Progenitor Cells Derived From Injured Cartilage Surface Respond to Damage Associated Molecular Patterns


 Objective: To elucidate how chondrogenic progenitor cells (CPCs) originated from mechanically injured cartilage surface respond to proinflammatory endogenous damage-associated-molecular-patterns (DAMPs). Design: Passage 1 bovine CPCs and non-CPCs isolated from injured articular cartilage either by blunt impaction or by scratches were treated with mitochondrial DAMPs (MTDs) composed of fMLF and CpG DNA, or HMGB1 (a nuclear DAMP), or IL-1b for 24 hrs. At the end of the experiments, the expression levels of matrix metalloproteinases (MMPs), chemokines, and cytokines that are associated with cartilage degeneration was examined with Western blotting and quantitative PCR. Results: Both HMGB1 and MTDs remarkably up-regulated expression level of pro-MMP-13 protein in CPCs while showed weak effect on non-CPCs. Compared to non-CPCs, CPCs expressed significantly higher baseline mRNA levels of MMP-13, CXCL12, and IL-6. MTDs further increased the expression levels of MMP-13 and IL-6 in CPCs while HMGB1 did not show such effect. When treated with MTDs, CPCs also expressed significantly higher levels of IL-8 mRNA than did non-CPCs. However, compared to non-CPCs, CPCs expressed much lower levels of MMP-3 and active MMP-13 proteins as well as lower mRNA levels of CCL2 and IL-6 in response to IL-1b. Conclusions: CPCs were more sensitive than non-CPCs in response to DAMPs, especially MTDs, to up-regulate expression of MMP-13, IL-6 and -8. When IL-1b was present, CPCs were less responsive than non-CPCs in terms of up-regulating MMP-3, CCL2, and IL-6 expression. The proinflammatory nature of CPCs implied their critical role in the early phase of PTOA development.


Introduction
Trauma to weight-bearing joint leads to post-traumatic osteoarthritis (PTOA). Focal mechanical damage in articular cartilage results in irreversible degradation of the tissue. Cartilage degeneration is partially responsible for the typical symptoms including joint pain and stiffness observed in patients diagnosed with PTOA. Extracellular matrix (ECM) destruction and cell death are the cause of cartilage degeneration [1][2][3][4][5][6][7]. How mechanical insults induce catabolic changes in cartilage cells remains poorly understood. Cells termed as "chondrocytes" have long been thought as the only cell type in articular cartilage [2,8].
However, studies have showed that cartilage harbored chondrogenic progenitor cells that could be induced to differentiate into mature chondrocytes. Dowthwaite and colleagues reported that there was 1-2% of chondrocyte population residing in the surface zone of articular cartilage of 7-day-old calves were progenitor-like cells which possessed a high colony-forming e ciency and could differentiate into chondrocytes [9]. Furthermore, they discovered chondrogenic progenitor cells (CPCs) in normal human femoral condyle cartilage which accounted for 0.7% of total cell population [10].
Correspondingly, our group discovered that injury to cartilage surface could induce migration of CPCs into the injury site and those cells could repopulate the damaged area. Migrating CPCs expressed high level of PRG4 indicating that they originated from the super cial zone. However, compared to mesenchymal stem cells (MSCs) or normal chondrocytes, CPCs expressed higher levels of cytokines, chemokines and MMPs involved in cartilage in ammation [11].
Instant cell death, a critical form of cartilage injury, immediately follows trauma to the joint, which may create pro-in ammatory environments. Several ex vivo studies revealed that signi cant amount of cell death quickly following mechanical insults was observed in the super cial tangential zone and then turned into a slower propagating "wave of cell death" [12][13][14][15]. Those observations were later con rmed in in vivo studies that revealed even more profound effect. They showed complete loss of cells spanning the full-thickness of injured cartilage only weeks after the insult [7,16,17]. Dead cells spill intracellular contents, some of which are proin ammatory and capable of priming immune cells including dendritic cells, T cells and macrophages [18]. Those derived endogenously from stressed or injured tissues are termed as "alarmins" or endogenous danger/damage-associated molecular patterns (DAMPs) [19,20].
DAMPs may come from ruptured mitochondria. Hajizadeh et al. reported that mitochondrial DNA (mtDNA) containing unmethylated CpG motif was detected in synovial uids of 70% patients diagnosed with rheumatoid arthritis (RA). Rheumatoid factor (RF) was detected with signi cantly higher frequency in patients whose synovial uid samples were mtDNA-positive [21]. Moreover, Dong et al. and Collins et al. reported that mtDNA was in ammatogenic both in vitro and in vivo and its proin ammatory effect was mediated by monocytes/macrophages, NF-kB, and TNF-a [22,23]. Zhang and colleagues further discovered that mtDNA and formyl peptides formed mitochondrial DAMPs (MTDs) which could trigger innate immunity leading to neutrophil-mediated organ injury usually observed in a systemic in ammatory response syndrome (SIRS) [24].
In addition to MTDs, high mobility group box 1 protein (HMGB1) is a nuclear DAMP (NuD) . HMGB1 is a non-histone protein and involved in DNA organization and regulation of transcription. Recent study showed that HMGB1 not only mediated sepsis but also played a critical role in pathophysiology of trauma, autoimmunity, ischemia perfusion injury, and cancer [25]. HMGB1 was demonstrated as a systemic in ammation mediator in a murine fracture model. It signaled through membrane toll-like receptor 4 (TLR4) located on the apical surface of intestinal cells to elicit in ammatory response and endorgan injury following bilateral femur fracture [26]. Our previous study revealed that HMGB1 synergized with IL-1b or Fn-f on up-regulation of ECM degrading enzymes including MMP-3, -13, ADAMTS-5, ADAM-8, and iNOS in articular chondrocytes [27].
However, it still remains unclear whether CPCs respond differently from remaining chondrocytes to DAMPs released from dead cells following joint trauma. We hypothesized that in injured cartilage CPCs were more responsive to DAMPs than the remaining cell population in terms of up-regulating expression of MMPs, chemokines, or cytokines involved in cartilage in ammatory degradation. To test this hypothesis, we employed an ex vivo cartilage injury model and established cultures of CPCs and non-CPCs, respectively. We then compared the expression levels of MMP-3, -13, CXCL12, CCL2, IL-6, and IL-8 in those cultures in response to MTDs or HMGB1 stimulation.

