Purification and characterization of human interleukin-1 alpha produced in Escherichia coli

The production of human interleukin-1 alpha (IL-1 alpha) in Escherichia coli is described together with a method for its purification. The isolated protein was shown to be pure and physically homogeneous. The in vitro biological activity of IL-1 alpha was tested with the mononuclear-cell factor and the lymphocyte-activating factor assays. The specific activity determined with both assays was about 3 X 10(7) units mg-1 and is similar to that observed with recombinant human IL-1 beta. The purified protein was resolved by chromatofocusing into two species of isoelectric points 5.45 and 5.20 (75% and 25%, respectively, of the total protein). Both species had similar chemical properties and biological activities to the unfractionated protein. The charge difference between the species was attributed to the deamidation of a single Asn or Gln residue.     

Site-directed mutagenesis of a conserved, extracellular aspartic acid residue affects the ouabain sensitivity of sheep Na,K-ATPase

Site-specific mutagenesis was used to study the function of a conserved, extracellular aspartic acid residue from the sheep Na,K-ATPase alpha subunit. This amino acid, Asp-121, is the penultimate residue of the first extracellular domain of the alpha subunit. The border residues of this particular extracellular loop of the alpha subunit have been shown to be determinants of ouabain sensitivity (Price, E. M., and Lingrel, J. B. (1988) Biochemistry 27, 8400-8408). In order to determine if Asp-121 is involved in ouabain binding, five different amino acid substitutions at this position were generated. Four of the five mutant alpha subunits, containing either Asn, Ala, Glu, or Ser in place of Asp-121, conferred ouabain resistance to HeLa cells when expressed in those cells. Cloned sublines of cells selected in ouabain were characterized in terms of ouabain-inhibitable cell growth and Na,K-ATPase activity. The cells expressing the mutant Na,K-ATPase alpha subunit containing either Asn, Ala, Glu, or Ser in place of Asp-121 contained a component of Na,K-ATPase activity that was nearly 100-times more resistant to ouabain than the endogenous HeLa (human) or sheep enzyme. Apparently, conservative (Glu for Asp), isosteric (Asn for Asp), and nonconservative (Ala or Ser for Asp) substitutions all significantly decreased ouabain sensitivity. These data suggest that Asp-121 of the sheep Na,K-ATPase alpha subunit participates in the binding interaction between the enzyme and ouabain.     

Effect of amino acid residues on conformational stability in eight mutant proteins variously substituted at a unique position of the tryptophan synthase alpha-subunit

To elucidate the role of individual amino acid residues in stabilizing the conformation of a protein, the stabilities of wild-type tryptophan synthase alpha-subunit from Escherichia coli and seven mutant proteins substituted by single amino acid residues at position 49, which is buried in the interior of the protein, were compared. The mutant proteins have Gln, Met, Val, Tyr, Leu, Ser, or Lys in place of Glu in the wild-type protein. The dissociation constant, pK, of the Glu residue at position 49 for the wild-type protein was determined to be 7.5 from a titration curve obtained by comparison of two-dimensional isoelectric focusing electrophoresis of the wild-type and mutant proteins. Our results indicate that 1) the conformational stabilities of the proteins studied increase linearly with hydrophobicity of the substituting residues (except Tyr), with the coefficient of this linear dependence being 2.0, 3.4, or 1.3 at pH 5.5, 7.0, or 9.0, respectively; and 2) Lys or Glu at position 49 serve as a destabilizing factor when ionized.     

