Differential Effect of Transforming Growth Factor β (TGF-β) on the Genes Encoding Hyaluronan Synthases and Utilization of the p38 MAPK Pathway in TGF-β-induced Hyaluronan Synthase 1 Activation*

Unfettered hyaluronan (HA) production is a hallmark of rheumatoid arthritis. The discovery of three genes encoding hyaluronan synthases (HASs) allows for the investigation of the signaling pathways leading to the activation of these genes. Our objective is to further understanding of the regulation of these genes as well as to find ways to prevent undesired gene activation. Human fibroblast-like synoviocytes were used in these experiments. mRNA levels of HAS were monitored by reverse transcriptase-PCR. A series of specific kinase inhibitors were used to investigate intracellular pathways leading to the up-regulation of HAS1. Our experiments, testing a series of stimuli including tumor necrosis factor α (TNFα), demonstrate that TGF-β is the most potent stimulus for HAS1 transcription. TGF-β activates HAS1 in a dose-dependent manner with a maximum effect at a concentration of 0.5–1 ng/ml. TGF-β-induced HAS1 mRNA can be detected within 60 min and reaches maximal levels at 6 h. Furthermore, TGF-β treatment leads to an increase in synthase activity as determined by HA ELISA and by in vitro HA synthase assays. In contrast to the activatory effect on HAS1, TGF-β dose-dependently suppresses HAS3 mRNA. As to the mode of action of TGF-β-induced HAS1 mRNA activation, our experiments reveal that blocking p38 MAPK inhibited the TGF-β effect by 90%, blocking the MEK pathway led to an inhibition by 40%, and blocking the JNK pathway had no effect. The presented data might contribute to a better understanding of the role of TGF-β and of HA in the pathology of diseases.

Today, this macromolecule is most frequently referred to as hyaluronan. Until recently, it was believed that HA was an inert compound found in many tissues that did not specifically interact with other macromolecules. It was in the early 1970s that the first reports emerged indicating that HA can specifically interact with other molecules (2). Subsequently, a series of reports confirmed and extended the findings that HA indeed interacts with many other molecules and that HA itself might play an important physiological role in a series of biological processes (e.g. inflammation (3,4) and cancer progression (1,(5)(6)(7)).
The discovery of a lymphocyte homing receptor termed CD44, which acts as a HA-binding protein, is indicative of the importance of HA in many physiological as well as pathological events (8). Subsequently, it has been demonstrated that functional activation of lymphocyte CD44 in peripheral blood is a marker of autoimmune disease activity (9). For example, specific ligation of CD44 to HA activates peripheral blood T cells, causing increased interleukin-2 (IL-2) levels, and stimulates monocytes to release higher levels of the proinflammatory cytokines IL-1 and tumor necrosis factor ␣ (TNF␣). Similarly, activation of T cell-associated CD44 is required for leukocyte extravasation into inflammatory sites (10). The discovery of a series of other specific HA-binding proteins termed hyaladherins further supports the concept of HA regulating cellular activity. In addition to the above, a large number of reports have been published in the last decade on the role of HA and hyaladherins in cellular migration, inflammation, and angiogenesis (11)(12)(13)(14)(15)(16).
A major breakthrough, namely the identification of three separate genes of vertebrate hyaluronan synthases (HASs) a few years ago (17), renewed scientific interest in HA. The discovery of the three genes encoding HAS (HAS1, HAS2, and HAS3) laid the foundation for further studies into the regulation and function of these genes. Although this discovery occurred only very recently, there is mounting evidence that the products of these three genes seem to have distinct functions. Therefore, studies investigating activation and regulation of these genes seem imperative for a better understanding of many forms of pathological disorders including RA. It is anticipated that a better understanding of intracellular signaling pathways, leading to up-or down-regulation of these genes might lay the foundation for specific intervention, thereby preventing unfettered disease progression. Undoubtedly, large amounts of HA are produced in joints of RA patients. Currently, it is not yet clear whether or to what degree one or the other form of intact HA might exert proinflammatory effects. Irrespectively, HA has been shown to undergo rapid degrada-tion at sites of inflammation (18,19). Furthermore, low molecular weight degradation products of HA have been found to elicit various proinflammatory responses such as the stimulation and invasion of macrophages in affected joints (20,21), as well as the functional maturation/activation of dendritic cells (22). Furthermore, fragments of HA induce plasminogen activator inhibitor-1, NO and IL-12, as well as other chemokines (23)(24)(25). More importantly, low molecular weight HA molecules have been shown to act as a chemoattractant for many cell types as well as an angiogenic factor inducing the growth of blood vessels into joints, skin, and tumors (26,27).
