Modulating the Nucleophile of a Glycoside Hydrolase through Site-Specific Incorporation of Fluoroglutamic Acids

Article abstract of DOI:10.1021/jacs.8b04235

Understanding the detailed mechanisms of enzyme-catalyzed hydrolysis of the glycosidic bond is fundamentally important, not only to the design of tailored cost-efficient, stable and specific catalysts but also to the development of specific glycosidase inhibitors as therapeutics. Retaining glycosidases employ two key carboxylic acid residues, typically glutamic acids, in a double-displacement mechanism involving a covalent glycosyl-enzyme intermediate. One Glu functions as a nucleophile while the other acts as a general acid/base. A significant part of enzymatic proficiency is attributed to a "perfect match" of the electrostatics provided by these key residues, a hypothesis that has been remarkably difficult to prove in model systems or in enzymes themselves. We experimentally probe this synergy by preparing synthetic variants of a model glycosidase Bacillus circulans β-xylanase (Bcx) with the nucleophile Glu78 substituted by 4-fluoro or 4,4-difluoroglutamic acid to progressively reduce nucleophilicity. These Bcx variants were semisynthesized by preparation of optically pure fluoroglutamic acid building blocks, incorporation into synthetic peptides, and ligation onto a truncated circular permutant of Bcx. By measuring the effect of altered electrostatics in the active site on enzyme kinetic constants, we show that lowering the nucleophile pKa by two units shits the pH-dependent activity by one pH unit. Linear free energy correlations using substrates of varying leaving group ability indicate that by reducing nucleophilic catalysis the concerted mechanism of the enzyme is disrupted and shifted toward a dissociative pathway. Our study represents the first example of site-specific introduction of fluorinated glutamic acids into any protein. Furthermore, it provides unique insights into the synergy of nucleophilic and acid/base catalysis within an enzyme active site.

Full text of DOI:10.1021/jacs.8b04235

Subscriber access provided by JAMES COOK UNIVERSITY LIBRARY
Article
Modulating the nucleophile of a glycoside hydrolase
through site-specific incorporation of fluoroglutamic acids
Miriam P. Koetzler, Kyle Robinson, Hong-Ming Chen, Mark
Okon, Lawrence P. McIntosh, and Stephen G. Withers
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b04235 • Publication Date (Web): 12 Jun 2018
Downloaded from http://pubs.acs.org on June 12, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 14
Journal of the American Chemical Society
1
2
3
4
5
6
7
Modulating the nucleophile of a glycoside hydrolase through site-spe-
cific incorporation of fluoroglutamic acids
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
1
1
1,2
1,2,3
Miriam P. Kötzler , Kyle Robinson , Hong-Ming Chen , Mark Okon , Lawrence P. McIntosh
and
1,2
Stephen G. Withers *
1
2
3
Department of Chemistry, Department of Biochemistry and Molecular Biology, Michael Smith Laboratories, University of
British Columbia, Vancouver, BC, Canada V6T 1Z1
Keywords: Bacillus circulans endo--1,4-xylanase, nucleophilic catalysis, 4-fluoroglutamic acid, 4,4-difluoroglutamic acid,
protein semisynthesis, unnatural amino acid, linear free energy relationship, enzyme engineering.
ABSTRACT: Understanding the detailed mechanisms of enzyme-catalyzed hydrolysis of the glycosidic bond is fundamentally im-
portant, not only to the design of tailored cost-efficient, stable and specific catalysts, but also to the development of specific glyco-
sidase inhibitors as therapeutics. Retaining glycosidases employ two key carboxylic acid residues, typically glutamic acids, in a
double-displacement mechanism involving a covalent glycosyl-enzyme intermediate. One Glu functions as a nucleophile while the
other acts as a general acid/base. A significant part of enzymatic proficiency is attributed to a “perfect match” of the electrostatics
provided by these key residues, a hypothesis that has been remarkably difficult to prove in model systems or in enzymes themselves.
We experimentally probe this synergy by preparing synthetic variants of a model glycosidase, Bacillus circulans -xylanase (Bcx)
with the nucleophile Glu78 substituted by 4-fluoro or 4,4-difluoroglutamic acid to progressively reduce nucleophilicity. These Bcx
variants were semi-synthesized by preparation of optically pure fluoroglutamic acid building blocks, incorporation into synthetic
peptides and ligation onto a truncated circular permutant of Bcx. By measuring the effect of altered electrostatics in the active site on
enzyme kinetic constants, we show that lowering the nucleophile pKa by two units shits the pH-dependent activity by one pH unit.
Linear free energy correlations using substrates of varying leaving group ability indicate that by reducing nucleophilic catalysis, the
concerted mechanism of the enzyme is disrupted and shifted towards a dissociative pathway. Our study represents the first example
of site-specific introduction of fluorinated glutamic acids into any protein. Furthermore, it provides unique insights into the synergy
of nucleophilic and acid/base catalysis within an enzyme active site.
Introduction
1
7
Glycoside hydrolases are among the most proficient catalysts known, accelerating reactions by up to ~10 fold above the sluggish,
uncatalyzed processes. They are also among the most widely used enzymes commercially, with applications ranging from food
processing and brewing through to the textiles, biofuel and pulp and paper industries. Some of these enzymes, among them the
1
2
–5
Bacillus circulans endo--1,4-xylanase (Bcx), have been engineered to optimize their stability and activity for use under harsh pro-
4,6,7
cessing conditions.
Further fine-tuning of catalytic properties, including shifting or broadening pH-dependent activity profiles,
would be of great interest. However, this requires a deep mechanistic understanding of factors underlying catalysis, such as protona-
tion equilibria in the active site and electrostatic stabilization of the transition states.
The general models of catalytic mechanisms employed by glycoside hydrolases were introduced by Koshland over half a century
8
9,10
ago and have been heavily refined in the interim. Inverting glycoside hydrolases (which effect inversion of anomeric stereochem-
istry) employ a direct displacement mechanism via a single oxocarbenium-like transition state. By contrast retaining glycoside hy-
drolases use a two-step mechanism involving two oxocarbenium-like transition states (ii and iv in Scheme 1). One carboxylate (Glu78
in Bcx) acts as a nucleophilic catalyst, attacking the anomeric carbon to form a covalent glycosyl enzyme intermediate (glycosylation
step (i) to (ii)). This transient covalent intermediate (iii) is then hydrolyzed in the second step, releasing the free reducing sugar with
net retention of anomeric configuration (deglycosylation step (iv) to (v)). The second carboxylic acid (Glu172 in Bcx) acts as a
general acid catalyst in the first step, transferring a proton onto the glycosidic oxygen as the nucleophile attacks and thereby enhancing
the leaving group ability of an otherwise unactivated aglycone (leaving group pK
covalent glycosyl enzyme intermediate and ablation of the charge on the nucleophilic carboxyl, the pK
This sets the carboxyl group in its deprotonated form, enabling it to function as the base catalyst in the second step and deprotonate,
a
~16 for a sugar alcohol). Upon formation of the
a
of the acid/base residue drops.
