Protein Engineering vol.5 no.4 pp.361-365, 1992
Cellulose-binding domains: potential for purification of complex proteins
Jeffrey M.Greenwood, Edgar Ong, Neil R.Gilkes, R.Antony J.Warren, Robert C.Miller,Jr and Douglas G.Kilburn1 Department of Microbiology, University of British Columbia, Vancouver, British Columbia, Canada V6T 1Z3 'To whom correspondence should be addressed
Introduction Many cellulases comprise a catalytic domain linked to a noncatalytic cellulose-binding domain (CBD) (Gilkes et al., 1991). The domains may function independently when separated by proteolysis (Tomme etai, 1988; Gilkes etai, 1988) or by manipulation of the genes encoding the enzymes (Warren et al., 1987; Greenwood etai., 1989; Ong etai, 1989). We have shown that CBDs function normally when fused to heterologous polypeptides and can be used as affinity tags for the purification or immobilization of fusion proteins (Greenwood et al., 1989; Ong etai., 1989). The CBDs of endoglucanase A (CenA) and exoglucanase (Cex) from Cellulomonas fimi are located at the amino- and carboxylterminus of the respective enzymes. This allows fusion of a CBD, in its normal orientation, to either end of a heterologous polypeptide. This increases me versatility of the CBDs as affinity tags for protein purification and enzyme immobilization, especially if fusion to one end of a heterologous polypeptide leads to its inactivation. Such applications have been demonstrated using Escherichia coli alkaline phosphatase (PhoA) (Greenwood et al., 1989) and a /3-glucosidase from an Agrobacterium sp (Abg) (Ong et al., 1989) as fusion partners for the CBDs. However, like CenA and Cex (Gilkes etai., 1988, 1989), these fusion polypeptides are prone to proteolysis between the domains. Because PhoA and Abg are dimeric proteins (Sowadski et al., 1985; Day and Withers, 1986), the observed proteolysis results © Oxford University Press
Materials and methods Strains and plasmids Escherichia coli strains CC118 (Manoil and Beckwith, 1985) and CAG456 (Baker et al., 1984), and plasmids pEOl (Ong et al., 1989) and pUC18-1.6 cenA::TnphoA (Greenwood et al., 1989) were described previously. E.coli CC118 has a deletion in the phoA gene. E.coli CAG456 has an amber mutation in the htpR gene, which codes for a heat shock sigma factor. Use of this strain, which has reduced expression of heat shock-related protease genes, yielded higher levels of Abg-CBDcex (Ong et al., 1991). The pUC18-based plasmid pEOl contains a gene fusion between abg and die CBDCex coding sequence (Ong et al., 1989). pUC18-1.6 cenAy.TnphoA IX-8 had a TnphoA insertion in the coding region for the CenA catalytic domain. This plasmid was digested with BglQ and the large fragment recircularized using T4 DNA ligase to give pIX-8A. This deleted IS50R from TnphoA in pUC18-1.6 cenA: .TnphoA IX-8, and protected the plasmid from further transposition events. pIX-8A was sequenced (Sanger et al., 1977) across the fusion junction to determine the exact point of TnphoA insertion, which was after Gin174 of mature CenA (Figure 1). In both pEOl and pIX-8A the gene fusions were expressed from the lac promoter. Preparation of cell-free extracts Culture growth and cell extract preparation of E.coli CAG456/ pEOl has been described previously (Ong et al., 1991). For E.coli CC118/pIX-8A, cultures were grown at 30°C in LB medium supplemented with ampicillin (100 /ig/ml). Cells were resuspended in 50 mM Tris-HCl pH 7.5 (Tris buffer), 0.02% NaN3 prior to breakage in a French press. Phenylmethylsulfonylfluoride (10~3 M) and pepstatin A (4 X 10~5 M) were added immediately following breakage. Nucleic acids were precipitated with streptomycin sulfate, then removed by centrifugation. Abg-CBD C e x was purified by affinity chromatography on CF1 cellulose (Whatman) (Ong et al., 1989). Protein electrophoresis Sodium dodecyl sulfate -polyacrylamide gel electrophoresis (SDS-PAGE) (Laemmli, 1970) was used to analyze polypeptides. Electroblotting of polypeptides onto Immobilon-P (Millipore Corp.) and amino-terminal amino acid sequencing was carried out as described (Gilkes et al., 1989). Gel staining for alkaline phosphatase activity was as described (Greenwood et al., 1989) except that bovine serum albumin was omitted from the gel. Non-denaturing PAGE was performed as described (Laemmli, 1970), omitting the SDS, /3-mercaptoethanol and heat treatment. Running buffer for non-denaturing PAGE was as 361
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
The endoglucanase CenA and the exoglucanase Cex from Cellulomonas fimi each contain a discrete cellulose-binding domain (CBD), at the amino-terminus or car boxy 1-terminus respectively. The gene fragment encoding the CBD can be fused to the gene of a protein of interest. Using this approach hybrid proteins can be engineered which bind reversibly to cellulose and exhibit the biological activity of the protein partner. Alkaline phosphatase (PhoA) from Escherichia coli, and a /3-glucosidase (Abg) from an Agrobacterium sp. are dimeric proteins. The fusion polypeptides CenA-PhoA and Abg-CBC Cex are sensitive to proteolysis at the junctions between the fusion partners. Proteolysis results in a mixture of homo- and heterodimers; these bind to cellulose if one or both of the monomers carry a CBD, e.g. CenA-PhoA/ CenA-PhoA and CenA-PhoA/PhoA. CBD fusion polypeptides could be used in this way to purify polypeptides which associate with the fusion partner. Key words: affinity purification/cellulose/fusion protein/cellulosebinding domain/protein complexes
in a mixture of homo- and heterodimers which contain two, one or no CBDs. This report shows that both PhoA and Abg bind to cellulose when one or bom of die constituent monomers carry a CBD. This has practical implications for the use of CBD fusion polypeptides in the detection of other polypeptides which associate widi the fusion partner, and for the purification or immobilization of multimeric proteins or protein complexes.
J.M.Greenwood et al.