Establishment of cartilage surface injury ex vivo model
Bovine osteochondral explants (2.0-2.5 cm W X 2.0-2.5 cm L X 0.5-1.0 cm H) were sawed under sterile conditions from medial or lateral tibial plateau of sti e joints of 18-month-old cows obtained from a local abattoir (Bud's Custom Meats, Inc., Riverside, IA, USA). Explants were pre-equilibrated for 48 hrs in serum free DMEM/F12 supplemented with 50 U/ml Penicillin, 50 mg/l Streptomycin, and 2.5mg/l Amphotericin B inside a humidi ed 37 C incubator supplied with 5% CO 2 and 5% O 2 .
To create an impact injury, cartilage surface was subjected to a single blunt impaction with a customized drop-tower reported in our previous studies [11,28]. Brie y, an osteochondral explant was rigidly xed in a chamber with a 5.0 mm diameter brass rod resting on the surface of cartilage. A 2-kg mass was dropped from 14 cm height onto the brass rod resulting in impact energy density of 14 J/cm 2 . This high energy impact created cracks and instant cell death in cartilage surface. To create a scratch injury, a sterile 26 G1/2 needle was dragged over cartilage surface. An X-shaped matrix tear in the super cial zone was generated.
To harvest non-CPCs, full thickness cartilage slices were shaved from osteochondral explants immediately after CPCs were harvested, and were weighed, minced, and subjected to 0.4% protease (Sigma-Aldrich ® , St. Louis, MO) for 90 mins and then to 0.02% collagenase (Sigma-Aldrich ® , St. Louis, MO) for 16 hrs to release chondrocytes. Same method was used in isolation of normal articular chondrocytes (ACs) from uninjured cartilage. Passage 1 CPCs, or non-CPCs, or ACs were seeded at a high density (3.0 X 10 5 cells/cm 2 ). The experimental design was summarized and illustrated in Figure 1.

Light microscopy and calculation of cell population doubling time (PDT)
The growth status of CPCs, or non-CPCs, or ACs was recorded with an Olympus CKX53 light microscope (Olympus Corporation of the Americas, PA, USA) equipped with a camera (Olympus Soft Imaging Solutions GMBH, Olympus LC30, Munster, Germany). PDT of CPCs, or non-CPCs, or ACs was calculated from total cell counts obtained at the seeding time of passage 0 and 1 cells, respectively.

