The S2 gene nucleotide sequences of prototype strains of the three reovirus serotypes: characterization of reovirus core protein sigma 2.

The S2 gene nucleotide sequences of prototype strains of the three reovirus serotypes were determined to gain insight into the structure and function of the S2 translation product, virion core protein sigma 2. The S2 sequences of the type 1 Lang, type 2 Jones, and type 3 Dearing strains are 1,331 nucleotides in length and contain a single large open reading frame that could encode a protein of 418 amino acids, corresponding to sigma 2. The deduced sigma 2 amino acid sequences of these strains are very conserved, being identical at 94% of the sequence positions. Predictions of sigma 2 secondary structure and hydrophobicity suggest that the protein has a two-domain structure. A larger domain is suggested to be formed from the amino-terminal three-fourths of sigma 2 sequence, which is separated from a smaller carboxy-terminal domain by a turn-rich hinge region. The carboxy-terminal domain includes sequences that are more hydrophilic than those in the rest of the protein and contains sequences which are predicted to form an alpha-helix. A region of striking similarity was found between amino acids 354 and 374 of sigma 2 and amino acids 1008 and 1031 of the beta subunit of the Escherichia coli DNA-dependent RNA polymerase. We suggest that the regions with similar sequence in sigma 2 and the beta subunit form amphipathic alpha-helices which may play a related role in the function of each protein. We have also performed experiments to further characterize the double-stranded RNA-binding activity of sigma 2 and found that the capacity to bind double-stranded RNA is a property of the sigma 2 protein of prototype strains and of the S2 mutant tsC447.     

Effects of cyclic GMP on the secondary structure of cyclic GMP dependent protein kinase and analysis of the enzyme's amino-terminal domain by far-ultraviolet circular dichroism

Far-UV circular dichroism spectra of bovine lung cyclic GMP dependent protein kinase (G-kinase) show that the enzyme contains alpha-helical and beta-pleated sheet elements. Binding of cyclic GMP changes the spectra in a way consistent with the induction of beta-sheet from random coil. Examination of the amino-terminal sequence of G-kinase indicates the presence of a strongly alpha-helical segment with several features in common with the leucine zipper motif. We propose that this sequence may be the important part of the dimerization domain of the enzyme. A synthetic peptide corresponding to amino acids 1-39 of G-kinase has a strongly alpha-helical CD spectrum, supporting the predicted secondary structure of this amino-terminal sequence. In contrast to the native enzyme, a structure reduced in alpha-helix was found when a constitutively active form of G-kinase, which lacks amino acids 1-77, was studied.     

Structural characterization of the SARS-coronavirus spike S fusion protein core

The spike (S) glycoprotein of coronaviruses mediates viral entry into host cells. It is a type 1 viral fusion protein that characteristically contains two heptad repeat regions, denoted HR-N and HR-C, that form coiled-coil structures within the ectodomain of the protein. Previous studies have shown that the two heptad repeat regions can undergo a conformational change from their native state to a 6-helix bundle (trimer of dimers), which mediates fusion of viral and host cell membranes. Here we describe the biophysical analysis of the two predicted heptad repeat regions within the severe acute respiratory syndrome coronavirus S protein. Our results show that in isolation the HR-N region forms a stable alpha-helical coiled coil that associates in a tetrameric state. The HR-C region in isolation formed a weakly stable trimeric coiled coil. When mixed together, the two peptide regions (HR-N and HR-C) associated to form a very stable alpha-helical 6-stranded structure (trimer of heterodimers). Systematic peptide mapping showed that the site of interaction between the HR-N and HR-C regions is between residues 916-950 of HR-N and residues 1151-1185 of HR-C. Additionally, interchain disulfide bridge experiments showed that the relative orientation of the HR-N and HR-C helices in the complex was antiparallel. Overall, the structure of the hetero-stranded complex is consistent with the structures observed for other type 1 viral fusion proteins in their fusion-competent state.     

