The Microenvironment of Decellularized Extracellular Matrix from Heart Failure Myocardium Alters the Balance between Angiogenic and Fibrotic Signals from Stromal Primitive Cells

Cardiac adverse remodeling is characterized by biological changes that affect the composition and architecture of the extracellular matrix (ECM). The consequently disrupted signaling can interfere with the balance between cardiogenic and pro-fibrotic phenotype of resident cardiac stromal primitive cells (CPCs). The latter are important players in cardiac homeostasis and can be exploited as therapeutic cells in regenerative medicine. Our aim was to compare the effects of human decellularized native ECM from normal (dECM-NH) or failing hearts (dECM-PH) on human CPCs. CPCs were cultured on dECM sections and characterized for gene expression, immunofluorescence, and paracrine profiles. When cultured on dECM-NH, CPCs significantly upregulated cardiac commitment markers (CX43, NKX2.5), cardioprotective cytokines (bFGF, HGF), and the angiogenesis mediator, NO. When seeded on dECM-PH, instead, CPCs upregulated pro-remodeling cytokines (IGF-2, PDGF-AA, TGF-β) and the oxidative stress molecule H2O2. Interestingly, culture on dECM-PH was associated with impaired paracrine support to angiogenesis, and increased expression of the vascular endothelial growth factor (VEGF)-sequestering decoy isoform of the KDR/VEGFR2 receptor. Our results suggest that resident CPCs exposed to the pathological microenvironment of remodeling ECM partially lose their paracrine angiogenic properties and release more pro-fibrotic cytokines. These observations shed novel insights on the crosstalk between ECM and stromal CPCs, suggesting also a cautious use of non-healthy decellularized myocardium for cardiac tissue engineering approaches.

Accordingly, the aim of the present study was to investigate the effects of human decellularized native ECM from normal or pathological HF hearts on a population of human cardiac stromal cells, including CPCs. Our results will be useful to better understand the influence of pathological ECM remodeling on cardiac stromal populations, also for translational purposes on suitable cell product candidates for heart regeneration.