Determination of MMP expression with Western blotting
Conditioned media were dialyzed against MilliQ water and then concentrated to dry powder with a speed vacuum. Equal volume of prepared reduced medium samples was resolved by 10% SDS-PAGE. Proteins were then blotted onto a nitrocellulose membrane. After been blocked with 3% BSA/TBS, blots were

Statistical analysis
Relative gene expression to β-actin was calculated using a comparative Ct method (2 −ΔΔCt ). Results are reported as mean ± standard deviation (SD). Relative expression of each target gene in CPCs from impact or scratch injury was compared to that in non-CPCs with Welch's t-test. P (T<=t) two-tail value was reported. P < 0.05 was considered statistically signi cant. The 95% con dence interval (CI) was also calculated by using this formula: CI = Mean ± t X (SD/√sample size). The sample size was determined by n = (z 2 × σ 2 )/MOE 2 .

Results
Compared to non-CPCs, CPCs displayed remarkably different morphology and much shorter PDT At day 1 post-seeding, passage 0 CPCs isolated either from blunt impacted or scratched cartilage displayed elongated spindle shape while corresponding non-CPCs showed rounded shape. CPCs was also notably larger than non-CPCs (Figure 2A-C). PDT was 1.7 days for CPCs from impact injury and 1.9 days for CPCs from scratch injury, both which was much shorter than that for non-CPCs (6 days) or ACs (3.6 days) ( Figure 2D). In cartilage of an osteochondral explant, CPCs only accounted for 9.5% while non-CPCs for 90.5% of the whole cell population.
CPCs were more responsive than non-CPCs to DAMPs stimulation in terms of up-regulation of MMP-13 expression while non-CPCs expressed moderately more MMP-3 mRNA than did CPCs When left untreated, CPCs from impact injury expressed more MMP-13 proform (pro-MMP-13) than did non-CPCs. When either DAMPs were present, CPCs secreted signi cantly more pro-MMP-13 than did non-CPCs with CPCs from impact injury showing stronger response than those from scratch injury ( Figure 3A, top blot).
Unlike MMP-13, MMP-3 protein expression was only detected in the group treated with IL-1b. Compared to non-CPCs, CPCs secreted much less MMP-3 ( Figure 3, bottom blot). Consistently, the fold increase of MMP-3 mRNA expression in groups either untreated or treated with DAMPs was less than 3 while in the group treated with IL-1b the fold increase was over 200. CPCs expressed much less MMP-3 mRNA than did non-CPCs when treated with IL-1b ( Figure 3D&E).
CPCs expressed higher levels of CXCL12 than did non-CPCs in response to DAMPs; however, CPCs expressed less CCL2 than did non-CPCs in response to IL-1b When DAMPs were added into cultures, CPCs from either source expressed more CXCL12 mRNA than did . Nonetheless, similar expression difference was detected in non-treated group (P = 0.001 for CPCs from scratch injury vs. non-CPCs; P = 0.004 for CPCs from impact injury vs. non-CPCs). This baseline expression difference was not enhanced by IL-1b ( Figure 4A).