Comparison of CD spectra in the aromatic region on a series of variant proteins substituted at a unique position of tryptophan synthase alpha-subunit

CD spectra in the aromatic region of a series of the mutant alpha-subunits of tryptophan synthase from Escherichia coli, substituted at position 49 buried in the interior of the molecule, were measured at pH 7.0 and 25 degrees C. These measurements were taken to gain information on conformational change produced by single amino acid substitutions. The CD spectra of the mutant proteins, substituted by Tyr or Trp residue in place of Glu residue at position 49, showed more intense positive bands due to one additional Tyr or Trp residue at position 49. The CD spectra of other mutant proteins also differed from that of the wild-type protein, despite the fact that the substituted residues at position 49 were not aromatic. Using the spectrum of the wild-type protein (Glu49) as a standard, the spectra of the other mutants were classified into three major groups. For 10 mutant proteins substituted by Ile, Ala, Leu, Met, Val, Cys, Pro, Ser, His, or Gly, their CD values of bands (due to Tyr residues) decreased in comparison with those of the wild-type protein. The mutant protein substituted by Phe also belonged to this group. These substituted amino acid residues are more hydrophobic than the original residue, Glu. In the second group, three mutant proteins were substituted by Lys, Gln, or Asn, and the CD values of tyrosyl bands increased compared to those of the wild-type proteins. These residues are polar. In the third group, the CD values of tyrosyl bands of two mutant proteins substituted by Asp or Thr were similar to those of the wild-type protein, except for one band at 276.5 nm. These results suggested that the changes in the CD spectra for the mutant proteins were affected by the hydrophobicity of the residues at position 49.     

Identification of essential amino acid residues for catalytic activity and thermostability of novel chitosanase by site-directed mutagenesis

The functional importance of a conserved region in a novel chitosanase from Bacillus sp. CK4 was investigated. Each of the three carboxylic amino acid residues (Glu-50, Glu-62, and Asp-66) was changed to Asp and Gln or Asn and Glu by site-directed mutagenesis, respectively. The Asp-66-->Asn and Asp-66-->Glu mutation remarkably decreased kinetic parameters such as Vmax and kcat to approximately 1/1,000 those of the wild-type enzyme, indicating that the Asp-66 residue was essential for catalysis. The thermostable chitosanase contains three Cys residues at positions 49, 72, and 211. The Cys-49-->Ser/Tyr and Cys-72-->Ser/Tyr mutant enzymes were as stable to thermal inactivation and denaturating agents as the wild-type enzyme. However, the half-life of the Cys-211-->Ser/Tyr mutant enzyme was less than 10 min at 80 degrees C, while that of the wild-type enzyme was about 90 min. Moreover, the residual activity of Cys-211-->Ser/Tyr enzyme was substantially decreased by 8 M urea; and it lost all catalytic activity in 40% ethanol. These results show that the substitution of Cys with any amino acid residues at position 211 seems to affect the conformational stability of the chitosanase.     

Activation of factor V during intrinsic and extrinsic coagulation. Inhibition by heparin, hirudin and D-Phe-Pro-Arg-Ch2Cl.

Ser130, Asp131 and Asn132 ('SDN') are highly conserved residues in class A beta-lactamases forming one wall of the active-site cavity. All three residues of the SDN loop in Streptomyces albus G beta-lactamase were modified by site-directed mutagenesis. The mutant proteins were expressed in Streptomyces lividans, purified from culture supernatants and their kinetic parameters were determined for several substrates. Ser130 was substituted by Asn, Ala and Gly. The first modification yielded an almost totally inactive protein, whereas the smaller-side-chain mutants (A and G) retained some activity, but were less stable than the wild-type enzyme. Ser130 might thus be involved in maintaining the structure of the active-site cavity. Mutations of Asp131 into Glu and Gly proved to be highly detrimental to enzyme stability, reflecting significant structural perturbations. Mutation of Asn132 into Ala resulted in a dramatically decreased enzymic activity (more than 100-fold) especially toward cephalosporin substrates, kcat. being the most affected parameter, which would indicate a role of Asn132 in transition-state stabilization rather than in ground-state binding. Comparison of the N132A and the previously described N132S mutant enzymes underline the importance of an H-bond-forming residue at position 132 for the catalytic process.     

