With the advent of directed evolution techniques, protein
engineering has received a fresh impetus. Engineering proteins
for thermostability is a particularly exciting and challenging
field, as it is crucial for broadening the industrial use of
recombinant proteins. In addition to directed evolution, a
variety of partially successful rational concepts for engineering
thermostability have been developed in the past. Recent
results suggest that amino acid sequence comparisons of
mesophilic proteins alone can be used efficiently to engineer
thermostable proteins. The potential benefits of the underlying,
semirational ‘consensus concept’ are compared with those of
rational design and directed evolution approaches.
F Hoffmann-La Roche Ltd., Vitamins and Fine Chemicals Division,
Department VFB, Building 203, CH-4070 Basel, Switzerland
Current Opinion in Biotechnology 2001, 12:371–375
0958-1669/01/$ — see front matter
© 2001 Elsevier Science Ltd. All rights reserved.
Because natural enzymes are adapted to their particular
function in a living cell, in most instances they are poorly
suited for industrial applications that often encounter
extremes of pH, temperature and/or salinity. Both
for broadening the industrial applicability of enzymes
and for furthering our understanding of protein
structure/function relationships, protein engineering is
currently a very active area of research. Interest in this
subject area has further increased following the emer-
gence of high-throughput directed evolution techniques,
which can be seen as test tube models of natural evolu-
tion. This strong interest in protein engineering is
reflected in two recent publications dedicated entirely to
this topic [1,2]. The present review focuses on engineer-
ing proteins for thermostability and compares the
advantages and benefits of classical rational design prin-
ciples, directed evolution, and a ‘consensus approach’
that extracts valuable information from sequence
comparisons of homologous (mesophilic) proteins alone.
Rational design principles
The stability of a protein is determined by a multitude of
both local and long-range interactions. In order to achieve
pronounced thermostabilisation, several substitutions
each with a relatively small effect usually need to be com-
bined in a multiple mutant. This endeavour is facilitated
by the fact that in many cases the thermostabilisation
effects of individual mutations are independent and nearly
additive (e.g., [3–8]).
One strategy for identifying thermostabilising mutations
involves the comparison of more stable proteins with less
stable ones, with the goal of identifying amino acid
sequence patterns that correlate with thermostability .
In a systematic study, Perl et al. [10••] compared the cold
shock proteins from the thermophile Bacillus caldolyticus
and the mesophile Bacillus subtilis, which differ in only 12
out of 67 residues but display a considerable difference in
stability (15.8 kJ/mol at 70°C). Site-directed mutagenesis
of all 12 residues in the Bacillus caldolyticus enzyme
revealed that the difference in thermostability can be fully
accounted for by only two amino acid substitutions
(Glu3Arg, Glu66Leu) on the surface of the molecule. In
this illustrative example, with small proteins that exhibit
high homology and a pronounced difference in thermosta-
bility, it is noted that less than 20% of the amino acid
substitutions actually contribute to the difference in stabil-
ity. This observation highlights the problem of identifying
the relevant thermostabilising mutations in larger and less
homologous sets of proteins.
In other studies published last year, higher thermostability
was suggested to correlate with more proline and less
asparagine and glutamate residues , more arginines and
tyrosines but less cysteine and serine residues, and with
increased numbers of salt bridges and sidechain–sidechain
hydrogen bonds . In addition, higher thermostability
correlated with a larger fraction of residues in α helices and
with more arginine and less proline, cysteine and histidine
residues in those α helices . The information accumu-
lated to date indicates that nature relies on no single
strategy for stabilisation (see ), and as a result many
publications in this area arrive at different and even con-
flicting conclusions. The availability of more complete
genome sequences may cause the reliability of the correla-
tions to improve, but sequence statistics of this kind, when
taken alone, are still unlikely to provide useful information
for the accurate prediction of thermostabilising mutations.
