elevated in patients with acute aortic dissection and also in patients with coronary artery disease 14,15 .
Here in our present study, for the first time, we aim to identify Adamts4 as a novel biomarker of adult cardiac injury under stress conditions. We have detected a strong expression of Adamts4 protein in the developing cardiac chamber myocardium in utero compared to very restricted expression in the adult murine hearts. To our surprise, we have also detected the expression of Adamts4 in vivo murine model of myocardial infarction (MI), localized in adult cardiomyocytes. To better understand the molecular insights of Adamts4 induction and associated affected signaling pathway activation, we have used H9c2, a rat ventricular myoblast cell line for several in vitro assays. Likewise, Adamts4 expression was induced in H9c2 cells, subjected to hypoxia (Hyp) and ROS injury inductions in vitro. Moreover, we manipulated the expression of Adamts4 with siRNA-mediated loss of function and TGF-β inhibitor studies in H9c2 cells to evaluate its regulation and dependency on TGF-β signaling since TGF-β has been long known to be a characteristic marker for inflammatory and fibrotic responses following pathological stress including MI, ischemia and reperfusion (I/R) injury [16][17][18][19][20] . Finally and most importantly, we also validated our hypothesis in human clinical samples and demonstrated the induced expression of ADAMTS4 in patients with indicated cardiac ailments.

Results
Dynamic expression pattern of murine Adamts4 protein observed in developing and adult hearts. Immunohistochemistry (IHC) with anti-Adamts4 antibody shows Adamts4 protein expresses strongly in developing murine cardiac chambers but the expression significantly wanes in adult murine hearts. In contrast to the adult heart (Fig. 1e) where Adamts4 expression is largely restricted at the edge of the interventricular septum (IVS) mostly adjacent to the left ventricle (LV), in the embryonic heart ( Fig. 1a-d) the expression of Adamts4 is more widespread throughout the left and right auricles (RA, LA), ventricles (LV and RV) and the inter-ventricular septum (IVS) of the chamber myocardium of E10.5, E12.5, E14.5, and E18.5 although its expression in endothelial-derived heart valves remained mostly inconspicuous. Further, the expression of Adamts4 at E13.5 (Fig. 1f) is also co-localized with cardiomyocyte-specific marker MF20 (MF20 is a myosin heavy chain-II marker in cardiac and skeletal muscle system, hence often used to label sections of myocardium and is therefore considered to be cardiomyocyte marker as well 21,22 .) throughout the LV of chamber myocardium.
Adamts4 reactivation in the ventricular chamber in the adult heart following injury in-vivo. While Adamts4 expression in adult murine hearts significantly waned in comparison to developing hearts, it is found that MI-induced adult murine heart shows significant reactivation of Adamts4 protein. IHC with anti-Adamts4 antibody shows a strong expression of Adamts4 in the infarct and border zone (Fig. 2a) fol- Figure 1. Dynamic expression of Adamts4 protein in embryonic and adult heart. Adamts4 expression is shown in developing E10.5 (a), E12.5 (b), E14.5 (c), E18.5 (d) adult (e) murine hearts. IHC with anti-Adamts4 antibody (green colour) and Topro3 (blue colour) used as nuclear stain, the expression pattern of Adamts4 is observed in (a)-(e). While the expression pattern of Adamts4 is more widespread in the embryonic stages throughout the RA, LA, RV, LV and IVS (a-d), the expression drastically reduces and is mainly only confined at the IVS in the adult stage (e) displaying a sharp contrast between the expression of Adamts4 in developing and adult stages. (f) Highlights the colocalization of Adamts4 (green) in the chamber myocardium