Results
Decellularized native extracellular matrix sections obtained from normal or pathological hearts with advanced HF (dECM-NH and dECM-PH, respectively) were seeded with human CPCs isolated from IHD patients ( Figure 1A,B) and cultured up to 7 days, with standard culture on fibronectin (FN)-coating as control. Cell viability of CPCs cultured on cardiac dECMs or on FN-coated dishes was quantified daily using a trypan blue exclusion assay ( Figure 1C). Cell death rate 2 days after seeding was significantly lower on both dECMs compared to FN (21.82 ± 1.36%), without any statistically significant difference between dECM-NH and dECM-PH (11.45 ± 1.68% and 11.41 ± 2.32%, respectively). Between 3 and 4 days after cell seeding, cell death rate resulted significantly lower on dECM-PH than FN (2.42 ± 0.75% versus 10.01 ± 0.85% after 3 days, and 2.16 ± 0.73% versus 7.27 ± 0.82% after 4 days). Nonetheless, starting at 5 days after seeding, cell death rate dramatically decreased on all substrates until it reached values as low as 0.4% after 7 days of culture, without any significant differences among dECMs and FN (1.05 ± 0.03% on FN, 0.88 ± 0.02% on dECM-NH, and 0.41 ± 0.36% on dECM-PH after 168 h). Therefore, cell viability increased with time on all substrates with differences that were statistically significant only until 4 days after cell seeding ( Figure 1C).  A comprehensive panel of genes was analyzed by realtime PCR for expression levels in CPCs after 7 days of culture on the different substrates. Hierarchical clustering analysis did not evidence any significant difference among sample groups ( Figure 2A). Nonetheless, analysis on specific genes evidenced a statistically significant increase in the expression of cardiac-specific transcription factor NK2 Homeobox 5 (NKX2-5) and Troponin T (TNNT2) on both dECM substrates compared with FN control ( Figure 2B,C). Upregulation of the gap junction protein connexin 43 (CX43) was significant only on dECM-NH sections ( Figure 2D), while GATA4 and kinase insert domain receptor (KDR, also named vascular endothelial growth factor receptor 2, VEGFR2) were significantly upregulated only on dECM-PH sections ( Figure 2E,F). Other genes analyzed, but not modulated in these conditions, include adhesion and ECM proteins, such as Vinculin and Collagen I (Supplementary Figure S1).
To assess possible differences in the ability of dECMs to direct the phenotype of CPCs, we evaluated the expression of typical markers of cardiomyocytes, endothelial, mesenchymal, and smooth muscle cells by immunofluorescence ( Figure 3). Consistently with real-time PCR data, immunofluorescence analysis revealed that CPCs cultured on cardiac dECM, regardless of its derivation, expressed Vimentin, GATA4, Cardiac alpha-Actin (ACTC1), and Smooth Muscle Actin (ACTA2). However, CX43 and NKX2.5 were detectable by immunofluorescence staining only when CPCs were cultured on dECM-NH, while their expression on dECM-PH was not detected in these conditions. This apparent difference from NKX2.5 gene expression data suggests regulation at post-transcriptional level, as previously described [36]. Conversely, CPCs were immunopositive for KDR/VEGFR2 only when cultured on dECM-PH, consistently with gene expression levels ( Figure 2F). This latter staining, though, was compatible with both intracellular and membrane signal. Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 5 of 18   Nitric oxide (NO, a pro-angiogenic mediator), hydrogen peroxide (H 2 O 2 , a reactive oxygen species), and growth factors release were assessed in 48 hour-conditioned media (CM) from all culture conditions. NO levels were significantly higher in CM from dECM-NH cultured cells, compared to both dECM-PH sections and control substrates ( Figure 4A), while H 2 O 2 concentration was increased in CMs collected from both dECMs versus FN, with higher levels on dECM-PH cultures ( Figure 4B). The comparative screening analysis of the profile of cytokines and growth factors released by CPCs on different substrates was performed by protein array ( Figure 4C). The analysis revealed that lower amounts of basic fibroblast growth factor (bFGF) were released on both dECMs compared to CPCs cultured on fibronectin, with a significant difference also in dECM-PH versus dECM-NH ( Figure 4D). Moreover, CPCs cultured on both normal and pathological cardiac dECM released significantly higher amounts of granulocyte-macrophage colony-stimulating factor (GM-CSF) ( Figure 4E) and hepatocyte growth factor (HGF) ( Figure 4F), this latter being also significantly upregulated in dECM-NH versus PH. Additionally, the release of insulin-like growth factor 2 (IGF- The comparative screening analysis of the profile of cytokines and growth factors released by CPCs on different substrates was performed by protein array ( Figure 4C). The analysis revealed that lower amounts of basic fibroblast growth factor (bFGF) were released on both dECMs compared to CPCs cultured on fibronectin, with a significant difference also in dECM-PH versus dECM-NH ( Figure 4D). Moreover, CPCs cultured on both normal and pathological cardiac dECM released significantly higher amounts of granulocyte-macrophage colony-stimulating factor (GM-CSF) ( Figure 4E) and hepatocyte growth factor (HGF) ( Figure 4F), this latter being also significantly upregulated in dECM-NH versus PH. Additionally, the release of insulin-like growth factor 2 (IGF-2) ( Figure 4G), platelet-derived growth factor AA (PDGF-AA) ( Figure 4H), transforming growth factor beta 2 (TGF-β2) ( Figure 4I), and vascular endothelial growth factor (VEGF) ( Figure 4J) increased significantly on both dECMs, particularly when CPCs were cultured on dECM-PH. Interestingly, very low levels of the receptor KDR/VEGFR2 were detectable in CMs from dECM cultures by protein array ( Figure 4C,K), suggesting the presence of a secreted isoform.
This last result, associated with the corresponding real-time PCR ( Figure 2F) and immunofluorescence data (Figure 3), required further investigation. We performed a PCR with specific primer design to distinguish transcripts for soluble and membrane bound KDR/VEGFR2, in order to understand which isoform was upregulated on pathological dECM. Electrophoretic analysis and densitometric quantification demonstrated that virtually all KDR/VEGFR2 transcripts corresponded to the decoy receptor soluble form (sKDR) in all samples ( Figure 5A,B), which is known to sequester bioavailable VEGF, thus reducing its paracrine angiogenic action [37]. We attempted a more accurate quantification of secreted KDR/VEGFR2 by ELISA, but the assay detection limit was not low enough to allow a reliable comparison among groups in our samples. 2) ( Figure 4G), platelet-derived growth factor AA (PDGF-AA) ( Figure 4H), transforming growth factor beta 2 (TGF-β2) ( Figure 4I), and vascular endothelial growth factor (VEGF) ( Figure 4J) increased significantly on both dECMs, particularly when CPCs were cultured on dECM-PH. Interestingly, very low levels of the receptor KDR/VEGFR2 were detectable in CMs from dECM cultures by protein array ( Figure 4C,K), suggesting the presence of a secreted isoform. This last result, associated with the corresponding real-time PCR ( Figure 2F) and immunofluorescence data (Figure 3), required further investigation. We performed a PCR with specific primer design to distinguish transcripts for soluble and membrane bound KDR/VEGFR2, in order to understand which isoform was upregulated on pathological dECM. Electrophoretic analysis and densitometric quantification demonstrated that virtually all KDR/VEGFR2 transcripts corresponded to the decoy receptor soluble form (sKDR) in all samples ( Figure 5A,B), which is known to sequester bioavailable VEGF, thus reducing its paracrine angiogenic action [37]. We attempted a more accurate quantification of secreted KDR/VEGFR2 by ELISA, but the assay detection limit was not low enough to allow a reliable comparison among groups in our samples. To test the pro-angiogenic paracrine function on differential substrates, CPC conditioned media (CM) were used to culture human umbilical vein endothelial cells (HUVECs) on Matrigel overnight. The quantification of total tubes length, total master segments length, and nodes were similar in all conditions, while in CM from dECM-NH cultures the number of closed loops was significantly higher than CM from pathological sections or control conditions ( Figure 6). These results were consistent with a higher release of the pro-angiogenic mediator NO on dECM-NH ( Figure 4A), and with KDR/VEGFR2 data ( Figures 2F, 3, and 5), indicating overall full angiogenic capacity by HUVECs only when exposed to CM from CPCs cultured on dECM-NH. To test the pro-angiogenic paracrine function on differential substrates, CPC conditioned media (CM) were used to culture human umbilical vein endothelial cells (HUVECs) on Matrigel overnight. The quantification of total tubes length, total master segments length, and nodes were similar in all conditions, while in CM from dECM-NH cultures the number of closed loops was significantly higher than CM from pathological sections or control conditions ( Figure 6). These results were consistent with a higher release of the pro-angiogenic mediator NO on dECM-NH ( Figure 4A), and with KDR/VEGFR2 data ( Figures 2F, 3 and 5), indicating overall full angiogenic capacity by HUVECs only when exposed to CM from CPCs cultured on dECM-NH.

