Retaining LeLoir glycosyltransferases catalyze the formation of glycosidic bonds between nucleotide sugar donors and carbohydrate acceptors. The anomeric selectivity of trehalose transferase from Thermoproteus uzoniensis was investigated for both d- and l-glycopyranose acceptors. The enzyme couples a wide range of carbohydrates, yielding trehalose analogues with conversion and enantioselectivity of >98%. The anomeric selectivity inverts from α,α-(1 → 1)-glycosidic bonds for d-glycopyranose acceptors to α,β-(1 → 1)-glycosidic bonds for l-glycopyranose acceptors, while (S)-selectivity was retained for both types of sugar acceptors. Comparison of protein crystal structures of trehalose transferase in complex with α,α-trehalose and an unnatural α,β-trehalose analogue highlighted the mechanistic rationale for the observed inversion of anomeric selectivity.
Retaining LeLoir glycosyltransferases catalyze the formation of glycosidic bonds between nucleotide sugar donors and carbohydrate acceptors. The anomeric selectivity of trehalose transferase from Thermoproteus uzoniensis was investigated for both d- and l-glycopyranose acceptors. The enzyme couples a wide range of carbohydrates, yielding trehalose analogues with conversion and enantioselectivity of >98%. The anomeric selectivity inverts from α,α-(1 → 1)-glycosidic bonds for d-glycopyranose acceptors to α,β-(1 → 1)-glycosidic bonds for l-glycopyranose acceptors, while (S)-selectivity was retained for both types of sugar acceptors. Comparison of protein crystal structures of trehalose transferase in complex with α,α-trehalose and an unnatural α,β-trehalose analogue highlighted the mechanistic rationale for the observed inversion of anomeric selectivity.
The synthesis of a glycosidic
bond is one of the most important reactions within glycochemistry.
Enzymes couple sugars to afford oligosaccharides with high selectivity
under mild reaction conditions. As enzymes are regarded to have evolved
toward the selective conversion of the naturally more abundant d-sugars, their l-sugar enantiomers are often not considered
as suitable substrates for enzymatic conversions. For this reason,
the coupling of both d- and l-glycopyranose acceptors
has rarely been compared for a single enzyme. However, the incorporation
of l-sugars offers a broad spectrum of diametrically opposed
glycosides or oligosaccharides, which might display new biological
activities.In one example, a retaining non-LeLoir glycosyltransferase
(GT)
coupled l-glycopyranose acceptors with sucrose as sugar donor
in an α,β-(1 → 2)-fashion, while α,α-(1
→ 2)-glycosidic bonds were observed with d-glycopyranose
acceptors.[1] The switch of anomeric selectivity
for the sugar acceptor was attributed to the 4C1 and 1C4 chair configuration for d- and l-glycopyranoses, which affect the position of the
nucleophilic hydroxyl group at the anomeric position. In general,
α-d and β-l anomers of the same sugar
are structurally more alike (Figure a), than the corresponding α-d and α-l anomers (Figure b).[2] This structural similarity allows
the conversion of both α-d and β-l configured
substrates by an (S)-selective enzyme.
Figure 1
Position of
the anomeric hydroxyl of α-d-glucopyranose
(4C1, cyan) when superimposed to β-l-glucopyranose (1C4, green) in panel
a is more similar than α-l-glucopyranose (1C4, purple) in panel b. The insets in panel a, b show
the representative overlay of the Newman projection of the anomeric
OH1. The SNi-like reaction mechanism with an oxocarbenium
transition state (4H3) that allows the approach
of the anomeric hydroxyl, α–OH of d-glucose
(4C1) in panel c, or the β–OH of l-glucose (1C4) in panel d, is guided
by hydrogen bonding from the same face as the NDP leaving group. R
= NDP.
