259
Stabilizing proteins.., with hybrid vigour ative approaches to both random and site-directed mutagenesis approaches have now been suggested in two recent reports 4.s from the Carlsberg Laboratory in Copenhagen. In the first report 4, Karl Thomsen and his colleagues set out to improve the thermostability of bacterial j~-glucanases. These enzymes are important in the malting process of beer manufacture, where they act on the glucan-rich cell wall Mutagenesis for stabilizing of barley endosperm. Breakdown proteim: site-directed or of this wall allows amylase to reserendipity? lease glucose from stored starch 'Improvement' of naturally ocand thus to provide sugar for the curring enzymes is not, however, without precedent. From intensive subsequent fermentation. Heat is site-directed amino acid changes of used for extraction during the mashing process and therefore the barnase (an extracellular RNase longevity of glucanases is of both from B. ara),loliquefaciens) 2, it has practical and economic signifibeen possible to effect an increase in stability against unfolding of 5 kcal cance. The enzymes in question, (1-3, (21 kJ) mol-I _ around 50%. How1-4)-~-glucanases, were toned ever, barnase is small, has no disulphide linkages and undergoes a from B. macerans (MAC) and B. simple, reversible, two-state de- amyloliquefaciens (AMY). These naturation, making it relatively proteins have 74% sequence homsimple to measure the effects of ology and the MAC enzyme is mutation on folding (hence its somewhat more heat-stable than AMY, but is much less able to selection for such studies). The tolerate low pH. No three-dimenstructure ofbarnase has been solved by both NMR and X-ray crystal- sional structure information is lography. However, the majority available and so the researchers undertook to make hybrid enof enzymes of commercial interest zymes, combining C-terminal porare larger than barnase and have tions of MAC with N-terminal more post-translational modifications or require co-factors. Such portions of AMY, retaining the 'frame' specified by the conserved proteins represent a significant sequence. Their aim was to see challenge for the site-directed enwhether such hybrids could comgineering approach. The threedimensional structure is known for bine the heat- land acid-tolerance of some, but by no means all, of these the respective parent proteins. PCR (polymerase chain reaction) proteins. was used to construct the hybrids, using a pair of oligonudeotide Alternative routes to more stable primers for one DNA strand flankenzymes ing both the AMY land MAC Random mutagenesis has been genes, and another pair consisting used with some success to obtain of the appropriate AMY-MAC enzymes with altered properties fusion sequences for the other and stability3. However, the clear lessons from the barnase studies are DNA strands. In the first PCR that significant improvement of reaction using all four primers, DNA encoding N- and C-terminal stability requires multi-site changes in amino acids. This means that a portions of AMY and MAC was random mutagenesis programme is produced (Fig. 1). Following denaturation of this DNA and relikely to be slow in yielding results, even if an effective method of hybridization, PCR using the flanking primers only, was used to screening large numbers of microorganisms could be devised. Altem- generate the hybrid molecules.
As discussed recently in the Biotopics column t, genetic engineering has, in principle, considerable potential to deliver enzymes which have altered stability or physicochemical properties for research or production processes. It was also noted that, in practice, it has been difficult to determine the rules underlying the stability of proteins and apply them successfully.
~)
1991,ElsevierSciencePublishersLtd (UK) 0167- 9430/91/$2.00
Hybrids consisting of amino acid residues 1-16, 1-36, 1-78 and 1-152 of the AMY sequence, followed by the appropriate MAC sequence were produced and secreted from E. coli using Bacillus promoters. All the hybrid molecules were active and had specific activities higher than the MAC parent enzyme. Interestingly, and perhaps rather unexpectedly, three of the hybrid enzymes were significantly more thermostable than either parent. Thus, at pH 4.1 and 65°C, the MAC parent was completely ina,ctive within 10 minutes, and the AMY parent lost 70% of its initial activity. The hybrids, however, showed little loss of activity under these conditions even after one hour.
