The glycan alphabet is not universal: a hypothesis

Several monosaccharides constitute naturally occurring glycans, but it is uncertain whether they constitute a universal set like the alphabets of proteins and DNA. Based on the available experimental observations, it is hypothesized herein that the glycan alphabet is not universal. Data on the presence/absence of pathways for the biosynthesis of 55 monosaccharides in 12 939 completely sequenced archaeal and bacterial genomes are presented in support of this hypothesis. Pathways were identified by searching for homologues of biosynthesis pathway enzymes. Substantial variations were observed in the set of monosaccharides used by organisms belonging to the same phylum, genera and even species. Monosaccharides were grouped as common, less common and rare based on their prevalence in Archaea and Bacteria. It was observed that fewer enzymes are sufficient to biosynthesize monosaccharides in the common group. It appears that the common group originated before the formation of the three domains of life. In contrast, the rare group is confined to a few species in a few phyla, suggesting that these monosaccharides evolved much later. Fold conservation, as observed in aminotransferases and SDR (short-chain dehydrogenase reductase) superfamily members involved in monosaccharide biosynthesis, suggests neo- and sub-functionalization of genes led to the formation of the rare group monosaccharides. The non-universality of the glycan alphabet begets questions about the role of different monosaccharides in determining an organism’s fitness.

Monosaccharide Biosynthesis Pathways Database

A distinctive feature of glycans vis-à-vis proteins and nucleic acids is its structural complexity which arises from the huge repertoire of monosaccharides, isomeric linkages and branching. A very large number of monosaccharides have so far been discovered in natural glycans. Experimentally, pathways for the biosynthesis have been characterized completely for 55 monosaccharides and partially for a few more. However, there is no single platform which provides information about monosaccharide biosynthesis pathways and associated enzymes We have gathered 572 experimentally characterized enzymes of 66 biosynthesis pathways from literature and set up a first of its kind database called the Monosaccharide Biosynthesis Pathways Database http://www.bio.iitb.ac.in/mbpd/). Annotations such as the reaction catalysed, substrate specificity, biosynthesis pathway and PubMed IDs are provided for all the enzymes in the database. Sequence homologs of the experimentally characterized enzymes found in nearly 13,000 completely sequenced genomes from Bacteria and Archaea have also been included in the database. This platform will help in the deduction of evolutionary relationships among enzymes such as aminotransferases, nucleotidyltransferases, acetyltransferases and SDR family enzymes. It can also facilitate experimental studies such as direct enzyme assays to validate putative annotations, establish structure-function relationship, expression profiling to determine the function, determine the phenotypic consequences of gene knock-out/knock-in and complementation studies.

Characterization of left-handed beta helix-domains, and identification and functional annotation of proteins containing such domains

Only about 0.3% of the entries in UniProt database have manually curated annotation. Annotation at the molecular level often relies on low-throughput one-protein-at-a-time approach. Computational methods bridge this gap by assigning function based on sequence and/or fold similarity. Left-handed beta helix (LbH) consists of three repeating six-stranded beta-strands forming an 18-mer turn of the helix. Analysis of LbH-domains showed that variations are found in the number of residues in a beta-strand (5-7, 6 being the most common), number of turns (4-10) of the helix, insertions of one or more loops of variable length (0-36 residues), and the location of loop insertion. An 18-mer HMM profile was created which identifies LbH-domain containing proteins using sequence as the only input; the number of false positives is zero when proteins tested were those with known 3D structures. 136 474 entries of TrEMBL database were found to contain LbH-domain. Rules developed by analyzing LbH-domain containing acyltransferases, gamma-class carbonic anhydrases, and nucleotidyltransferases have led to the annotation of 17 389 TrEMBL entries which currently have no functional tag.

EpsN from Bacillus subtilis 168 has UDP-2,6-dideoxy 2-acetamido 4-keto glucose aminotransferase activity in vitro

The gene epsN of Bacillus subtilis 168 was cloned and overexpressed in Escherichia coli. Purified recombinant EpsN is shown to be a pyridoxal 5'-phosphate (PLP)-dependent aminotransferase by absorption spectroscopy, l-cycloserine inhibition and reverse phase HPLC studies. EpsN catalyzes the conversion of UDP-2,6-dideoxy 2-acetamido 4-keto glucose to UDP-2,6-dideoxy 2-acetamido 4-amino glucose. Lys190 was found by sequence comparison and site-directed mutagenesis to form Schiff base with PLP. Mutagenesis studies showed that, in addition to Lys190, Ser185, Glu164, Gly58 and Thr59 are essential for aminotransferase activity.

EpsM from Bacillus subtilis 168 has UDP-2,4,6-trideoxy-2-acetamido-4-amino glucose acetyltransferase activity in vitro

Bacillus subtilis 168 EpsM (UniProt id P71063) has been electronically annotated as putative acetyltransferase in the UniProt database. The gene epsM was cloned and overexpressed in E. coli with an N-terminal GST tag. The purified fusion protein was shown by absorption spectroscopy, autoradiography and reverse phase HPLC to catalyse the conversion of UDP-2,4,6-trideoxy-2-acetamido-4-amino glucose to UDP-2,4,6-trideoxy-2,4-diacetamido glucose, commonly known as N,N'-diacetylbacillosamine, using acetyl coenzyme A as the donor substrate. His146 was shown by site-directed mutagenesis to be essential for acetyltransferase activity. It is hypothesized that EpsC (NAD⁺ dependent UDP GlcNAc 4,6-dehydratase), EpsN (PLP dependent aminotransferase) and EpsM, all of which are part of the eps operon, are involved in the biosynthesis of N,N'-diacetylbacillosamine.

In vitro characterization of N-terminal truncated EpsC from Bacillus subtilis 168, a UDP-N-acetylglucosamine 4,6-dehydratase

Bacillus subtilis 168 EpsC is annotated as "Probable polysaccharide biosynthesis protein" in the SwissProt database. epsC is part of the eps operon, thought to be involved in the biosynthesis of exopolymeric substances (EPS). The present study was undertaken to determine the molecular function of EpsC. Sequence analysis of EpsC suggested the presence of a transmembrane domain. Two N-terminal deletion mutants in which residues 1-89 (EpsC₈₉) and 1-115 (EpsC₁₁₅) are deleted were cloned and overexpressed. Enzyme activity and substrate preferences were investigated by reverse phase HPLC, surface plasmon resonance (SPR) spectroscopy and absorption spectroscopy. These data show that EpsC has UDP-GlcNAc 4,6-dehydratase activity in vitro. Purified recombinant proteins were found to utilise UDP-Glc and TDP-Glc also as substrates. In addition, EpsC₁₁₅ could utilise UDP-Gal and UDP-GalNAc as substrates whereas EpsC₈₉ could only bind these two sugar nucleotides. These results show that deletion of a longer N-terminal region broadens substrate specificity. These broadened specificity is perhaps an outcome of the deletion of the putative transmembrane domain and may not be present in vivo. EpsC, together with the aminotransferase EpsN (Kaundinya CR et al., Glycobiology, 2018) and acetyltransferase EpsM (unpublished data), appears to be involved in the biosynthesis of N,N'-diacetylbacillosamine.

