`Soluble' and `insoluble' mucins - Identification of distinct populations

Mucins: Their Structure and Biology

32 Podolsky, D. K. (1985) J. Biol. Chem. 260, 8262-8271 33 Podolsky, D. K. (1985) J. Biol. Chem. 260, 15510-15515 34 King, M.-J., Chan, A. and Roe, R. et al. (1994) Glycobiology 4, 267-279 35 Williams, A. C., Harper, S. A. and Paraskeva, C. (1990) Cancer Res. 50, 4724-4730 36 Allen, A. and Hoskins, L. C. (1988) in Disease of the Colon and Rectum (Kirsner, J. B. and Shorter, R. G., eds.), pp. 65-94, Williams & Williams, Baltimore 37 Hoskins, L. C., Boulding, E. T., Gerken, T . A., Harouny, V. R. and Kriaris, M. (1992) Microb. Ecol. Health. Dis. 5. 193-207

38 Corfield, A. P., Wagner, S. A., Clamp, J. R., Kriaris, M. S. and Hoskins, L. C. (1992) Infect. Immun. 60, 3971-3978 39 Corfield, A. P., Williams, A. J. K., Clamp, J. R., Wagner, S. A. and Mountford, R. A. (1988) Clin. Sci. 74, 71-78 40 Roberton, A. M., McKenzie, C., Scharfe, N. and Stubbs, L. (1993) Biochem. J. 293, 683-689 41 Tsai, H. H., Sunderland, D., Gibson, G. R., Hart, C. A. and Rhodes, J. M. (1992) Clin. Sci. 82, 447-454 42 Corfield, A. P., Wagner, S. A. and Clamp, J. R. (1987) Biochem. SOC.Trans. 15, 1089 Received 19 June 1995

‘Soluble’ and ‘insoluble’ mucins - Identification of distinct populations 1. Carlstedt*$, A. Herrrnann*, H . Hovenberg*, G. Lindellt, H. Nordrnan*, C. Wickstrom* and J. R. Davies* *Department of Cell and Molecular Biology, Section for Molecular Pathogenesis, Lund University, S-22 I 00 Lund, Sweden, and tDepartment of Surgery, University Hospital, 5-22 I 85 Lund, Sweden

T h e mucosal surfaces represent an impressive barrier between the ‘inside’ of the body and the ‘outside’ world. This ‘front line’ is under constant threat from micro-organisms, degradative enzymes and other noxious agents and is, therefore, as a first line of defence, covered by a gellike secretion - mucus. T h e polymer matrix of this biofilm is provided by high-M, glycoproteins referred to as mucus glycoproteins or mucins.

The mucin superfamily Mucins constitute a family of glycoproteins traditionally regarded as secretory products from epithelial surfaces. T h e apoprotein is substituted with a large number of oligosaccharides attached via an 0-glycosidic linkage between serine and/or threonine and GalNAc. In a typical mucin, the 0-linked glycans - often referred to as mucintype oligosaccharides - are enriched within serine/threonine-rich domains. These often contain proline, which is believed to give the protein core an extended conformation, in particular after substitution with the link GalNAc [ l ] , and it is therefore likely that one major function of the serinelthreonine-rich regions is to provide a matrix for the presentation of carbohydrate structures. Similar sequences have been identified in, for example, some membrane-bound glycoproteins, and a ‘mucin-like stretch’ has become an expression used to denote such domains with multiple sites for substitution with $To whom correspondence should be addressed.

0-linked oligosaccharides in any protein. For example, glycoproteins carrying receptor structures for selectins have been referred to as endothelial mucins [Z]. From being a relatively ill-defined group of glycoproteins secreted onto epithelial surfaces, mucins are now recognized as a family of molecules sharing distinctive structural motifs. Epithelial mucins can be divided into two main groups - those that are membrane bound and those that are secreted. T h e very large, biofilm-forming species produced by specialized secretory cells constitute a distinct subgroup of the latter. After synthesis, these ‘large secreted mucins’ are usually stored in granules before secretion, which occurs mainly via the regulated pathway. Molecular biology studies have shown the existence of several mucin genes (MUCIMUG‘@, although only MUC1 and MUC7 have been both sequenced completely and unequivocally identified biochemically. MUC 1, whose biochemical equivalent is known as episialin or PEM, is a membrane-bound glycoprotein [3], whereas MUC7 is a low-M, secreted mucin from saliva previously known as MG2 [4]. T h e other genetically defined mucins are probably large and belong to the biofilm-forming family. Of these, the complete sequence is known only for the human MUC2 mucin [ S ] . T h e published apoprotein sequences have disclosed many interesting structural features of mucins. T h e heavily glycosylated serinehhreo-