Position of
the anomeric hydroxyl of α-d-glucopyranose
(4C1, cyan) when superimposed to β-l-glucopyranose (1C4, green) in panel
a is more similar than α-l-glucopyranose (1C4, purple) in panel b. The insets in panel a, b show
the representative overlay of the Newman projection of the anomeric
OH1. The SNi-like reaction mechanism with an oxocarbenium
transition state (4H3) that allows the approach
of the anomeric hydroxyl, α–OH of d-glucose
(4C1) in panel c, or the β–OH of l-glucose (1C4) in panel d, is guided
by hydrogen bonding from the same face as the NDP leaving group. R
= NDP.For retaining glycosyltransferases
(GTs) with an internal nucleophilic
substitution (SNi) mechanism, the anomeric selectivity
can be expected to invert when (S)-selectivity is
retained (Figure c,d).[3−5] The “same-face” attack of the nucleophile (i.e., sugar
acceptor) on the leaving group (i.e., sugar donor) is guided by hydrogen
bonding and proceeds with high anomeric selectivity for the sugar
donor and acceptor.[4] The position of the
anomeric hydroxyl of the sugar acceptor might affect the type of glycosidic
bond formed.Here, the anomeric selectivity of the retaining
LeLoir GT trehalose
transferase (TreT) was investigated. TreT is particularly suitable
for the screening of l-glycopyranoses, as it couples nucleotide
diphosphate (NDP) sugar donors to a wide spectrum of nonphosphorylated d-sugar acceptors, resulting in an α,α-(1 →
1)-glycosidic linkage.[5] We focused on the
recently described TreT from Thermoproteus uzoniensis (TuTreT) fused to mCherry for the systematic screening
of d- and l-glycopyranoses as sugar acceptors.[6,7] mCherry TuTreT is an interesting enzyme because
of a high thermostability, high activity, the possibility of fluorometric
detection that is due to mCherry, and performance as an immobilized
catalyst.[8]Initially, the reaction conditions were optimized to exclude any
possible side reactions or promiscuous activities. TuTreT did not display any phosphorylase or hydrolase activity. The
use of glucose-1-phosphate as sugar donor did not result in the formation
of trehalose (excluding phosphorylase activity), and no hydrolase
activity was observed when the enzyme was incubated solely with trehalose.
However, slow hydrolysis of uridine diphosphate-glucose (UDP-glucose)
to UDP and glucose by TuTreT was observed, resulting
in the subsequent formation of trehalose from glucose and UDP-glucose
(Figure S3). To minimize the undesired
formation of trehalose as a side product via UDPglucose hydrolysis
during the screening of other sugar acceptors, the reaction time was
limited to 60 min using 1.0 mg mL–1 of TuTreT.[9,10]Using these optimized conditions,
the substrate tolerance of TuTreT was probed in a
HPLC-based screening of d- and l-sugars (Figure ). Conversion of l-glycopyranoses resulted
in the hypothesized β-selectivity for TuTreT.
Successful enzymatic conversions were repeated on preparative scale
and the obtained trehalose analogues were analyzed by NMR and HR-MS
(Supporting Information). d-Glucose, d-mannose, d-galactose, and d-xylose exclusively
led to the formation of α,α-(1 → 1)-linked trehalose
derivatives, while l-glucose, l-galactose, and l-gulose led to the formation of α,β-(1 →
1)-linked trehalose derivatives. The long-range C–H coupling
over the glycosidic linkage confirmed the direct coupling of the C1acceptor with the H1′donor and vice versa
in gHMBC experiments. The 4C1 configuration
of α-d-α-d-glycopyranosides was confirmed
by J1,2 coupling (∼4 Hz) of the
anomeric protons, which are gauche configured. The
anomeric protons of β-l-glycopyranosides with a 1C4 chair conformation are anti configured, resulting in larger J1,2 coupling constants (∼8 Hz).[1]
Figure 2
mCherry TuTreT catalyzed conversion of d- and l-sugar acceptors with UDP-d-glucose as donor
for the screening for the formation of product. The conversion was
determined using external calibration curves of sugar acceptor with
HPLC. 1n.d: not detected including a trehalose analogue
product. Reaction conditions: substrate (10 mM), UDP-d-glucose
(40 mM), HEPES (50 mM), MgCl2 (20 mM), mCherry TuTreT (13.5 nM), pH 7.0, 60 °C, 1 h of reaction time.