1
1
in a concerted fashion, a water molecule that attacks the anomeric carbon. Beyond the well-established framework of this double-
1
7
displacement mechanism, a deep understanding of how the 10 fold rate enhancement is achieved, or how the pH optimum of a
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 14
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
glycoside hydrolase is set, remains to be fully established. However, computational approaches are contributing to addressing these
fundamental questions.
12,13
Central to this desired understanding is the nature of the interplay between the two key catalytic carboxylic acids, a phenomenon
that is challenging to study experimentally. In contrast to what is seen for spontaneous hydrolysis of glycosides in aqueous solution,
enzymatic cleavage proceeds through steps involving concerted proton transfers. In Bcx, the first step (glycosylation) is rate-limiting
1
4
M
in kcat and kcat/K , even for aryl glycosides. Thus the stability of the first transition state relative to the ES complex or free enzyme,
1
3
respectively, determines the observed rates of hydrolysis. As shown by NMR-monitored titrations of selectively C-labelled Bcx,
the protonation equilibria of Glu78 and Glu172 in the free enzyme are responsible for the observed profile of kcat/K vs pH, which is
M
1
1,15
bell-shaped with a maximum at pH 5.9.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1: Mechanism of retaining glycoside hydrolases. Hydrolysis of the glycosidic bond proceeds via two oxocarbenium
ion-like transition states, stabilized by two carboxylic acid/carboxylate residues. In Bcx, the first step is rate-limiting. Modifi-
cation of the 4-position of the catalytic nucleophile Glu78 (R
well as charge distribution at the transition states.
3
, R ) will therefore modulate initial protonation equilibria, as
4
The nature of transition state (ii) is not intuitive, since carboxylate anions are poor nucleophiles in water and the anomeric carbon
1
6
is likewise weakly electrophilic. Furthermore, proton transfer from a weak acid (pK
a
Glu172 = 6.7) to a weakly basic exocyclic
It is the interplay between the two catalytic processes that results in the observed
That is, the proton transfer enhances the electrophilicity of the anomeric carbon enough to render the nucle-
ophilic attack viable. Vice versa, the accumulating negative charge on the leaving exocyclic oxygen renders proton transfer from a
17–19
oxygen is thermodynamically unfavorable.
20,21
catalytic efficiency.
1
2,13
weak acid favorable, as is confirmed computationally.
At the core of both the pH-dependent activity and catalytic efficiency of Bcx lies the ability of Glu78 and Glu172 to stabilize
positive charge in the oxocarbenium-like transition states. The pK values of these residues largely determine these experimental
characteristics. It would therefore be extremely useful to be able to tune their pK values independently. Unfortunately the repertoire
of 20 common proteinogenic amino acids does not enable this through simple mutagenesis, though it has allowed probing of charge
a
1
1
a
2
1
placement. Unnatural amino acids provide a promising alternative, particularly 4-fluoro and 4,4-difluoro glutamic acid (Glu
F
and
GluFF respectively), since fluorine is the most electronegative element, yet small enough to minimize steric burden within the active
2
2
a
site. By analogy to fluoro- and difluoroacetic acid, 4-fluoro substituents in glutamic acid are expected to lower the pK value of the
sidechain carboxylic acid groups by roughly 2 and 3 units, respectively.
The specific introduction of Glu
acids can be introduced via hijacking of amber codons,
pair that can distinguish between glutamic acid and its 4-fluorinated derivatives (P. Schultz, Pers. Comm.). We have found that the
natural translation apparatus of E. coli incorporates Glu and GluFF at a significant level in place of Glu (unpublished results). Alt-
hough wild type Bcx contains only two Glu residues, this approach would still preclude individual modification of the catalytic
nucleophile or general acid/base residue. We therefore chose a semi-synthetic route to access Glu and GluFF ‘mutants’ at the position
F
and GluFF into the active site of Bcx is by no means trivial. Whereas fluorinated aromatic amino
2
3,24
it has thus far not proven possible to develop a tRNA/amino acyl synthetase
F
F
of the catalytic nucleophile of Bcx. This route involved introduction of the unnatural amino acid into a short synthetic peptide, fol-
lowed by native chemical ligation with a bacterially expressed counterpart to form a full-length functional enzyme containing the
2
5–27
desired modification.
While synthetic routes for both 4-fluoroglutamic acid28–30 and 4,4-difluoroglutamic acid31–33 have been established, to the best of
our knowledge there is no report of the site-selective incorporation of these unnatural amino acids into a peptide or an active enzyme.
In this manuscript we describe the methodology we developed to synthesize these molecules, as well as the kinetic characterization
of resultant Bcx constructs carrying 4-fluoro- and 4,4-difluoroglutamic acids as their catalytic nucleophile. These studies revealed
substantial effects of the modifications upon the pH-dependent activity of Bcx and upon stabilization of the oxocarbenium ion-like
transition states.
2
ACS Paragon Plus Environment

Page 3 of 14
1
Journal of the American Chemical Society
Experimental Section
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Results and Discussion
Circular permutant cpP75 allows semisynthesis of Glu78 variants.
Development of a strategy to introduce fluoroglutamic acids into position 78 of Bcx via native chemical ligation of a synthetic
peptide and a recombinant truncated protein required several limitations to be taken into account. Firstly, the synthetic peptide needs
to be < 15 residues. Even though longer peptides are often accessible via solid phase peptide synthesis, the very hydrophobic nature
of the residues flanking Glu78 and their inclination to form insoluble aggregates limits the size that can be reasonably assembled.
Indeed, these properties severely hampered the synthesis, purification and characterization of the peptides in this study. Consequently,
the site of modification needs to be close to one terminus of the final protein. Fortunately, Bcx is very tolerant towards circular
permutation, and a number of active "circular permutants" (cp) have been characterized. The newly generated termini of those
circular permutants primarily lie in the loops between the -strands of the -jellyroll fold of Bcx.
