Glycosylation of the primary binding pocket of a subtilisin protease causes a remarkable broadening in stereospecificity in peptide synthesis

Article abstract of DOI:10.1039/b010021h

Site-selective glycosylation at position 166 at the base of the primary specificity S₁ pocket in the serine protease subtilisin Bacillus lentus (SBL) created glycoproteins that are capable of catalyzing the coupling reactions of not only L- ami

Full text of DOI:10.1039/b010021h

View Article Online / Journal Homepage / Table of Contents for this issue
Glycosylation of the primary binding pocket of a subtilisin protease
causes a remarkable broadening in stereospecificity in peptide
synthesis
a
b
c
Kazutsugu Matsumoto, Benjamin G. Davis* and J. Bryan Jones*
a
Department of Applied Chemistry and Biotechnology, Faculty of Engineering, Fukui University,
Bunkyo 3-9-1, Fukui 910-8507, Japan
Department of Chemistry, University of Durham, South Road, Durham, UK DH1 3LE.
E-mail: Ben.Davis@durham.ac.uk
Department of Chemistry, University of Toronto, 80 St. George St, Toronto, Ontario, Canada,
M5S 3H6
b
c
Received (in Liverpool, UK) 8th December 2000, Accepted 8th March 2001
First published as an Advance Article on the web 1st May 2001
Site-selective glycosylation at position 166 at the base of the
primary specificity S pocket in the serine protease subtilisin
1
Bacillus lentus (SBL) created glycoproteins that are capable
of catalyzing the coupling reactions of not only - amino acid
esters but also -amino acid esters to give the corresponding
dipeptides in good yields as a result of greatly broadened
substrate specificities that can be rationalized by the
interaction of the glycans acting as chiral auxiliaries in
stereochemically mismatched pairs.
L
D
Scheme 1
3
effective chiral auxiliaries and we wished to probe the effect of
introducing the homochiral groups a–d (Scheme 1) into the
primary binding region of an enzyme as a tactic for broadening
substrate stereospecificity, perhaps by increasing the potential
for stereochemically mismatched pairs.4
Enzymatic peptide coupling requires minimal protection of the
substrate, proceeds under mild condition and without racemiza-
1
tion. In spite of these advantages, two major problems have
Representative carbohydrates were attached to the interior of
limited the use of serine proteases in peptide synthesis. One is
their efficient proteolytic (amidase) activity which causes
hydrolysis of the coupled peptide product, and the other is their
stringent structural and stereo-specificity which typically con-
5
the primary S
1
binding pocket, which modulates P
1
specificity,
of the serine protease subtilisin Bacillus lentus (SBL, EC
.4.21.14) by site-selective glycosylation2 at position 166.
,6
3
Gratifyingly, this also resulted in enhanced E/A values relative
fines their use in synthesis to a limited range of -amino acid
L
to unglycosylated SBL-WT (Scheme 1).2
substrates. Controlled site-selective glycosylation can be used
to create glycosylated proteases with greatly improved esterase
activities and enhanced esterase–amidase activity ratios (E/As)
that are up to 8.4- and 17.2-fold enhanced over their
Firstly, we probed the structural breadth of the P
-amino acids,
1)–(3) as acyl donors, with glycinamide (7) as an acyl acceptor
1
specificity
of S166C-S-a–d by examining the ligation of
L
(
(
2
1
7
Scheme 2, Table 1)† using 1 mg ml of active enzyme in
2
unglycosylated wildtype (WT) counterparts, respectively. By
simple 1+1 water–DMF solutions. In accord with our goals of
not reducing the existing substrate breadth of SBL, these results
indicated that the introduction of carbohydrate groups a–d did
virtue of their higher esterase and lower amidase activities,
these glycosylated enzymes are excellent candidates for
efficient amide bond formation as they possess enhanced
acylating properties and yet substantially reduced hydrolytic
activity towards the peptide products of coupling. Moreover,
these catalysts can be constructed on scales of hundreds of
milligrams, thereby allowing their use in multigram scale
syntheses. We reasoned that such glycosylated enzymes would
not only be efficient catalysts for peptide ligation but that the
presence of an internally bound carbohydrate might influence
the stereospecificity of such ligations. Carbohydrates are
not affect the essential, inherent ability of SBL to accept -
L
amino acids as acyl donors in peptide coupling reactions. Good
Scheme 2
Table 1 Glyco-SBL catalyzed peptide coupling reactionsa
b
Yield (%) with
Acyl donor
Acyl acceptor
Product
time/h
SBL-WT S166C-a S166C-b S166C-c S166C-d
Z-
Z-
Z-
1
L
L
L
-PheOBn (1)
-AlaOBn (2)
-GluOMe(3)
GlyNH
7
2
(7)
Z-
Z-
Z-
Z-
Z-
Z-
Z-
Z-
Z-
L
L
L
L
L
L
-PheGlyNH
-AlaGlyNH
-GluGlyNH
-Phe-
-Ala-
-Glu-
-PheGlyNH
-AlaGlyNH
-GluGlyNH
2
(9)
(10)
(11)
1
5
5
92
91
62
57
0
0
0
0
0
95
85
58
28
15
48
6
93
77
65
34
16
50
8
91
92
54
31
22
51
7
95
83
67
32
11
55
8
2
7
2
c
L
8
8
7
7
7
-AlaNH
2
(8)
L
-AlaNH
-AlaNH
-AlaNH
2
(12)
(13)
(14)
24
c
2
L
2
24
c
3
L
2
24
d
Z-
Z-
Z-
a
D
D
D
-PheOBn (4)
-AlaOBn (5)
-GluOBn (6)
D
D
D
2
(15)
(16)
(17)
48
d
2
48
80
63
77
62
72
64
b
70
64
d
2
48
DMF–water, 1+1, 0.1 M donor, 0.3 M acceptor, 0.036 mol% enzyme. Under the same conditions, spontaneous reaction did not occur. Isolated yields. c 0.2
M acceptor. d Further 0.036 mol% added at 24 h.
DOI: 10.1039/b010021h
Chem. Commun., 2001, 903–904
This journal is © The Royal Society of Chemistry 2001
903