CenA 1
418
111 134 \wv\w CBD
VVV\
CAT
PT
kwww
CenA-PhoA 111 134
174 191
636
170 I
A
175 L
T
P
Q
A
D
180 S
Y
T
Q
cen A
JnphoA
Fig. 1. Schematic diagram of CenA-PhoA fusion polypeptide encoded by plasmid pIX-8A. Mature CenA is shown for comparison. The striped area represents the CBD; stippled area, PhoA; black area, Tn5-encoded amino acids; PT, proline-threonine linker; CAT, CenA catalytic domain. Numbers refer to amino acid residues at the mature amino-terminus of each polypeptide and the carboxyl-terminus of each domain structure. The DNA sequence and deduced amino acid sequence of the fusion junction in CenA —PhoA are shown. cenA DNA is underlined and amino acid residues are numbered.
described (Schagger and von Jagow, 1987), omitting the SDS. Polypeptides with /3-glucosidase activity were detected following non-denaturing PAGE by soaking the gel in 2 mM methylumbelliferyl-/3-D-glucoside (MUG) for 5 min at room temperature and visualizing fluorescent bands under UV light. Polypeptides were recovered from non-denaturing gel slices (5 mm X 2 mm) by incubation overnight at 4°C in 20 jtl 50 mM potassium phosphate buffer, pH 7.0. Then 45 jtl 2 X SDS-loading dye (Laemmli, 1970) was added and the sample agitated on a vortex mixer. The gel was removed by centrifugation at 15 000 r.p.m. for 10 min at room temperature. The supernatants were analyzed by SDS-PAGE. Binding of polypeptides to cellulose Avicel™ PH-101 (FMC International) was the microcrystalline cellulose used in binding studies. Abg—CBDCex polypeptides were bound to cellulose as described previously (Ong et al., 1991) and eluted from cellulose with 0.1 M Tris-HCl, pH 9.6. For binding of CenA—PhoA polypeptides Avicel was first washed three times with water and twice with Tris buffer. Cell extract was added to the Avicel and incubated for 30-60 min on ice with occasional mixing. The Avicel was recovered by centrifugation and washed once with 1 M NaCl in Tris buffer and twice with Tris buffer. Bound polypeptides were extracted into SDS-loading dye (Greenwood etal, 1989). For acid dissociation of Avicel-bound PhoA dimers, the Avicel was suspended in 50 mM KC1-HC1, pH 2.0 and incubated for 30 min on ice, then washed twice with 50 mM KC1-HC1, pH 2.0, 5 mM EDTA and once with Tris buffer, 5 mM EDTA. E.cpli PhoA (Sigma) was adjusted to pH 2.0 with 1 N HC1 and diluted into 50 mM KC1-HC1, pH 2.0, then incubated for 30 min on ice. Following incubation, EDTA was added to 5 mM and the solution was neutralized with 2 N NaOH. Mixtures of acid dissociated PhoA and Avicel-bound CenA—PhoA were made 10 mM in ZnCl2 or diluted with water to the same volume, then incubated at 37°C for 1 h with regular mixing. The Avicel was recovered by centrifugation and washed twice with Tris buffer (for samples with added zinc) or Tris buffer, 5 mM EDTA (for 362
'74 • -
'24 • -
Fig. 2. SDS-PAGE of Avicel-bound fraction of E.coli CC118/pIX-8A clarified cell extract, electroblotted onto Immobilon-P. Lanes 1 and 2 show different gel loadings of the same sample. Sizes of major bands are shown in kilodaltons. Unambiguous amino-terminal amino acid sequences were obtained for bands marked with an asterisk.
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
ATC GCG CTC ACC CCG CAG GCT GAC TCT TAT ACA CAA
Use of cellulose-binding fusion proteins Table I. Molecular mass determinations for CenA-PhoA polypeptidesa Polypeptide
Total amino acid residues11 Apparent molecular massc (kDa) Predicted molecular massd (kDa) Apparent minus predicted molecular mass (kDa)
CenA-PhoA 636 (639)e 'PhoA 467 CenA' 169 (172)
74 51 24
66.4 (66.7) 49.0 17.4 (17.7)
7.6 (7.3) 2.0 6.6 (6.3)
"Molecular mass data for monomer forms of dimeric polypeptides. b From the deduced amino acid sequence. c Deduced by SDS-PAGE. d Calculated from the deduced amino acid sequence. c Values in parentheses refer to amino-terminal variant of the polypeptide (Gilkes et al., 1988; Guo et al., 1988).
control samples without zinc). Bound polypeptides were extracted into SDS-loading dye (Greenwood et al., 1989).
1 2 97.4 _
3
4
5
6
7
3
4
5
6
7
"-
68 53 — =* 45 — 36 — = 29
1 2 106
—
80
—
«•
—
_
^
_
—
495— 32.5 — Fig. 3. Formation of CenA — PhoA/PhoA heterodimers on Avicel. Aciddissociated PhoA or Tris buffer, 5 mM EDTA (control) was added to aciddissociated CenA-PhoA bound to Avicel. Samples were incubated with or without 10 mM ZnCI2. Polypeptides binding to Avicel were analyzed by SDS —PAGE followed by: A, staining with Coomassie blue; B, alkaline phosphatase activity staining. Lanes: 1, calibration standards (sizes shown in kDa); 2 and 3, incubation with Tris buffer, 5 mM EDTA (lane 2, no ZnCl2; lane 3, 10 mM ZnCl2); 4—7, incubation with acid-dissociated PhoA (lanes 4 and 6, no ZnCl2; lanes 5 and 7, 10 mM ZnCI2). Approximate molar ratios of PhoA:CenA —PhoA were 2.5:1 (lanes 4 and 5) and 1:1 (lanes 6 and 7).