The second limitation arises from a fundamental requirement of native chemical ligation: The N-terminal component needs to enter
the ligation as a C-terminal thioester. Despite tremendous effort on our part, it did not prove possible to generate a thioester derivative
of any truncated, recombinantly expressed Bcx fragment since such species cannot adopt a native fold and are highly insoluble in
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
4
3
5
aqueous solution. This ruled out conventional approaches using intein systems. Various chemical approaches to the generation of
C-terminal thioesters or hydrazides
caused Bcx to either aggregate or hydrolyze.
3
6,37
failed also since the long reaction times (> 24 h), elevated temperatures or low pH conditions
Consequently, the synthetic peptide needed to be the N-terminal partner in the native chemical ligation reaction and the counterpart
expressed protein fragment needed to have an N-terminal cysteine that ends up proximal to the site of modification, Glu78, in the
final reconstructed circular permutant (Fig. 1; wt numbering is used throughout for consistency). We generated several circular per-
mutants with their N-termini in the loop near Glu78 and identified cpP75 (i.e., the new N-terminus is Pro75 and the new C-terminus
is Ser74, with Ala1-Trp185 linked via a diglycine) as the most promising candidate (data not shown). A suitable ligation site com-
patible with the size of the synthetic peptide was then required. Fortunately, Bcx has no cysteines, and thus the simplest approach
was to mutate a solvent-exposed threonine or serine. We chose Thr87 over Ser84, since the former is preceded by a Gly (facilitating
3
8
fast ligation), whereas the latter is preceded by Asp, which renders thioesterification and ligation of the peptide impossible.
Figure 1: Schematic representations (left, 3D model based on PDB entry 3LB9; right, sequence) of the Bcx construct designed to incorporate
an unnatural amino acid at the position of the catalytic nucleophile. The circular permutant Bcx cpP75 T87C (Bcx , cartoon) is formed by
ligation of a 12-mer synthetic peptide (green) containing Glu
cp
F
or GluFF (Glu78*, magenta) with a suitably truncated recombinant protein
(blue) carrying an N-terminal cysteine. Ligation site (Cys87, orange), N- and C-termini and catalytic residues are represented by stick models.
cp
wt
We recombinantly expressed the resultant control enzyme Bcx cpP75 T87C (henceforth abbreviated as Bcx ). Compared to Bcx ,
this protein was produced in E. coli at lower, albeit acceptable, expression yields and exhibited reduced thermal stability and catalytic
wt
efficiency (kcat/K
M
= 33% of Bcx , Table 1). Otherwise the variant behaved like the wild type enzyme in terms of long-term stability
3
4
and substrate specificity. These findings are in line with those on the previously characterized circular permutants. We therefore set
out to semi-synthesize this enzyme by ligating the peptide hydrazide P75LIEYYVVDSWG86-N
Bcx cpP75 T87C  (henceforth Bcx ).
2
H
3
with the recombinant fragment
cp
cp
Successful expression of the recombinant fragment Bcx in E. coli required iterative optimization of the first ten codons
by silent mutations (Supporting Information), This presumably prevented the 5’-end of the mRNA derived from the wt sequence from
forming stable secondary structures that impeded initiation of the translation process. The expression product was exclusively found
in the insoluble fraction of the cell lysate, as expected given that the truncated polypeptide is unable to adopt its native fold. The
insolubility of the product unfortunately prohibited its purification. Even though the truncation fragment could be solubilized in strong
denaturants such as 6 M guanidinium chloride or 8 M urea at concentrations up to 10 mg/mL, prolonged (> 24 h) handling in solution,
mechanical stress, or changes in ionic strength or pH resulted in its aggregation to form a transparent gel. We therefore decided to
work with the crude protein (~75% pure judged by SDS PAGE) contained in inclusion bodies that had been thoroughly washed to
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 14
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
remove cell debris. To characterize the resulting sample, we performed a digestion with Lys-C. Subsequent MALDI TOF analysis of
the digested product exclusively showed the expected mass of the N-terminal peptide C75YRPTGTYK among possible N-terminal
peptides. This indicated that the majority of the expression product contained the desired unmodified N-terminal cysteine residue.
We synthesized the synthetic peptide hydrazide 3 (Scheme 2), which can be converted into an active thioester in situ, by using
39
standard Fmoc strategy and a hydrazinyl carboxylate-modified 4- hydroxybenzyl (PHB) resin. The peptide hydrazide was converted
into a transient mercaptophenyl acetic acid (MPAA) thioester by oxidation and subsequent addition to a solution containing MPAA
cp
as a thiol catalyst in 6 M guanidinium chloride buffer. Even though the ligation with Bcx was rapid, conversion halted after
1-2 hours reaction time at ~ 50% of the protein precursor in all our experiments, judged by comparing the intensity of the bands on
the gel. Possibly the N-terminus is blocked in a significant portion of the polypeptides, either by chemical modification or aggregation.
Since separation of unfolded product and the remaining precursor is impractical, we directly subjected the reaction mixture to refold-
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
0
ing. Our reasoning was that only the ligation product would be able to fold correctly and remain in solution, whereas any nonreactive
protein aggregates would precipitate and can be removed by centrifugation. After desalting and concentration of the refolded product,
cation-exchange chromatography yielded 300 g of pure, soluble active enzyme. This was an 11% yield relative to the protein frag-
cp
ment used. As expected, the semi-synthetic Bcx exhibited the same characteristics and catalytic properties as its recombinant coun-
terpart (data not shown). Having established the applicability of our semi-synthetic approach, we moved on to introduce the fluo-
roglutamic acids at position 78.
Synthesis of optically pure fluoroglutamic acid building blocks.
The optimized large-scale synthesis of orthogonally protected, optically pure N-Cbz-OtBu-L-4,4-difluoroglutamic acid from chlo-
3
3
rodifluoroacetic acid published by Miller et al. proved straightforward in our hands, with an overall yield of 11%. Exchange of the
amine protection group (Cbz) for Fmoc to produce building block 2 was accomplished using standard protocols (Supporting Infor-
mation).
Unprotected 4-fluoroglutamic acid is commercially available as a mixture of all four stereoisomers from Synthequest Labs (Fig.