just add a little sugar- in the right place In the second report s, the same research group expressed the MAC enzyme and an AMY-MAC fusion (1-107 AMY, 108-214 MAC) in both E. coli and S. cerevisiae. MAC secreted from the yeast host was Nglycosylated at asparagine residues (a large change in mobility on SDSPAGE), while the hybrid molecule was O-glycosylated to limited extent (with no mass change obvious on SDS-PAGE). The hybrid molecule contained a potential Nlinked glycosylation site (Asn-ArgSer) which was absent from the MAC enzyme. In view of the apparent absence of N-linked glycosylation of the hybrid on SDSPAGE, the authors expected that it was not utilized. This proved to be so since mutation of the MAC enzyme to the hybrid sequence at this site was equally without effect on SDS-PAGE, In addition to possessing more than one site of Nglycosylation, the N-linked oligosat~charide chains on yeast-expressed MAC were heterogeneous, giving rise to a number of glycoforms. Non-glycosylated and glycosylated forms, both of MAC and the hybrid enzymes, were compared for thermostability at pH 6 and 70°C. Glycosylation of MAC approximatdy doubled its half-life. However, the rdatively limited O-glycosyhtion of the hybrid TIBTECHAUGUST1991(VOL9)
260
biotopics 1 m
AMYgene
MAC gene m
4 1
I
4
Rgure 1 Stages in the formation of hybrid bacillus glycanase genes. The oligonucleotide primers 1-4 are used to form the intermediate products with an overlap. After re-hybridization, the final product is formed by PCR using primers 1 and 4.
molecule extended its half-life 20fold. In this case, it appears that glycosylation generally increases glucanase half-life, but the nature and site of glycosylation play a more significant role than the extent ofglycosylation in this respect.
Take-home messages The stabilization of enzymes is a common requirement in industry. Protein engineering potentially provides a rational way to achieve this goal, but the ground rules are not clear and the three-dimensional structure of the enzyme may not be available. Thomsen offers a pragmatic approach to effecting a
change in the stability or other properties of industrially important enzymes. One should of course be cautious, because glucanases could prove to be a special case and the construction of homologous enzyme hybrids may not, in general, lead to desirable products. However, the careful work done on 'protein test beds', such as barnase, indicates that multiple discrete amino acid substitutions are needed to provide a significantly more stable enzyme. Hybridization is a rational way to make just such a large number of conservative changes in the absence of any knowledge of three-dimensional structure. The stabilizing effect of sugars is interesting. Since the normal bacterial hosts do not glycosylate glucanases this effect may well be applicable to other bacterial enzymes of commercial interest. The qualitative rather than quantitative effect of the sugar chains indicates that steric shielding or additional bonding interactions at specific sites is what is important here. Given the relative east with which site-directed mutage:~esis may be used to generate an N-linked glycosylation site (or simple expression in a host with the ability to glycosylate pre-existing acceptor sites), experiments on stabilization of other enzymes would appear to be well worthwhile.
References I Gcisow, M.J. (1991) Trends Bioteclmol. 9, 149-150 2 Serrano, L. and Fersht, A. R. (1989) Nature 342, 296-299 3 Bryan, P. N., Rollence, M. L., Pantollano, M. W., Wood,J., Finzel,B. C., Gilliand, G. L., Howard, A. J, and Poulos, T. L. (1986) Proteins 1,326-334 4 0 l s e n , O., Borriss, R., Simon, O. and Thomsen, K. K. (1991) Mot. Gen. Ge, et. 225, 177-185 $ Olsen, O. and Thomsen, K. K. (1991) J. Gem Microbiol. 137, 579-585
Michael J. Geisow Biodigm, 81 Kneeton Road, East Bridg[ord, Nottingham NG13 8PH, UK.