Overproduction of HA in RA patients seems to be so high that it can easily be detected even in the serum of such patients. Hence, serum levels of HA were demonstrated to be a potential diagnostic marker for the progression of RA (28,29). Taken together, it seems a worthwhile endeavor to investigate mechanisms leading to the up-regulation of HASs, since insight into regulatory mechanisms and the signaling pathways involved in the regulation of the three HAS genes might eventually lead to ways of preventing undesired HA production. To date, intracellular signaling pathways involved in the regulation of HAS have not been reported. Since nearly all cell surface receptors utilize one or more of the mitogen-activated protein kinase (MAPK) cascades, we focused in this study on MAPK pathways leading to the activation of HAS and report our findings on the regulation of HAS1 and HAS3. Cell Culture-Human fibroblast-like synoviocytes (FS) were a gift from Dr. G. Partsch (Vienna). FS were cultured as previously described (30). In brief, FS were propagated in T75 tissue culture flasks or culture dishes (Nalge Nunc International, Rochester, NY) in RPMI 1640 medium (Sigma) supplemented with 10% heat-inactivated fetal bovine serum (Sigma), L-glutamine, and 50 units/ml penicillin/streptomycin. Medium was changed every 3 days. For experiments, FS were detached using trypsin and transferred to 6-well plates (Iwaki, Funabasi, Chiba, Japan).

Reagents-Recombinant
RNA Isolation and Reverse Transcription-Polymerase Chain Reaction (RT-PCR)-RNA isolation from cultured cells was as follows. 1 ml of TRI-reagent was added to cells (maximum of 10 6 /ml). Subsequently, the cell suspension was homogenized by repetitive taking up into a syringe through a 26-gauge needle. The resulting solution was stored for 5 min at room temperature. After that, 0.2 ml of chloroform was added. Each vial was shaken vigorously by hand and subsequently incubated for 5 min at room temperature and centrifuged at 12,000 ϫ g for 15 min at 4°C, after which the aqueous phase was transferred to a new vial. Mixing with isopropyl alcohol (0.5 ml), incubating the vials for 10 min at room temperature and subsequent centrifugation at 15,000 ϫ g for 10 min at 4°C precipitated the RNA. After two washes in 75% EtOH, the resulting RNA was briefly dried in air and subsequently dissolved in H 2 O, aliquoted, and stored at Ϫ80°C. Furthermore, small aliquots of RNA were used to check the quality of the resulting RNA using agarose gel and ethidium bromide or Vistagreen (Molecular Probes, Inc., Eugene, OR) for visualization.
First-strand cDNA synthesis was performed using a bulk mix (11 l) (Amersham Biosciences, Freiburg, Germany) to which pd(N) hexamers (0.2 g), 1 l of dithiothreitol (1 M stock) solution, and 1 g of RNA were added. H 2 O was added to a final volume of 33 l. RNA solutions were heated to 65°C for 10 min and placed on ice. Subsequently, the synthesis mix was added, and the solution was incubated at 37°C for 1 h. Aliquots were used for PCR. A Techne cycler (Techgene, Duxford, Cambridge, UK) and an Eppendorf cycler (Eppendorf, Hamburg, Germany) were used for PCR under the following standard conditions: initial denaturation, 4 min at 94°C; annealing, 55 or 62°C (HAS3); amplification, 20 s at 72°C; denaturation, 20 s at 94°C. 21-35 cycles, followed by final extension for 10 min at 72°C. Care was taken to work out exact PCR conditions in order to ensure that the amplificationreaction was stopped in the log phase of amplification. As control for equal usage of mRNA, either actin or GAPDH or both were used.