Discussion
The ECM is a dynamic and complex environment capable of significantly regulating cell behavior [38][39][40]. Understanding how the interplay between ECM and CPCs can affect the balance between regeneration and fibrosis will provide new insights on myocardial microenvironment, and innovative thrust towards the improvement of regenerative therapies. We have previously shown that the in vitro substrate produced by human cardiac fibroblasts isolated from failing hearts brings CPCs to a pro-fibrotic and pro-remodeling paracrine profile, compared to the substrate synthesized by cardiac fibroblasts derived from healthy hearts [35]. To perform a more comprehensive evaluation of the effects of the naturally occurring cardiac ECM on cardiac stromal cells, in the present study we used three-dimensional decellularized human cardiac ECM.
The survival and phenotype of CPCs was significantly affected by the dECMs in vitro ( Figures  1C, 2, and 3). As compared to standard FN culture conditions, cardiac commitment-related genes were either significantly upregulated on both dECMs (NKX2.5 and TNNT2), or exclusively on dECM-NH (CX43) or dECM-PH (GATA4 and KDR/VEGFR2). These results show that dECMs represent a more physiological and cardiogenic culture microenvironment for CPCs compared to standard 2D culture conditions. Nonetheless, protein expression for NKX2.5 and CX43 was detectable by immunofluorescence only on dECM-NH (Figure 3), suggesting a stronger cardiac commitment drive only on the physiologic healthy architecture and composition of dECM-NH.
Interestingly, gene expression of the endothelial-related receptor KDR/VEGFR2 was significantly upregulated in CPCs cultured on pathological dECM sections ( Figure 2F). Specific PCR