with E13.5 chamber cardiomyocytes shows colocalization of Adamts4 with MF20 (shown in yellow colour) across the LV of chamber myocardium. The arrowheads point towards cells that have colocalized with both the markers ie, Adamts4 and MF20. This confirms the expression of Adamts4 in embryonic cardiomyocytes. n = 4. Adamts4, Tgf-β1 upregulation in H9c2 cells following injury induction. To better understand the detailed mechanistic insights into Adamts4 function in post injury; further studies were done in H9c2, a rat cardiomyocyte cell line. H9c2 cells were subjected to H 2 O 2 and Hypoxia (Hyp) treatment. Figure 3a, b show upregulation of ROS and hypoxia injury-related markers Catalase [23][24][25] and Hif-1α 26 respectively validated by real-time qPCR, showing upregulation after injury for both the markers. Also, Adamts4 and Collagen-III 27,28 expression at mRNA levels shows elevation after both stress treatments (Fig. 3c, d). Furthermore, Tgf-β1 mRNA shows upregulation following H 2 O 2 and Hypoxia treatment 18,29 as quantified by real-time PCR (Fig. 3e). Moreover, Adamts4, α-SMA and Vimentin (a type-III intermediate filament marker that is known to induce repair mesenchymal cells following severe stress or injury to cells to initiate wound healing and provide structural support network 30,31 ) expression at protein levels assessed by western blot (WB) shows significant upregulation following H 2 O 2 and Hypoxia treatments ( Fig. 3f-i). Figure 3j depicts H9c2 morphology after injury induction as compared to control taken in DIC mode and cell size measurements post injury induction is shown in Supplementary  Fig. S1. Overall, this figure shows the upregulation of ECM markers like Col-III, α-SMA, Vimentin along with Adamts4 and Tgf-β1 following H 2 O 2 and Hypoxia treatment. All qPCRs were normalized with β-actin and total protein was used as loading control for western blot assays.
Adamts4 and Tgf-β1 expression is suppressed by SB431542 (ALKI) pre-treatment before H 2 O 2 and Hypoxia stress induction. Immunofluorescence (IF) staining with anti-Adamts4 and anti-Tgf-β1 antibody shows an elevated expression of both the markers following H 2 O 2 and Hypoxia treatment assessed by quantifying fluorescence mean intensity (via ImageJ software, NIH). Interestingly, this elevated expression significantly reduced for both Adamts4 and Tgf-β1 in presence of ALKI pre-treatment. Interesting to note, a non-specific nuclear presence of Adamts4 is observed irrespective of treatment ( Fig. 4a-c) Further, the WB also shows a similar trend in the expression pattern of Adamts4 and Tgf-β1 proteins where the expression of both markers falls after ALKI pre-treatment in comparison to only injury groups ( Fig. 4 d, e, g and h). Also, Tgf-β1 measured at mRNA levels by qPCR also validates the same finding (Fig. 4f). Overall, ALKI shows successful significant inhibition of Tgf-β1 as also stated previously 20,32,33 along with inhibition of Adamts4. This finding is indicative of Tgf-β1 dependent activation of Adamts4. ALKI pre-treatment prior to H 2 O 2 and Hypoxia treatment inhibits Periostin expression. Periostin expression is detected by IF staining with anti-Periostin antibody and shows upregulation of the same following H 2 O 2 and Hyp treatments group in comparison to the control set but this elevation is significantly reduced in ALKI + H 2 O 2 and ALKI + Hyp groups respectively in comparison to the control group (Fig. 6a, b). Adamts4 loss of function mediated by Adamts4 siRNA transfection (ATSsi Tr) was validated by WB (Fig. 6c, d), IF (Fig. 6e, f) and qPCR (Fig. 6g), all of which validated successful knockdown of Adamts4 compared to the scrambled siRNA (Scsi Tr) treated group. Overall, these data suggest that Periostin activity may be Adamts4 and Tgf-β1 dependent.
