In silico analysis of the cyanobacterial lectin scytovirin: new insights into binding properties

Scytovirin is a lectin isolated from the cyanobacterium Scytonema varium that has shown activity against HIV, SARS coronavirus and Zaire Ebola virus. Its 95 amino acids are divided into two structural domains (SD), the first spanning amino acids 1–48 (SD1) and the second 49–95 (SD2). Interestingly, the domains are nearly identical but differ in their affinities for carbohydrates. With the aim of enhancing understanding of the binding properties of scytovirin, we performed molecular dynamics (MD) simulations of scytovirin complexed with Man4. We set up three systems: (i) Man4 bound to both domains (SD1?+?SD2) using the full-length protein; (ii) Man4 bound to an incomplete protein, containing only SD1 and (iii) Man4 bound to an incomplete protein containing only SD2. Contrary to other reports, binding free energy results suggest that Man4 can bind simultaneously to SD1 and SD2 binding regions, but SD1 individually has the best values of energy and the best affinity for Man4. Decomposition of the binding free energy showed that the residues that interact with Man4 were different in the three systems, suggesting that the binding mechanism of Man4 varies between full-length protein, SD1 and SD2. The results presented here may help to formulate strategies to use scytovirin and promote mutagenesis studies to improve the antiviral activity of scytovirin.     

A potent novel anti-HIV protein from the cultured cyanobacterium Scytonema varium

A new anti-HIV protein, scytovirin, was isolated from aqueous extracts of the cultured cyanobacterium Scytonema varium. The protein displayed potent anticytopathic activity against laboratory strains and primary isolates of HIV-1 with EC50 values ranging from 0.3 to 22 nM. Scytovirin binds to viral coat proteins gp120, gp160, and gp41 but not to cellular receptor CD4 or other tested proteins. This unique protein consists of a single 95-amino acid chain with significant internal sequence duplication and 10 cysteines forming five intrachain disulfide bonds.     

Multiplicity of carbohydrate-binding sites in beta-prism fold lectins: occurrence and possible evolutionary implications

The beta-prism II fold lectins of known structure, all from monocots, invariably have three carbohydrate-binding sites in each subunit/domain. Until recently, beta-prism I fold lectins of known structure were all from dicots and they exhibited one carbohydrate-binding site per subunit/domain. However, the recently determined structure of the beta-prism fold I lectin from banana, a monocot, has two very similar carbohydrate-binding sites. This prompted a detailed analysis of all the sequences appropriate for two-lectin folds and which carry one or more relevant carbohydrate-binding motifs. The very recent observation of a beta-prism I fold lectin, griffthsin, with three binding sites in each domain further confirmed the need for such an analysis. The analysis demonstrates substantial diversity in the number of binding sites unrelated to the taxonomical position of the plant source. However, the number of binding sites and the symmetry within the sequence exhibit reasonable correlation. The distribution of the two families of beta-prism fold lectins among plants and the number of binding sites in them, appear to suggest that both of them arose through successive gene duplication, fusion and divergent evolution of the same primitive carbohydrate-binding motif involving a Greek key. Analysis with sequences in individual Greek keys as independent units lends further support to this conclusion.It would seem that the preponderance of three carbohydrate-binding sites per domain in monocot lectins, particularly those with the beta-prism II fold, is related to the role of plant lectins in defence.     

Structural basis for chitotetraose coordination by CGL3, a novel galectin-related protein from Coprinopsis cinerea

Recent advances in genome sequencing efforts have revealed an abundance of novel putative lectins. Among these, many galectin-related proteins, characterized by many conserved residues but intriguingly lacking critical amino acids, have been found in all corners of the eukaryotic superkingdom. Here we present a structural and biochemical analysis of one representative, the galectin-related lectin CGL3 found in the inky cap mushroom Coprinopsis cinerea. This protein contains all but one conserved residues known to be involved in β-galactoside binding in galectins. A Trp residue strictly conserved among galectins is changed to an Arg in CGL3 (R81). Accordingly, the galectin-related protein is not able to bind lactose. Screening of a glycan array revealed that CGL3 displays preference for oligomers of β1-4-linked N-acetyl-glucosamines (chitooligosaccharides) and GalNAcβ1-4GlcNAc (LacdiNAc). Carbohydrate-binding affinity of this novel lectin was quantified using isothermal titration calorimetry, and its mode of chitooligosaccharide coordination not involving any aromatic amino acid residues was studied by X-ray crystallography. Structural information was used to alter the carbohydrate-binding specificity and substrate affinity of CGL3. The importance of residue R81 in determining the carbohydrate-binding specificity was demonstrated by replacing this Arg with a Trp residue (R81W). This single-amino-acid change led to a lectin that failed to bind chitooligosaccharides but gained lactose binding. Our results demonstrate that, similar to the legume lectin fold, the galectin fold represents a conserved structural framework upon which dramatically altered specificities can be grafted by few alterations in the binding site and that, in consequence, many metazoan galectin-related proteins may represent lectins with novel carbohydrate-binding specificities.     