Identification of carbohydrate-binding domains in the attachment proteins of type 1 and type 3 reoviruses

The reovirus attachment protein, sigma1, is responsible for strain-specific patterns of viral tropism in the murine central nervous system and receptor binding on cultured cells. The sigma1 protein consists of a fibrous tail domain proximal to the virion surface and a virion-distal globular head domain. To better understand mechanisms of reovirus attachment to cells, we conducted studies to identify the region of sigma1 that binds cell surface carbohydrate. Chimeric and truncated sigma1 proteins derived from prototype reovirus strains type 1 Lang (T1L) and type 3 Dearing (T3D) were expressed in insect cells by using a baculovirus vector. Assessment of expressed protein susceptibility to proteolytic cleavage, binding to anti-sigma1 antibodies, and oligomerization indicates that the chimeric and truncated sigma1 proteins are properly folded. To assess carbohydrate binding, recombinant sigma1 proteins were tested for the capacity to agglutinate mammalian erythrocytes and to bind sialic acid presented on glycophorin, the cell surface molecule bound by type 3 reovirus on human erythrocytes. Using a panel of two wild-type and ten chimeric and truncated sigma1 proteins, the sialic acid-binding domain of type 3 sigma1 was mapped to a region of sequence proposed to form the more amino terminal of two predicted beta-sheet structures in the tail. This unit corresponds to morphologic region T(iii) observed in computer-processed electron micrographs of sigma1 protein purified from virions. In contrast, the homologous region of T1L sigma1 sequence was not implicated in carbohydrate binding; rather, sequences in the distal portion of the tail known as the neck were required. Results of these studies demonstrate that a functional receptor-binding domain, which uses sialic acid as its ligand, is contained within morphologic region T(iii) of the type 3 sigma1 tail. Furthermore, our findings indicate that T1L and T3D sigma1 proteins contain different arrangements of receptor-binding domains.     

Exploring the conformational roles of signal sequences: synthesis and conformational analysis of lambda receptor protein wild-type and mutant signal peptides

Secretion of the Escherichia coli lambda receptor protein (LamB protein) appears from genetic evidence to be correlated with the predicted tendency of its signal sequence to adopt an alpha-helical conformation [Emr, S. D., & Silhavy, T. J. (1983) Proc. Natl. Acad. Sci. U. S. A. 80, 4599]. We have tested this hypothesis by synthesizing major portions of signal sequences from the wild-type and mutant LamB proteins and analyzing their conformations by circular dichroism. The wild-type signal sequence contains a seven-residue hydrophobic region flanked by a proline and a glycine. Chou-Fasman rules predict that this segment will adopt an alpha-helical conformation. An export-deficient mutant is missing four residues from this region; the helix-breaking glycine and proline are thus separated by only three residues, and an alpha helix is not predicted to form. In each of the export-restored revertants, either the glycine or the proline is replaced with a residue which promotes helix formation. The helix content of the synthetic signal sequence fragments on the basis of CD measurements supports the secondary structure hypothesis described above. The relative helicity in aqueous sodium dodecyl sulfate, lysolecithin, or trifluoroethanol is as follows: wild type greater than R2 (Pro----Leu) greater than R1 (Gly----Cys) much greater than deletion mutant.     

Two genes encode the major membrane-associated protein of methanol-induced peroxisomes from Candida boidinii

A massive proliferation of peroxisomes occurs in the yeast Candida boidinii when methanol is utilized as the sole carbon source; these peroxisomes contain the enzymes which catalyze the initial steps of methanol utilization. The most abundant peroxisomal membrane-associated protein has an apparent molecular mass of 20 kDa and is termed PMP20. We report the isolation of two genes that encode very similar forms of PMP20; this is the first report of genes that encode proteins associated with peroxisomal membranes. Southern analysis demonstrates that the two genes are on different loci, although there are several homologous regions of both 5'- and 3'-untranslated sequence. One of the areas of 5' homology is within the untranslated region of the mRNA. Within the coding region there are 35 base differences between the two genes that are reflected in only five amino acid differences. The mRNAs representing both genes of PMP20 are induced in cells grown in methanol-containing medium and are below detection in cells grown in glucose. S1 nuclease protection analysis indicates that there is a 2.5-fold difference in mRNA expression between the two genes when induced. The predicted sequences of both PMP20 genes show the absence of a cleaved amino-terminal leader sequence and the presence of only 1 cysteine residue. In agreement with previous biochemical data suggesting a peripheral association of this protein with the membrane (Goodman, J. M., Maher, J., Silver, P. A., Pacifico, A., and Sanders, D. (1986) J. Biol. Chem. 261, 3464-3468), there are no obvious membrane spanning regions predicted in the sequences. Both PMP20 gene products contain the carboxyl-terminal sequence AKL, similar to the putative SKL peroxisomal sorting sequence (Gould, S. J., Keller, G.-A., and Subramani, S. (1988) J. Cell Biol. 107, 897-905).     