Adamts4 siRNA mediated gene knockdown before H 2 O 2 and Hypoxia treatment inhibits Adamts4 but does not affect Tgf-β1 expression. Now to better understand the regulatory hierarchy and interaction between Tgf-β signaling and Adamts4; Adamts4 knockdown experiments are performed. Tgf-β1, α-SMA and Vimentin indicated development of injury related fibrosis. ß-actin was used to normalize gene expression for qPCR assay and total protein was used as loading control for WB. n = 3. Data analyzed and expressed as mean ± SD. Differences were considered statistically significant for p < 0.05. (j) H9c2 morphology after injury induction as compared to control taken with the help of a DIC microscope.    The induced expression of Adamts4 and α-SMA in humans with cardiac anomalies. To better correlate our in vivo murine animal model and in vitro findings, in humans, we have used human patients' serum samples with indicated cardiac anomalies. Adamts4 and α-SMA protein levels were assessed by western blot. Both Adamts4 and α-SMA proteins show significantly enhanced expression in patients who have suffered DCM, MI, either AWMI (anterior wall MI) or inferior wall MI (IWMI) (Fig. 10a-c) Furthermore, Adamts4 specific ELISA also validates the same findings (Fig. 10d). Patients with DCM, IWMI or AWMI show significant expression of Adamts4 as compared to the control group. This confirms that Adamts4 is also induced in humans who have a history of cardiac diseases like DCM or have suffered MI. Figure 10e shows a proposed model based on our findings for the interaction and inter-relationship between Adamts4 and Tgf-β1 in post cardiac injury conditions; when the Tgf-β1 expression is inhibited by blocking the binding of Tgf-β1 to one of its two binding receptors (ALK receptors, more specifically ALK 4 and 5), Adamts4 expression is inhibited since it is regulated downstream of Tgf-β1. So, the elevated fibrosis like conditions that escalated following pathological stress induction is inhibited and eventually fibrosis related markers (Col-III, α-SMA and periostin) along with Adamts4 are downregulated following inhibition of Adamts4.

Discussion
The salient finding of this study is that Adamts4 is upregulated in injury to cardiac muscle. This finding is consistent with previous studies where Adamts4 expression was shown to be enhanced in atherosclerotic plaques 13 and patients with acute coronary syndrome 15 . Importantly, Adamts4 being a secretory protein in nature, it resides both intracellularly and is also secreted from cardiomyocytes out into the ECM where its function is to regulate the turnover of other proteins by directly affecting other resident cells including but not limited to cardiac fibroblasts, endothelial and smooth muscle cells in the same ECM milieu. However, injury induced function of Adamts4 in other adult cardiac resident cell types is unknown. The expression of Adamts4 significantly DAPI (shown in blue) was used as nuclear stain. Finally, from qPCR Adamts4 mRNA levels were downregulated from onefold observed for Scsi tr set to 0.25-fold for ATSsi tr group (g). n = 3, data analyzed and expressed as mean ± SD. Differences were considered statistically significant for p < 0.05. www.nature.com/scientificreports/ decreased from being widespread in the chamber myocardium of developing hearts to very restricted expression in normal adult hearts (Fig. 1) which could imply that its roles in developing and adult hearts are different but the reactivation of Adamts4 following cardiac injury (MI) (Fig. 2) implied that it could be another marker for adult cardiac injury. Since, this expression co-localized with cardiomyocytes, manipulation of Adamts4 was easier to experiment with within H9c2, a cardiomyocyte cell line. Injury inductions through H 2 O 2 and hypoxia treatment showed upregulation of Adamts4 along with a couple of other fibrosis-related markers like