CBDCenA to cellulose were exploited in an experiment to reconstitute the heterodimer. Cellulose-bound polypeptides from a clarified cell extract of E.coli CC118/pIX-8A were treated at pH 2.0 to dissociate the PhoA dimers. Under these conditions the intact CenA —PhoA fusion polypeptide remained bound to cellulose (Figure 3) while PhoA polypeptides lacking a CBD were released into the supernatant. The Avicel was neutralized in the presence of EDTA to chelate zinc and prevent reassociation of PhoA subunits. E.coli PhoA which had been acid-dissociated in a similar fashion was added to the Avicel. The suspension was incubated in the presence or absence of excess zinc ions, and the Avicel-bound fraction was analyzed by SDS-PAGE 363
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
Results Cellulose-binding polypeptides from E.coli CC118/pIX-8A Several polypeptides in clarified cell extracts of E.coli CC118/ pIX-8A could bind to cellulose (Figure 2). Of these the 74 kDa and 24 kDa polypeptides contained the amino-terminal amino acid sequence of mature CenA, indicating that they were intact CenA—PhoA and an amino-terminal fragment of the fusion polypeptide containing CBDCenA, respectively. The 51 kDa polypeptide appeared to be the corresponding carboxy-terminal fragment comprising PhoA as it had the amino-terminal sequence Ala-Leu-Thr-Pro-Gln which matched amino acids 170-175 of CenA-PhoA (Figure 1), and it had an Mr close to that expected for such a fragment (Table I). Amino-terminal sequencing on the 49 and 44 kDa polypeptides yielded ambiguous sequence data. The PhoA polypeptide of 51 kDa was formed presumably by proteolysis of the CenA-PhoA fusion polypeptide between amino acids 169 and 170, within the catalytic domain fragment of CenA (Figure 1). Native PhoA does not bind to cellulose (Greenwood et al., 1989). Since the 51 kDa polypeptide lacked a CBD, it could bind to cellulose only by forming a dimer with an intact CenA-PhoA fusion polypeptide. In the absence of definitive amino-terminal sequences the origin of the 49 and 44 kDa polypeptides is unclear. While the 49 kDa polypeptide may be the result of further truncation of the 51 kDa polypeptide, previous results suggest that the 44 kDa polypeptide is an E.coli protein which binds non-specifically to cellulose (Greenwood et al., 1989). The apparent molecular masses of the 74 and 24 kDa polypeptides are ~ 7 kDa greater than expected from their amino acid sequences (Table I). However, the conformation of the proline-threonine linker impedes the migration of polypeptides containing it during SDS-PAGE, giving them apparent molecular masses 4 - 6 kDa greater than expected (Gilkes et al. ,1989; Shen et al., 1991). Amino-terminal sequencing of the 74 and 24 kDa polypeptides indicated that the CenA leader peptide had been processed at two related sites three amino acids apart. This is consistent with previous observations of CenA expressed in E.coli (Gilkes et al., 1988; Guo et al., 1988). Association of heterodimers on cellulose The active PhoA enzyme is a metalloprotein with one magnesium and two zinc atoms per monomer (Sowadski et al., 1985). The identical subunits of E. coli PhoA can be reversibly dissociated by treatment with dilute acid (Schlesinger and Barrett, 1965). Acid treatment releases the zinc atoms, which are required for the formation of the active dimer at neutral pH. The reversible dissociation of the PhoA dimer and the binding stability of
A.
J.M.Greenwood et al.
kDa
B
7
B
9
Fig. 4. Analysis of purified Abg-CBDCex by non-denaturing PAGE (lanes 1-5), MUG zymogram (lanes 4 and 5) and SDS-PAGE (lanes 6 - 9 ) , following storage of the protein for several months at 4°C. Purified Abg-CBDCex was rebound to cellulose (Avicel) following prolonged storage. The purified protein and the Avicel-bound fraction were analyzed by non-denaturing PAGE. Enzymatically active bands from MUG zymogram of Abg-CBDCex were further characterized by excision from the gel and analysis by SDS-PAGE. Lanes: 1, 4 and 6, purified Abg; 2 and 5, purified Abg-CBDCex; 3, fraction of lane 2 bound to Avicel (overloaded relative to lane 2); 7, lane 5 upper band; 8, lane 5 middle band; 9, lane 5 lower band.
Purification and properties of Abg-CBDCex Abg-CBD Cex was purified from an extract of E.coli CAG456/pEOl by affinity chromatography on cellulose (Ong et al., 1991). Analysis of the purified enzyme by SDS-PAGE revealed polypeptides of 68 and 51 kDa. Both polypeptides had amino-terminal amino acid sequences (Met-Thr-Asp-Pro-Asn) identical to that of native Abg (Wakarchuk et al., 1988). Both polypeptides reacted with anti-Abg serum, but only the 68 kDa polypeptide reacted with anti-Cex serum (Ong etal., 1989). Thus, the 68 kDa polypeptide appeared to be intact Abg-CBDCex and the 51 kDa polypeptide its proteolytic degradation product lacking CBDCex. Analysis of purified Abg-CBDCex following storage Following storage at 4°C for several months, analysis of purified Abg-CBDCex by non-denaturing PAGE revealed three distinct species (Figure 4, lane 2). All of the species showed enzymatic activity on a MUG zymogram (Figure 4, lane 5). Excision of the three protein bands from the non-denaturing gel and analysis by SDS-PAGE showed that the upper band contained only the 68 kDa polypeptide (Figure 4, lane 7), the middle band contained both the 68 and 51 kDa polypeptides (Figure 4, lane 8) and the lower band contained only the 51 kDa polypeptide (Figure 4, lane 9). Because Abg is a dimeric protein, the upper band represents the intact homodimer Abg-CBDCex/Abg-CBDCex, the middle band represents the heterodimer Abg-CBDCex/Abg, and the lower band represents the homodimer Abg/Abg. When the affinity-purified Abg-CBDCex preparation was rebound to cellulose only dimers containing at least one CBD were adsorbed (Figure 4, lane 3), as was observed with the CenA —PhoA fusion. The Abg/Abg homodimer in the purified Abg-CBDCex preparation did not bind appreciably to cellulose, and therefore must have been formed during prolonged storage following the initial purification. Relative levels of the 68 and 51 kDa polypeptides remained constant during prolonged storage 364