mCherry TuTreT catalyzed conversion of d- and l-sugar acceptors with UDP-d-glucose as donor
for the screening for the formation of product. The conversion was
determined using external calibration curves of sugar acceptor with
HPLC. 1n.d: not detected including a trehalose analogue
product. Reaction conditions: substrate (10 mM), UDP-d-glucose
(40 mM), HEPES (50 mM), MgCl2 (20 mM), mCherry TuTreT (13.5 nM), pH 7.0, 60 °C, 1 h of reaction time.Further analysis of the HPLC screening demonstrates
that d- and l-enantiomers of glucose and galactose
were accepted,
but l-mannose was not. For l-gulose, l-allose,
and l-altrose, their d-glycopyranoses enantiomers
were not accepted. Anomeric selectivity is dictated by more than the
anomeric configuration, and the overall conformation of the sugar
acceptor is important as well. The structural variants of d-glycopyranoses with a 4C1 configuration were
readily converted, such as d-glucose, d-mannose,
and d-galactose. This is in line with results for other TreTs.[11,12] C5 sugars were generally less well accepted, with the exception
of d-xylose, which lacks a CH2OH group in comparison
to d-glucose. The ketohexopyranosesd-fructose and d- and l-tagatose were not converted under the conditions
provided, which display dissimilar overall structural conformation
of the cyclic ring structure as well as the anomeric configuration.
Interactions between active site residues and the carbohydrate substrate
were investigated by extending the substrate screening to fluorodeoxy-carbohydrates.
Unlike hydroxyl groups, fluorine can exclusively function as a hydrogen
bond acceptor. All fluoro-deoxy-d-glucopyranoses were quantitatively
converted as acceptor substrates, with the exception of 4-deoxy-4-fluoro-d-glucose.[11] The interaction of hydrogen
bond donor 4-OH of the sugar acceptor with the deprotonated Asp254
of TuTreT is possibly important for acceptor substrate
recognition.The substrate tolerance toward the glycopyranose
moiety of the
sugar donor and the sugar acceptor is distinct from one another.[13] For instance, the coupling of UDP-d-glucose with N-acetyl-d-glucosamine (GlcNAc)
did not result in any observable conversion. The sugar donor UDP-d-GlcNAc and d-glucose is readily converted by TuTreT with >98% conversion and α,α-(1 →
1)-selectivity.As this is the first report that the α-d-selective
retaining glycosyltransferase TreT catalyzes the glycosidic bond formation
with β-l-glycopyranose acceptors, the protein crystal
structure of TuTreT was studied. As the mCherry TuTreT fusion construct did not crystallize satisfactorily,
the glycine-rich linker of the fusion protein was cleaved using “in situ” proteolysis with retention of enzyme activity
(Figure S1), and the protein was purified
(Figure S2). The protein without the mCherry
tag subsequently crystallized as apo (PDB: 6ZJ4, 2.1 Å resolution),
cocrystallized with magnesium(II) (PDB: 6ZJ7, 2.15 Å resolution), or soaked with d-trehalose (PDB: 6ZJH, 2.1 Å resolution), d-glucopyranosyl-l-galactopyranose (PDB: 6ZN1, 1.75 Å resolution), and UDP-α-d-glucose (PDB: 6ZMZ, 1.9 Å resolution). The latter three are shown
in Figure a–c.
Figure 3
Protein
crystal structure of TuTreT containing
UDP-d-glucose in panel a (PDB: 6ZMZ, inset in panel d), TuTreT bound with α,α-trehalose in panel b (PDB: 6ZJH, inset shown in
panel e), and TuTreT in complex with enzymatically
synthesized α-d-glucopyranosyl-β-l-galactopyranose
in panel c (PDB: 6ZN1, inset shown in panel f). The nature of the glycosidic bond is demonstrated
showing a high overlap for the sugar donor and the glycosidic linkages
between panels d and e, or panels d and f, while the orientation of
the sugar acceptor changes slightly between panels e and f.