Tgf-β1, Collagen-III, α-SMA, and Periostin which is notably known as a marker for detecting fibroblast to myofibroblast switch 36 were found to be elevated (Figs. 3 and 6) implying that an injury mediated ECM remodeling could be a cause for the elevation of these markers. Further pre-treatment with SB431542, an inhibitor of ALK4 and 5 (one of the two binding receptors of Tgf-β1) receptor that eventually leads to inhibition of Tgf-β1 also showed inhibition of Adamts4, this inhibition further extended to the expression of Collagen-III, α-SMA and Periostin proteins (Figs. 4, 5 and 6). To better understand the hierarchy supremacy between Adamts4 and Tgf-β1, Adamts4 loss of function study was mediated by Adamts4 siRNA transfection. In groups where Adamts4 knockdown was performed before stress induction, Tgf-β1 expression remained quite unaffected whereas the expression of the other 3 mentioned markers-Collagen-III, α-SMA, and Periostin along with Adamts4 was found to reduce to somewhat similar levels when ALKI pre-treatment with stress induction was done (Figs. 7, 8 and 9). These findings indicated that Adamts4 expression is mediated by Tgf-β1 and also that other ECM and fibrosis related markers including Col-III, α-SMA, and Periostin seem to be regulated by Adamts4 since loss of function of Adamts4 significantly inhibited the expression of these markers following injury to H9c2 cells. Finally, not only our marker of interest, Adamts4 but also α-SMA showed significantly enhanced expression in patients with MI and DCM injury (Fig. 10) indicating a cardiac fibrosis like condition following adult cardiac injury. Although . DAPI (shown in blue) was used as nuclear stain. Differences between groups H 2 O 2 and ATSsi + H 2 O 2 , and Hyp and ATSsi + Hyp were not found to be significant. n = 3, data analyzed and expressed as mean ± SD. Differences were considered statistically significant for p < 0.05. www.nature.com/scientificreports/ not done here but our future studies are focussed on validating the direct interaction of Adamts4 with Col-III α-SMA, and Periostin through binding/ChIP assays. The ECM is an integral part of the myocardium. The dynamic composition of cardiac ECM consisting of both structural and non-structural components plays a critical role in cellular events and during pathogenicity 37,38 . Under stress conditions such as Ischemia or MI, the chamber myocardium undergoes intensive ECM remodeling which remains relatively inconspicuous in healthy individuals. Cardiac Fibrosis is often viewed as an expansion of ECM remodelling 39 . Following MI or I/R, there is an acute inflammatory response that suffices for the overexpression of pro-inflammatory cytokines like Tgf-β and interleukins. Tgf-β1 most notably does so by the canonical Smad 2/3 signaling cascade which leads to fibrosis 16,17,39 . This, in turn, activates and synthesizes matrix macromolecule like Adamts 40,41 family among others as a part of the pro-inflammatory mediated cell signaling cascade and these MMPs then take over the centre stage post-cardiac injury such as MI which leaves behind a pool of necrotic myocytes, it is then that the MMPs like Adamts4 takeover to regulate a turn-over in the synthesis of matrix macromolecules like Collagen-I/III, α-SMA, and Periostin, Tenascin-C to synthesize more matrix macromolecules and fibroblasts which is required to fill in the scar left by necrosis of myocardial cells in order to maintain the physiology of the myocardium since adult cardiomyocytes have very limited proliferation capacity. However, prolonged expansion of ECM could lead to extensive fibrosis and as a resulting stiffness of the myocardium following which ventricular dysfunction could occur which itself may turn fatal and thus inhibition of MMPs like Adamts4 could be one of the possible targets to reduce fibrosis post-cardiac injury 42 like the one proposed in our model (Fig. 10e) by inhibiting Tgf-β1.