Corrigendum In the biotopics article entitled 'Annexins - forms without function - but not without fun' (TIBTECH9, 180-181, 1991) the author refers to phospholipase C (p. 180, 3rd col.) - this should have been phospholipase A2. We apologize for any confusion caused by this error.
Biosensors- a swell response to sugar Biosensor technology is slowly becoming more familiar to the biotechnology consumer and even to some sectors of the paying public. Diabetics will be aware of the glucose monitor which uses an enzyme electrode (glucose oxidase), to measure glucose concentration directly in a drop of blood. Biochemists, too, are starting to utilize commercial biosensor instrumentation, especially immunosensors. The first generation of immunosensors use surface plasmon resonance (SPR, an optical method), to detect changes in mass in a thin film containing immobilized antibody or antigen. In this manner, protein interactions can be followed in real TIBTECHAUGUST199] (VOL9)
time without the usual requirement of labelling one of the biochemical reactants. Instruments, more exotic still, are on their way, utilizing artificial membranes and engineered membrane proteins (see Ref. 1). One of the major difficulties in the development of commercial biosensors has been to make the systems sufficiently robust to provide reasonable shelf- and operating life, At present, most o f the devices have been electrochemical or electro-optical. Now a research group at MIT have demonstrated a functional device that undergoes a volume change in response to a specific biological interaction: sugar-lectin binding2.
Forces of expansion That separation medium familiar to biochemists, the polyacrylamide gel, undergoes substantial and often discontinuous volume changes in response to non-specific changes in solvent composition or temperature. The gel used by Kokufata et al. contained N-isopropylacrylamide groups and showed a large volume phase-shift at about 34°C. The idea was to couple this physical change to a specific biological interaction, and thereby to prove the feasibility of a volume-shift-based biosensor. This was achieved by polymerizing the gel with concanavalin A (ConA, a jack-bean-meal lectin) in situ. After polymerization, the ConA was trapped and the addition of a charged polysaccharide (i, this case
(~) 199], ElsevierSciencePublishersLtd (UK) 0167- 9430/91/$2.00
Stabilizing proteins.., with hybrid vigour ative approaches to both random and site-directed mutagenesis approaches have now been suggested in two recent reports 4.s from the Carlsberg Laboratory in Copenhagen. In the first report 4, Karl Thomsen and his colleagues set out to improve the thermostability of bacterial j~-glucanases. These enzymes are important in the malting process of beer manufacture, where they act on the glucan-rich cell wall Mutagenesis for stabilizing of barley endosperm. Breakdown proteim: site-directed or of this wall allows amylase to reserendipity? lease glucose from stored starch 'Improvement' of naturally ocand thus to provide sugar for the curring enzymes is not, however, without precedent. From intensive subsequent fermentation. Heat is site-directed amino acid changes of used for extraction during the mashing process and therefore the barnase (an extracellular RNase longevity of glucanases is of both from B. ara),loliquefaciens) 2, it has practical and economic signifibeen possible to effect an increase in stability against unfolding of 5 kcal cance. The enzymes in question, (1-3, (21 kJ) mol-I _ around 50%. How1-4)-~-glucanases, were toned ever, barnase is small, has no disulphide linkages and undergoes a from B. macerans (MAC) and B. simple, reversible, two-state de- amyloliquefaciens (AMY). These naturation, making it relatively proteins have 74% sequence homsimple to measure the effects of ology and the MAC enzyme is mutation on folding (hence its somewhat more heat-stable than AMY, but is much less able to selection for such studies). The tolerate low pH. No three-dimenstructure ofbarnase has been solved by both NMR and X-ray crystal- sional structure information is lography. However, the majority available and so the researchers undertook to make hybrid enof enzymes of commercial interest zymes, combining C-terminal porare larger than barnase and have tions of MAC with N-terminal more post-translational modifications or require co-factors. Such portions of AMY, retaining the 'frame' specified by the conserved proteins represent a significant sequence. Their aim was to see challenge for the site-directed enwhether such hybrids could comgineering approach. The threedimensional structure is known for bine the heat- land acid-tolerance of some, but by no means all, of these the respective parent proteins. PCR (polymerase chain reaction) proteins. was used to construct the hybrids, using a pair of oligonudeotide Alternative routes to more stable primers for one DNA strand flankenzymes ing both the AMY land MAC Random mutagenesis has been genes, and another pair consisting used with some success to obtain of the appropriate AMY-MAC enzymes with altered properties fusion sequences for the other and stability3. However, the clear lessons from the barnase studies are DNA strands. In the first PCR that significant improvement of reaction using all four primers, DNA encoding N- and C-terminal stability requires multi-site changes in amino acids. This means that a portions of AMY and MAC was random mutagenesis programme is produced (Fig. 1). Following denaturation of this DNA and relikely to be slow in yielding results, even if an effective method of hybridization, PCR using the flanking primers only, was used to screening large numbers of microorganisms could be devised. Altem- generate the hybrid molecules.