Electrostatic interactions in wild-type and mutant recombinant human myoglobins

Residue Val68 in human myoglobin has been replaced by Asn, Asp, and Glu with site-directed mutagenesis. Purified proteins were characterized by isoelectric focusing and by absorption, CD, and NMR spectroscopy. These studies demonstrated that Mb is able to tolerate substitution of the buried hydrophobic residue Val68 by Asn, Asp, and Glu. In the metaquo derivatives of the Glu and Asp mutants, the negative charge at residue 68 is stabilized by a favorable Coulombic interaction with the heme iron. In the absence of this interaction, as in the metcyano and ferrous deoxy derivatives, the relatively nonpolar protein interior cannot stabilize an isolated buried negative charge, and the carboxylate is either protonated or stabilized via a salt bridge with the nearby distal histidine. Hence in the Asp and Glu mutant proteins, both reduction and cyanide binding are accompanied by proton uptake by the protein. The apoproteins were prepared and reconstituted with the chlorophyll derivative zinc pyrochlorophyllide a. Absorption and fluorescence spectra were quite similar for wild-type and all mutant proteins reconstituted with this derivative. These results do not support the point charge model for the red shifts observed in the spectra of chlorophylls associated with photosynthetic proteins. From the pH dependence of the absorption spectrum of zinc pyrochlorophyllide a in the Glu mutant, the apparent pKa of the buried glutamate residue was estimated to be 8.9. This increase of 4.4 pH units, over the value for Glu in aqueous solution, provides a measure of the polarity of the protein interior.     

Contribution of an aspartate residue, D114, in the active site of clostridial glutamate dehydrogenase to the enzyme's unusual pH dependence

Glutamate dehydrogenase from Clostridium symbiosum displays unusual kinetic behaviour at high pH when compared with other members of this enzyme family. Structural and sequence comparisons with GDHs from other organisms have indicated that the Asp residue at position 114 in the clostridial enzyme may account for these differences. By replacing this residue by Asn, a mutant protein has been created with altered functional properties at high pH. This mutant protein can be efficiently overexpressed in Escherichia coli, and several criteria, including mobility in non-denaturing electrophoresis, circular dichroism (CD) spectra and initial crystallisation studies, suggest a folding and an assembly comparable to those of the wild-type protein. The D114N mutant enzyme shows a higher optimum pH for activity than the wild-type enzyme, and both CD data and activity measurements show that the distinctive time-dependent reversible conformational inactivation seen at high pH in the wild-type enzyme is abolished in the mutant.     

A conserved deamidation site at Asn 2 in the catalytic subunit of mammalian cAMP-dependent protein kinase detected by capillary LC-MS and tandem mass spectrometry.

The N-terminal sequence myr-Gly-Asn is conserved among the myristoylated cAPK (protein kinase A) catalytic subunit isozymes Calpha, Cbeta, and Cgamma. By capillary LC-MS and tandem MS, we show that, in approximately one third of the Calpha and Cbeta enzyme populations from cattle, pig, rabbit, and rat striated muscle, Asn 2 is deamidated to Asp 2. This deamidation accounts for the major isoelectric variants of the cAPK C-subunits formerly called CA and CB. Deamidation also includes characteristic isoaspartate isomeric peptides from Calpha and Cbeta. Asn 2 deamidation does not occur during C-subunit preparation and is absent in recombinant myristoylated Calpha (rCalpha) from Escherichia coli. Deamidation appears to be the exclusive pathway for introduction of an acidic residue adjacent to the myristoylated N-terminal glycine, verified by the myristoylation negative phenotype of an rCalpha(Asn 2 Asp) mutant. This is the first report thus far of a naturally occurring myr-Gly-Asp sequence. Asp 2 seems to be required for the well-characterized (auto)phosphorylation of the native enzyme at Ser 10. Our results suggest that the myristoylated N terminus of cAPK is a conserved site for deamidation in vivo. Comparable myr-Gly-Asn sequences are found in several signaling proteins. This may be especially significant in view of the recent knowledge that negative charges close to myristic acid in some proteins contribute to regulating their cellular localization.     

