Snapshot of a Michaelis complex in a sulfuryl transfer reaction: Crystal structure of a mouse sulfotransferase, mSULT1D1, complexed with donor substrate and accepter substrate

We report the crystal structure of mouse sulfotransferase, mSULT1D1, complexed with donor substrate 3′-phosphoadenosine 5′-phosphosulfate and accepter substrate p-nitrophenol. The structure is the first report of the native Michaelis complex of sulfotransferase. In the structure, three proposed catalytic residues (Lys48, Lys106, and His108) were in proper positions for engaging in the sulfuryl transfer reaction. The data strongly support that the sulfuryl transfer reaction proceeds through an S_N2-like in-line displacement mechanism.     

On the similar spatial arrangement of active site residues in PAPS-dependent and phenolic sulfate-utilizing sulfotransferases

Mammalian sulfotransferases (STs) utilize exclusively the sulfuryl group donor 3′-phosphoadenosine 5′-phosphosulfate (PAPS) to catalyze the sulfurylation reactions based on a sequential transfer mechanism. In contrast, the commensal intestinal bacterial arylsulfate sulfotransferases (ASSTs) do not use PAPS as the sulfuryl group donor, but instead catalyze sulfuryl transfer from phenolic sulfate to a phenol via a Ping-Pong mechanism. Interestingly, structural comparison revealed a similar spatial arrangement of the active site residues as well as the cognate substrates in mouse ST (mSULT1D1) and Escherichia coli CFT073 ASST, despite that their overall structures bear no discernible relationship. These observations suggest that the active sites of PAPS-dependent SULT1D1 and phenolic sulfate-utilizing ASST represent an example of convergent evolution.     

Crystal structures of SULT1A2 and SULT1A1∗3: Insights into the substrate inhibition and the role of Tyr149 in SULT1A2

The cytosolic sulfotransferases (SULTs) in vertebrates catalyze the sulfonation of endogenous thyroid/steroid hormones and catecholamine neurotransmitters, as well as a variety of xenobiotics, using 3′-phosphoadenosine 5′-phosphosulfate (PAPS) as the sulfonate donor. In this study, we determined the structures of SULT1A2 and an allozyme of SULT1A1, SULT1A1∗3, bound with 3′-phosphoadenosine 5′-phosphate (PAP), at 2.4 and 2.3 Å resolution, respectively. The conformational differences between the two structures revealed a plastic substrate-binding pocket with two channels and a switch-like substrate selectivity residue Phe247, providing clearly a structural basis for the substrate inhibition. In SULT1A2, Tyr149 extends approximately 2.1 Å further to the inside of the substrate-binding pocket, compared with the corresponding His149 residue in SULT1A1∗3. Site-directed mutagenesis study showed that, compared with the wild-type SULT1A2, mutant Tyr149Phe SULT1A2 exhibited a 40 times higher K_m and two times lower V_max with p-nitrophenol as substrate. These latter data imply a significant role of Tyr149 in the catalytic mechanism of SULT1A2.     

Molecular determinants of substrate specificity in plant 5'-methylthioadenosine nucleosidases

5′-Methylthioadenosine (MTA)/S-adenosylhomocysteine (SAH) nucleosidase (MTAN) is essential for cellular metabolism and development in many bacterial species. While the enzyme is found in plants, plant MTANs appear to select for MTA preferentially, with little or no affinity for SAH. To understand what determines substrate specificity in this enzyme, MTAN homologues from Arabidopsis thaliana (AtMTAN1 and AtMTAN2, which are referred to as AtMTN1 and AtMTN2 in the plant literature) have been characterized kinetically. While both homologues hydrolyze MTA with comparable kinetic parameters, only AtMTAN2 shows activity towards SAH. AtMTAN2 also has higher catalytic activity towards other substrate analogues with longer 5′-substituents. The structures of apo AtMTAN1 and its complexes with the substrate- and transition-state-analogues, 5′-methylthiotubercidin and formycin A, respectively, have been determined at 2.0-1.8 Å resolution. A homology model of AtMTAN2 was generated using the AtMTAN1 structures. Comparison of the AtMTAN1 and AtMTAN2 structures reveals that only three residues in the active site differ between the two enzymes. Our analysis suggests that two of these residues, Leu181/Met168 and Phe148/Leu135 in AtMTAN1/AtMTAN2, likely account for the divergence in specificity of the enzymes. Comparison of the AtMTAN1 and available Escherichia coli MTAN (EcMTAN) structures suggests that a combination of differences in the 5′-alkylthio binding region and reduced conformational flexibility in the AtMTAN1 active site likely contribute to its reduced efficiency in binding substrate analogues with longer 5′-substituents. In addition, in contrast to EcMTAN, the active site of AtMTAN1 remains solvated in its ligand-bound forms. As the apparent pK_a of an amino acid depends on its local environment, the putative catalytic acid Asp225 in AtMTAN1 may not be protonated at physiological pH and this suggests the transition state of AtMTAN1, like human MTA phosphorylase and Streptococcus pneumoniae MTAN, may be different from that found in EcMTAN.     