Discussion
The ECM is a dynamic and complex environment capable of significantly regulating cell behavior [38][39][40]. Understanding how the interplay between ECM and CPCs can affect the balance between regeneration and fibrosis will provide new insights on myocardial microenvironment, and innovative thrust towards the improvement of regenerative therapies. We have previously shown that the in vitro substrate produced by human cardiac fibroblasts isolated from failing hearts brings CPCs to a pro-fibrotic and pro-remodeling paracrine profile, compared to the substrate synthesized by cardiac fibroblasts derived from healthy hearts [35]. To perform a more comprehensive evaluation of the effects of the naturally occurring cardiac ECM on cardiac stromal cells, in the present study we used three-dimensional decellularized human cardiac ECM.
The survival and phenotype of CPCs was significantly affected by the dECMs in vitro ( Figures 1C,  2 and 3). As compared to standard FN culture conditions, cardiac commitment-related genes were either significantly upregulated on both dECMs (NKX2.5 and TNNT2), or exclusively on dECM-NH (CX43) or dECM-PH (GATA4 and KDR/VEGFR2). These results show that dECMs represent a more physiological and cardiogenic culture microenvironment for CPCs compared to standard 2D culture conditions. Nonetheless, protein expression for NKX2.5 and CX43 was detectable by immunofluorescence only on dECM-NH (Figure 3), suggesting a stronger cardiac commitment drive only on the physiologic healthy architecture and composition of dECM-NH.
Interestingly, gene expression of the endothelial-related receptor KDR/VEGFR2 was significantly upregulated in CPCs cultured on pathological dECM sections ( Figure 2F). Specific PCR experiments revealed that the proportion of transcript for membrane-associated KDR/VEGFR2 was negligible in all conditions, while its soluble form was the absolute majority in all conditions ( Figure 5). Moreover, KDR/VEGFR2 protein was detectable by immunofluorescence only on dECM-PH; albeit, we could not distinguish between intracellular and membrane signals in our conditions. Overall, data suggest that CPCs cultured on pathological dECM express significantly higher levels of decoy KDR/VEGFR2 compared to dECM-NH. Despite consistent VEGF release also on dECM-PH ( Figure 4J), functional results ( Figure 6) are coherent with sequestering and low bioavailability of VEGF. Moreover, cells cultured on pathological dECM sections also released significantly lower levels of NO (a well-known mediator of angiogenesis [41][42][43]), HGF (an angiogenic growth factor with multiple functions, including inhibiting fibrosis and activating tissue regeneration [44]), and GM-CSF, which also possess pro-angiogenic properties [45]. These changes were consistently associated to reduced pro-angiogenic paracrine support to endothelial cells: in fact, HUVEC angiogenic functional assay demonstrated that only conditioned media from dECM-NH was able to stimulate complete angiogenesis within 18 h, including mature closed loops formation ( Figure 6). Overall, our combined observation of decreased NO, increased decoy KDR/VEGFR2 expression, and reduced pro-angiogenic activity points to the existence of a block in VEGF-mediated angiogenic signaling in CPCs exposed to pathological ECM.
It is well established that GATA4 plays an important role in cardiac hypertrophy, acting as a rescue agent. In stress conditions, GATA4 is overexpressed in cardiomyocytes, leading to compensatory paracrine mechanisms [46]. In our conditions, GATA4 gene expression was significantly higher on dECM-PH; albeit, no qualitative difference was appreciable by immunofluorescence staining. Moreover, conditioned media analyses underlined that H 2 O 2 release in CM from CPCs on pathological dECM was higher than the other two conditions. H 2 O 2 can act as a rescue agent in stress conditions; in fact, it is released from myocardial cells during ischemia/reperfusion injury, and it is able to interact with membrane KDR/VEGFR2 and stimulate tissue vascularization [47]. For the evidence presented above, we can hypothesize that when embedded in pathological dECM, CPCs are detrimentally affected by the altered microenvironment, and consequently forced to activate a "rescue-like plan". This plan is nonetheless defective due to the increased production of the soluble form of KDR/VEGFR2, which indeed acts as a decoy for secreted VEGF, paralleled to reduced production of NO, impairing overall pro-angiogenic signals. These results suggest a novel mechanism by which the cardiac stromal primitive compartment may partially lose its paracrine ability to sustain angiogenesis in IHD and HF myocardium subjected to pathological ECM remodeling.
Concerning the paracrine profile, the microenvironment created by dECM-PH seemed also to support pro-fibrotic signaling from CPCs. In particular, PDGF-AA and TGF-β2 release in CM from dECM-PH was increased compared to dECM-NH. PDGF-AA promotes fibrosis through multiple mechanisms, such as inducing proliferation, myofibroblast polarization, and collagen I synthesis in cardiac fibroblasts [48,49]. TGFβ2, instead, is involved in inducing endothelial-to-mesenchymal transition (EndMT), which is a main mechanism of cardiac fibrosis [50,51]. Basic FGF release, instead, was significantly reduced on dECMs, particularly on dECM-PH; this growth factor is known to attenuate myofibroblast-mediated ECM remodeling and fibrosis progression in the heart [52], and its signaling sustains cardiomyocyte homeostasis through CX43 phosphorylation [53]. Finally, high IGF-2 levels have been associated to reduced recovery from HF due to reverse remodeling [54]; interestingly, CPCs cultured on pathological dECM indeed released higher amounts of IGF-2, further supporting a pro-fibrotic paracrine profile from CPCs in this condition. Based on our results, one might speculate that TGFβ2 secreted by CPCs in the presence of dECM-PH may sustain interstitial fibrosis, that is, the typical post-infarction compensatory response of the uninjured myocardium, remote to the reparative fibrosis [55]. Moreover, in a kind of vicious circle, CPCs respond to pathological remodeled cardiac ECM, likely through signals transduced by β1 integrin [56,57], releasing TGFβ2 that, in turn, induces the activation of resident fibroblasts, promotes the persistence of myofibroblasts, and induces the synthesis of fibrillar collagens, thus possibly further worsening ventricular fibrosis and compliance.
Despite this shift in paracrine signaling, though, CPCs did not display features of direct contribution to ECM deposition, as for collagen expression levels observed (Supplementary Figure S1).
In conclusion, CPCs exposed in vitro to dECM-PH from HF myocardium displayed partially reduced cardiac commitment drive, defective paracrine support to angiogenesis, and increased release of pro-fibrotic cytokines, despite displaying some features of a rescue-like mechanism, similar to those described in vivo in pathological conditions. These observations shed novel insights on the crosstalk between ECM remodeling and cardiac stromal primitive cells, suggesting also caution and limitations in the use of non-healthy decellularized myocardium for cardiac tissue engineering approaches.

