Abstract

Complex glycans have important roles in biological recognition processes and considerable pharmaceutical potential. The synthesis of novel glycans can be facilitated by engineering glycosyltransferases to modify their substrate specificities. The choice of sites to modify requires the knowledge of the structures of enzyme–substrate complexes while the complexity of protein structures necessitates the exploration of a large array of multisite mutations. The retaining glycosyltransferase, α-1,3-galactosyltransferase (α3GT), which catalyzes the synthesis of the α-Gal epitope, has strict specificity for UDP-galactose as a donor substrate. Based on the structure of a complex of UDP-galactose with α3GT, the specificity for the galactose moiety can be partly attributed to residues that interact with the galactose 2-OH group, particularly His280 and Ala282. With the goal of engineering a variant of bovine α3GT with GalNAc transferase activity, we constructed a limited library of 456 α3GT mutants containing 19 alternative amino acids at position 280, two each at 281 and 282 and six at position 283. Clones (1500) were screened by assaying partially purified bacterially expressed variants for GalNAc transferase activity. Mutants with the highest levels of GalNAc transferase activity, AGGL or GGGL, had substitutions at all four sites. The AGGL mutant had slightly superior GalNAc transferase activity amounting to about 3% of the activity of the wild-type enzyme with UDP-Gal. This mutant had a low activity with UDP-Gal; its crystallographic structure suggests that the smaller side chains at residues 280–282 form a pocket to accommodate the larger acetamido group of GalNAc. Mutational studies indicate that Leu283 is important for stability in this mutant.

Introduction

The structures of glycans depend on the specificity of glycosyltransferases (GTs), the enzymes that catalyze their synthesis. GTs catalyze the transfer of a sugar residue from activated donor molecules (e.g., UDP-galactose) to a target group on a specific acceptor molecule, forming a glycosidic bond and are specific for the glycosyl donor, the acceptor substrate, and for the glycosidic linkage in the product. Since their specificities determine the structures of oligomers that contain high levels of biological information, an understanding of the structural basis of their specificity and catalytic activities is a core issue in the field of glycobiology (Breton et al. 2006).

UDP-galactose:β-galactosyl α-1,3-galactosyltransferase (α- 3GT, EC 2.4.1.151), a member of family 6 of the 90 GT families in the CAZY database (Campbell et al. 1997; Coutinho et al. 2003), is a retaining GT that catalyzes the transfer of galactose from UDP-α-d-galactose into an α-1,3-linkage with β-galactosyl groups in glycoconjugates. Its homologs include the histo-blood group A and B glycosyltransferases, Forssman glycolipid synthase, isogloboside 3 (iGb3) synthase, and some mammalian, bacterial, and viral proteins of unknown function (Turcot-Dubois et al. 2007). α3GT and its products are present in most mammals but not in Old World primates including humans because of inactivating mutations (Galili and Swanson 1991). The absence of active enzymes in these species results in the production of natural antibodies directed toward the products of enzyme action, collectively designated as the α-Gal epitope. Such antibodies are thought to provide a barrier to infection by pathogenic microorganisms and viruses (Galili et al. 1988).

In previous studies, structures have been determined for wild-type bovine α3GT and various mutants, in the free state and in complexes with substrates and inhibitors (Boix et al. 2001, 2002; Zhang et al. 2003, 2004; Jamaluddin et al. 2007; Tumbale et al. 2008). When wild-type α3GT is cocrystallized with UDP-galactose, its structure was found to contain the UDP and β-galactose, reflecting the relatively high UDP-Gal hydrolase activity of α3GT (Boix et al. 2002). Recently, we have determined the structure of an inhibitory analog of UDP-Gal, UDP-2F-galactose, and Mn2+ in a complex with a variant of α3GT that contains a conservative mutation in the C-terminal region, R365K (Jamaluddin et al. 2007), and also the structure of a complex of the intact donor substrate UDP-Gal with a mutant of α3GT (E317Q) that has very low galactosyltransferase and UDP-Gal hydrolase activities (Tumbale et al. 2008). In these structures, NE2 of the imidazole ring in the side chain of His280 and the peptide nitrogen of Ala282 are within hydrogen bonding (H-bond) distance of the galactose 2-OH; the peptide oxygen of Ala281 is also within H-bond distance of the galactose 3-OH. The role of this region in donor substrate specificity is highlighted by the fact that the homologous human blood group A and B glycosyltransferases differ in sequence at only four positions and can be interconverted in their respective specificities for UDP-GalNAc and UDP-Gal as donor substrates by substitutions for the residues corresponding to His280 and Ala282, namely Leu/Met 266 and Gly/Ala268 (Seto et al. 1997). A naturally occurring substitution at a different site (Pro234Ser) in the blood group B glycosyltransferase produces an enzyme with an increased kcat for UDP-GalNAc, but this mutation is in the acceptor binding site, and the enhanced kcat appears to result from an increase in free energy of the enzyme complex with the two substrates through weaker binding of the acceptor substrate (Marcus et al. 2003). To attempt to identify a mutant of α3GT with a level of GalNAc transferase activity that could be useful for glycan synthesis, we selected residues 280–283, adjacent to His280 and Ala282, that differ between different members of the α3GT family, for variation, and constructed a restricted combinatorial library based on the knowledge of variability at these sites. Activity screening identified two variants with sufficient GalNAc transferase activity to be useful for glycan synthesis (3% of the galactosyltransferase activity of the wild-type enzyme) but low galactosyltransferase activity. The structure of this mutant provides insights into the basis of donor specificity.

Results and discussion

Library screening

The combinatorial mutant library was constructed to introduce substitutions for residues 280–283; in the library, residue 280 had 19 alternatives, 281 and 282 2 alternatives each (Ala and Gly), and residue 283 6 different amino acids. To check the complexity of the library, plasmid DNA from 21 randomly selected mutant colonies was sequenced. Of these, 19 encoded different amino acids at the variable sites, 280–283, are as follows: KGGL, WGGL, AGGL, HAAI, LGGV, GGGV, GAGF, SGGI, YGAF, HGGV, IGGV, DAGI, HGGF, Stop-AAF, TGGM, EAGF, WGGF, KAAI, Stop-GGM.