Asparaginyl β-Hydroxylation of Proteins Containing Ankyrin Repeat Domains Influences Their Stability and Function

Recent reports have provided evidence that the β-hydroxylation of conserved asparaginyl residues in ankyrin repeat domain (ARD) proteins is a common posttranslational modification in animal cells. Here, nuclear magnetic resonance (NMR) and other biophysical techniques are used to study the effect of asparaginyl β-hydroxylation on the structure and stability of ‘consensus’ ARD proteins. The NMR analyses support previous work suggesting that a single β-hydroxylation of asparagine can stabilize the stereotypical ARD fold. A second asparaginyl β-hydroxylation causes further stabilization. In combination with mutation studies, the biophysical analyses reveal that the stabilizing effect of β-hydroxylation is, in part, mediated by a hydrogen bond between the asparaginyl β-hydroxyl group and the side chain of a conserved aspartyl residue, two residues to the N-terminal side of the target asparagine. Removal of this hydrogen bond resulted in reduced stabilization by hydroxylation. Formation of the same hydrogen bond is also shown to be a factor in inhibiting binding of hydroxylated ARDs to factor-inhibiting hypoxia-inducible factor (FIH). The effects of hydroxylation appear to be predominantly localized to the target asparagine and proximal residues, at least in the consensus ARD protein. The results reveal that thermodynamic stability is a factor in determining whether a particular ARD protein is an FIH substrate; a consensus ARD protein with three ankyrin repeats is an FIH substrate, while more stable consensus ARD proteins, with four or five ankyrin repeats, are not. However, NMR studies reveal that the consensus protein with four ankyrin repeats is still able to bind to FIH, suggesting that FIH may interact in cells with natural ankyrin repeats without resulting hydroxylation. Overall, the work provides novel biophysical insights into the mechanism by which asparaginyl β-hydroxylation stabilizes the ARD proteins and reduces their binding to FIH.     

A method for identifying a proposed carbohydrate-binding motif of proteins.

An examination of the binding sites of four carbohydrate binding proteins (Escherichia coli lactose repressor, E. coli arabinose-binding protein, yeast hexokinase A and Concanavalin A) revealed certain similarities of amino acid sequences and residues forming hydrogen bonds and hydrophobic interactions with the bound carbohydrate. These were: (i) Asx-Asx, hydrogen bonding to the pyranose ring oxygen and anomeric-OH group; (ii) Arg-X-X-X-(Ser/Thr), or the reverse sequence, with the Arg hydrogen bonding to the pyranose ring oxygen; (iii) Lys-(Ser/Thr)-X-X-Asp, or the reverse sequence and with interchange of the Lys-(Ser/Thr) positions, with hydrogen bonding of either or both the Lys and Asp residues to the -OH groups at carbons 2, 3, 4 or 6; (iv) a diaromatic sequence with possible hydrophobic interactions to the faces of the pyranose ring structure. An algorithm was devised to search the amino acid sequences of a large number of proteins, those known to bind carbohydrates as well as those without known carbohydrate-binding activities, for the four amino acid sequence criteria. The algorithm incorporated a weighted distance value (WDV) to assess the approximate distance between any two criteria, with the WDV being based on the predicted secondary structure of the protein amino acid sequence. When the algorithm using criteria 1 and 2 plus the WDV was applied to the sequences of 125 proteins, the method indicated the presence of the potential carbohydrate-binding site motif for 42% of proteins with known carbohydrate binding, only 8% of proteins were predicted as false positives, and the accuracy of the method was calculated to be 61.6%.(ABSTRACT TRUNCATED AT 250 WORDS)     

Multiplicity of carbohydrate-binding sites in <Emphasis Type="Italic">β</Emphasis>-prism fold lectins: occurrence and possible evolutionary implications