Crystal structures of two Bacillus carboxylesterases with different enantioselectivities

Naproxen esterase (NP) from Bacillus subtilis Thai I-8 is a carboxylesterase that catalyzes the enantioselective hydrolysis of naproxenmethylester to produce S-naproxen (E > 200). It is a homolog of CesA (98% sequence identity) and CesB (64% identity), both produced by B. subtilis strain 168. CesB can be used for the enantioselective hydrolysis of 1,2-O-isopropylideneglycerol (solketal) esters (E > 200 for IPG-caprylate). Crystal structures of NP and CesB, determined to a resolution of 1.75 Å and 2.04 Å, respectively, showed that both proteins have a canonical α/β hydrolase fold with an extra N-terminal helix stabilizing the cap subdomain. The active site in both enzymes is located in a deep hydrophobic groove and includes the catalytic triad residues Ser130, His274, and Glu245. A product analog, presumably 2-(2-hydroxyethoxy)acetic acid, was bound in the NP active site. The enzymes have different enantioselectivities, which previously were shown to result from only a few amino acid substitutions in the cap domain. Modeling of a substrate in the active site of NP allowed explaining the different enantioselectivities. In addition, Ala156 may be a determinant of enantioselectivity as well, since its side chain appears to interfere with the binding of certain R-enantiomers in the active site of NP. However, the exchange route for substrate and product between the active site and the solvent is not obvious from the structures. Flexibility of the cap domain might facilitate such exchange. Interestingly, both carboxylesterases show higher structural similarity to meta-cleavage compound (MCP) hydrolases than to other α/β hydrolase fold esterases.     

Step-wise enzymatic preparation and structural characterization of singly and doubly substituted arabinoxylo-oligosaccharides with non-reducing end terminal branches

Shearzyme (GH10 endo-1,4-β-d-xylanase) and two different α-l-arabinofuranosidases (AXH-m and AXH-d3) were used stepwise to manufacture arabinoxylo-oligosaccharides (AXOS) with α-l-Araf (1→2)-monosubstituted β-d-Xylp residues or α-l-Araf (1→2)- and (1→3) doubly substituted β-d-Xylp residues from wheat arabinoxylan (AX) in a rather straightforward way. Four major AXOS (d-I, d-II, m-I and m-II) were formed in two separate hydrolyses. The AXOS were purified and the structures were confirmed using TLC, HPAEC-PAD, MALDI-TOF-MS and 1D and 2D NMR spectroscopy. The samples were identified as d-I: α-l-Araf-(1→2)-[α-l-Araf-(1→3)]-β-d-Xylp-(1→4)-β-d-Xylp-(1→4)-d-Xylp, d-II: α-l-Araf-(1→2)-[α-l-Araf-(1→3)]-β-d-Xylp-(1→4)-d-Xylp, m-I: α-l-Araf-(1→2)-β-d-Xylp-(1→4)-β-d-Xylp-(1→4)-d-Xylp and m-II: α-l-Araf-(1→2)-β-d-Xylp-(1→4)-d-Xylp. To our knowledge, this is the first report on structural ¹H and ¹³C NMR analysis of xylobiose-derived AXOS d-II and m-II. The latter compound has not been reported previously. The doubly substituted AXOS were produced for the first time in good yields, as d-I and d-II corresponded to 11.8 and 5.6 wt% of AX, respectively. Singly α-l-Araf (1→2)-substituted AXOS could also be prepared in similar yields by treating the doubly substituted AXOS further with AXH-d3.     