Identification and characterization of human rhinovirus-14 3C protease deamidation isoform

A purified recombinant human rhinovirus-14 3C protease preparation contained only approximately 50% active enzyme as titrated using specifically designed irreversible 3C protease inhibitors. Analysis of the purified 3C protein by isoelectric focusing showed differently charged 3C isoforms that had isoelectric points (pI) of 8.3 (55%) and 9.0 (45%), with the latter one being consistent with the predicted pI of the human rhinovirus-14 3C protein. Further analysis indicated that the pI 8.3 protein was the deamidated form of 3C, and it displayed approximately 10-fold reduced cleavage activity relative to the original 3C protease sample. Peptide mapping followed by sequence analysis revealed that a single asparagine, Asn-164, was deamidated to aspartic acid in the pI 8.3 isoform. Converting Asn-164 to Asp by site-directed mutagenesis resulted in a mutated 3C protease with extremely low activity, as seen with the pI 8.3 isoform, indicating a role of Asn-164 in substrate recognition and binding. In addition, the deamidated 3C protease was found to be present in vivo, and its abundance was related to the viral replication cycle. Moreover, mutant virus carrying Asp-164 showed reduced viability in infected cells. Taken together, our data suggest that 3C protein deamidation plays a role in the regulation of its enzymatic activity.     

An increase in apparent affinity for sucrose of mung bean sucrose synthase is caused by in vitro phosphorylation or directed mutagenesis of Ser11

A mutational analysis of mung bean (Vigna radiata Wilczek) sucrose synthase was performed by site-directed mutagenesis of the recombinant protein expressed in Escherichia coli, in which two different acidic amino acid residues (Asp or Glu) were introduced at Ser11 (S11D, S11E). Only the wild-type enzyme (Ser11) was phosphorylated in vitro by a Ca(2+)-dependent protein kinase from soybean root nodules, suggesting that this is the specific target residue in mung bean sucrose synthase. The apparent affinity for sucrose was increased in this phosphorylated enzyme and also in the S11D and S11E mutant enzymes, although the affinities for UDP-glucose and fructose were similar in the wild-type, phosphorylated wild-type, and mutant enzymes. These results suggest that a monoanionic (1-) side chain at position 11 mimics the Ser11-P2- residue to bind and cleave sucrose for the synthesis of UDP-glucose. Since the S11E mutant enzyme showed the lowest K(m) (sucrose) and the highest catalytic efficiency of the recombinant proteins, the enzymic properties of this S11E mutant were further characterized. The results showed that replacement of Ser11 with Glu11 modestly protected the sucrose synthesis activity against phenolic glycosides and altered the enzyme nucleotide specificity. We postulate that the introduction of negative charge at Ser11 is possibly involved in the enzymatic perturbation of sucrose synthase.     

Differential rate constants of racemization of aspartyl and asparaginyl residues in human alpha A-crystallin mutants

Asp58 and Asp151 in alpha A-crystallin of human eye lenses become highly inverted and isomerized to d-beta-Asp residues with age. Racemization was previously shown to proceed rapidly when the residue on the carboxyl side of the Asp residue is small. Asn was also demonstrated to be more susceptible to racemization than Asp in protein. In this study, the changes of rate constants for racemization at Asp58 and Asp151 and at Asn58 and Asn151 were investigated using D58N, S59T, D151N and A152V mutants obtained through site-directed mutagenesis. The rate constant of racemization at Asn151 in D151N was found to be 1.5 times more rapid than Asp151 in the wild-type. For A152V, the rate constant at Asp151 was 1/4 that of the wild-type. There were no significant differences in the rate constants of racemization for both Asp58 and Asn58 residues. The aggregate size of D58N, S59T and D151N mutants increased or increased in polydispersity and their chaperone activities decreased. The size and chaperone activity of A152V was unchanged. These results suggest that structures close to Asp58 and Asp151 residues in the protein affect the rate constant of Asp racemization and the size and chaperone function of alpha A-crystallin.     