The β-prism II fold lectins of known structure, all from monocots, invariably have three carbohydrate-binding sites in each subunit/domain. Until recently, β-prism I fold lectins of known structure were all from dicots and they exhibited one carbohydrate-binding site per subunit/domain. However, the recently determined structure of the β-prism fold I lectin from banana, a monocot, has two very similar carbohydrate-binding sites. This prompted a detailed analysis of all the sequences appropriate for two-lectin folds and which carry one or more relevant carbohydrate-binding motifs. The very recent observation of a β-prism I fold lectin, griffithsin, with three binding sites in each domain further confirmed the need for such an analysis. The analysis demonstrates substantial diversity in the number of binding sites unrelated to the taxonomical position of the plant source. However, the number of binding sites and the symmetry within the sequence exhibit reasonable correlation. The distribution of the two families of β-prism fold lectins among plants and the number of binding sites in them, appear to suggest that both of them arose through successive gene duplication, fusion and divergent evolution of the same primitive carbohydrate-binding motif involving a Greek key. Analysis with sequences in individual Greek keys as independent units lends further support to this conclusion. It would seem that the preponderance of three carbohydrate-binding sites per domain in monocot lectins, particularly those with the β-prism II fold, is related to the role of plant lectins in defence.     

Sequence and structural analysis of fibronectin‐binding protein reveals importance of multiple intrinsic disordered tandem repeats

The location of certain amino acid sequences like repeats along the polypeptide chain is very important in the context of forming the overall shape of the protein molecule which in fact determines its function. In gram‐positive bacteria, fibronectin‐binding protein (FnBP) is one such repeat containing protein, and it is a cell wall‐attached protein responsible for various acute infections in human. Several studies on sequence, structure, and function of fibronectin‐binding regions of FnBPs were reported; however, no detailed study was carried out on the full‐length protein sequence. In the present study, we have made a thorough sequence and structure analysis on FnBP_A of Staphylococcus aureus and explored the presence of dual ligand‐binding ability of fibrinogen (fg)‐binding region and its molecular recognition processes. Multiple sequence alignment and protein‐protein docking analysis reveal the regions which are likely involved in dual ligand binding. Further analysis of docking of FnBP_A fg‐binding region and fn N‐terminal modules suggests that if the latter binds to the fg‐binding region of FnBP_A, it would inhibit the subsequent binding of fg because of steric hindrance. The sequence analysis further suggests that the abundance of disorder promoting residue glutamic acid and dual personality (both order/disorder promoting) residue threonine in tandem repeats of FnBP_A and B proteins possibly would help the molecule to undergo a conformational change while binding with fn by β‐zipper mechanism. The segment‐based power spectral analysis was carried out which helps to understand the distribution of hydrophobic residues along the sequence particularly in intrinsic disordered tandem repeats. The results presented here will help to understand the role of internal repeats and intrinsic disorder in the molecular recognition process of a pathogenic cell surface protein.     

Mutational analysis of the carbohydrate binding activity of the tobacco lectin

At present the three-dimensional structure of the tobacco lectin, further referred to as Nictaba, and its carbohydrate-binding site are unresolved. In this paper, we propose a three-dimensional model for the Nictaba domain based on the homology between Nictaba and the carbohydrate-binding module 22 of Clostridium thermocellum Xyn10B. The suggested model nicely fits with results from circular dichroism experiments, indicating that Nictaba consists mainly of β-sheet. In addition, the previously identified nuclear localization signal is located at the top of the protein as a part of a protruding loop. Judging from this model and sequence alignments with closely related proteins, conserved glutamic acid and tryptophan residues in the Nictaba sequence were selected for mutational analysis. The mutant DNA sequences as well as the original Nictaba sequence have been expressed in Pichia pastoris and the recombinant proteins were purified from the culture medium. Subsequently, the recombinant proteins were characterized and their carbohydrate binding properties analyzed with glycan array technology. It was shown that mutation of glutamic acid residues in the C-terminal half of the protein did not alter the carbohydrate-binding activity of the lectin. In contrast, mutation of tryptophan residues in the N-terminal half of the Nictaba domain resulted in a complete loss of carbohydrate binding activity. These results suggest that tryptophan residues play an important role in the carbohydrate binding site of Nictaba.     