The library contained 456 possible combinations and 1500 clones were screened for GalNAc transferase activity. The variants were expressed, purified, and assayed for glycosyltransferase activity, as described, in 250 groups, each containing six clones. Thirty groups that had UDP-GalNAc transferase activity exceeding 0.2 nmol/min were identified as positive and their component clones (180) were cultured individually. The proteins purified from the cultures of these variants were assayed separately and nine clones were identified that express proteins with GalNAc transferase activity exceeding 0.1 nmol/min. The relative amounts of protein in the different samples, as revealed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE), were taken into account in selecting these clones. DNA sequencing showed that six clones encoded variants with the translated sequence A280GGL283 and one clone had the sequence G280GGL283. The GalNAc transferase activities of these proteins exceeded their galactosyltransferase activity. Two additional clones were identified as having significant GalNAc transferase activity. These had the translated sequences, S280AGL283 and M280GGM283, but their activities with UDP-GalNAc were less than those of the AGGL and GGGL variants, and in both cases, the activity was higher with UDP-Gal than with UDP-GalNAc.

Properties of A280GGL283 and G280GGL283 mutants

When the two mutant enzymes were expressed on a larger scale, each was found to be present in a soluble form and, when purified using a Ni-chelate column, both were obtained in yields of about 10 mg/L of bacterial culture. Activity measurements indicated that both are less stable than the wild-type enzyme as shown by the lack of linearity of activity with time under the assay conditions used for screening (10 mM Tris–HCl, pH 7.4, at 37°C). However, when assays were conducted in a 10 mM MES buffer, pH 6.0, at a reduced temperature of 30°C, the activity was linear with time for at least 20 min. The lack of linearity at the higher pH and temperature does not appear to reflect substrate depletion or product inhibition because it was observed regardless of the level of product formation. The AGGL mutant showed a higher GalNAc transferase activity than the GGGL mutant and was selected for a more detailed study. Steady-state kinetic studies were carried at pH 6 and 30°C using both the mutant and wild-type enzyme to determine the apparent Km values for both substrates at a fixed concentration of the second substrate and to assess the kcat value. These revealed that the GalNAc transferase activity was about 3% of the galactosyltransferase activity of the wild-type α3GT when LacNAc was used as an acceptor under the same conditions (Table I). The galactosyltransferase activity is about 20% of the GalNAc transferase activity. Besides having a lower kcat than that of wild-type enzyme with UDP-Gal, the binding of the acceptor substrate, LacNAc was weaker as reflected in a 23-fold increase in Km. The catalytic activity was also characterized using lactose as a substrate. Wild-type α3GT has a higher kcat and Km with lactose relative to LacNAc when UDP-Gal was used as a donor substrate (Table I). With the AGGL mutant, the Km of lactose, determined using UDP-GalNAc as a donor substrate, is about 3-fold greater than that of the wild-type enzyme (determined with UDP-Gal), although there is some uncertainty in the exact value with AGGL because the highest concentration achieved in the assay was comparable to the estimated Km. The kcat of AGGL with lactose and UDP-GalNAc as substrates was about 2% of that of the wild-type enzyme with lactose and UDP-Gal. We attempted to determine the activity of wild-type α3GT with UDP-GalNAc as a substrate. As shown in Figure 1, wild-type α3GT has insignificant activity when directly compared with AGGL under similar conditions. Assays at higher enzyme concentrations indicate that there is a low level of activity but this is more than three orders of magnitude lower than the activity with UDP-Gal and was insufficient for meaningful kinetic studies.

Table I

Kinetic parameters of α3GT and the AGGL mutant with UDP-Gal and UDP-GalNAc as donor substrates

Acceptor substrate Enzyme Wild-type AGGL mutant 
 Donor substratea UDP-GalNAc UDP-Gal UDP-GalNAc UDP-Gal 
LacNAc kcat (s−1)a n.d.b 1.91 0.054 0.0067 
 Ka (mM)c  0.12 0.64 0.15 
 Kb (mM)d  0.6 15 13 
Lactose kcat (s−1)e n.d.b 4.8 0.095 0.0099 
 Ka (mM)f  0.16 0.25 0.04 
 Kb (mM)e  30 98 100 
Acceptor substrate Enzyme Wild-type AGGL mutant 
 Donor substratea UDP-GalNAc UDP-Gal UDP-GalNAc UDP-Gal 
LacNAc kcat (s−1)a n.d.b 1.91 0.054 0.0067 
 Ka (mM)c  0.12 0.64 0.15 
 Kb (mM)d  0.6 15 13 
Lactose kcat (s−1)e n.d.b 4.8 0.095 0.0099 
 Ka (mM)f  0.16 0.25 0.04 
 Kb (mM)e  30 98 100 

kcat is the turnover number; Ka is the Km for the donor substrate; Kb is the Km for the acceptor substrate (N-acetyllactosamine or lactose).

akcat was calculated from the apparent Vm at a saturating concentration of the acceptor substrate, corrected by multiplying by (1 + Ka/[UDP-sugar]). See Zhang et al. (2004).

bn.d., no activity detected.

cDetermined at 3 mM LacNAc.

dDetermined at 0.3 mM UDP-GalNAc or UDP-Gal.

eDetermined at 1 mM UDP-GalNAc or UDP-Gal.

fMeasured at 30 mM lactose.

Fig. 1

Comparison of the activities of the AGGL mutant and wild-type α3GT with UDP-GalNAc as a donor substrate at different concentrations of lactose. Symbols: forumla, AGGL mutant; ◯, wild-type α3GT. Each assay contained 3.6 μg of enzyme and was carried out as described in Material and methods.

