Synergistic metalloproteinase-based remodeling of matrix by pancreatic tumor and stromal cells

The process by which tumor cells mechanically invade through the surrounding stroma into peripheral tissues is an essential component of metastatic dissemination. Matrix metalloproteinase (MMP)-mediated extracellular matrix (ECM) degradation plays an important role in this invasive process. Defining the contribution and interaction between these MMPs during invasion remains a key interest in the development of targeted anti-metastatic therapies. In this study we have utilized multiple different stromal fibroblasts and tumor cells to define the relative contributions between cancer cells and stromal cells during MMP-dependent matrix remodeling and pancreatic (PDAC) tumor cell invasion. We find that tumor cells co-cultured with the conditioned medium from stromal fibroblasts exhibited a substantial increase in invadopodial-based matrix degradation and transwell invasion. This increase is dependent on pro-MMP2 expressed and secreted by stromal fibroblasts. Further, the pro-MMP2 from the stromal fibroblasts is activated by MT1-MMP expressed on the tumor cells. Depletion of MT1-MMP, the known activator of MMP2, in tumor cells largely blocked matrix remodeling, even in the presence of stromal cell medium. In summary, these findings implicate an important interplay between MT1-MMP from tumor cells and MMP2 from fibroblasts as a key component for ECM remodeling and invasion.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) is a lethal and highly metastatic cancer. The high mortality rate for pancreatic cancer is largely due to the advanced stage at which the disease is diagnosed, resistance to therapies, and particularly early metastasis [1][2][3]. Pancreatic tumors are exceptionally invasive and are believed to invade peripheral tissues via a combination of stromal remodeling and active migration that facilitates the intravasation of this tumor into the adjacent vasculature [4,5]. This remodeling of the surrounding tumor environment is achieved in part through the secretion of a variety of different matrix metalloproteinases (MMPs).

Fibroblasts stimulate the invasive properties of pancreatic cancer cells
As fibroblasts are known to promote invasive metastasis of various tumor types by a variety of processes, we tested if stromal cells might alter the ability of pancreatic cancer cells to degrade extracellular matrix components in vitro. To this end, BxPC3 PDAC cells were seeded on fluorescent gelatin-coated coverslips alone, with or without human pancreatic stellate cells (hPSC), which are one source of CAFs in pancreatic tumors, as well as normal human or rat fibroblasts.
To distinguish the tumor cells from the fibroblasts, the BxPC3 tumor cells were stably transfected with GFP (pEGFP N1) vector. The capacity of PDAC cells to degrade the surrounding substrate was quantified by both the percentage of PDAC cells degrading the matrix and the area of degradation per PDAC cell area. Importantly, coculture with stromal cells enhanced the percentage of BxPC3 cells degrading the matrix over 2 fold and the area of degradation per BxPC3 cell area 6-9 fold (Fig 1a-1f). The experiment was also performed using PANC-1 cells, which are normally unable to degrade surrounding substrates. Strikingly, the PANC-1 cells exhibited a substantial increase in both the percentage of cells capable of degrading gelatin and the area of degradation per PANC-1 cell area upon coculturing with fibroblasts (S1a-S1d Fig).
Consistent with our prior studies, the hPSCs, HFs, and RFs did not degrade significant amounts of matrix at this timepoint when plated alone (Fig 1g-1i) [36]. These findings suggest that fibroblasts stimulate the invasive properties of pancreatic cancer cells.
To test if stromal cells contribute to tumor cell ECM degradation by a direct physical interaction or by the secretion of trans-acting factors, we collected serum-free conditioned medium (CM) from the hPSCs, HFs, or RFs. This media was applied to 4 different PDAC cell lines (BxPC3, PANC-1, DanG, HPAF-II cells) and the level of matrix degradation exhibited by these cells was quantitated. Incubation with stromal cell CM induced substantial levels of ECM degradation that mimicked the levels observed from a direct coculturing of the PDAC cells with fibroblasts directly (Fig 1j-1o, S1e-S1j and S2 Figs). These findings indicate that fibroblasts stimulate pancreatic cancer cell-mediated ECM degradation by secreting factors into the environment. The ECM remodeling capacity of MDA-MB-231 breast cancer cells is also promoted by CM from fibroblasts, suggesting that this important process is utilized by multiple tumor types (S2 Fig).
