M. Tammi, Institute of Biomedicine, School of Medicine, University of Eastern Finland, PO Box 1627, FIN-70211, Kuopio, Finland Fax: +358 17 163032 Tel: +358 40 7674826 E-mail: email@example.com
Hyaluronan, a ubiquitous high-molecular-mass glycinoglycan on cell surfaces and in extracellular matrices, has a number of specific signaling functions in cell–cell communication. Changes in its content, molecular mass and turnover rate are crucial for cell proliferation, migration and apoptosis, processes that control tissue remodeling during embryonic development, inflammation, injury and cancer. To maintain tissue homeostasis, the synthesis of hyaluronan must therefore be tightly controlled. In this review, we highlight some recent data on the transcriptional regulation of hyaluronan synthase (Has1–3) expression and on the post-transcriptional control of hyaluronan synthase activity, which, in close association with the supply of the UDP-sugar substrates of hyaluronan synthase, adjust the rate of hyaluronan synthesis.
The synthesis of hyaluronan (HA) takes place in selected bacteria and Archaea, including Gram-positive streptococci (Streptococcus pyogenes, S. equisimilis, S. uberis and S. zooepidemicus) and Gram-negative Pasteurella multocida, and increases their virulence. HA appeared only about 500 Ma in chordates . In mammals, HA can be synthesized by three isoenzymes (hyaluronan synthases, HAS1–3), which transfer both the glucuronic acid (GlcUA) and N-acetylglucosamine (GlcNAc) moieties that alternate in the linear glycosaminoglycan chain . These isoenzymes show a structural identity of about 55–70%  and HAS2 has a critical role for animal survival . HA often reaches a molecular mass of 107 Da and an extended length of more than 20 μm. HAS is an integral membrane protein that is normally active in the plasma membrane. Interestingly, HASs show a peculiar structure, possessing two distinct binding domains for UDP-sugars, whereas most glycosyl transferases have only one. Moreover, it is likely that HA remains attached to HAS when it is synthesized and simultaneously extruded into the extracellular space . Therefore, a large proportion of cell surface HA can be bound to HAS, rather than to HA receptors such as CD44 . Whether cell surface HA is bound to HAS or to CD44 and other receptors may strongly modify its biological influence. Likewise, the molecular mass of the bound HA and the specific plasma membrane locations in which HAS is inserted are probably crucial. Therefore, a knowledge of the regulatory factors governing HAS expression, trafficking, enzymatic activation, location and quality of the HA product are key to an understanding of the biological effects of HA.
The expression of HAS enzymes is the first and perhaps most important determinant of the HA synthesis rate in a given cell type under specific circumstances. There are several examples of cell types in which HA synthesis changes in parallel with the level of Has mRNA. However, studies comparing the quantities of HAS proteins with HA synthesis are very few, apparently because of technical difficulties in the extraction and western blotting of HAS.
The next level influencing HA synthesis is the post-translational modification of HAS proteins and the traffic of HAS to the plasma membrane. New data are emerging, particularly on phosphorylation, ubiquitination and O-GlcNAcylation of HAS, post-translational modifications that can modify its enzymatic activity. The traffic of HAS to the plasma membrane may also be coupled to its post-translational modifications, but little is known about this.
HA synthesis consumes large quantities of UDP-GlcUA and UDP-GlcNAc, the substrates for HAS enzymes. The cellular concentration of either of these UDP-sugars can become limiting in HA synthesis . The content of the UDP-sugars varies between cell types and responds to metabolic modifiers in ways that are not well known, suggesting that the contributions of the UDP-GlcUA and UDP-GlcNAc contents to overall HA synthesis need to be assessed.
Thus, a number of new control points have emerged that potentially regulate HA synthesis, described in more detail below. In addition, through its interaction with the CD44 receptor, HA feeds back towards its own synthesis, as discussed in another minireview of this series (Misra et al. ).