As discussed recently in the Biotopics column t, genetic engineering has, in principle, considerable potential to deliver enzymes which have altered stability or physicochemical properties for research or production processes. It was also noted that, in practice, it has been difficult to determine the rules underlying the stability of proteins and apply them successfully.
~)
1991,ElsevierSciencePublishersLtd (UK) 0167- 9430/91/$2.00
Hybrids consisting of amino acid residues 1-16, 1-36, 1-78 and 1-152 of the AMY sequence, followed by the appropriate MAC sequence were produced and secreted from E. coli using Bacillus promoters. All the hybrid molecules were active and had specific activities higher than the MAC parent enzyme. Interestingly, and perhaps rather unexpectedly, three of the hybrid enzymes were significantly more thermostable than either parent. Thus, at pH 4.1 and 65°C, the MAC parent was completely ina,ctive within 10 minutes, and the AMY parent lost 70% of its initial activity. The hybrids, however, showed little loss of activity under these conditions even after one hour.
just add a little sugar- in the right place In the second report s, the same research group expressed the MAC enzyme and an AMY-MAC fusion (1-107 AMY, 108-214 MAC) in both E. coli and S. cerevisiae. MAC secreted from the yeast host was Nglycosylated at asparagine residues (a large change in mobility on SDSPAGE), while the hybrid molecule was O-glycosylated to limited extent (with no mass change obvious on SDS-PAGE). The hybrid molecule contained a potential Nlinked glycosylation site (Asn-ArgSer) which was absent from the MAC enzyme. In view of the apparent absence of N-linked glycosylation of the hybrid on SDSPAGE, the authors expected that it was not utilized. This proved to be so since mutation of the MAC enzyme to the hybrid sequence at this site was equally without effect on SDS-PAGE, In addition to possessing more than one site of Nglycosylation, the N-linked oligosat~charide chains on yeast-expressed MAC were heterogeneous, giving rise to a number of glycoforms. Non-glycosylated and glycosylated forms, both of MAC and the hybrid enzymes, were compared for thermostability at pH 6 and 70°C. Glycosylation of MAC approximatdy doubled its half-life. However, the rdatively limited O-glycosyhtion of the hybrid TIBTECHAUGUST1991(VOL9)
260
biotopics 1 m
AMYgene
MAC gene m
4 1
I
4
Rgure 1 Stages in the formation of hybrid bacillus glycanase genes. The oligonucleotide primers 1-4 are used to form the intermediate products with an overlap. After re-hybridization, the final product is formed by PCR using primers 1 and 4.
molecule extended its half-life 20fold. In this case, it appears that glycosylation generally increases glucanase half-life, but the nature and site of glycosylation play a more significant role than the extent ofglycosylation in this respect.