To extend the matrix degradation assays into a more biological context, the effects of CM collected from stromal cells were tested using transwell invasion assays. In these experiments BxPC3 cells were seeded in chemotactic transwell invasion chambers on the top side of porous filters coated with 0.3% gelatin and incubated in either serum-free DMEM control media or CM. Cell invasion from the top to the bottom chamber was initiated by the inclusion of 10% FBS in the media in the bottom chamber acting as a chemo-attractant. BxPC3 cells exhibited a 2-3 fold increase in invasion when incubated with fibroblast CM compared to the cells cultured with control DMEM only, particularly with the conditioned medium from the hPSCs (Fig 1p). Taken together these findings suggest that cancer cells can sense and respond to trans-acting factors secreted by nearby stroma to enhance ECM degradation.
To test if the number of invadopodia is increased by exposure to stromal-generated CM, BxPC3 cells were plated on green fluorescent gelatin-coated coverslips and stained for the invadopodial markers Tks5 and cortactin. Consistent with data shown in Fig 1j-1o, BxPC3 cells co-cultured with the CM from fibroblasts exhibited a substantial increase in matrix degradation (Fig 2a-2d). Importantly, these cells also displayed a marked increase in the number of both Tks5 and cortactin-positive invadopodia. A 3-10 fold increase in the number of invadopodia were observed in the cells co-cultured with CM from HFs (Fig 2b), hPSCs (Fig 2c), or RFs (Fig 2d) compared to the DMEM control treated cells (2a, e). Remarkably, the CM also (a-f) ECM degradation by BxPC3 cells is increased by coculturing with fibroblasts. BxPC3 cells stably expressing GFP were seeded onto Cy3-fluorescent gelatin-coated coverslips alone (a, a',a") or together with HFs (b, b', b"), hPSCs (c, c', c"), or RFs (d, d', d"). Cells were stained for actin and gelatin degradation was quantified after 8 h. Tumor cells were distinguished by GFP expression. Scale bar = 10μm. Co-culture with any of these stromal cells results in an over 2-fold increase in the percentage of cells degrading the gelatin substrate by BxPC3 cells (�100 cells per condition, e) and a 6 fold increase in the area degraded per cell in BxPC3 cells (�10 cells per condition, f). (g-i) Fibroblasts were plated on Cy3-fluorescent gelatin-coated coverslips and stained for actin using phalloidin. The HF (g, g'), hPSC (h, h'), and RF (i,i'), degrade minimal matrix after 8 hours. (j-o) Conditioned medium (CM) from fibroblasts promotes ECM degradation by PDAC cells. BxPC3 cells were seeded onto green fluorescent gelatin-coated coverslips with DMEM only (j, j') or CM collected from HFs (k, k'), hPSCs (l, l'), and RF cells (m, m'). Cells were stained for actin and gelatin degradation was quantified after 8 h. Coculture of BxPC3 cells with CM from fibroblasts increased the percent of BxPC3 cells degrading the gelatin substrate by 2 fold (�100 cells per condition, n) and the area degraded per cell in BxPC3 cells by 4-6 fold (�10 cells per condition, o). (p) The conditioned medium from fibroblasts increased BxPC3 cell transwell invasion. BxPC3 cells were seeded in transwell invasion assays for 6 h with DMEM or CM from fibroblasts. The presence of CM from HF, hPSC, and RF all increased the number of BxPC3 cells migrating through the filters. All graphed data represent the mean ± SEM of 3 independent experiments. � p <0.05. �� p <0.01. Scale bars = 10μm.
https://doi.org/10.1371/journal.pone.0248111.g001 induced a remarkable increase in invadopodia formation in the PANC-1 cells (S1k Fig), which do not normally form invadopodia. No functional invadopodia were detected in the presence of the MMP inhibitor BB-94 (Fig 2e), and the conditioned medium by itself did not lead to any degradation of the matrix (Fig 2f). These observations suggest that fibroblasts may secrete factors/proteins to stimulate invadopodia accumulation and function in pancreatic cancer cells while enhancing their matrix remodeling ability.