The cDNA sequence of three hemocyanin subunits from the garden snail Helix lucorum

Hemocyanins are blue copper containing respiratory proteins residing in the hemolymph of many molluscs and arthropods. They can have different molecular masses and quaternary structures. Moreover, several molluscan hemocyanins are isolated with one, two or three isoforms occurring as decameric, didecameric, multidecameric or tubule aggregates. We could recently isolate three different hemocyanin isopolypeptides from the hemolymph of the garden snail Helix lucorum (HlH). These three structural subunits were named α_D-HlH, α_N-HlH and β-HlH. We have cloned and sequenced their cDNA which is the first result ever reported for three isoforms of a molluscan hemocyanin. Whereas the complete gene sequence of α_D-HlH and β-HlH was obtained, including the 5′ and 3′ UTR, 180 bp of the 5′ end and around 900 bp at the 3′ end are missing for the third subunit. The subunits α_D-HlH and β-HlH comprise a signal sequence of 19 amino acids plus a polypeptide of 3409 and 3414 amino acids, respectively. We could determine 3031 residues of the α_N-HLH subunit. Sequence comparison with other molluscan hemocyanins shows that α_D-HlH is more related to Aplysia californicum hemocyanin than to each of its own isopolypeptides. The structural subunits comprise 8 different functional units (FUs: a, b, c, d, e, f, g, h) and each functional unit possesses a highly conserved copper-A and copper-B site for reversible oxygen binding. Potential N-glycosylation sites are present in all three structural subunits. We confirmed that all three different isoforms are effectively produced and secreted in the hemolymph of H. lucorum by analyzing a tryptic digest of the purified native hemocyanin by MALDI-TOF and LC-FTICR mass spectrometry.     

Enzymatic synthesis of unique sialyloligosaccharides using marine bacterial α-(2→3)- and α-(2→6)-sialyltransferases

We investigated the acceptor substrate specificities of marine bacterial α-(2→3)-sialyltransferase cloned from Photobacterium sp. JT-ISH-224 and α-(2→6)-sialyltransferase cloned from Photobacterium damselae JT0160 using several saccharides as acceptor substrates. After purifying the enzymatic reaction products, we confirmed their structure by NMR spectroscopy. The α-(2→3)-sialyltransferase transferred N-acetylneuraminic acid (Neu5Ac) from cytidine 5′-monophospho-N-acetylneuraminic acid (CMP-Neu5Ac) to the β-anomeric hydroxyl groups of mannose (Man) and α-Manp-(1→6)-Manp, and α-(2→6)-sialyltransferase transferred N-acetylneuraminic acid to the 6-OH groups of the non-reducing end galactose residues in β-Galp-(1→3)-GlcpNAc and β-Galp-(1→6)-GlcpNAc.     

Reaction product affinity regulates activation of human sulfotransferase 1A1 PAP sulfation

Cytosolic sulfotransferase (SULT)-catalyzed sulfation regulates the activity of bio-signaling molecules and aids in metabolizing hydroxyl-containing xenobiotics. The sulfuryl donor for the SULT reaction is adenosine 3′-phosphate 5′-phosphosulfate (PAPS), while products are adenosine 3′,5′-diphosphate (PAP) and a sulfated alcohol. Human phenol sulfotransferase (SULT1A1) is one of the major detoxifying enzymes for phenolic xenobiotics. The mechanism of SULT1A1-catalyzed sulfation of PAP by pNPS was investigated. PAP was sulfated by para-nitrophenyl sulfate (pNPS) in a concentration-dependent manner. 2-Naphthol inhibited sulfation of PAP, competing with pNPS, while phenol activated the sulfation reaction. At saturating PAP, a ping pong kinetic mechanism is observed with pNPS and phenol as substrates, consistent with phenol intercepting the E–PAPS complex prior to dissociation of PAPS. At high concentrations, phenol competes with pNPS, consistent with formation of the E–PAP–phenol dead-end complex. Data are consistent with the previously reported mechanism for sulfation of 2-naphthol by PAPS, and its activation by pNPS [14]. Overall, data are consistent with release of PAP from E–PAP and PAPS from E–PAPS contributing to rate-limitation in both reaction directions.     