Rv3634c from Mycobacterium tuberculosis H37Rv encodes an enzyme with UDP-Gal/Glc and UDP-GalNAc 4-epimerase activities

A bioinformatics study revealed that Mycobacterium tuberculosis H37Rv (Mtb) contains sequence homologs of Campylobacter jejuni protein glycosylation enzymes. The ORF Rv3634c from Mtb was identified as a sequence homolog of C. jejuni UDP-Gal/GalNAc 4-epimerase. This study reports the cloning of Rv3634c and its expression as an N-terminal His-tagged protein. The recombinant protein was shown to have UDP-Gal/Glc 4-epimerase activity by GOD-POD assay and by reverse phase HPLC. This enzyme was shown to have UDP-GalNAc 4-epimerase activity also. Residues Ser121, Tyr146 and Lys150 were shown by site-directed mutagenesis to be important for enzyme activity. Mutation of Ser121 and Tyr146 to Ala and Phe, respectively, led to complete loss of activity whereas mutation of Lys150 to Arg led to partial loss of activity. There were no gross changes in the secondary structures of any of these three mutants. These results suggest that Ser121 and Tyr146 are essential for epimerase activity of Rv3634c. UDP-Gal/Glc 4-epimerases from other organisms also have a catalytic triad consisting of Ser, Tyr and Lys. The triad carries out proton transfer from nucleotide sugar to NAD⁺ and back, thus effecting the epimerization of the substrate. Addition of NAD⁺ to Lys150 significantly abrogates the loss of activity, suggesting that, as in other epimerases, NAD⁺ is associated with Rv3634c.

Interplay of sequence, topology and termini charge in determining the stability of the aggregates of GNNQQNY mutants: a molecular dynamics study

This study explores the stabilities of single sheet parallel systems of three sequence variants of ¹GNNQQNY⁷, N2D, N2S and N6D, with variations in aggregate size (5–8) and termini charge (charged or neutral). The aggregates were simulated at 300 and 330 K. These mutations decrease amyloid formation in the yeast prion protein Sup35. The present study finds that these mutations cause instability even in the peptide context. The protonation status of termini is found to be a key determinant of stabilities; other determinants are sequence, position of mutation and aggregate size. All systems with charged termini are unstable, whereas both stable and unstable systems are found when the termini are neutral. When termini are charged, the largest stable aggregate for the N2S and N6D systems has 3 to 4 peptides whereas N2D mutation supports oligomers of larger size (5-and 6-mers) as well. Mutation at 2^nd position (N2S and N2D) results in fewer H-bonds at the mutated as well as neighboring (Gly1/Gln4) positions. However, no such effect is found if mutation is at 6^th position (N6D). The effect of Asn→Asp mutation depends on the position and termini charge: it is more destabilizing at the 2^nd position than at the 6^th in case of neutral termini, however, the opposite is true in case of charged termini. Appearance of twist in stable systems and in smaller aggregates formed in unstable systems suggests that twist is integral to amyloid arrangement. Disorder, dissociation or rearrangement of peptides, disintegration or collapse of aggregates and formation of amorphous aggregates observed in these simulations are likely to occur during the early stages of aggregation also. The smaller aggregates formed due to such events have a variety of arrangements of peptides. This suggests polymorphic nature of oligomers and presence of a heterogeneous mixture of oligomers during early stages of aggregation.

Comparative genomics analysis of completely sequenced microbial genomes reveals the ubiquity of N-linked glycosylation in prokaryotes

Glycosylation of proteins in prokaryotes has been known for the last few decades. Glycan structures and/or the glycosylation pathways have been experimentally characterized in only a small number of prokaryotes. Even this has become possible only during the last decade or so, primarily due to technological and methodological developments. Glycosylated proteins are diverse in their function and localization. Glycosylation has been shown to be associated with a wide range of biological phenomena. Characterization of the various types of glycans and the glycosylation machinery is critical to understand such processes. Such studies can help in the identification of novel targets for designing drugs, diagnostics, and engineering of therapeutic proteins. In view of this, the experimentally characterized pgl system of Campylobacter jejuni, responsible for N-linked glycosylation, has been used in this study to identify glycosylation loci in 865 prokaryotes whose genomes have been completely sequenced. Results from the present study show that only a small number of organisms have homologs for all the pgl enzymes and a few others have homologs for none of the pgl enzymes. Most of the organisms have homologs for only a subset of the pgl enzymes. There is no specific pattern for the presence or absence of pgl homologs vis-à-vis the 16S rRNA sequence-based phylogenetic tree. This may be due to differences in the glycan structures, high sequence divergence, horizontal gene transfer or non-orthologous gene displacement. Overall, the presence of homologs for pgl enzymes in a large number of organisms irrespective of their habitat, pathogenicity, energy generation mechanism, etc., hints towards the ubiquity of N-linked glycosylation in prokaryotes.

Quantification of binding affinities of essential sugars with a tryptophan analogue and the ubiquitous role of C-H···π interactions

The role of noncovalent interactions in carbohydrate recognition by aromatic amino acids has long been reported. To develop a molecular understanding of noncovalent interactions in the recognition process, we have examined a series of binary complexes between 3-methylindole (3-MeIn) and sugars. In particular, the geometries and binding affinities of 3-MeIn with α/β-D-glucose, β-D-galactose, α-D-mannose and α/β-L-fucose are obtained using the MP2(full)/6-31G(d,p) and the M06/TZV2D//MP2/6-31G(d,p) level of theories. The conventional hydrogen bonding such as N-H···O and C-H···O as well as nonconventional O-H···π and C-H···π type of interactions is, in general, identified as responsible for the moderately strong interaction energies. Large variations in the position-orientations of 3-MeIn with respect to saccharide are noticed, within the same sugar family, as well as across different sugar series. Furthermore, complexes with large differences in their geometries are recognized as capable of exhibiting very similar interaction energies, underscoring the significance of exhaustive conformation sampling, as carried out in the present study. These observations are readily attributed to the differences in the efficiency of the type of interactions enlisted above. The highest and lowest interaction energies, upon inclusion of 50% BSSE correction, are found to be -16.02 and -6.22 kcal mol(-1), respectively, for α-D-glucose (1a) and α-L-fucose (5j). While more number of prominent conventional hydrogen bonding contacts remains as a characteristic feature of the strongly bound complexes, the lower end of the interaction energy spectrum is dominated by multiple C-H···π interactions. The complexes exhibiting as many as four C-H···π contacts are identified in the case of α/β-D-glucose, β-D-galactose, and α/β-L-fucose with an interaction energy hovering around -8 kcal mol(-1). The presence of effective C-H···π interactions is found to be dependent on the saccharide configuration as well as the area of the apolar patch constituted by the C-H groups. The study offers a comprehensive set of binary complexes, across different saccharides, which serves as an illustration of the significance and ubiquitous nature of C-H···π interactions in carbohydrate binding in saccharide-protein complexes.