Fibroblasts secrete MMP2 to enhance the invasive properties of pancreatic cancer cells
The data presented above indicate that stromal fibroblasts stimulate tumor cell invadopodial degradation through secretion of a soluble factor. We hypothesized that stromal cells might secrete MMPs that can cooperate with PDAC cells to accentuate ECM degradation. To test this prediction, we analyzed the proteases found in both stromal and tumor cells with a focus on the soluble MMP2 and MMP9 proteases and the transmembrane MT1-MMP proteases that are known to drive matrix degradation in many tumor types, including breast, prostate, and pancreatic cancer [11,[37][38][39][40]. The expression levels of these 3 proteases were compared using western blot and zymography (Fig 3a and 3b) in 6 different stromal cell lines (RF, HF, hPSC, ITAF, imPSCc2, imPSCc3), as well as 6 different PDAC cell lines (DanG, BxPC3, CFPAC, L3.6, HPAF-II, PANC-1) and 1 breast cancer cell line (MDA-MB-231). Within the stromal cells, RF and HF are normal fibroblasts while hPSC, ITAF, imPSCc2 and imPSCc3 are pancreatic stellate cells from either human or mice [41,42]. Interestingly, the stromal cell lines, particularly RF, HF, hPSC, and ITAF cells, displayed relatively high levels of MMP2, especially pro-MMP2, in comparison to the tumor cells. Of all the cells we tested, MMP9 was detected in two cells lines (ITAF and BxPC3 cells). MT1-MMP on the other hand was ubiquitously expressed in PDAC and stromal cells.
Because of the elevated MMP2 expression observed in the stromal cells, we tested if this soluble protease might act as the trans-acting factor responsible for promoting PDAC cell matrix degradation. To this end, CM from the different stromal cells was added to BxPC3 cells seeded on fluorescent gelatin and analyzed for changes in ECM degradation. Correlative measurements were made to test if the stromal cells with the greatest MMP2 levels induced greater matrix degradation than that induced by cells with lower MMP2 expression/activity. Accordingly, BxPC3 cells exhibited a higher degradative response correlated with the MMP2 levels of the stromal cells (iTAF, imPSCc2, imPSCc3) from which the CM was derived (S3d-S3h In order to address the role of MMP2 within fibroblast CM more directly, hPSC and HF cells were treated for 72 h with control siRNA or siRNA to reduce the levels of MMP2, and we tested if CM from these cells had a reduced ability to stimulate PDAC cell degradation. A decrease in the MMP2 levels and activity in the CM from the transfected cells was confirmed by western blot and zymography (Fig 3c and   To test if MMP2 secreted from stromal cells contributes to PDAC cell invasion in vitro, we employed the transwell invasion assays described above in Fig 1p. BxPC3 cells were seeded in the top chamber with CM from control-treated stromal cells or MMP2-depleted CM. The invasiveness of BxPC3 cells was analyzed after 8 h by counting the percentage of cells invading from the top to the bottom side of the chamber. Cells incubated with the CM from control cells exhibited 12-fold more invasion than did the cells presented with the MMP2-depleted CM (Fig 3i), consistent with the prediction that MMP2 from CM contributes to PDAC cell invasion.
As an additional approach to the siRNA-mediated MMP2 depletion studies described above, the MMP2/9 inhibitor SB-3CT was utilized to inhibit stromal cell MMP2 activity directly. To first verify the activity of this inhibitor, a dose-response analysis was performed by adding increasing concentrations of this inhibitor to CM collected from RFs (RF CM). As shown by the zymograph in S4g Fig

The importance of PDAC cell MT1-MMP in matrix remodeling
MMP2 is initially secreted as a pro-form and becomes activated outside the cell via the action of the MT1-MMP residing on the plasma membrane [37]. Based on the protease western blotting and zymography in Fig 3a and 3b, we predicted that MMP2 secreted by the stromal fibroblasts is cleaved and activated by MT1-MMP on the PDAC tumor cells. To test this premise, BxPC3 (Fig 4)  Conversely, we tested if overexpression of MT1-MMP was sufficient to induce matrix degradation in PANC-1 cells, which do not degrade gelatin. PANC-1 express pro-MMP2, but have very low levels of MT1-MMP (Fig 3a and 3b). Consistent with our prior studies, overexpression of MT1-MMP in PANC-1 induced a dramatic increase in matrix degradation. To test if this degradation required the pro-MMP2 secreted by PANC-1, we depleted MMP2 expression in PANC-1 cells using siRNA. Indeed, knockdown of MMP2 blunted the matrix degradation induced by overexpression of MT1-MMP, supportive of cross-talk between these two pro-invasive proteinases in pancreatic tumor cells (S5g-S5m Fig).