Crystal Structure of Human Poly(A) Polymerase Gamma Reveals a Conserved Catalytic Core for Canonical Poly(A) Polymerases

In eukaryotes, the poly(A) tail added at the 3′ end of an mRNA precursor is essential for the regulation of mRNA stability and the initiation of translation. Poly(A) polymerase (PAP) is the enzyme that catalyzes the poly(A) addition reaction. Multiple isoforms of PAP have been identified in vertebrates, which originate from gene duplication, alternative splicing or post-translational modifications. The complexity of PAP isoforms suggests that they might play different roles in the cell. Phylogenetic studies indicate that vertebrate PAPs are grouped into three clades termed α, β and γ, which originated from two gene duplication events. To date, all the available PAP structures are from the PAPα clade. Here, we present the crystal structure of the first representative of the PAPγ clade, human PAPγ bound to cordycepin triphosphate (3′dATP) and Ca^2 +. The structure revealed that PAPγ closely resembles its PAPα ortholog. An analysis of residue conservation reveals a conserved catalytic binding pocket, whereas residues at the surface of the polymerase are more divergent.     

Crystal structure of glycoside hydrolase family 78 alpha-L-Rhamnosidase from Bacillus sp. GL1

α-L-Rhamnosidase (EC 3.2.1.40) catalyzes the hydrolytic release of rhamnose from polysaccharides and glycosides. Bacillus sp. GL1 α-L-rhamnosidase (RhaB), a member of glycoside hydrolase (GH) family 78, is responsible for degrading the bacterial biofilm gellan, and also functions as a debittering agent for citrus fruit in the food and beverage industries through the release of rhamnose from plant glycoside, naringin. The X-ray crystal structure of RhaB was determined by single-wavelength anomalous diffraction using a selenomethionine derivative and refined at 1.9 Å resolution with a final R-factor of 18.2%. As is seen in the homodimeric form of the active enzyme, the structure of RhaB in crystal packing is a homodimer containing 1908 amino acids (residues 3-956), 43 glycerol molecules, four calcium ions, and 1755 water molecules. The overall structure consists of five domains, four of which are β-sandwich structures designated as domains N, D1, D2, and C, and an (α/α)₆-barrel structure designated as domain A. Structural comparison by DALI showed that RhaB shares its highest level of structural similarity with chitobiose phosphorylase (Z score of 25.3). The structure of RhaB in complex with the reaction product rhamnose (inhibitor constant, K_i = 1.8 mM) was also determined and refined at 2.1 Å with a final R-factor of 19.5%. Rhamnose is bound to the deep cleft of the (α/α)₆-barrel domain, as is seen in the clan-L GHs. Several negatively charged residues, such as Asp567, Glu572, Asp579, and Glu841, conserved in GH family 78 enzymes, interact with rhamnose, and RhaB mutants of these residues have drastically reduced enzyme activity, indicating that the residues are crucial for enzyme catalysis and/or substrate binding. To our knowledge, this is the first report on the determination of the crystal structure of α-L-rhamnosidase and identification of its clan-L (α/α)₆-barrel as a catalytic domain.     

Improvement of the efficiency of transglycosylation catalyzed by α-galactosidase from Thermotoga maritima by protein engineering

At high concentrations of p-nitrophenyl-α-D-galactopyranoside (pNPGal) as a substrate, its hydrolysis catalyzed by α-galactosidase from Thermotoga maritima (TmGalA) is accompanied by transglycosylation resulting in production of a mixture of (α1,2)-, (α1,3)-, and (α1,6)-p-nitrophenyl (pNP)-digalactosides. Molecular modeling of the reaction stage preceding the formation of the pNP-digalactosides within the active site of the enzyme revealed amino acid residues which modification was expected to increase the efficiency of transglycosylation. Upon the site-directed mutagenesis to the predicted substitutions of the amino acid residues, genes encoding the wild type TmGalA and its mutants were expressed in E. coli, and the corresponding enzymes were isolated and tested for the presence of the transglycosylating activity in synthesis of different pNP-digalactosides. Three mutants, F328A, P402D, and G385L, were shown to markedly increase the total transglycosylation as compared to the wild type enzyme. Moreover, the F328A mutant displayed an ability to produce a regio-isomer with the (α1,2)-bond at yield 16-times higher than the wild type TmGalA.     

