Protein Kinase Cζ Is Up-regulated in Osteoarthritic Cartilage and Is Required for Activation of NF-κB by Tumor Necrosis Factor and Interleukin-1 in Articular Chondrocytes*

Protein kinase Cζ (PKCζ) is an intracellular serine/threonine protein kinase that has been implicated in the signaling pathways for certain inflammatory cytokines, including interleukin-1 (IL-1) and tumor necrosis factor α (TNF-α), in some cell types. A study of gene expression in articular chondrocytes from osteoarthritis (OA) patients revealed that PKCζ is transcriptionally up-regulated in human OA articular cartilage clinical samples. This finding led to the hypothesis that PKCζ may be an important signaling component of cytokine-mediated cartilage matrix destruction in articular chondrocytes, believed to be an underlying factor in the pathophysiology of OA. IL-1 treatment of chondrocytes in culture resulted in rapidly increased phosphorylation of PKCζ, implicating PKCζ activation in the signaling pathway. Chondrocyte cell-based assays were used to evaluate the contribution of PKCζ activity in NF-κB activation and extracellular matrix degradation mediated by IL-1, TNF, or sphingomyelinase. In primary chondrocytes, IL-1 and TNF-α caused an increase in NF-κB activity resulting in induction of aggrecanase-1 and aggrecanase-2 expression, with consequent increased proteoglycan degradation. This effect was blocked by the pan-specific PKC inhibitors RO 31-8220 and bisindolylmaleimide I, partially blocked by Gö 6976, and was unaffected by the PKCζ-sparing inhibitor calphostin C. A cell-permeable PKCζ pseudosubstrate peptide inhibitor was capable of blocking TNFand IL-1-mediated NF-κB activation and proteoglycan degradation in chondrocyte pellet cultures. In addition, overexpression of a dominant negative PKCζ protein effectively prevented cytokine-mediated NF-κB activation in primary chondrocytes. These data implicate PKCζ as a necessary component of the IL-1 and TNF signaling pathways in chondrocytes that result in catabolic destruction of extracellular matrix proteins in osteoarthritic cartilage.

joint disease in man (1). More than 20 million individuals in the United States alone have symptomatic OA, and it has been estimated that many more, greater than 50% of people over 65 years of age and 80% of those over age 75, have radiographic evidence of this disease (2). Yet, despite the widespread incidence of the disease in the human population, the etiology of OA is still largely unknown. OA is characterized by a slow focal destruction of articular cartilage, causing a roughening and thinning of the weight-bearing regions of the articular surface resulting in progressive immobility and pain. Articular cartilage is the tissue that provides shock-absorptive resiliency as well as low friction articulation to joints. It contains only a single cell type, the chondrocyte, which is responsible for the homeostasis of the tissue by synthesizing extracellular matrix that surrounds the cells and provides the important biophysical characteristics of the tissue. However, chondrocytes are also capable of producing catabolic factors capable of destroying the matrix components. Thus, extracellular matrix synthesis and degradation are dynamic processes that must be balanced by the chondrocytes for proper homeostasis of the tissue. In osteoarthritic cartilage, this balance appears to be shifted toward degradation, resulting in progressive loss of matrix due to up-regulation of proteolytic activities such as matrix metalloproteinases and aggrecanases. There is some debate whether osteoarthritis is a noninflammatory arthrosis or an inflammatory arthritis; however, synovial inflammation has been documented in OA (3) and inflammatory cytokines, especially interleukin-1 (IL-1) and tumor necrosis factor (TNF), have been implicated as important mediators of the disease (4 -8). These proinflammatory cytokines are known to be major regulators of chondrocytic expression of downstream proteases (such as aggrecanases and collagenases) that are ultimately involved in matrix breakdown resulting in the formation of osteoarthritic lesions (6,8). Therefore, these cytokines themselves, or components of their intracellular signaling pathways, constitute possible therapeutic intervention points that could mitigate the destruction of articular cartilage in OA.
One group of signaling proteins that are shared by both the IL-1 and TNF signaling pathways are the members of the protein kinase C (PKC) family of intracellular serine/threonine kinases. In the course of their convergence on the activation of the nuclear factor B (NF-B) transcription factor, both the IL-1 and TNF pathways reportedly signal through PKC family members in various cell types (9 -13). The PKC family is made up of several isoforms that are divided into three basic classes ("conventional," "novel," and "atypical"), by virtue of the structure of their regulatory domains and (consequently) their methods of activation (14). The conventional or cPKC isoforms contain two characteristic membrane targeting domains called C1 and C2 that are capable of binding diacylglycerol (or the synthetic analog phorbol ester) or calcium, respectively, ultimately resulting in activation of the kinase. The novel or nPKC family members also contain these domains, but they are reversed in their orientation, and the C2 is modified such that it is unresponsive to calcium. The human atypical or aPKC group consists of only two members, and (called in mouse), both of which lack a C2 domain and possess only a modified C1 domain. Atypical PKC isoforms are insensitive to both diacylglycerol and calcium but are activated by phosphatidylserine.
In an effort to gain understanding of the pathological processes that underlie the development and progression of OA, we performed experiments to identify transcriptional alterations in articular chondrocytes that distinguish patients with end stage OA from normal subjects. This work revealed that PKC was the only member of the protein kinase C family with dysregulated expression in chondrocytes from human osteoarthritic cartilage. PKC has been implicated previously in the NF-B signaling pathway in some cell types, but its function in chondrocytes has not been well characterized. In this study, we utilized chondrocyte-based assay systems to investigate the role of PKC on TNF-and IL-1-mediated activation of NF-B, and on the consequent proteoglycan degradation that results from the activation of the NF-B pathway. We find that NF-B activation by TNF and IL-1 in chondrocytes requires PKC activity and, most importantly, that inhibition of PKC blocks cytokinemediated up-regulation of aggrecanase expression and the resulting destruction of articular cartilage extracellular matrix proteoglycans that is a hallmark of the OA disease process. Thus, PKC constitutes a pivotal signaling molecule in the catabolic pathways initiated by the proinflammatory cytokines IL-1 and TNF in articular chondrocytes and may represent an important therapeutic target.
Primary Chondrocyte Isolation and Culture-Bovine cartilage was obtained from the metacarpophalangeal joint of calves (2-10 days old), and chondrocytes were isolated by serial enzymatic digestion using Pronase (1 mg/ml, 37°C for 30 min) and collagenase (1 mg/ml, 37°C for overnight) in Dulbecco's modified Eagle's medium (DMEM) with 10 mM HEPES, and 100 units/ml penicillin, 100 g/ml streptomycin. The digest was filtered through a 70-m nylon cell strainer (Falcon) and processed as described (24). Cells were resuspended in growth media (HL-1 media, Cambrex catalog number 77201) containing 2 M L-glutamine, 50 g/ml ascorbate, antibiotics, and 10% fetal bovine serum (FBS) and aliquoted into 15-ml Falcon centrifuge tubes at a concentration of 1 ϫ 10 6 cells per tube. The tubes were centrifuged at 200 ϫ g at room temperature to allow formation of cell pellets and subsequently incubated at 37°C in a humidified atmosphere with low oxygen tension (5% O 2 ) to retain their differentiated chondrocytic phenotype and to maximize the production of extracellular matrix (25). For inhibitor studies, chondrocyte pellets were weaned from the serum gradually by replacing the media every 3-4 days with decreasing concentrations of FBS (5, 2.5, and 0%). Chondrocytes were then cultured in pellet form for 3 weeks in the absence of serum to allow the accumulation of the proteoglycan-and collagen-containing extracellular matrix. Chondrocyte cell pellets were preincubated for 2 h with inhibitors prior to cytokine stimulation for 18 h. The level of sulfated glycosaminoglycan in the culture media and cartilage/pellet extracts was determined by the dimethylmethylene blue (DMMB) assay (26). Shark and whale chondroitin sulfate (Fluka Biochemika, Switzerland) was used as a standard. Cytotoxicity testing was performed for each of the inhibitors at the concentrations used in the pellet culture assays by measuring lactate levels in culture media as an indicator of cellular metabolism and viability using a kit from Sigma.