To conclude, Adamts4 is upregulated in response to cardiac stress in in-vivo, in-vitro and human studies, and this upregulation is mediated by Tgf-β1. This elevation of Adamts4 leads to fibrosis induction as markers related  (a and c). DAPI (shown in blue) was used as nuclear stain. n = 3, p < 0.05 is considered as significant for differences among groups. Data analysed and expressed as mean ± SD. www.nature.com/scientificreports/ to fibrosis including Collagen-III, α-SMA, and Periostin along with Tgf-β1 is also found to be upregulated but when Adamts4 loss of function is performed, these markers were also found to be inhibited following injury except for Tgf-β1 which remains unaffected indicating that it is upstream of Adamts4 in the signaling cascade. Our findings suggest that Adamts4 after being activated by Tgf-β1 under stress inducing conditions, mediates ECM remodeling and thereafter fibrosis through the functioning of Periostin and α-SMA as determined by our findings. It is possible that Periostin, another secretory ECM molecule known to either regulate or work in tandem with other MMPs 43,44 , works in synchronisation with Adamts4 to regulate ECM remodeling and induce fibrosis. Our findings show an Adamts4 dependent Periostin functioning (depicted by Adamts4 loss of function assay) but to state whether Adamts4 and Periostin directly interact, requires further experimentation to establish any direct association between these two ECM molecules. α-SMA has been known to induce fibroblast contractility and is highly expressed in infarct myofibroblasts to prevent extreme remodeling and thereafter cardiac rupture 45,46 . As our data shows the hallmark expression of Adamts4 in cardiac stress conditions, it can be considered as a novel biomarker for cardiac related injuries leading way for therapeutics to manipulate its expression following cardiac injury for improvement of cardiac functioning. Finally, our work hypothesises that Adamts4 being an ECM secretory protein, after being secreted from chamber myocardium regulates ECM and its remodeling, the details of this mechanism require further depth, understanding and experimentations.

Methods
Animal study and experimentation. All experiments involving animals were carried out with the experimental protocols and procedures reviewed and approved by the Cincinnati Children's Hospital Medical Center Biohazard Safety Committee and Institutional Animal Care and Use Committee. The study was per- ATSsi + Hyp groups following siRNA mediated knockdown of Adamts4 (a). DAPI (shown in blue) was used as nuclear stain. Its quantification is depicted in the graph (b). n = 3, data analysed and expressed as mean ± SD. Differences were considered statistically significant for p < 0.05. www.nature.com/scientificreports/ formed in accordance with ARRIVE guidelines and all animal experiments were conducted in compliance with the relevant guidelines. Timed-matings were established, with the morning of an observed copulation plug set at E0.5. For embryonic studies, whole embryos or hearts were harvested on E10.5, E12.5, E14.5, and E18.5 days and proceeded with IHC. For MI induction, 8-10 weeks old adult male Swiss albino mice were used. MI was performed as described previously 24 .
Immunohistochemistry. Whole embryos and embryonic or adult hearts were harvested, washed in 1X PBS, fixed in 4% paraformaldehyde, dehydrated, and embedded in paraffin. Later the tissue sectioning (5-7 μm) was done using a microtome (Leica Biosystems). For immunohistochemistry, the tissue sections were deparaffinized and hydrated after which antigen retrieval was done using citrate buffer (pH 6) washed in 1X PBS and thereafter, blocking solution (2% BSA with 0.1% Tween20) was added to the sections for 1 h. Later after blocking,  a and b) was observed. The same was true for α-SMA which showed increased fold change varied from 3.7 to 4.7, 3.3 to 5.3, 3.9 to 5.4 for DCM, IWMI, and AWMI groups in comparison with the control set where fold levels for α-SMA varied from 1 to 1.8. (a and c). Total protein was used as loading control. Further, Adamts4 specific ELISA showed concentration gradient varying from 4666 to 6500 (pg/ml), 5200 to 7333 (pg/ml), and 4000 to 5500 (pg/ml) for DCM, IWMI, and AWMI groups respectively as compared to the concentration gradient observed for the control group which varied from 