Reduced chondrogenic matrix accumulation by 4-methylumbelliferone reveals the potential for selective targeting of UDP-glucose dehydrogenase.

4-Methylumbelliferone (4-MU) is described as a selective inhibitor of hyaluronan (HA) production. It is thought that 4-MU depletes UDP-glucuronic acid (UDP-GlcUA) substrate for HA synthesis and also suppresses HA-synthase expression. The possibility that 4-MU exerts at least some of its actions via regulation of UDP-glucose dehydrogenase (UGDH), a key enzyme required for both HA and sulphated-glycosaminoglycan (sGAG) production, remains unexplored. We therefore examined the effects of 4-MU on basal and retroviral UGDH-driven HA and sGAG release in cells derived from chick articular cartilage and its influence upon UGDH protein and mRNA expression and HA and sGAG production. We found that 4-MU: i) suppressed UGDH mRNA and protein expression and chondrogenic matrix accumulation in chick limb bud micromass culture, ii) significantly reduced both HA and sGAG production and iii) more selectively reversed the potentiating effects of UGDH overexpression on the production of HA than sGAG. Understanding how GAG synthesis is controlled and the mechanism of 4-MU action may inform its future clinical success.

Introduction

Specific inhibition of hyaluronan (HA) synthesis is an attractive therapeutic option for the treatment of diseases associated with elevated HA levels, such as atherosclerosis, hepatitis and tumour growth (Fischer and Schrör, 2007; McHutchison et al., 2000; Simpson et al., 2001). It has been reported that 4-methylumbelliferone (4-MU) can selectively inhibit HA synthesis and pericellular HA-rich matrix formation in a number of cell types (Nakamura et al., 1995; Morohashi et al., 2006; Yoshihara et al., 2005). The attractiveness of 4-MU as a therapeutic option has also been supported by the demonstration that it shows anti-metastatic actions in vivo (Yoshihara et al., 2005).

Kakizaki et al. described a mechanism of action for the inhibition of HA synthesis by 4-MU in rat 3Y1 fibroblasts. This was shown to involve glucuronidation of 4-MU by endogenous UDP-glucuronyltransferase (UGT), resulting in a depletion of UDP-glucuronic acid (UDP-GlcUA). It was concluded that excess glucuronidation of 4-MU by endogenous UGT depleted the UDP-GlcUA pool, which in turn restricted the availability of this essential substrate for HA synthesis. Such depletion of UDP-GlcUA in the cellular pool may, however, be expected to affect the biosynthesis of other GlcUA-containing glycosaminoglycans (GAGs), such as heparan and chondroitin sulphate (CS). It has been shown, however, that 4-MU has no affect on the biosynthesis of sulphated GAGs (sGAGs) in human skin fibroblasts (Nakamura et al., 1995, 1997). For this reason the mechanism underpinning the specificity demonstrated by 4-MU for inhibiting production of only non-sulphated GlcUA-containing GAG, HA, remains somewhat enigmatic.

Several possible explanations for the selective targeting of HA synthesis by 4-MU have been proposed. These include the specific targeting of plasma membrane-located HAS over the Golgi-located glycosyltransferases required in sGAG biosynthesis. Similarly, the relative cell membrane enrichment of UGT activity and, therefore, differential restriction of UDP-GlcUA supply close to HAS have also been proposed as a possible explanation. It has, been shown that the extent of the inhibition of HA synthesis by 4-MU can be reduced by an excess of exogenous UDP-GlcUA (Kakizaki et al., 2004), raising the relatively unexplored possibility that the cellular supply of UDP-GlcUA may modify the influence of 4-MU.

UDP-GlcUA is the product of UDP-glucose dehydrogenase (UGDH) activity. UGDH is a key enzyme required for the conversion of UDP-glucose into UDP-GlcUA and is considered both rate-limiting in GAG synthesis and pivotal in determining the specific forms of GAGs synthesised (Hickery et al., 2003; Pitsillides, 2003). Indeed, our recent studies have demonstrated that direct modulation of UGDH expression levels is sufficient to promote both marked increases in HA as well as sGAG production and also to enhance chondrogenesis in micromass cultures in vitro (Clarkin et al., 2011). Thus we propose that UGDH could act as a potential target for the actions of 4-MU.