Characterization of the Endoribonuclease Active Site of Human Apurinic/Apyrimidinic Endonuclease 1

Apurinic/apyrimidinic endonuclease 1 (APE1) is the major mammalian enzyme in DNA base excision repair that cleaves the DNA phosphodiester backbone immediately 5′ to abasic sites. Recently, we identified APE1 as an endoribonuclease that cleaves a specific coding region of c-myc mRNA in vitro, regulating c-myc mRNA level and half-life in cells. Here, we further characterized the endoribonuclease activity of APE1, focusing on the active-site center of the enzyme previously defined for DNA nuclease activities. We found that most site-directed APE1 mutant proteins (N68A, D70A, Y171F, D210N, F266A, D308A, and H309S), which target amino acid residues constituting the abasic DNA endonuclease active-site pocket, showed significant decreases in endoribonuclease activity. Intriguingly, the D283N APE1 mutant protein retained endoribonuclease and abasic single-stranded RNA cleavage activities, with concurrent loss of apurinic/apyrimidinic (AP) site cleavage activities on double-stranded DNA and single-stranded DNA (ssDNA). The mutant proteins bound c-myc RNA equally well as wild-type (WT) APE1, with the exception of H309N, suggesting that most of these residues contributed primarily to RNA catalysis and not to RNA binding. Interestingly, both the endoribonuclease and the ssRNA AP site cleavage activities of WT APE1 were present in the absence of Mg²⁺, while ssDNA AP site cleavage required Mg²⁺ (optimally at 0.5-2.0 mM). We also found that a 2′-OH on the sugar moiety was absolutely required for RNA cleavage by WT APE1, consistent with APE1 leaving a 3′-PO₄²⁻ group following cleavage of RNA. Altogether, our data support the notion that a common active site is shared for the endoribonuclease and other nuclease activities of APE1; however, we provide evidence that the mechanisms for cleaving RNA, abasic single-stranded RNA, and abasic DNA by APE1 are not identical, an observation that has implications for unraveling the endoribonuclease function of APE1 in vivo.     

Structural insight into the mechanism of substrate specificity and catalytic activity of an HD-domain phosphohydrolase: the 5'-deoxyribonucleotidase YfbR from Escherichia coli

HD-domain phosphohydrolases have nucleotidase and phosphodiesterase activities and play important roles in the metabolism of nucleotides and in signaling. We present three 2.1-Å-resolution crystal structures (one in the free state and two complexed with natural substrates) of an HD-domain phosphohydrolase, the Escherichia coli 5′-nucleotidase YfbR. The free-state structure of YfbR contains a large cavity accommodating the metal-coordinating HD motif (H33, H68, D69, and D137) and other conserved residues (R18, E72, and D77). Alanine scanning mutagenesis confirms that these residues are important for activity. Two structures of the catalytically inactive mutant E72A complexed with Co²⁺ and either thymidine-5′-monophosphate or 2′-deoxyriboadenosine-5′-monophosphate disclose the novel binding mode of deoxyribonucleotides in the active site. Residue R18 stabilizes the phosphate on the Co²⁺, and residue D77 forms a strong hydrogen bond critical for binding the ribose. The indole side chain of W19 is located close to the 2′-carbon atom of the deoxyribose moiety and is proposed to act as the selectivity switch for deoxyribonucleotide, which is supported by comparison to YfdR, another 5′-nucleotidase in E. coli. The nucleotide bases of both deoxyriboadenosine-5′-monophosphate and thymidine-5′-monophosphate make no specific hydrogen bonds with the protein, explaining the lack of nucleotide base selectivity. The YfbR E72A substrate complex structures also suggest a plausible single-step nucleophilic substitution mechanism. This is the first proposed molecular mechanism for an HD-domain phosphohydrolase based directly on substrate-bound crystal structures.     

A sulfated glucuronofucan containing both fucofuranose and fucopyranose residues from the brown alga Chordaria flagelliformis

A fucoidan fraction composed of l-fucose, sulfate, and d-glucuronic acid in a molar proportion of about 1:1:0.25 and small amount of acetyl groups was isolated from the brown alga Chordaria flagelliformis. Several modified polysaccharides were prepared from the native fucoidan using solvolytic desulfation, carboxyl reduction, and partial acid hydrolysis. Polysaccharide structures were elucidated by methylation analysis and 1D and 2D NMR spectroscopy. The fucoidan was shown to contain a backbone of 3-linked α-l-fucopyranose residues, about one-third of which are glycosylated at C-2 by α-d-glucopyranosyluronic acid residues. About half of the latter residues are glycosylated at C-4 by single α-l-fucofuranose residues or by disaccharides α-l-Fucf-(1→2)-α-l-Fucf-(1→. Fucofuranose residues are mono- and disulfated at different positions, whereas some additional sulfate groups occupy C-2 and C-4 of the backbone, the latter position being also partially acetylated.     