Fig. 1

Comparison of the activities of the AGGL mutant and wild-type α3GT with UDP-GalNAc as a donor substrate at different concentrations of lactose. Symbols: forumla, AGGL mutant; ◯, wild-type α3GT. Each assay contained 3.6 μg of enzyme and was carried out as described in Material and methods.

Role of Leu283

It is interesting that residue 283 is Leu in both mutants since, while residues 280–282 make contacts with the galactose, residue 283 is buried and appears to have a structural as opposed to the functional role. To investigate the possible role of this residue, variants of the AGGL mutant with substitutions of Ile and Val for Leu283 were constructed and expressed. These variants were obtained in a soluble, active form, but in lower yields than the AGGL and GGGL variants (3 and 5 mg/mL, respectively, versus 10 mg/mL). They are also less soluble and precipitated during dialysis against a 20 mM Tris buffer, pH 7.4. The GalNAc transferase activity of both mutants was much lower than the wild-type protein and declined rapidly during assays. Although quantitative stability studies were not conducted, these properties suggest that the mutants are less stable than the parent AGGL mutant.

Crystal structure of the AGGL and RAAI mutants

To investigate the structural basis of specificity and catalysis, we determined crystallographic structures for the AGGL mutant, identified here, and for a RAAI mutant that was previously found to be devoid of catalytic activity (Zhang et al. 2003). The AGGL mutant was crystallized in the trigonal space group with one molecule in an asymmetric unit (Table II). The crystal structure showed the presence of one Mn2+ ion and UDP but there was no electron density corresponding to a GalNAc moiety, even though the enzyme was crystallized in the presence of UDP-GalNAc, suggesting that the substrate had been hydrolyzed during crystallization or that the GalNAc is present but disordered. The interactions between the enzyme and UDP and Mn2+ are essentially identical with those previously observed in the wild-type enzyme (Boix et al. 2001) (Figure 2A); however, the A280GGL283 sequence, replacing HAAI, was clearly observed in the electron density map. To attempt to determine the structural basis of the altered donor substrate specificity, a model of the AGGL mutant in a complex with UDP-GalNAc was constructed using the structure of complex of the Glu317Gln mutant (E317Q) with UDP-Gal (Tumbale et al. 2008) and the UDP complex of AGGL. In the complex of UDP-Gal with the E317Q mutant, it is observed that the O2′ atom of the galactose moiety forms an H-bond with the side chain of His280, a residue that had been previously implicated in substrate recognition and specificity of α3GT (Zhang et al. 2003). A previous study by Sujino et al. (2000) indicates that UDP-2′deoxy-galactose is a good substrate for bovine α3GT, so this interaction may be more important for excluding substrates such as UDP-GalNAc rather than enhancing binding and catalysis with UDP-Gal. In the AGGL mutant, the substitutions leave a pocket in the donor substrate recognition site that is not present in the wild-type protein. Modeling the complex of the AGGL mutant with UDP-GalNAc bound in the same configuration as UDP-Gal in its complex with E317Q mutant suggests that there is no H-bond between Ala280 and the GalNAc, but the pocket allows the mutant enzyme to accommodate the larger 2-acetamido group of the GalNAc moiety (Figure 2B and D). It is possible that this mode of binding results in increased flexibility of the GalNAc moiety or a conformation that is less amenable for acceptor substrate binding, accounting for the increased Kb values and lower kcat.

Table II

X-ray crystallographic data

 AGGL mutant RAAI mutant 
Substrates observed in the crystal structure Manganese UDP None 
Disordered regions 193–197, 340–368 193–197 (mol A), 
  355–368 (mol A), 
  193–198 (mol B), 
  342–368 (mol B), 
  193–197 (mol C), 
  337–368 (mol C), 
  193–195 (mol D), 
  348–368 (mol D) 
Space group P3121 C2 
Number of molecules/asymmetric unit 
Crystal parameters   
a (Å) 70.2 130.7 
b (Å) 70.2 65.3 
c (Å) 127.0 163.7 
α (°) 90.0 90.0 
β (°) 90.0 105.4 
γ (°) 120.0 90.0 
Resolution (Å) 50.0–2.7 50.0–2.8 
Rsymm (%)a (outermost shell)b 13.7 (30.3) 10.9 (35.7) 
Completeness (%) (outermost shell)b 99.7 (98.3) 83.8 (25.7) 
I/σI (outermost shell)b 12.1 (4.2) 8.38 (2.04) 
Total reflections 89,132 142,960 
Unique reflections 11,624 33,047 
Rcrystc/Rfreed (%) 19.2/26.0 21.9/29.0 
Ramachandran plot   
 % core/allowed 88.6/11.4 80.2/19.8 
RMSD from ideal   
 Bond angles (°) 1.199 1.4 
 Bond lengths (Å) 0.010 0.008 
Number of water molecules 44 101 
B-factor statistics (Å2  
 Protein 24.4 30.3 
 Ligand in the active site 20.6 (UDP)  
 Metal 20.8 (MN1)  
 AGGL mutant RAAI mutant 
Substrates observed in the crystal structure Manganese UDP None 
Disordered regions 193–197, 340–368 193–197 (mol A), 
  355–368 (mol A), 
  193–198 (mol B), 
  342–368 (mol B), 
  193–197 (mol C), 
  337–368 (mol C), 
  193–195 (mol D), 
  348–368 (mol D) 
Space group P3121 C2 
Number of molecules/asymmetric unit 
Crystal parameters   
a (Å) 70.2 130.7 
b (Å) 70.2 65.3 
c (Å) 127.0 163.7 
α (°) 90.0 90.0 
β (°) 90.0 105.4 
γ (°) 120.0 90.0 
Resolution (Å) 50.0–2.7 50.0–2.8 
Rsymm (%)a (outermost shell)b 13.7 (30.3) 10.9 (35.7) 
Completeness (%) (outermost shell)b 99.7 (98.3) 83.8 (25.7) 
I/σI (outermost shell)b 12.1 (4.2) 8.38 (2.04) 
Total reflections 89,132 142,960 
Unique reflections 11,624 33,047 
Rcrystc/Rfreed (%) 19.2/26.0 21.9/29.0 
Ramachandran plot   
 % core/allowed 88.6/11.4 80.2/19.8 
RMSD from ideal   
 Bond angles (°) 1.199 1.4 
 Bond lengths (Å) 0.010 0.008 
Number of water molecules 44 101 
B-factor statistics (Å2  
 Protein 24.4 30.3 
 Ligand in the active site 20.6 (UDP)  
 Metal 20.8 (MN1)  

aRsymm = ΣhΣi[∣Ii(h) − 〈I(h)〉∣/ΣhΣiIi(h)], where Ii is the ith measurement and 〈I(h)〉 is the weighted mean of all measurements of I(h).