Recent studies suggest that these actions of 4-MU on post-translational control of UDP-GlcUA substrate supply, are complemented by a more complex mechanism of action. Thus, 4-MU has been shown to influence the mRNA expression for other components of the HA synthetic pathway, such as HA-synthase (HAS) (Kakizaki et al., 2004; Kultti et al., 2009). Despite this, the possibility that 4-MU exerts at least some of its actions by regulating the expression of UGDH, another essential up-stream component of this HA synthetic pathway, remains unexplored. Herein, we examine whether 4-MU selectively modulates in vitro chondrogenic matrix accumulation by targeting HA production, whether it modifies UGDH expression and whether retrovirally-driven overexpression of UGDH can effectively overcome the inhibition of HA production by 4-MU in chick articular surface (AS) cells.

Results

4-MU treatment inhibits both HA and sGAG production in chick limb bud micromass cultures

4-MU has previously been found to suppress the release of HA, but not sGAG, from a range of cell types. It has been proposed that 4-MU achieves this inhibition by depleting the UDP-GlcUA substrate supply. If this is the case, then the UDP-GlcUA supply that is also required in sGAG synthesis, may also be influenced by 4-MU. We therefore investigated this possibility using chick limb bud micromass cultures, which produce both HA and sGAGs during the process of chondrogenesis. Treatment with 4-MU (200 μM) for 6 days reduced sGAG content and chondrogenic nodule formation in the micromass cultures (visualised by intensity of Alcian blue staining; Fig. 1A–B). To examine whether this reduction in Alcian blue staining intensity by 4-MU treatment was due only to failure of sGAG retention in chondrogenic nodules, we also measured the HA and sGAGs released into the medium. This highlighted a significant reduction in sGAG release (50%) but a much greater reduction in HA released into the media (88%; Fig. 1C–D).

Fig. 1

Effects of 4-MU on accumulation of Alcian blue-positive cartilage matrix proteoglycans, HA and sGAG release from micromass cultures prepared from stage 23/24 whole limb bud mesenchyme and on UGDH protein and mRNA expression. Mesenchymal cells were plated at high density in 10 μl suspensions. Each panel (A/B) depicts a light micrograph (×0.2 objective) of single micromass spot cultures. In stage 23/24 whole limb mesenchyme cultures, 4-MU decreased deposition of Alcian blue-stainable matrix proteoglycans relative to vehicle (methanol) control cultures following 6 days of treatment. Medium was changed once and collected at the end of each culture period and assayed for either HA (C) or sGAGs (D) at day 6. Data are expressed as mean ± SEM and are from a total of 6 individual micromass cultures per treatment (***p < 0.001 and *****p < 0.00001). Chick AS cells were serum-deprived for 16 h and treated 4-MU (200 μM) in the absence of serum for 24 h prior to lysis for either RNA or protein extraction. Whole cell lysates were analysed by SDS-PAGE and immunoblotting using a UGDH antibody (F); and equal protein loading was confirmed with ERK1 antibody (F). Densitometric analyses of UGDH protein (E) results are obtained from three independent experiments (*p < 0.05 versus control). RNA isolates (G) were reverse transcribed and cDNA amplified by RT-PCR for UGDH, HAS-2 and GAPDH.

4-MU suppresses UGDH mRNA and protein expression

Our previous studies revealed a regulatory role for UGDH in controlling chondrogenesis (Clarkin et al., 2011). Due to the negative influence of 4-MU upon chondrogenic matrix accumulation (Fig. 1A/B) we therefore investigated whether this was also associated with modulation of the UGDH protein and mRNA expression levels. We found that 4-MU treatment reduced UGDH mRNA and HAS-2 expression levels in AS cells as visualised by semi-quantitative PCR (Fig. 1E). In parallel studies, 4-MU treatment for 24 h produced a significant but modest suppression of UGDH protein levels in AS cells when assessed by Western blotting and densitometric analysis (Fig. 1F/G). Media from these experiments were also collected and assayed for HA and sGAG release. In association with the reduced UGDH protein and mRNA expression described, 4-MU also decreased release of both HA (control, 460 ± 12.1 ng/ml; 4-MU, 175 ± 38.2 ng/ml) and sulphated GAGs (control, 88 ± 0.76 ng/ml; 4-MU, 6.11 ± 1.4 ng/ml) from AS cells after 24 h of treatment. These results suggest that the described effects of 4-MU on HA/sGAG production may be at least partly achieved via modulation of UGDH levels.