Thermostabilising mutations either increase the thermody-
namic stability of a protein (i.e., they increase the free-energy
difference between the unfolded and the folded state) or
they decrease the rate of unfolding by increasing the free-
energy difference between the folded state and the transition
state of unfolding (see ). To achieve either of these,
various rational concepts have been proposed: to decrease
the entropy of the unfolded state by introducing additional
disulfide bridges or by X→Pro mutations; to increase α-helix
propensity by Gly→Ala substitutions or by stabilisation of
α-helix macrodipoles; to improve electrostatic interactions
between charged surface residues by introducing additional
salt bridges or even salt-bridge networks, or by predicting
Engineering proteins for thermostability: the use of sequence
alignments versus rational design and directed evolution
Martin Lehmann* and Markus Wyss†
thermostabilising mutations on the basis of calculations of
electrostatic potentials. Again, because the predictive power
of these rational concepts is rather limited , the supposed
thermostabilising mutations need to be tested individually
by site-directed mutagenesis. This reduces speed and
throughput and thereby limits the sequence space amenable
The term ‘directed evolution’ encompasses a series of
experimental techniques that reproduce, on an accelerated
timescale in the test tube, the evolution of natural diversi-
ty and environmental adaptation. This is achieved through
mutation and recombination and by giving the process a
‘direction’ towards the optimisation of one or more proper-
ties of interest. Either a selective pressure is applied, or in
each round of mutagenesis and/or recombination the
library of variants obtained is screened for the desired trait.
One frequently applied strategy comprises repeated
rounds of random mutagenesis, starting with a given par-
ent gene of interest. After each round, the best mutant(s)
is (are) selected and used as parent sequence(s) in the
following round of random mutagenesis. Typically, rather
low mutation frequencies are employed to suppress accu-
mulation of neutral or even deleterious mutations [13,14].
In a recent example of this type, two cycles of random
mutagenesis and screening yielded two mutants of phos-
pholipase A1 with six and seven amino acid substitutions
each; the mutants displayed increases in their temperature
of half-inactivation of 7 and 11°C, respectively .
Since the introduction of DNA shuffling in 1994 [17,18],
most directed evolution experiments encompass one or
more cycles of recombination between a set of homologous
sequences. These experiments may use a set of improved
variants of a given enzyme, obtained by random mutagene-
sis, or a set of naturally occurring, homologous genes isolated
from wild-type organisms. DNA shuffling was shown to be
very powerful in recombining favourable mutations and in
eliminating deleterious or neutral mutations.
Combinations of random mutagenesis and/or DNA (family)
shuffling have been used successfully to engineer protein
thermostability [19–21,22•,23–27]. In these examples,
increases in ‘thermostability’ of 14–20°C were attained with
7–19 amino acid substitutions (relative to the parent
protein), which were accumulated over up to eight genera-
tions of directed evolution. Two important conclusions
can be drawn from these results. Firstly, the increases in
thermostability obtained by directed evolution have, as yet,
been no more impressive than the best examples of rational
design (e.g., [28,29]), although they may have required less
time and/or less effort. Secondly, despite improved equip-
ment that allows higher throughput and increased
automation, the sequence space amenable to testing is still
rather limited. Key prerequisites for larger steps and longer
walks in sequence space include the efficient elimination
of neutral and deleterious mutations, higher frequencies of
recombination between homologous genes, recombination
events in stretches of low amino acid sequence identity
(e.g., [30••]), and powerful selection tools for improved
traits. Although longer walks in sequence space might not
actually be required for increases in thermostability of
20–30°C, they may be desirable for more dramatic increases
in stability and will definitely be advantageous for optimis-
ing other enzyme properties, such as reaction mechanism or
The consensus concept
A third, semirational approach for engineering thermosta-
bility is based on the hypothesis that at a given position in
an amino acid sequence alignment of homologous pro-
teins, the respective consensus amino acid contributes
more than average to the stability of the protein than the
nonconsensus amino acids. Consequently, substitution of
nonconsensus by consensus amino acids may be a feasible
approach for improving the thermostability of a protein.
Although Pantoliano et al.  were the first to apply the
‘consensus concept’, showing that the consensus-type
mutation Met50Phe increased the unfolding temperature
(Tm) of subtilisin BPN′ by 1.8°C, it was five years later that
Steipe et al.  offered a possible theoretical explanation,
based on statistical thermodynamics, for the feasibility of
this approach. Steipe et al.  predicted ten individual,
stabilising mutations for the immunoglobulin variable VL
domain, of which six were indeed stabilising. Further
experiments on the immunoglobulin VLand VHdomains
as well as on a catalytic single-chain Fv fragment con-
firmed these results and showed that the combination of a
set of stabilising mutations in a multiple variant provided
additive thermostabilising effects [5,32–34].