1533 to 2170 (pg/ml) (d). (n = 5 for WB for each group, N = 10 and n = 2 for ELISA for each study group). Data analyzed and expressed as median, 1st and 3rd quartile and range. p < 0.05 was considered a significant difference. The proposed model (e) shows the hypothetical hierarchy and inter-relationship between Tgf-β1 and Adamts4 based on our findings. ALKI acting as an inhibitor of Tgf-β1 inhibits the stimulation of Adamts4 by Tgf-β1 which normally is activated under stress or injury conditions and activates downstream molecule Adamts4 and Adamts4 then proceeds with pathological ECM remodeling to restore the damaged physiology of the cells and tissues.   Protein isolation and western blotting. Protein isolation was done using ice-cold mammalian lysis buffer (250 mM NaCl, 50 mM Tris pH 7.5, 0.1% SDS, 1% TritonX, and 5 Mm EDTA) containing protease inhibitor cocktail (Genetix GX-2811AR) and phosphatase inhibitor (Genetix GX-0211AR). The proteins were stored in small aliquots at − 80 °C until further use for Western Blotting (WB). Western blotting was performed as previously described 48 . The immunoblots developed using Clarity™ Western ECL substrate (1705060, Bio-Rad) and scanned using ChemiDocMP (Bio-Rad) as previously described 49 . The antibodies used were Adamts4 (PA1-1749A, Invitrogen) used at a concentration of 2 μg/ml, α-SMA (14-976082, Invitrogen) used at a dilution of 1:500, Tgf-β1 (ab64715, Abcam) used at a concentration of 1 μg/ml and Vimentin (ab17321, Abcam) at a dilution of 1:1000. After overnight incubation at 4 °C with these primary antibodies, the corresponding secondary antibodies-HRP conjugated secondary goat anti-rabbit (ab97051, Abcam) and HRP conjugated goat anti-mouse secondary antibodies (ab97023, Abcam) were added against their respective primary antibodies at a dilution of 1:3000. The SDS gels were also stained with 2.5% coomassie (Brilliant blue G, SRL) and destained with coomassie de-stainer to obtain total protein intensities. The coomassie stained gels were scanned using ChemiDoc MP mentioned earlier. Quantification of intensities was done using ImageJ software (NIH). Figures for unprocessed blots and gels are also included in Supplementary Figs. S5, S6, S7 and S8 with highlighted regions (in red box) corresponding to the data depicted in the main manuscript figures. Full-length blots could not be submitted since the blots were trimmed (according to molecular size of the proteins) before hybridisation with respective antibodies to save on the antibodies and reagents.
Immunostaining. Immunostaining was performed in H9c2 cells following the protocol previously described 25 except for the omission of permeabilization steps for staining with antibodies. Adamts4 (PA1-1749A, Invitrogen) used as mentioned earlier, Tgf-β1 (ab 64715, Abcam) used at a concentration of 2 μg/ml, α-SMA (14-976082, Invitrogen) used at 1:400 dilution, collagen III (ab 7778, Abcam) at a dilution of 1:1000 and Periostin (ab14041, Abcam) used at a dilution of 1:500. After overnight incubation at 4 °C with these primary antibodies, the corresponding secondary antibodies Alexa flour 488 goat anti-rabbit secondary antibody (ab150077, Abcam) and Alexa flour 594 goat anti-mouse secondary antibody (ab150116, Abcam) were added against their respective primary antibodies. All cell nuclei were stained with DAPI (D9542, Sigma). Images were acquired using Leica confocal microscope and LasX software. The fluorescence intensities were measured and quantified with the help of ImageJ software (NIH).
Enzyme linked immunosorbent assay (ELISA). ELISA kit (ab213753, Abcam) was used to perform ELISA. Serum proteins were diluted with sample buffer in the ratio of 1:10 and samples were loaded onto the wells in duplicates. ELISA was performed as stated by the manufacturer's protocol.
Statistical analysis. All the results are mostly represented as mean ± Standard Deviation of mean (SD) or as median, quartiles and range (Fig. 10). Statistical analyses were done using student's unpaired two-tailed T-test and one-way ANOVA for more than 2 groups. GraphPad Prism 9.3.1 was used for statistical analysis. Differences among the groups were considered statistically significant for P < 0.05.

Data availability
All the generated data used to support the findings of this study are either included within the article or in the supplementary file.