S-1), and thus we used this as a precursor for building block 1. The orthogonal protection and kinetic resolution of this mix presented
two possible complications that were not an issue for 2: the additional stereocenter and the chemoselective protection of the side-
chain carboxylate, which was not expected to be as well discriminated from the -carboxy group in 1. However, to our surprise, the
methodology of orthogonal protection developed by Miller et al. for the 4,4-difluoro derivative was applicable to the 4-fluoroglutamic
acid with only minor modifications (Supporting Information). Furthermore, during kinetic resolution, the B. subtilis protease em-
ployed to selectively hydrolyze the -methyl ester of the L-isomers in the mixture was also found to prefer the L-erythro over the L-
threo isomer (Fig. S-3). This allowed isolation of the optically pure Cbz-protected precursor, which was reprotected with Fmoc to
yield 1. The integrity of the L-configuration in 1 and 2 was confirmed by modification with Marfey’s Reagent and subsequent HPLC
analysis and found to be 94 ee,, 99 de for 1 and 92 ee for 2, respectively.
Scheme 2: Fluoroglutamic acid building blocks and synthetic peptides prepared for assembly of the modified enzymes. Opti-
cally pure and orthogonally protected threo-L-4-fluoroglutamic acid (1) and L-4,4-difluoroglutamic acid (2) were introduced
into the 12-mer peptide hydrazides (4,5) by Fmoc-based solid phase peptide synthesis.
Synthesis and characterization of fluoroglutamic acid-containing peptides.
Successful solid phase peptide synthesis of the fluoroglutamic acid-containing peptide hydrazides 4 and 5 required modifications
to the standard procedure used to obtain 3. Of the two building blocks, only 1 tolerated the standard coupling and Fmoc deprotection
conditions without notable side reactions. In contrast, the 4,4-difluoro glutamic acid building block 2 exhibited a number of fatal side
reactions, most notably fast formation of a pyroglutamate during deprotection with piperidine, DBU and similar reagents that are
commonly used for Fmoc deprotection (Supporting Scheme S-2). Efficient elongation of 4,4-difluoroglutamic acid could only be
achieved by use of the milder base piperazine, supplemented with 0.1 M HOBt as acidic modifier, and keeping deprotection times
under 5 min with immediate coupling of the subsequent residue. This protocol yielded peptide 5 of a satisfactory purity (Supporting
4
ACS Paragon Plus Environment

Page 5 of 14
1
Journal of the American Chemical Society
Fig. S-5). However, it should be noted that we found it impossible to introduce 2 into a peptide bearing a Gly residue next to GluFF
as is the case with the sequence around the Bcx general acid/base catalyst, Glu172). In this case glutarimide formation during solid
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
phase peptide synthesis diminished the yield. Even when glutarimide formation was prevented by tedious introduction of dimethox-
ybenzyl (Dmb) backbone protection of the Gly amide, the peptide cleaved at various backbone amides around GluFF during depro-
tection in TFA solution (Supporting Fig. S-9). With the bulky, hydrophobic residues flanking Glu78 however, no such side reactions
were observed for peptide 5.
Modifications to our method were also needed to generate the C-terminal hydrazide. We had found that 2-3 rounds of coupling
with 5 eq. Fmoc amino acid per residue were necessary to obtain 3 in satisfactory purity, probably due to the noted hydrophobicity
of the sequence. However, such large excesses would have been too wasteful given our limited supply of building blocks 1 and 2.
Consequently we moved to the use of a TentaGel resin with low (0.2 mmol/g) substitution to facilitate synthesis. Unexpectedly, the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
9
well-established hydrazinyl-carboxylate-PHB linker used to generate a C-terminal hydrazide proved to be unstable on this resin,
resulting in very low yields and a large share of carboxy-terminated product. We therefore resorted to a protocol recently described
4
1
by Bello et al. in which a classic carboxyl-linked peptide is assembled on a TentaGel PHB resin. The protected, resin-linked peptide
is then treated with hydrazine solution, resulting in selective cleavage of the linking ester to form a C-terminal hydrazide. In our
hands, treatment with anhydrous hydrazine in THF resulted in a cleavage yield of > 80%. Fortunately, when treatment times were
kept short (< 12 h) and fresh anhydrous reagent was used, no hydrazinolysis of the highly electrophilic sidechain ester of 4,4-glutamic
acid was observed. Peptide hydrazides 4 and 5 were purified by reversed-phase HPLC and their identities confirmed by MALDI TOF
1
MS and homonuclear NOESY and TOCSY H-NMR spectroscopy.
F
Semisynthesis of Bcx cpP75 Glu 78 and GluFF78.
cp
We subjected peptide hydrazides 4 and 5 to native chemical ligation with Bcx in the same fashion as with peptide 3. The
two peptides could only be dissolved up to 0.5 mM in the 6 M guanidinium chloride buffer at pH 3.5 required to convert the hydrazide
to a thioester. The ligation nonetheless proceeded rapidly, terminating in less than one hour (Fig. 2A). However, again for unknown
reasons, only ~ 30% of the precursor protein was consumed as judged by SDS PAGE, even though the peptide was provided in 20-
fold excess over the protein fragment. The reaction solution was subjected to refolding after 6 h and the folded enzymes were isolated
40
and purified as described previously. Following this protocol, 50 g of each purified enzyme was obtained per synthesis. The
cp
cp
F
identities of Bcx Glu 78 and Bcx GluFF78 were confirmed by high resolution ESI MS of the intact proteins (Figure 2B and C) and
activity confirmed by measurement using the substrate 2,5-dinitrophenyl -xylobioside.
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 14
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2: Protein semisynthesis and characterization of Bcx cpP75 Glu
chemical ligation monitored by SDS PAGE. Intact protein ESI MS spectra of the refolded and purified products carrying (B) Glu 78 (calc.
F
F
78 T87C and Bcx cpP75 GluFF78 T87C. (A) Progress of the native
2
0,529, found 20,529 and (C) GluFF78 (calc. 20,547, found 20,547). Additional signals in the expanded insets are attributed to methionine
oxidation.
F
Bcx cpP75 Glu 78 and Bcx cpP75GluFF78 show a shift in pH-dependent activities.