In addition to applying the consensus concept to sets of
homologous amino acid sequences, Steipe and coworkers
also tried to apply the concept to structural motifs. The
hypothesis that the most frequently occurring residues
in specific positions of β-turn motifs increase the folding
stability of a protein was experimentally confirmed for
an immunoglobulin VLdomain . In contrast, when
the concept was applied to surface-exposed loops and
turns, which were assumed to be stabilised primarily
by local interactions, only less stable variants were
observed, suggesting that this particular application of
the consensus concept does not allow reliable prediction
of thermostabilising mutations.
A set of systematic studies of the consensus concept has
also been provided by Fersht’s group. In one study ,
p53 homologues from 23 species were aligned, and 20
nonconsensus residues of human p53 were mutated
individually to the respective consensus residue. The
changes in stability ranged from +1.27 to –1.49 kcal/mol,
and the theoretical sum of the stability changes of the 20
individual mutations was no more than –0.48 kcal/mol.
372Protein technologies and commercial enzymes
Four stabilising mutations were combined in a quadruple
mutant (Met133Leu, Val203Ala, Asn239Tyr, Asn268Asp),
which was stabilised by 2.65 kcal/mol and displayed an
increase in Tmof 5.6°C.
In two other studies [36•,37], the consensus concept was
applied to GroEL minichaperones (i.e., fragments of
GroEL capable of facilitating protein folding). A sequence
alignment of 130 sequences of homologous chaperonin 60
proteins revealed that 31 (out of approximately 150)
amino acids of the Escherichia coli minichaperone
GroEL(193–345) occur with a frequency of <35% at their
respective positions in the alignment. Each of these
residues was replaced individually with the most fre-
quently occurring residue(s), yielding a total of 34 single
mutants which displayed differences in stability relative
to wild type ranging from +1.55 to –1.78 kcal/mol.
Summing up the stability effects of all the individual
mutations yields a theoretical overall difference in stabili-
ty of –0.30 kcal/mol, which is unlikely to be significantly
different from zero. Among the 34 amino acid substitu-
tions analysed, 13 were stabilising (38% success rate), five
were neutral, and 16 were destabilising. However, when
considering for substitution only those residues that occur
with a frequency of <20% (rather than <35%) at the
respective position in the alignment, 13 out of 18 muta-
tions were stabilising (72% success rate), two were neutral,
and only three were destabilising. Two multiple variants
each combining a set of six stabilising mutations were sta-
bilised by 6.99 and 6.15 kcal/mol, and displayed Tm
increases of 18.6 and 14.2°C relative to wild-type
No attempt has been made by either Steipe’s or Fersht’s
group to apply the consensus concept over the entire
sequence of a protein. In fact, on the basis of the data of
Nikolova et al.  and Wang et al. [36•] (see above), one
might be inclined to assume that such an attempt will fail,
because stabilising and destabilising mutations may coun-
terbalance each other. Very much to the contrary are the
results of Lehmann et al. [38••] describing the design of a
consensus phytase. In this study an appropriate computer
program was used to calculate an entire consensus
sequence from 13 homologous amino acid sequences of
wild-type phytases from mesophilic fungi. A synthetic
gene was constructed from the consensus sequence, and
recombinant expression of this gene gave rise to a consen-
sus phytase (consensus phytase-1) that was 15–26°C more
thermostable than all of its parents [38••]. Subsequently,
incorporation of additional wild-type sequences in the
alignment yielded consensus phytase-10, which displayed
32 amino acid differences relative to consensus phytase-1
and a further 7.4°C increase in Tm(M Lehmann et al.,
unpublished data). Both consensus phytases differ in at
least 80 amino acids from any of their parents. This clearly
shows that the consensus concept allows multiple amino
acid exchanges to be combined in a single step in order to
provide a significantly improved variant of the enzyme.
Remarkably, the overall correlation often observed
between increases in thermostability and decreases in
catalytic turnover rate at ambient temperature does not
apply to the consensus phytases; instead, the opposite is
true (M Lehmann et al., unpublished data).
In summary, the consensus approach allows much larger
steps in sequence space than most directed evolution tech-
niques reported to date. In addition, at present, increases in
thermostability of up to 30°C seem possible in a single shot.
In the case of engineering proteins for thermostability,
researchers are in the enviable situation of being able to
choose between three different, apparently equally
successful, strategies: rational design, directed evolution,
and the construction of (semirational) synthetic consensus
genes. Rather than playing off one approach against the
others, future efforts should focus on how to best combine
these alternative approaches in order to access a larger frac-
tion of the total sequence space and to approach the global
optimum for a given trait in a consistent and direct way. A
first successful study of this type has been reported by
Cherry et al. [39•]. In their endeavour to improve the
stability of a haem peroxidase for laundry applications, four
mutations were rationally designed: one to increase the
enzyme’s thermostability and three to increase resistance
to oxidative damage. The combination of these mutations
with favourable amino acid exchanges identified in direct-
ed evolution experiments yielded a final mutant with 174
times the thermal stability and 100 times the oxidative
stability of the wild-type haem peroxidase.
We expect that the consensus concept will prove to be, and
will thus be generally recognised as, an additional valuable
instrument in our tool box for designing more ther-
mostable enzymes for all types of industrial applications.
Construction of a single consensus protein has the poten-
tial to provide a similar increase in thermostability as many
months or years of rational design or as directed evolution
campaigns involving multiple generations and analysis of
thousands of mutants.
References and recommended reading
Papers of particular interest, published within the annual period of review,
have been highlighted as:
• of special interest
••of outstanding interest
1.Dalboge H, Borchert TV (Guest Eds): Protein engineering of
enzymes. Biochim Biophys Acta 2000, 1543:203-455.
2.Arnold FH (Guest Ed): Evolutionary protein design. Adv Protein
Chem 2001, 55:1-438.
3.Pantoliano MW, Whitlow M, Wood JF, Dodd SW, Hardman KD,
Rollence ML, Bryan PN: Large increases in general stability for
subtilisin BPN′ ′ through incremental changes in the free energy of
unfolding. Biochemistry 1989, 28:7205-7213.
4. Serrano L, Day AG, Fersht AR: Step-wise mutation of barnase to
binase. A procedure for engineering increased stability of proteins
and an experimental analysis of the evolution of protein stability.
J Mol Biol 1993, 233:305-312.
Engineering proteins for thermostability Lehmann and Wyss 373
5. Ohage E, Steipe B: Intrabody construction and expression. I. The
critical role of VLdomain stability.J Mol Biol 1999, 291:1119-1128.
von der Osten C, Branner S, Hastrup S, Hedegaard L, RasmussenMD,
Bisgard-Frantzen H, Carlsen S, Mikkelsen JM: Protein engineering of
subtilisins to improve stability in detergent formulations.J
Biotechnol 1993, 28:55-68.
7. Akasako A, Haruki M, Oobatake M, Kanaya S: High resistance of
Escherichia coli ribonuclease HI variant with quintuple
thermostabilizing mutations to thermal denaturation, acid
denaturation, and proteolytic degradation. Biochemistry 1995,
8. Shih P, Kirsch JF: Design and structural analysis of an engineered
thermostable chicken lysozyme. Protein Sci 1995, 4:1063-2072.
9.Haney PJ, Badger JH, Buldak GL, Reich CI, Woese CR, Olsen GJ:
Thermal adaptation analyzed by comparison of protein sequences
from mesophilic and extremely thermophilic Methanococcus
species.Proc Natl Acad Sci USA 1999, 96:3578-3583.
10. Perl D, Müller U, Heinemann U, Schmid FX: Two exposed amino acid
residues confer thermostability on a cold shock protein. Nat Struct
Biol 2000, 7:380-383.
The cold shock proteins from the thermophile Bacillus caldolyticus and the
mesophile Bacillus subtilis differ in only 12 out of 67 residues. Systematic
site-directed mutagenesis revealed that only two of these 12 residues
account for the considerable difference in stability between the two proteins.
For some of the other 10 residues, the B. caldolyticus protein was even sta-
bilised when the respective amino acid from the less thermostable B. sub-
tilis protein was introduced. These results question the usefulness of
sequence comparisons between thermophilic and mesophilic or psy-
chrophilic proteins for accurately predicting (thermo-) stabilising mutations.
11. Sriprapundh D, Vieille C, Zeikus JG: Molecular determinants of
xylose isomerase thermal stability and activity: analysis of
thermozymes by site-directed mutagenesis. Protein Eng 2000,
12. Kumar S, Tsai CJ, Nussinov R: Factors enhancing protein
thermostability. Protein Eng 2000, 13:179-191.