4-MU inhibits basal HA release from AS cells and retroviral UGDH overexpression accentuates this inhibitory influence

To examine whether direct modulation of UGDH expression levels can overcome the effects of 4-MU on sGAG and HA production, we used an established retroviral expression system (RCAS) to drive the production of UGDH (Fig. 2A) (Clarkin et al., 2011). Consistent with previous findings we showed that AS cells released significantly more HA than sGAGs in control-RCAS transfected cultures and that 4-MU modifies UGDH mRNA expression levels and exerts greater inhibition on HA production and release than on sGAG (44% decrease and 15%, respectively; Fig. 2A–D).

Fig. 2

Effects of 4-MU on HA and sGAG release in UGDH transfectants. Chick AS cells were infected with either control-RCAS or UGDH-RCAS and overexpression was verified by western blotting and PCR showing UGDH mRNA expression levels in the presence and absence of 4-MU (A). Cells were serum-deprived for 16 h, treated in the absence of serum with 200 μM 4-MU for 24 h and HA (B) and sGAG (C) release quantified. Percentage reduction in HA and sGAG release in response to 4MU in control and UGDH transfectants is also tabulated (D). Results are mean ± SEM of n = 6 transfections (ns; nonsignificant, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001).

Retroviral overexpression of UGDH was found to produce significant increases (~ 2 fold) in the release of both HA and sGAG (Fig. 2B and C). Furthermore, 4-MU treatment significantly reduced HA and sGAG release from AS cell over-expressing UGDH (Fig. 2B–C), inferring that 4-MU can effectively block the effects of UGDH gain-of-function on both HA and sGAG synthesis. Again, 4MU appears to exert a much more pronounced inhibition of UGDH-mediated HA than sGAG release (70% decrease and 22%, respectively; Fig. 2B–D). Our results therefore show that synthesis and release of HA and sGAGs are differentially affected by 4-MU. Specifically, the HA release that presumably is selectively driven by ectopic UGDH is completely abrogated by treatment with 4-MU. In contrast, sGAG release is significantly diminished by 4-MU treatment but remains elevated over baseline (2 fold) in the presence of ectopic UGDH (Fig. 2B). This differential influence on HA and sGAG release suggests that 4-MU targets UGDH directly to mediate its selective actions on HA production. Alternatively, these data may suggest that retroviral UGDH overexpression fails to overcome the inhibitory influence of 4-MU on the HA release, but to a limited extent achieves such rescue of sGAG production by AS cells. Thus, our results imply that 4-MU is not capable of completely inhibiting UGDH-retrovirally-induced increases in sGAG release and suggest the existence of divergent mechanisms of action in this context.

Discussion

Previous studies have shown that 4-MU inhibits HA synthesis by decreasing HAS activity and hence polymerisation of its constituent UDP-sugar substrates (UDP-GlcNAc and UDP-GlcUA). We have examined whether 4-MU also modifies expression of UGDH, an enzyme involved in producing one of these sugar substrates (UDP-GlcUA). As UDP-GlcUA is also a vital component of sGAGs, UGDH activity is putatively a rate-limiting step in the synthesis of both HA and sGAGs. We therefore also monitored HA and sGAG production following the exposure of limb bud micromass cultures to 4-MU. We found that in addition to its known function in blocking HA release, 4-MU has a novel role in modulating sGAG production. We found that 4-MU suppresses in vitro chondrogenic matrix accumulation and also diminishes UGDH protein and mRNA levels. With regard to its mechanism of action, we show that 4-MU more selectively reverses the potentiating effects of UGDH overexpression on HA (versus sGAG) production. This suggests that 4-MU may inhibit GAG synthesis via UGDH, which itself is more intimately linked with HA, than sGAG production in this specific cell system.