The first step in the kinetic characterization of these novel forms of the enzyme was to determine the impact of the substitution of
Glu78 by Glu and GluFF on pH-dependent activity. Since the modifications lower the pKa values of the fluorinated carboxylic acid
they therefore should reduce the share of protonated species at a given pH value, potentially changing the bell-shaped pH vs. activity
profile. To explore this the pH profiles (Figure 3) were derived by measuring the second order rate constants (kcat/K ) for hydrolysis
of the fluorogenic substrate 6-chloro-4-methylumbelliferyl -xylobioside (ClMU-X ) at 13 pH values. In control experiments, we
confirmed that all three enzymes were stable over the assay time used at each pH value (data not shown). As expected, the pH profile
F
M
2
cp
wt
of Bcx carrying Glu as nucleophilic residue was essentially identical to that of Bcx , with a maximum at pH 6, a pK
attributed to the nucleophile (left hand inflection point of Figure 3A) and a pK value of 7.1 attributed to the acid/base (Glu172). In
comparison, the pH profiles of Glu 78 (Figure 3B) and GluFF78 (Figure 3C) exhibited strong shifts in maxima and both inflection
points towards more acidic pH values, with the difluoro variant undergoing slightly greater perturbations. At least for the nucleophile,
this is expected and consistent with the relative sidechain pK values of free Glu, Glu and GluFF. However, it is quite possible that
a
value of 4.7
a
F
a
F
some other ionization equilibria within the proteins at these relatively acidic conditions resulted in the observed pH-profile changes.
6
ACS Paragon Plus Environment

Page 7 of 14
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3: pH-activity profiles for hydrolysis of ClMU-X
C). Inflection points and maxima resulting from fitting a double-ionization model to the profiles (red) are indicated at the respective positions.
by Bcxcp in its unmodified (Glu78, A) and modified forms (Glu
78, B and GluFF78,
2 F
Interestingly, the pK value attributed to the unmodified acid/base residue Glu172 (right hand inflection points in Figure 3) also
a
shifted to more acidic values in the fluorinated enzymes. This was not expected. However, as seen in previous measurements of active
1
1
site residue pK
alter the interactions between these residues, and thus the pK
tional perturbations of a tightly linked network of active site residues. For example, the ablation of the charge on Glu78 by mutation
a
values in Bcx, the ionizations are closely coupled. Consequently modification of the acidity of Glu78 is certain to
a
value of Glu172, either electrostatically or through subtle conforma-
1
1
to Gln reduces the pK
a
of Glu172 by 2.5 units. Thus a drop of approximately one unit in response to a reduction in partial negative
charge on Glu78 upon fluorination is perhaps quite reasonable.
Introduction of Glu
The overall catalytic efficiencies (kcat/K
terpart at their respective pH maxima. This raises the question of which step is now rate-limiting in the two cases. As noted earlier,
F
and GluFF at position 78 does not alter the rate-determining step.
M
) of the fluorinated enzymes are ~25-40 times lower than that of their unfluorinated coun-
4
2
for the wild type enzyme, the glycosylation step is rate-limiting in kcat for aryl glycoside substrates. Further, it is generally expected
that glycosylation is rate-limiting in kcat/K for most retaining glycosidases, including Bcx since this represents the first irreversible
M
2
1
step. In theory, one might expect that the glycosylation step should remain rate-limiting after lowering the pK
a
value of Glu78 by
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 14
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
fluorination, since this reduces nucleophilicity. By the same token, the deglycosylation step should be accelerated since the fluoroglu-
tamic acid will be a better leaving group than the natural residue.
To test whether this general trend is indeed followed, we studied a reaction in which the two steps are clearly separated. This
reaction involved the formation and then the hydrolysis of the 2-fluoroxylobiosyl-enzyme using the covalent probe 2,4-dinitrophenyl
2
2
-deoxy 2-fluoro--xylobioside (2,4DNP-2FX ). In this probe the electron-withdrawing fluorine at the 2-position of the proximal
xylose destabilizes both oxocarbenium-ion-like transition states, slowing both steps. However, the excellent leaving group DNP en-
sures that the glycosylation step is fast enough for the intermediate to accumulate in Bcx . If our prediction is correct, inactivation
wt 15
kinetics should be markedly slowed in the Glu
(reactivation) should be faster.
F
78 and, even more so, the GluFF78 variants, while turnover of the covalent intermediate
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4: Inactivation of the control enzyme (Bcxcp with Glu78) and the two modified enzymes by 5 mM 2,4DNP-2FX
As shown in Figure 4, inactivation was indeed 20 times slower for the monofluoro glutamic acid variant with Glu
.
2
F
78 than for the
Glu78 Bcx . No reaction was seen over 30 minutes with the GluFF78 enzyme. The 20-fold drop in inactivation rate constant is
cp
consistent with the ~ 40-fold lower kcat/K
M
value for ClMU-X
2
hydrolysis by the Glu
F
78 variant versus the Glu variant (Fig. 3) and
confirms a slower glycosylation step.
wt
-1
Reactivation of Bcx (the deglycosylation step) occurs with a rate constant of kreact = 0.0021 min , corresponding to a half-life for
1
5
F
the 2-fluoroxylobiosyl enzyme of t1/2 = 330 min. We therefore attempted to measure the rate of reactivation for the Glu 78 enzyme,
which could be partially inactivated. We first incubated it with a high inactivator concentration (5 mM) for a prolonged time (6 h),
and then removed the free inactivator from the enzyme by rapid filtration on a spin filter (MWCO 10,000) at low temperature (4 °C).
However, by the time this had been completed, the enzyme had regained activity. Although this precluded the opportunity to measure
reactivation kinetics, this observation demonstrates that the half-life for reactivation was considerably less than the 330 min. measured
wt
for Bcx . These studies therefore confirmed predictions concerning the effects of fluorination on the rates of the individual steps.
This set the stage for further kinetic studies to probe the structure of the glycosylation transition state.
a
Lowering the pK value of Glu78 breaks the concerted nature of catalysis in the glycosylation step.
To probe the effects of nucleophile fluorination on the structure of the glycosylation transition state we measured kinetic parameters
for hydrolysis of a set of aryl xylobioside substrates with aglycones of differing leaving group ability. This allowed us to vary both
the “pull” of the leaving group and the “push” of the nucleophile, and to do so at a pH that is optimal for each enzyme form. We
carried out these studies using the seven -1,4-linked xylobioside substrates shown in Scheme 3, for which the conjugate acids of the
a a
aglycones cover a pK range of 4.2 to 7.8 (Table 1). Since the pK values of the acid/base catalyst Glu172 lie between 6 and 7 in our
enzyme variants (Figure 3) we used four substrates (6-9) that hydrolyze without general acid catalysis (because the leaving group pK
is lower than that of the general acid). In contrast, three substrates (10-12) potentially require general acid assistance.
a
8
ACS Paragon Plus Environment

Page 9 of 14
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 3: The aryl xylobioside substrates used in kinetic characterization of Bcx constructs modified at position 78 by fluo-
roglutamic acid derivatives.