Isolation of RNA from Primary Cartilage Tissue and from Chondrocytes in Culture-RNA was isolated from human osteoarthritic articular cartilage samples obtained from patients (n ϭ 18, mean age ϭ 66.2 years, range 49 -84 years) undergoing total knee replacement surgery (New England Bap-tist Hospital) or from nonosteoarthritic cartilage obtained from above-knee amputations (n ϭ 10, mean age ϭ 71.6 years, range 43-100) (Clinomics). The OA cartilage samples were obtained as whole joints within 2 h of surgery, and the articular cartilage was shaved from the joint surfaces taking great care to avoid any pannus, fibrotic tissues, subchondral bone, and other noncartilaginous regions of the joint (27). Nonosteoarthritic cartilage samples were obtained from individuals without a clinical diagnosis nor symptoms of OA, and the specimens were evaluated histologically to confirm the classification prior to inclusion in this study. Cartilage pieces were flash-frozen in liquid nitrogen and stored at Ϫ80°C until processed for RNA isolation. The frozen cartilage was pulverized using a Spex Certiprep freezer mill (model 6750) at 15 Hz two times for 1 min each under liquid nitrogen. The frozen powdered cartilage was resuspended in 4 M guanidinium isothiocyanate (Invitrogen) containing 8.9 mM 2-mercaptoethanol and homogenized on ice with a Polytron homogenizer at maximum speed setting twice for 1 min each time, with a 1 min "rest" between homogenizations. The homogenate was centrifuged at 1500 ϫ g for 10 min, and the supernatant was saved. The gelatinous pellet was resuspended in guanidinium isothiocyanate/2-mercaptoethanol and homogenized a second time as described above. The pellet was then discarded, and the two resulting supernatant fractions were combined and incubated with Triton X-100 (2% final concentration) and sodium acetate (pH 5.5, 1.5 M final concentration) sequentially for 15 min each. The samples were extracted once with an equal volume of acid phenol chloroform, pH 4.5, and twice with acid phenol, pH 4.5, phenol, pH 7.5, chloroform mix (1:1:1). RNA was subsequently precipitated by the addition of isopropyl alcohol, and further purified using an RNeasy mini Kit (Qiagen, Valencia, CA) according to the manufacturer's protocol. RNA quantity and purity were measured by ultraviolet absorbance at A 260 /A 280 , and RNA quality was assessed by the RNA6000 assay using the Agilent BioAnalyzer 2100 (Palo Alto, CA). RNA yields averaged between 5 and 10 g of total RNA per g of cartilage tissue.
For isolation of RNA from chondrocyte pellet cultures and from chondrocytes in monolayer culture, no pulverization was required. Pellets were digested with collagenase (2.5 mg/ml, Sigma), and RNA was subsequently prepared using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. Primary chondrocytes and chondrocytic cell lines in monolayer culture were lysed by direct addition of TRIzol reagent followed by standard TRIzol RNA purification methodologies.
Microarray Analysis of Osteoarthritic and Normal Cartilage-Gene expression changes in RNA from lesional (n ϭ 14) and adjacent nonlesional (n ϭ 13) osteoarthritic cartilage compared with nonosteoarthritic cartilage (n ϭ 10) were analyzed using the Human Genome U95Av2 (HG-U95Av2) GeneChip Array (Affymetrix, Santa Clara, CA) for expression profiling. The HG-U95Av2 chip contains 25-mer oligonucleotide probes representing ϳ12,000 primarily full-length sequences (ϳ16 probe pairs/sequence) derived from the human genome. For each probe that is designed to be perfectly complementary to a target sequence, a partner probe is generated that is identical except for a single base mismatch in its center. These probe pairs allow for signal quantitation and subtraction of nonspecific noise.
RNA was extracted from individual articular cartilage tissue samples, converted to biotinylated cRNA, and fragmented according to the Affymetrix protocol. The fragmented cRNAs were diluted in 1ϫ MES buffer containing 100 g/ml herring sperm DNA and 500 g/ml acetylated bovine serum albumin and denatured for 5 min at 99°C followed immediately by 5 min at 45°C. Insoluble material was removed from the hybridization mixtures by a brief centrifugation, and the hybridization mixture was added to each array and incubated at 45°C for 16 h with continuous rotation at 60 rpm. After incubation, the hybridization mixture was removed, and the chips were extensively washed and stained with streptavidin (R)-phycoerythrin (Molecular Probes, Eugene, OR) using the GeneChip Fluidics Station 400 following the manufacturer's specifications. The raw fluorescent intensity value of each transcript was measured at a resolution of 6 m with a Hewlett-Packard Gene Array Scanner. GeneChip software 3.2 (Affymetrix), which uses an algorithm to determine whether a gene is "present" or "absent," as well as the specific hybridization intensity values or "average differences" of each gene on the array, was used to evaluate the fluorescent data. The average difference for each gene was normalized to frequency values by referral to the average differences of 11 control transcripts of known abundance that were spiked into each hybridization mixture according to the procedure of Hill et al. (28). First, the frequency of each gene was calculated and represents a value equal to the total number of individual gene transcripts per 10 6 total transcripts. Transcripts which were called present by the GeneChip software in at least one of the arrays for both arthritis and normal cartilage, were included in the analysis. Second, for comparison between arthritis and normal cartilage, a t test was applied to identify the subset of transcripts that had a significant ( p Ͻ 0.05) increase or decrease in frequency values. Third, average fold changes in frequency values across the statistically significant subset of transcripts were required to be 2.0-fold or greater. These criteria were established based upon replicate experiments that estimated the intra-array reproducibility.
Quantitative RT-PCR (TaqMan)-RNA for TaqMan analysis (ABI PRISM 7700 sequence detection; PerkinElmer Life Sciences) was isolated from primary cartilage tissue as described above and then further purified with two more rounds of phenol/chloroform extraction followed by RNeasy (Qiagen) column binding and elution. To ensure the elimination of genomic DNA, RNA was treated with DNase (Qiagen) during RNeasy column purification (as recommended by the supplier), and following the RNA purification any residual genomic DNA was removed using DNA-free (Ambion, Austin, TX), following that manufacturer's instructions.