Structure of a mannan-specific family 35 carbohydrate-binding module: evidence for significant conformational changes upon ligand binding

Enzymes that digest plant cell wall polysaccharides generally contain non-catalytic, carbohydrate-binding modules (CBMs) that function by attaching the enzyme to the substrate, potentiating catalytic activity. Here, we present the first structure of a family 35 CBM, derived from the Cellvibrio japonicus beta-1,4-mannanase Man5C. The NMR structure has been determined for both the free protein and the protein bound to mannopentaose. The data show that the protein displays a typical beta-jelly-roll fold. Ligand binding is not located on the concave surface of the protein, as occurs in many CBMs that display the jelly-roll fold, but is formed by the loops that link the two beta-sheets of the protein, similar to family 6 CBMs. In contrast to the majority of CBMs, which are generally rigid proteins, CBM35 undergoes significant conformational change upon ligand binding. The curvature of the binding site and the narrow binding cleft are likely to be the main determinants of binding specificity. The predicted solvent exposure of O6 at several subsites provides an explanation for the observed accommodation of decorated mannans. Two of the key aromatic residues in Man5C-CBM35 that interact with mannopentaose are conserved in mannanase-derived CBM35s, which will guide specificity predictions based on the primary sequence of proteins in this CBM family.     

Solution structure of a CCHHC domain of neural zinc finger factor-1 and its implications for DNA binding

The structure of a CCHHC zinc-binding domain from neural zinc finger factor-1 (NZF-1) has been determined in solution though the use of NMR methods. This domain is a member of a family of domains that have the Cys-X(4)-Cys-X(4)-His-X(7)-His-X(5)-Cys consensus sequence. The structure determination reveals a novel fold based around a zinc(II) ion coordinated to three Cys residues and the second of the two conserved His residues. The other His residue is stacked between the metal-coordinated His residue and a relatively conserved aromatic residue. Analysis of His to Gln sequence variants reveals that both His residues are required for the formation of a well-defined structure, but neither is required for high-affinity metal binding at a tetrahedral site. The structure suggests that a two-domain protein fragment and a double-stranded DNA binding site may interact with a common two-fold axis relating the two domains and the two half-sites of the DNA-inverted repeat.     

Relationship between sequence determinants of stability for two natural homologous proteins with different folds

In the Cro protein family, an evolutionary change in secondary structure has converted an alpha-helical fold to a mixture of alpha-helix and beta-sheet. P22 Cro and lambda Cro represent the ancestral all-alpha and descendant alpha+beta folds, respectively. The major structural differences between these proteins are at the C-terminal end of the domain (residues 34-56), where two alpha-helices in P22 Cro align with two beta-strands in lambda Cro. We sought to assess the possibility that smooth evolutionary transitions could have converted the all-alpha structure to the alpha+beta structure through sequences that could adopt both folds. First, we used scanning mutagenesis to identify and compare patterns of key stabilizing residues in the C-terminal regions of both P22 Cro and lambda Cro. These patterns exhibited little similarity to each other, with structurally important residues in the two proteins most often occurring at different sequence positions. Second, "hybrid scanning" studies, involving replacement of each wild-type residue in P22 Cro with the aligned wild-type residue in lambda Cro and vice versa, revealed five or six residues in each protein that strongly destabilized the other. These results suggest that key stability determinants for each Cro fold are quite different and that the P22 Cro sequence strongly favors the all-alpha structure while the lambda Cro sequence strongly favors the alpha+beta structure. Nonetheless, we were able to design a "structurally ambivalent" sequence fragment (SASF1), which corresponded to residues 39-56 and simultaneously incorporated most key stabilizing residues for both P22 Cro and lambda Cro. NMR experiments showed SASF1 to stably fold as a beta-hairpin when incorporated into the lambda Cro sequence but as a pair of alpha-helices when incorporated into P22 Cro.     

Fold independent structural comparisons of protein-ligand binding sites for exploring functional relationships

The rapid growth in protein structural data and the emergence of structural genomics projects have increased the need for automatic structure analysis and tools for function prediction. Small molecule recognition is critical to the function of many proteins; therefore, determination of ligand binding site similarity is important for understanding ligand interactions and may allow their functional classification. Here, we present a binding sites database (SitesBase) that given a known protein-ligand binding site allows rapid retrieval of other binding sites with similar structure independent of overall sequence or fold similarity. However, each match is also annotated with sequence similarity and fold information to aid interpretation of structure and functional similarity. Similarity in ligand binding sites can indicate common binding modes and recognition of similar molecules, allowing potential inference of function for an uncharacterised protein or providing additional evidence of common function where sequence or fold similarity is already known. Alternatively, the resource can provide valuable information for detailed studies of molecular recognition including structure-based ligand design and in understanding ligand cross-reactivity. Here, we show examples of atomic similarity between superfamily or more distant fold relatives as well as between seemingly unrelated proteins. Assignment of unclassified proteins to structural superfamiles is also undertaken and in most cases substantiates assignments made using sequence similarity. Correct assignment is also possible where sequence similarity fails to find significant matches, illustrating the potential use of binding site comparisons for newly determined proteins.     