13. Wintrode PL, Arnold FH: Temperature adaptation of enzymes:
lessons from laboratory evolution. Adv Protein Chem 2001,
14. Steipe B: Evolutionary approaches to protein engineering.
Curr Top Microbiol Immunol 1999, 243:55-86.
15. Spector S, Wang A, Carp SA, Robblee J, Hendsch ZS, Fairman R,
Tidor B, Raleigh DP: Rational modification of protein stability by
the mutation of charged surface residues. Biochemistry 2000,
16. Song JK, Rhee JS: Simultaneous enhancement of thermostability
and catalytic activity of phospholipase A1 by evolutionary
molecular engineering. Appl Environ Microbiol 2000, 66:890-894.
17. Stemmer WPC: Rapid evolution of a protein in vitro by DNA
shuffling. Nature 1994, 370:389-391.
18. Stemmer WPC: DNA shuffling by random fragmentation and
reassembly: in vitro recombination for molecular evolution. Proc
Natl Acad Sci USA 1994, 91:10747-10751.
19. Giver L, Gershenson A, Freskgard PO, Arnold FH: Directed evolution
of a thermostable esterase. Proc Natl Acad Sci USA 1998,
20. Hoseki J, Yano T, Koyama Y, Kuramitsu S, Kagamiyama H: Directed
evolution of thermostable kanamycin-resistance gene: a
convenient selection marker for Thermus thermophilus. J Biochem
21. Kikuchi M, Ohnishi K, Harayama S: Novel family shuffling methods
for the in vitro evolution of enzymes. Gene 1999, 236:159-167.
22. Spiller B, Gershenson A, Arnold FH, Stevens RC: A structural view
of evolutionary divergence. Proc Natl Acad Sci USA 1999,
The three-dimensional structure of the parent wild-type p-nitrobenzyl
esterase was compared with that of a thermostabilised mutant that was
obtained by eight generations of directed evolution; the mutant displayed
13 amino acid substitutions and a 17°C increase in Tmrelative to the wild
type. Remarkably, the mutations exerted their influence on the esterase
structure over large distances, in a manner that would be difficult (or in
some cases even impossible at present) to predict. The loops with the
largest structural changes were not generally the sites of mutations. These
results therefore question the reliability of rational approaches that often
neglect long-range effects.
23. Zhao H, Arnold FH: Directed evolution converts subtilisin E into a
functional equivalent of thermitase. Protein Eng 1999, 12:47-53.
24. Gershenson A, Schauerte JA, Giver L, Arnold FH: Tryptophan
phosphorescence study of enzyme flexibility and unfolding in
laboratory-evolved thermostable esterases. Biochemistry 2000,
25. Kim GJ, Cheon YH, Kim HS: Directed evolution of a novel N-
carbamylase/D-hydantoinase fusion enzyme for functional
expression with enhanced stability. Biotechnol Bioeng 2000,
26. Miyazaki K, Wintrode PL, Grayling RA, Rubingh DN, Arnold FH:
Directed evolution study of temperature adaptation in a
psychrophilic enzyme. J Mol Biol 2000, 297:1015-1026.
27.Arnold FH: Combinatorial and computational challenges for
biocatalyst design. Nature 2001, 409:253-257.
28. van den Burg B, Vriend G, Veltman OR, Venema G, Eijsink VGH:
Engineering an enzyme to resist boiling. Proc Natl Acad Sci USA
29. Malakauskas SM, Mayo SL: Design, structure and stability of a
hyperthermophilic protein variant. Nat Struct Biol 1998, 5:470-475.
30. Coco WM, Levinson WE, Crist MJ, Hektor HJ, Darzins A, Pienkos PT,
Squires CH, Monticello DJ: DNA shuffling method for generating
highly recombined genes and evolved enzymes. Nat Biotechnol
The authors describe a novel strategy for the molecular evolution of proteins,
termed random chimeragenesis on transient templates (RACHITT). The
strategy relies on the ordering, trimming and joining of randomly cleaved
parental DNA fragments annealed to a transient, full-length, single-stranded
polynucleotide scaffold. This method may lead to a significant improvement
of recombination frequency and a more comprehensive exploitation of
sequence space than previously reported DNA shuffling strategies.