For comparison of PKC and PKC mRNA levels in human OA chondrocytes, human primary chondrocytes were isolated as described above for bovine primary chondrocytes. The primary chondrocytes were plated in a 24-well format (2 ϫ 10 6 / well) in DMEM/F-12 ϩ 10% FBS ϩ 100 units/ml penicillin, 100 g/ml streptomycin for 2 days. The cells were lysed, and RNA was isolated using the RNeasy mini kit (Qiagen). 200 ng of RNA was used for each TaqMan assay in the TaqMan one-step RT-PCR method (Applied Biosystems). The human PKC and PKC probe/primer sets were "Assays on Demand" from Applied Biosystems, assay identification Hs00177051_m1 and Hs00702254_s1, respectively. Assay mixtures and cycling conditions were per the manufacturer's recommendation.
Phosphoprotein Analysis of PKC-For the chondrocyte cell line, T/C-28a2 cells (23) (kindly provided by Mary Goldring) were plated in a 24-well format in DMEM/F-12 ϩ 10% FBS ϩ 1% antimycotic antibiotic solution and grown to confluence. After the cells became confluent, they were changed to serumfree medium, allowed to adapt to serum-free conditions overnight, and were then stimulated with 20 ng/ml recombinant human IL-1␤ (R & D Systems) for varying lengths of time (0 -30 min). The cells were immediately lysed at the conclusion of each time point with 1ϫ Cell Lysis buffer (Cell Signaling Technologies). PKC was immunoprecipitated from the cell lysates with a monoclonal antibody to p62 Lck ligand (BD Transduction Laboratories). The immunoprecipitated proteins were then run on 10% SDS-PAGE under reducing conditions, and Western analysis was performed using either a monoclonal antibody that specifically recognizes PKC phosphorylated at position Thr 410 (Cell Signaling Technologies) or with a monoclonal antibody raised to the carboxyl-terminal 20 amino acids of PKC (C-20, Santa Cruz Biotechnology). The C-20 antibody recognizes total PKC regardless of its phosphorylation state. For bovine primary chondrocytes, primary bovine chondrocytes were isolated from fresh bovine metacarpophalangeal joints as described above. The primary chondrocytes were plated in a 24-well format (2 ϫ 10 6 cells/well) in DMEM/F-12 ϩ 10% FBS ϩ 1% antimycotic antibiotic solution for 2 days and then switched to serum-free media overnight prior to cytokine induction. The cells were treated with IL-1␣ (20 ng/ml) for various time points; the cells were lysed, and the lysate was analyzed as described above for the T/C-28a2 cells.
Western blots were visualized using a goat anti-mouse horseradish peroxidase conjugate followed by detection using the ECL Western blotting detection kit (Amersham Biosciences). Band intensities were determined by scanning the developed Western blots with a Gel Doc 2000 PC using the Quantity One quantitation software (Bio-Rad). The degree of phosphorylation for each sample was determined by calculating the ratio of phospho-PKC to total PKC.
Immunodetection of Aggrecan Cleavage Product-Detection of aggrecan cleavage sites in pellet culture conditioned media was performed using neoepitope monoclonal antibody Agg-C1 (anti-NITGE 373 , detects aggrecanase cleavage at aggrecan interglobular domain site (24)). Equal volumes of conditioned media from pellet cultures following incubation with or without cytokines and/or inhibitors were deglycosylated with chondroitinase ABC and keratanase (Calbiochem) and separated by 4 -12% NOVEX Tris-glycine gels (Invitrogen). Subsequently, the samples were electrophoretically transferred to Hybond membrane (Amersham Biosciences) and incubated with Agg-C1 antibody overnight at 4°C in TSA (50 mM Tris-Cl, pH 7.4; 0.2 M NaCl; 0.02% sodium azide). Unbound antibody was removed by washing the membrane in 1ϫ TSA buffer three times for 5 min, followed by incubation with goat anti-mouse IgG-alkaline phosphate conjugate secondary antibody (1:5000, Novagen) for 1 h at room temperature in 1ϫ TSA buffer. Following a second set of three 5-min washes in 1ϫ TSA, immunoreactive products were detected as described previously (28) by developing the Western blot with bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (Promega), and the image was digitized using a Hewlett-Packard flatbed scanner.
Quantitation of Atypical PKC Protein Levels in Human Chondrocytes-Specific rabbit polyclonal antibodies against PKC / and PKC were generated by immunizing rabbits with peptides spanning either residues 184 -234 (PKC /; GenBank TM accession number NM_008857.2) or 185-244 (PKC; GenBank TM accession number NM_008860.2). Primary human chondrocytes were isolated from live cartilage using Pronase and collagenase digestion as described above for bovine cartilage. Whole cell lysates from these primary human OA chondrocytes (ϳ150 g per sample) were size-fractionated by 10% SDS-PAGE and electroblotted onto a nitrocellulose membrane (Hybond ECL; Amersham Biosciences). The primary antibody reaction was performed overnight at 4°C. Unbound antibodies were removed by washing the nitrocellulose membrane three times for 15 min in washing buffer (PBS, pH 7.4, 0.1% Tween 20) at room temperature. Subsequently the membrane was incubated with secondary antibody (goat antirabbit horseradish peroxidase, 1:5,000; Dianova) for 2 h at room temperature followed by washing as described above. Antibodies were detected by chemiluminescence using ECL Western blotting detection reagents (Amersham Biosciences).
NF-B-Luciferase Cell Line Construction-To construct a chondrocytic cell line that stably expresses an NF-B-luciferase reporter gene, the T/C-28a2 cell line (23) was transfected with vectors pIRESpuro3 (catalog number 6986, Clontech) and pNF-B-Luc (catalog number 6053, Clontech), and cells were selected for resistance to puromycin. Cells that survived the selection were screened by a luciferase reporter assay (Promega) after IL-1␤ (10 ng/ml) induction. Puromycin-resistant T/C-28a2/NF-B-luciferase cell lines were first selected for response to IL-1␤ induction with a minimum signal/background ratio of 5 in the luciferase reporter assay (Promega). The clones that passed this primary screen were further tested using TNF-␣ (5 and 20 ng/ml)/IL-1␤ (5 and 20 ng/ml). In this secondary screening, a T/C-28a2/NF-B-luciferase clone was selected that showed the highest signal:background ratio and dose-dependent response to both TNF and IL-1 compared with other clones at the same cell density. Optimal conditions for using this cell line to test NF-B response to cytokines Ϯ inhibitors were determined empirically. The conditions found to be optimal were plating density of 30,000 cells/well in a 96-well format, 10 ng/ml IL-1␤ concentration, 25 ng/ml TNF-␣ concentration, 1 h of preincubation with inhibitors prior to cytokine induction, and a 3-h incubation with cytokines before luciferase assay. Cytotoxicity testing was performed for each of the inhibitors at the concentrations used in the NF-B-luciferase assays either by measuring lactate levels in culture media as a measure of cellular metabolism and viability using a kit from Sigma, or by directly measuring cellular proliferation using the WST-1 assay (Roche Applied Science).