Previous studies using human aortic smooth muscle cells (AoSMC) found that down-regulation of UGDH reduced HA production but did not influence CS (Vigetti et al., 2006). Later studies in AoSMC highlighted similar specificity of 4-MU as an inhibitor of HA production (with no appreciable attenuation in CS production); this was verified using the particle exclusion assay and hyaluronidase digestion confirmed that AoSMC GAG content was comprised principally of HA. These elegant studies supported the view that although central to the synthesis of both HA and sGAGs, UGDH specifically contributes to HA production and that the latter is also selectively targeted by 4-MU (Vigetti et al., 2009).

It is possible, however, that such differential effects on GAG production and release may differ in distinct cell types (Clarkin et al., 2011). Indeed, it has been shown that human skin fibroblasts which produce an HA-deficient extracellular matrix, can also increase production of the small proteoglycan, decorin (with increased size) when treated with 4-MU (Funahashi et al., 2009). For this reason we have chosen a connective-tissue cell type derived from chick limb bud mesenchyme for their capacity to differentiate into chondrocytes and secrete abundant amounts of both sulphated and unsulphated GAGs. The high density micromass cell system for culturing chick limb bud mesenchyme is known to trigger chondrogenesis, promoting secretion of a cartilagenous matrix containing both HA and large amounts of sGAG, mainly CS connected to the proteoglycan, aggrecan. Our use of this micromass culture system to augment production of both sGAG and HA, enabled us to conclude that 4-MU not only targets HA release but also diminishes sGAG production. Commensurate reduction in UGDH mRNA and protein levels suggests that some effects of 4-MU may be achieved by modulating UGDH.

The mechanism of action for 4-MU is known to be complex. A recent study using melanoma and carcinoma cells (Kultti et al., 2009) has proposed that 4-MU inhibits HA synthesis by depleting the cellular UDP-GlcUA pool, as well as by down-regulating HAS2 and 3 expression. Unfortunately, however, this study did not determine the effects of 4MU on UGDH expression in these cells. It is possible that 4-MU-induced decreases in UGDH mRNA levels described by Vigetti et al. (2006), and the decreased UGDH mRNA and protein expression levels described using semi-quantitative methods herein, would almost certainly result in reduced conversion of UDP-glucose to UDP-GlcUA and explain the 4-MU associated depletion of the UDP-GlcUA pool seen by Kultti et al. (2009).

To investigate whether 4-MU elicits its suppression of GAG synthesis by posttranslational influences, the effect of 4-MU was monitored in chick AS cells over-expressing UGDH through retroviral transfection. Such increases in UGDH expression are likely to increase the amount of UDP-GlcUA precursors available, making increases in GAG production feasible. We confirm this possibility by showing that UGDH over-expression can indeed increase both HA and sGAG release from AS cells. One of the most important findings of this current study is that even when these increases are driven selectively by UGDH over-expression, 4-MU is still capable of completely and relatively selectively abolishing all of the UGDH-dependent increases in HA release. One possible explanation for this selective targeting of HA synthesis by 4-MU includes the relative insulation against changes in substrate availability in the Golgi versus the cytosol. Thus, the Golgi environment will be more controlled and less susceptible to large fluxes of substrates than the cytosol, due to the existence of its antiporter mechanism for substrate entry (Schwartz et al., 1998). Our data are therefore consistent with the notion that UGDH-mediated changes in UDP-GlcUA substrate availability and targeting by 4-MU will more effectively limit the activity of plasma membrane-located HAS over the Golgi-located glycosyltransferases required in sGAG biosynthesis (Philipson and Schwartz, 1984). In this way, the contribution made by UGDH to HA production would be subject to targeting by 4-MU, but the production of sGAGs would remain less affected.