Take-home messages The stabilization of enzymes is a common requirement in industry. Protein engineering potentially provides a rational way to achieve this goal, but the ground rules are not clear and the three-dimensional structure of the enzyme may not be available. Thomsen offers a pragmatic approach to effecting a
change in the stability or other properties of industrially important enzymes. One should of course be cautious, because glucanases could prove to be a special case and the construction of homologous enzyme hybrids may not, in general, lead to desirable products. However, the careful work done on 'protein test beds', such as barnase, indicates that multiple discrete amino acid substitutions are needed to provide a significantly more stable enzyme. Hybridization is a rational way to make just such a large number of conservative changes in the absence of any knowledge of three-dimensional structure. The stabilizing effect of sugars is interesting. Since the normal bacterial hosts do not glycosylate glucanases this effect may well be applicable to other bacterial enzymes of commercial interest. The qualitative rather than quantitative effect of the sugar chains indicates that steric shielding or additional bonding interactions at specific sites is what is important here. Given the relative east with which site-directed mutage:~esis may be used to generate an N-linked glycosylation site (or simple expression in a host with the ability to glycosylate pre-existing acceptor sites), experiments on stabilization of other enzymes would appear to be well worthwhile.
References I Gcisow, M.J. (1991) Trends Bioteclmol. 9, 149-150 2 Serrano, L. and Fersht, A. R. (1989) Nature 342, 296-299 3 Bryan, P. N., Rollence, M. L., Pantollano, M. W., Wood,J., Finzel,B. C., Gilliand, G. L., Howard, A. J, and Poulos, T. L. (1986) Proteins 1,326-334 4 0 l s e n , O., Borriss, R., Simon, O. and Thomsen, K. K. (1991) Mot. Gen. Ge, et. 225, 177-185 $ Olsen, O. and Thomsen, K. K. (1991) J. Gem Microbiol. 137, 579-585
Michael J. Geisow Biodigm, 81 Kneeton Road, East Bridg[ord, Nottingham NG13 8PH, UK.
Corrigendum In the biotopics article entitled 'Annexins - forms without function - but not without fun' (TIBTECH9, 180-181, 1991) the author refers to phospholipase C (p. 180, 3rd col.) - this should have been phospholipase A2. We apologize for any confusion caused by this error.
Biosensors- a swell response to sugar Biosensor technology is slowly becoming more familiar to the biotechnology consumer and even to some sectors of the paying public. Diabetics will be aware of the glucose monitor which uses an enzyme electrode (glucose oxidase), to measure glucose concentration directly in a drop of blood. Biochemists, too, are starting to utilize commercial biosensor instrumentation, especially immunosensors. The first generation of immunosensors use surface plasmon resonance (SPR, an optical method), to detect changes in mass in a thin film containing immobilized antibody or antigen. In this manner, protein interactions can be followed in real TIBTECHAUGUST199] (VOL9)
time without the usual requirement of labelling one of the biochemical reactants. Instruments, more exotic still, are on their way, utilizing artificial membranes and engineered membrane proteins (see Ref. 1). One of the major difficulties in the development of commercial biosensors has been to make the systems sufficiently robust to provide reasonable shelf- and operating life, At present, most o f the devices have been electrochemical or electro-optical. Now a research group at MIT have demonstrated a functional device that undergoes a volume change in response to a specific biological interaction: sugar-lectin binding2.
Forces of expansion That separation medium familiar to biochemists, the polyacrylamide gel, undergoes substantial and often discontinuous volume changes in response to non-specific changes in solvent composition or temperature. The gel used by Kokufata et al. contained N-isopropylacrylamide groups and showed a large volume phase-shift at about 34°C. The idea was to couple this physical change to a specific biological interaction, and thereby to prove the feasibility of a volume-shift-based biosensor. This was achieved by polymerizing the gel with concanavalin A (ConA, a jack-bean-meal lectin) in situ. After polymerization, the ConA was trapped and the addition of a charged polysaccharide (i, this case
(~) 199], ElsevierSciencePublishersLtd (UK) 0167- 9430/91/$2.00