Table 1: Measured kinetic constants for Bcx constructs carrying either Glu, Glu
F
or GluFF at position 78. Assays were con-
78) and 4.5 (Bcx GluFF78). Standard deviation of the last signif-
values far above their solubility limit, and therefore individual,
isolated kinetic constants could not be determined (n.d.). However, second-order rate constants could be obtained at [S] ~
.1K
cp
cp
cp
ducted at 30 °C and at pH 6.0 (Bcx Glu78), 5.0 (Bcx Glu
icant digit is given in brackets. *Some substrates exhibit K
F
M
0
M
.
Substrate
Position
78
K
M
k
cat
k
cat/K
M
-
1
-1 -1
Aglycone pK
a
(mM)
(s )
(mM s )
6 (2,4-DNP)
Glu
Glu
GluFF
1.0(0)
2.2(3)
1.5(1)
14(0)
1.2(1)
1.8(1)
13.9(5)
0.52(7)
1.2(1)
4.2
F
7
(F
4
2
-MU)
.7
Glu
Glu
F
GluFF
21(2)
>20
>20
n.d.
n.d.
n.d.
2.9*
2.2*
2.2*
8
(2,5-DNP)
.2
Glu
0.85(9)
0.9(2)
1.1(2)
7.7(5)
0.19(1)
0.20(1)
9.0(15)
0.22(6)
0.18(3)
5
Glu
F
GluFF
9
(Cl-MU)
.0
Glu
1.7(3)
1.4(3)
0.83(6)
9.8(7)
0.11(1)
0.060(1)
5.8(9)
0.08(1)
0.072(5)
6
Glu
F
GluFF
10 (CF
3
-MU)
Glu
1.3(2)
3.1(2)
3.5(2)
0.55(4)
0.30(1)
0.30(1)
0.42(7)
0.097(7)
0.086(5)
7.0
Glu
F
GluFF
1
1 (pNP)
Glu
9.6
(85) n.d.
2.2*
7
.2
Glu
GluFF
F
>20
>20
n.d.
n.d.
0.17*
0.063*
1
2 (MU)
.8
Glu
11(2)
9.2(16)
27(6)
2.5(3)
0.55(7)
1.2(3)
0.090*
0.068*
0.052*
7
Glu
F
GluFF
The steady state kinetic data shown in Table 1 revealed that, for each individual substrate, similar K
M
values were measured for all
cp
three Bcx forms. Thus, the introduced fluorine modifications of Glu78 did not interfere significantly with substrate binding. The
same is not true when comparing the K values for each individual enzyme with the seven substrates. Rather, these values fell into
two regimes, with the majority around ~2 mM, while the K
M
M
values of 7, 11 and 12 were >10 mM. This suggested that the aglycone
9
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 14
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
m a
structure affected the binding behavior significantly. As a consequence, Brønsted plots of log kcat or log kcat/K versus aglycone pK ,
while showing correlations (Figure 6A), have considerable scatter that limited their utility or at least masked clear interpretation
wt 14
(
Supporting Fig. S-10). Similar scatter was also seen in the previous such study on Bcx .
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5: Selected Brønsted plots illustrating varying sensitivity to the nucleophile pK
pK
values in parentheses and fit slopes (NUC) indicated). Left-to-right, the data points in each plot correspond to Bcxcp with GluFF, Glu
and Glu at position 78.
a
value versus the quality of the leaving group (aglycone
a
F
Alternatively, it is possible to inspect the data from the perspective of changes in the pK
a
value of the nucleophile at position 78.
The sterically minimal changes in nucleophile structure do not seem to affect substrate binding, as indicated by the consistent K
values across Bcx variants processing the same substrate. Assuming that the catalytically optimal structure of the active site is
M
cp
likewise undisturbed, we can compare kinetic parameters across these three enzymes with each substrate individually and sidestep
the above issues. Accordingly, we present a set of six 3-point Brønsted plots of log kcat/K versus nucleophile pK value for each
M a
substrate (Figure 5). The slopes of these plots (NUCreflect the sensitivity of the reaction to the nucleophile. This in turn relates to
the extent of interaction of the nucleophile with the anomeric center at the transition state of the glycosylation step. Indeed, the plots
of the nuc values versus substrate leaving group pK
a
have an interesting biphasic form, peaking around an aglycone pK
a
of 6 (Fig.
6
B). This tells us that for substrates of low or high aglycone pK
a
values, there is little transition state bond formation between the
a
nucleophile and the sugar anomeric center. In contrast, for substrates with aglycone pK values around 6, the interaction is substantial.
1
0
ACS Paragon Plus Environment

Page 11 of 14
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6: (A) Brønsted plot for Bcxcp Glu
from the single-step mechanism (A
Fig. 5 plots of kcat/K versus nucleophile pK
leaving group pK values as reported for human O-GlcNAcase , showing the same trend in decreasing leaving group sensitivity above pK
, where general acid catalysis becomes viable for this enzyme.
Thus substrates, such as 7, 8 and 9, that do not require general acid assistance show a progressive need for more nucleophilic push
larger NUC) as the aglycone pK increases (poorer leaving group). This observation is in line with a single-step mechanism involving
an anionic nucleophile displacing an anionic leaving group (A , Fig. S-11A). Conversely, substrates such as 10 and 12 that require
acid catalysis, show decreasing sensitivity towards the nucleophile pK value. Interestingly (although cautiously given the sparse
data) this latter phase corresponds to one in which kcat values for the fluorinated enzymes rise with increasing aglycone pK (Figure
A and Fig S-10). Such a response implies positive charge accumulation on the glycosidic oxygen at the reaction transition state since
78 reporting a break in kcat with leaving group (LG) sensitivity LG. This likely indicates a shift
F
‡
N
D
N
) to a dissociative mechanism DN *A
N
(see Fig S-10 for all plots). (B) NUC values derived from the
. (C) NUC values for the same range of
M
a
value for six substrates with differing aglycone pK
a
4
4
a
a
7
(
a
N N
D
a
a
6
relatively electron-donating substituents increase rates. This observation probably reports on almost complete protonation of the
glycosidic oxygen at the transition state, with partial fission of the glycosidic bond (Fig. S-11D). In this dissociative mechanism
‡
DN *A
N
., the nucleophile attacks after the rate-limiting protonation/dissociation, hence the level of concertedness between proton
transfer and nucleophilic attack is decreased. Similar behavior was seen in the spontaneous hydrolysis of electron rich aryl glycosides
where protonation is facilitated by electron donating substituents on the leaving group.