The following primers were from MWG-Biotech AG (Ebersberg, Germany) and were dissolved at a concentration of 100 pmol/l in Tris-EDTA: HAS1 forward primer (5Ј-GCG GGC TTG TCA GAG CTA-3Ј) and HAS1 reverse primer (5Ј-AGA GCG ACA GAA GCA CCA-3Ј); HAS2 forward primer (5Ј-GTG ATG ACA GGC ATC TCA-3Ј) and HAS2 reverse primer (5Ј-GCG GGA AGT AAA CTC GA-3Ј); HAS3 forward primer (5Ј-CAG CCT GCA CCA TCG A-3Ј) and HAS3 reverse primer HA Measurements-FS were grown to high density in 6-well plates. Immediately before experiments, FS were washed two times with culture medium (37°C) to completely remove HA accumulated during cell growth. Subsequently, FS were cultured with or without stimuli in 1 ml of complete medium for times ranging from 8 to 24 h. At the indicated times, aliquots of medium were removed, centrifuged (5 min at 2000 ϫ g), and tested for the presence of HA via a procedure provided by Corgenix (Corgenix, Westminster, CO). In short, plates coated with HAbinding protein were purchased from Corgenix. Such plates were incubated with supernatant (10 l of supernatant diluted with 90 l of reaction buffer) and standards, respectively, for 1 h (room temperature) in duplicates, washed five times with washing buffer, incubated with a solution containing horseradish peroxidase-conjugated HA-binding protein for 1 h at room temperature, washed again five times, and incubated with 100 l of the provided substrate solution. After 20 min, the reaction was stopped by adding an equal amount of sulfuric acid (0.36 N), and after that the OD was measured at 450 nm (630-nm reference). OD values were used to calculate HA levels using a third-order polynominal regression analysis performed with a universal assay calculation program (AssayZap, Biosoft, Cambridge, UK).
In Vitro Hyaluronan Synthase Assay-Hyaluronan synthase activity was monitored using a modification of previously described methods (31,32). Briefly, FS, cultured in 15-cm tissue culture dishes, were washed, incubated in hypotonic lysis buffer (LB) (10 min), and harvested into 1 ml of LB supplemented with aprotinin, leupeptin, and phenylmethylsulfonyl fluoride. A Dounce homogenizer (pestle B) was used to disrupt cells. Nuclei were pelleted by spinning tubes at 1000 ϫ g for 4 min. Samples were centrifuged at 16,000 g for 25 min in order to pellet membrane fragments. LB (50 l) with protease inhibitors was used to resuspend membrane pellets. An aliquot, diluted in LB plus 1% SDS was used for protein measurement utilizing the BCA assay (Pierce). All of the above steps were performed in the cold room using prechilled solutions. The in vitro HA synthase assays using 25 g of cell membrane extract were assembled exactly as described (32). After incubation for 1 h at 37°C, the reaction was stopped by boiling, and the mixtures were further incubated with or without Streptomyces hyaluronate lyase (40 turbidity-reducing units per 50-l aliquot) at 37°C overnight and then treated with 200 mg/ml Pronase at 37°C for 5 h for deproteinization (33).
After boiling the samples in the presence of 1% SDS (w/v), mixtures were transferred to Microcon centrifugal filter devices that retained molecules larger than 100,000 daltons (Microcon YM-100; Millipore Corp., Bedford MA). Unincorporated [ 14 C]glucuronic acid was removed by filtration (5 min at 5000 ϫ g). Subsequently, LB (200 l) was added to the sample reservoir. Centrifugation and resuspension of the retentate in LB was repeated three times to ensure complete removal of unincorporated [ 14 C]glucuronic acid. After the final spin, sample reservoirs were placed upside down in a new vial and centrifuged at 1000 ϫ g to recover polysaccharides. Scintillation mixture was added for determination of radioactivity. [ 14 C]glucuronic acid incorporated into the hyaluronan polymer was calculated from the Streptomyces hyaluronidase-sensitive radioactivity.
Western Blotting Experiments-Cells were washed twice in ice-cold phosphate-buffered saline and subsequently dissolved in SDS sample buffer (SB) (62.5 mM Tris/HCl (pH 6.8), 2% (w/v) SDS, 10% glycerol, 50 mM dithiothreitol, 0.01% (w/v) bromphenol blue). For 3-cm culture dishes, 100 l of SB, and for 10 cm dishes 500 l of SB, respectively, was used. Aliquots of whole cell protein extract (10 -25 l/well) were separated on a minigel (10%). Proteins were blotted onto polyvinylidene difluoride membranes (Amersham Biosciences) using a semidry apparatus (Bio-Rad). Blots were flashed with TB, dipped into MetOH, and dried for 20 min before proceeding with the next steps. Subsequently, blots were transferred to a blocking buffer solution (1ϫ Tris-buffered saline, 0.1% Tween 20, 5% (w/v) nonfat dry milk) and incubated at 4°C overnight. Membranes were then incubated with the appropriate diluted primary antibody in 5% bovine serum albumin, 1ϫ Tris-buffered saline, and 0.1% Tween 20, again at 4°C overnight in a roller bottle. Following three washing steps in wash buffer (1ϫ Tris-buffered saline, 0.1% Tween 20), blots were incubated with appropriate secondary antibodies (Pierce) diluted in phosphate-buffered saline. After 45 min of gentle agitation, blots were washed five times for 5 min in wash buffer, and proteins were made visible using either LumiGLO (New England Biolabs) or Renaissance Plus (PerkinElmer Life Sciences) and Kodak BioMax MR films.