Sequence diversity in S1 genes and S1 translation products of 11 serotype 3 reovirus strains.

The S1 gene nucleotide sequences of 10 type 3 (T3) reovirus strains were determined and compared with the T3 prototype Dearing strain in order to study sequence diversity in strains of a single reovirus serotype and to learn more about structure-function relationships of the two S1 translation products, sigma 1 and sigma 1s. Analysis of phylogenetic trees constructed from variation in the sigma 1-encoding S1 nucleotide sequences indicated that there is no pattern of S1 gene relatedness in these strains based on host species, geographic site, or date of isolation. This suggests that reovirus strains are transmitted rapidly between host species and that T3 strains with markedly different S1 sequences circulate simultaneously. Comparison of the deduced sigma 1 amino acid sequences of the 11 T3 strains was notable for the identification of conserved and variable regions of sequence that correlate with the proposed domain organization of sigma 1 (M.L. Nibert, T.S. Dermody, and B. N. Fields, J. Virol. 64:2976-2989, 1990). Repeat patterns of apolar residues thought to be important for sigma 1 structure were conserved in all strains examined. The deduced sigma 1s amino acid sequences of the strains were more heterogeneous than the sigma 1 sequences; however, a cluster of basic residues near the amino terminus of sigma 1s was conserved. This analysis has allowed us to investigate molecular epidemiology of T3 reovirus strains and to identify conserved and variable sequence motifs in the S1 translation products, sigma 1 or sigma 1s.     

Characterization of the genes for the hexagonally arranged surface layer proteins in protein-producing Bacillus brevis 47: complete nucleotide sequence of the middle wall protein gene.

Bacillus brevis 47 contains two surface (S)-layer proteins, termed the outer wall protein (OWP) and the middle wall protein (MWP), which form a hexagonal array in the cell wall. The MWP and OWP genes are contained in the 9-kilobase-pair (kbp) BclI fragment and constitute an operon under coordinate control of their expression. The nucleotide sequence of a 3.8-kbp EcoRI-SacI fragment containing the entire MWP gene has been determined in this study. Together with the DNA sequence of the promoter region for the MWP-OWP gene operon (H. Yamagata, T. Adachi, A. Tsuboi, M. Takao, T. Sasaki, N. Tsukagoshi, and S. Udaka, J. Bacteriol. 169:1239-1245, 1987) and that of the OWP gene (A. Tsuboi, R. Uchihi, R. Tabata, Y. Takahashi, H. Hashiba, T. Sasaki, H. Yamagata, N. Tsukagoshi, and S. Udaka, J. Bacteriol. 168:365-373, 1986), the complete nucleotide sequence of the MWP-OWP gene operon has been determined. The MWP gene encodes a secretory precursor of the MWP, consisting of a total of 1,053 amino acid residues with a signal peptide of 23 amino acid residues at its amino-terminal end. Bacillus subtilis harboring the MWP gene synthesized an immunoreactive polypeptide with almost the same molecular weight as the authentic MWP, as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The amino acid compositions deduced from the MWP and OWP genes were similar to the chemical amino acid compositions of other S-layer proteins in the predominance of acidic amino acids compared with basic amino acids and in the very low content of sulfur-containing amino acids. The acidic nature of the MWP and OWP was confirmed by isoelectric focusing on polyacrylamide gels. In addition, circular dichroism spectra indicated that the S-layer proteins in B. brevis 47 were composed of approximately 30% beta-sheet and 5% alpha-helical structures, with the remainder of the polypeptide backbone being aperiodic in nature.     

The C terminus of the B5 receptor for herpes simplex virus contains a functional region important for infection

The expression of a previously uncharacterized human hfl-B5 cDNA confers susceptibility for herpes simplex virus (HSV) to porcine cells and fulfills criteria as an HSV entry receptor (A. Perez, Q.-X. Li, P. Perez-Romero, G. DeLassus, S. R. Lopez, S. Sutter, N. McLaren, and A. Oveta Fuller, J. Virol. 79:7419-7430, 2005). Heptad repeats found in the B5 C terminus are predicted to form an alpha-helix for coiled coil structure. We used mutagenesis and synthetic peptides with wild-type and mutant sequences to examine the function of the heptad repeat motif in HSV binding and entry into porcine cells that express B5 and for infection of naturally susceptible human HEp-2 cells. B5 with point mutations predicted to disrupt the putative C-terminal coiled coil failed to mediate HSV binding and entry into porcine cells. Synthetic peptides that contain the single amino acid changes lose the blocking activity of HSV entry. We concluded that the C terminus of B5 contains a functional region that is important for the B5 receptor to mediate events in HSV entry. Structural evidence that this functional region forms coiled coil structures is under investigation. Blocking of HSV interaction with the C-terminal region of the B5 receptor is a new potential target site to intervene in the virus infection of human cells.     

A new alpha-helical coiled coil protein encoded by the Salmonella typhimurium virulence plasmid.

A new protein of Salmonella typhimurium was identified and characterized. The gene (tlpA) encoding this protein (TlpA) was isolated from the large virulence-associated plasmid of S. typhimurium and sequenced in order to predict the primary structure of TlpA. tlpA encodes a 371-amino acid soluble protein with a calculated M(r) of 41600 and pI of 4.63. Secondary structure predictions and sequence statistics of TlpA indicated a predominant alpha-helical configuration and presence of heptapeptide repeat motifs characteristic of coiled coil proteins. Purified TlpA was shown to have biochemical properties similar to those of coiled coil proteins, including adoption of an alpha-helical configuration and a tendency to form homodimers. Furthermore, TlpA possessed heat resistance, evidence for a chain register and altered mobility in urea/sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels which are characteristics of tropomyosins. TlpA shows 32% overall sequence similarity with rat cardiac myosin and 36% similarity with horse platelet beta-tropomyosin over 226 residues, whereas selected regions possessed significant sequence identities with myosins, tropomyosins, and alpha-helical surface proteins of Streptococcus pyogenes. Our results indicate that TlpA represents a new member of prokaryotic coiled coil proteins.     

Cooperative involvement of the S1 and S2 subunits of the murine coronavirus spike protein in receptor binding and extended host range

To study the process of spike (S)-receptor interaction during coronavirus entry, we evaluated the contributions of mutations in different regions of the murine hepatitis virus (MHV) S protein to natural receptor murine carcinoembryonic antigen-related cell adhesion molecule 1a (CEACAM1a) dependence and to the acquisition of extended host range. Extended-host-range variants of MHV strain A59 were previously obtained from persistently infected cells (J. H. Schickli, B. D. Zelus, D. E. Wentworth, S. G. Sawicki, and K. V. Holmes, J. Virol. 71:9499-9504, 1997). These variant viruses contain several mutations in the S protein that confer to the viruses the ability to enter cells in a heparan sulfate-dependent manner (C. A. de Haan, Z. Li, E. te Lintelo, B. J. Bosch, B. J. Haijema, and P. J. M. Rottier, J. Virol. 79:14451-14456, 2005). While the parental MHV-A59 is fully dependent on murine CEACAM1a for its entry, viruses carrying the variant mutations in the amino-terminal part of their S protein had become dependent on both CEACAM1a and heparan sulfate. Substitutions in a restricted, downstream part of the S protein encompassing heptad repeat region 1 (HR1) and putative fusion peptide (FP) did not alter the CEACAM1a dependence. However, when the mutations in both parts of the S protein were combined, the resulting viruses became independent of CEACAM1a and acquired the extended host range. In addition, these viruses showed a decreased binding to and inhibition by soluble CEACAM1a. The observations suggest that the amino-terminal region of the S protein, including the receptor-binding domain, and a region in the central part of the S protein containing HR1 and FP, i.e., regions far apart in the linear sequence, communicate and may even interact physically in the higher-order structure of the spike.     

Identification of adenovirus type 2 early region 1B proteins that share the same amino terminus as do the 495R and 155R proteins.

Adenovirus type 2 early region 1B (E1B) proteins synthesized in vitro were fractionated chromatographically and characterized by peptide and sequence analysis and by reaction with peptide-specific antisera targeted to either the N or C terminus of either of two overlapping E1B reading frames (175 or 495 codons). In addition to the previously identified E1B-495R, E1B-175R, and E1B-155R species, two other E1B proteins of similar electrophoretic mobility to the 175R protein were identified. E1B-82R is an abundant product in vitro and in vivo that has the same N terminus as that of the 495R and 155R proteins but a different C terminus. The structure of 82R is predicted by the structure of the abundant 13S (1.02-kilobase) E1B mRNA. E1B-168R is a novel minor species consisting of the 24 amino-terminal residues of the 495R protein fused to the entire polypeptide IX sequence. An additional, minor 16,000-molecular-weight polypeptide was detected that may correspond to a predicted 92R E1B protein, but definitive identification was not possible. These observations establish that the leftmost portion (78 codons) of the 495-codon reading frame, which overlaps the right half of the 175-codon reading frame, is expressed as an abundant protein that does not contain other 495R sequences. This region, which may participate in the regulation of region E1A expression, may thus constitute a functional domain distinct from the rightward portion of the 495R protein.     

Molecular structure of the cell-attachment protein of reovirus: correlation of computer-processed electron micrographs with sequence-based predictions.

The receptor-recognition interaction that initiates reovirus infection is mediated by the sigma 1 protein, located at the vertices of the icosahedral virion. We have applied computer-based image-averaging techniques to electron micrographs of negatively stained preparations of sigma 1 purified from virions (serotype 2 Jones). Combining these results with inferences based on the amino acid sequence has led to a molecular model in which the overall folding of the chains is described; its conformation embodies motifs, coiled-coil alpha-helices and nodular multichain elements rich in beta-sheets, previously detected in the corresponding proteins of other viruses, but with some novel variations. Sigma 1 is a filamentous lollipop-shaped molecule with an overall length of approximately 48 nm; it has a flexible "tail," approximately 40 nm long by 4 to 6 nm wide, terminating at its distal end in a globular "head," approximately 9.5 nm in diameter. The purified protein is a tetramer (4 by 50 kilodaltons) consisting of two similarly oriented dimers bonded side by side and in register. For each chain, a cluster of hydrophobic residues at its amino terminus resides at the proximal end of the tail; next, an alpha-helical domain (residues 25 to 172) participates in a two-chained coiled coil, 22 nm long, with two such coiled coils pairing laterally to form the proximal half of the tail. The remainder of the tail (residues 173 to approximately 316) is less uniform in width and is expected to be rich in beta-sheet; the interdimer bonding is evidently sustained through this portion of the molecule. Finally, the globular head consists of the carboxy-terminal domains (which contain the receptor-binding sites) folded into compact globular conformations; in appropriate side views, the head is resolved into two subunits, presumably contributed by the respective dimers. This model for how the four sigma 1 polypeptide chains are threaded in parallel through the fiber is supported by the observed match between an empirical curvature profile, which identifies the locations of relatively flexible sites along the tail, and the flexibility profile predicted on the basis of the model. Appraisal of the interactions that stabilize the coiled coils suggests that (i) the alpha-helices are individually only marginally stable, a property that may be of significance with regard to the retracted conformation in which sigma 1 is accommodated in the intact virion, and (ii) the predominant interactions between the two coiled coils are likely to involve hydrogen bonding between patches of uncharged residues.     

Sequence comparisons of intermediate filament chains: evidence of a unique functional/structural role for coiled-coil segment 1A and linker L1

A comprehensive analysis of the sequences of all types of intermediate filament chains has been undertaken with a particular emphasis on those of segment 1A and linker L1. This has been done to assess whether structural characteristics can be recognized in the sequences that would be consistent with the role of each region in the recently proposed "swinging head" hypothesis. The analyses show that linker L1 is the most flexible rod domain region, that it is the most elongated structure (on a per residue basis), and that it is the most variable region as regards sequence and length. Segment 1A has one of the two most highly conserved regions of sequence in the rod domain (the other being at the end of segment 2B), with seven particular residues conserved across all chain types. It also contains one of the very few potential interchain ionic interactions that could be conserved across all chain types. However, the aggregation of chains in segment 1A is specified less precisely overall by interchain ionic interactions than are the other coiled-coil segments. The apolar residue contents in positions a and d of the heptad substructure are the highest of any coiled-coil segment in the intermediate filament family. Segment 1A also displays an amino acid composition atypical of not only coiled-coil segments 1B and 2B, but indeed of two-stranded coiled coils in general. Nonetheless, molecular modeling based on the crystal structure of the monomeric 1A fragment from human vimentin shows that coiled-coil formation is plausible. The most extensive regions of apolar/aromatic residues lie at the C-terminal end of segment 2B in the helix termination motif and in segment 1A in and close to the helix initiation motif. The predicted stability of the individual alpha-helices in segment 1A is greater than in those comprising segments 1B and 2B, though potential intrachain ionic interactions are either lacking or are minimal in number. Analysis of the 1A sequence and those regions immediately N- and C-terminal to it has shown that the capping residues are near optimal close to the previously predicted ends, thus adding to the likely stability of the alpha-helical structure. However, a second terminating sequence is predicted in 1A (about 10 residues back from the C-terminus). This allows the possibility of some unwinding of the alpha-helical structure of 1A immediately adjacent to linker L1 when the head domains no longer stabilize the coiled-coil structure. All of these data are consistent with the concept of a flexible hinge at L1 and with the ability of the two alpha-helical coiled-coil strands to separate under appropriate conditions and partly unwind at their C-terminal ends to allow the head domains a greater degree of mobility, thus facilitating function.     

Mutations in type 3 reovirus that determine binding to sialic acid are contained in the fibrous tail domain of viral attachment protein sigma1.

The reovirus attachment protein, sigma1, determines numerous aspects of reovirus-induced disease, including viral virulence, pathways of spread, and tropism for certain types of cells in the central nervous system. The sigma1 protein projects from the virion surface and consists of two distinct morphologic domains, a virion-distal globular domain known as the head and an elongated fibrous domain, termed the tail, which is anchored into the virion capsid. To better understand structure-function relationships of sigma1 protein, we conducted experiments to identify sequences in sigma1 important for viral binding to sialic acid, a component of the receptor for type 3 reovirus. Three serotype 3 reovirus strains incapable of binding sialylated receptors were adapted to growth in murine erythroleukemia (MEL) cells, in which sialic acid is essential for reovirus infectivity. MEL-adapted (MA) mutant viruses isolated by serial passage in MEL cells acquired the capacity to bind sialic acid-containing receptors and demonstrated a dependence on sialic acid for infection of MEL cells. Analysis of reassortant viruses isolated from crosses of an MA mutant virus and a reovirus strain that does not bind sialic acid indicated that the sigma1 protein is solely responsible for efficient growth of MA mutant viruses in MEL cells. The deduced sigma1 amino acid sequences of the MA mutant viruses revealed that each strain contains a substitution within a short region of sequence in the sigma1 tail predicted to form beta-sheet. These studies identify specific sequences that determine the capacity of reovirus to bind sialylated receptors and suggest a location for a sialic acid-binding domain. Furthermore, the results support a model in which type 3 sigma1 protein contains discrete receptor binding domains, one in the head and another in the tail that binds sialic acid.     

Molecular cloning and characterization of the Endo B cytokeratin expressed in preimplantation mouse embryos

A cDNA clone of a keratin-related, intermediate filament protein, designated Endo B, was constructed from size-fractionated parietal endodermal mRNA and characterized. The 1466-nucleotide cDNA insert contains an open reading frame of 1272 nucleotides that would result in 5' and 3' noncoding sequences of 54 and 60 nucleotides, respectively. The predicted amino acid composition, molecular weight (47,400), and peptide pattern correlate well with data obtained on the isolated protein. The predicted amino acid sequence fits easily into the general domain structure suggested for all intermediate filament proteins with a unique amino-terminal head domain, a large conserved central domain of predominantly alpha-helical structure, and a relatively unique carboxyl-terminal or tail domain. Over the entire molecule, Endo B is 43% identical with human 52-kDa epidermal type I keratin. However, over two of the three regions contained in the central domain that are predicted to form coiled-coil structures, the Endo B is 54-68% identical with other type I keratin sequences. This homology, along with the presence of the completely conserved sequence DNARLAADDFR-KYE, which is found in all type I keratins, permits the unambiguous identification of Endo B as a type I keratin. Comparison of the Endo B sequence to other intermediate filament proteins reveals 22 residues which are identical in all intermediate filament proteins regardless of whether filament formation requires only one type of protein subunit (vimentin, desmin, glial fibrillar acidic protein, or a neurofilament protein) or two dissimilar types (type I and type II keratins). Endo B mRNA was detectable in RNA isolated from F9 cells treated with retinoic acid for 48 h. Approximately three to five genes homologous to Endo B were detected in the mouse genome.