Functional Importance of Bacteriophage ?29 DNA Polymerase Residue Tyr148 in Primer-terminus Stabilisation at the 3′-5′ Exonuclease Active Site

Recent crystallographic resolution of ?29 DNA polymerase complexes with ssDNA at its 3′-5′ exonuclease active site has allowed the identification of residues Pro129 and Tyr148 as putative ssDNA ligands, the latter being conserved in the Kx₂h motif of proofreading family B DNA polymerases. Single substitution of ?29 DNA polymerase residue Tyr148 to Ala rendered an enzyme with a reduced capacity to stabilize the binding of the primer terminus at the 3′-5′ exonuclease active site, not having a direct role in the catalysis of the reaction. Analysis of the 3′-5′ exonuclease on primer/template structures showed a critical role for residue Tyr148 in the proofreading of DNA polymerisation errors. In addition, Tyr148 is not involved in coupling polymerisation to strand displacement in contrast to the catalytic residues responsible for the exonuclease reaction, its role being restricted to stabilisation of the frayed 3′ terminus at the exonuclease active site. Altogether, the results lead us to extend the consensus sequence of the above motif of proofreading family B DNA polymerases into Kx₂hxA. The different solutions adopted by proofreading DNA polymerases to stack the 3′ terminus at the exonuclease site are discussed. In addition, the results obtained with mutants at ?29 DNA polymerase residue Pro129 allow us to rule out a functional role as ssDNA ligand for this residue.     

Crystal structure of mSULT1D1, a mouse catecholamine sulfotransferase

In mammals, sulfonation as mediated by specific cytosolic sulfotransferases (SULTs) plays an important role in the homeostasis of dopamine and other catecholamines. To gain insight into the structural basis for dopamine recognition/binding, we determined the crystal structure of a mouse dopamine-sulfating SULT, mouse SULT1D1 (mSULT1D1). Data obtained indicated that mSULT1D1 comprises of a single α/β domain with a five-stranded parallel β-sheet. In contrast to the structure of the human SULT1A3 (hSULT1A3)-dopamine complex previously reported, molecular modeling and mutational analysis revealed that a water molecule plays a critical role in the recognition of the amine group of dopamine by mSULT1D1. These results imply differences in substrate binding between dopamine-sulfating SULTs from different species.     

The structure of the O-glycosylically-linked oligosaccharide chains of LPG-I,a glycoprotein present in articular lubricating fraction of bovine synovial fluid

Periodate oxidation of LPG-1 established that N-acetylneuraminic acid residues are linked preponderantly α-(2→3) to D-galactose residues. The resistance of 2-acetamido-2-deoxyD-galactose residues to periodate oxidation suggests that they are linked at either O-3 or O-4 to D-galactose residues. After treatment of LPG-I with alkaline sulfite, ≈80% of 2-acetamido-2-deoxygalactose was recovered as the sulfonic acid derivative. The Gal→GalNAc disaccharide released from sialic-acid-free LPG-I by digestion with endo-2-acetamido-2-deoxy-α-D-galactosidase (which suggests an α-D-GalNAc→-L-Ser or -L-Thr linkage) gave a high color-yield in the Morgan—Elson reaction, indicating that 2-acetamido-2-deoxy-D-galactose residues are linked at C-3 to D-galactose residues. The migration of the released Gal-GalNAc disaccharide was the same as that of a standard sample of O-β-D-galactosyl-(1→3)-2-acetamido-2-deoxy-D-galactose. Treatment of sialic acid-free LPG-I with Streptococcus pneumoniae β-D-galactosidase, which hydrolyzes only galactosides linked β-D-(1→4) gave no free D-galactose, whereas treatment of LPG-I with bovine testes β-D-galactosidase released > 90% of D-galactose. These results provide evidence for β-D-Galp-(1→3)-α-D-GalNAcp-(1→3)-L-Ser or -L-Thr and α-NeuAc-(2→3)-β-D-Galp-(1→3)-α-D- GalNAcp-(1→3)-L-Ser or -L-Thr structures. The sensitivity of the methods used and the recovery of constituents following treatment of LPG-I do not rule out the occurrence of small amounts of other tri- or tetra-saccharide chains.