Data Analysis-Agarose gels were stained with ethidium bromide and scanned on Fluorimager 595 (Amersham Biosciences). Data were analyzed and quantitated using ImageQuant software (Amersham Biosciences). mRNA for GAPDH, actin, or both were used as controls for RT-PCR, and scanner readings were used to recalculate PCR data. Films from Western blot experiments were scanned on an Agfa Duoscan T 1200 or scanned and quantitated on a densitometer (Amersham Biosciences).

RESULTS
mRNA Levels of HAS1, HAS2, and HAS3 in Stimulated and Unstimulated FS-RT-PCR experiments were carried out in order to determine basal mRNA levels of the genes encoding HAS. Such experiments demonstrated that in unstimulated cells, mRNA levels of HAS1 were very low or undetectable, levels of mRNA for HAS2 were intermediate, and levels of HAS3 in unstimulated SF were highest. Subsequently, a series of stimuli (TNF␣ (10 ng/ml), IL-1 (10 ng/ml), TGF-␤ (1 ng/ml), IL-8 (10 ng/ml), and PMA (20 ng/ml)) were tested for their ability to induce HAS1. As shown in Fig. 1A, among the physiologically relevant stimuli tested, TGF-␤ had the most pronounced effect. In such experiments, FS were treated with the indicated amounts of stimuli for 6 h or were left untreated (lane MED).
Time course experiments revealed that elevated levels of mRNA can be detected as early as 60 min following the addition of TGF-␤ (1 ng/ml) and that maximal levels of HAS1 mRNA in FS are reached at 6 h (data not shown). Shown in Fig.  1B are data demonstrating that the RT-PCR conditions were properly adjusted to ensure that RT-PCR experiments were terminated in the early log phase of product accumulation. Shown in Fig. 1 is a representative experiment used to determine optimal cycle numbers for the amplification of the housekeeping gene actin. As demonstrated here, using 23 cycles for the amplification of actin will allow for semiquantitative analysis of PCR products.
Elevated Levels of HA in Supernatant of FS Stimulated with TGF-␤ or TNF␣-Enzyme-linked immunosorbent assay exper- iments were performed in order to determine whether the observed elevated levels of HAS1 mRNA following stimulation of FS also result in increased HA production. Cells cultured in 6-well plates were washed two times with medium (37°C) before each experiment in order to remove HA accumulated prior to stimulation. FS were cultured in 1 ml of medium for the duration of the experiments. HA levels were measured as described under "Experimental Procedures." At all time points measured (10, 16, and 28 h), TGF-␤ (1 ng/ml)-treated cells produced higher levels of detectable HA than cells stimulated with TNF␣ (10 ng/ml). A comparison of HA levels detectable in culture medium (MEDIUM) and the medium of unstimulated cells (FS) reveals that untreated cells also produce HA, albeit to a lesser degree. Fig. 2 shows one of three experiments, each performed in duplicate, were cells were left untreated (FS), treated with 1 ng/ml of TGF-␤ (FS ϩ TGF), or treated with 10 ng/ml of TNF␣ (FS ϩ TNF). In the representative experiment shown in Fig. 2, where HA was measured after 16 h, HA levels were as follows. The HA in the medium used for cell culture was 1.9 ng/ml (STD 0.77), HA in unstimulated cells was 81 ng/ml (STD 4.1), and HA levels in cells treated with TGF-␤ and TNF␣ were 135 (STD 1.5) and 114 ng/ml (STD 3.9) respectively. Furthermore, three experiments terminated at 16 h revealed that, compared with unstimulated cells, treating FS with TGF-␤ leads to 70 Ϯ 9% higher levels of HA, whereas exposing FS to TNF␣ results in 47 Ϯ 7% higher HA levels. These enzyme-linked immunosorbent assay data seem to correlate well with changes in mRNA observed in treated and unstimulated cells.