Decellularized Extracellular Matrix (dECM) Sections Production
Cardiac tissue samples were collected from normal and pathological adult human hearts. Waste fragments of atrial appendages from heart of donors (n = 6, mean age 36.4 ± 8.9 years) who died for reasons other than cardiovascular diseases, and samples from atrial appendages of explanted hearts of patients (n = 6; mean age 59.3 ± 5.2 years; 4 males, 2 females; mean ejection fraction 16.7 ± 2.6%) with end-stage HF associated with ischemic cardiomyopathy, were collected during heart transplant procedures performed at the Monaldi Hospital (Naples, Italy). Specimens were snap-frozen and stored at −80 • C. Informed consent for use of heart tissue for experimental studies was obtained from all patients, and samples were collected and classified without patient identifiers, in accordance with protocols approved by the Ethical Committee of Monaldi Hospital and the "Federico II" Hospital (79/18, May 11th, 2018), and in conformity with the principles outlined in the Declaration of Helsinki. Frozen specimens were mounted on a cryostat chuck using Tissue Freezing Medium (Leica Microsystems, Wetzlar, Germany). Then, 200 µm thick sections were cut by a Leica CM1950 cryostat (Leica Microsystems, Wetzlar, Germany). Cryosections were collected in sterile 15 mL plastic tubes containing sterile sodium chloride solution (Sigma-Aldrich, St. Louis, MO, USA). Afterwards, cryosections were decellularized, as previously described [58]. Briefly, cryosections were immersed in a 1% SDS, 1% Triton solution in bidistilled water for 24 h, then rinsed for 24 h in antibiotic solution (100 U/mL penicillin, 50 U/mL streptomycin, and 0.25 µg/mL amphotericin B, all from Sigma-Aldrich, St. Louis, MO, USA) in PBS and finally for an additional 30 min in sterile bi-distilled water. All steps were performed with agitation on an orbital shaker. Sections of cardiac dECM were then mounted on sterile 35 mm cell culture dishes and on 96-well cell culture plates, sterilized by exposure to ultraviolet radiation for two cycles of 20 min each, and rehydrated for 5 days with Iscove's modified Dulbecco's medium (IMDM) supplemented with 20% FBS (both from Sigma-Aldrich, St. Louis, MO, USA) and antibiotics, in an incubator at 37 • C with 5% CO 2 . Once rehydrated, sections of dECM from normal and pathological hearts (dECM-NH and dECM-PH, respectively) were used as substrates for cell culture.