To pursue this concept further, zymography was employed to test if the MMP2 derived from the stromal cells was activated by the tumor cells. To this end, CM from hPSCs, HFs, or RFs was collected and applied to BxPC3 or CFPAC PDAC cells for incubation for 0, 24, 48, or 72 h. Subsequently, CM from the PDAC cells was collected for zymography as depicted in Fig  5a. As predicted, a steady increase in the active form of MMP2 was observed over increasing co-incubations (Fig 5b-5g; S6a-S6d Fig). This finding suggests that MMP2 secreted by stromal cells can be activated by tumor cells. To confirm the requirement for tumor cell MT1-MMP in this process, we used siRNA to deplete MT1-MMP in BxPC3 tumor cells. Importantly, knockdown of MT1-MMP blocked the activation of MMP2 upon incubation with BxPC3 cells, demonstrating that MT1-MMP is crucial for this activation (Fig 5f and 5g). Notably, we have not observed any active MMP2 secreted by BxPC3 or CFPAC cells by testing their conditioned medium collected at 24h, 48h, or 72h (S6e Fig). These data support the concept that the stimulation of pancreatic cancer cell matrix degradation by fibroblast-secreted MMP2 requires the activation by MT1-MMP residing on the tumor cells. Thus, crosstalk between fibroblasts and cancer cells appears critical for promoting tumor stromal remodeling, as illustrated in Fig 5j.

The tumor microenvironment in cancer metastasis
In this study we have focused on the interactive contributions of stromal and PDAC tumor cells to pro-invasive matrix remodeling. The specific tumor microenvironment for pancreatic cancer has two major features that include fibrotic desmoplasia as well as an immunosuppressive component [26,43]. Here we focus on desmoplasia where PSCs are a central contributing factor [20]. PSCs are resident mesenchymal cells located in the pancreas. Quiescent PSCs store lipid droplets, express MMPs including MMP2, MMP9 and MMP13, as well as MMP inhibitors, and maintain normal tissue homeostasis [32]. Upon activation by growth factors and cytokines that are secreted by immune cells, pancreatic cancer cells, and endothelial cells, PSCs can differentiate into cancer-associated fibroblasts (CAFs), leading to a loss of lipid droplets with a concomitant upregulation of MMPs and ECM components [44]. Importantly, PSCs can differentiate into different subtypes of CAFs, including those adjacent to neoplastic cells expressing high levels of α-smooth muscle actin (α-SMA), while others may secrete IL6 and other inflammatory mediators without SMA expression [20,21,26,45]. Activated CAFs can then promote metastasis through the production and remodeling of ECM, the secretion of growth factors and cytokines that promote cancer cell survival and invasion, as well as angiogenesis [46]. Thus, the biology of the supportive PDAC stroma is complex and critical for invasion and metastasis. Accordingly, this study utilized both PSCs and normal fibroblasts to cast a wide net in an attempt to test for conserved interactive processes between these stromal cells and PDAC tumor cell lines, and determine how these processes contribute to cancer cell invasion.
Through a combination of the cell models described above, we report a mechanism of matrix metalloprotease activation in PDAC through a crosstalk between the tumor cells themselves and the adjacent stroma (Fig 1). Importantly, this invasion-promoting cross communication appears to be mediated by the synergistic interaction of membrane-bound MT1-MMP residing on tumor cells with the soluble MMP2 secreted by PSCs and normal fibroblasts ( Fig  3). As normal human and rat fibroblasts similarly provided MMP2 to promote matrix degradation and tumor cell invasion, these data suggest that tumor cell invasion could be supported by the stroma even prior to the conversion to CAFs, for example, early in tumor development, or as invading tumor cells interact with normal fibroblasts in other tissues. Interestingly, we observed that different populations of PDAC tumor cells can also activate other epithelial PDAC tumor cells to become more invasive (S3i-S3q Fig). As PDAC can have significant intratumoral heterogeneity, this finding suggests that cross-talk among tumor cells with complementary MMP expression may also promote invasion [47].