Chimeric libraries averaging 14 cross-overs per gene were obtained, com-
pared with the one to four cross-overs obtained with previous techniques. In
addition, four cross-overs per gene occurred in regions of 10 or fewer bases
of sequence identity. No unshuffled parental clones or duplicated sibling
chimeras were detected, and only a few inactive clones were identified.
31. Steipe B, Schiller B, Plückthun A, Steinbacher S: Sequence
statistics reliably predict stabilizing mutations in a protein
domain. J Mol Biol 1994, 240:188-192.
32. Ohage EC, Graml W, Walter MM, Steinbacher S, Steipe B: β β-Turn
propensities as paradigms for the analysis of structural motifs to
engineer protein stability. Protein Sci 1997, 6:233-241.
33. Ohage EC, Wirtz P, Barnikow J, Steipe B: Intrabody construction
and expression. II. A synthetic catalytic Fv fragment. J Mol Biol
34. Wirtz P, Steipe B: Intrabody construction and expression III:
engineering hyperstable VHdomains.Protein Sci1999, 8:2245-2250.
35. Nikolova PV, Henckel J, Lane DP, Fersht AR: Semirational design of
active tumor suppressor p53 DNA binding domain with enhanced
stability. Proc Natl Acad Sci USA 1998, 95:14675-14680.
36. Wang Q, Buckle AM, Foster NW, Johnson CM, Fersht AR: Design of
highly stable functional GroEL minichaperones. Protein Sci 1999,
An amino acid sequence alignment of 130 homologous chaperonin 60 pro-
teins was used to identify amino acids of an E. coli GroEL minichaperone
that occur with a frequency of <35% or <20% at their respective position in
the alignment. When each of these residues was replaced individually with
the most frequently occurring residue(s), stability differences relative to wild
type of +1.55 to –1.78 kcal/mol were observed, and 13 out of 34 (for amino
acids with a frequency of <35%; 38% success rate) and 13 out of 18 amino
acid substitutions (<20%; 72% success rate) were actually found to be sta-
bilising. Only five of the stabilising amino acid substitutions could have also
been predicted by comparison of E. coli GroEL with the sequences of the
five thermophilic chaperonin 60 proteins known at the time. When six of the
stabilising amino acid substitutions were combined in a multiple mutant, the
energetic effects were approximately additive, yielding an increase in Tmof
up to 18.6°C. Finally, a small-to-large stabilising mutation, Ala223Thr, was
identified that is one of the rare examples where the introduction of a polar
hydroxyl group is accompanied by stabilisation of a hydrophobic core.
37.Wang Q, Buckle AM, Fersht AR: Stabilization of GroEL
minichaperones by core and surface mutations. J Mol Biol 2000,
374Protein technologies and commercial enzymes
38. Lehmann M, Kostrewa D, Wyss M, Brugger R, D’Arcy A, Download full-text
Pasamontes L, van Loon APGM: From DNA sequence to improved
functionality: using protein sequence comparisons to rapidly design
a thermostable consensus phytase.Protein Eng 2000, 13:49-57.
Whereas Steipe’s and Fersht’s groups only tested their versions of the ‘consen-
sus concept’ on individual residues, the authors of this paper were the first to apply
the consensus approach over the entire sequence of a protein. A consensus
amino acid sequence was calculated from an alignment of 13 homologous fungal
phytases (myo-inositol hexakisphosphate phosphohydrolases), followed by con-
struction of a synthetic gene and recombinant expression. Astonishingly enough,
the resulting consensus phytase-1 was 15–26°C more thermostable than all its
parents but displayed catalytic properties closely resembling those of its parents.
39. Cherry JR, Lamsa MH, Schneider P, Vind J, Svendsen A, Jones A,
Pedersen AH: Directed evolution of a fungal peroxidase. Nat
Biotechnol 1999, 17:379-384.
To increase the stability of a haem peroxidase under laundry conditions (i.e.,
at pH 10.5, 50°C and 5–10 mM peroxide concentration), rational and direct-
ed evolution approaches were merged. Four rationally designed amino acid
substitutions were randomly (re-) combined with several thermostabilising
mutations obtained by directed evolution to yield a sevenfold mutant with
174 times the thermal stability and 100 times the oxidative stability of wild
type. These impressive improvements were compromised, however, by a 20-
fold lower specific activity at pH 10.5.
Engineering proteins for thermostability Lehmann and Wyss 375