43
Thus, it appears that, for our fluorinated enzymes, during enzymatic cleavage of substrates with low aglycone pK
leaving groups), the rate-limiting glycosylation reaction occurs without proton donation from the acid/base catalyst and with minimal
participation from the enzymatic nucleophile. As the aglycone pK increases towards 6 (thus leaving group ability decreases) nucle-
ophilic participation becomes increasingly important. For substrates with aglycone pK values above 6, acid catalysis can start to
contribute to transition state stabilization and consequently protonation of the glycosidic oxygen runs ahead of glycosidic bond cleav-
age at the transition state. This is reflected in increasing kcat values with higher aglycone pK values, and indeed this pre-protonation
a
values (good
a
a
a
increasingly dominates, with correspondingly decreasing nucleophilic participation (Figure 6A).
Due to the substrate-inherent binding effects that obscure the data (Fig. S-10) these interpretations have to be made with care, even
if the observed values of NUC and LG are high enough (>0.2) to render any pre-chemical step (like sugar ring distortion) reporting
4
4
on kinetic constants unlikely. Our findings are also very much in line with a similar study conducted by the Vocadlo group . They
demonstrated that the retaining -N-acetylglucosaminidase OGA acts via a substrate-assisted mechanism in which the neutral car-
bonyl of the acetamide moiety acts as the nucleophilic catalyst, forming an oxazoline intermediate. In this system the strength of the
1
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 14
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
nucleophile could be modulated by introduction of fluorines into the amide moiety of the substrate, thereby avoiding laborious mod-
ification of the enzyme. The authors observed several regimes in their Brønsted plots. Substrates with good leaving groups (general
acid catalysis not required) showed increased sensitivity towards the strength of the nucleophile with decreasing leaving group ability
up to the point at which acid catalysis was effective (Figure 6C). Beyond that, in the regime of acid-catalyzed leaving group departure,
the same decreasing sensitivity to the strength of the nucleophile with increasing leaving group pK was observed. The authors like-
a
wise interpreted this observation as evidence for the formation of a protonated intermediate becoming rate-limiting.
Presumably, the active site of the enzyme has evolved to maximize catalysis of the cleavage of unactivated disaccharide linkages.
Accordingly, both acid catalysis and nucleophilic catalysis have been honed to minimize localized charge development. This is
achieved by effecting nucleophilic and acid catalysis in a highly concerted fashion, with bond formation from the nucleophile to the
anomeric center occurring simultaneously with glycosidic bond cleavage to minimize positive charge development on the sugar ring.
Meanwhile proton donation from the acid catalyst occurs with glycosidic bond cleavage to minimize negative charge development
on the glycosidic oxygen. By dampening nucleophilic catalysis, we have forced the concerted mechanism to become more asynchro-
nous and dissociative with little nucleophilic participation, thereby exposing these rate dependencies on substrate reactivity.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Conclusions
In this study we report, for the first time, the site-specific incorporation of fluorinated glutamic acid residues into the active site of
an enzyme. Beyond the confirmation of the expected effects on pH-dependent activity, this study has provided unique insights into
the fine tuning of nucleophilicity of an anionic nucleophile within an enzyme active site in a sterically and structurally conservative
manner. We have shown that the highly concerted nature of the mechanism that has evolved in the wild type enzyme to minimize
local charge development on the substrate is achieved by matching the enzymatic nucleophilicity with the acid/base properties of the
responsible residue.
ASSOCIATED CONTENT
Supporting Information. Experimental procedures, additional analytical data. This material is available free of charge via the Internet at
http://pubs.acs.org.”
AUTHOR INFORMATION
Corresponding Author
*Stephen G. Withers, Department of Chemistry, University of British Columbia, 2036 Main Mall, Vancouver, B.C., Canada V6T 1Z1;
withers@chem.ubc.ca
ACKNOWLEDGMENT
We thank Dr. Lars Baumann and Dr. Ian Greig for helpful discussions. We acknowledge financial support from the Natural Sciences and
Engineering Research Council of Canada (to L.P.M. and S.G.W.) and a fellowship from the Deutsche Forschungsgemeinschaft (to M.P.K.).
Instrument support was provided by the Canada Foundation for Innovation and the British Columbia Knowledge Development Fund.
ABBREVIATIONS
cp
wt
MPAA, mercapto phenylacetic acid; Bcx, Bacillus circulans endo -1,4-xylanase; cp, circular permutant, Bcx , Bcx cpP75 T87C; Bcx ,
wild type Bcx; DBU, 1,8-Diazabicyclo(5.4.0)undec-7-ene; Glu , 4-fluoroglutamic acid; GluFF, 4,4-difluoroglutamic acid; MPAA, mercapto
phenylacetic acid; TFA, trifluoro acetic acid; ClMU 6-chloro-4-methylumbelliferyl; DNP-2FX , 2,4-dinitrophenyl 2-deoxy 2-fluoro xylo-
bioside; HOBt, 1-hydroxybenzotriazole; LG, leaving group; NUC, nucleophile, PHB, 4- hydroxybenzyl alcohol; Dmb, dimethoxybenzyl;
, -xylobioside; DNP, dinitrophenyl; MU, 4-methylumbelliferyl
F
,
2
X
2
REFERENCES
(1) Wolfenden, R. Annu. Rev. Biochem. 2011, 80, 645–667.
(2) Beg, Q.; Kapoor, M.; Mahajan, L.; Hoondal, G. Appl. Microbiol. Biotechnol. 2001, 56, 326–338.
(3) Gírio, F. M.; Fonseca, C.; Carvalheiro, F.; Duarte, L. C.; Marques, S.; Bogel-Łukasik, R. Bioresour. Technol. 2010, 101, 4775–
4800.
4) Juturu, V.; Wu, J. C. Biotechnol. Adv. 2012, 30, 1219–1227.
5) Sweeney, M. D.; Xu, F. Biomass Converting Enzymes as Industrial Biocatalysts for Fuels and Chemicals: Recent Develop-
ments; Catalysts 2, 2012.
6) Dumon, C.; Varvak, A.; Wall, M. A.; Flint, J. E.; Lewis, R. J.; Lakey, J. H.; Morland, C.; Luginbühl, P.; Healey, S.; Todaro,
T.; DeSantis, G.; Sun, M.; Parra-Gessert, L.; Tan, X.; Weiner, D. P.; Gilbert, H. J. J. Biol. Chem. 2008, 283, 22557–22564.