Adenovirus Constructs and Infection Conditions-The vectors used to produce adenovirus were replication-defective human adenovirus type 5 with complete deletion of the E1a and E1b regions and partial deletion of the E3 regions of the viral genome. Separate adenoviral expression constructs were created that contained cDNAs encoding full-length active human PKC (FL-PKC; GenBank TM accession number Q05513); a mutant human PKC (DNϪ PKC) in which the alteration of a key residue in the ATP-binding site (K281W) results in a dominant negative kinase (29,30); an NF-B-luciferase reporter gene construct; and a green fluorescent protein virus as a transfection normalization standard (Ad-GFP). In experiments utilizing infection of primary bovine chondrocytes with the NF-B-luciferase reporter virus, cells were infected 4 days prior to the experiment with equal m.o.i. (typically 100 m.o.i.) that resulted in 100% infectivity as determined by titrated Ad-GFP infections. These m.o.i. levels were tested and confirmed to be noncytotoxic using the lactate assay. All constructs (except the NFB-luciferase reporter gene virus) expressed cDNAs under the control of the cytomegalovirus (CMV) promoter. These vectors were used to propagate recombinant viruses in human embryonic kidney cells (HEK293) (ATCC, Manassas, VA), which were then purified by two rounds of cesium chloride centrifugation (31). The purified virus was dialyzed in phosphate-buffered saline (PBS) and stored at Ϫ80°C in 10% glycerol in PBS at a concentration of 10 9 plaque-forming units/l. Adenovirus were generated, purified, and titered by ViraQuest Inc. (North Liberty, IA).

Protein Kinase C Is Transcriptionally Up-regulated in
Human Osteoarthritic Cartilage-Gene expression changes in articular chondrocytes from lesional (n ϭ 14) and adjacent nonlesional (n ϭ 13) osteoarthritic cartilages compared with nonosteoarthritic cartilages (n ϭ 10) were analyzed using Affymetrix GeneChip U95Av.2 arrays. Analysis of these data to identify the global transcriptional changes in articular chondrocytes that were associated with osteoarthritis is beyond the scope of this paper, and will be published separately. 3 However, a focused analysis of the data was performed to specifically identify OA-associated expression changes in protein kinase C family members. Probe sets for nine PKC isoforms were represented on the U95Av.2 array as follows:␣, ␤, ␥, ␦, ⑀, , , , and . Only four of these PKC isoforms, ␦, , ,and , were judged to be present by virtue of their hybridization intensity levels and the consistent hybridization performance of their probes on the arrays. Of these four PKC isoforms, only PKC␦ and PKC exceeded an empirically determined signal intensity threshold of 50 signal units allowing reliable quantitation of their transcript levels on the gene chips, and only PKC appeared to be transcriptionally altered in OA articular cartilage compared with normal articular cartilage ( Fig. 1A and data not shown). Transcript levels for PKC, the only other human atypical PKC isoform, were weakly detected on the gene chips and did not appear to show disease-related expression changes, but it was expressed at levels too low for accurate quantitation by this methodology. When the raw gene chip signal intensity for PKC hybridization was converted to normalized mRNA quantities (in parts per million) using an intrinsic standard curve on the chips (28), the PKC transcript levels were found to be ϳ2.5-3.5-fold higher in the nonlesional and lesional OA cartilage samples compared with normal cartilage (Fig. 1B). The up-regulation of PKC mRNA in the lesional OA samples reached statistical significance ( p Ͻ 0.004), but variability in the PKC values in the nonlesional samples limited the significance of the up-regulation in those samples on the gene chips. To confirm the gene chip results and to increase the sensitivity of the transcriptional analysis, PKC transcript levels were measured in RNA from these same human samples by TaqMan quantitative RT-PCR (Fig. 1C). These TaqMan data supported the gene chip results, confirming that PKC mRNA was up-regulated 3.5-fold in nonlesional OA articular cartilage ( p Ͻ 0.002 versus normal cartilage) and 2.0-fold in lesional OA articular cartilage ( p Ͻ 0.02 versus normal cartilage).
PKC Is the Predominant Atypical PKC Isoform in Articular Cartilage-Because PKC shares a high degree of sequence similarity with PKC (72% amino acid identity overall and even greater within the catalytic domain (32,33)), differentiating between PKC and PKC using most available biochemical reagents is difficult or impossible. Therefore, we were interested in determining how PKC transcript levels compared with PKC in an effort to judge the potential extent or proportion of the functional contribution of PKC in subsequent experiments. Because PKC mRNA levels were too low for accurate quantitation in the gene chip transcriptional profiling, more sensitive TaqMan assays were performed with probe/primer sets designed to distinguish PKC from PKC. These assays were first calibrated with known input amounts of PKC and PKC cDNA to allow direct comparison of transcript abun-dance between genes (data not shown). The correction factor based upon assay efficiencies in these calibration assays was 0.988 (PKC ϭ PKC/0.988). RNA was isolated from articular chondrocytes from three separate donors with end stage OA, and PKC and PKC transcript levels were measured in each donor sample by TaqMan Q-PCR. The threshold cycle (Ct) value for PKC averaged 9 cycles lower than for PKC ( Fig. 2A). In Fig. 2B, transcript abundance for each gene was converted to TaqMan units (raw Ct value normalized by comparison to GAPDH in the same samples), and the normalized value was corrected for assay efficiency by calculation of absolute transcript levels using the cDNA calibration curves. The result showed that the abundance of PKC mRNA in human OA articular cartilage was more than 800 times greater than PKC, which was detectable but present at consistently low levels in all three of the OA articular cartilage samples (Fig. 2B).
To determine whether the protein level for PKC in chondrocytes was as predominant compared with PKC as the mRNA levels would indicate, antibodies capable of distinguishing between PKC and PKC ("Experimental Procedures") were utilized. Cell lysates from human primary OA chondrocytes were run on separate 10% SDS-polyacrylamide gels and immunoblotted with either anti-PKC antibody, anti-PKC antibody, or the C-20 polyclonal antibody that recognizes both isoforms. These Western blots, shown in Fig. 2C, clearly show that PKC protein accounts for virtually all of the detectable aPKC protein in the human chondrocyte cell lysates, supporting the mRNA abundance data and providing additional evidence that PKC is not present to any significant extent in articular chondrocytes, whereas PKC is relatively abundant.