Assay of CPC Viability
An amount of 1 × 10 4 CPCs were seeded on dECM-NH or dECM-PH mounted on 96-well plates and cultured for 7 days under standard culture conditions in the same cell culture medium used for hydration. As a control, 1 × 10 4 cells were seeded on fibronectin-coated wells. Cells were checked daily by an Olympus CKX41 inverted microscope equipped with a Colorview IIIu digital camera (Olympus Corporation, Tokyo, Japan). Beginning at 48 h after seeding, and then every day for one week, cell death rate was assessed using trypan blue exclusion assay adapting a previously described protocol [58,61]. Specifically, every day cells were detached from a subset of wells in the multi-well plates by incubation with 0.25% trypsin-EDTA solution (Sigma-Aldrich, St. Louis, MO, USA) for 10 min. Detached cells were then stained with trypan blue stain (0.4% in PBS) (Lonza, Walkersville, MD, USA) for 2 min at room temperature and counted using a hemocytometer. The percentage of dead cells and of alive cells over total cells for each time point was expressed as the mean percentages.

Immunostaining and Fluorescence Microscopy Analyses
An amount of 2.5 × 10 5 CPCs/cm 2 were seeded on dECM-NH or dECM-PH mounted on 35 mm culture dishes and cultured under standard culture conditions in the same cell culture medium used for hydration. As a control, 2.5 × 10 5 cells/cm 2 were seeded on fibronectin-coated 35 mm culture dishes and cultured in the same conditions. Cells were checked daily by an Olympus CKX41 inverted microscope equipped with a Colorview IIIu digital camera (Olympus Corporation, Tokyo, Japan). After five days of culture, cells were immunostained as previously described [34,62]

RNA Extraction and Real-Time PCR
Total RNA was extracted using the miRNeasy Micro Kit (Qiagen, Hilden, Germany) and quantified using a spectrophotometer. cDNA was synthesized using 0.5µg RNA, with the High Capacity cDNA Reverse Transcription Kit (Life Technologies, Thermo Fisher Scientific, Waltham, MA, USA). Real-time qPCR was performed to assess gene expression, using Power SYBR Green PCR Master Mix (Life Technologies, Thermo Fisher Scientific, Waltham, MA, USA) and standard thermocycling conditions according to the manufacturer's protocol. The relative ratio for each section versus culture on fibronectin was calculated using the comparative Ct method (2ˆ-∆∆Ct) for each patient's sample. The set of genes were analyzed, and the primer sequences are listed in Table 1. GAPDH was selected as housekeeping gene, according to the Norm Finder software (MOMA, Aarthus, Denmark).