bFigures in parentheses refer to the outermost shell (2.77–2.7 and 2.9–2.8 Å for AGGL and RAAI mutant structures, respectively).

cRcryst = ΣhFoFc∣/ΣhFo, where Fo and Fc are the observed and calculated structure factor amplitudes of reflection h, respectively.

dRfree is equal to Rcryst for a randomly selected 5% subset of reflections not used in the refinement.

Fig. 2

Comparison of the structures of AGGL and RAAI mutants with wild-type α3GT. (A) The interaction of H280 with UDP in native α3GT. The neighboring residues A281, A282, I283, and W356 are shown as ball-and-stick models. Hydrogen bonds are displayed as black dotted lines and the distances are shown. The bound manganese ion and UDP are labeled. (B) Residues 280–283 (AGGL) and the adjacent residues D340 and W356 in the structure of the AGGL mutant of α3GT are shown as ball-and-stick models. The bound manganese ion and UDP are labeled. (C) Residues 280–283 (RAAI) and the surrounding residues D340 and K350 in the structure of the RAAI mutant of the α3GT structure are shown. The interaction of R280 with D340 is represented by black dotted lines. Note: because of the H280R mutation, a significant rearrangement of C-terminal region was observed in the mutant enzyme and no bound UDP or manganese ion are present. (D) Stereo representation of the superposition of the structures of the 280HAAI283 region of wild-type α3GT and the corresponding region of the AGGL mutant. Wild-type α3GT is represented in green, and the AGGL mutant is in pink. The bound manganese ion and UDP are labeled. (E) Stereo figure representation of superposition of the structures of wild-type α3GT and H280R from the structure of the RAAI mutant. The H280R mutant is represented in orange, and wild-type α3GT is colored green. The UDP and manganese ion (purple sphere) from wild-type α3GT are shown. A drastic rearrangement of the C-terminal region is evident in the RAAI-α3GT mutant, as compared with the wild-type enzyme.

Fig. 2

Comparison of the structures of AGGL and RAAI mutants with wild-type α3GT. (A) The interaction of H280 with UDP in native α3GT. The neighboring residues A281, A282, I283, and W356 are shown as ball-and-stick models. Hydrogen bonds are displayed as black dotted lines and the distances are shown. The bound manganese ion and UDP are labeled. (B) Residues 280–283 (AGGL) and the adjacent residues D340 and W356 in the structure of the AGGL mutant of α3GT are shown as ball-and-stick models. The bound manganese ion and UDP are labeled. (C) Residues 280–283 (RAAI) and the surrounding residues D340 and K350 in the structure of the RAAI mutant of the α3GT structure are shown. The interaction of R280 with D340 is represented by black dotted lines. Note: because of the H280R mutation, a significant rearrangement of C-terminal region was observed in the mutant enzyme and no bound UDP or manganese ion are present. (D) Stereo representation of the superposition of the structures of the 280HAAI283 region of wild-type α3GT and the corresponding region of the AGGL mutant. Wild-type α3GT is represented in green, and the AGGL mutant is in pink. The bound manganese ion and UDP are labeled. (E) Stereo figure representation of superposition of the structures of wild-type α3GT and H280R from the structure of the RAAI mutant. The H280R mutant is represented in orange, and wild-type α3GT is colored green. The UDP and manganese ion (purple sphere) from wild-type α3GT are shown. A drastic rearrangement of the C-terminal region is evident in the RAAI-α3GT mutant, as compared with the wild-type enzyme.

Previous mutational studies of His280 showed that while the substitution of amino acids with smaller side chains, alanine and glycine, introduced low levels of GalNAc transferase activity, mutation to an amino acid with a larger side chain, arginine, rendered the enzyme completely inactive (Zhang et al. 2003) in contrast with a report on a study on porcine α3GT (Lazarus et al. 2002). Crystals of His280Arg mutant (H280R) diffracted only to 2.8 Å resolution (C2 space group with four molecules in the asymmetric unit). This structure is interesting because it shows a major conformational rearrangement and significant disorder in several regions (Figure 2C). No electron density was found corresponding to the donor substrate UDP-Gal or UDP even though the crystals had been grown in the presence of 10 mM UDP-Gal. This suggests that this mutant has a very low affinity for UDP-Gal explaining the observation that it is devoid of galactosyltransferase activity (Zhang et al. 2003).

In the structure of H280R mutant, there is a general rearrangement in the section of polypeptide chain that contains residues 338–350 and a complete absence of electron density for the polypeptide chain containing residues 351–368 indicating that this region is highly disordered. This appears to reflect potential steric clashes in the wild-type structure that would be generated by replacing His280 with arginine, which has a longer side chain, that results in major structural rearrangements. It appears that the extended side chain of Arg280 may push Trp356 away from the active site cleft, but this is not observed in the mutant structure because of disorder. In contrast, in the wild-type structure Trp356 makes van der Waal interactions with His280 (Figure 2A).