I995

84 5

Biochemical Society Transactions

846

nine/proline-rich regions are composed of tandemly repeated structures and are, as was predicted from biochemical studies, flanked by cysteine-rich domains (see, for example [5-91). T h e number of tandem repeats may vary within a mucin apoprotein and the presence of these VNTR (variable number tandem repeats) regions explains, in part, the size polydispersity that is a characteristic feature of mucins. T h e domains flanking the tandem repeat regions of MUC2 show significant homology with the D-domains in prepro von Willebrand factor [ 5 ] , but nothing is yet known about whether the MUC2 mucin monomer is processed and oligomerized in the same way as described for this protein (see, for example, [lo]). However, if that is the case, it can be predicted that MUC2 subunits isolated from human tissue are truncated compared with the published sequence.

Figure I Extraction of mucins from (a) rat small intestine, (b) the gel phase of bovine tracheal secretion and (c) saliva from human submandibular/sublingualglands Samples were subjected to three or four rounds of extraction in 6 M guanidinium chloride supplemented with proteinase inhibitors. The final residue (R) was solubilized by reduction, For rat small intestine, the extraction was evaluated by isolating high-M, glycopeptides from the various supernatants (S l -53) and the residue (R) using gel chromatography. For the other two samples, mucins and mucin subunits were isolated with isopycnic density-gradient centrifugation in CsC1/4 M guanidinium chloride.

75

50

25

‘Soluble’ and ‘insoluble’ biofilm-forming m u c h

Many different procedures have been used to extract mucins from mucus gels and tissue samples, including proteolysis, reduction of disulphide bonds, high-speed homogenization and slow stirring in denaturing solvents. We have favoured the use of careful agitation in 6 M guanidinium chloride supplemented with proteinase inhibitors to solubilize mucins in order to ensure, as far as possible, the isolation of undegraded molecules [ 111. This procedure has been used successfully for cervical, respiratory and gastric mucins but, when this approach was used to prepare mucins from rat small intestine, we discovered that only a small proportion of the macromolecules were ‘soluble’ [ 121. T h e major part of the glycoproteins could not be brought into solution even after several rounds of extraction. However, the ‘insoluble’ mucins in the residual pellet could be solubilized by cleavage of disulphide bonds. T h e approach we are currently using to evaluate the ‘solubility’ of mucins is illustrated in Figure 1. Mucins were prepared from three different sources - rat small intestine, the secreted mucus gel from bovine trachea and saliva from human submandibular/sublingual glands. T h e samples were subjected to three or four rounds of extraction with 6 M guanidinium chloride followed by high-speed centrifugation. Mucins in the final residue were solubilized by reduction and the amount of mucinslmucin subunits in the three/four individual supernatants (S1-S4) and the residue (R) was evaluated using density-

Volume 23

75

c0

b

CI

.-ahl

5c

s

.c

0

2

25

c

X

W

7c

C

5c

25

s1

s2

s3

s4

R

gradient centrifugation in CsCVguanidinium chloride for whole mucins [13] (bovine tracheal and human salivary mucins) or by gel chromatography for the cognate high-Mr mucin glycopolypeptides (rat small intestine). In the rat small intestine, approximately 20% of the mucins were found in the first supernatant (Sl) and very little material was solubilized by further extractions (Figure l a ) . Most of the glycoproteins are thus present in the insoluble residue and treatment with denaturing solvents for several weeks does