The solution structure of a domain from the Neisseria meningitidis lipoprotein PilP reveals a new beta-sandwich fold

Type IV pili are long, thin fibres, which extend from the surface of the bacterial pathogen Neisseria meningitidis; they play a key role in adhesion and colonisation of host cells. PilP is a lipoprotein, suggested to be involved in the assembly and stabilization of an outer membrane protein, PilQ, which is required for pilus formation. Here we describe the expression of a recombinant fragment of PilP, spanning residues 20 to 181, and determination of the solution structure of a folded domain, spanning residues 85 to 163, by NMR. The N-terminal third of the protein, from residues 20 to 84, is apparently unfolded. Protease digestion yielded a 113 residue fragment that contained the folded domain. The domain adopts a simple beta-sandwich type fold, consisting of a three-stranded beta-sheet packed against a four-stranded beta-sheet. There is also a short segment of 3(10) helix at the N-terminal part of the folded domain. We were unable to identify any other proteins that are closely related in structure to the PilP domain, although the fold appears to be distantly related to the lipocalin family. Over 40 homologues of PilP have been identified in Gram-negative bacteria and the majority of conserved residues lie within the folded domain. The fourth beta-strand and adjacent loop regions contain a high proportion of conserved residues, including three glycine residues, which seem to play a role in linking the two beta-sheets. The two beta-sheets pack together to form a crevice, lined with conserved hydrophobic residues: we suggest that this feature could act as a binding site for a small ligand. The results show that PilP and its homologues have a conserved, folded domain at the C-terminal end of the protein that may be involved in mediating binding to hydrophobic ligands.     

Crystal structure of the glutaredoxin-like protein SH3BGRL3 at 1.6 Angstrom resolution

We report the 1.6 Angstrom resolution crystal structure of SH3BGRL3, a member of a new mammalian protein family of unknown function. The observed "thioredoxin fold" of SH3BGRL3 matches the tertiary structure of glutaredoxins, even in the N-terminal region where the sequence similarity between the two protein families is negligible. In particular, SH3BGRL3 displays structural modifications at the N-terminal Cys-x-x-Cys loop, responsible for glutathione binding and catalysis in glutaredoxins. The loop hosts a six residue insertion, yielding an extra N-terminal-capped helical turn, first observed here for the thioredoxin fold. This, together with deletion of both Cys residues, results in a substantial reshaping of the neighboring cleft, where glutathione is hosted in glutaredoxins. While not active in redox reaction and glutathione binding, SH3BGRL3 may act as an endogenous modulator of glutaredoxin activities by competing, with its fully conserved thioredoxin fold, for binding to yet unknown target proteins.     

ACCA phosphopeptide recognition by the BRCT repeats of BRCA1

The tumour suppressor gene BRCA1 encodes a 220 kDa protein that participates in multiple cellular processes. The BRCA1 protein contains a tandem of two BRCT repeats at its carboxy-terminal region. The majority of disease-associated BRCA1 mutations affect this region and provide to the BRCT repeats a central role in the BRCA1 tumour suppressor function. The BRCT repeats have been shown to mediate phospho-dependant protein-protein interactions. They recognize phosphorylated peptides using a recognition groove that spans both BRCT repeats. We previously identified an interaction between the tandem of BRCA1 BRCT repeats and ACCA, which was disrupted by germ line BRCA1 mutations that affect the BRCT repeats. We recently showed that BRCA1 modulates ACCA activity through its phospho-dependent binding to ACCA. To delineate the region of ACCA that is crucial for the regulation of its activity by BRCA1, we searched for potential phosphorylation sites in the ACCA sequence that might be recognized by the BRCA1 BRCT repeats. Using sequence analysis and structure modelling, we proposed the Ser1263 residue as the most favourable candidate among six residues, for recognition by the BRCA1 BRCT repeats. Using experimental approaches, such as GST pull-down assay with Bosc cells, we clearly showed that phosphorylation of only Ser1263 was essential for the interaction of ACCA with the BRCT repeats. We finally demonstrated by immunoprecipitation of ACCA in cells, that the whole BRCA1 protein interacts with ACCA when phosphorylated on Ser1263.     