(
(
(
(
(
(
7) Qian, C.; Liu, N.; Yan, X.; Wang, Q.; Zhou, Z.; Wang, Q. Enzyme Microb. Technol. 2015, 70, 35–41.
8) Koshland, D. E. Biol. Rev. 1953, 28, 416–436.
9) Vocadlo, D. J.; Davies, G. J.; Laine, R.; Withers, S. G. Nature 2001, 412, 835 EP -.
(10) Davies, G. J.; Mackenzie, L.; Varrot, A.; Dauter, M.; Brzozowski, A. M.; Schülein, M.; Withers, S. G. Biochemistry 1998, 37,
1707–11713.
1
1
2
ACS Paragon Plus Environment

Page 13 of 14
1
Journal of the American Chemical Society
(
11) McIntosh, L. P.; Hand, G.; Johnson, P. E.; Joshi, M. D.; Körner, M.; Plesniak, L. A.; Ziser, L.; Wakarchuk, W. W.; Withers,
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
S. G. Biochemistry 1996, 35, 9958–9966.
(
12) Raich, L.; Borodkin, V.; Fang, W.; Castro-López, J.; van Aalten, D. M. F.; Hurtado-Guerrero, R.; Rovira, C. J. Am. Chem.
Soc. 2016, 138, 3325–3332.
13) Knott, B. C.; Haddad Momeni, M.; Crowley, M. F.; Mackenzie, L. F.; Götz, A. W.; Sandgren, M.; Withers, S. G.; Ståhlberg,
J.; Beckham, G. T. J. Am. Chem. Soc. 2014, 136, 321–329.
(
(14) Lawson, S. L.; Wakarchuk, W. W.; Withers, S. G. Biochemistry 1997, 36, 2257–2265.
(15) Sidhu, G.; Withers, S. G.; Nguyen, N. T.; McIntosh, L. P.; Ziser, L.; Brayer, G. D. Biochemistry 1999, 38, 5346–5354.
(16) Whitfield, D. M. Carbohydr. Res. 2015, 403, 69–89.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(17) Hartwell, E.; Hodgson, D. R. W.; Kirby, A. J. J. Am. Chem. Soc. 2000, 122, 9326–9327.
(
(
(
(
(
18) Kirby, A. J. Acc. Chem. Res. 1997, 30, 290–296.
19) Shan, S.-o.; Herschlag, D. Proc. Natl. Acad. Sci. U. S. A. 1996, 93, 14474–14479.
20) Dean, K. E. S.; Kirby, A. J. J. Chem. Soc., Perkin Trans. 2 2002, 428–432.
21) Zechel, D. L.; Withers, S. G. Curr. Opin. Chem. Biol. 2001, 5, 643–649.
22) Berger, A. A.; Völler, J.-S.; Budisa, N.; Koksch, B. Acc. Chem. Res. 2017, 50, 2093–2103.
(23) Wang, L.; Brock, A.; Herberich, B.; Schultz, P. G. Science 2001, 292, 498.
(
(
(
(
(
24) Wang, L.; Schultz, P. G. Angew. Chem. Int. Ed. 2005, 44, 34–66.
25) Hackeng, T. M.; Griffin, J. H.; Dawson, P. E. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 10068–10073.
26) Humphrey, J. M.; Chamberlin, A. R. Chem. Rev. 1997, 97, 2243–2266.
27) Muir, T. W. Annu. Rev. Biochem. 2003, 72, 249–289.
28) Kokuryo, Y.; Nakatani, T.; Kobayashi, K.; Tamura, Y.; Kawada, K.; Ohtani, M. Tetrahedron: Asymmetry 1996, 7, 3545–
3
551.
(29) Konas, D. W.; Coward, J. K. J. Org. Chem. 2001, 66, 8831–8842.
(30) Qu, W.; Zha, Z.; Ploessl, K.; Lieberman, B. P.; Zhu, L.; Wise, D. R.; B. Thompson, C.; Kung, H. F. J. Am. Chem. Soc. 2011,
133, 1122–1133.
(31) Ding, Y.; Wang, J.; Abboud, K. A.; Xu, Y.; Dolbier, W. R.; Richards, N. G. J. J. Org. Chem. 2001, 66, 6381–6388.
(
(
(
32) Konas, D. W.; Coward, J. K. Org. Lett. 1999, 1, 2105–2107.
33) Li, Y.; Miller, S. J. J. Org. Chem. 2011, 76, 9785–9791.
34) Reitinger, S.; Yu, Y.; Wicki, J.; Ludwiczek, M.; D’Angelo, I.; Baturin, S.; Okon, M.; Strynadka, N. C. J.; Lutz, S.; Withers,
S. G.; McIntosh, L. P. Biochemistry 2010, 49, 2464–2474.
(35) Mootz, H. D. ChemBioChem 2009, 10, 2579–2589.
(36) Adams, A. L.; Cowper, B.; Morgan, R. E.; Premdjee, B.; Caddick, S.; Macmillan, D. Angew. Chem. Int. Ed. 2013, 52, 13062–
1
3066.
(37) Kang, J.; Macmillan, D. Org. Biomol. Chem. 2010, 8, 1993–2002.
(38) Johnson, E. C. B.; Kent, S. B. H. J. Am. Chem. Soc. 2006, 128, 6640–6646.
(
(
(
(
(
39) Fang, G.-M.; Li, Y.-M.; Shen, F.; Huang, Y.-C.; Li, J.-B.; Lin, Y.; Cui, H.-K.; Liu, L. Angew. Chem. 2011, 123, 7787–7791.
40) Kötzler, M. P.; McIntosh, L. P.; Withers, S. G. Protein Sci. 2017, 26, 1555–1563.
41) Bello, C.; Kikul, F.; Becker, C. F.W. J. Pept. Sci. 2015, 21, 201–207.
42) Lawson, S. L.; Wakarchuk, W. W.; Withers, S. G. Biochemistry 1996, 35, 10110–10118.
43) Nath, R. L.; Rydon, H. N. Biochem. J. 1954, 57, 1.
(44) Greig, I. R.; Macauley, M. S.; Williams, I. H.; Vocadlo, D. J. J. Am. Chem. Soc. 2009, 131, 13415–13422.
1
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 14
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table of Contents Graphic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
4
ACS Paragon Plus Environment