E93R substitution of Escherichia coli FtsZ induces bundling of protofilaments, reduces GTPase activity, and impairs bacterial cytokinesis

Recently, we found that divalent calcium has no detectable effect on the assembly of Mycobacterium tuberculosis FtsZ (MtbFtsZ), whereas it strongly promoted the assembly of Escherichia coli FtsZ (EcFtsZ). While looking for potential calcium binding residues in EcFtsZ, we found a mutation (E93R) that strongly promoted the assembly of EcFtsZ. The mutation increased the stability and bundling of the FtsZ protofilaments and produced a dominating effect on the assembly of the wild type FtsZ (WT-FtsZ). Although E93R-FtsZ was found to bind to GTP similarly to the WT-FtsZ, it displayed lower GTPase activity than the WT-FtsZ. E93R-FtsZ complemented for its wild type counterpart as observed by a complementation test using JKD7-1/pKD3 cells. However, the bacterial cells became elongated upon overexpression of the mutant allele. We modeled the structure of E93R-FtsZ using the structures of MtbFtsZ/Methanococcus jannaschi FtsZ (MjFtsZ) dimers as templates. The MtbFtsZ-based structure suggests that the Arg(93)-Glu(138) salt bridge provides the additional stability, whereas the effect of mutation appears to be indirect (allosteric) if the EcFtsZ dimer is similar to that of MjFtsZ. The data presented in this study suggest that an increase in the stability of the FtsZ protofilaments is detrimental for the bacterial cytokinesis.

Length and composition analysis of the cytoplasmic, transmembrane and stem regions of human Golgi glycosyltransferases

A dataset of experimentally characterized, human Golgi GlyTs with type II membrane topology was created. Based on the experimentally observed acceptor substrate preferences, the GlyTs were classified into five functional categories: biosynthesis of blood group antigens, glycolipids, N-glycans, O-glycans and glycosaminoglycans. The cytoplasmic, transmembrane and stem (CTS) regions were predicted and their length and composition were analyzed. The stem region of GlyTs involved in the biosynthesis of glycolipids and blood group antigens appear to have a shorter stem region compared to those GlyTs which participate in the biosynthesis of N- and O-linked glycans and glycosaminoglycans. The stem regions of all the GlyTs, irrespective of the functional category to which they belong, were found to be rich in disorder-promoting amino acid residues. Thus, the stem region is largely devoid of any regular secondary structure thereby facilitating its tethering role. A higher frequency of occurrence of basic amino acids is observed towards the N-terminus of the transmembrane domain and this is suggested to be important for topogenesis of these enzymes.

Characterization of the conformational and orientational dynamics of ganglioside GM1 in a dipalmitoylphosphatidylcholine bilayer by molecular dynamics simulations

The structure and dynamics of a single GM1 (Gal5-beta1,3-GalNAc4-beta1,4-(NeuAc3-alpha2,3)-Gal2-beta1,4-Glc1-beta1,1-Cer) embedded in a DPPC bilayer have been studied by MD simulations. Eleven simulations, each of 10 ns productive run, were performed with different initial conformations of GM1. Simulations of GM1-Os in water and of a DPPC bilayer were also performed to delineate the effects of the bilayer and GM1 on the conformational and orientational dynamics of each other. The conformation of the GM1 headgroup observed in the simulations is in agreement with those reported in literature; but the headgroup is restricted when embedded in the bilayer. NeuAc3 is the outermost saccharide towards the water phase. Glc1 and Gal2 prefer a parallel, and NeuAc3, GalNac4 and Gal5 prefer a perpendicular, orientation with respect to the bilayer normal. The overall characteristics of the bilayer are not affected by the presence of GM1; however, GM1 does influence the DPPC molecules in its immediate vicinity. The implications of these observations on the specific recognition and binding of GM1 embedded in a lipid bilayer by exogenous proteins as well as proteins embedded in lipids have been discussed.

The sequence TGAAKAVALVL from glyceraldehyde-3-phosphate dehydrogenase displays structural ambivalence and interconverts between α-helical and β-hairpin conformations mediated by collapsed conformational states

The peptide TGAAKAVALVL from glyceraldehyde-3-phosphate dehydrogenase adopts a helical conformation in the crystal structure and is a site for two hydrated helical segments, which are thought to be helical folding intermediates. Overlapping sequences of four to five residues from the peptide, sample both helical and strand conformations in known protein structures, which are dissimilar to glyceraldehyde-3-phosphate dehydrogenase suggesting that the peptide may have a structural ambivalence. Molecular dynamics simulations of the peptide sequence performed for a total simulation time of 1.2 micros, starting from the various initial conformations using GROMOS96 force field under NVT conditions, show that the peptide samples a large number of conformational forms with transitions from alpha-helix to beta-hairpin and vice versa. The peptide, therefore, displays a structural ambivalence. The mechanism from alpha-helix to beta-hairpin transition and vice versa reveals that the compact bends and turns conformational forms mediate such conformational transitions. These compact structures including helices and hairpins have similar hydrophobic radius of gyration (Rgh) values suggesting that similar hydrophobic interactions govern these conformational forms. The distribution of conformational energies is Gaussian with helix sampling lowest energy followed by the hairpins and coil. The lowest potential energy of the full helix may enable the peptide to take up helical conformation in the crystal structure of the glyceraldehyde-3-phosphate dehydrogenase, even though the peptide has a preference for hairpin too. The relevance of folding and unfolding events observed in our simulations to hydrophobic collapse model of protein folding are discussed.