In contrast, whilst 4-MU significantly reduced AS cell sGAG release, it failed to reduce the retroviral UGDH-related sGAG production to basal levels. These data suggest less effective inhibition of sGAG by 4-MU and imply the existence of other mechanisms by which 4-MU reduces sGAG synthesis and specifically targets HA production for inhibition. We have not measured UDP-sugar precursor concentrations, however, other studies have shown that the UDP-GlcUA pool available for HA synthesis is depleted by 4-MU (Kakizaki et al., 2004; Kultti et al., 2009; Vigetti et al., 2009). Indeed, this and the fact that both HAS and UDP-glucuronyl transferases (UGTs) utilise UDP-GlcUA as a substrate (Kakizaki et al., 2004; Kultti et al., 2009; Vigetti et al., 2009), have contributed to the view that 4-MU, via its UGT-mediated glucuronidation (to 4MU-glucuronide), acts to diminish the UDP-GlcUA sugar precursors available for GAG synthesis. T his is also supported by our studies which suggest that retroviral UGDH over-expression may increase UDP-GlcUA levels in AS cells and that increased UGT-mediated glucuronidation of 4MU acts similarly to deplete the available UDP-GlcUA pool for GAG synthesis. However, this proposed mechanism of 4-MU action alone cannot account for the greater selectivity exhibited for HA production in the context where UGDH is over-expressed. Clearly there appear to be multiple mechanisms by which 4-MU selectively regulates GAG synthesis and studies aimed at defining how 4-MU affects kinetics of the many enzymes that are involved in HA and GAG production are still required. Down regulation of HAS expression and the 4-MU-mediated decreases in UGDH expression revealed herein suggest multiple additional functions for this inhibitor at the pre- and post-transcriptional levels to target HA synthesis.

Several studies have alluded to the fact that 4-MU has minimal effect on sGAG synthesis and is therefore specific for HA synthesis (Nakamura et al., 1995, 1997; Nakazawa et al., 2006). This has led to the use of 4-MU to specifically block HA synthesis in certain anticancer therapies and recently it was shown to reduce tumour growth and metastasis by preventing HA synthesis and HA-associated signalling in a mouse model of prostate tumorigenesis (Lokeshwar et al., 2010). Our studies support the notion that a range of different cell systems show blockade of HA synthesis in response to 4-MU, but we find that cells derived from the chondrogenic cell lineage also show reduced synthesis of sGAG. In this regard, it is particularly pertinent that our studies show that retrovirally-driven increases in UGDH expression act to markedly promote the selectivity of 4-MU for HA production. These findings can be interpreted to suggest either that increases in UGDH expression are more intimately linked to HA than to sGAG production, or that some alternative mechanism by which 4-MU specifically targets UGDH is responsible for this selectivity. It remains that 4-MU clearly influences UGDH expression and cartilage formation in cells derived from the chondrogenic lineage and that further study into its mechanism of action in this cell type, producing a broad GAG repertoire, is warranted.

Experimental procedures

Chick limb bud micromass culture

Mesenchymal cells from stages 20 to 22 embryonic chick wing buds were extracted as previously described and plated at 1 × 105 cells in 10 μl (Clarkin et al., 2011). Following culture for 6 days, cells were fixed in 4% PFA and media retained for analysis. The accumulation of cartilage matrix by micromass cultures is a classic index used to monitor chondrogenesis in vitro and was monitored histochemically by staining fixed cells with 1% Alcian blue (pH 1.0) as previously described (Ogihara et al., 2001).

Isolation of chick articular surface (AS) cells

AS cells were extracted from embryonic stage 42 (18 days) chick tibiotarsi by collagenase digestion as previously described (Bastow et al., 2005; Lewthwaite et al., 2006). Primary cells extracted by two sequential collagenase digestions were seeded onto 75 mm2 flasks (Nunc, Rochester, NY) and grown to confluence in Dulbecco's modified Eagle's medium (DMEM) containing 5% chick serum, 50 μg/ml−1 ascorbic acid, 5 mg/ml d-glucose and gentamycin. At first passage, cells were trypsinised as described (Bastow et al., 2005; Hwang et al., 2003) and plated onto the appropriate tissue culture dishes/trays. Cells were then cultured in serum-containing DMEM medium for 24 h and subsequently in serum-free DMEM for 16 h prior to experimentation in serum-free DMEM.

Measurement of HA concentration in conditioned medium

A competitive enzyme-linked immunosorbance-based assay was used to measure HA concentration in culture medium (Fosang et al., 1990). Briefly, the assay is based upon the competition between HA absorbed onto the plate and HA free in solution, for binding to biotinylated cartilage proteoglycan binding region (G1 domain) (Garcia-Garcia and Anderson, 2003).

Measurement of sGAG concentration in conditioned medium

Sulphated glycosaminoglycans were quantified using an ELISA based method adapted from Farndale et al. (1986) which is dependent on the enzymatic hydrolysis of specific polysaccharidases. Chondrotin sulphate (shark cartilage, mixed isomers, Sigma) standards were prepared 0, 10, 20, 30, 40, and 50 μg/ml to generate a standard curve. 40 μl of media samples was added in duplicate to a 96 well plate and control wells supplemented with 40 μl dH2O. DMMB dye in solvent (4.6 μM, dimethyl methylene blue chloride, Serva 0.5% ethanol, 0.2% sodium formate, 0.2% formic acid; pH 3.5) was added to each of the wells (200 μl) and analysed immediately at a wavelength of 600 nm. The DMMB dye complexes with sulphated GAGs and causes a metachromatic shift, which is detected by a change in absorbance from 600 nm to 535 nm.

RNA extraction and reverse transcription-PCR

AS cells were treated as indicated in the figure legends and RNA was extracted using a Qiagen RNeasy minikit according to the manufacturer's instructions. Total RNA was isolated and as it consists of 95–98% ribosomal RNA we used RNA measurements to normalise for sGAG and HA release in our experiments, where parallel studies used RNA for cDNA synthesis and PCR. A typical group of wells from the same treatment had mean values of 2.37 μg RNA ± 0.06 SEM and represents minimal (~ 2.5% coefficient of variation) change within treatments and undetectable changes in mRNA levels.

Total AS cell RNA was reverse transcribed by incubating 1 μg of RNA with 100 ng/μl of random primers and 200 units of M-Superscript reverse transcriptase for 1 h at 37 °C, 15 min at 42 °C, and 3 min at 98 °C. Gallus gallus UGDH primers were designed and sequenced and the 5′ to 3′ sequences used were as follows: CCATCTACGAGCCAGGGTTA (forward)CGTCTAGCACAAGCTTCAATGT(reverse) HAS2 : GGAATCACCGCTGCTTAC(forward)CAGCAATACAAAGGGCAACA(reverse),GAPDH: GCATTGTGGAGGGTCTTATGA(forward)TCATCATACTTGGCTGGTTTCTC(reverse). The PCR was performed using the following parameters: 30 cycles of denaturation (94 °C for 15 s), annealing (62 °C for 30 s) and elongation (72 °C for 30 s). Each analysis contained a range of standards (known concentrations of the same target sequence).

Preparation of UGDH antibody and immunoblotting protocol

A novel rabbit polyclonal antibody was raised against an antigenic, well-conserved hydrophilic sequence located towards the C-terminus of the chick UGDH peptide (Sigma Genosys, Suffolk, UK) and validated for use by Clarkin et al. (2011). The UGDH peptide sequence used was conserved across species (421KELDYERIHKKMLK434) and was rendered immunogenic by conjugation to keyhole limpet hemocyanin (KLH) at the N-terminus through a cysteine residue. Protein levels in whole cell lysates were measured using a BCA protein assay (Pierce, Rockford Illinois). Proteins (40 μg/lane) were resolved by SDS-PAGE (10%) and transferred to a polyvinylidene difluoride (PVDF) membrane. Blots were incubated with antibodies raised against total ERK1 or UGDH antiserum was used at a 1:1000 dilution. Optical densities of bands were quantified using Quantity One software.

Preparation of the UGDH retroviral expression vector and in vitro infection protocol

cDNA containing the full coding sequence of chicken ugdh was cloned into the avian replication competent retroviral expression vector RCASBP(A), and concentrated ugdh retrovirus (109 cfu/ml) was prepared using standard established protocols as described (Ferrari and Kosher, 2002; Ferrari et al., 1998; Fisher et al., 2006; Kavanagh et al., 2006). For transfection chick AS cells at passage 1 were plated at sub-confluent densities in the presence of UGDH- or control-RCAS (1 μl/105 cells, MOI of between 2 and 5). After 48 h and upon reaching confluence cells were subsequently cultured for 16 h in serum-free DMEM prior to experimentation in serum-free medium.

Statistical analysis

Student's t-test or ANOVA, as appropriate, was used to compare the means of groups of data. Values of p < 0.05 were considered statistically significant.