In addition, the steric clash with Trp356 arising from the His280 to Arg mutation is reflected in the noticeable structural reorientation of Asp340 in the structure (Figure 2E). In the structure of H280R mutant, the side chain of Asp340 is rotated by 90° and makes a direct H-bond with NH2 atom of Arg280 (Figure 2E). Another major structural change in the mutant structure is associated with Lys350, which interacts with Arg280 through a water molecule (Figure 2E). There is almost an ∼17 Å shift in the position of Cα of this residue compared with the wild-type structure. Lys350 is solvent-exposed in the wild-type structure whereas it points into the active site in the H280R structure. The drastic structural changes, produced by the H280R mutation combined with the large C-terminal rearrangement, make the active site less accessible providing a possible explanation for the absence of bound UDP and manganese ion in the mutant enzyme.

Conclusion

The present study supports the perspective that combinatorial libraries that are designed based on the knowledge of enzyme structure and function can provide a relatively efficient route for identifying variants with altered specificity. The assay of groups of six variants for GalNAc transferase activity, which is essentially absent from the wild-type protein, facilitated the initial screening. Family 6 glycosyltransferases provide an interesting test case for this approach because the blood group A and B transferases for UDP-GalNAc and UDP-Gal, respectively, are determined by the amino acids at positions 266 and 268, corresponding to His280 and Ala282 in α3GT. Previously, changes in donor substrate specificity have been successfully engineered in other glycosyltransferases, including the blood group enzymes (Seto et al. 1997; Marcus et al. 2003), β-1,4 galactosyltransferase (Ramakrishnan and Qasba 2007) and Clostridium difficile toxin B (Jank et al. 2005); the sequence changes used in these studies were homology based, reflecting substitutions in related proteins that differ in specificity, or in a naturally occurring mutant. In the present work, two-stage functional screening of the library identified two variants with sequences, AGGL and GGGL, reflecting changes at all four sites from the wild-type enzyme; the AGGL variant was present in multiple clones and had superior activity. The selection of these sequences appears to be based on stability as well as catalytic activity since two substitutions for Leu283, in the AGGL mutant based on residues present at this site homologs, were found to be destabilizing. In retrospect, it seems possible that additional GalNAc-specific mutants might have been identified if the activity screening had been carried out at a lower temperature. It is interesting that neither sequence found by screening is present in any currently known GT6 enzyme from eukaryotes. The GTs with the most similar sequences are histo-blood group glycosyltransferases from pig, cow, and rat which have the sequence AGGF; among these, the rat enzyme has been functionally characterized and found to possess A transferase activity (Turcot et al. 2003) reflecting the catalysis of GalNAc transfer to H-antigen-like substrates such as 2′fucosyllactose. The variant of α3GT identified here has a distinct specificity since it produces a GalNAc α1,3-gal β-1,4-OR structure that is not generated by any known GT-6 enzyme; the activity level of this variant is less than that of the wild-type enzyme with UDP-Gal (3%) but is nevertheless sufficiently high for enzyme-catalyzed or chemo-enzymatic oligosaccharide synthesis.

Material and methods

Expression, mutagenesis, and enzyme activity measurements

Mutants of the catalytic domain of α3GT were constructed using the PCR megaprimer method and expressed and purified as described previously (Zhang et al. 2001). Activity measurements and steady-state kinetic measurements were performed as previously described (Zhang et al. 2001, 2003).

Design of primers for combinatorial library construction

Primers were designed to randomly substitute 19 amino acids at position 280 (all except Cys), 2 amino acids (Ala and Gly) for residues 281 and 282, and 6 amino acids (Leu, Ile, Val, Met, Tyr, and Phe) for residue 283. To minimize the potential bias arising from overrepresentation of amino acids with higher degeneracy, six random primers were designed and used in which each variable amino acid was represented by a single codon, except for Arg at position 280 and Val at position 283, each of which was encoded by two codons. One stop codon at position 280 was unavoidably included in primers (ii) and (v). The primers used to introduce the different substitutions are listed in Table III. The primers were

  • TTCGGCGAAGGGGATTTTTATTACnwCGsCGsCdTkTTTGGGGGAACACCCACTCAGGTC

  • TTCGGCGAAGGGGATTTTTATTACnvGGSCGsCdTkTTTGGGGGAACACCCACTCAGGTC

  • TTCGGCGAAGGGGATTTTTATTACATGGsCGsCTkTTTGGGGGAACACCCACTCAGGTC

  • TTCGGCGAAGGGGATTTTTATTACnwCGsCGsCTACTTTGGGGGAACACCCACTCAGGTC

  • TTCGGCGAAGGGGATTTTTATTACnvGGsCGsCTACTTTGGGGGAACACCCACTCAGGTC

  • TTCGGCGAAGGGGATTTTTATTACATGGsCGsCTACTTTGGGGGAACACCCACTCAGGTC,

where n is A+C+G+T, v is A+C+G, d is A+T+G, w is A+T, s is C+G, and k is T+G.

Table III

Amino acids at positions 280–283 encoded in the six mixed primers used for library construction

Primer/     
position 280 281 282 283 
Primer (i) I, L, V, F, Y, N, H, D A, G A, G I, L, V, F, M 
Primer (ii) A, G, P, T, S, W, Q, E, D, K, R, STOP A, G A, G I, L, V, F, M 
Primer (iii) A, G A, G I, L, V, F, M 
Primer (iv) I, L, V, F, Y, N, H, D A, G A, G 
Primer (v) A, G, P, T, S, W, Q, E, D, K, R, STOP A, G A, G 
Primer (vi) A, G A, G 
Primer/     
position 280 281 282 283 
Primer (i) I, L, V, F, Y, N, H, D A, G A, G I, L, V, F, M 
Primer (ii) A, G, P, T, S, W, Q, E, D, K, R, STOP A, G A, G I, L, V, F, M 
Primer (iii) A, G A, G I, L, V, F, M 
Primer (iv) I, L, V, F, Y, N, H, D A, G A, G 
Primer (v) A, G, P, T, S, W, Q, E, D, K, R, STOP A, G A, G 
Primer (vi) A, G A, G 