not bring this material into solution. In the secreted gel from bovine trachea, most of the mucins are solubilized during the first (Sl) extraction step (Figure lb) and the amount of extractable mucins decreases with every round of extraction. However, a significant proportion is left in the final residue. In some preparations of salivary samples, the mucins are very difficult to disentangle from the gel (Figure lc). In this particular case, there seems to be a slow ‘leakage’ of mucins from the gel and the insoluble residue is not as well defined as for the rat small intestinal and bovine tracheal mucins. In our hands, salivary samples are usually solubilized to approximately 65-80%, but the example given here clearly shows that it is important to evaluate the extraction yield in the individual experiment. Trpically 80-90% of human respiratory, cervical and gastric mucins are extracted by slow stirring in guanidinium chloride. Finally, it should be pointed out that ‘soluble’ and ‘insoluble’ are used here in an ‘operational’ sense describing the behaviour of the glycoproteins during highspeed centrifugation after extraction in a denaturing solvent. One obvious function of all biofilm-forming mucins is to appear as a mucus gel and thus be ‘insoluble’ under physiological conditions! T h e structural basis for the ‘insolubility’ is not clear. One possibility is that this fraction represents a population of very large glycoproteins that are not disentangled even during prolonged extraction. However, the extraction data for the intestinal and tracheal mucins suggest that the ‘insoluble’ species represent a distinct subpopulation that remains after the removal of all ‘soluble’ species. Therefore, we favour the idea that these subunits are more densely ‘cross-linked’ than the end-to-end assembly (see below) that is the structural design for the ‘soluble’ mucins. Electron microscopy of shear-induced fragments of the ‘insoluble’ complex from rat small intestine suggests the presence of branched structures (A. Herrmann, M. Morgelin and I. Carlstedt, unpublished work). T h e ‘insolubility’ may therefore reflect a fundamental difference in the way that the subunits are assembled into these structures.

The ‘soluble’ biofilm-forming mucins are long, flexible chains composed of subunits linked end to end T h e dominating populations of ‘soluble’ mucins are very large molecules (Mr between 5 and

25 x lo6), which can be fragmented into subunits (monomers) with M , 2-3 x lo6 by reduction of disulphide bonds. T h e macromolecules behave as random coils [14], suggesting that subunits are linked end to end, and this structural architecture for the major mucin populations from the airways, stomach and cervix proposed previously [ 11,151 has been verified using molecular electron microscopy [ 16-19]. T h e size distributions of these mucins indicated that the macromolecules occur as dimershrimers up to structures corresponding to approximately decamers (for example, see, [20] ). Thus, the size polydispersity of oligomeric whole mucins is to a larger extent dependent on a Variable Number of Monomers (‘VNM’) than on the VNTR mentioned above. Mucin apoproteins are substituted with oligosaccharides constituting approximately 80% of the total weight of the molecule. T h e glycans are concentrated into domains in which they are so tightly packed that they protect the protein core from proteolysis. These oligosaccharide-rich domains can be isolated as high-M, (200800 x lo3) glycopolypeptides (see [21] for references) and correspond to the tandem repeat regions described above. Each subunit from the predominant mucin populations from the airways, stomach and cervix contains several such oligosaccharide domains flanked by stretches of the protein core that are susceptible to proteolytic digestion. So far, only the partial sequences available for the MUC5AC mucin predict an apoprotein structure compatible with this model [221*

Gastric mucins can be separated into distinct populations Of the putative large secreted mucins, MUC4 and MUC5 are strongly expressed in the airways, MUC3, MUC4, MUC5 and MUC6 in the stomach, MUC2 and MUC3 in the intestine and MUC4 and MUC5 in the cervix [23]. It is thus likely that the mucin populations isolated from these tissues are in fact mixtures of several distinct gene products, but it has not been possible, so far, to separate these species from each other with biochemical techniques, assign a ‘MUC number’ to them and investigate whether, for example, the degree of oligomerization and glycosylation differs. However, mucin subunits from respiratory mucins were recently separated into two distinct populations using agarose electrophoresis, providing one example of a successful approach to this important problem [24].

I995

847


848

Histochemical studies suggest that glycoproteins with distinctly different glycosylation are produced by the surface mucosa and the glands in the stomach, and structural differences have also been observed between the regions (see, for example, [25,26]). In the antrum region of human stomach, the Solantim tuberosum (potato)

Figure 2 Mucins from human gastric mucosa (a) Formalin-fixed tissue sections from the antrum region of human stomach were probed with a biotinylated Solonurn tuberosurn

(potato) lectin Horseradish peroxidase-conjugatedstreptavidin was then used with diaminobenzidine as a substrate to detect bound lectins The bar represents 200 prn (b) Mucins purified by densitygradient centrifugation in CsC1/4 M guanidinium chloride from the subrnucosal tissue of human antrum were subjected to a second density gradient step in CsCI/O 5 M guanidinium chloride (initial density I 45 g h l ) Fractions were analysed for density, carbohydrate (Boehringer glycan detection kit), reactivity with the 801onum tuberosum (potato) lectin and a polyclonal antibody raised in rabbits against a synthetic peptide from the MUC5AC sequence Lectin and antibody reactivities were analysed using ELISA after coating aliquots of the fractions onto rnicrotitre plates 0 Carbohydrate, o potato lectin reactivity A MUCSAC reactivity