Protein
crystal structure of TuTreT containing
UDP-d-glucose in panel a (PDB: 6ZMZ, inset in panel d), TuTreT bound with α,α-trehalose in panel b (PDB: 6ZJH, inset shown in
panel e), and TuTreT in complex with enzymatically
synthesized α-d-glucopyranosyl-β-l-galactopyranose
in panel c (PDB: 6ZN1, inset shown in panel f). The nature of the glycosidic bond is demonstrated
showing a high overlap for the sugar donor and the glycosidic linkages
between panels d and e, or panels d and f, while the orientation of
the sugar acceptor changes slightly between panels e and f.The overall three-dimensional fold observed in
all determined crystal
structures are similar to the one found in trehalose phosphate synthase
(OtsA, PDB: 1GZ5, RMSD of 2.0
Å for 304 Cα)[14] from E. coli and in trehalose transferase from Pyrococcus
horikoshii (PDB: 2X6Q, RMSD of 2.5 Å for 363 Cα),[15] showing
in each domain a characteristic Rossmann fold (Figure S4). Furthermore, all TuTreT structures
show a monomer in the asymmetric unit, and this state was confirmed
by size-exclusion small-angle X-ray scattering (SEC-SAXS) measurements
in aqueous solution (Figure S9–S10).[16] We conclude that the functional unit
of TuTreT is a monomer, whereas a dimer has been
described for PhTreT[15] and a tetramer for OtsA from E. coli.[14] The overall protein conformation remained unchanged
when bound to ligands in aqua, according to SEC-SAXS
(Figure S9–S10). The protein crystal
structures also show high structural similarity (Table S3); however, a minor conformational change was observed
for the sugar donor binding site of TuTreT when it
was soaked with UDP-d-glucose. Hydrogen bond interactions
between the uracil moiety and a disordered loop region of the protein
(Figure S6, res. 250–262), were
inducing a shift of an α-helix by 2.0 Å (Figure S5). This finding is not in agreement for what has
been found for PhTreT, where a larger conformational
change was observed for the whole domain after soaking the crystals
with trehalose.[15]The active site
of TuTreT is located between the N- or C-terminal domains of the acceptor
and donor binding sites (Figure S4). Substrate-bound
structures demonstrated clear electron densities at the active site
(Figure S6). The active site residues for
the sugar donor and acceptor binding sites are conserved for TuTreT, PhTreT,[15] and OtsA (Figure S7).[14] In TuTreT, the pyrophosphate moiety of
the nucleotide sugar donor interacts with Arg221 and Lys226, and the
active site can accommodate pyrimidine or purine nucleobases (Figure a,d). This leads
to the ability of the enzyme to convert nucleotide sugar donors with
different nucleotides, which holds for TreTs in general,[17,18] as was observed with UDP- and ADP-d-glucose with TuTreT previously.[6,7]The natural product
α,α-d-trehalose shows
an α,α-(1 → 1)-glycosidic bond when bound to TuTreT (Figure b,e). In trehalose, the α-d-glucopyranosedonor moiety binds at the same sugar binding site as the UDP-α-d-glucopyranose donor (Figure S8a–c). The sugar donor binding site of UDP-d-glucose of TuTreT
(shown in Figure S8d) is similar to what
has been reported for TreT from Pyrococcus horikishii.[15] Interestingly, the α,α-
and α,β-trehalose derivatives the α-d-glucopyranoseacceptor interacts with TuTreT in a different
binding mode than the β-l-galactopyranoseacceptor (Figure S8e,f). Arg221 might bind with
the 2OH and 3OH of the β-l-galactopyranoseacceptor (Figure S8e), while the 3OH and 4OH of
α-d-glucopyranoseacceptor is at closer distance
to Arg221 (Figure S8f). The movement of