PKC Is Activated by IL-1 Signaling in Chondrocytes-The role of IL-1 in the initiation of extracellular matrix destruction in chondrocytes has been well established (24, 34 -37). IL-1 has been shown to trigger the phosphorylation of PKC in some cell types (38,39). To investigate whether PKC may be downstream of IL-1 signaling in chondrocytes, a human immortalized chondrocyte cell line T/C-28a2 (23) and pri-mary bovine articular chondrocytes were treated separately in culture with 20 ng/ml IL-1␤ or IL-1␣, respectively. We have shown previously that at equivalent doses the recombinant human IL-1␣ is more potent than recombinant human IL-1␤ on bovine chondrocytes in terms of activation of NF-B and degradation of matrix proteoglycan (24), whereas human-and porcine-derived chondrocytes show greater response to IL-1␤. 4 Because IL-1␣ and IL-␤ signal through the same cell-surface receptor and elicit the same downstream responses (40), we assumed that the differences in potency reflected differences in cross-species binding affinities for the two isoforms of IL-1 to the cognate receptors and therefore chose the most active cytokine for each system. At various time points up to 10 min after addition of IL-1, cultures were lysed and the cell lysates were immunoprecipitated using an antibody to p62 (41). p62 is a scaffolding protein known to associate with PKC when it is activated by upstream protein kinases (42,43). Thus, immunoprecipitation of p62 provides enrichment for PKC that has been phosphorylated by upstream kinases. p62 protein, along with proteins from the cell lysates that were bound to it, was recovered and run on an SDS-polyacrylamide gel, and a Western blot was performed either with an antibody that specifically binds PKC when it is phosphorylated at threonine 410 (Fig. 3, top panels of A and C) or with an antibody raised to a peptide representing the carboxyl-terminal 20 amino acids of PKC that binds to PKC regardless of its phosphorylation state (Fig. 3, bottom panels of A and C). Phosphorylation of PKC at Thr 410 is known to be due to the activity of PDK-1, and this phosphorylation event initiates the activation of PKC enzymatic activity (44). In the T/C-28a2 cell line, basal levels of phospho-Thr 410 PKC were low (Fig. 3, A and B). Upon addition of IL-1␤, phosphorylation of PKC at the Thr 410 position was significantly increased after 4 E. R. LaVallie and M. Arai, unpublished results. 1 min ( p Ͻ 0.001), and peaked at 3 min of IL-1␤ exposure. This increase in PKC phosphorylation remained at roughly the same significantly elevated level throughout 10 min of IL-1␤ exposure (Fig. 3, A and B). The timing and persistence of PKC phosphorylation in the IL-1␣-treated primary bovine chondrocytes were very similar to T/C-28a2 cells that were stimulated with IL-1␤ (Fig. 3, C and D). The increase in PKC phosphorylation elicited by IL-1␣ treatment of the primary bovine chondrocytes was also detectably increased after 1 min but did not reach statistical significance ( p Ͻ 0.05) until 3 min of IL-1␣ exposure; this significant increase in bovine chondrocyte phospho-PKC levels then persisted throughout the remaining 10-min time course of the experiment. The greater variability of response of the primary bovine chondrocytes to IL-1␣ treatment compared with the human chondrocytic cell line probably arose from variability between donors, because the bovine chondrocytes were harvested from articular cartilage of different individual calves for each of the replicate experiments.

Up-regulation of Aggrecanase Expression by TNF-␣ and IL-1␣ in Primary Articular Chondrocytes Is Dependent upon PKC and NF-B-Aggrecanases
(ADAMTS-4 and ADAMTS-5) are metalloproteinases that are believed to be responsible for the increased cleavage of aggrecan (the most abundant cartilage extracellular matrix proteoglycan) at specific sites that are characteristic of OA (45)(46)(47). Both ADAMTS-4 and ADAMTS-5 are expressed by articular chondrocytes (48), and although there has been some discrepancy in the literature, there is cumulative evidence that both are transcriptionally up-regulated when chondrocytes are treated with either IL-1 or TNF-␣ (24,48,49). Using a primary bovine articular chondrocyte pellet culture assay system that we devised (24), the effect of PKC inhibitors and an NF-B inhibitor on TNF-␣and IL-1␣-mediated induction of ADAMTS-4 and ADAMTS-5 was tested. Bovine chondrocyte three-dimensional pellet cultures surrounded by self-synthesized extracellular matrix were pretreated with the following inhibitors: 10 M RO 31-8220 (3-[1- [3-(amidinothio) propyl-1H-indol-3-yl]-3-(1-methyl-1H-indol-3-yl)maleimide; bisindolylmaleimide IX); 40 M myristoylated PKC pseudosubstrate (PS) peptide (50,51); 100 M SN50, a cell-permeable peptide inhibitor of NF-B nuclear translocation; and an equivalent concentration of the control peptide SN50M, a mutated, inactive derivative of SN50 (19). RO 31-8220 reportedly inhibits all PKC isoforms with varying potency and also inhibits mitogen-activated protein kinase phosphatase-1 (MKP-1) expression, induces c-Jun expression, and activates Jun amino-terminal kinase (15). Following pretreatment with inhibitors (control cultures received no inhibitor pretreatment), TNF-␣ or IL-1␣(or no cytokine) was added to the cultures. After 18 h, cytokine-mediated matrix destruction was assessed by measuring the amount of total proteoglycan released from the pellet cultures to the conditioned media by DMMB assay (24). Fig. 4A shows that both TNF-␣ and IL-1␣ significantly increased proteoglycan degradation in the pellet cultures in the absence of inhibitors, resulting in ϳ4-fold more proteoglycan released to the media when compared with the "no cytokine" control. RO 31-8220 treatment without addition of cytokines had no effect on proteoglycan release; however, addition of RO 31-8220 prior to TNF-␣ or IL-1␣ effectively blocked the cytokine-mediated proteoglycan degradation seen in the control cultures ( p Ͻ 0.01). Addition of the PKC PS peptide also resulted in significant reduction of TNF-␣ and IL-1␣-mediated proteoglycan release ( p Ͻ 0.01), as did the NF-B blocker SN50 but not the SN50M negative control peptide (Fig. 4A). Evaluation of proteoglycan levels remaining in the pellet showed that the proteoglycan

. Comparison of PKC and PKC mRNA and protein levels in human articular chondrocytes.
Quantitative RT-PCR assays (TaqManா) were used to evaluate the relative abundance of mRNA for PKC and PKC in human osteoarthritic cartilage samples. A, threshold cycle values (mean Ϯ S.D.) for PKC and PKC in articular cartilage RNA from three different human donors with end stage OA. B, "TaqMan units" were calculated for each transcript by comparing Ct values for each sample (assayed in triplicate) to a standard curve consisting of known quantities of cDNA for each gene. The extrapolated transcript abundance from the standard curve was then normalized to an internal standard (GAPDH) for each assay. The calculated values are shown within the bar graph because the PKC levels were too low for graphing on this scale. C, immunoblot analysis of primary chondrocyte cell lysates from two different ("H3" and "H4") human donors with end stage OA. Identical blots were prepared from 10% SDS-polyacrylamide gels loaded with 150 g of cell lysate protein per lane, and the immobilized proteins were individually subjected to Western blotting with either a rabbit polyclonal antibody that recognizes both PKC and PKC ("aPKC C-20", left panel ), or with rabbit polyclonal antibodies that specifically recognize PKC (middle panel ) or PKC (right panel ). Data shown for these two donors typify the results seen for additional samples that were tested.
released to the conditioned media upon cytokine treatment resulted in a concomitant decrease in pellet proteoglycan content, and inhibitor pretreatment preserved the proteoglycan content of the pellets (data not shown). Therefore, the observed decrease in cytokine-mediated proteoglycan release to the conditioned media resulting from inhibitor pretreatment of pellet cultures appeared to be attributable to decreased proteoglycan degradation and not to decreased proteoglycan synthesis.