De novo protein fold families expand the designable ligand binding site space

A major challenge in designing proteins de novo to bind user-defined ligands with high affinity is finding backbones structures into which a new binding site geometry can be engineered with high precision. Recent advances in methods to generate protein fold families de novo have expanded the space of accessible protein structures, but it is not clear to what extend de novo proteins with diverse geometries also expand the space of designable ligand binding functions. We constructed a library of 25,806 high-quality ligand binding sites and developed a fast protocol to place (“match”) these binding sites into both naturally occurring and de novo protein families with two fold topologies: Rossman and NTF2. Each matching step involves engineering new binding site residues into each protein “scaffold”, which is distinct from the problem of comparing already existing binding pockets. 5,896 and 7,475 binding sites could be matched to the Rossmann and NTF2 fold families, respectively. De novo designed Rossman and NTF2 protein families can support 1,791 and 678 binding sites that cannot be matched to naturally existing structures with the same topologies, respectively. While the number of protein residues in ligand binding sites is the major determinant of matching success, ligand size and primary sequence separation of binding site residues also play important roles. The number of matched binding sites are power law functions of the number of members in a fold family. Our results suggest that de novo sampling of geometric variations on diverse fold topologies can significantly expand the space of designable ligand binding sites for a wealth of possible new protein functions.     

The structure of Aip1p,a WD repeat protein that regulates Cofilin-mediated actin depolymerization

Actin-interacting protein 1 (Aip1p) is a 67-kDa WD repeat protein known to regulate the depolymerization of actin filaments by cofilin and is conserved in organisms ranging from yeast to mammals. The crystal structure of Aip1p from Saccharomyces cerevisiae was determined to a 2.3-A resolution and a final crystallographic R-factor of 0.204. The structure reveals that the overall fold is formed by two connected seven-bladed beta-propellers and has important implications for the structure of Aip1 from other organisms and WD repeat-containing proteins in general. These results were unexpected because a maximum of 10 WD repeats had been reported in the literature for this protein using sequence data. The surfaces of the beta-propellers formed by the D-A and B-C loops are positioned adjacent to one another, giving Aip1p a shape that resembles an open "clamshell." The mapping of conserved residues to the structure of Aip1p reveals dense patches of conserved residues on the surface of one beta-propeller and at the interface of the two beta-propellers. These two patches of conserved residues suggest a potential binding site for F-actin on Aip1p and that the orientation of the beta-propellers with respect to one another plays a role in binding an actin-cofilin complex. In addition, the conserved interface between the domains is mediated by a number of interactions that appear to impart rigidity between the two domains of Aip1p and may make a large substrate-induced conformational change difficult.     

Motifs and structural fold of the cofactor binding site of human glutamate decarboxylase.

The pyridoxal-P binding sites of the two isoforms of human glutamate decarboxylase (GAD65 and GAD67) were modeled by using PROBE (a recently developed algorithm for multiple sequence alignment and database searching) to align the primary sequence of GAD with pyridoxal-P binding proteins of known structure. GAD's cofactor binding site is particularly interesting because GAD activity in the brain is controlled in part by a regulated interconversion of the apo- and holoenzymes. PROBE identified six motifs shared by the two GADs and four proteins of known structure: bacterial ornithine decarboxylase, dialkylglycine decarboxylase, aspartate aminotransferase, and tyrosine phenol-lyase. Five of the motifs corresponded to the alpha/beta elements and loops that form most of the conserved fold of the pyridoxal-P binding cleft of the four enzymes of known structure; the sixth motif corresponded to a helical element of the small domain that closes when the substrate binds. Eight residues that interact with pyridoxal-P and a ninth residue that lies at the interface of the large and small domains were also identified. Eleven additional conserved residues were identified and their functions were evaluated by examining the proteins of known structure. The key residues that interact directly with pyridoxal-P were identical in ornithine decarboxylase and the two GADs, thus allowing us to make a specific structural prediction of the cofactor binding site of GAD. The strong conservation of the cofactor binding site in GAD indicates that the highly regulated transition between apo- and holoGAD is accomplished by modifications in this basic fold rather than through a novel folding pattern.