Transcription of HASs
Embryonic development and Has expression
The distribution of HAS during development may yield an answer to the question of why three different isoenzymes exist for the synthesis of the simple HA polymer. Using an invivo model in Xenopus laevis, the first HAS in a vertebrate was discovered (XHAS1) . It was demonstrated  that three synthases are expressed in Xenopus at different stages of development. Whereas XHas1 and XHas2 are widely expressed in the embryo, the transcription of XHas3 is restricted to certain areas of the embryo, including the inner ear and the cement gland [9,10]. Similarly, the three synthases are expressed in different temporal patterns during mouse development . In particular, Has2 has been identified as a major source of HA during initial organogenesis. Mice with a homozygous deletion of the Has2 gene manifest severe cardiac and vascular abnormalities leading to death at mid-gestation (E9.5–10) . It has been demonstrated that HAS1 and HAS2 are able to produce large-sized HA, whereas HAS3 produces HA of a lower molecular mass. When considering the important role of size in HA function, it is clear that the differential expression of these enzymes can be critical for cell behavior. Usually, the most common HAS in mammalian tissue is HAS2, whereas HAS3 is highly expressed in specific conditions, as in tumor cells or during inflammation.
HA is abundant in fetal human tissues [12,13], but is depleted during development, and is replaced by collagen and proteoglycans, possibly as a matrix adaptation to more severe mechanical requirements . Therefore, comparative gene expression analysis of HASs suggests that the different roles for HA during development are fulfilled by the spatio-temporally regulated transcription of the three different synthases.
Multiple growth factors and cytokines influence Has expression
The expression of Has genes undergoes rapid and dramatic changes during embryonic development, corresponding to the migration of cells to their final sites in organs [4,14], the formation of specific matrices such as that in cartilages  and general cell proliferation in expanding tissues . In adult tissues, HA synthesis is stimulated by injury, inflammation and neoplastic tumors. A number of cytokines and growth factors, such as platelet-derived growth factor (PDGF) [16,17], fibroblast growth factor-2 , keratinocyte growth factor , epidermal growth factor (EGF) , transforming growth factor-β , interleukin-1β , tumor necrosis factor-α  and interferon-γ , are released from local cells, and also from platelets and leukocytes arriving in the area, and increase Has expression. In addition, Has expression and HA synthesis are sensitive to local mediators, such as prostaglandins , and hormone-type effectors, such as corticosteroids , the latter downregulating Has2, and retinoids, which induce its expression [27,28]. Some of the effectors are shown in Fig. 1. There are large differences between different cell types concerning the stimulants to which they respond. In addition, some of the effectors modulate the expression of all Has genes, whereas others just modulate one or two. For example, Has2 in epidermal keratinocytes is particularly sensitive to EGF receptor ligands, whereas MCF7 breast cancer cells respond weakly, if at all, to EGF. Transforming growth factor-β downregulates Has2 and Has3 in keratinocytes, but enhances the expression of Has1 in fibroblasts  and synoviocytes . These findings suggest that the three Has genes have promoters reacting to common transcriptional signals in addition to their specific responses.
Response elements and transcription factors in the Has2 promoter
Of the three Has genes, deletion of only Has2 has a severe (lethal) phenotype, and particularly large changes in the rate of synthesis of HA are noted by external stimulants that affect Has2 expression [20,25,31,32]. Therefore, the promoter area of Has2 has been most actively studied for response elements that bind regulatory transcription factors (Fig. 1).
In epidermal keratinocytes, the EGF-family growth factors, especially HB-EGF , acting in autocrine and paracrine fashions, are major inducers of Has2 expression. Inhibitors of EGFR (ErbB1) and phosphatidylinositol 3-kinase prevent this stimulation, which is mediated by the transcription factor STAT3 (Fig. 1).
Stimulants of HA synthesis mediated by G-protein-associated receptors act via protein kinase A and the CREB1 response elements on the Has2 promoter , whereas all-trans-retinoic acid, an important developmental signal that supports normal epithelia, acts through its nuclear receptor with a functional response element 1300 bp upstream of the Has2 transcription start site in keratinocytes (Fig. 1).
Among the transcription factors, nuclear factor kappa B has a central role in the induction and resolution of inflammation. In this latter process, HA has multiple effects depending on its molecular mass, and it has been shown that HA can participate with other adhesive molecules (i.e. selectins and integrins) in the recruitment of circulating immune cells to the site of inflammation. For example, cultured endothelial cells synthesize very little HA, but their stimulation with tumor necrosis factor-α and interleukin-1β induces Has2 transcription via nuclear factor kappa B , and the newly synthesized HA mediates monocyte adhesion through CD44 . In vivo, endothelial cells display an HA-containing glycocalyx with an important influence on leukocyte binding and vascular disease .
It has been found recently that intracellular glucose metabolism, through UDP-GlcNAc content, controls Has2 transcription, probably through the transcription factors YY1 and SP1 as described below (T. Jokela et al., University of Eastern Finland, Kuopio, unpublished results) (Figs 1 and 2).
UDP-sugar substrates as limiting factors in HA synthesis
UDP-GlcUA and UDP-GlcNAc contents control HA synthesis
HA is produced from glucose through the metabolic steps shown in Fig. 2. Relatively specific inhibitors of HA synthesis were not known until the discovery of 4-methylumbelliferone (4-MU) , a derivative of coumarin that is common in flowering plants. 4-MU inhibits HA synthesis by depleting the HAS substrate UDP-GlcUA, which is consumed by 4-MU glucuronidation [38–40], a reaction common in the detoxification of foreign substances (Fig. 2). Exploration of the 4-MU mechanism revealed the crucial role of the supply of UDP-GlcUA for HA synthesis. Interestingly, other glycosaminoglycans, such as chondroitin and heparan sulfates, are less sensitive to UDP-GlcUA deficiency because they are synthesized in the Golgi, a privileged compartment as a result of the high-affinity transporters that pump UDP-sugars from the cytosol into the Golgi (see the accompanying minireview by Wang et al. ). Increasing the content of UDP-GlcUA by the overexpression of UDP-Glc dehydrogenase enhances HA production , confirming the importance of the cellular level of UDP-GlcUA in HA production.
The cellular content of UDP-GlcNAc is usually three to seven times higher than that of UDP-GlcUA, and was previously not considered to be a limiting factor in HA synthesis. However, the Km value for UDP-GlcNAc is also up to ∼ 10 times higher than that for UDP-GlcUA in all HAS enzymes . Therefore, cells readily respond to mannose-induced UDP-GlcNAc depletion by an inhibition of HA synthesis . The reduction of UDP-GlcNAc by mannose is probably a result of reduced formation or enhanced catabolism of glucosamine-6-phosphate (Fig. 2). Again, the syntheses of other glycosaminoglycans, operating in the presumably higher UDP-GlcNAc concentration in the Golgi, are not affected . Raising the content of UDP-GlcNAc by adding glucosamine to the growth medium increases HA synthesis, indicating that the cellular content of UDP-GlcNAc is as important as that of UDP-GlcUA in controlling the level of HA synthesis .
It is interesting to note that the factors regulating HA size are still poorly understood in eukaryotic cells , whereas, in P. multocida, the UDP-sugar availability can influence the size of HA .
UDP-sugars control HAS access to the plasma membrane
Transfection of cells with green fluorescent protein (GFP)-labeled Has constructs has enabled the examination of HAS localization within live cells, and its movements between the cellular compartments. When HA synthesis is inhibited by the depletion of UDP-GlcNAc with 4-MU, GFP-Has3 gradually disappears from its regular location in plasma membrane protrusions, and the protrusions also wither . In the same way, when mannose-induced reduction of the UDP-GlcNAc pool inhibits HA synthesis, it also prevents HAS3 localization in the plasma membrane (T. Jokela et al., unpublished results), and the same response occurs with glucose deprivation. Cells cultured in the absence of glucose produce very little HA (Fig. 3A) and show GFP-HAS3 mainly in the Golgi area, whereas the same cells, after the addition of glucose to the medium, present HAS in the plasma membrane, typically in microvillous projections (Fig. 3B). Thus, the synthesis of HA is coupled to HAS localization in the plasma membrane, suggesting that either the availability of UDP-sugars stabilizes HAS in the plasma membrane, or that sufficient amounts of the UDP-sugars are required to trigger the transport of HAS into the plasma membrane. These findings may offer clues for future research on the enigma of why HA synthesis normally takes place only in the plasma membrane, whereas HAS inserted in the endoplasmic reticulum and Golgi membranes normally remains inactive.