Aggrecanase neoepitope Western analysis of conditioned media from the different pellet cultures in the experiment shown in Fig. 4A revealed that the proteoglycan fragments released from IL-1-or TNF-treated chondrocyte pellet cultures in the absence of inhibitors contained a substantial amount of aggrecanase cleavage products that was readily detectable on Agg-C1 neoepitope Western blots (Fig. 4B, lanes 1 and 2). These IL-1-and TNF-induced aggrecanase cleavage products were markedly decreased in abundance by pretreatment with RO 31-8220, PKC pseudosubstrate peptide, or the SN50 NF-B blocking peptide in a manner that closely mirrored the DMMB data in Fig. 4A, suggesting that blocking PKC or NF-B activity prior to cytokine treatment resulted in inhibition of aggrecanase expression and activity.
To test this assumption, total RNA was extracted from the pellet cultures at the end of the culture period, and quantitative RT-PCR was performed using probe/primer sets designed to bovine ADAMTS-4 (Agg-1) and bovine ADAMTS-5 (Agg-2) mRNA sequences (24). Bovine GAPDH was used as a normalization control. Both TNF-␣ and IL-1␣ treatment induced Agg-1 and Agg-2 mRNA levels in these chondrocyte cultures (Fig. 4,C and D). Agg-1 mRNA induction by TNF-␣ and IL-1␣ was significantly suppressed by RO 31-8220 and SN50 ( p Ͻ 0.01), and the PKC PS peptide also appeared to reduce Agg-1 mRNA levels, but the effect did not reach statistical significance (Fig. 4C). However, up-regulation of Agg-2 mRNA by TNF and IL-1 in the primary chondrocyte cultures was significantly suppressed by all three inhibitors (Fig. 4D), demonstrating that PKC inhibition can effectively block induction of aggrecanase expression in chondrocytes by these inflammatory cytokines. Thus, the proteoglycan degradation caused by exposure of primary chondrocytes to IL-1 or TNF involves PKC-dependent aggrecanase up-regulation that results in accumulation of aggrecan fragments in the conditioned media that are cleaved at the Glu 373 -Ala 374 site in the interglobular domain of aggrecan. Because aggrecan is the predominant proteoglycan in cartilage matrix (52), these data support a model in which suppression of cytokine-mediated aggrecanase induction by inhibition of PKC activity is the mechanism by which proteoglycan is preserved in this system.
To further explore the effects of specific inhibition of PKC, primary bovine chondrocyte pellet cultures were pretreated with increasing concentrations of the myristoylated (cell-permeable) PKC PS peptide prior to addition of 10 ng/ml IL-1␣ and overnight (18 h) incubation. Doses of peptide as low as 10 M significantly ( p Ͻ 0.05) reduced induction of proteoglycan degradation by IL-1␣, and doses of 20 M or more of the PKC PS peptide led to even more significant reduction ( p Ͻ 0.01) of the IL-1␣ effect on proteoglycan release from the pellet cultures (Fig. 5A). The inhibition of cytokine-induced proteoglycan release by the PKC PS peptide was not attributable to decreased synthesis of proteoglycan due to cytotoxicity, because lactate assays on conditioned media from these same pellet cultures showed no decrease in cellular metabolism (Fig.  5B). In addition, evaluation of the proteoglycan content of the cell pellets following papain digestion at the end of the experiment showed that total proteoglycan synthesis was not impaired (data not shown). In this same cytokine-induced chondrocyte proteoglycan release assay using TNF-␣, comparison of the myristoylated PKC PS peptide to a nonmyristoylated PKC PS peptide with the identical sequence, and to a myristoylated peptide containing the pseudosubstrate sequence of PKC␣/␤, revealed that the inhibitory effect was specific to PKC and required cell permeability (Fig. 5C). An additional myristoylated control peptide containing the same amino acid content as the PKC PS peptide but with the order of the amino acids scrambled has also been repeatedly tested in this assay and showed no inhibitory effect on cytokine-induced matrix degradation (data not shown). An NF-B-luciferase reporter gene construct was stably integrated into the immortalized human costal chondrocyte cell line T/C-28a2 (23), and a reporter cell line was developed to evaluate whether PKC was responsible (directly or indirectly) for the activation of NF-B in chondrocytes by IL-1␣ and TNF-␣, as it is in some other cell types (43,(53)(54)(55)(56). The resulting cell line was selected to be highly responsive to IL-1␣ and TNF-␣ treatment as judged by its luciferase expression via NF-B-mediated transcription (see "Experimental Procedures"). As shown in Fig. 5D, the induction of the NF-B-luciferase reporter gene by either IL-1␣ or TNF-␣ in this human chondrocytic cell line was blocked by both triptolide (an NF-B inhibitor (20)) and by the myristoylated PKC PS peptide inhibitor ( p Ͻ 0.01), but not by the nonmyristoylated PKC peptide nor by the myristoylated PKC␣/␤ PS peptide. These experiments provide evidence that TNF-␣ and IL-1␣ elicit comparable effects in bovine primary chondrocytes with regard to PKCdependent activation of NF-B-mediated transcription and proteoglycan degradation, and that the T/C-28a2 human chondrocyte cell line responds in a very similar fashion to primary bovine chondrocytes.

TNF or IL-1 Induction of NF-B and Resulting Proteoglycan Degradation in Chondrocytes Is Blocked by Pan-PKC Inhibitors but Not by an Atypical-sparing PKC Inhibitor-
Additional PKC inhibitors with different specificities were used with the T/C-28a2 NF-B-luciferase cell line in an effort to further implicate the isoform as the responsible protein kinase C family member in the pathways by which TNF and IL-1 signal through NF-B. Bisindolylmaleimide I (BIS) inhibits all PKC isoforms with a potency rank order of cPKC Ͼ nPKC Ͼ aPKC (16,57). IL-1␤ induction of the NF-B-luciferase reporter gene in the T/C-28a2 cell line was inhibited by almost 80% by 12.5 M BIS, with less inhibition noted with decreasing concentrations of BIS (Fig. 6A). This level of inhibition is consistent with the reported IC 50 of 5.8 M for BIS on PKC, which is almost 300-fold higher than the IC 50 for BIS on cPKCs (0.02 M) and more than 30-fold higher than the IC 50 for BIS on nPKCs (58). Gö 6976 reportedly is also most potent against the Ca 2ϩ -requiring (conventional) PKC isoforms and is less potent against the nPKCs and aPKCs, with an IC 50 for PKC of Ͼ10 M (21,59). Gö 6976 showed some inhibition of both IL-1 and TNF induc-  (SN50, 100 M), or a mutated inactive SN50 control peptide (SN50M, 100 M). TNF or IL-1 was then added, and the cells were incubated prior to assay. All culture treatments were performed in triplicate. A, total proteoglycan released to the conditioned media from the pellet culture extracellular matrix after 18 h of cytokine treatment, as measured by DMMB assay. DMMB assays were performed in duplicate and averaged. Data are expressed as the mean Ϯ S.D. of the replicate cultures (n ϭ 3). B, aliquots of culture media from the experiment shown in A were subjected to Western blotting using the monoclonal neoepitope antibody Agg-C1 to allow detection of aggrecanase-cleaved aggrecan fragment as a measure of aggrecanase activity in the pellet cultures. The figure is a composite of two separate blots prepared at the same time, immunostained, and developed together for the same amount of time. C and D, RNA harvested 18 h after cytokine addition was assayed for ADAMTS-4 (Agg-1; C ) and ADAMTS-5 (Agg-2; D) mRNA levels using TaqManா probe/primer sets designed to the bovine sequences. Data are expressed as the mean Ϯ S.D. of replicate experiments (n ϭ 3), and significance was determined by Student's t test.