View Article Online
the glyco-SBLs, Fig. 1(b), the presence of a homochiral
auxiliary at position 166 with correct stereochemical disposition
of its C-6 hydroxymethyl group and ring oxygen creates a
hydrogen bond acceptor and donor motif which allows binding
of the NHZ group in the S
operate either by binding mode (a) or (b). However, in these
glyco-SBLs, neither -amino acids with the (a) mode nor -
1
pocket. Thus, glyco-SBLs may
L
D
amino acids with the (b) mode represent optimally matched
diastereomeric pairs. This is consistent with their broadened
specificity. By introducing a carbohydrate moiety at position
Fig. 1 Summarized representation of molecular mechanics analysis which
showed different binding modes of (a) -amino acids in SBL-WT or glyco-
SBLs and (b) -amino acids only in glyco-SBLs
1
66 we have therefore tailored an enzyme with previously
exclusive -stereospecificity and excellent efficiency (matched
pair) into an enzyme that has broadened - and -ster-
L
L
D
L
D
to excellent yields of 9, (91–95%), 10, (77–92%)
,
and fair yields
eospecificity and reduced, but still good, efficiency due to
mismatched pairing. This modified enzyme clearly has greater
synthetic utility. This molecular mechanics analysis therefore
raises the interesting possibility that the stereochemistry of the
sidechain at 166 may influence stereospecificity through a
double diastereoselective process. Further studies, including the
synthesis of diastereomeric protein glycosylating reagents, to
explore whether this is correct or whether location of just polar
achiral groups at this position will have the same effect are now
in progress. In addition, we are probing the precise effects of
of 11, (54–67%) compared well, or were superior to, those for
SBL-WT.
Our next goal was to investigate any secondary effects of S
A specificity using the a-branched
-alaninamide (8) as a challenging acyl acceptor probe. The
restricted S A specificity of the unglycosylated SBL-WT
enzyme is demonstrated by its ability to catalyze only the
coupling of 8 with the preferred -phenylalaninate acyl donor 1.
Remarkably, in spite of the small S
SBLs S166C-S-a–d were catalysts for the coupling of
1
pocket glycosylation upon P
1
L
1
L
8
1
A pocket all of the glyco-
-amino
L
these modifications upon kinetic parameters kcat and K
M
using
acids 1–3 with 8. In the synthesis of 13, low absolute yields
were observed yet these represent a significant breakthrough in
comparison with the absence of coupling activity prior to
glycosylation. Excitingly, the yields of the coupling of 3 with 8
were similar to those obtained using unhindered 7 (48–55%).
L
- and -amino acid substrates as both acceptors and donors in
D
order to correlate the observed yield enhancements with true
kinetic specificity.
In conclusion, glyco-SBLs S166C-S-a–d accept a wide range
of substrates including -amino acids as acyl donors and an a-
D
These results represent a dramatic improvement of the S
1
A
branched acyl acceptor to give a variety of dipeptides, many in
very high yields, that cannot be synthesized with SBL-WT.
Furthermore, these dramatic improvements have been achieved
specificity of SBL. Interestingly, chemically-modified mutant
enzymes bearing non-polar modifications at position 166 do not
9
catalyze this ligation. Possibly by exploiting polar interactions,
without the loss of the natural ability of SBL to handle -amino
L
we have obtained our goal of broadening substrate specificity,
acids. It therefore represents a true broadening of synthetic
utility and demonstrates that this site-selective glycosylation
technology is a powerful tool for enhancing the application of
enzymes in organic synthesis.
here of P
1
A towards -Ala to allow the synthesis of 12–14,
L
without diminishing the natural breadth of SBL. Indeed,
previous molecular mechanics analysis of similarly internally
glycosylated enzymes have indicated that such internally-bound