Discussion
We previously reported a group of CPCs in the super cial zone of mechanically injured articular cartilage in weight-bearing joint. Compared to normal chondrocytes, those cells were migratory in response to de ned or unde ned chemotactic factors resulted from cartilage damage and cell death. One de ned chemotactic factor examined in that study was HMGB1 that is a NuD and can be passively released from the dead cells in damaged cartilage [11]. In order to test our hypothesis that CPCs were more responsive to DAMPs than the remaining cell population in terms of up-regulating expression of MMPs, chemokines, or cytokines involved in cartilage in ammatory degradation, we examined and compared the expression levels of MMPs, chemokines, and cytokines involved in PTOA pathogenesis between CPCs and non-CPCs in response to MTDs or NuD, two major proin ammatory DAMPs. Our study revealed that CPCs were morphologically different from non-CPCs in monolayer cultures, more active in proliferation, and more responsive to DAMPs, especially MTDs, than non-CPCs to up-regulate in ammatory genes including MMP-13, IL-6, and IL-8. To our knowledge, this is the rst investigation into the in ammatory response to DAMPs by CPCs emerged from mechanically injured articular cartilage.
Compared to non-CPCs in mechanically injured cartilage, CPCs expressed signi cantly more MMP-13 at both mRNA and protein levels in response to HMGB1 or MTD stimulation. MMP-13 is a zinc-dependent collagenase that is a major cartilage ECM degrading metalloproteinase implicated in OA pathogenesis. The ECM of articular cartilage is mainly composed of type II collagen and proteoglycan. MMP-13 also named as collagenase-3 can speci cally cleave type II collagen triple helix into characteristic ¾ and ¼ fragments [30][31][32]. Besides type II collagen, MMP-13 degrades other types of collagen, such as type IV and IX, and proteoglycan in cartilage ECM as well [33]. Secreted as a zymogen by chondrocytes, synoviocytes, or immune cells in a normal or an in amed joint, the proform of MMP-13 consists of 471 amino acids and is 53.8 kDa in size. The removal of the N-terminal 84 amino acids by other MMPs, such as MMP-3, exposes the catalytic domain of MMP-13 and turns it into the active form that is 48 kDa in size [34]. Due to glycosylation modi cation, pro-MMP-13 displayed a MW around 64 kDa and the active form was around 54 kDa when their expressions were examined with Western blotting under reducing conditions as seen in our study [35].
Although we observed much higher levels of pro-MMP-13 produced by CPCs than non-CPCs when either DAMPs were present in the cultures, we did not detect any active form of this metalloprotease in the conditioned media of CPCs. This result was well backed up by observations that MMP-3 protein secretion by CPCs was undetectable when either DAMPs were present in the cultures. On the other hand, MMP-3 expression could be strongly induced by IL-1b in either CPCs or non-CPCs cultures accompanied by the expression of both pro-and active MMP-13. This implied that MMP-3 might be the major protease responsible for the removal of the N-terminal peptide from pro-MMP-13 structure and generation of the active form of MMP-13 in CPCs and non-CPCs. Moreover, the observation of MMP-3 expression in either CPCs or non-CPCs could only be signi cantly upregulated by IL-1b not by HMGB1 or MTDs was consistent with our previous discovery showing that those types of DAMPs could not upregulate expression of MMP-3 in normal chondrocytes but either of them could synergize with proin ammatory cytokines in up-regulating MMP-3 and other cartilage-degrading proteases [27]. Interestingly, we also observed that CPCs expressed much lower levels of MMP-3 mRNA and protein than did non-CPCs when IL-1b was present in the cultures. This may suggest that the majority of MMP-3 was secreted by non-CPCs rather than CPCs whose cell population was only one tenth of that of non-CPCs and mainly resided in the surface zone of the injured cartilage.
Studies have shown that chemokine CXCL12 or stromal cell-derived factor-1a (SDF-1a) is one of the upstream effectors of MMPs in OA pathogenesis. Kanbe et al. reported that in OA patients CXCL12 secreted by synovial broblasts could up-regulate expression of MMP-3 in chondrocytes via interaction with its speci c receptor CXCR4 [36]. Wang and colleagues further showed that MMP-3 up-regulation and cartilage degradation induced by CXCL12-CXCR4 axis could be attenuated by AMD3100, a CXCR4 antagonist [37]. Our data showed that in mechanically injured cartilage CPCs expressed ~110-fold higher baseline levels of CXCL12 mRNA than did non-CPCs. This result was consistent with our previous ndings that CPCs expressed 28-fold higher CXCL12 than did normal chondrocytes or 12-fold higher than did mesenchymal stem cells (MSCs) [11]. However, the relatively higher CXCL12 mRNA expression in CPCs we observed was not accompanied by higher MMP-3 mRNA level. This might imply that chemokine CXCL12 was not a main inducer of MMP-3 expression in CPCs emerged from mechanically injured cartilage, i.e. CXCL12 could not up-regulate MMP-3 in CPCs via autocrine mode. We also observed that neither HMGB1 nor MTDs further enhanced the fold increase of CXCL12 mRNA expression in CPCs over non-CPCs. The fold increase of CXCL12 mRNA expression was only slightly enhanced by IL-1b. This indicated that those two types of DAMPs or IL-1b might have little effect on further up-regulating CXCL12 mRNA expression in CPCs which already showed much higher baseline expression than did non-CPCs.