Trans-activation of MMP2 by MT1-MMP
From the findings presented here it appears that MMP2 and MT1-MMP are directly involved in the degradation of extracellular matrix components. It has been reported that MMP2 activation by MT1-MMP contributes to pancreatic cancer progression and invasion, as both fibroblasts and cancer cells can express these proteases [48]. However, the specific contributions and actions of the stroma versus epithelial tumor cells in protease-based ECM degradation are undefined. In this study, a comparison of MMP2 and MT1-MMP expression in 6 stromal cells and 7 tumor cell lines revealed a strong correlation between the levels of protease expression and the capacity of these cells to degrade gelatin. We observed that MMP2 is highly expressed Quantitative graphs depicting the ratio of pro/active MMP2 at each timepoint, demonstrating the increase in activity of MMP2 from the zymograms of at least 3 independent experiments. (f) MT1-MMP was depleted in BxPC3 cells by siRNA transfection. MMP2 in CM from hPSCs is also activated over time when incubated with control BxPC3 cells, but this is inhibited following MT1-MMP knockdown. (g) Quantitation of the ratio of pro/active MMP2 at each timepoint from zymograms from 3 independent experiments. Averages ± SEM. � p <0.05. ��  and secreted by fibroblasts and PSCs, whereas MT1-MMP is expressed in both fibroblasts and cancer cells (Fig 3 and 3b).
Surprisingly, MMP2 from stromal fibroblasts appears to be largely inactive, despite the coexpression of MT1-MMP by these cells (Fig 3a and 3b and S3i Fig). Importantly, this indicates that these stromal cells have only a modest capacity for self-activation of their own MMP2 and require a trans-activation of this protease by the MT1-MMP expressed by the nearby tumor cells (Fig 5 and S6 Fig). Reciprocal trans-activation of tumor cell matrix remodeling by the nearby stroma is also demonstrated by the fact that PANC-1 cells, epithelial tumor cells that exhibit very modest matrix degradation, can be activated to increase matrix remodeling via incubation with media from MMP2-secreting stromal cells, even though PANC-1 also express pro-MMP2 (S1 Fig). Similarly, other pancreatic cancer cells such as DanG, HPAF-II, L3.6, and BxPC3 cells all express various levels of both MMP2 and MT1-MMP (Fig 3a and 3b). However, similar to fibroblasts they appear incapable of self-activating MMP2 as minimal active MMP2 is detected by zymography (Fig 3b and S6e Fig). Thus, trans-activation of MMP from one cell type depends upon the activity of MT1-MMP from another. These findings further support the premise that MMP2 activation is a trans-acting event requiring the participation of both cell populations.
It remains unclear why MT1-MMP based activation of secreted MMP2 does not occur in a cis-fashion at the surface of tumor or stromal cells that generates both proteases. Perhaps additional cofactors or the topological orientation of these two proteases play a key role. It is noteworthy that the soluble protease MMP9, while not expressed in the majority of the pancreatic cancer cells and stromal cells tested here, has been implicated in PDAC progression as well (Fig 3b) [10,38]. While there are undoubtedly additional factors mediating cross-talk between fibroblasts and tumor cells, which may regulate invadopodial dynamics, migration, protein expression, proliferation, and survival, our data indicate the requirement for the MT1-MMP/ MMP2 tumor/stromal cell axis for pro-invasive matrix degradation. The findings reported here underscore the need for further understanding of this complex crosstalk between the cell types of this lethal tumor. They also emphasize the importance of defining the biology of this stroma that drives the desmoplastic nature of PDAC and may represent over 80% of PDAC tumor mass. Such understanding will prove helpful toward developing new therapies toward attenuating the metastatic invasion of this lethal cancer.  -1918). HPAF-II (ATCC CRL-1997) was maintained in MEM containing 10% FBS and 1% P/S. hPSC was from ScienCell (#3830, Carlsbad, CA, USA) and maintained in stellate cell medium (SteCM) as described previously [22]. A patient-derived PSC cell line, ITAF (hCAF), and two C57BL/6 mouse-derived immortalized PSC cell lines, imPSC c2 and imPSC c3, were provided by Dr. Raul Urrutia, Medical College of Wisconsin and maintained as described previously [41,42]. BxPC3 (GFP) is pEGFP N1 vector stable expressed in BxPC3 and maintained with G418 sulfate (400 μg/ml, CORNING, Mediatech Manassas,VA) in RPMI1640 containing 10% FBS and 1% P/S. All cells were grown in a 5% CO 2 incubator at 37˚C.