Correlation between the structural stability and aggregation propensity of proteins

Protein aggregation, being an outcome of improper protein folding, is largely dependent on the folding kinetics of a protein. Previous studies have reported a positive correlation between the stability of the secondary structural elements of a protein and their rate of folding/unfolding. In this in silico study, the secondary and tertiary structures of proteins a) that form inclusion bodies on overexpression in Escherichia coli, b) that form amyloid fibrils and c) that are soluble on overexpression in E. coli are analyzed for certain features that are known to be associated with structural stability. The study revealed that the soluble proteins seem to have a higher rate of folding (based on contact order) and a lower percentage of exposed hydrophobic residues as compared to the inclusion body forming or amyloidogenic proteins. The soluble proteins also seem to have a more favored helix and strand composition (based on the known secondary structural propensities of amino acids). The secondary structure analyses also reveal that the evolutionary pressure is directed against protein aggregation. This understanding of the positive correlation between structural stability and solubility, along with the other parameters known to influence aggregation, could be exploited in the design of mutations aimed at reducing the aggregation propensity of the proteins.

Fold-recognition and comparative modeling of human beta3GalT I, II, IV, V and VI and beta3GalNAcT I: prediction of residues conferring acceptor substrate specificity

beta3GalTs are type II transmembrane proteins that transfer galactose from UDP-Gal donor substrate to acceptor GlcNAc, GalNAc or Gal in beta1-->3-linkage. beta1-->3-linked galactose have been found to be a part of many glycans like glycosphingolipids, core tetrasaccharide of proteoglycans, type 1 chains. The 3-D structure of none of the beta3GalTs is known to date. In this study, the 3-D structures of human beta3GalT I, II, IV, V, VI and beta3GalNAcT I have been modeled using fold-recognition and comparative modeling methods. Residues that constitute the UDP-Gal binding site have been predicted. The models are able to qualitatively rationalize data from the site-directed mutagenesis experiments reported in the literature. Residues likely to be involved in conferring differential acceptor substrate specificity have been predicted by a combination of specificity determining positions prediction (SDPs) and subsequent mapping on the generated 3-D models.

β-Hairpins with native-like and non-native hydrogen bonding patterns could form during the refolding of staphylococcal nuclease

Refolding of staphylococcal nuclease has been studied recently by hydrogen-deuterium exchange and NMR spectroscopy. These studies infer that beta-hairpin formed by strand 2 and strand 3 connected by reverse turn forms early during the refolding of nuclease. Typically, hydrogen-deuterium exchange NMR techniques are usually carried out on a time scale of milliseconds whereas beta-hairpins are known to fold on a much shorter time scale. It follows that in the experiments, the hydrogen-deuterium exchange protection patterns could be arising from a significant population of fully formed hairpins. In order to demonstrate it is the fully formed hairpins which gives rise to the hydrogen-deuterium exchange protection patterns, we have considered molecular dynamics simulation of the peptide (21)DTVKLMYKGQPMTFR(35) from staphylococcal nuclease corresponding to the beta-hairpin region, using GROMOS96 force field under NVT conditions. Starting from unfolded conformational states, the peptide folds into hairpin conformations with native-like and non-native hydrogen bonding patterns. Subsequent to folding, equilibrium conditions prevail. The computed protection factors and atom depth values, at equilibrium, of the various amide protons agree qualitatively with experimental observations. A collection of molecules following the trajectories observed in the simulations can account for experimental observations. These simulations provide a molecular picture of the formed hairpins and their conformational features during the refolding experiments on nuclease, monitored by hydrogen-deuterium exchange.

Fold-recognition and comparative modeling of human alpha2,3-sialyltransferases reveal their sequence and structural similarities to CstII from Campylobacter jejuni

Background

The 3-D structure of none of the eukaryotic sialyltransferases (SiaTs) has been determined so far. Sequence alignment algorithms such as BLAST and PSI-BLAST could not detect a homolog of these enzymes from the protein databank. SiaTs, thus, belong to the hard/medium target category in the CASP experiments. The objective of the current work is to model the 3-D structures of human SiaTs which transfer the sialic acid in α2,3-linkage viz., ST3Gal I, II, III, IV, V, and VI, using fold-recognition and comparative modeling methods. The pair-wise sequence similarity among these six enzymes ranges from 41 to 63%.

Results

Unlike the sequence similarity servers, fold-recognition servers identified CstII, a α2,3/8 dual-activity SiaT from Campylobacter jejuni as the homolog of all the six ST3Gals; the level of sequence similarity between CstII and ST3Gals is only 15–20% and the similarity is restricted to well-characterized motif regions of ST3Gals. Deriving template-target sequence alignments for the entire ST3Gal sequence was not straightforward: the fold-recognition servers could not find a template for the region preceding the L-motif and that between the L- and S-motifs. Multiple structural templates were identified to model these regions and template identification-modeling-evaluation had to be performed iteratively to choose the most appropriate templates. The modeled structures have acceptable stereochemical properties and are also able to provide qualitative rationalizations for some of the site-directed mutagenesis results reported in literature. Apart from the predicted models, an unexpected but valuable finding from this study is the sequential and structural relatedness of family GT42 and family GT29 SiaTs.

Conclusion

The modeled 3-D structures can be used for docking and other modeling studies and for the rational identification of residues to be mutated to impart desired properties such as altered stability, substrate specificity, etc. Several studies in literature have focused on the development of tools and/or servers for the large-scale/automated modeling of 3-D structures of proteins. In contrast, the present study focuses on modeling the 3-D structure of a specific protein of interest to a biochemist and illustrates the associated difficulties. It is also able to establish a sequence/structure relationship between sialyltransferases of two distinct families.

A support vector machine-based method for predicting the propensity of a protein to be soluble or to form inclusion body on overexpression in Escherichia coli

MOTIVATION

Inclusion body formation has been a major deterrent for overexpression studies since a large number of proteins form insoluble inclusion bodies when overexpressed in Escherichia coli. The formation of inclusion bodies is known to be an outcome of improper protein folding; thus the composition and arrangement of amino acids in the proteins would be a major influencing factor in deciding its aggregation propensity. There is a significant need for a prediction algorithm that would enable the rational identification of both mutants and also the ideal protein candidates for mutations that would confer higher solubility-on-overexpression instead of the presently used trial-and-error procedures.

RESULTS

Six physicochemical properties together with residue and dipeptide-compositions have been used to develop a support vector machine-based classifier to predict the overexpression status in E.coli. The prediction accuracy is approximately 72% suggesting that it performs reasonably well in predicting the propensity of a protein to be soluble or to form inclusion bodies. The algorithm could also correctly predict the change in solubility for most of the point mutations reported in literature. This algorithm can be a useful tool in screening protein libraries to identify soluble variants of proteins.