TGF-␤ Induces HA Synthesis in Membrane Extracts-Although the demonstrated increase of HA in the culture medium of TGF-␤-treated cells (Fig. 2) is an indicator that the observed elevated mRNA levels of HAS1 resulted in elevated HA levels, we measured HA synthase activity in untreated as well as TGF-␤-treated cell membrane extracts. An in vitro HA synthase assay was established using published protocols (31,32). FS were grown to high density in 15-cm tissue culture dishes. Cells were washed three times with prewarmed medium. Subsequently, FS were left untreated or incubated with TGF-␤ (1 ng/ml) for 8 h. Membrane extracts were prepared as described under "Experimental Procedures." Cell membrane extracts (25 g) were incubated with UDP-[ 14 C]glucuronic acid and UDP-N-acetylglucosamine. After terminating the synthase activity by heating the reaction mix, half of the reaction was treated with Streptomyces hylauronidase before deproteinization. Instead of separating long chain HA molecules by gel filtration, a procedure that results in elution of the HA in the void volume of columns as well as in lamentably large amounts of liquid radioactive waste, we separated high molecular weight HA from [ 14 C]glucuronic acid monomers utilizing a filtration device that separates molecules according to their size. Unincorporated [ 14 C]glucuronic acid was removed by filtration through a Microcon YM-100 unit, a device that retains high molecular weight molecules (Ն100,000 daltons). Shown in Fig. 3 are data that demonstrate that TGF-␤ leads to an increase in HA synthesis as measured by [ 14 C]glucuronic acid incorporation into HA polymers. The y axis shows [ 14 C]glucuronic acid incorporation (dpm). Given on the x-axis are dpm measured in samples containing membrane extract of nontreated cells (lane MED) and in extracts of FS treated with 1 ng/ml of TGF-␤ for 8 h (lane TGF). Experiments were done in duplicate. In the graph shown, the remaining counts in dpm after hyaluronidase treatment were subtracted from corresponding non-hyaluronidase-treated samples.
Dose Effect of TGF-␤ on HAS1 and HAS3 mRNA Levels-After establishing the time requirements for TGF-␤-induced HAS1 mRNA activation, dose effects of TGF-␤ were studied. TGF-␤, at doses ranging from 0.01 to 30 ng/ml, was used to induce HAS mRNA. Such dose-finding experiments revealed that maximal levels of HAS1 mRNA were observed when doses of 0.5-1 ng/ml TGF-␤ were used to stimulate cells. Interestingly, such experiments also revealed that TGF-␤ dose-dependently suppressed HAS3 mRNA levels. Fig. 4 shows a comparison of mRNA levels of HAS1 and HAS3 in FS following stimulation of FS with increasing amounts of TGF-␤ for 6 h. Cells were left untreated (lanes labeled 0) or stimulated with increasing amounts of TGF-␤. The left part of Fig. 4 shows a representative experiment monitoring HAS1 mRNA levels following stimulation with 0.15, 0.6, 2.5, and 10 ng/ml TGF-␤. Shown on the right side of Fig. 4 is a representative experiment demonstrating dose-dependent inhibition of HAS3 mRNA accumulation by TGF-␤. mRNA for actin or GAPDH was co-amplified as a control and used for adjustments in calculating HAS mRNA levels. As shown in Fig.  4, such experiments demonstrated again that HAS1 levels in unstimulated FS are very low, yet mRNA of HAS3 is abundant and easily detectable. More importantly, such experiments demonstrated a differential effect of TGF-␤ on HAS1 and HAS3 mRNA accumulation. Whereas TGF-␤ induces HAS1 levels in cultured FS, levels of HAS3 mRNA are dose-dependently suppressed. Whereas 10 ng of TGF-␤ suppressed HAS3 accumulation by more than 70% (72% in this experiment), 2.5 ng/ml TGF-␤ was less inhibitory (47% inhibition in this particular experiment), and the lowest amount of TGF-␤ used in these experiments (0.6 ng/ml) had the least suppressive effect (31% inhibition).

Utilization of MAPK Inhibitors to Determine TGF-␤-induced Signaling Pathways in FS-
Our working hypothesis is based on the assumption that in RA patients, one or several of the HAS genes are activated in an uncontrolled manner. In subsequent experiments, we focused on TGF-␤-induced activation mechanisms of HAS1, since this gene is the only one that could be readily activated by the stimuli tested.
In all of the experiments shown in Fig. 5, FS were pretreated with the indicated inhibitors for 30 min. After that, TGF-␤ (1 ng/ml) was added, and cells were incubated for an additional 6 h. Cells were subsequently washed with ice-cold phosphatebuffered saline and placed on ice, after which mRNA was extracted and RT-PCR was performed as described under "Experimental Procedures." Inhibitors were dissolved according to the manufacturers' suggestions and used at 2 and 5 times the IC 50 , respectively. In lanes labeled A, cells were left untreated; in lanes B, inhibitors at 2 times IC 50 were added; and in lanes C, inhibitors at 5 times IC 50 were added. Lanes E and F were pretreated with inhibitors at 2 times IC 50 and 5 times IC 50 , respectively, for 30 min; in addition to that, TGF-␤ was added in lanes D-F. Fig. 5A shows a representative experiment in which a specific JNK1 inhibitor was used to test the involvement of JNK in TGF-␤-induced activation of HAS1. Fig. 5 again demonstrates low basal levels of HAS1 in unstimulated cells (lane A) and significantly elevated levels of HAS1 mRNA following stimulation of FS by TGF-␤ (1 ng/ml) (lane D). More importantly, this figure demonstrates that the cell-permeable and selective JNK inhibitor (JNK inhibitor II) affected neither basal nor TGF-␤induced HAS1 mRNA accumulation significantly. mRNA for actin, shown in the lower section of Fig. 5, was used as a control and for adjustments for the calculation of inhibition.