Reversing Evolution: Re-establishing Obligate Metal Ion Dependence in a Metal-independent KDO8P Synthase

Two distinct groups of 3-deoxy-d-manno-octulosonate 8-phosphate synthase (KDO8PS), a key enzyme of cell-wall biosynthesis, differ by their requirement for a divalent metal ion for enzymatic activity. The unique difference between these groups is the replacement of the metal-binding Cys by Asn. Substitution of just this Asn for a Cys in metal-independent KDO8PS does not create the obligate metal-ion dependency of natural metal-dependent enzymes. We describe how three or four mutations of the metal-independent KDO8PS from Neisseria meningitidis produce a fully functional, obligately metal-dependent KDO8PS. For the substitutions Asn23Cys, Asp247Glu (this Asp binds to the metal ion in all metal-dependent KDO8PS) and Pro249Ala, and for double and triple combinations, mutant enzymes that contained Cys in place of Asn showed an increase in activity in the presence of divalent metal ions. However, combining these mutations with substitution by Ser of the Cys residue in the conserved ²⁴⁶CysAspGlyPro²⁴⁹ motif of metal-independent KDO8PS created enzymes with obligate metal dependency. The quadruple mutant (Asn23Cys/Cys246Ser/Asp247Glu/Pro249Ala) showed comparable activity to wild-type enzymes only in the presence of metal ions, with maximum activity with Cd²⁺, the metal ion that is strongly inhibitory at micromolar concentrations for the wild-type enzyme. In the absence of metal ions, activity was barely detectable for this quadruple mutant or for triple mutants bearing both Cys246Ser and Asn23Cys mutations. The structures of NmeKDO8PS and its Asn23Cys/Asp247Glu/Pro249Ala and quadruple mutants at pH 4.6 were characterized at resolutions better than 1.85 Å. Aged crystals of the Asn23Cys/Asp247Glu/Pro249Ala mutant featured a Cys23-Cys246 disulfide linkage, explaining the spectral bleaching observed when this mutant was incubated with Cu²⁺. Such bleaching was not observed for the quadruple mutant. Reverse evolution to a fully functional obligately metal-dependent KDO8PS has been achieved with just three directed mutations for enzymes that have, at best, 47% identity between metal-dependent and metal-independent pairs.     

Asn183 in alpha5 is essential for oligomerisation and toxicity of the Bacillus thuringiensis Cry4Ba toxin

The proposed toxicity mechanism of the Bacillus thuringiensis Cry insecticidal proteins involves membrane penetration and lytic pore formation of the alpha4-alpha5 hairpins in the target larval midgut cell membranes. In this study, alanine substitutions of selected polar residues (Tyr(178), Gln(180), Asn(183), Asn(185), and Asn(195)) in the hydrophobic helix-alpha5 of the Cry4Ba mosquito-larvicidal protein were initially conducted via PCR-based directed mutagenesis. Upon IPTG induction, all the 130-kDa mutant protoxins were highly expressed in Escherichia coli as cytoplasmic inclusions, with yields similar to the wild-type protoxin. When E. coli cells expressing each mutant toxin were tested against Stegomyia aegypti mosquito larvae, the larvicidal activity of the N183A mutant was almost completely abolished whereas the four other mutants showed only a small reduction in toxicity. Additionally, replacements of this critical residue with various amino acids revealed that the uncharged polar residue at position 183 in alpha5 is crucial for larvicidal activity. Further characterisation of the N183K bio-inactive mutant revealed that the 65-kDa activated toxin was unable to form oligomers in lipid vesicles and its ability to induce the release of entrapped calcein from liposomes was much weaker than that of the wild-type toxin. These results suggest that the highly conserved Asn(183) located in the middle of the transmembrane alpha5 of Cry4Ba plays a crucial role in toxicity and toxin oligomerisation in the lipid membranes.     

Isolation and structural characterization of three isoforms of recombinant consensus alpha interferon

Recombinant DNA-derived consensus alpha interferon was expressed in Escherichia coli and purified. Isoelectric focusing of this purified protein indicated the presence of three isoelectric subforms of pI 6.1, 6.0, and 5.7. These three subforms were preparatively separated by isoelectric focusing using Immobiline polyacrylamide gel and did not exhibit apparent differences in biological activity and tertiary structure. The pI 5.7 subform could also be separated from the pI 6.1 and 6.0 subforms by reverse-phase HPLC. Automated N-terminal amino acid sequence analysis of the pI 6.1 and 6.0 subforms yielded sequences corresponding to the methionyl and des-methionyl forms of the protein, respectively. Sequence analysis of the pI 5.7 subform indicated that its N terminus is blocked. To further determine the structure of the blocking moiety in the pI 5.7 subform, a blocked N-terminal tryptic peptide was isolated from HPLC peptide mapping of the S-carboxymethylated derivative. Results obtained from mass spectroscopic and amino acid analyses of this peptide suggest that it is blocked with an acetyl group at the N-terminal cysteine residue.     

Expression and characterization of active site mutants of hevamine, a chitinase from the rubber tree Hevea brasiliensis.

Hevamine is a chitinase from the rubber tree Hevea brasiliensis. Its active site contains Asp125, Glu127, and Tyr183, which interact with the -1 sugar residue of the substrate. To investigate their role in catalysis, we have successfully expressed wild-type enzyme and mutants of these residues as inclusion bodies in Escherichia coli. After refolding and purification they were characterized by both structural and enzyme kinetic studies. Mutation of Tyr183 to phenylalanine produced an enzyme with a lower k(cat) and a slightly higher K(m) than the wild-type enzyme. Mutating Asp125 and Glu127 to alanine gave mutants with approximately 2% residual activity. In contrast, the Asp125Asn mutant retained substantial activity, with an approximately twofold lower k(cat) and an approximately twofold higher K(m) than the wild-type enzyme. More interestingly, it showed activity to higher pH values than the other variants. The X-ray structure of the Asp125Ala/Glu127Ala double mutant soaked with chitotetraose shows that, compared with wild-type hevamine, the carbonyl oxygen atom of the N-acetyl group of the -1 sugar residue has rotated away from the C1 atom of that residue. The combined structural and kinetic data show that Asp125 and Tyr183 contribute to catalysis by positioning the carbonyl oxygen of the N-acetyl group near to the C1 atom. This allows the stabilization of a positively charged transient intermediate, in agreement with a previous proposal that the enzyme makes use of substrate-assisted catalysis.     

Mutation of the glycosylated asparagine residue 286 in human CLN2 protein results in loss of enzymatic activity

Late infantile neuronal ceroid lipofuscinosis (LINCL) is caused by the deficiency of the lysosomal tripeptidyl peptidase-I encoded by CLN2. We previously detected in two LINCL patients a homozygous missense mutation, p.Asn286Ser, that affects a potential N-glycosylation site. We introduced the p.Asn286Ser mutation into the wild-type CLN2 cDNA and performed transient expression analysis to determine the effect on the catalytic activity, intracellular targeting, and glycosylation of the CLN2 protein. Expression of mutant p.Asn286Ser CLN2 in HEK293 cells revealed that the mutant was enzymatically inactive. Western blot analysis demonstrated that at steady state the amounts of expressed p.Asn286Ser CLN2 were reduced compared with wild-type expressing cells. The rate of synthesis and the sorting of the newly synthesized p.Asn286Ser CLN2 in the Golgi was not affected compared with wild-type CLN2 protein. The electrophoretic mobility of the immunoprecipitated mutant p.Asn286Ser CLN2 was increased by approximately 2 kDa compared with the wild-type CLN2 protein, whereas deglycosylation led to the generation of polypeptides of the same apparent size. The data suggest that mutant p.Asn286Ser CLN2 lacks one oligosaccharide chain resulting in enzymatic inactivation.     

Functional topology of a surface loop shielding the catalytic center in lipoprotein lipase.

Lipoprotein lipase (LPL), hepatic lipase, and pancreatic lipase show high sequence homology to one another. The crystal structure of pancreatic lipase suggests that it contains a trypsin-like Asp-His-Ser catalytic triad at the active center, which is shielded by a disulfide bridge-bounded surface loop that must be repositioned before the substrate can gain access to the catalytic residues. By sequence alignment, the homologous catalytic triad in LPL corresponds to Asp156-His241-Ser132, absolutely conserved residues, and the homologous surface loop to residues 217-238, a poorly conserved region. To verify these assignments, we expressed in vitro wild-type LPL and mutant LPLs having single amino acid mutations involving residue Asp156 (to His, Ser, Asn, Ala, Glu, or Gly), His241 (to Asn, Ala, Arg, Gln, or Trp), or Ser132 (to Gly, Ala, Thu, or Asp) individually. All 15 mutant LPLs were totally devoid of enzyme activity, while wild-type LPL and other mutant LPLs containing substitutions in other positions were fully active. We further replaced the 22-residue LPL loop which shields the catalytic center either partially (replacing 6 of 22 residues) or completely with the corresponding hepatic lipase loop. The partial loop-replacement chimeric LPL was found to be fully active, and the complete loop-replacement mutant had approximately 60% activity, although the primary sequence of the hepatic lipase loop is quite different. In contrast, replacement with the pancreatic lipase loop completely inactivated the enzyme. Our results are consistent with Asp156-His241-Ser132 being the catalytic triad in lipoprotein lipase.(ABSTRACT TRUNCATED AT 250 WORDS)     

Alpha-helix stability in proteins. I. Empirical correlations concerning substitution of side-chains at the N and C-caps and the replacement of alanine by glycine or serine at solvent-exposed surfaces.

The importance of amino acid side-chains in helix stability has been investigated by making a series of mutations at the N-caps, C-caps and internal positions of the solvent-exposed faces of the two alpha-helices of barnase. There is a strong positional and context dependence of the effect of a particular amino acid on stability. Correlations have been found that provide insight into the physical basis of helix stabilization. The relative effects of Ala and Gly (or Ser) may be rationalized on the basis of solvent-accessible surface areas: burial of hydrophobic surface stabilizes the protein as does exposure to solvent of unpaired hydrogen bond donors or acceptors in the protein. There is a good correlation between the relative stabilizing effects of Ala and Gly at internal positions with the total change in solvent-accessible hydrophobic surface area of the folded protein on mutation of Ala----Gly. The relationship may be extended to the N and C-caps by including an extra term in hydrophilic surface area for the solvent exposure of the non-intramolecularly hydrogen-bonded main-chain CO, NH or protein side-chain hydrogen bonding groups. The requirement for solvent exposure of the C-cap main-chain CO groups may account for the strong preference for residues having positive phi and psi angles at this position, since this alpha L-conformation results in the largest solvent exposure of the C-terminal CO groups. Glycine in an alpha L-conformation results in the greatest exposure of these CO groups. Further, the side-chains of His, Asn, Arg and Lys may, with positive phi and psi-angles, form a hydrogen bond with the backbone CO of residue in position C -3 (residues are numbered relative to the C-cap). The preferences at the C-cap are Gly much greater than His greater than Asn greater than Arg greater than Lys greater than Ala approximately Ser approximately greater than Asp. The preferences at the N-cap are determined by hydrogen bonding of side-chains or solvent to the exposed backbone NH groups and are: Thr approximately Asp approximately Ser greater than Gly approximately Asn greater than Gln approximately Glu approximately His greater than Ala greater than Val much greater than Pro. These general trends may be obscured when mutation allows another side-chain to become a surrogate cap.(ABSTRACT TRUNCATED AT 400 WORDS)