1.0

3.8 c

ln

s 3.6 e

k 3.4

0.2

0

Volume 23

g

z 0

lectin shows a strong reactivity with the glands and has apparently no affinity for the glycoproteins in the surface mucosa (Figure 2a). Mucins were extracted from the same tissue sample (a single individual was studied here since we do not know how general this observation is) with the procedure outlined above and purified from proteins with density-gradient centrifugation in CsC1/4 M guanidinium chloride. When the mucin band was subjected to a second density-gradient centrifugation step in CsCI/ 0.5 M guanidinium chloride (originally devised to separate mucins from DNA [13]), the carbohydrate analysis suggested the presence of two populations of glycoproteins (Figure Analysis with the potato lectin was positive only over the high-density population, suggesting that this is derived from the gland cells and not from the surface epithelium. In contrast, the low-density population reacted avidly with a putative MUC5AC antibody raised to a synthetic peptide with the sequence RNQDQQGPFKMC. This sequence appears in two ‘unique’ stretches flanking a tandem repeat region [22], as well as in the C-terminal part of the molecule [27], and is thus likely to be present in non-glycosylated parts of the molecule. Similar probes for the other relevant mucin apoproteins are not available to us at the moment, but the example shows clearly how undegraded mucins may be separated into distinct populations, localized histochemically and the apoprotein given a ‘genetic’ identity. It should be pointed out, however, that the MUCSAC-positive population could well be a mixture of mucins also containing other apoproteins, and further investigations are needed before it can be stated that this population is only MUCSAC. Nothing is currently known about how the different populations of gastric mucins co-operate in forming the biofilm protecting the gastric mucosa. However, secretions with different histochemical staining properties form laminated structures in the mucus gel [28].

a).

Rat small intestinal mucins occur as an ‘insoluble’ glycoprotein complex Since mucins from rat small intestine could not be solubilized by extraction in guanidinium chloride, a technique was developed to isolate these macromolecules as an insoluble glycoprotein complex [ 121. Two oligosaccharide-rich domains were identified and isolated as glycopeptides A and B from subunits of this complex. Antibodies against glycopeptide A were used to


isolate a clone (VR-1A) that showed homology with human MUC2 [29]. cDNA probes for this clone and probes for rat intestinal mucin-like peptide (rat-MLP; [S]) identified the same band after hybridization with both mRNA and genomic DNA after cleavage with restriction enzymes. Since rat-MLP shows significant similarity with the C-terminal domain of human MUCZ, it is likely that the insoluble mucin complex from rat small intestine is a MUC2 homologue. Recently, the N-terminal domain of rat MUC2 has been cloned and sequenced [9], and a clear picture of this mucin apoprotein is now emerging. Since MUC2 seems to be the apoprotein in the insoluble mucin complex - the predominant gel-forming mucin from rat small intestine this appears to be the first biofilm-forming mucin with both a genetic and a biochemical identity.

Figure 3 Density-gradient centrifugation of material ‘insoluble’ (a) and ‘soluble’ (b) in 6 M guanidinium chloride isolated from human colon Mucosal scrapings from human colonic tissue were dispersed in 6 M guanidinium chloride supplemented with proteinase inhibitors and subjected to gentle stirring in the cold. After high-speed centrifugation ( I 8 000 rev./min; approx. I h in the cold), the pellet was re-extracted twice as above and the final residue was solubilized by reduction with dithiothreitol. These mucin subunits, as well as soluble material from the extraction, were subjected to densitygradient centrifugation in CsC1/4 M guanidinium chloride (initial density I .37 g/ml). The tubes were emptied from the bottom and sialic acid (A) fractions analysed for absorbance at 280 nm (-), and reactivity against a polyclonal antibody (A) raised in rabbit against a synthetic peptide from the MUCZ sequence. Five times more sample was used when analysing the ‘soluble’ mucins (b) than the ‘insoluble’ ones (a) for MUCZ readivity

1

120

a

The major part of human colonic mucins forms an insoluble glycoprotein complex containing MUCZ subunits In order to evaluate whether human intestinal mucins are also ‘insoluble’ and to provide the human MUC2 mucin with a biochemical identity, colonic tissue was extracted with several rounds of guanidinium chloride and the final residue solubilized with reduction. Isopycnic density-gradient centrifugation of the pooled supernatants and the subunits from the residue clearly showed that more than 90% of the glycoproteins behaving as mucins in the gradient are ‘insoluble’ by the criteria put forward above (Figure 3). T h e buoyant density and the ‘sharpness’ of the subunit peak from the insoluble complex varied significantly between preparations, and we postulate that this is due to differences in glycosylation. A polyclonal antibody raised to a synthetic peptide with the sequence NGLQPVRVEDPDGC, C-terminal to the tandem repeat region in MUC2 [5], reacted strongly with the subunits from the ‘insoluble’ complex but did not recognize anything in the solubilized material (also after reduction). We conclude that MUC2 is virtually absent in the ‘soluble’ fraction and that MUC2 is the building block of the ‘insoluble’ glycoprotein complex or at least one of them. MUC2 subunits from human colon can be separated into two populations with gel chromatography, suggesting that proteolytic ‘processing’ of the apoprotein is a distinct possibility (A. Hermann et al., this

2

1

40

0

0 1.5

1.4

1.3

Density (g/ml)

issue). Other investigators have also shown that MUC2 is a major mucin in the human colon ~301.

Apoproteins and glycoforms T h e oligosaccharide domains of gel-forming mucins are 100-200nm long and may contain several hundred oligosaccharides. Studies of the free oligosaccharides have shown that mucins may be substituted with a vast number of different oligosaccharides (see, for example [31-331). Studies with hybridization in situ show that the same mucin genes are expressed in several different tissues [23] and, since these

I995

849


850

tissues may express different glycosyltransferases, it is likely that there are several glycofoms of the same apoprotein. How the expression of apoproteins and glycosyltransferases is used to modulate the properties of the mucins and the epithelial biofilm is currently not understood.

Towards genetic and biochemical ID cards for the biofilm-forming mucins As mentioned above, a number of mucins have now been completely or partially cloned and sequenced and polynucleotide probes used to study the expression of the cognate genes in different tissues. However, since structural molecules may vary substantially in their turnover rates, it is not necessarily true that the most strongly expressed mucin is the predominant species in the secretion. Furthermore, the possibility that some of the biofilm-forming mucins lack a genetic identity must be considered. It is therefore important to isolate the predominant populations and describe their biochemical/ macromolecular properties, give them a MUC number by identifying their apoprotein and decide whether or not they occur as distinct glycoforms. For example, it seems likely that the MUCSAC gene product is the apoprotein in one of the dominating mucins in the stomach, airways and cervix. By identifymg the corresponding biochemical entities, it would be possible to subject them to a comparative study concerning their degree of oligomerization, post-translational modifications of the apoprotein, glycosylation and macromolecular properties. Such investigations would show how the ‘global’ properties of the mucins are matched to the particular needs of the tissue, shed light on the structure-function relationships of these highly complex glycoconjugates and thus provide an understanding of their ‘glycobiology’. Only when all major species of biofilm-forming mucins have been assigned both a genetic and a biochemical ID card and tools to identify them in physiological and biosynthetic studies are available can a true understanding of their biological significance be anticipated. This work was supported by grants from the Swedish Medical Research Council (grants 7902, 9711 and 9823), The Medical Faculty of Lund, The Smokeless Tobacco Research Council Inc., Greta och Johan Kocks Stiftelse, Alfred h e r l u n d s Stiftelse, Stiftelsen Lars Hiertas Minne and Tore Nilsons Fond for Medicinsk Forskning & Stiftelsen Riskforbundet for Cystisk Fibros.

Volume 23

1 Shogren, R., Gerken, T. and Jentoft, N. (1989) Biochemistry 28,5525-5536 2 Shirnizu, Y. (1993) Nature 366, 630-631 3 Hilkens, J., Ligtenberg, M. J. L., Vos, H. L. and Litvinov, S. V. (1992) Trends Biochern. Sci. 17, 359-363 4 Bobek, L. A., Tsai, H., Biesbrock, A. R. and Levine, M. J. (1993) J. Biol. Chern. 268, 20563-20569 5 Gum, J. R., Hicks, J. W., Toribara, N. W., Siddiki, B. and Kim, Y. S. (1994) J. Biol. Chern. 269, 2440-2446 6 Gum, J. R., Hicks, J. W., Toribara, N. W., Rothe, E.-M., Lagace, R. E. and Kim, Y. S. (1992) J. Biol. Chem. 267,21375-21383 7 Eckhardt, A. E., Tirnpte, C. S., Abernethy, J. L., Zhao, Y. and Hill, R. L. (1991) J. Biol. Chem. 266, 9678-9686 8 Xu, G., Huan, L.-J., Khatri, I. A. et al. (1992) J. Biol. Chern. 267,5401-5407 9 Ohrnori, H., Dorhrnan, A. F., Gallup, M. et al. (1994) J. Biol. Chern. 269, 17833-17840 10 Mayadas, T. N. and Wagner, D. D. (1991) Ann. N.Y. Acad. Sci. 614, 153-166 1 1 Carlstedt, I. and Sheehan, J. K. (1984) Ciba Found. Symp. 109, 157-172 12 Carlstedt, I., Herrrnann, A., Karlsson, H., Sheehan, J. K., Fransson, L.-.k and Hansson, G. C. (1993) J. Biol. Chern. 268, 18771-18781 13 Carlstedt, I., Lindgren, H., Sheehan, J. K., Ulrnsten, U. and Wingerup, L. (1983) Biochem. J. 211, 13-22 14 Sheehan, J. K. and Carlstedt, I. (1984) Biochern. J. 217,93-101 15 Carlstedt, I., Lindgren, H. and Sheehan, J. K. (1983) Biochern. J. 213,427-435 16 Mikkelsen, A., Stokke, B. T., Christensen, B. E. and Elgsaeter, A. (1985) Biopolymers 24, 1683- 1704 17 Sheehan, J. K., Oates, K. and Carlstedt, I. (1986) Biochem. J. 239, 147-153 18 Marianne, T., Perini, J.-M., Lafitte, J.-J. et al. (1987) Biochem. J. 248, 189-195 19 Sheehan, J. K. and Carlstedt, I. (1990) Biochern. J. 265, 169-178 20 Sheehan, J. K. and Carlstedt, I. (1987) Biochern. J. 239, 757-762 21 Carlstedt, I., Sheehan, J. K., Corfield, A. P. and Gallagher, J. T. (1985) Essays Biochern. 20,40-76 22 Guyonnet Duperat, V., Audie, J.-P., Debailleul, V. et al. (1995) Biochem. J. 305, 211-219 23 Audie, J. P., Janin, A., Porchet, N., Copin, M. C., Gosselin, B. and Aubert, J. P. (1993) J. Histochem. Cytochern. 41, 1479-1485 24 Thornton, D. J., Devine, P. L., Hanski, C., Howard, M. and Sheehan, J. K. (1994) Am. J. Respir. Crit. Care Med. 150, 823-832 25 Filipe, M. I. (1979) Invest. Cell Pathol. 2, 195-216


26 Suganuma, T., Katsuyama, T., Tsukahara, M., Tatematsu, M., Sakakura, Y. and Murata, F. (1981) Am. J. Anat. 161,219-238 27 Meerzaman, D., Charles, P., Daskal, E., Polymeropoulos, M. H., Martin, B. M. and Rose, M. C. (1994) J. Biol. Chem. 269, 12932-12939 28 Ota, H. and Katsuyama, T. (1992) Histochem. J. 24,86-92 29 Hansson, G. C., Baeckstrom, D., Carlstedt, 1. and Klinga-Levan, K. (1994) Biochem. Biophys. Res. Commun. 198, 181-190 30 Tytgat, K. M. A. J., Buller, H. A., Opdam, F. J. M., Kim, Y. S., Einerhand, A. W. C. and Dekker, J.

(1994) Gastroenterology 107, 1352-1363 31 Breg, J., Van Halbeek, H., Vliegenthart, F. G., Klein, A., Lamblin, G. and Roussel, P. (1988) Eur. J. Biochem. 171,643-654 32 Klein, A., Carnoy, C., Lamblin, G., Roussel, P., Kuik, J. A., de Waard, P. and Vliegenthart, F. G. (1991) Eur. J. Biochem. 198, 151-168 33 Van Halbeek, H., Breg, J., Vliegenthart, F. G., Klein, A., Lamblin, G. and Roussel, P. (1988) Eur. J. Biochem. 177,443-460 Received 21 June 1995

I995

85 I