In addition to CXCL12, chemokine CCL2 was implicated in OA pathogenesis since it could recruit monocytes to cause in ammatory damage of cartilage [38]. In our study, MTDs enhanced CCL2 mRNA expression in CPCs but to a much lesser extent than did IL-1b which greatly increased the expression by more than 100-fold. Moreover, when IL-1b was present, CPCs expressed signi cantly less CCL2 mRNA than did non-CPCs, which was comparable to MMP-3 expression. This concurrent up-regulation of CCL2 and MMP-3 by IL-1b observed mainly in non-CPCs was consistent with what Robert et al. reported in hepatic stellate cells involved in chronic liver in ammation [39]. Interestingly, Zarebska and colleagues discovered that CCL2 (-/-) mice displayed strongly suppressed MMP-3 expression in joint tissue 6 hrs postdestabilization of the medial meniscus (DMM) surgical procedure which is frequently employed to create PTOA in an animal model [40]. However, more evidence is required to verify whether MMP-3 could be upregulated by CCL2 in CPCs in mechanically injured cartilage via autocrine pathway.
As one of the main proin ammatory cytokines involved in OA pathophysiology, IL-6 level elevates in synovial uids and serum of OA patients and this elevation is correlated with up-regulation MMP-1 and -13 expression and with reduction of type II collagen synthesis which result in cartilage damage [41]. Our data indicated that CPCs expressed signi cantly higher baseline level of IL-6 mRNA than did non-CPCs.
This result was consistent with our previous ndings made from comparing baseline IL-6 expression between CPCs and normal chondrocytes [11]. Furthermore, we observed that MTDs remarkably enhanced IL-6 expression in CPCs. Similarly, Schwacha et al. and Hu et al. reported that MTDs could induce IL-6 expression in gammadelta T cells which play critical role in sterile in ammation like OA and could stimulate hepatocytes to increasingly produce IL-6 leading to liver injury, respectively [42,43]. Similar to CCL2 and MMP-3, when IL-1b was present, CPCs transcribed less IL-6 mRNA than did non-CPCs. Those results implied that, in response to IL-1b, different signaling mechanisms were employed by CPCs and non-CPCs to up-regulate expression of IL-6, CCL2 and MMP-3 genes.
Studies showed that chondrocytes could secret IL-8, another member of interleukins, which attracted neutrophils to joint tissue and caused degranulation of neutrophils leading to cartilage degradation [44,45]. A recent study further demonstrated that serum IL-8 level positively associated with knee OA symptoms, with levels of cartilage degradation markers, with up-regulation of MMP-3 & -13, and with radiographic characteristics [46]. Our data indicated MTDs could signi cantly up-regulate IL-8 mRNA expression in CPCs, not in non-CPCs. This suggested that in the early phase of cartilage injury CPCs might be the major player that attracts neutrophils to the injury site via IL-8 production upon stimulation mainly by MTDs released from dead cells. IL-8 might act on CPCs via autocrine pathway to up-regulate MMP-13 expression in response to MTD stimulation.
Taken together, our study demonstrated that CPCs emerged from the super cial zone of mechanically injured articular cartilage and accounted for only ~10% of total cell population not only divided two times faster than did non-CPCs but also expressed signi cantly higher baseline mRNA levels of MMP-13, CXCL12, and IL-6. More importantly, CPCs responded to endogenous DAMPs differently from non-CPCs.
Between two types of DAMPs examined in this study, MTDs were more stimulatory than HMGB1 to CPCs in terms of up-regulating expression of MMP-13, IL-6, and -8. Those ndings shed light on the dynamics of in ammatory response inside mechanically injured articular cartilage which may lead to PTOA. Future studies are needed to elucidate the complicated role of CPCs playing in the early phase of PTOA development.

Conclusions
Mechanical injury to articular cartilage surface stimulated migration of CPCs to the lesion. Due to this migratory property, CPCs could be easily separated from non-CPCs in injured cartilage with trypsin treatment. In addition to shorter PDT and more elongated shape in monolayer cultures, CPCs displayed were more sensitive than non-CPCs in response to DAMPs, especially to MTDs, that were released from damaged mitochondria and nuclei. Compared to non-CPCs, signi cant up-regulation of MMP-13, IL-6 and -8 was detected in CPCs when challenged by MTDs. Nonetheless, CPCs were less responsive to IL-1b than did non-CPCs in terms of up-regulating MMP-3, CCL2, and IL-6 expression. The proin ammatory nature of CPCs implied their critical role in the early phase of PTOA development.

Consent for publication
Not applicable.

Availability of supporting data
All data generated or analyzed during this study are included in this published article.