Library construction

The combinatorial library was constructed using the PCR megaprimer method with a pET-42b-α3GT plasmid (Zhang et al. 2001) as a template. To reduce bias arising from the greater complementarity of some primers to the wild-type sequence, two additional DNA templates were included containing sequences encoding 280A-A-G-I283 and 280H-G-A-L283. In the first PCR, a mixture of primers, templates, T7 terminator, dNTPs (Eppendorf, NJ), thermo buffer (NEB, MA), and Vent polymerase was used for amplification under the following conditions: 3 min at 94°C followed by 30 cycles of 94°C for 1 min, 60°C for 1 min, 72°C for 1 min, and 72°C for 10 min. The product was gel-purified (Qiagen) and used as the reverse primer in a second amplification reaction that also included T7 promoter, dNTPs, thermo buffer, and Vent polymerase under the following conditions: 94°C for 3 min followed by 30 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min. Finally, the sample was incubated at 72°C for 10 min.

The second PCR product and pET-42b vector were digested separately with Nde1 and BamH1. The digestion products were gel-purified (Qiagen, CA) and ligated with T4 DNA ligase in a thermocycler using 30 cycles of 10°C for 3 min, 12°C for 3 min, 14°C for 3 min, 16°C for 3 min, and 18°C for 1 min with a final cycle of 65°C for 10 min. The ligation product was first desalted using Minielute (Qiagen, CA) and then transformed into Escherichia coli DH5α competent cells by electroporation. The cells were grown in 1 mL SOC medium with rapid shaking (250 rpm) at 37°C for 60 min before being plated on LB agar plates containing 50 μg/mL kanamycin. The plates were placed in an incubator at 37°C for 18–20 h.

Screening

Colonies were individually inoculated in the wells of a 96-well plate, and 1 mL of terrific broth containing kanamycin (50 μg/mL) was added to each. The colonies were then incubated at 37°C for 16–18 h. To extract the plasmid DNA from each clone, the plates were centrifuged at 3000 rpm for 10 min at 24°C. The pellet in each well was resuspended in a P1 buffer (100 μL) containing RNAse (Qiagen, CA); a P2 buffer (100 μL) was then added to each well followed by an N3 buffer (150 μL). The mixture was centrifuged at 3000 rpm at 24°C for 20 min. Each supernatant was transferred to a well in a fresh plate and the plasmid DNA was precipitated by adding 0.5 mL 95% ethanol and incubating at −20°C overnight. The precipitate was collected by centrifugation and washed twice with 500 μL 70% ethanol. The plasmid DNA from each well was resuspended in sterile double distilled water (50 μL).

The vector DNA isolated from individual colonies was used to transform BL21 (DE3) Gold competent cells. The cells were heat shocked at 42°C for 50 s. One milliliter of LB medium was added to each well and the cells were grown at 37°C for 1 h. A sample (200 μL) from each well was transferred into 1.5 mL of selective medium (LB Kan 50 μg/mL) and the cultures were grown at 37°C until the optical density at 600 nm reached 0.6. The temperature was then lowered to 24°C to allow leaky expression overnight. Cell cultures were centrifuged at 3000 × g for 5 min and the cell pellets were lysed using EasyLyse (Epicenter, WI). The lysates from groups of six mutants were transferred to a 1.5 mL Eppendorf tube with 100 μL of 50% suspended Ni-NTA resin that had been previously equilibrated with 20 mM Tris–HCl, 0.5 M NaCl, pH 7.8, and were gently shaken at 4°C for 1 h. The resin was washed with 1 mL of 20 mM Tris–HCl, pH 7.9, containing 0.5 M NaCl and 5 mM imidazole followed by 0.5 mL of 20 mM Tris–HCl, pH 7.9, containing, 0.5 M NaCl and 30 mM imidazole. Proteins were finally eluted with 200 μL of 20 mM Tris–HCl, pH 7.9, containing 0.5 M NaCl and 250 mM imidazole.

Activity screening

Assays contained 50 μL eluate, 50 mM Tris–HCl buffer, pH 7.4, 20 mM lactose, 0.3 mM UDP- [3H] GalNAc (specific activity 500 cpm/nmol), 0.1% bovine serum albumin, and 20 mM MnCl2 in a total volume of 100 μL and were incubated at 30°C for 20 min. Blanks contained same mixture except that the acceptor substrate was omitted. Ice-cold 0.1 M EDTA solution (0.1 mL) was added to terminate the reaction. The reaction mixture was then applied to a 1.5 mL Dowex (1 × 800) anion exchange resin, and the radioactive product was eluted with 1 mL followed by an additional 1 mL of water. The flow-through containing the uncharged product was collected in a plastic vial to which 10 mL of EcoLume (ICN Biomedicals, Costa Mesa, CA) was added. The vials were counted in a liquid scintillation counter (LKB). The standard assay described above was performed with the combined eluates from six colonies. The individual clones present in the groups positive for GalNAc transferase activity were identified and cultured. The proteins purified from these variants were again assayed to identify the positive variants individually. Samples of the expressed proteins were also analyzed by SDS–PAGE to assess the relative quantity of α3GT protein in each sample, for comparison with the level of activity. This provided a semi-quantitative estimate of protein levels and did not allow a reliable assessment of the specific activity of the recombinant enzymes. Clones expressing the variants with the highest levels of GalNAc transferase activity were grown to generate plasmid DNA for sequence analysis.

