Does DG42 synthesize hyaluronan or chitin?: A controversy about oligosaccharides in vertebrate development.

Science generally progresses in slow but deliberate increments, which are punctuated by major advances in concept or fact. However, the latter are rare and not infrequently go unrecognized when they first occur. Another event that can add spice to a field, and attract the attention of scientists from outside the discipline, is a genuine controversy. One such controversy is presented by two reports in this issue of the Proceedings (1, 2). To be asked to referee such a controversy is an interesting but difficult task, since both groups have significant data to back their claims. The story begins in 1983, when Igor Dawid and colleagues (1) reported the isolation of several genes that are Differentially expressed at Gastrulation (DG) in embryos of the frog, Xenopus laevis (3). One of these, the endoderm-specific DG42, is expressed in a short window during embryogenesis, being first detected after the midblastula stage, peaking at late gastrula, and decaying by the end of neuralation (4–6). Appropriate probes were used to show that the messenger RNA and predicted protein product move in a wave or gradient through the embryo, with the last remnants seen in the ventral regions of the gut at the tailbud stage. For a while thereafter, DG42 remained an interesting gene in search of a function. As often happens, the first clues came from unexpected sequence homology information. When it was first cloned, DG42 showed no obvious homologies to any previously known protein or gene. Subsequently, some similarities were found with fungal chitin synthases (7) and with the rhizobium NodC gene that is known to synthesize chitin oligomers (8–10). What is chitin? It is a repeating b1-4-linked homopolymer of the monosaccharide GlcNAc (see Fig. 1) that is one of the most widespread and abundant molecules in the biosphere, providing, for example, a major component of the cell walls of fungi and the shells of crustaceans and arthropods (11, 12). This important structural role for the extended polysaccharide may seem of little relevance to vertebrate development. However, shorter oligomers of the same repeating sequence are known to be soluble ‘‘oligosaccharins,’’ mediating short range hormonal responses between Rhizobium bacteria and leguminous plants during the process of nitrogen-fixing root-nodule formation (13–15). Indeed, complex structural variations on the theme of the basic chitin backbone are well known to mediate a variety of specific interactions between bacteria and plants (for some examples, see refs. 16–20). Intrigued by these homologies, Semino and Robbins (21) then showed that when generated in an in vitro transcriptiony translation system, the DG42 gene product was capable of synthesizing both short chitin oligomers and some larger products. The required sugar nucleotide donor was UDPGlcNAc; the products had the correct chromatographic properties, and they were degraded appropriately by a bacterial chitinase. Thus, DG42 was proposed to be the first recognized vertebrate chitooligosaccharide synthase (21). However, another interesting homology had also appeared between DG42 and the hasA gene of Streptococci (22, 23). The latter is responsible for the synthesis of another repeating polymer of sugars called hyaluronan. What is hyaluronan? It is a polymer consisting of alternating units of b1-4-linked GlcNAc and b1-3-linked glucuronic acid (GlcA, see Fig. 1). At first glance, these may seem to be very similar structures. Indeed, the linkages are very similar, and the donor nucleotides for both units are based on UDP (UDP-GlcNAc and UDP-GlcA). However, the similarity ends there (24, 25). Partly by virtue of its carboxylate groups, hyaluronan has physical properties that are almost diametrically opposite to those of chitin, being capable of retaining large amounts of water to form a gel. Furthermore, unlike chitin, hyaluronan expression is primarily reported in vertebrates, and in a few pathogenic bacteria such as group A and C Streptococci (24, 25). In view of these homologies, Semino and Robbins had also checked to see if the DG42 protein had hyaluronan synthase activity in vitro, but did not find any (21). This seemed to settle the issue that the DG42 gene product was primarily a chitin synthase. Enter the new study of Meyer and Kreil (1), which shows that rabbit kidney and human osteosarcoma cells induced to express the DG42 gene with a vaccinia virus system synthesize increased amounts of hyaluronan. Lysates and membranes from such transfected cells showed markedly increased hyaluronan synthase activity, which required the addition of both UDP-GlcNAc and UDP-GlcA donors. The product of the reaction was sensitive to hyaluronidases, but not to chitinases, and appropriate controls showed that the overexpression of hyaluronan synthesis was clearly related to DG42 expression (1). These authors conclude that their results are at variance with the earlier report of Semino and Robbins (21). Meanwhile, the latter group have an update to their story that is also published in this issue (2). They now show that DG42 homologues and their protein products are expressed in early embryos of zebrafish and mouse during the gastrula–early neuralation stages, and that chitin-oligosaccharide synthesis can be detected in extracts from these sources as well. Furthermore, this activity was immunoprecipitated by a DG42specific antibody (4) provided by Dawid. Also, overexpression of DG42 in a different cell type (mouse 3T3 cells) gives the synthesis of chitooligosaccharides, but no increase in background levels of hyaluronan synthesis. Finally, these authors show a physical separation of chitin synthase activity from most (but not all) of the hyaluronan synthase activity in embryo extracts (2). How can one reconcile the findings of the two studies and determine the true role of DG42? Semino et al. do make one preliminary attempt to do so (2). They state that commercial preparations of hyaluronan have chitin oligomers at their reducing end core region (further details are evidently to be published elsewhere). They suggest that DG42 might function to produce chitin oligomers that act as templates for hyaluronan synthesis (see Fig. 1). In this regard, it is interesting that Meyer and Kreil note a requirement for high concentrations of UDP-GlcNAc in their reactions (1). To consider this possibility further, let us review what is known about hyaluronan synthesis in vertebrate systems. The biosynthesis of this polysaccharide is peculiar, in that it follows a route different from that taken by most other molecules