Cell culture
For co-culture experiments, fibroblasts:cancer cells were plated in the following ratios: human fibroblasts 7:1, hPSCs 5:1, rat fibroblasts 3:1. To generate conditioned medium, cells were cultured at confluence for 24-48 hours in the absence of serum. Cells incubated with conditioned medium were compared to cells in serum-free medium.
The following inhibitors were used as indicated:

Gelatin zymography
Conditioned medium from cultured cells was collected as previously described [22]. Briefly, cells were cultured in medium without serum for 24-48 hours at similar confluency. Upon collection, the culture medium was clarified by centrifugation at 6000rpm for 2 min. Then the conditioned medium was mixed with zymogram sample buffer (BioRad, Hercules, CA) and incubated at room temperature for 10 minutes. The conditioned medium was resolved on 7.5% SDS-PAGE gels containing 1 mg/ml gelatin. After electrophoresis, the SDS-PAGE gel was then transferred into 2.5% Triton X-100 and incubated at room temperature for 40 minutes with shaking. After incubation, the gel was rinsed with incubation buffer (50 mM Tris-B pH8.0; 150 mM NaCl; 10 mM CaCl; 0.05% NaN 3 ) before being soaked in incubation buffer at 37˚C for 20~24 hours. After incubation, the gel was rinsed with dH 2 O 3 times before staining with Coomassie blue for 40 minutes at room temperature. Finally, destaining was performed at room temperature for 2 hours before the gels were imaged. Active and Pro-MMP2 bands were quantified by the area density measurement using Chemi Doc-IT 2 Imager software (UVP).

Immunofluorescence
Cells were fixed as described previously [22]. TRITC-phalloidin and Phalloidin-Atto 390 (Sigma Aldrich) were used to visualize Actin. The coverslips were mounted with Prolong mounting medium (Life Technologies) before imaging. Fluorescence images were acquired using epifluorescence microscopes (Axio Observer and Axiovert 200; Carl Zeiss MicroImaging) using a 63x oil objective with iVision software or Zen software. Adobe Photoshop software (Adobe) was applied to process and adjust the images uniformly.

Matrix degradation assays
Gelatin-coated coverslips were prepared using 0.2% gelatin (Sigma Aldrich) and diluted Oregon Green-conjugated gelatin (Invitrogen) and Cy3-conjugated gelatin (EMD Millipore), as previously described [50,51]. Cells were seeded on the gelatin-coated coverslips in the presence of the MMP inhibitor BB-94 overnight (2μM) and in the presence of 10% FBS. The inhibitor was washed out the following day to allow matrix degradation by MMPs. The cells were incubated in the appropriate serum-free medium for the indicated time before fixation. After TRITC-phalloidin staining (Sigma Aldrich) or immunofluorescence for the invadopodial markers Tks5 (Millipore-Sigma), and cortactin (4F11, Millipore-Sigma) the cells were imaged with an AxioObserver D.1 epifluorescence microscope (Carl Zeiss, Thornwood, NY, USA). The percentage of cells degrading the matrix was determined with at least 100 randomly imaged cells. The degradation area per cell area was quantified with at least 10 cells per condition with ImageJ software, and was normalized to the cell area [52]. The number of invadopodia per cell was quantified manually using ImageJ and Adobe Photoshop by counting the number of spots that were positive for cortactin/Tks5 staining and gelatin degradation (Fig 2), or actin staining and gelatin degradation (S1k Fig). Transwell assays 24-well transwell chambers containing 8-micron diameter pores (Millipore) were coated with 0.3% gelatin prior to the seeding of cells. 2x10 4 BxPC3 cells cultured in the indicated conditioned medium were seeded on the top of the transwell. The bottom of the culture dish was filled with culture medium containing 10% FBS to promote invasion over 6 h. After 6 h, the cells on the top and bottom of the permeable membrane were fixed and stained with DAPI. Cell invasiveness was determined by measuring the number of DAPI-positive nuclei at the top versus the bottom of the membrane.
Statistical analysis. Data were analyzed using Microsoft Excel, and are represented as the mean +/-standard error. A two-tailed unpaired student's t-test was used to calculate statistical significance, with p<0.05 indicating a statistically significant difference.