Nevertheless, it has been reported that, in in vitro systems, HASs are able to produce HA when microsomal preparations of different cell lines are incubated in the presence of substrates . These findings shed light on the possibility of HA synthesis inside the cells in specific situations, such as endoplasmic reticulum stress, inflammation and hyperglycemia  (see accompanying minireview by Wang et al. ).
An unexpected and interesting feedback regulation of HA synthesis has been discovered recently that is associated with the transcription of Has2. When the UDP-GlcNAc content was increased in keratinocytes by feeding with glucosamine, the expression of Has2 was reduced by ∼ 50%, whereas, when the cellular UDP-GlcNAc content was reduced by feeding with mannose, Has2 mRNA was increased. Checking the glucosamine-induced changes in transcription factors bound to the Has2 promoter revealed an accumulation of YY1. Treatment with mannose reduced the promoter binding of SP1, another transcription factor. Following siRNA-mediated reduction of YY1 and SP1 levels, Has2 mRNA was increased, confirming the suppressive role of YY1 and SP1 on Has2 expression.
The changes in Has2 promoter binding of YY1 and SP1 were associated with increased O-GlcNAc modifications to these proteins. This is a signaling system in which GlcNAc from UDP-GlcNAc is transferred to certain Ser and Thr residues of cytosolic and nuclear proteins by a specific enzyme (O-GlcNAc transferase). This dynamic addition and removal of O-GlcNAc has been shown to control protein functions, sometimes by supporting the effects of phosphorylation, or by counteracting the influence of phosphorylation, or by competing with phosphorylation for certain Ser and Thr residues. The supply of UDP-GlcNAc to O-GlcNAc transferase influences the extent of O-GlcNAc modification. Accordingly, the levels of O-GlcNAcylation of YY1 and SP1 were increased and decreased by glucosamine and mannose, respectively, which can explain the alterations in their promoter binding and Has2 expression (Fig. 1). It can be assumed that the Has2 response is a way to dampen excessive impacts of UDP-GlcNAc fluctuation on HA synthesis.
Post-translational processing of HAS
Post-translational modulation of HAS activity
The quantification of rapid changes in HA synthesis, such as those expected as a result of the post-translational regulation of HAS activity, is technically difficult in intact cells. However, plasma membrane and cytoplasmic fractions (the latter including the endoplasmic reticulum and Golgi membranes) derived from cell homogenates have been successfully utilized for in vitro assays of HAS activity . In this system, a 5–10-min treatment of cells with phorbol 12-myristate-13-acetate, interleukin-1β and PDGF-BB induced a several-fold increase in HAS activity in both the plasma membrane and cytoplasmic fractions . Multiple receptor types, such as those of cytokines, growth factors and protein kinase C activators, thus have immediate influences on the activity of HAS, in addition to their stimulation of Has transcription, as discussed above.
Phosphorylation influences HAS activity
Receptor-mediated signaling mostly works through kinases, and HAS protein itself is a possible target . The pretreatment of membrane preparations with alkaline phosphatase reduced the HAS activity induced by phorbol 12-myristate-13-acetate and interleukin-1β, but not that by PDGF-BB , demonstrating the diversity of the activation pathways. All three HAS isoenzymes can be phosphorylated following activation of ErbB2/extracellular signal-regulated kinase, and associates to increased synthetic activity . Interestingly, when membranes prepared from COS-7 cells transfected with human HAS2-expressing vector were treated with alkaline phosphatase, the HAS activity increased several fold when compared with control membranes (D. Vigetti et al., unpublished results), suggesting that HASs could be finely regulated at a post-translational level. At present, it is unclear whether there is an association between the phosphorylation of HAS and its enzymatic activity, and it could be argued that the same post-translational modification (i.e. phosphorylation) in different residues of the protein could increase or decrease enzymatic activity. To support this possibility, cells treated with inducers of energetic stress (i.e. AICAR or 2-deoxyglucose) phosphorylated HAS2 with a resultant reduced HA synthetic activity (D. Vigetti et al., unpublished results). Interestingly, tunicamycin, which causes endoplasmic reticulum stress by inhibiting N-glycosylation, produced a shift of HAS activity from the plasma membrane to the cytoplasmic fraction, suggesting interference in HAS trafficking .