Identification of linkage-specific sequence motifs in sialyltransferases

Eukaryotic sialyltransferases (SiaTs) comprise a superfamily of enzymes catalyzing the transfer of sialic acid (Sia) from a common donor substrate to various acceptor substrates in different linkages. These enzymes have been classified as ST3Gal, ST6Gal, ST6GalNAc, and ST8Sia families based on linkage- and acceptor monosaccharide-specificities and sequence similarities. It was recognized early on that SiaTs contain certain well-conserved motifs, and these were denoted as L (large)-, S (small)-, and VS (very small)-motifs; recently, a fourth motif, denoted as motif III, was identified. These four motifs are common to all the SiaTs, irrespective of the linkage- and acceptor saccharide-specificities. In this study, the sequences of the various families have been analyzed, and sequence motifs that are unique to the various families have been identified. These unique motifs are expected to contribute to the characteristic linkage- and acceptor saccharide-specificities of the family members. One of the linkage specific motifs is contiguous to L-motif. Members of ST3Gal and ST8Sia families share significant sequence similarities; in contrast, the ST6Gal family is distinct from the ST6GalNAc family. The latter consists of two subfamilies, one comprising ST6GalNAc I and ST6GalNAc II, and the other comprising ST6GalNAc III, ST6GalNAc IV, ST6GalNAc V, and ST6GalNAc VI. Each of these subfamilies has characteristic sequence motifs not present in the other subfamily.

Effect of the Choice of the Pressure Coupling Method on the Spontaneous Aggregation of DPPC Molecules

The self-assembly of DPPC molecules starting from a random, solution-like configuration in the presence of water molecules is described in the present MD simulation study. Simulations were performed with either anisotropic or isotropic pressure coupling. Use of anisotropic pressure coupling led to the formation of a bilayer/bilayerlike aggregate; the features of the bilayer are in agreement with those reported from earlier simulation and experimental studies. In contrast, simulating the same system with isotropic pressure coupling led to the formation of a cylindrical micelle/lamellar structure with a large water hole. The formation of micelles seems unrealistic since diacylphosphatidylcholines having hydrocarbon tails with nine or more carbon atoms have been shown to form only bilayers. Simulations were also performed with preformed bilayerlike configurations with either anisotropic or isotropic pressure coupling. The bilayer characteristics deduced from simulations using anisotropic pressure coupling are in better agreement with those reported from earlier experimental and simulation studies. Thus, the choice of the pressure coupling method has a significant effect on the spontaneous aggregation of DPPC molecules but makes relatively lesser effect if the bilayer has formed already.

Insights into the Role of the Aromatic Residue in Galactose-Binding Sites: MP2/6-311G++** Study on Galactose− and Glucose−Aromatic Residue Analogue Complexes

The presence of an aromatic residue (Trp, Phe, Tyr) facing the nonpolar face of galactose is a common feature of galactose-specific lectins. The interactions such as those between the C-H groups of galactose and the pi-electron cloud of aromatic residues have been characterized as weak hydrogen bonds between soft acids and soft bases, largely governed by dispersive and charge transfer interactions. An analysis of the binding sites of several galactose-specific lectins revealed that the spatial position-orientation of galactose relative to the binding site aromatic residue varies substantially. The effect of variations in position-orientations of galactose on the interaction energies of galactose-aromatic residue complexes has not been determined so far. In view of this, MP2/6-311G++** calculations were performed on galactose- and glucose-aromatic residue analogue complexes in eight position-orientations. The results show that the strength of the C-H...pi interactions in galactose-aromatic residue complexes is comparable to that of a hydrogen bond. Rather than the type of aromatic residue, the position-orientation of the saccharide appears to be more critical in determining the strength of their interactions. Earlier studies have found the binding site aromatic residue to be critical, but its role was not clear. This study shows that the aromatic residue is important for discriminating galactose from glucose, in addition to its contribution to binding energy.

Understanding the relationship between the primary structure of proteins and their amyloidogenic propensity: clues from inclusion body formation

Amyloid formation is dependent to a considerable extent on the amino acid sequence of the protein. The present study delineates certain sequence-dependent features that are correlated with amyloidogenic propensity. The analyses indicate that amyloid formation is favored by lower thermostability and increased half-life of the protein. There seems to be a certain degree of bias in the composition of order-promoting amino acids in the case of amyloidogenic proteins. Based on these parameters, a prediction function for the amyloidogenic propensity of proteins has been created. The prediction function has been found to rationalize the reported effect of certain mutations on amyloid formation. It seems that a higher sheet propensity of residues that constitute the first seven residues of a helical structure in a protein might increase the propensity for a helix to sheet transition in that region under denaturing conditions.

Energetics of galactose- and glucose-aromatic amino acid interactions: implications for binding in galactose-specific proteins

An aromatic amino acid is present in the binding site of a number of sugar binding proteins. The interaction of the saccharide with the aromatic residue is determined by their relative position as well as orientation. The position-orientation of the saccharide relative to the aromatic residue was found to vary in different sugar-binding proteins. In the present study, interaction energies of the complexes of galactose (Gal) and of glucose (Glc) with aromatic residue analogs have been calculated by ab initio density functional (U-B3LYP/ 6-31G**) theory. The position-orientations of the saccharide with respect to the aromatic residue observed in various Gal-, Glc-, and mannose-protein complexes were chosen for the interaction energy calculations. The results of these calculations show that galactose can interact with the aromatic residue with similar interaction energies in a number of position-orientations. The interaction energy of Gal-aromatic residue analog complex in position-orientations observed for the bound saccharide in Glc/Man-protein complexes is comparable to the Glc-aromatic residue analog complex in the same position-orientation. In contrast, there is a large variation in interaction energies of complexes of Glc- and of Gal- with the aromatic residue analog in position-orientations observed in Gal-protein complexes. Furthermore, the conformation wherein the O6 atom is away from the aromatic residue is preferred for the exocyclic -CH2OH group in Gal-aromatic residue analog complexes. The implications of these results for saccharide binding in Gal-specific proteins and the possible role of the aromatic amino acid to ensure proper positioning and orientation of galactose in the binding site have been discussed.

Understanding the relationship between the primary structure of proteins and its propensity to be soluble on overexpression in Escherichia coli

Solubility of proteins on overexpression in Escherichia coli is a manifestation of the net effect of several sequence-dependent and sequence-independent factors. This study aims to delineate the relationship between the primary structure and solubility on overexpression. The amino acid sequences of proteins reported to be soluble or to form inclusion bodies on overexpression in E. coli under normal growth conditions were analyzed. The results show a positive correlation between thermostability and solubility of proteins, and an inverse correlation between the in vivo half-life of proteins and solubility. The amino acid (Asn, Thr, Tyr) composition and the tripeptide frequency of the protein were also found to influence its solubility on overexpression. The amino acids that were seen to be present in a comparatively higher frequency in inclusion body-forming proteins have a higher sheet propensity, whereas those that are seen more in soluble proteins have a higher helix propensity; this is indicative of a possible correlation between sheet propensity and inclusion body formation. Thus, the present analysis shows that thermostability, in vivo half-life, Asn, Thr, and Tyr content, and tripeptide composition of a protein are correlated to the propensity of a protein to be soluble on overexpression in E. coli. The precise mechanism by which these properties affect the solubility status of the overexpressed protein remains to be understood.

Identification of common structural features of binding sites in galactose-specific proteins

Galactose-binding proteins characterize an important subgroup of sugar-binding proteins that are involved in a variety of biological processes. Structural studies have shown that the Gal-specific proteins encompass a diverse range of primary and tertiary structures. The binding sites for galactose also seem to vary in different protein-galactose complexes. No common binding site features that are shared by the Gal-specific proteins to achieve ligand specificity are so far known. With the assumption that common recognition principles will exist for common substrate recognition, the present study was undertaken to identify and characterize any unique galactose-binding site signature by analyzing the three-dimensional (3D) structures of 18 protein-galactose complexes. These proteins belong to 7 nonhomologous families; thus, there is no sequence or structural similarity across the families. Within each family, the binding site residues and their relative distances were well conserved, but there were no similarities across families. A novel, yet simple, approach was adopted to characterize the binding site residues by representing their relative spatial dispositions in polar coordinates. A combination of the deduced geometrical features with the structural characteristics, such as solvent accessibility and secondary structure type, furnished a potential galactose-binding site signature. The signature was evaluated by incorporation into the program COTRAN to search for potential galactose-binding sites in proteins that share the same fold as the known galactose-binding proteins. COTRAN is able to detect galactose-binding sites with a very high specificity and sensitivity. The deduced galactose-binding site signature is strongly validated and can be used to search for galactose-binding sites in proteins. PROSITE-type signature sequences have also been inferred for galectin and C-type animal lectin-like fold families of Gal-binding proteins.

Crystal structure of beta1,4-galactosyltransferase complex with UDP-Gal reveals an oligosaccharide acceptor binding site

The crystal structure of the catalytic domain of bovine beta1,4-galactosyltransferase (Gal-T1) co-crystallized with UDP-Gal and MnCl(2) has been solved at 2.8 A resolution. The structure not only identifies galactose, the donor sugar binding site in Gal-T1, but also reveals an oligosaccharide acceptor binding site. The galactose moiety of UDP-Gal is found deep inside the catalytic pocket, interacting with Asp252, Gly292, Gly315, Glu317 and Asp318 residues. Compared to the native crystal structure reported earlier, the present UDP-Gal bound structure exhibits a large conformational change in residues 345-365 and a change in the side-chain orientation of Trp314. Thus, the binding of UDP-Gal induces a conformational change in Gal-T1, which not only creates the acceptor binding pocket for N-acetylglucosamine (GlcNAc) but also establishes the binding site for an extended sugar acceptor. The presence of a binding site that accommodates an extended sugar offers an explanation for the observation that an oligosaccharide with GlcNAc at the non-reducing end serves as a better acceptor than the monosaccharide, GlcNAc. Modeling studies using oligosaccharide acceptors indicate that a pentasaccharide, such as N-glycans with GlcNAc at their non-reducing ends, fits the site best. A sequence comparison of the human Gal-T family members indicates that although the binding site for the GlcNAc residue is highly conserved, the site that binds the extended sugar exhibits large variations. This is an indication that different Gal-T family members prefer different types of glycan acceptors with GlcNAc at their non-reducing ends.

Conformation, orientation and dynamics of dodecylphosphocholine in micellar aggregate: a 3.2 ns molecular dynamics simulation study

A dodecylphosphocholine micelle of 86 monomers with 5776 water molecules has been simulated under NPT conditions for 3.2 ns using GROMACS2.0. The micelle was found to be very dynamic. Some of the C-C bonds, independent of their position in the DPC monomer, adopt gauche conformation and the trans <--> gauche transitions are quite frequent. An average of about 11% of the C-C bonds in the micelle are observed to be in the gauche conformation (i.e., |dihedral angle|< 120 degrees). The terminal methyl groups are randomly distributed all over the micelle whereas the nitrogen atom of phosphocholine headgroup atoms is restricted to the interface region. Some of the monomers were found to lie on the surface. The shape of micelle, influenced by the packing considerations, shows deviations from spherical shape. The phosphocholine headgroup is well solvated and there is no water penetration into the micelle core. The overall features of the micelle of 86 DPC monomers conforms to the lattice model of micelle proposed by Dill and Flory [Dill K A, Flory P J (1981) Proc Natl Acad Sci USA 78, 676-680] and is similar to DPC micelles of smaller aggregate sizes except for the positional preference of the C-C bonds for the gauche conformation and the trans<-->gauche transition times [Tieleman D P, van der Spoel D, Berendsen H J C (2000) J Phys Chem B 104, 6380-6388; Wymore T, Gao X F, Wong T C (1999) J Mol Struct (Theochem) 485-486, 195-210]. It appears that packing considerations play a predominant role in determining the shape and dynamics of the micelle.

Molecular dynamics simulations of alpha2 --> 8-linked disialoside: conformational analysis and implications for binding to proteins

Computational methods have played a key role in elucidating the various three-dimensional structures of oligosaccharides. Such structural information, together with other experimental data, leads to a better understanding of the role of oligosaccharide in various biological processes. The disialoside Neu5Ac-alpha2-->8-Neu5Ac appears as the terminal glycan in glycoproteins and glycolipids, and is known to play an important role in various events of cellular communication. Neurotoxins such as botulinum and tetanus require Neu5Ac-alpha2 --> 8-Neu5Ac for infecting the host. Glycoconjugates containing this disialoside and the enzymes catalyzing their biosynthesis are also regulated during cell growth, development, and differentiation. Unlike other biologically relevant disaccharides that have only two linkage bonds, the alpha2-->8-linked disialoside has four: C2-O, O-C8', C8'-C7', and C7'-C6'. The present report describes the results from nine 1 ns MD simulations of alpha2-->8-linked disialoside (Neu5Ac-alpha2-->8-Neu5Ac); simulations were run using GROMOS96 by explicitly considering the solvent molecules. Conformations around the O-C8' bond are restricted to the +sc/+ap regions due to stereochemical reasons. In contrast, conformations around the C2-O and C8'-C7' bonds were found to be largely unrestricted and all the three staggered regions are accessible. The conformations around the C7'-C6' bond were found to be in either the -sc or the anti region. These results are in excellent agreement with the available NMR and potential energy calculation studies. Overall, the disaccharide is flexible and adopts mainly two ensembles of conformations differing in the conformation around the C7'-C6' bond. The flexibility associated with this disaccharide allows for better optimization of intermolecular contacts while binding to proteins and this may partially compensate for the loss of conformational entropy that may be incurred due to disaccharide's flexibility.

Molecular dynamics simulations of hybrid and complex type oligosaccharides

Conformational preferences of hybrid (GlcNAc1Man5GlcNAc2) and complex (GlcNAc1Man3GlcNAc2; GlcNAc2Man3GlcNAc2) type asparagine-linked oligosaccharides and the corresponding bisected oligosaccharides have been studied by molecular dynamics simulations for 2.5 ns. The fluctuations of the core Man-alpha 1,3-Man fragment are restricted to a region around (-30 degrees, -30 degrees) due to a 'face-to-face' arrangement of bisecting GlcNAc and the beta 1,2-GlcNAc on the 1,3-arm. However, conformations where such a 'face-to-face' arrangement is disrupted are also accessed occasionally. The orientation of the 1,6-arm is affected not only by changes in chi, but also by changes in phi and psi around the core Man-alpha 1,6-Man linkage. The conformation around the core Man-alpha 1,6-Man linkage is different in the hybrid and the two complex types suggesting that the preferred values of phi, psi, and chi are affected by the addition or deletion of saccharides to the alpha 1,6-linked mannose. The conformational data are in agreement with the available experimental studies and also explain the branch specificity of galactosyltransferases.

Functional domains of bovine beta-1,4 galactosyltransferase

A number of N- and C-terminal deletion and point mutants of bovine beta-1,4 galactosyltransferase (beta-1,4GT) were expressed in E. coli to determine the binding regions of the enzyme that interact with N-acetylglucosamine (NAG) and UDP-galactose. The N-terminal truncated forms of the enzyme between residues 1-129, do not show any significant difference in the apparent Kms towards NAG or linear oligosaccharide acceptors e.g. for chitobiose and chitotriose, or for the nucleotide donor UDP-galactose. Deletion or mutation of Cys 134 results in the loss of enzymatic activity, but does not affect the binding properties of the protein either to NAG- or UDP-agarose. From these columns the protein can be eluted with 15 mM NAG and 50 mM EDTA, like the enzymatically active protein, TL-GT129, that contains residues 130-402 of bovine beta-1,4GT. Also the N-terminus fragment, TL-GT129NAG, that contains residues 130-257 of the beta-1,4GT, binds to, and elutes with 15 mM NAG and 50 mM EDTA from the NAG-agarose column as efficiently as the enzymatically active TL-GT129. Unlike TL-GT129, the TL-GT129NAG binds to UDP-columns less efficiently and can be eluted from the column with only 15 mM NAG. The C-terminus fragment GT-257UDP, containing residues 258-402 of beta-1,4GT, binds tightly to both NAG- and UDP-agarose columns. A small fraction, 5-10% of the bound protein, can be eluted from the UDP-agarose column with 50 mM EDTA alone. The results show that the binding behaviour of N- and C-terminal fragments of beta-1,4GT towards the NAG- and UDP-agarose columns differ, the former binds preferentially to NAG-columns, while the latter binds to UDP-agarose columns via Mn2+.

Molecular dynamics simulations of oligosaccharides and their conformation in the crystal structure of lectin-carbohydrate complex: importance of the torsion angle psi for the orientation of alpha 1,6-arm

The conformation of the heptasaccharide Man-alpha 1,6-(Man-alpha 1,3)(Xyl-beta 1,2)-Man-beta 1,4-GlcNAc2-beta 1,4-(L- Fuc-alpha 1,3)-GlcNAc1, the carbohydrate moiety of Erythrina corallodendron lectin (EcorL), the hexasaccharide Man-alpha 1,6-(Man-alpha 1,3) (GlcNAc-beta 1,4)-Man-beta 1,4-GlcNAc-beta 1,4-GlcNAc and their disaccharide fragments have been studied by molecular dynamics (MD) simulations for 1000 ps with different initial conformations. In the isolated heptasaccharide, the most frequently accessed conformation during MD has a psi value of 180 degrees around Man-alpha 1,6-Man linkage. This conformation is stabilized by the formation of a hydrogen bond between the carbonyl oxygen of GlcNAc2 with the O3/O4 hydroxyls of the alpha 1,6-linked mannose residue. The conformation of the heptasaccharide found in the crystal structure of the EcorL-lactose complex (Shaanan et al., Science, 254, 862, 1991), that has a psi value of approximately 76 degrees around Man-alpha 1,6-Man linkage, is accessed, although less frequently, during MD of the isolated oligosaccharide. The phi, psi, chi = 58 degrees, -134 degrees, -60 degrees conformation around Man-alpha 1,6-Man fragment observed in the crystal structure of the Lathyrus ochrus lectin complexed with a biantennary octasaccharide (Table I in Homans, S.W., Glycobiology, 3, 551, 1993) has also been accessed in the present MD simulations. These psi values for the alpha 1,6-linkage, which are observed in the protein-carbohydrate crystal structures and are accessed in the MD simulations, though occasionally, have not been predicted from NMR studies. Furthermore, these different values of psi lead to significantly different orientations of the alpha 1,6-arm for the same value of chi. This contrasts with the earlier predictions that only different values of chi can bring about significant changes in the orientation of the alpha 1,6-arm. The MD simulations also show that the effects of bisecting GlcNAc or beta 1,2-xylose are very similar on the alpha 1,3-arm and slightly different on the alpha 1,6-arm.

Molecular dynamics simulations of high-mannose oligosaccharides

Conformations of several high-mannose-type oligosaccharides that are generated during the biosynthetic degradation of Man9GlcNAc2 to Man5GlcNAc2 have been studied by molecular dynamics (MD). Simulations were performed on NCI-FCRDC's Cray Y-MP 8D/8128 supercomputer using Biosym's CVFF force field for 1000 ps with different initial conformations. The conformations of the two alpha 1,3- and the two alpha 1,6-linkages in each oligomannose were different, suggesting that deriving oligosaccharide conformations based on the conformational preferences of the constituent disaccharide fragments will not always yield correct results. Unlike other oligomannoses, Man9GlcNAc2 appears to take more than one distinct conformation around the core alpha 1,6-linkage. These various conformations may play an important role in determining the processing pathways. Using the data on the preferred conformations of these oligomannoses and the available experimental results, possible pathways for processing Man9GlcNAc2 to Man5GlcNAc2 by alpha 1,2-linkage-specific mannosidases have been proposed. Conformational analysis of Man5GlcNAc2 indicates that the addition of beta 1,2-GlcNAc to the alpha 1,3-linked core mannose, besides serving as a prerequisite for mannosidase II action as suggested earlier, may also prevent the removal of alpha 1,3-mannose. The MD simulations also suggest that the processing of the precursor oligosaccharide during Asn-linked complex and hybrid glycan biosynthesis proceeds in a well-defined pathway involving more than one alpha 1,2-linkage-specific mannosidase. Knowledge of the conformation of the processing intermediates obtained from the present study can be used to design highly specific substrate analogues to inhibit a particular mannosidase, thereby blocking one processing pathway without interfering with the others.

Molecular dynamics simulations of asialoglycoprotein receptor ligands

Several recent studies have implicated carbohydrates in cell adhesion, inflammation, clearance of glycoproteins from blood circulation, embryonic development, and metastasis among others. Understanding the conformation of these carbohydrate recognition elements and their interaction at the molecular level is essential for the design of oligosaccharide inhibitors/drugs. Given the difficulty in solving carbohydrate structures by X-ray crystallography and since NMR experiments give only time-averaged conformation, molecular dynamics simulations are well suited to determine all the accessible conformations of oligosaccharides. Present communication reports the simulation of some of the oligosaccharide ligands of asialoglycoprotein receptor for 1 ns using Biosym's Insight II molecular modeling package on NCI-FCRDC's Y-MP 8D/8128 supercomputer. Results obtained from these simulations, in addition to explaining the observed differences in the binding affinities of these ligands to the asialoglycoprotein receptor, have led to a modified model for the recognition of the oligosaccharides by the receptor. Accordingly, only the two terminal galactose residues on the 1,3-arm of the triantennary oligosaccharide (GlcNAc2Man3 core of the N-linked oligosaccharides with N-acetyllactosamine in beta 1,2- and beta 1,4-linkages on the 1,3-linked core mannose) are primarily required for recognition, and the terminal galactose on the 1,6-arm (N-acetyllactosamine in beta 1,2-linkage on the 1,6-linked core mannose) provides additional binding energy. It has been shown that the oligosaccharides studied here have significant flexibility and the flexibility is more around the 1,3-linkage than the 1,6-linkage. The need for simulation for longer periods and with multiple initial conformations is also discussed in the present report.

Computer modelling studies of ribonuclease A-pyrimidine nucleotide complexes

Different modes of binding of pyrimidine monophosphates 2'-UMP, 3'-UMP, 2'-CMP and 3'-CMP to ribonuclease (RNase) A are studied by energy minimization in torsion angle and subsequently in Cartesian coordinate space. The results are analysed in the light of primary binding sites. The hydrogen bonding pattern brings out roles for amino acids such as Asn44 and Ser123 apart from the well known active site residues viz., His12,Lys41,Thr45 and His119. Amino acid segments 43-45 and 119-121 seem to be guiding the ligand binding by forming a pocket. Many of the active site charged residues display considerable movement upon nucleotide binding.

Expression of deletion constructs of bovine beta-1,4-galactosyltransferase in Escherichia coli: importance of Cys134 for its activity

Bovine beta-1,4-galactosyltransferase (beta-1,4-GT; EC 2.4.1.90) belongs to the glycosyltransferase family and as such shares a general topology: an N-terminal cytoplasmic tail, a signal anchor followed by a stem region and a catalytic domain at the C-terminal end of the protein. cDNA constructs of the N-terminal deleted forms of beta-1,4-GT were prepared in pGEX-2T vector and expressed in E. coli as glutathione-S-transferase (GST) fusion proteins. Recombinant proteins accumulated within inclusion bodies as insoluble aggregates that were solubilized in 5 M guanidine HCl and required an 'oxido-shuffling' reagent for regeneration of the enzyme activity. The recombinant beta-1,4-GT, devoid of the GST domain, has 30-85% of the sp. act. of bovine milk beta-1,4-GT with apparent Kms for N-acetylglucosamine and UDP-galactose similar to those of milk enzyme. Deletion analyses show that both beta-1,4-GT and lactose synthetase activities remain intact even in the absence of the first 129 residues (pGT-d129). The activities are lost when either deletions extend up to residue 142 (pGT-d142) or Cys134 is mutated to Ser (pGT-d129C134S). These results suggest that the formation of a disulfide bond involving Cys134 holds the protein in a conformation that is required for enzymatic activity.

Mutational analysis of the Golgi retention signal of bovine beta-1,4-galactosyltransferase

To examine the role of the NH2-terminal region of the 402-residue-long beta-1,4-galactosyltransferase (beta-1,4-GT), a series of mutants and chimeric cDNA were constructed by polymerase chain reaction and transiently expressed in COS-7 cells, the enzyme activities were measured, and the protein was localized in the cells by subcellular fractionation or indirect immunofluorescence microscopy. We showed earlier that the deletion of the amino-terminal cytoplasmic tail and transmembrane domain from GT abolishes the stable expression of this protein in mammalian cells (Masibay, A.S., Boeggeman, E., and Qasba, P.K. (1992) Mol. Biol. Rep. 16, 99-104). Further deletion analyses of the amino-terminal region show that the first 21 amino acids of beta-1,4-GT are not essential for the stable production of the protein and are consistently localized in the Golgi apparatus. In addition, analysis of hybrid constructs showed that residues 1-25 of alpha-1,3-galactosyltransferase can functionally replace the beta-1,4-GT amino-terminal domain (residues 1-43). This fusion protein also showed Golgi localization. On the other hand, the alpha-2,6-sialyltransferase/beta-1,4-GT fusion protein (alpha-2,6-ST/beta-1,4-GT) needed additional COOH-terminal sequences flanking the transmembrane domain of the alpha-2,6-ST for stability and Golgi localization. Substitution of Arg-24, Leu-25, Leu-26, and His-33 of the beta-1,4-GT transmembrane by Ile (pLFM) or substitution of Tyr by Ile at positions 40 and 41 coupled with the insertion of 4 Ile residues at position 43 (pLB) released the mutant proteins from the Golgi and was detected on the cell surface. Our results show that (a) the transmembrane domains of beta-1,4-GT, alpha-1,3-galactosyltransferase, and alpha-2,6-ST, along with its stem region, all play a role in Golgi targeting and participate in a common mechanism that allows the protein to be processed properly and not be degraded in vivo; (b) increasing the length of the transmembrane domain overrides the Golgi retention signal and directs the enzyme to the plasma membrane; and (c) the length of the hydrophobic region of the transmembrane domain of beta-1,4-GT is an important parameter but is not sufficient by itself for Golgi retention.