Discussion CBDs provide useful affinity tags for the purification of target polypeptides. The binding to cellulose of CenA-PhoA and Abg-CBDCex dimers in which only one of the monomers has an attached CBD shows that the CBDs should also be useful for detecting, and perhaps purifying, polypeptides which associate with a particular target polypeptide. This will require that at least one of the termini of the target polypeptide is neither buried nor required for such interactions. Association of polypeptides with the target could occur in vitro, for example on a cellulose column containing the immobilized target polypeptide, or in vivo, with the CBD—target polypeptide fusion expressed directly in the cell. This paper gives an example of the former of these possibilities, with formation of CenA —PhoA/PhoA heterodimers on microcrystalline cellulose. The susceptibility of CenA-PhoA and Abg-CBDCex to proteolytic cleavage between the fused domains highlights the importance of careful design in the construction of CBD fusion polypeptides. Ideally, the CBD and the polypeptides to which it is fused should be separated by a linker large enough to allow correct folding and independent function of the fusion partners but short enough and of such a nature as to be resistant to proteolytic attack in vivo. The presence of a section of the CenA catalytic domain in CenA — PhoA is the primary cause of the proteolytic sensitivity of the fusion polypeptide. This stretch of 40 amino acids following the proline-threonine linker together with the following 17 Tn5-encoded amino acids would probably have a disordered structure, sensitive to proteolytic attack. Amino-terminal sequencing of the 51 kDa polypeptide confirmed a site of proteolysis within the catalytic domain segment at a position which is normally resistant to proteolytic attack in CenA (Gilkes et al., 1989). Similarly, the presence of 37 amino acids of the Cex catalytic domain in Abg-CBDCex may contribute to the proteolytic sensitivity of this fusion polypeptide. However, deletion of the catalytic domain region and the proline-threonine linker from Abg-CBDCex did not result in a significant decrease in proteolysis (unpublished data), suggesting that sequences flanking the fusion linker can also contribute to proteolytic sensitivity. This is true for CenA, which is sensitive to proteolysis in vitro at both ends of the proline-threonine linker but not within the linker itself (Gilkes et al., 1989; Shen et al., 1991). It is not known if the adsorption of CenA-PhoA and AbgCBDCex homodimers to cellulose is mediated by just one CBD or if the simultaneous binding of both CBDs can also occur. Clearly, the possibility of simultaneous binding complicates modeling of the adsorption process and the determination of meaningful adsorption parameters. Although PhoA and Abg provide useful enzymic markers to quantitate unbound enzyme concentrations in adsorption analyses, these considerations preclude the use of CenA —PhoA and Abg-CBDCex to quantitatively analyze the adsorption of the CBD to cellulose. The usefulness of these and other similar fusions lies in the strength of binding of the CBD to cellulose, allowing the affinity purification of heterologous proteins or immobilization of specific enzymes or multi-enzyme complexes. The notion of multienzyme systems generated by gene fusion (Billow and Mosbach, 1991)
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
(Figure 3). When zinc was present, reassociation of PhoA monomers took place, and the presence of a 43 kDa PhoA polypeptide together with the 74 kDa CenA-PhoA polypeptide showed that CenA—PhoA/PhoA heterodimers had been formed which bound to Avicel. Increasing the amount of added PhoA increased the yield of bound heterodimer. It is not surprising that an excess of PhoA was necessary for significant heterodimer formation, as there would be direct competition from homodimer formation in solution (PhoA/PhoA) and possibly also on the cellulose (CenA-PhoA/CenA-PhoA).
(data not shown); this suggests that further proteolysis is not the reason for the appearance of the Abg/Abg homodimer. This species was probably formed by subunit exchange between heterodimers (i.e. Abg-CBDCex/Abg + Abg-CBDCex/Abg ^ Abg-CBDCex/Abg-CBDCex + Abg/Abg).
Use of cellulose-binding fusion proteins
can be carried one step further by using CBDs as affinity tags to immobilize such genetically derived polyfunctional enzymes to a cellulose support matrix. The results presented here point towards the use of CBD fusions for in vivo affinity tagging of multimeric proteins and protein complexes for subsequent purification or immobilization and as a tool for studying protein-protein interactions in vivo. Acknowledgements We thank Emily Kwan and Sandy Kielland for expert technical assistance, and Steve Withers for providing purified Abg. This research was supported by the Natural Sciences and Engineering Research Council of Canada.
References
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
Baker.T.A., Grossman.A.D. andGross.C.A. (1984) Proc. Natl. Acad. Sci. USA, 81, 6779-6783. Bulow.L. and Mosbach.K. (1991) Trends Biotechnol., 9, 226-231. Day,A.G. and Withers,S.G. (1986) Biochem. Cell. Biol., 64, 914-922. Gilkes,N.R., Warren.R.A.J., Miller,R.C.,Jr. and Kilburn.D.G. (1988)7. Biol. Chem., 263, 10401-10407. Gilkes,N.R., Kilburn.D.G., Miller,R.C.,Jr. and Warren.R.A.J. (1989)7. Biol. Chem., 264, 17802-17808. Gilkes,N.R., Henrissat,B., Kilburn.D.G., Miller,R.C.,Jr. and Warren.R.A.J. (1991) Microbiol. Rev., 55, 305-315. Greenwood.J.M., Gilkes.N.R., Kilbum.D.G., Miller,R.C.,Jr. and Warren.R.A.J. (1989) FEBS Lett., 244, 127-131. Guo.Z , Arfman.N., Ong,E., Gilkes,N.R., Kilburn,D.G., Warren,R.A.J. and Miller,R.C.,Jr. (1988) FEMS Microbiol. Lett., 49, 279-283. Laemmli.K. (1970) Nature, 227, 680-685. Manoil,C. and Beckwith,J. (1985) Proc. Natl. Acad. Sci. USA, 82, 8129-8133. Ong,E., Gilkes.N.R., Warren.R.A.J., Miller,R.C.,Jr. and Kilbura,D.G. (1989) Bio/Technology, 7, 604-607. Ong.E., Gilkes.N.R., Miller.R.C.Jr., Warren.R.A.J. and Kilburn.D.G. (1991) Enzyme Microb. Technol., 13, 59—65. Sanger.F., Nicklen.S. and Coulson.A.R. (1977) Proc. Natl. Acad. Sci. USA, 74, 5463-5467. Schagger,H. and von Jagow,G. (1987) Anal. Biochem., 166, 368-379. Schlesinger,M.J. and Barrett.K. (1965) 7. Biol. Chem., 240, 4284-4292. Shen.H., Schmuck.M., Pilz.I., Gilkes,N.R., Kilburn.D.G., Miller.R.C.Jr. and Warren.R.A.J. (1991)7. Biol. Chem., 266, 11335-11340. Sowadski.J.M., Handschumacher.M.D., Murthy.H.M.K., Foster.B.A. and Wyckoff.H.W. (1985)7. Mol. Biol., 186, 417-433. Tomme,P., Van Tilbeurgh.H., Pettersson.G., Van Damme.J., VandekerckhoveJ., Knowles,J., Teeri.T. and Claeyssens,M. (1988) Eur. 7. Biochem., 170, 575-581. Wakarchuk,W.W., Greenberg,N.M., Kilburn.D.G., Miller,R.C.,Jr. and Warren,R.A.J. (1988)7. Bacterioi, 170, 301-307. Warren.R.A.J., Gerhard.B., Gilkes.N.R., Owolabi.J B., Kilburn.D.G. and Miller,R.C.,Jr. (1987) Gene, 61, 421-427. Received on December 3, 1991; revised and accepted on March 26, 1992