Enzyme kinetics

Steady-state velocities were determined using a radiochemical assay, similar to that described above for screening but with different assay conditions, modified to ensure linearity of activity with time for the mutant enzyme. Assays were conducted in a 20 mM MES buffer, pH 6.0, containing 0.1% BSA, 5 mM MnCl2 at 30°C for 10 min. For the determination of kinetic parameters, enzyme activity was determined by varying the concentration of UDP-Gal (0.12–0.6 mM for the wild-type enzyme and 0.24– 1.2 mM for the AGGL mutant) or UDP-GalNAc (0.12–0.6 mM) at a fixed concentration of LacNAc (3 mM for wild type and 15 mM for AGGL) or varying the concentration of LacNAc (6– 30 mM for the AGGL mutant and 3–30 mM for wild type) or lactose (0–95 mM) at a fixed concentration of the donor substrate (1 mM). Generally, less than 10% of donor substrate was converted to product during the assay. Data were analyzed by fitting to the single substrate Michaelis–Menten equation (1):  

(1)
formula

This provides apparent Km values for the donor (Ka) or acceptor (Kb) substrate and two apparent Vm values. The apparent Vm value obtained by extrapolating to saturating concentrations of LacNAc or lactose was adjusted by multiplying by (1 + [UDP-sugar]/Ka), where Ka is the apparent Km for UDP-Gal or UDP-GalNAc, to get a more accurate estimation of Vm and kcat.

X-ray crystallography

The purified bovine α3GT mutants were stored at −20°C in a 20 mM MES–NaOH buffer (pH 6.0) containing 50% glycerol. Crystals of the AGGL mutant in complex with UDP-GalNAc were grown at 16°C by the vapor diffusion, hanging drop method by mixing 1 μL of the protein at 5 mg/mL in a 20 mM MES–NaOH buffer, pH 6.0, 10% glycerol, containing 10 mM MnCl2 and 10 mM UDP-GalNAc, with an equal volume of reservoir solution containing 10–15% PEG 6000, 0.1 M Tris–HCl, pH 8.0, and 15–25% MPD. Crystals of the RAAI mutant were grown in the presence of UDP-Gal in 0.1 M Tris–HCl, pH 8.0, containing 10–50% PEG 4000 and 0.2 M sodium acetate, by mixing 1 μL of the protein at 5 mg/mL in the presence of 10 mM MnCl2 and 10 mM UDP-Gal with an equal volume of the reservoir solution. Before data collection the crystals were flash-cooled at 100 K. No cryoprotectant was used for the AGGL mutant crystal (crystals were grown at a high concentration of MPD) while for the RAAI mutant a cryoprotectant containing the reservoir solution and 25% glycerol was used during X-ray data collection. Diffraction data from single crystals were collected on station 10.1 of the Synchrotron Radiation Source (Daresbury, UK), which was equipped with a Quantum-4 CCD detector (Area Detector Systems Corporation). Two datasets were collected for the AGGL and RAAI mutants at 2.7 and 2.8 Å, respectively. Raw data images were indexed and scaled using DENZO and SCALEPACK modules of the HKL suite (Otwinowski and Minor 1997).

The structures of α3GT mutants were determined by the molecular replacement method using the native form II structure (PDB code: 1K4V (Boix et al. 2001)), as a search model with the program MOLREP (Vagin and Teplyakov 1997). Crystallographic refinement was performed using the program package REFMAC (CCP4 1994) for the complex structures. After the initial refinement, the difference electron density maps revealed the presence of the mutated residues at their respective positions and a bound UDP and Mn2+. Several rounds of energy minimization, simulated annealing, individual B-factor refinement, and model building were performed using the program COOT (Emsley and Cowtan 2004) until the Rfree could not be improved. Some visible water molecules were gradually included into the model at positions corresponding to peaks in the ∣Fo∣ − ∣Fc∣ electron density map with heights greater than 3σ and at H-bond distance from appropriate atoms. Residues with poor side-chain density were modeled as alanine, and regions with very poor main chain electron density were excluded from refinement to avoid bias in the model. Refinement was carried out until the Rcryst and Rfree could not be improved. The data collection statistics and the final refinement statistics are listed in Table II. All figures were generated using PyMOL (http://www.pymol.org).

Protein data bank accession codes

The atomic coordinates and the structure factors that have been deposited with the RCSB Protein Data Bank are as follows: 2VXL and 2VXM for the AGGL and RAAI mutant structures, respectively.

We thank the scientists at synchrotron radiation source Station 10.1 (Daresbury-UK) and Shalini Iyer for their support during X-ray data collection. H.J. was supported by a postgraduate studentship by the Ministry of Higher Education Malaysia and Universiti Teknologi, Malaysia.

Conflict of interest statement

There are no conflicts of interest.