O-GlcNAc as a modifier of HAS function
It is quite interesting that the cellular concentration of UDP-GlcNAc is a strong determinant of both the O-GlcNAc level of proteins and the presence of HAS in the plasma membrane. Indeed, UDP-GlcNAc-dependent trafficking of HAS to the plasma membrane, and its enzymatic activation as discussed above, could be a result of the O-GlcNAc modification of HAS or other proteins involved. Interestingly, a lectin that binds to O-GlcNAc can be used to capture a membrane fraction with HAS activity (S. Deleonibus et al., University of Insubria, Varese, unpublished results). If O-GlcNAcylation of HAS can be shown to control HAS trafficking and activation, the cellular content of UDP-GlcNAc, in addition to its effect on transcription and its use as a substrate, is obviously a central coordinator of HA synthesis.
HAS2 and HAS3 can dimerize
It has long been assumed that HAS proteins work as monomers without functional interaction with other proteins . However, it has been demonstrated recently that HAS2 tagged with either 6myc or Flag is co-immunoprecipitated with antibodies against either tag , suggesting dimerization (Fig. 4). Furthermore, an antibody against the 6myc-HAS2 tag also co-immunoprecipitates Flag-HAS3, suggesting heterodimerization of HAS2 and HAS3. Dimerization (or multimerization) could explain the problem of how a single 65-kDa polypeptide can form a membrane pore for the extrusion of the bulky HA chain, in addition to its two different UDP-transferase activities and an ability to hold the growing chain . Indeed, most of the glycosyl transferases, as well as pore-forming proteins in plasma membranes, are multimers.
Monoubiquitination and HAS2 activity
Polyubiquitin chains built on cytoplasmic proteins form a label that leads the proteins into proteasomal degradation. In contrast, monoubiquitin is a signal that has been shown to change the function of proteins or their trafficking to certain subcellular compartments. A portion of 6myc-HAS2 and Flag-HAS2 was monoubiquitinated on Lys190 of HAS2 . The enzymatic activity of Has2 is lost with a site-directed Lys190Arg mutation (Fig. 4). The 190Arg construct was able to dimerize with the wild-type form of HAS2, but inhibited HA synthesis by the latter in a dominant negative fashion. These data strongly suggest that monoubiquitination is required for HAS2 activity and that HAS2 is only active as a dimer. A number of interesting vistas are now open for future research concerning the activation process of different HAS isoforms.
Cell stress changes the structure of HA on the cell surface
It has been shown that cells undergoing various inflammatory stresses can form ‘cable’ structures extending far away from the cell. Functionally, the formation of cables is associated with the enhanced binding of leukocytes, which changes their behavior and presumably the course of inflammation. This altered surface structure of HA may depend on the specific trafficking and localization of HAS, and has widespread influences on cell behavior, as described in an accompanying minireview of this series (Wang et al. ).
It is still unclear whether HA cable formation is linked to a particular HAS isoenzyme expression. However, it has been shown that HAS3 overexpression leads to cable formation, whereas HAS2 overexpression reduces such filamentous structures .
The authors thank Tiina Jokela, Moira Clerici, Sara Deleonibus, Sanna Oikari and Juha Hyttinen for sharing some of their unpublished data, and Dr Vincent Hascall for critical reading of the manuscript. The Academy of Finland (MIT), Sigrid Juselius Foundation (RHT), Kuopio University Hospital (MIT), Mizutani Foundation (MIT), Fondazione Comunitaria del Varesotto-ONLUS, Centro Insubre di Biotecnologie per la Salute Umana and FEBS Fellowship supported the authors’ original research that forms part of this minireview. We apologize that only a few examples of the original work forming the basis of this minireview could be cited because of space limitations.