365
Cellulose-binding domains: potential for purification of complex proteins
Jeffrey M.Greenwood, Edgar Ong, Neil R.Gilkes, R.Antony J.Warren, Robert C.Miller,Jr and Douglas G.Kilburn1 Department of Microbiology, University of British Columbia, Vancouver, British Columbia, Canada V6T 1Z3 'To whom correspondence should be addressed
Introduction Many cellulases comprise a catalytic domain linked to a noncatalytic cellulose-binding domain (CBD) (Gilkes et al., 1991). The domains may function independently when separated by proteolysis (Tomme etai, 1988; Gilkes etai, 1988) or by manipulation of the genes encoding the enzymes (Warren et al., 1987; Greenwood etai., 1989; Ong etai, 1989). We have shown that CBDs function normally when fused to heterologous polypeptides and can be used as affinity tags for the purification or immobilization of fusion proteins (Greenwood et al., 1989; Ong etai., 1989). The CBDs of endoglucanase A (CenA) and exoglucanase (Cex) from Cellulomonas fimi are located at the amino- and carboxylterminus of the respective enzymes. This allows fusion of a CBD, in its normal orientation, to either end of a heterologous polypeptide. This increases me versatility of the CBDs as affinity tags for protein purification and enzyme immobilization, especially if fusion to one end of a heterologous polypeptide leads to its inactivation. Such applications have been demonstrated using Escherichia coli alkaline phosphatase (PhoA) (Greenwood et al., 1989) and a /3-glucosidase from an Agrobacterium sp (Abg) (Ong et al., 1989) as fusion partners for the CBDs. However, like CenA and Cex (Gilkes etai., 1988, 1989), these fusion polypeptides are prone to proteolysis between the domains. Because PhoA and Abg are dimeric proteins (Sowadski et al., 1985; Day and Withers, 1986), the observed proteolysis results © Oxford University Press
Materials and methods Strains and plasmids Escherichia coli strains CC118 (Manoil and Beckwith, 1985) and CAG456 (Baker et al., 1984), and plasmids pEOl (Ong et al., 1989) and pUC18-1.6 cenA::TnphoA (Greenwood et al., 1989) were described previously. E.coli CC118 has a deletion in the phoA gene. E.coli CAG456 has an amber mutation in the htpR gene, which codes for a heat shock sigma factor. Use of this strain, which has reduced expression of heat shock-related protease genes, yielded higher levels of Abg-CBDcex (Ong et al., 1991). The pUC18-based plasmid pEOl contains a gene fusion between abg and die CBDCex coding sequence (Ong et al., 1989). pUC18-1.6 cenAy.TnphoA IX-8 had a TnphoA insertion in the coding region for the CenA catalytic domain. This plasmid was digested with BglQ and the large fragment recircularized using T4 DNA ligase to give pIX-8A. This deleted IS50R from TnphoA in pUC18-1.6 cenA: .TnphoA IX-8, and protected the plasmid from further transposition events. pIX-8A was sequenced (Sanger et al., 1977) across the fusion junction to determine the exact point of TnphoA insertion, which was after Gin174 of mature CenA (Figure 1). In both pEOl and pIX-8A the gene fusions were expressed from the lac promoter. Preparation of cell-free extracts Culture growth and cell extract preparation of E.coli CAG456/ pEOl has been described previously (Ong et al., 1991). For E.coli CC118/pIX-8A, cultures were grown at 30°C in LB medium supplemented with ampicillin (100 /ig/ml). Cells were resuspended in 50 mM Tris-HCl pH 7.5 (Tris buffer), 0.02% NaN3 prior to breakage in a French press. Phenylmethylsulfonylfluoride (10~3 M) and pepstatin A (4 X 10~5 M) were added immediately following breakage. Nucleic acids were precipitated with streptomycin sulfate, then removed by centrifugation. Abg-CBD C e x was purified by affinity chromatography on CF1 cellulose (Whatman) (Ong et al., 1989). Protein electrophoresis Sodium dodecyl sulfate -polyacrylamide gel electrophoresis (SDS-PAGE) (Laemmli, 1970) was used to analyze polypeptides. Electroblotting of polypeptides onto Immobilon-P (Millipore Corp.) and amino-terminal amino acid sequencing was carried out as described (Gilkes et al., 1989). Gel staining for alkaline phosphatase activity was as described (Greenwood et al., 1989) except that bovine serum albumin was omitted from the gel. Non-denaturing PAGE was performed as described (Laemmli, 1970), omitting the SDS, /3-mercaptoethanol and heat treatment. Running buffer for non-denaturing PAGE was as 361
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
The endoglucanase CenA and the exoglucanase Cex from Cellulomonas fimi each contain a discrete cellulose-binding domain (CBD), at the amino-terminus or car boxy 1-terminus respectively. The gene fragment encoding the CBD can be fused to the gene of a protein of interest. Using this approach hybrid proteins can be engineered which bind reversibly to cellulose and exhibit the biological activity of the protein partner. Alkaline phosphatase (PhoA) from Escherichia coli, and a /3-glucosidase (Abg) from an Agrobacterium sp. are dimeric proteins. The fusion polypeptides CenA-PhoA and Abg-CBC Cex are sensitive to proteolysis at the junctions between the fusion partners. Proteolysis results in a mixture of homo- and heterodimers; these bind to cellulose if one or both of the monomers carry a CBD, e.g. CenA-PhoA/ CenA-PhoA and CenA-PhoA/PhoA. CBD fusion polypeptides could be used in this way to purify polypeptides which associate with the fusion partner. Key words: affinity purification/cellulose/fusion protein/cellulosebinding domain/protein complexes
in a mixture of homo- and heterodimers which contain two, one or no CBDs. This report shows that both PhoA and Abg bind to cellulose when one or bom of die constituent monomers carry a CBD. This has practical implications for the use of CBD fusion polypeptides in the detection of other polypeptides which associate widi the fusion partner, and for the purification or immobilization of multimeric proteins or protein complexes.
J.M.Greenwood et al.