carbohydrates may act as hydrogen-bonding motifs that give
We acknowledge NSERC (Canada) and Genencor Inter-
national, Inc for generous funding. We are also grateful to
Genencor International, Inc., for providing WT- and S166C-
SBL enzymes, and to Drs Rick Bott and Kanjai Khumtaveeporn
for helpful discussions. We also thank the EPSRC (UK) for
access to the Mass Spectrometry Service at Swansea and the
Chemical Database Service at Daresbury.
2
rise to enhanced kinetic parameters.
Finally, we examined these powerful glyco-SBLs in the
coupling of -amino acids (4)–(6) as acyl donors with acyl
D
acceptor 1. The number of examples of enzyme catalyzed
coupling of -amino acids at the C-terminus of peptides by
using
D
D
-amino acid acyl donors is vanishingly small and even
1
0
then proceed with typically low efficiencies. For example, to
the best of our knowledge, yields above 10% for the
References and notes
†
Electronic supplementary information (ESI) available: detailed experi-
9
incorporation of
with this, SBL-WT did not accept
D
-Glu have never been achieved. Consistent
mental procedures of peptide couplings and selected chemical data of the
products. See http://www.rsc.org/suppdata/cc/b0/b010021m/
D
-amino acids as acyl donors
and starting materials 4–6 were recovered. In dramatic contrast,
all of the glyco-SBLs S166C-S-a–d were able to catalyze the
1
C.-H. Wong and G. M. Whitesides, Enzymes in Synthetic Organic
Chemistry, Pergamon Press, Oxford, 1994, pp 41–130; K. Faber,
Biotransformations in Organic Chemistry, 2nd ed. Springer-Verlag,
Heidelberg, 1995, pp. 298–305 and references cited therein.
coupling of all three of these -amino acid esters with acyl
D
acceptor 1. The reactions of 4 in all cases were slow and gave 15
in low yields, the best being 8% using S166C-S-b,d, and
starting material 4 was recovered after 48 h in all cases. Peptide
couplings of 5, 6 with 7 proceeded more rapidly, and the yields
2 R. C. Lloyd, B. G. Davis and J. B. Jones, Bioorg. Med. Chem., 2000, 8,
1
537.
3
4
H. Kunz and K. Ruck, Angew. Chem., Int. Ed. Engl., 1993, 32, 336.
S. Masamune, W. Choy, J. S. Peterson and L. R. Sita, Angew. Chem.,
Int. Ed. Engl., 1985, 24, 1.
were dramatically improved. In fact, the good yields of the
dipeptide 17 (62–64%) were, surprisingly, superior to those
found for coupling of -Glu. Indeed, S166C-S-c showed a 1.2+1
stereochemical preference for -glutamate 6 over -glutamate 3.
-Ala dipeptide (16) were also
D
-
L
5
6
Nomenclature according to I. Schechter and A. Berger, Biochem.
Biophys. Res. Commun., 1967, 27, 152.
D
L
Very high yields (up to 80%) of
observed.
D
B. G. Davis, R. C. Lloyd and J. B. Jones, J. Org. Chem., 1998, 63, 9614;
B. G. Davis and J. B. Jones, Synlett, 1999, 1495; B. G. Davis, M. A. T.
Maughan, M. P. Green, A. Ullman and J. B. Jones, Tetrahedron:
Asymmetry, 2000, 11, 245; B. G. Davis, R. C. Lloyd and J. B. Jones,
Bioorg. Med. Chem., 2000, 8, 1527; B. G. Davis, Chem. Commun.,
Such is the remarkable nature of these broadened P
stereospecificities that we speculated that these glyco-SBLs
may bind -amino acids in a different mode from -amino acids.
1
D
L
2
001, 351.
Molecular mechanics analyses of SBL-WT and S166C-a–d
with substrates 1–6 resulted in the models summarized in Fig. 1.
In SBL-WT the normal binding mode for the acyl-enzyme
intermediate, which is shown in Fig. 1(a), operates. The amino
7
8
9
Determined by titration with PMSF: C. Y. Hsia, G. Granshaw, C. Paech
and C. J. Murray, Anal. Biochem., 1996, 242, 221.
For 3D-structure of SBL-WT see M. Knapp, J. Daubermann and R. R.
Bott, RCSB-PDB entry 1jea.
K. Khumtaveeporn, G. DeSantis and J. B. Jones, Tetrahedron:
Asymmetry, 1999, 10, 2563.
acid side chain (R) binds in the S
group covalently linked to O of the side chain of Ser221. This
binding mode is not available to -amino acids. In contrast, in
1
pocket with the reacting acyl
g
D
10 Y. Asano and T. L. Lübbehüsen, J. Biosci. Bioeng., 2000, 89, 295.
904
Chem. Commun., 2001, 903–904