tion of NF-B-luciferase activity, but it was no more potent in this assay than BIS, despite the fact that Gö 6976 has an IC 50 ϳ4-fold lower than BIS on cPKCs (21). Calphostin C is a compound that competes for the diacylglycerol/phorbol ester-binding site in the regulatory domain of the conventional and novel PKCs and competitively inhibits their activity (17). Because the atypical PKCs ( and ) lack this domain, their activity is unaffected by calphostin C. Calphostin C was totally ineffective at blocking IL-1 or TNF induction of NF-B (Fig. 6A), even at a concentration 100 times higher than its IC 50 for cPKC and nPKC (17). Similar results were obtained in the bovine primary chondrocyte pellet culture assay (Fig. 6B). Increased degradation and release of extracellular matrix proteoglycan by addition of TNF-␣ was totally blocked in this assay by 40 M BIS, whereas calphostin C had no effect at concentrations up to 100 nM. Similar data were obtained in bovine and porcine chondrocytes treated with either IL-1 or TNF-␣, even at calphostin C concentrations well above the IC 50 for cPKC and nPKC (data not shown). Previous studies have shown that calphostin C is capable of inhibiting PKC␣ activation in primary chondrocytes (18), and 100 nM calphostin C significantly inhibited proteoglycan synthesis induced by CCN2 in the chondrocytic cell line HCS-2/8 (60), proving that calphostin C is active on chondrocytes in culture. These data strongly implicate an atypical PKC as a necessary component of IL-1 and TNF signal transduction to NF-B in chondrocytes.

Sphingomyelinase-induced Transcription via NF-B in
Chondrocytes Is Dependent on PKC-Ceramide is a second messenger that is liberated by hydrolysis of sphingomyelin (a cell membrane-derived sphingolipid) by sphingomyelinases (61). Sphingomyelinase (SMase) activity has been shown to be up-regulated in some cell types by treatment with TNF or IL-1 (62)(63)(64)(65). Furthermore, ceramide is capable of directly activating PKC without stimulation of upstream signal transduction components (66 -68). Direct treatment of cells with sphingomyelinase has been shown to increase intracellular ceramide levels resulting in the activation of PKC (66,69). Therefore, if PKC is an important signaling component of the NF-B pathway in chondrocytes, it would be expected that increasing ceramide levels within these cells would result in NF-B activation, and this activation should be blocked by addition of PKC inhibitors. Fig. 7 shows the results of such an experiment, in which primary bovine chondrocytes expressing an NF-B-luciferase reporter gene were induced with either 10 or 40 milliunits of sphingomyelinase, or with TNF-␣, or were uninduced. Prior to addition of SMase or TNF, cells received either no pretreatment or a 1-h pretreatment with the pan-PKC inhibitor BIS (20 M), the atypical PKC-sparing calphostin C (CalC, 100 nM), or 0.5% Me 2 SO as a vehicle control. Even without addition of SMase or TNF-␣, BIS decreased basal levels of NF-B transcription significantly, whereas calphostin C did not. Without inhibitor pretreatment, addition of SMase to the primary chondrocytes increased NF-B-luciferase expression significantly (Ͼ4-fold for 10 milliunits of SMase and Ͼ6-fold for 40 milliunits of SMase), comparable with TNF-␣ treatment (ϳ8-fold induction). Addition of BIS totally blocked the induction of NF-B-luciferase expression by either SMase or TNF-␣, demonstrating the necessity of a protein kinase C activity in the pathway. However, pretreatment with calphostin C had no significant effect on either SMase or TNF-␣ induction of NF-Bluciferase expression compared with the Me 2 SO control ( p ϭ 0.19 for 10 milliunits of SMase, p ϭ 0.26 for 40 milliunits of SMase, and p ϭ 0.39 for 1 ng/ml TNF-␣; comparison of CalC changes to the untreated control also resulted in no significant decreases). These data further implicate PKC as the atypical PKC family member acting as an essential component of the NF-B signaling pathway in chondrocytes.

Effect of Expression of a Dominant Negative (K281W) PKC Mutant on NF-B Signaling in
Chondrocytes-A kinase-defective mutant form of PKC was constructed in which a key residue in the ATP-binding region of the catalytic domain (lysine 281) was changed to a tryptophan. This mutation has been shown to create a dominant negative form of PKC (DN-PKC) that specifically and nonproductively competes for components in the signaling pathway, resulting in a specific competitive inhibitor for the kinase (29,30,43). The DN-PKC cDNA was placed under the control of a CMV promoter in an adenoviral vector, and adenovirus stocks were generated as described previously (24). Primary bovine chondrocytes were prepared and infected with an equal m.o.i. of viral stocks expressing either full-length active human PKC (FL-PKC), the DN-PKC, or an "empty" control virus without a cDNA inserted downstream of the CMV promoter. In some experiments, comparisons were made to a co-infected GFP-expressing adenovirus instead of the empty control virus, with comparable results (data not shown). These viruses were co-infected in combination with an NF-B-luciferase reporter virus. After 4 days in culture to allow time for viral expression of the PKC constructs, cultures were treated with either IL-1␣, with two different concentrations of TNF-␣, or were not treated with any cytokine. As shown in Fig. 8, expression of FL-PKC resulted in ; or triptolide (6.3 nM-50 nM) prior to addition of either IL-1␤ (10 ng/ml) or TNF-␣ (25 ng/ml). All compounds were shown to by noncytotoxic on these cells in the concentration ranges used in these assays by WST-1 proliferation assays. Luciferase expression was measured by luciferase assays on cell lysates 3 h after addition of cytokines. Control cultures without inhibitor added were used to define 100% NF-B-luciferase activity. B, bovine pellet cultures received either no inhibitor (none) or were treated with BIS (40 M) or with increasing concentrations of calphostin C (25-100 nM) prior to addition of 100 ng/ml TNF-␣. Proteoglycan degradation was measured 18 h later using the DMMB assay on the culture conditioned media. Values represent the mean Ϯ S.D. of three replicate experiments, and p values were derived from comparisons using the Student's t test.
a small but consistent increase in the basal levels of luciferase activity (in the absence of cytokine addition) when compared with the empty control virus. In some experiments this trend reached statistical significance, but in most experiments it did not (data not shown). No significant increase in cytokine-induced NF-B-luciferase activity was observed when PKC was overexpressed, suggesting that normal intracellular levels of PKC protein are not limiting. However, expression of DN-PKC consistently resulted in significant attenuation of both IL-1-and TNF-mediated induction of NF-B-luciferase expression when compared with the control virus ( p Ͻ 0.002 for 0.1 ng/ml TNF-␣ versus control and p Ͻ 0.001 for 0.1 ng/ml IL-1␣ and 1 ng/ml TNF-␣ versus control). Comparable results were seen in similar experiments performed in primary human articular chondrocytes (data not shown). These data provide further evidence that PKC is a key mediator of both IL-1 and TNF signaling through NF-B in chondrocytes.

DISCUSSION
The initial intent of these studies was to identify genes that display dysregulated expression in human articular chondrocytes from osteoarthritic cartilage when compared with nonosteoarthritic cartilage by transcriptional profiling on Affymetrix GeneChip microarrays. Cartilage is well suited for this type of transcriptional analysis, because the tissue contains only a single cell types (chondrocytes), and there are no blood vessels, nerves, or other cell types to contribute RNA that could confound the results. This is especially relevant because it is believed that the pathological changes in OA cartilage are manifested primarily by dysregulated gene expression in the chondrocytes (70).
Many genes were detected on the gene chips that displayed significantly altered transcript levels in OA cartilage compared with normal cartilage. 3 The focus of this paper is one of those genes, the intracellular serine/threonine kinase PKC. A search for PKC family members in the human OA articular cartilage transcriptional profiling data revealed that PKC was the only PKC family member that showed altered expression in OA cartilage, and in fact only one other family member (PKC␦) was expressed at quantifiable levels on the gene chips. Many studies have demonstrated that PKC is involved in the pathway for activation of the NF-B transcriptional factor (29, 30, 38, 41, 43, 53-55, 66, 67, 71, 72) as well as in the MAPK signaling cascade (71,73). The NF-B pathway was of particular interest to us because certain inflammatory cytokines thought to be intrinsically linked to OA pathophysiology, especially TNF and IL-1, are known to cause activation of NF-B and result in increased expression of catabolic factors with destruction of articular cartilage that is characteristic of the OA disease process. In our studies, treatment of chondrocyte pellet cultures with RO 31-8220 (which has been shown to act as an activator of MAPK and c-Jun (15)) did not appear to cause an increase in proteoglycan degradation in the absence of TNF or IL-1. In addition, calphostin C has been shown to be capable of activating AP1 and inducing c-Jun transactivation (18), but treatment of chondrocyte pellet cultures with calphostin C in the absence of cytokines did not increase proteoglycan release. These results pointed us to the NF-B pathway to evaluate a potential role of PKC in chondrocytes.
Recently, the characterization of a PKC knock-out (KO) mouse has been described (54). This KO mouse appeared to be grossly normal, but at the cellular and molecular level it displayed defects in NF-B signaling of varying severity depending on the cell type. However, the impact of a lack of PKC on NF-B signaling in chondrocytes from this KO mouse was not FIGURE 7. Sphingomyelinase-induced NF-B transcription in chondrocytes is dependent on PKC. Bovine primary chondrocytes expressing an NF-B-luciferase reporter gene were treated with either 10 or 40 milliunits (mU) of sphingomyelinase (SMase), 1 ng/ml TNF-␣, or were untreated. All cultures were pretreated for 1 h with either the pan-PKC inhibitor BIS (20 M), the PKC-sparing inhibitor CalC (100 nM), or with Me 2 SO (DMSO) (vehicle control; 0.5% final concentration). Activation of the NF-B reporter gene was monitored by luciferase activity 4 h after the addition of SMase or TNF. Bars represent the mean Ϯ S.D. of three culture replicates; these data are from a single experiment that was representative of at least three independent experiments, and significance was determined by Student's t test.  (DN-PKC), or adenovirus without an inserted gene (Control ). Each bar represents the mean Ϯ S.D. of 3 culture replicates; these data are from a single experiment that was representative of at least three independent experiments. Differences were determined to be significant by Student's t test.
reported. A possible explanation for the variability of the NF-B phenotype in different tissues in the PKC KO mouse was that cell types that normally expressed higher levels of PKC might show a greater impact in NF-B signaling (54,55). Another possible explanation was that PKC might compensate for loss of PKC to varying degrees in different cell types. A third and very significant possibility arose after subsequent publications revealed that the gene disruption strategy that was used to generate the PKC KO allele still allowed the expression of a truncated, constitutively active form of PKC (called PKM) from an alternative, intronic promoter (74,75). This alternative promoter was reported to be brain-specific; however, we have found PKM to be expressed in cartilage and in some other tissues by quantitative RT-PCR analysis. 4 Therefore, the impact of PKC activity on NF-B signaling may be even more important (and perhaps even essential) in more cell types than the initial analysis of this PKC KO mouse suggested.
The experiments described here utilized various PKC activators and inhibitors in chondrocyte cell-based assays to confirm the essential role of an atypical PKC in activation of NF-B by TNF and IL-1 in this cell type. Importantly, these data showed that atypical PKC activity is not only essential for chondrocytic NF-B activation, but also that inhibition of atypical PKC activity results in protection from the degradative effects of TNF and IL-1 signaling on extracellular matrix in articular chondrocyte cultures. However, from our data it was difficult to conclusively pinpoint PKC as the responsible atypical isoform, and a role for PKC could not be definitively ruled out. The pseudosubstrate sequence of PKC is identical to the pseudosubstrate sequence of PKC, so it would be expected that the myristoylated PKC pseudosubstrate peptide inhibitor used in this study would also inhibit PKC. Antibody reagents to phospho-PKC and total PKC also bind PKC, as does p62, so immunoprecipitation of p62 prior to phosphoprotein Western analysis does not help to distinguish between atypical isoforms (42). Also, it is possible that the dominant negative PKC construct used in these studies may interfere with activity as well as with activity. However, our findings that PKC is expressed at extremely low levels in chondrocytes and does not appear to be dysregulated in OA cartilage suggested to us that even if PKC was capable of functionally substituting for PKC, the large disparity in expression levels for the two isoforms in chondrocytes meant that PKC activity probably accounted for most if not all of the effect that we saw in our studies.
A recent publication by Soloff et al. (10) has provided important evidence to support this assumption. These investigators provided compelling evidence that the mouse ortholog of PKC (called PKC) is not involved in the NF-B signal transduction pathway. This group created a PKC KO mouse that exhibited an embryonic lethal phenotype prior to day 9.5, which prevented the study of cells from the adult mouse. However, they were able to select for PKC-deficient embryonic stem cells from PKC ϩ/Ϫ precursors and created immortalized PKCdeficient mouse embryonic fibroblasts in which to study the effects of the loss of PKC on signaling pathways. These PKCdeficient fibroblasts showed no defect in NF-B activation, as judged by the unimpaired degradation of IB␣ and induction of a NF-B-luciferase reporter construct in response to TNF treatment in these cells (10). It is important to note that Leitges et al. (54) had shown previously that TNF-treated mouse embryonic fibroblasts from the PKC KO mouse were one cell type that showed severe impairment of NF-B signaling compared with PKC ϩ/ϩ mouse embryonic fibroblasts, providing a direct comparison of PKC versus PKC contributions to NF-B signaling in the same cell type.
Based upon these and our own observations, we conclude that PKC is the atypical PKC isoform that serves as a critical link to NF-B activation by TNF and IL-1 in chondrocytes. In this regard, PKC represents a potentially important diseaseassociated gene responsible for OA cartilage destruction and an attractive therapeutic target. Orally active, small molecule inhibitors of PKC might potentially provide an attractive and powerful therapeutic option for the chronic treatment of OA, because such an approach may serve to block a common signaling constituent for multiple inflammatory cytokines as well as interrupt the synthesis of other mediators of both OA and rheumatoid arthritis that lie downstream of NF-B (76).