CenA 1
418
111 134 \wv\w CBD
VVV\
CAT
PT
kwww
CenA-PhoA 111 134
174 191
636
170 I
A
175 L
T
P
Q
A
D
180 S
Y
T
Q
cen A
JnphoA
Fig. 1. Schematic diagram of CenA-PhoA fusion polypeptide encoded by plasmid pIX-8A. Mature CenA is shown for comparison. The striped area represents the CBD; stippled area, PhoA; black area, Tn5-encoded amino acids; PT, proline-threonine linker; CAT, CenA catalytic domain. Numbers refer to amino acid residues at the mature amino-terminus of each polypeptide and the carboxyl-terminus of each domain structure. The DNA sequence and deduced amino acid sequence of the fusion junction in CenA —PhoA are shown. cenA DNA is underlined and amino acid residues are numbered.
described (Schagger and von Jagow, 1987), omitting the SDS. Polypeptides with /3-glucosidase activity were detected following non-denaturing PAGE by soaking the gel in 2 mM methylumbelliferyl-/3-D-glucoside (MUG) for 5 min at room temperature and visualizing fluorescent bands under UV light. Polypeptides were recovered from non-denaturing gel slices (5 mm X 2 mm) by incubation overnight at 4°C in 20 jtl 50 mM potassium phosphate buffer, pH 7.0. Then 45 jtl 2 X SDS-loading dye (Laemmli, 1970) was added and the sample agitated on a vortex mixer. The gel was removed by centrifugation at 15 000 r.p.m. for 10 min at room temperature. The supernatants were analyzed by SDS-PAGE. Binding of polypeptides to cellulose Avicel™ PH-101 (FMC International) was the microcrystalline cellulose used in binding studies. Abg—CBDCex polypeptides were bound to cellulose as described previously (Ong et al., 1991) and eluted from cellulose with 0.1 M Tris-HCl, pH 9.6. For binding of CenA—PhoA polypeptides Avicel was first washed three times with water and twice with Tris buffer. Cell extract was added to the Avicel and incubated for 30-60 min on ice with occasional mixing. The Avicel was recovered by centrifugation and washed once with 1 M NaCl in Tris buffer and twice with Tris buffer. Bound polypeptides were extracted into SDS-loading dye (Greenwood etal, 1989). For acid dissociation of Avicel-bound PhoA dimers, the Avicel was suspended in 50 mM KC1-HC1, pH 2.0 and incubated for 30 min on ice, then washed twice with 50 mM KC1-HC1, pH 2.0, 5 mM EDTA and once with Tris buffer, 5 mM EDTA. E.cpli PhoA (Sigma) was adjusted to pH 2.0 with 1 N HC1 and diluted into 50 mM KC1-HC1, pH 2.0, then incubated for 30 min on ice. Following incubation, EDTA was added to 5 mM and the solution was neutralized with 2 N NaOH. Mixtures of acid dissociated PhoA and Avicel-bound CenA—PhoA were made 10 mM in ZnCl2 or diluted with water to the same volume, then incubated at 37°C for 1 h with regular mixing. The Avicel was recovered by centrifugation and washed twice with Tris buffer (for samples with added zinc) or Tris buffer, 5 mM EDTA (for 362
'74 • -
'24 • -
Fig. 2. SDS-PAGE of Avicel-bound fraction of E.coli CC118/pIX-8A clarified cell extract, electroblotted onto Immobilon-P. Lanes 1 and 2 show different gel loadings of the same sample. Sizes of major bands are shown in kilodaltons. Unambiguous amino-terminal amino acid sequences were obtained for bands marked with an asterisk.
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
ATC GCG CTC ACC CCG CAG GCT GAC TCT TAT ACA CAA
Use of cellulose-binding fusion proteins Table I. Molecular mass determinations for CenA-PhoA polypeptidesa Polypeptide
Total amino acid residues11 Apparent molecular massc (kDa) Predicted molecular massd (kDa) Apparent minus predicted molecular mass (kDa)
CenA-PhoA 636 (639)e 'PhoA 467 CenA' 169 (172)
74 51 24
66.4 (66.7) 49.0 17.4 (17.7)
7.6 (7.3) 2.0 6.6 (6.3)
"Molecular mass data for monomer forms of dimeric polypeptides. b From the deduced amino acid sequence. c Deduced by SDS-PAGE. d Calculated from the deduced amino acid sequence. c Values in parentheses refer to amino-terminal variant of the polypeptide (Gilkes et al., 1988; Guo et al., 1988).
control samples without zinc). Bound polypeptides were extracted into SDS-loading dye (Greenwood et al., 1989).
1 2 97.4 _
3
4
5
6
7
3
4
5
6
7
"-
68 53 — =* 45 — 36 — = 29
1 2 106
—
80
—
«•
—
_
^
_
—
495— 32.5 — Fig. 3. Formation of CenA — PhoA/PhoA heterodimers on Avicel. Aciddissociated PhoA or Tris buffer, 5 mM EDTA (control) was added to aciddissociated CenA-PhoA bound to Avicel. Samples were incubated with or without 10 mM ZnCI2. Polypeptides binding to Avicel were analyzed by SDS —PAGE followed by: A, staining with Coomassie blue; B, alkaline phosphatase activity staining. Lanes: 1, calibration standards (sizes shown in kDa); 2 and 3, incubation with Tris buffer, 5 mM EDTA (lane 2, no ZnCl2; lane 3, 10 mM ZnCl2); 4—7, incubation with acid-dissociated PhoA (lanes 4 and 6, no ZnCl2; lanes 5 and 7, 10 mM ZnCI2). Approximate molar ratios of PhoA:CenA —PhoA were 2.5:1 (lanes 4 and 5) and 1:1 (lanes 6 and 7).