Soluble KDR/VEGFR2 mRNA Analysis
We performed semi-quantitative PCR in order to detect the presence of the soluble VEGFR2 isoform. We designed a primer pair specific for the VEGFR2 spanning exon 13 in order to detect the inclusion of this exon by agarose gel electrophoresis [63]. We used GAPDH as reference gene. The primer sequences are listed in Table 1. The PCR was performed using Bestaq DNA polymerase (ABM Inc, New York, NY, USA) as follows: 2 at 95 • C followed by 35 cycles of 30 s at 95 • C, 45 s at 58 • C, and 45 s at 72 • C. The samples were separated on a 1% agarose gel with Gel Red Nucleic Acid gel stain (Biotium, San Francisco, CA, USA) for bands detection, and the image was acquired and analyzed using Typhoon FLA 9500 imager and analysis software (GE Healthcare Life Sciences, Sheffield, UK).

Conditioned Media Collection
After 5 days of culture on dECM-NH, dECM-PH, or fibronectin (2.5 × 10 5 CPCs/cm 2 ), media was changed for the last 48 h of culture, in the presence of 0.1% of FBS (Sigma-Aldrich, St. Louis, MO, USA), and then collected to be analyzed for the quantification of cytokines, NO, and H 2 O 2 . Media were centrifuged at 2000rcf for 5 min, and then stored at −80 • C until analysis. Basal non-conditioned media was used as blank control for all assays.

Quantification of H 2 O 2 and NO
Assays on conditioned media were performed as previously described [64]. Briefly, H 2 O 2 was evaluated by a Colorimetric Detection Kit (Arbor Assays, Ann Arbor, MI, USA) and expressed as µmol/L. Intra-assay and inter-assay coefficients of variation were 2.1% and 3.7%, respectively. A colorimetric assay kit (Tema Ricerca, Castenaso, Italy) was used to determine the nitric oxide metabolites nitrite and nitrate (NOx) in cell culture supernatants. Intra-assay and inter-assay coefficients of variation were 2.9% and 1.7%, respectively. Assays were performed according to the manufacturers' instructions.

Growth Factor Array
Culture medium was assayed by the Human Growth Factor Array C1 (Raybiotech, Norcross, GA, USA) to simultaneously detect 41 targets. The procedure was performed as previously described [65]. Briefly, array membranes were blocked with blocking buffer for 30 min at room temperature, and then 1 mL of culture medium was added to each membrane and incubated at room temperature for 2.5 h. Membranes were then washed 3 times in wash buffer I, twice in wash buffer II, and then incubated with biotin-conjugated antibody overnight at 4 • C. After further washes, membranes were incubated for 2 h at room temperature with horseradish peroxidase (HRP)-conjugated streptavidin and washed one last time to remove unbound reagents. All incubation steps were performed with agitation on orbital shaker. Membranes were then developed with the detection buffer, exposed to film, and processed by autoradiography. Quantitative comparison of the signal densities was performed following the guidelines supplied with the array protocol. Briefly, images of arrays were scanned, and spot signal densities were obtained using ImageJ software (NIH, Bethesda, WA, USA). The background was then subtracted from the densitometry data, and the obtained values were normalized to the positive control signals.

HUVEC Angiogenic Assay
Human umbilical vein endothelial cells (HUVECs) were cultured for 18 h on Matrigel-coated 96-well plates (Growth Factor Reduced Matrigel Matrix Phenol Red Free, BD, San Jose, CA, USA) at a density of 2 × 10 4 cells/well in the presence of the CMs from each condition, using basal non-conditioned media as a negative control. Assay quantification, in terms of total master segment length, total length, and number of loops and nodes, was performed with the Angiogenesis Analyzer Plugin of the ImageJ Software (NIH, Bethesda, WA, USA) on randomly captured images with a 4X objective.

Statistical Analysis
All results are presented as mean value ± standard error of the mean, unless specified. Significance of difference between any two groups was determined by one-way ANOVA test, while multiple comparison correction was applied when necessary. A final value of p < 0.05 was considered significant.