the highly conserved Arg221 when UDP-d-glucose is bound was
notable (Figure S5f), which has been postulated
to play a role in substrate recognition in PhTreT.[15] Also, Asp256 is at a hydrogen bonding distance
to 4OH of α-d-galactopyranoseacceptor (Figure S8e), while no clear electron density
for this loop could be found for the natural α-d-glucopyranoseacceptor. The α-d-glucopyranoseacceptor moiety of trehalose is directed into the wide cavity of TuTreT.Within the protein crystal structure, the
conformation of the carbohydrates
demonstrated for the enzymatically synthesized α,β-(1
→ 1)-l-galactotrehalose (Figure c,f), a 1C4 conformation
of the β-l-galactopyranose moiety. As was shown in Figure , the anomeric β–OHacceptor hydroxyl group points into the same direction as the
α-OHacceptor hydroxyl group in the natural substrate d-trehalose (Figure f). Hence, the overall geometry of α-d-glucopyranose
with a 4C1 conformation and β-l-galactopyranose with a 1C4 conformation are
highly similar (Figure S8g,h).On
the basis of these findings, we hypothesized that highly (S)-selective TuTreT guides the anomeric hydroxyl of the sugar acceptor
according to a SNi-like mechanism. More specifically, the
α-OH of d-glucose (4C1), or the
β-OH of l-galactose (1C4) are
guided by hydrogen bonding from the same face as the NDP leaving group.
As the SNi-like mechanism requires the same-face participation
with the anomeric hydroxyl of the nucleotide phosphate of the glycosyl
donor, the sugar coupling does not readily proceed for the equatorial
β-OH of d-glucose (4C1) or the
axial α-OH of l-galactose (1C4). This allows TuTreT to retain its (S)-selectivity for the anomeric hydroxyl group, while the anomeric
configuration inverts. This mechanistic rationale explains the inversion
of anomeric selectivity of TuTreT with l-glycopyranose acceptors, emphasizing that understanding the structural
conformations of hexopyranoses is important for understanding enzyme
selectivity when glycosidic bonds are formed. This inversion of anomeric
selectivity might not be limited to TuTreT and could
occur in other GTs as well.In conclusion, mCherry TuTreT catalyzes the formation
of trehalose derivatives with a large substrate spectrum. The switch
in anomeric selectivity for d- and L-sugar acceptors
can be explained on the basis of structural conformations of carbohydrates,
leading to the formation of distinctive α-d-α-d- or α-d-β-l-glycosidic linkages.
This paves the way for further studies of utilizing rare l-glycopyranoses with retaining LeLoir glycosyltransferases, which
are especially interesting for the production of oligosaccharides
and glycans with unnatural glycosidic linkages.
Authors: Jürgen Seibel; Roxana Moraru; Sven Götze; Klaus Buchholz; Shukrallah Na'amnieh; Alice Pawlowski; Hans-Jürgen Hecht Journal: Carbohydr Res Date: 2006-07-25 Impact factor: 2.104
Authors: Seung Seo Lee; Sung You Hong; James C Errey; Atsushi Izumi; Gideon J Davies; Benjamin G Davis Journal: Nat Chem Biol Date: 2011-08-07 Impact factor: 15.040
Authors: Jessica M Groenevelt; Lisa M Meints; Alicyn I Stothard; Anne W Poston; Taylor J Fiolek; David H Finocchietti; Victoria M Mulholand; Peter J Woodruff; Benjamin M Swarts Journal: J Org Chem Date: 2018-07-23 Impact factor: 4.354
Authors: Clement E Blanchet; Alessandro Spilotros; Frank Schwemmer; Melissa A Graewert; Alexey Kikhney; Cy M Jeffries; Daniel Franke; Daniel Mark; Roland Zengerle; Florent Cipriani; Stefan Fiedler; Manfred Roessle; Dmitri I Svergun Journal: J Appl Crystallogr Date: 2015-03-12 Impact factor: 3.304