CBDCenA to cellulose were exploited in an experiment to reconstitute the heterodimer. Cellulose-bound polypeptides from a clarified cell extract of E.coli CC118/pIX-8A were treated at pH 2.0 to dissociate the PhoA dimers. Under these conditions the intact CenA —PhoA fusion polypeptide remained bound to cellulose (Figure 3) while PhoA polypeptides lacking a CBD were released into the supernatant. The Avicel was neutralized in the presence of EDTA to chelate zinc and prevent reassociation of PhoA subunits. E.coli PhoA which had been acid-dissociated in a similar fashion was added to the Avicel. The suspension was incubated in the presence or absence of excess zinc ions, and the Avicel-bound fraction was analyzed by SDS-PAGE 363
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
Results Cellulose-binding polypeptides from E.coli CC118/pIX-8A Several polypeptides in clarified cell extracts of E.coli CC118/ pIX-8A could bind to cellulose (Figure 2). Of these the 74 kDa and 24 kDa polypeptides contained the amino-terminal amino acid sequence of mature CenA, indicating that they were intact CenA—PhoA and an amino-terminal fragment of the fusion polypeptide containing CBDCenA, respectively. The 51 kDa polypeptide appeared to be the corresponding carboxy-terminal fragment comprising PhoA as it had the amino-terminal sequence Ala-Leu-Thr-Pro-Gln which matched amino acids 170-175 of CenA-PhoA (Figure 1), and it had an Mr close to that expected for such a fragment (Table I). Amino-terminal sequencing on the 49 and 44 kDa polypeptides yielded ambiguous sequence data. The PhoA polypeptide of 51 kDa was formed presumably by proteolysis of the CenA-PhoA fusion polypeptide between amino acids 169 and 170, within the catalytic domain fragment of CenA (Figure 1). Native PhoA does not bind to cellulose (Greenwood et al., 1989). Since the 51 kDa polypeptide lacked a CBD, it could bind to cellulose only by forming a dimer with an intact CenA-PhoA fusion polypeptide. In the absence of definitive amino-terminal sequences the origin of the 49 and 44 kDa polypeptides is unclear. While the 49 kDa polypeptide may be the result of further truncation of the 51 kDa polypeptide, previous results suggest that the 44 kDa polypeptide is an E.coli protein which binds non-specifically to cellulose (Greenwood et al., 1989). The apparent molecular masses of the 74 and 24 kDa polypeptides are ~ 7 kDa greater than expected from their amino acid sequences (Table I). However, the conformation of the proline-threonine linker impedes the migration of polypeptides containing it during SDS-PAGE, giving them apparent molecular masses 4 - 6 kDa greater than expected (Gilkes et al. ,1989; Shen et al., 1991). Amino-terminal sequencing of the 74 and 24 kDa polypeptides indicated that the CenA leader peptide had been processed at two related sites three amino acids apart. This is consistent with previous observations of CenA expressed in E.coli (Gilkes et al., 1988; Guo et al., 1988). Association of heterodimers on cellulose The active PhoA enzyme is a metalloprotein with one magnesium and two zinc atoms per monomer (Sowadski et al., 1985). The identical subunits of E. coli PhoA can be reversibly dissociated by treatment with dilute acid (Schlesinger and Barrett, 1965). Acid treatment releases the zinc atoms, which are required for the formation of the active dimer at neutral pH. The reversible dissociation of the PhoA dimer and the binding stability of
A.
J.M.Greenwood et al.
kDa
B
7
B
9
Fig. 4. Analysis of purified Abg-CBDCex by non-denaturing PAGE (lanes 1-5), MUG zymogram (lanes 4 and 5) and SDS-PAGE (lanes 6 - 9 ) , following storage of the protein for several months at 4°C. Purified Abg-CBDCex was rebound to cellulose (Avicel) following prolonged storage. The purified protein and the Avicel-bound fraction were analyzed by non-denaturing PAGE. Enzymatically active bands from MUG zymogram of Abg-CBDCex were further characterized by excision from the gel and analysis by SDS-PAGE. Lanes: 1, 4 and 6, purified Abg; 2 and 5, purified Abg-CBDCex; 3, fraction of lane 2 bound to Avicel (overloaded relative to lane 2); 7, lane 5 upper band; 8, lane 5 middle band; 9, lane 5 lower band.
Purification and properties of Abg-CBDCex Abg-CBD Cex was purified from an extract of E.coli CAG456/pEOl by affinity chromatography on cellulose (Ong et al., 1991). Analysis of the purified enzyme by SDS-PAGE revealed polypeptides of 68 and 51 kDa. Both polypeptides had amino-terminal amino acid sequences (Met-Thr-Asp-Pro-Asn) identical to that of native Abg (Wakarchuk et al., 1988). Both polypeptides reacted with anti-Abg serum, but only the 68 kDa polypeptide reacted with anti-Cex serum (Ong etal., 1989). Thus, the 68 kDa polypeptide appeared to be intact Abg-CBDCex and the 51 kDa polypeptide its proteolytic degradation product lacking CBDCex. Analysis of purified Abg-CBDCex following storage Following storage at 4°C for several months, analysis of purified Abg-CBDCex by non-denaturing PAGE revealed three distinct species (Figure 4, lane 2). All of the species showed enzymatic activity on a MUG zymogram (Figure 4, lane 5). Excision of the three protein bands from the non-denaturing gel and analysis by SDS-PAGE showed that the upper band contained only the 68 kDa polypeptide (Figure 4, lane 7), the middle band contained both the 68 and 51 kDa polypeptides (Figure 4, lane 8) and the lower band contained only the 51 kDa polypeptide (Figure 4, lane 9). Because Abg is a dimeric protein, the upper band represents the intact homodimer Abg-CBDCex/Abg-CBDCex, the middle band represents the heterodimer Abg-CBDCex/Abg, and the lower band represents the homodimer Abg/Abg. When the affinity-purified Abg-CBDCex preparation was rebound to cellulose only dimers containing at least one CBD were adsorbed (Figure 4, lane 3), as was observed with the CenA —PhoA fusion. The Abg/Abg homodimer in the purified Abg-CBDCex preparation did not bind appreciably to cellulose, and therefore must have been formed during prolonged storage following the initial purification. Relative levels of the 68 and 51 kDa polypeptides remained constant during prolonged storage 364
Discussion CBDs provide useful affinity tags for the purification of target polypeptides. The binding to cellulose of CenA-PhoA and Abg-CBDCex dimers in which only one of the monomers has an attached CBD shows that the CBDs should also be useful for detecting, and perhaps purifying, polypeptides which associate with a particular target polypeptide. This will require that at least one of the termini of the target polypeptide is neither buried nor required for such interactions. Association of polypeptides with the target could occur in vitro, for example on a cellulose column containing the immobilized target polypeptide, or in vivo, with the CBD—target polypeptide fusion expressed directly in the cell. This paper gives an example of the former of these possibilities, with formation of CenA —PhoA/PhoA heterodimers on microcrystalline cellulose. The susceptibility of CenA-PhoA and Abg-CBDCex to proteolytic cleavage between the fused domains highlights the importance of careful design in the construction of CBD fusion polypeptides. Ideally, the CBD and the polypeptides to which it is fused should be separated by a linker large enough to allow correct folding and independent function of the fusion partners but short enough and of such a nature as to be resistant to proteolytic attack in vivo. The presence of a section of the CenA catalytic domain in CenA — PhoA is the primary cause of the proteolytic sensitivity of the fusion polypeptide. This stretch of 40 amino acids following the proline-threonine linker together with the following 17 Tn5-encoded amino acids would probably have a disordered structure, sensitive to proteolytic attack. Amino-terminal sequencing of the 51 kDa polypeptide confirmed a site of proteolysis within the catalytic domain segment at a position which is normally resistant to proteolytic attack in CenA (Gilkes et al., 1989). Similarly, the presence of 37 amino acids of the Cex catalytic domain in Abg-CBDCex may contribute to the proteolytic sensitivity of this fusion polypeptide. However, deletion of the catalytic domain region and the proline-threonine linker from Abg-CBDCex did not result in a significant decrease in proteolysis (unpublished data), suggesting that sequences flanking the fusion linker can also contribute to proteolytic sensitivity. This is true for CenA, which is sensitive to proteolysis in vitro at both ends of the proline-threonine linker but not within the linker itself (Gilkes et al., 1989; Shen et al., 1991). It is not known if the adsorption of CenA-PhoA and AbgCBDCex homodimers to cellulose is mediated by just one CBD or if the simultaneous binding of both CBDs can also occur. Clearly, the possibility of simultaneous binding complicates modeling of the adsorption process and the determination of meaningful adsorption parameters. Although PhoA and Abg provide useful enzymic markers to quantitate unbound enzyme concentrations in adsorption analyses, these considerations preclude the use of CenA —PhoA and Abg-CBDCex to quantitatively analyze the adsorption of the CBD to cellulose. The usefulness of these and other similar fusions lies in the strength of binding of the CBD to cellulose, allowing the affinity purification of heterologous proteins or immobilization of specific enzymes or multi-enzyme complexes. The notion of multienzyme systems generated by gene fusion (Billow and Mosbach, 1991)
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
(Figure 3). When zinc was present, reassociation of PhoA monomers took place, and the presence of a 43 kDa PhoA polypeptide together with the 74 kDa CenA-PhoA polypeptide showed that CenA—PhoA/PhoA heterodimers had been formed which bound to Avicel. Increasing the amount of added PhoA increased the yield of bound heterodimer. It is not surprising that an excess of PhoA was necessary for significant heterodimer formation, as there would be direct competition from homodimer formation in solution (PhoA/PhoA) and possibly also on the cellulose (CenA-PhoA/CenA-PhoA).
(data not shown); this suggests that further proteolysis is not the reason for the appearance of the Abg/Abg homodimer. This species was probably formed by subunit exchange between heterodimers (i.e. Abg-CBDCex/Abg + Abg-CBDCex/Abg ^ Abg-CBDCex/Abg-CBDCex + Abg/Abg).
Use of cellulose-binding fusion proteins
can be carried one step further by using CBDs as affinity tags to immobilize such genetically derived polyfunctional enzymes to a cellulose support matrix. The results presented here point towards the use of CBD fusions for in vivo affinity tagging of multimeric proteins and protein complexes for subsequent purification or immobilization and as a tool for studying protein-protein interactions in vivo. Acknowledgements We thank Emily Kwan and Sandy Kielland for expert technical assistance, and Steve Withers for providing purified Abg. This research was supported by the Natural Sciences and Engineering Research Council of Canada.
References
Downloaded from http://peds.oxfordjournals.org/ at University of Strathclyde on January 9, 2015
Baker.T.A., Grossman.A.D. andGross.C.A. (1984) Proc. Natl. Acad. Sci. USA, 81, 6779-6783. Bulow.L. and Mosbach.K. (1991) Trends Biotechnol., 9, 226-231. Day,A.G. and Withers,S.G. (1986) Biochem. Cell. Biol., 64, 914-922. Gilkes,N.R., Warren.R.A.J., Miller,R.C.,Jr. and Kilburn.D.G. (1988)7. Biol. Chem., 263, 10401-10407. Gilkes,N.R., Kilburn.D.G., Miller,R.C.,Jr. and Warren.R.A.J. (1989)7. Biol. Chem., 264, 17802-17808. Gilkes,N.R., Henrissat,B., Kilburn.D.G., Miller,R.C.,Jr. and Warren.R.A.J. (1991) Microbiol. Rev., 55, 305-315. Greenwood.J.M., Gilkes.N.R., Kilbum.D.G., Miller,R.C.,Jr. and Warren.R.A.J. (1989) FEBS Lett., 244, 127-131. Guo.Z , Arfman.N., Ong,E., Gilkes,N.R., Kilburn,D.G., Warren,R.A.J. and Miller,R.C.,Jr. (1988) FEMS Microbiol. Lett., 49, 279-283. Laemmli.K. (1970) Nature, 227, 680-685. Manoil,C. and Beckwith,J. (1985) Proc. Natl. Acad. Sci. USA, 82, 8129-8133. Ong,E., Gilkes.N.R., Warren.R.A.J., Miller,R.C.,Jr. and Kilbura,D.G. (1989) Bio/Technology, 7, 604-607. Ong.E., Gilkes.N.R., Miller.R.C.Jr., Warren.R.A.J. and Kilburn.D.G. (1991) Enzyme Microb. Technol., 13, 59—65. Sanger.F., Nicklen.S. and Coulson.A.R. (1977) Proc. Natl. Acad. Sci. USA, 74, 5463-5467. Schagger,H. and von Jagow,G. (1987) Anal. Biochem., 166, 368-379. Schlesinger,M.J. and Barrett.K. (1965) 7. Biol. Chem., 240, 4284-4292. Shen.H., Schmuck.M., Pilz.I., Gilkes,N.R., Kilburn.D.G., Miller.R.C.Jr. and Warren.R.A.J. (1991)7. Biol. Chem., 266, 11335-11340. Sowadski.J.M., Handschumacher.M.D., Murthy.H.M.K., Foster.B.A. and Wyckoff.H.W. (1985)7. Mol. Biol., 186, 417-433. Tomme,P., Van Tilbeurgh.H., Pettersson.G., Van Damme.J., VandekerckhoveJ., Knowles,J., Teeri.T. and Claeyssens,M. (1988) Eur. 7. Biochem., 170, 575-581. Wakarchuk,W.W., Greenberg,N.M., Kilburn.D.G., Miller,R.C.,Jr. and Warren,R.A.J. (1988)7. Bacterioi, 170, 301-307. Warren.R.A.J., Gerhard.B., Gilkes.N.R., Owolabi.J B., Kilburn.D.G. and Miller,R.C.,Jr. (1987) Gene, 61, 421-427. Received on December 3, 1991; revised and accepted on March 26, 1992
365