Abbreviations

    Abbreviations
  • α3GT

    bovine β-galactosyl α-1,3-galactosyltransferase

  • Gal

    galactose

  • Glc

    glucose

  • GT

    glycosyltransferase

  • H-bond

    hydrogen bonding

  • Lac

    lactose

  • LacNAc

    N-acetyllactosamine

  • SDS–PAGE

    sodium dodecyl sulfate–polyacrylamide gel electrophoresis

  • UDP-Gal

    UDP-galactose

  • UDP-GalNAc

    UDP-N-acetylgalactosamine

References

Boix
E
Swaminathan
GJ
Zhang
Y
Natesh
R
Brew
K
Acharya
KR
Structure of UDP complex of UDP-Galactose:β-galactoside-α-1,3-galactosyltransferase at 1.53 Å resolution reveals a conformational change in the catalytically important C-terminus
J Biol Chem
 , 
2001
, vol. 
276
 (pg. 
48608
-
48614
)
Boix
E
Swaminathan
GJ
Zhang
Y
Natesh
R
Brew
K
Acharya
KR
Structural basis of ordered binding of donor and acceptor substrates to the retaining glycosyltransferase: α-1,3-Galactosyltransferase
J Biol Chem
 , 
2002
, vol. 
277
 (pg. 
28310
-
28318
)
Breton
C
Snajdrova
L
Jeanneau
C
Koca
J
Imberty
A
Structures and mechanisms of glycosyltransferases
Glycobiology
 , 
2006
, vol. 
16
 (pg. 
29R
-
37R
)
Campbell
JA
Davies
GJ
Bulone
V
Henrissat
B
A classification of nucleotide-diphospho-sugar glycosyltransferases based on amino acid sequence similarities
Biochem J
 , 
1997
, vol. 
326
 (pg. 
929
-
939
)
Coutinho
PM
Deleury
E
Davies
GJ
Henrissat
B
An evolving hierarchical family classification for glycosyltransferases
J Mol Biol.
 , 
2003
, vol. 
328
 (pg. 
307
-
317
)
Emsley
P
Cowtan
K
Coot: Model-building tools for molecular graphics
Acta Crystallogr
 , 
2004
, vol. 
D 60
 (pg. 
2126
-
2132
)
Galili
U
Shohet
SB
Kobrin
E
Stults
CL
Macher
BA
Man, apes, and old world monkeys differ from other mammals in the expression of α-galactosyl epitopes on nucleated cells
J Biol Chem
 , 
1988
, vol. 
263
 (pg. 
17755
-
17762
)
Galili
U
Swanson
K
Gene sequences suggest inactivation of α-1,3-galactosyltransferase in catarrhines after the divergence of apes from monkeys
Proc Natl Acad Sci USA
 , 
1991
, vol. 
88
 (pg. 
7401
-
7404
)
Jamaluddin
H
Tumbale
P
Withers
SG
Acharya
KR
Brew
K
Conformational changes induced by binding UDP-2F-galactose to α-1,3-galactosyltransferase-implications for catalysis
J Mol Biol
 , 
2007
, vol. 
369
 (pg. 
1270
-
1281
)
Jank
T
Reinert
DJ
Giesemann
T
Schulz
GE
Aktories
K
Change of the donor substrate specificity of Clostridium difficile toxin B by site-directed mutagenesis
J Biol Chem
 , 
2005
, vol. 
280
 (pg. 
37833
-
37838
)
Lazarus
BD
Milland
J
Ramsland
PA
Mouhtouris
E
Sandrin
MS
Histidine 271 has a functional role in pig alpha-1,3-galactosyltransferase enzyme activity
Glycobiology
 , 
2002
, vol. 
12
 (pg. 
793
-
802
)
Marcus
SL
Polakowski
R
Seto
NOL
Leinala
E
Borisova
S
Blancer
A
Roubinet
F
Evans
SV
Palcic
MM
A single point mutation reverses the donor specificity of human blood group B-synthesizing galactosyltransferase
J Biol Chem
 , 
2003
, vol. 
278
 (pg. 
12403
-
12405
)
Otwinowski
Z
Minor
W
Processing of X-ray diffraction data collected in oscillation mode
Methods Enzymol
 , 
1997
, vol. 
276
 (pg. 
307
-
326
)
Ramakrishnan
B
Qasba
PK
Role of a single amino acid in the evolution of glycans of invertebrates and vertebrates
J Mol Biol
 , 
2007
, vol. 
365
 (pg. 
570
-
576
)
Seto
NOL
Palcic
MM
Compston
CA
Li
H
Bundle
DR
Narang
SA
Sequential interchange of four amino acids from blood group B to blood group A glycosyltransferase boosts catalytic activity and progressively modifies substrate recognition in human recombinant enzymes
J Biol Chem
 , 
1997
, vol. 
272
 (pg. 
14133
-
14138
)
Sujino
K
Uchiyama
T
Hindsgaul
O
Seto
NOL
Wakarchuk
WW
Palcic
MM
Enzymatic synthesis of oligosaccharide analogues: Evaluation of UDP-Gal analogues as donors for three retaining α-galactosyltransferases
J Am Chem Soc
 , 
2000
, vol. 
122
 (pg. 
1261
-
1269
)
Tumbale
P
Jamaluddin
H
Thiyagarajan
N
Brew
K
Acharya
KR
Structural basis of UDP-galactose binding by α-3 galactosyltransferase: Role of negative charge on Asp316 in structure and activity
Biochemistry
 , 
2008
, vol. 
247
 (pg. 
8711
-
8718
)
Turcot
AL
Blancher
A
Le Moullac-Vaidye
B
Despiau
S
Rocher
J
Roubinet
F
Szpirer
C
Le Pendu
J
Cloning of a rat gene encoding the histo-blood group B enzyme: Rats have more than one Abo gene
Glycobiology
 , 
2003
, vol. 
13
 (pg. 
919
-
928
)
Turcot-Dubois
A-L
Le Moullac-Vaidye
B
Despiau
S
Roubinet
F
Bovin
N
Le Pendu
J
Blancher
A
Long-term evolution of the CAZY glycosyltransferase 6 (ABO) gene family from fishes to mammals—a birth-and-death evolution model
Glycobiology
 , 
2007
, vol. 
17
 (pg. 
516
-
528
)
Vagin
A
Teplyakov
A
MOLREP: An automated program for molecular replacement
J Appl Cryst
 , 
1997
, vol. 
30
 (pg. 
1022
-
1025
)
Zhang
Y
Deshpande
A
Xie
Z
Natesh
R
Acharya
KR
Brew
K
Roles of active site tryptophans in substrate binding and catalysis by α-1,3-galactosyltransferase
Glycobiology
 , 
2004
, vol. 
14
 (pg. 
1295
-
1302
)
Zhang
Y
Swaminathan
GJ
Deshpande
A
Natesh
R
Boix
E
Xie
Z
Acharya
KR
Brew
K
Roles of individual enzyme–substrate interactions by α-1,3-galactosyltransferase in catalysis and specificity
Biochemistry
 , 
2003
, vol. 
42
 (pg. 
13512
-
13521
)
Zhang
Y
Wang
PG
Brew
K
Specificity and mechanism of metal ion activation in UDP-galactose:β-galactoside-α-1,3-galactosyltransferase
J Biol Chem
 , 
2001
, vol. 
276
 (pg. 
11567
-
11574
)

Author notes

2
Present address: Faculty of Biosciences and Engineering, Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia.