A thermodynamic scale for leucine zipper stability and dimerization specificity: e and g interhelical interactions.

The leucine zipper is a dimeric coiled-coil protein structure composed of two amphipathic alpha-helices with the hydrophobic surfaces interacting to create the dimer interface. This structure has been found to mediate the dimerization of two abundant classes of DNA binding proteins: the bZIP and bHLH-Zip proteins. Several workers have reported that amino acids in the e and g positions of the coiled coil can modulate dimerization stability and specificity. Using the bZIP protein VBP as a host molecule, we report a thermodynamic scale (delta delta G) for 27 interhelical interactions in 35 proteins between amino acids in the g and the following e positions (g<==>e') of a leucine zipper coiled coil. We have examined the four commonly occurring amino acids in the e and g positions of bZIP proteins, lysine (K), arginine (R), glutamine (Q), glutamic acid (E), as well as the only other remaining charged amino acid aspartic acid (D), and finally alanine (A) as a reference amino acid. These results indicate that E<==>R is the most stable interhelical pair, being 0.35 kcal/mol more stable than E<==>K. A thermodynamic cycle analysis shows that the E<==>R pair is 1.33 kcal/mol more stable than A<==>A with -1.14 kcal/mol of coupling energy (delta delta Gint) coming from the interaction of E with R. The E<==>K coupling energy is only -0.14 kcal/mol. E interacts with more specificity than Q. The R<==>R pair is less stable than the K<==>K by 0.24 kcal/mol. R interacts with more specificity than K. Q forms more stable pairs with the basic amino acids K and R rather than with E. Changing amino acids in the e position to A creates bZIP proteins that form tetramers.     

Novel actin-related proteins Arp-T1 and Arp-T2 as components of the cytoskeletal calyx of the mammalian sperm head

The calyx is a large cytoskeletal component of the perinuclear theca of the mammalian sperm head, displaying remarkable morphological interspecies differences, which is biochemically characterized by resistance to high ionic strength and detergents and by a special protein composition, including the basic proteins calicin, cylicin I and II, and two major actin-capping proteins. In our calyx preparations from bull spermatozoa we have noted two major acidic components which upon partial amino acid sequencing have been identified as novel members of the subfamily of actin-related proteins (Arps). Antibodies raised against the corresponding human proteins, termed Arp-T1 and Arp-T2, have been used to detect the proteins by immunoblotting and immunofluorescence microscopy, demonstrating their specific synthesis in the testis, late in spermatid differentiation, and their localization in the calyx. The discovery of two novel Arps as major components in a cytoskeletal, nonmotile structure of mammalian spermatozoa suggests that certain members of this family of proteins may serve functions other than nucleation of actin filaments, and possible biological roles of such Arps in spermatozoa are discussed.     

Identification of a common hyaluronan binding motif in the hyaluronan binding proteins RHAMM, CD44 and link protein.

We have previously identified two hyaluronan (HA) binding domains in the HA receptor, RHAMM, that occur near the carboxyl-terminus of this protein. We show here that these two HA binding domains are the only HA binding regions in RHAMM, and that they contribute approximately equally to the HA binding ability of this receptor. Mutation of domain II using recombinant polypeptides of RHAMM demonstrates that K423 and R431, spaced seven amino acids apart, are critical for HA binding activity. Domain I contains two sets of two basic amino acids, each spaced seven residues apart, and mutation of these basic amino acids reduced their binding to HA--Sepharose. These results predict that two basic amino acids flanking a seven amino acid stretch [hereafter called B(X7)B] are minimally required for HA binding activity. To assess whether this motif predicts HA binding in the intact RHAMM protein, we mutated all basic amino acids in domains I and II that form part of these motifs using site-directed mutagenesis and prepared fusion protein from the mutated cDNA. The altered RHAMM protein did not bind HA, confirming that the basic amino acids and their spacing are critical for binding. A specific requirement for arginine or lysine residues was identified since mutation of K430, R431 and K432 to histidine residues abolished binding. Clustering of basic amino acids either within or at either end of the motif enhanced HA binding activity while the occurrence of acidic residues between the basic amino acids reduced binding. The B(X7)B motif, in which B is either R or K and X7 contains no acidic residues and at least one basic amino acid, was found in all HA binding proteins molecularly characterized to date. Recombinant techniques were used to generate chimeric proteins containing either the B(X7)B motifs present in CD44 or link protein, with the amino-terminus of RHAMM (amino acids 1-238) that does not bind HA. All chimeric proteins containing the motif bound HA in transblot analyses. Site-directed mutations of these motifs in CD44 sequences abolished HA binding. Collectively, these results predict that the motif of B(X7)B as a minimal binding requirement for HA in RHAMM, CD44 and link protein, and occurs in all HA binding proteins described to date.     

Mutagenic analysis of the a subunit of the F1F0 ATP synthase in Escherichia coli: Gln-252 through Tyr-263.

The a subunit of F1F0 ATP synthase contains a highly conserved region near its carboxyl terminus which is thought to be important in proton translocation. Cassette site-directed mutagenesis was used to study the roles of four conserved amino acids Gln-252, Phe-256, Leu-259, and Tyr-263. Substitution of basic amino acids at each of these four sites resulted in marked decreases in enzyme function. Cells carrying a subunit mutations Gln-252-->Lys, Phe-256-->Arg, Leu-259-->Arg, and Tyr-263-->Arg all displayed growth characteristics suggesting substantial loss of ATP synthase function. Studies of both ATP-driven proton pumping and proton permeability of stripped membranes indicated that proton translocation through F0 was affected by the mutations. Other mutations, such as the Phe-256-->Asp mutation, also resulted in reduced enzyme activity. However, more conservative amino acid substitutions generated at these same four positions produced minimal losses of F1F0 ATP synthase. The effects of mutations and, hence, the relative importance of the amino acids for enzyme function appeared to decrease with proximity to the carboxyl terminus of the a subunit. The data are most consistent with the hypothesis that the region between Gln-252 and Tyr-263 of the a subunit has an important structural role in F1F0 ATP synthase.     

Vesicular stomatitis virus glycoprotein mutations that affect membrane fusion activity and abolish virus infectivity.

We have introduced amino acid substitutions into two regions of the extracellular domain of the vesicular stomatitis virus (VSV) glycoprotein (G protein) and examined the effect of these mutations on protein transport, low-pH-induced stability of G protein oligomers, and membrane fusion activity. We suggested previously that the region between amino acids 118 and 139 may be important for the membrane fusion activity of G protein, on the basis of the characterization of a fusion-defective G protein mutant (M. A. Whitt, P. Zagouras, B. Crise, and J. K. Rose, J. Virol. 64:4907-4913, 1990). It has also been postulated by others that this region as well as the region between amino acids 181 and 212 may constitute putative internal fusion domains of VSV G protein. In this report, we show that three different amino acids substitutions between residues 118 and 139 (G-124-->E, P-127-->D, and A-133-->K) either altered or abolished low-pH-dependent membrane fusion activity. In contrast, substitutions between residues 192 and 212 resulted either in G proteins that had wild-type fusion activity or in mutant proteins in which the mutation prevented transport of G protein to the cell surface. Two of the substitutions between residues 118 and 139 (G-124-->E and P-127-->D) resulted in G proteins that were fusion defective at pH 5.7, although syncytia were observed after cells were treated with fusion buffer at pH 5.5, albeit at levels significantly less than that induced by wild-type G protein. Interestingly, when either G-124-->E or P-127-->D was incorporated into tsO45 virions, the resulting particles were not infectious, presumably because the viral envelope was not able to fuse with the proper intracellular membrane. These results support the hypothesis that the region between amino acids 118 and 139 is important for the membrane fusion activity of VSV G protein and may constitute an internal fusion domain.     

Determination and analysis of the primary structure of the nerve terminal specific phosphoprotein, synapsin I.

A rat brain cDNA clone containing an open reading frame encoding the neuron-specific protein synapsin I has been sequenced. The sequence predicts a protein of 691 amino acids with a mol. wt of 73 kd. This is in excellent agreement with the size of rat brain synapsin Ib measured by SDS--polyacrylamide gel electrophoresis. Inspection of the predicted primary structure has revealed the probable sites for synapsin I phosphorylation by the cAMP-dependent and Ca2+/calmodulin-dependent protein kinases. All of the biochemically observed intermediates of synapsin I digestion by collagenase can be verified by inspection of the sequence, and the collagenase-resistant fragment has been defined as the amino-terminal 439 amino acids of the molecule. Predictions of sequence secondary structure and hydrophobicity suggest that a central domain of approximately 270 amino acids may exist as a folded, globular core. The carboxyl-terminal domain of the protein (the region sensitive to collagenase digestion) contains sites for Ca2+/calmodulin-dependent protein kinase phosphorylation. These sites are flanked by three regions of repeating amino acid sequence that are proposed to be the synaptic vesicle-binding domain of synapsin I. This region also shares homology with the actin-binding proteins profilin and villin. The characteristics of the synapsin I sequence do not support extensive homology with the erythrocyte cytoskeletal protein 4.1.     

The phosphorylation status of the serine-rich region of the human cytomegalovirus 86-kilodalton major immediate-early protein IE2/IEP86 affects temporal viral gene expression

The 86-kDa major immediate-early protein (IE2/IEP86) of human cytomegalovirus (HCMV) contains a serine-rich region (amino acids 258 to 275) with several consensus casein kinase II (CKII) sites. We performed extensive mutational analysis of this region, changing serines to alternating alanines and glycines. Mutation of the serines between amino acids 266 and 275 eliminated in vitro phosphorylation by CKII. In vitro CKII phosphorylation of the serines between amino acids 266 and 269 or between amino acids 271 and 275 inhibited the ability of IE2/IEP86 to bind to TATA-binding protein. Correspondingly, nonphosphorylatable mutants in these regions showed increased activation of specific HCMV gene promoters in transfection studies. Viruses containing mutations of the serines throughout the entire region (amino acids 258 to 275) or the second half (amino acids 266 to 275) of the region showed delayed expression of all viral proteins tested and, correspondingly, delayed growth compared to wild-type HCMV. Mutation of the serines in the first half of the serine-rich region (amino acids 258 to 264) or between amino acids 266 and 269 propagated very slowly and has not been further studied. In contrast, mutation of the serines between amino acids 271 and 275 resulted in accelerated virus growth and accelerated temporal expression of viral proteins. These results suggest that the serine-rich region is structurally complex, possibly affecting multiple functions of IE2/IEP86. The data show that the phosphorylation state of the serine-rich region, particularly between amino acids 271 and 275, modulates the temporal expression of viral genes.     

Genetic Code-guided Protein Synthesis and Folding in Escherichia coli

Universal genetic codes are degenerated with 61 codons specifying 20 amino acids, thus creating synonymous codons for a single amino acid. Synonymous codons have been shown to affect protein properties in a given organism. To address this issue and explore how Escherichia coli selects its “codon-preferred” DNA template(s) for synthesis of proteins with required properties, we have designed synonymous codon libraries based on an antibody (scFv) sequence and carried out bacterial expression and screening for variants with altered properties. As a result, 342 codon variants have been identified, differing significantly in protein solubility and functionality while retaining the identical original amino acid sequence. The soluble expression level varied from completely insoluble aggregates to a soluble yield of ∼2.5 mg/liter, whereas the antigen-binding activity changed from no binding at all to a binding affinity of > 10⁻⁸ m. Not only does our work demonstrate the involvement of genetic codes in regulating protein synthesis and folding but it also provides a novel screening strategy for producing improved proteins without the need to substitute amino acids.     

Import of a new chloroplast inner envelope protein is greatly stimulated by potassium phosphate

A cDNA clone encoding a major chloroplast inner envelope membrane protein of 96 kDa (IEP96) was isolated and characterized. The protein is synthesized as a larger-molecular-weight precursor (pIEP96) which contains a cleavable N-terminal transit sequence of 50 amino acids. The transit peptide exhibits typical stromal targeting information. It is cleaved in vitro by the stromal processing peptidase, though the mature protein is clearly localized in the inner envelope membrane. Translocation of pIEP96 into chloroplasts is greatly stimulated in the presence of 80 mM potassium phosphate which results in an import efficiency of about 90%. This effect is specific for potassium and phosphate, but cannot be ascribed to a membrane potential across the inner envelope membrane. Protein sequence analysis reveals five stretches of repeats of 26 amino acids in length. The N-terminal 300 amino acids are 45% identical (76% similarity) to the 35 kDa -subunit of acetyl-CoA carboxyl-transferase from Escherichia coli. The C-terminal 500 amino acids share significant similarity (69%) with USOI, a component of the cytoskeleton in yeast.Abbreviations P_i phosphate - IEP inner envelope membrane protein - pIEP precursor form of IEP - SSU small subunit of ribulose-1,5-bisphosphate carboxylase oxygenase - IEP96_pep peptide specific antiserum to IEP96 - IEP96_pol polyspecific antiserum to IEP96     

The primary structure of the core protein of the small, leucine-rich proteoglycan (PG I) from bovine articular cartilage

Two forms of small, interstitial proteoglycans have been isolated from bovine articular cartilage and have different core proteins, based on NH2-terminal analysis and peptide mapping (Choi, H. U., Johnson, T. L., Pal, S., Tang, L-H., Rosenberg, L. C., and Neame, P. J. (1989) J. Biol. Chem. 264, 2876-2884). These proteoglycans have been called PG I and PG II. Since they were first described, they have also been called "biglycan" (PG I), "decorin," and "DS-PG" (PG II). This report describes the primary structure of PG I from bovine articular cartilage. The protein core consists of 331 amino acids with a molecular mass of 37,280 Da. The amino acid sequence shows 55% identity to the cDNA-derived sequence of PG II from bovine bone. There are four discrete domains in the amino acid sequence. Domain 1, at the NH2 terminus (approximately 23 amino acids), contains two sites of attachment of dermatan sulfate, both of which match the consensus sequence of Asp/Glu-X-X-Ser-Gly-hydrophobic. Neither of these sites is substituted to 100% with glycosaminoglycan in native PG I. Domain 2, near the NH2 terminus and containing approximately 28 amino acids, has a cysteine pattern similar to a domain near the COOH terminus of mouse metallothionein and contains at least one disulfide bond (between the first and fourth cysteine residues). The majority of the core protein of PG I (domain 3) is a leucine-rich domain containing ten repeating units (approximately 231 amino acids). Patthy [1987) J. Mol. Biol. 198, 567-577) has shown that for PG II, the majority of domain 3 shows considerable similarity to leucine-rich alpha 2-glycoprotein (LRG) from serum. Domain 2 of PG I or PG II also has an analog in LRG, in that it has two cysteines in a similar place. The major motif in the PG I described here, in PG II and in LRG, is a series of leucine-rich repeats. PG I and PG II both contain 10 leucine-rich repeats which are 14 amino acids long and which are somewhat irregularly spaced, while LRG contains 9 leucine-rich repeats spaced 10 amino acids apart. Other proteins which contain leucine repeats are the platelet glycoprotein Ib, which is involved in platelet adherence to subendothelium (eight repeats in the alpha chain and two in the beta chain), the protein encoded by the Toll gene (involved in lateral and ventral spatial organization in Drosophila) and chaoptin (a protein involved in Drosophila photoreceptor morphogenesis).(ABSTRACT TRUNCATED AT 400 WORDS)     

Three basic regions in adenovirus DNA polymerase interact differentially depending on the protein context to function as bipartite nuclear localization signals.

Adenovirus DNA polymerase (AdPol) contains three clusters of basic amino acids within the N-terminal 48 amino acids: RARR, which begins at amino acid 8, RRRVR, which begins at amino acid 25, and RARRRR, which begins at amino acid 41. These clusters are designated BS I, BS II, and BS III, respectively. (The amino acid codes are: R, arginine; A, alanine; V, valine.) Mutational analysis of these noncontiguous clusters showed that AdPol contains a novel organization of bipartite nuclear localization signals (NLS) that interact differentially to serve in the nuclear targeting of AdPol or of chimeric proteins in which AdPol is linked to Escherichia coli beta-galactosidase (beta-gal). The region containing BS I and BS II functioned interdependently as an NLS for the nuclear targeting of AdPol, for which BS III was dispensible. However, the region containing BS II and BS III constituted a second and more efficient bipartite NLS for the nuclear targeting of the AdPol-E. coli beta-gal fusion protein. Moreover, deletion or limited insertion of amino acids in the spacer region between BS II and BS III did not affect their nuclear targeting function for these fusion proteins. Chou-Fasman predictive analysis of protein secondary structure in the vicinity of the bipartite NLS sequences supports a model in which protein conformation in the spacer region may play an important role in bringing these clusters of basic amino acids into close proximity, allowing them to function as nuclear targeting signals for this class of nuclear proteins.     

B-Phycoerythrin from Rhodella violacea: characterization of two isoproteins.

Two isoproteins of the "native" B-phycoerythrin of the red alga, Rhodella violacea, were purified from crude extracts by preparative polyacrylamide gel electrophoresis and subsequently characterized. The slower moving pigment in gel electrophoresis was designated B-PE I, the faster as B-PE II. Both were found to occur in about equal amounts. B-PE I has a molecular weight of about 280000 and an IEP at 4.39, B-PE II a molecular weight of nearly 265000 and an IEP at 4.23. B-PE I and II are characterized by absorption maxima at 568 and 542 nm and a shoulder at 500 nm in the visible part of the absorption spectra. Their absorption coefficients at 542 nm differ with values of 5.54 and 5.63, respectively. The fluorescence emission spectra show a single maximum at 575 for B-PE I and at 578 nm for B-PE II. Both spectra have a shoulder at 630 nm. The fluorescence yield of B-PE II is lower by 25%. In calibrated SDS gel electrophoresis of the purified pigments B-PE I and II show two subunits of molecular weights of 18900 and 29200 and 18500 and 29900, respectively. Quantitative amino acid analyses indicated, that the isoproteins are very similar. B-PE II, however, has a significantly higher content of acidic amino acids and a lower percentage of basic residues, which is in keeping with its lower isoelectric point. Functional aspects of the occurrence of two isoproteins of B-phycoerythrin are discussed.     

The EICP22 protein of equine herpesvirus 1 physically interacts with the immediate-early protein and with itself to form dimers and higher-order complexes

The EICP22 protein (EICP22P) of Equine herpesvirus 1 (EHV-1) is an early protein that functions synergistically with other EHV-1 regulatory proteins to transactivate the expression of early and late viral genes. We have previously identified EICP22P as an accessory regulatory protein that has the ability to enhance the transactivating properties and the sequence-specific DNA-binding activity of the EHV-1 immediate-early protein (IEP). In the present study, we identify EICP22P as a self-associating protein able to form dimers and higher-order complexes during infection. Studies with the yeast two-hybrid system also indicate that physical interactions occur between EICP22P and IEP and that EICP22P self-aggregates. Results from in vitro and in vivo coimmunoprecipitation experiments and glutathione S-transferase (GST) pull-down studies confirmed a direct protein-protein interaction between EICP22P and IEP as well as self-interactions of EICP22P. Analyses of infected cells by laser-scanning confocal microscopy with antibodies specific for IEP and EICP22P revealed that these viral regulatory proteins colocalize in the nucleus at early times postinfection and form aggregates of dense nuclear structures within the nucleoplasm. Mutational analyses with a battery of EICP22P deletion mutants in both yeast two-hybrid and GST pull-down experiments implicated amino acids between positions 124 and 143 as the critical domain mediating the EICP22P self-interactions. Additional in vitro protein-binding assays with a library of GST-EICP22P deletion mutants identified amino acids mapping within region 2 (amino acids [aa] 65 to 196) and region 3 (aa 197 to 268) of EICP22P as residues that mediate its interaction with IEP.     

T-antigen kinase inhibits simian virus 40 DNA replication by phosphorylation of intact T antigen on serines 120 and 123.

Simian virus 40 (SV40) DNA replication begins after two large T-antigen hexamers assemble on the viral minimal origin of replication and locally unwind the template DNA. The activity of T antigen in this reaction is regulated by its phosphorylation state. A form of casein kinase I purified from HeLa nuclear extracts (T-antigen kinase) phosphorylates T antigen on physiologic sites and inhibits its activity in the unwinding reaction (A. Cegielska and D. M. Virshup, Mol. Cell. Biol. 13:1202-1211, 1993). Using a series of mutant T antigens expressed by recombinant baculoviruses in Sf9 cells, we find that the origin unwinding activities of both TS677-->A and TS677,679-->A are inhibited by the T-antigen kinase, as is wild-type T antigen. In contrast, mutants TS120-->A and TS123,679-->A are resistant to inhibition by the kinase. Thus, phosphorylation of serines 120 and 123 is necessary for inhibition of T-antigen activity. Previous studies of casein kinase I substrate specificity have suggested that acidic residues or a phosphorylated amino acid amino terminal to the target residue are required to create a casein kinase I recognition site. However, we find that the T-antigen kinase can add more than 3 mol of Pi per mol to full-length bacterially produced T antigen and that it inhibits the unwinding activity of p34cdc2-activated bacterially produced T antigen. Since no prior phosphorylation is present in this bacterially produced T antigen, and no acidic residues are present immediately amino terminal to serines 120 and 123, other structural elements of T antigen must contribute to the recognition signals for T-antigen kinase. In support of this conclusion, we find that while T-antigen kinase phosphorylates amino-terminal residues in bacterially produced full-length T antigen, it cannot phosphorylate bacterially produced truncated T antigen containing amino acids 1 to 259, a 17-kDa amino-terminal tryptic fragment of T antigen, nor can it phosphorylate denatured T antigen. These findings strongly suggest that the carboxy-terminal domain of T antigen is an important modifier of the recognition signals for phosphorylation of the critical amino-terminal sites by the T-antigen kinase. This conclusion is consistent with previous studies suggesting close apposition of amino- and carboxy-terminal domains of T antigen in the native protein. The three-dimensional conformation of the substrate appears to make a significant contribution to T-antigen kinase substrate specificity.     

Barrier-free paths of directed protein motion in the erythrocyte plasma membrane.

A model is presented for the steric interaction between a plasma membrane protein and the membrane cytoskeleton in the human erythrocyte. The cytoskeleton is treated as a network of polymer chains attached to a flat bilayer, and the membrane protein is a hemisphere of effective radius R(e) with center on the bilayer edge. The simulation is used to investigate the barrier-free path L for linear guided motion of a protein in the bilayer plane. It is shown that the barrier-free paths of small proteins can be used to extract the effective in-plane diameter of cytoskeletal components. For example, the in-plane diameter of an ankyrin attachment site is found to be approximately 12 nm in the simulation, or twice the computational spectrin diameter. The barrier-free paths of large proteins (R(e) > 23 nm) vanish when the proteins are corralled by the cytoskeleton. For intermediate size proteins, L decreases approximately as L is directly proportional to S-1.4 where S is proportional to the sum of the protein and cytoskeleton chain radii.     

Deduced protein sequence of bone small proteoglycan I (biglycan) shows homology with proteoglycan II (decorin) and several nonconnective tissue proteins in a variety of species

The small proteoglycans (PG) of bone consist of two different molecular species: one containing one chondroitin sulfate chain (PG II) and the other, two chains (PG I). These two proteoglycans are found in many connective tissues and have Mr = 45,000 core proteins with clear differences in their NH2-terminal sequences. Using antisera produced against synthetic peptides derived from the human PG I and PG II NH2 termini, we have isolated several cDNA clones from a lambda gt11 expression library made against mRNA isolated from human bone-derived cells. The clones, which reacted with antisera to the PG II peptide, were sequenced and found to be identical with the PG II class of proteoglycan from human fibroblasts known as PG-40 or decorin. The clones reacting to the PG I antisera, however, had a unique sequence. The derived protein sequence of PG I showed sufficient homology with the PG II sequence (55% of the amino acids are identical, with most others involving chemically similar amino acid substitutions) to strongly suggest that the two proteins were the result of a gene duplication. PG II (decorin) contains one attached glycosaminoglycan chain, while PG I probably contains two chains. For this reason, we suggest that PG I be called biglycan. The biglycan protein sequence contains 368 residues (Mr = 42,510 for the complete sequence and Mr = 37,983 for the secreted form) that appears to consist predominantly of a series of 12 tandem repeats of 24 residues. The repeats are recognized by their conserved leucines (and leucine-like amino acids) in positions previously reported for a diverse collection of proteins (none of which is thought to be proteoglycans) including: two morphogenic proteins (toll and chaoptin) in the fruit fly; a yeast adenylate cyclase; and two human proteins, the von Willebrand Factor-binding platelet membrane protein, GPIb, and a rare serum protein, leucine-rich glycoprotein.     

Nuclear basic protein transition during sperm differentiation. Primary structure of the spermatid-specific protein S2 from the dog-fish Scylliorhinus caniculus

The remodeling of nucleoproteins during dog-fish spermiogenesis involves two successive nuclear protein transitions: the first from somatic-type histones to transition proteins during the nuclear elongation of spermatids and the second leading to protamine-DNA association in mature spermatozoa. The chromatin of elongating spermatids contains two transition proteins called S1 and S2. The amino acid sequence of protein S1, a polypeptide of 87 residues was determined previously [Chauvière, M., Martinage, A., Briand, G., Sautière, P. & Chevaillier, Ph. (1987) Eur. J. Biochem. 169, 105-111]. In the present paper, we report the elucidation of the primary structure of the minor transition protein S2 established by automated Edman degradation of the protein and of its fragments generated by cleavage at methionine and aspartate residues. S2 contains 80 residues and has a molecular mass of 9726 Da. S2 is mainly characterized by a high content of basic amino acids mostly represented by lysine, a relatively high level of hydrophobic residues, the presence of six phosphorylatable residues and the lack of cysteine. Its amino acid sequence shows that the N-terminal half is highly basic, while the acidic residues are located in the C-terminal part of the protein where more diversity in amino acids is noticed. The two transition proteins S1 and S2 share striking structural similarities. Few but significative similarities have been detected with the mammalian transition protein TP1 [Kistler, W. S., Noyes, C., Hsu, R. & Heinrikson, R. L. (1975) J. Biol. Chem. 250, 1847-1853], suggesting similar functions for all these proteins in chromatin remodeling during sperm differentiation. By contrast, the two dog-fish spermatid-specific proteins are structurally unrelated to sperm protamines and cannot be considered as their precursors.