Next, PD 98059, a specific MAPK/extracellular signal-regulated kinase (MEK) inhibitor was used to study the involvement of the MEK pathway in TGF-␤-induced HAS1 activation. Again, this substance was used at 2 and 5 times IC 50 , respectively. As shown in Fig. 5B, blocking MEK leads to lower levels of mRNA in FS stimulated with TGF-␤. Whereas the use of 2 times the IC 50 (lane E) prior to stimulation with TGF-␤ (1 ng/ml for 6 h) leads to ϳ45% lower levels of HAS1 mRNA, using higher amounts of this inhibitor (5ϫ IC 50 ; lane F) results in a more pronounced inhibition. In the experiment shown in Fig. 5B, the inhibition was calculated to be ϳ52%. Interestingly, this inhibitor also lowered noninduced levels of HAS1, as a comparison of the HAS1 mRNA levels in untreated (lane A) and inhibitor only-treated cells (lanes B and C, respectively) reveals.
SB 202190 was used to investigate the involvement of the p38 MAPK pathway in TGF-␤-induced HAS1 mRNA accumulation in FS. As shown in Fig. 5C, blocking p38 MAPK is the most efficient way to suppress TGF-␤-induced HAS1 mRNA synthesis. As demonstrated in this and similar experiments, the use of 0.6 M (2 times IC 50 ) is sufficient to repeatedly achieve a reduction of close to 90%. As with the inhibition of the MEK pathway, blocking the p38 pathway also led to lower levels of HAS1 mRNA in unstimulated cells (lane B (2 times IC 50 ) and lane C (5 times IC 50 ), respectively).
Viability of FS was not affected in these experiments by the doses of inhibitors used as judged by visual inspection of the cells under a phase microscope as well as through the use of a trypan blue exclusion assay.
Activation of the p38 MAPK Pathway by TGF-␤ -That p38 MAPK in certain cells can be activated by TGF-␤ has been FIG. 4. TGF-␤ induces HAS1 but suppresses HAS3 mRNA in FS in a dose-dependent manner. Cells were incubated with the indicated amounts (0, 0.15, 0.6, 2.5, and 10 ng/ml) of TGF-␤ for 6 h. Following RT-PCR and separation of the reaction mix, gels were scanned on a Fluorimager and quantitated using Im-ageQuant software. Gels for quantitation of HAS1, HAS3, and actin are shown in the lower section, and graphs resulting from HAS1 and HAS3 mRNA are shown in the upper section. Whereas HAS1 mRNA is dose-dependently activated by TGF-␤ (left area), constitutively high levels of HAS3 mRNA in FS are dose-dependently suppressed by TGF-␤ as shown in the right section. Actin-adjusted inhibition of HAS3 mRNA in this experiment was calculated to be 72% using 10 ng/ml TGF-␤, 47% using 2.5 ng/ml TGF-␤, and 31% using 0.6 ng/ml.
shown (34). We confirmed TGF-␤-induced activation of p38 in FS by Western blot experiments (data not shown). FS were left untreated, stimulated with TGF-␤ (1 ng/ml) for 60 min, or preincubated with a specific p38 MAPK inhibitor (SB 202190) (2 times IC 50 ) for 30 min followed by TGF-␤ for 60 min or with inhibitor only. After that, cells were washed three times with ice-cold phosphate-buffered saline and placed on ice. Protein extraction and Western blot procedures were carried out as described under "Experimental Procedures." Blots were stained with an antibody specific for the phosphorylated form of p38 MAPK. The use of equal amounts of protein in these experiments was confirmed by staining membranes with Ponceau red and/or with antibodies recognizing tubulin (data not shown). Our data implicated that phosphorylation of p38 is readily detected in FS following stimulation of these cells with TGF-␤. Furthermore, the p38 MAPK inhibitor SB 202190 prevented TGF-␤-induced phosphorylation of p38 MAPK. DISCUSSION Since unfettered HA synthesis in affected joints of RA patients is one of the early signs of the onset of the disease as well as for other reasons outlined below, it is tempting to speculate that pathological HA release itself might be involved in onset and progression of RA. Additionally, a series of observations, made over the past decades, support such a hypothesis. HA or HA degradation products have been demonstrated to be potent chemoattractants for leukocytes (35). HA has also been shown to promote vascularization and to exert angiogenic properties (16,35). Furthermore, HA is the principle binding molecule for CD44 as well as a series of other so-called hyaladherins, underlining the importance of the role of HA in controlling cell migration. Based on the above data, it seems appealing to postulate that HA itself might well be amplifying the vicious cycle that seems to be so detrimental in RA, namely, a series of events that might be initiated by unfettered HA synthesis followed by subsequent HA degradation, leading to activation/ migration of leukocytes and the release of cytokines. This in turn leads to more HA production/release and further activation of leukocytes as well as to the formation of new blood vessels that, in turn, facilitate migration of even more cells into affected joints. Therefore, a better understanding of the regulatory pathways of HAS activation might allow us to investigate the involvement of HA in RA. For these reasons, we used cultured FS to study expression patterns and signaling pathways of the genes encoding HAS.
We first investigated mRNA levels of HAS1, HAS2, and HAS3 in resting cells and found that levels of HAS3 mRNA were highest in these cells. mRNA levels of HAS2 in resting cells are readily detectable but are consistently lower than mRNA levels of HAS3. Whereas HAS3 and HAS2 are constitutively expressed in the cells tested, levels of HAS1 mRNA in unstimulated FS were uniformly low. IL-1␤, TNF␣, and TGF-␤ are among the cytokines that can be detected in the synovial fluid of inflamed joints (36,37). We used these stimuli plus IL-8 and PMA and tested them for their ability to activate HAS genes. The outcome of such experiments was unexpected but consistent with other reports demonstrating the potential of TGF-␤ as an important regulator of HA (38,39). As demonstrated in Fig. 1A, TGF-␤ was by far the most potent activator of HAS1 mRNA synthesis among the stimuli tested. Furthermore, as demonstrated in Fig. 2, changes in levels of HAS1 mRNA following stimulation with TGF-␤ and TNF␣ are mirrored in similar changes of HA. These data demonstrate that, similar to results of RT-PCR experiments, levels of HA were consistently higher in cells treated with TGF-␤ than in cells treated with TNF␣. The interpretation that TGF-␤ not only increases mRNA levels but also results in increased HA syntheses is supported by the in vitro HA synthase assays shown in Fig. 3. Such assays demonstrated increased HA polymerization in membrane extracts of TGF-␤-treated cells. These exper- iments seem a clear indication that TGF-␤-induced changes at the mRNA level do result in similar changes at the protein as well as at the product level.
Since TGF-␤ might contribute to elevated HA levels by suppressing the synthesis of HA-degrading enzymes, we tested the effect of TGF-␤ (1 ng/ml) on mRNA levels of a series of known hyaluronidases and found no differences in mRNA levels of HYAL1, HYAL2, HYAL3, HYAL4, and PH-20 when comparing mRNA levels in untreated and TGF-␤-treated cells; a representative experiment demonstrating this is shown in Fig. 6.
These findings are also in accordance with published data from a time where no distinction between the different HAS genes could be made (36). Besides the well known effects of TNF␣ in RA, TGF-␤ levels seem of higher relevance for HAS activation and might indeed play an important role in the progression of RA. Such a concept is supported by the elevated levels of latent and active TGF-␤ found in synovial fluid of RA patients (37,40,41).
Our experiments further demonstrated that none of the stimuli used led to a significant increase in HAS2 mRNA. Our observation that the gene for HAS2, which seems essential in embryogenesis (42,43), is activated neither by TNF␣, TGF-␤, nor by IL-1␤ confirms similar findings by Oguchi et al. (39). Interestingly, whereas TGF-␤ activated HAS1 mRNA synthesis in a dose-dependent manner, mRNA levels for HAS3 decreased with increasing amounts of TGF-␤. Although higher doses of TGF-␤, namely 10 ng/ml, had to be used to achieve nearly total suppression of HAS3 mRNA, only about 1 ng/ml of TGF-␤ resulted in maximal activation of HAS1 mRNA accu-mulation. Taken together, such modulatory effects of TGF-␤ will result in a shift in the composition of total HA and might be important for a better understanding of biological effects of this cytokine. However, these findings also indicate that the qualitative properties of total HA could be altered without being reflected in quantitative changes. Furthermore, the observation of the differential effect of TGF-␤ on the genes encoding HAS could be of importance in fields other than rheumatic diseases (e.g. in cancer biology). It is well documented that HA is essential for the progression of certain forms of cancer and, at the same time, it has been noted that TGF-␤ exerts intriguing dose effects on cancer growth (44 -46).
With regard to HA synthesis in RA, in joints of RA patients HA is clearly produced in excess. Elevated levels of HA can also be detected in peripheral blood of these patients (47). It was therefore suggested that measuring elevated levels of HA in serum might be of diagnostic value (48). FS are a major source of HA production in RA (36). Since our in vitro experiments determined that HAS1 was the only gene that was up-regulated, it became the focus of subsequent experiments.
One of many ways to test the biological function of a gene is to block signaling pathways leading to the activation of this gene. Prerequisite for such experiments is an understanding of the intracellular signaling pathways that are activated by a given stimulus. The second part of our work focused on TGF-␤-induced signaling cascades leading to the activation of the HAS1 gene. The family of TGF cytokines are recognized as important signaling molecules, seemingly having myriad functions (49,50). The best defined pathway of the TGF-␤ signaling cascade is the TGF-␤-Smad network (51,52). Although to date, Smads are the only known TGF-␤ receptor substrates, it becomes increasingly clear that there are other signaling pathways involved in TGF-␤-induced gene regulation. TGF-␤ has been shown to activate the transcription factor activator protein-1 through the extracellular signal-regulated kinase pathway (34). Also, a transcriptional cross-talk between Smad, extracellular signal-regulated kinase 1/2, and p38 MAPK has been demonstrated in the TGF-␤-induced aggrecan activation (53) as well as in many other MAPK pathways (54 -57).
Nearly all cell surface receptors utilize one or more of the MAPK cascades (58). There are three principle MAPK pathways in mammalian cells, all of which have been implicated in TGF-␤-induced signaling and have specific inhibitors commercially available for them (59,60). Using such specific inhibitors, we demonstrated that the p38 MAPK pathway is the main pathway utilized by TGF-␤-induced activation of HAS1. Whereas the use of SB 202190, a p38 MAPK inhibitor, nearly completely abolished TGF-␤-induced HAS1 activation in the FS used, inhibiting the MEK pathway was significantly less effective in suppressing the observed TGF-␤ effect. Furthermore, experiments blocking the JNK/stress-activated protein kinase indicated that this pathway is not utilized in these cells. The activation of signaling pathways by TGF-␤ seems to be cell-specific. Whereas some cell types show rapid and marked activation of JNK/stress-activated protein kinase following TGF-␤ treatment, such a response is absent or minimal in others (61), whereas still others show patterns of TGF-␤-induced MAPK activation identical to those observed in our experiments (e.g. TGF-␤ treatment of pulmonary fibroblasts resulting in a rapid activation of p38 MAPK, a delayed activation of the extracellular signal-regulated kinase pathway, and no activation of JNK) (34).
Although to date there are no reports demonstrating a causal link between the up-regulation of HAS genes and the subsequent infiltration of leukocytes, our working hypothesis that unfettered HA production might well be the initiating and/or  TGF). HYAL1, HYAL2, and PH-20 mRNA can be detected in FS. As shown here, TGF-␤ treatment does not lead to changes in mRNA levels of these genes. mRNA levels of HYAL3 are undetectable in FS but readily detectable in carcinoma cells used as control (lowest panel). Such experiments confirm the use of functioning primer pair as well as correct RT-PCR conditions for the detection of HYAL3. Of the remaining hyaluronidase genes, HYAL4 is a chondroitinase that has no activity against hyaluronan, and HYALP1 is a pseudogene. In addition, mRNA levels of actin are presented in each panel as a demonstration of equal use of mRNA.
propagating event in RA is supported by a recently published report (62). This group tested a series of glycosaminoglycans in an in vivo model and found that, among others, HA injection can cause RA.
In summary, we demonstrate that TGF-␤ treatment can lead to elevated levels of HA. Furthermore, TGF-␤ exerts potent effects on HAS gene regulation. Although TGF-␤ strongly activates HAS1, this cytokine is a powerful suppressor of HAS3. Whereas blocking MEK does have a significant inhibitory effect on TGF-␤-induced HAS1 activation, it is the p38 MAPK pathway that seems to be the main route utilized by TGF-␤, since blocking this pathway led to a nearly complete suppression of TGF-␤-induced HAS1 activation. The dissection of TGF-␤-induced MAPK pathways might allow us to test parts of our working hypothesis, namely that HAS are indeed involved in the progression of RA.