Global elemental maps of the moon: the Lunar Prospector gamma-Ray spectrometer.

Lunar Prospector gamma-ray spectrometer spectra along with counting rate maps of thorium, potassium, and iron delineate large compositional variations over the lunar surface. Thorium and potassium are highly concentrated in and around the nearside western maria and less so in the South Pole-Aitken basin. Counting rate maps of iron gamma-rays show a surface iron distribution that is in general agreement with other measurements from Clementine and the Lunar Prospector neutron detectors.

Cloning and sequence of an infectious bovine rhinotracheitis virus (BHV-1) gene homologous to glycoprotein H of herpes simplex virus.

A homologue to the glycoprotein H (gH) gene of herpes simplex virus (HSV) has been identified in the genome of infectious bovine rhinotracheitis virus (IBR, BHV-1). The gene is located immediately downstream from the thymidine kinase gene, and codes for an open reading frame (orf) of 842 amino acids. The orf has the characteristics of a membrane glycoprotein, including an N-terminal hydrophobic region resembling a signal sequence, a C-terminal region which is probably a transmembrane domain, and six potential sites for N-linked glycosylation. This orf shows significant homology to the gH sequences of both HSV and pseudorabies virus (PRV). We conclude that this gene encodes BHV-1 gH.

Assembly of the herpes simplex virus capsid: characterization of intermediates observed during cell-free capsid formation.

The herpes simplex virus-1 (HSV-1) capsid is an icosahedral shell approximately 15 nm thick and 125 nm in diameter. Three of its primary structural components are a major capsid protein (VP5; coded by the UL19 gene) and two minor proteins, VP19C (UL38 gene) and VP23 (UL18 gene). Assembly of the capsid involves the participation of two additional proteins, the scaffolding protein (UL26.5 gene) and the maturational protease (UL26 gene). With the goal of identifying morphological intermediates in the assembly process, we have examined capsid formation in a cell-free system containing the five HSV-1 proteins mentioned above. Capsids and capsid-related structures formed during progressively longer periods of incubation were examined by electron microscopy of thin-sectioned specimens. After one minute, 90 minutes and eight hours of incubation the structures observed, respectively, were partial capsids, closed spherical capsids and polyhedral capsids. Partial capsids were two-layered structures consisting of a segment of external shell partially surrounding a region of scaffold. They appeared as wedges or angular segments of closed spherical capsids, the angle ranging from less than 30 degrees to greater than 270 degrees. Partial capsids are suggested to be precursors of closed spherical capsids because, whereas partial capsids were the predominant assembly product observed after one minute of incubation, they were rare in reactions incubated for 45 minutes or longer. Closed spherical capsids were highly uniform in morphology, consisting of a closed external shell surrounding a thick scaffold similar in morphology to the same layers seen in partial capsids. In negatively stained specimens, closed spherical capsids appeared round in profile, suggesting that they are spherical rather than polyhedral in shape. A three-dimensional reconstruction computed from cryoelectron micrographs confirmed that closed spherical capsids are spherical with T = 16 icosahedral symmetry. The reconstruction showed further that, compared to mature HSV-1 capsids, closed spherical capsids are more open structures in which the capsid floor layer is less pronounced. In contrast to closed spherical capsids, polyhedral capsids exhibited distinct facets and vertices, indicating that they are icosahedral like the capsids in mature virions. Upon incubation in vitro, purified closed spherical capsids matured into polyhedral capsids, indicating that the latter arise by angularization of the former. Partial capsids, closed spherical capsids and polyhedral capsids were all found to contain VP5, VP19C, VP23, VP21 and the scaffolding protein; the scaffolding protein being predominantly in the immature, uncleaved form in all cases. Polyhedral capsids and closed spherical capsids were found to differ in their sensitivity to disruption at 2 degrees C. Closed spherical capsids were disassembled while polyhedral capsids were unaffected. Our results suggest that HSV-1 capsid assembly begins with the partial capsid and proceeds through a closed, spherical, unstable capsid intermediate to a closed, icosahedral form similar to that found in the mature virion. Structures resembling HSV-1 partial capsids have been described as capsid assembly intermediates in Salmonella typhimurium bacteriophage P22. HSV-1 capsid maturation from a fragile, spherical state to a robust polyhedral form resembles the prohead maturation events undergone by dsDNA bacteriophages including lambda, T4 and P22. Because of this similarity, we propose the name procapsid for the closed spherical capsid intermediate in HSV-1 capsid assembly.

The herpes simplex virus procapsid: structure, conformational changes upon maturation, and roles of the triplex proteins VP19c and VP23 in assembly.

The proteins coded by the five major capsid genes of herpes simplex virus 1, VP5 (gene UL19), VP19c (UL38), VP23 (UL18), pre-VP22a (UL26.5), and pre-VP21 (UL26), assemble into fragile roundish "procapsids", which mature into robust polyhedral capsids in a transition similar to that undergone by bacteriophage proheads. Here we describe the HSV-1 procapsid structure to a resolution of approximately 2.7 nm from three-dimensional reconstructions of cryo-electron micrographs. Comparison with the mature capsid provides insight into the large-scale conformational changes that take place upon maturation. In the procapsid, the elongated protomers (VP5 subunits) make little contact with each other except around the bases of the hexons and pentons, whereas they are tightly clustered into capsomers in the mature state; the axial channels, which are constricted or blocked in the mature capsid, are fully open; and unlike the well observed 6-fold symmetry of mature hexons, procapsid hexons are distorted into oval and triangular shapes. These deformations reveal a VP5 domain in the inner part of the protrusion wall which participates in inter-protomer bonding in the procapsid and is close to the site where the channel closes upon maturation. Remarkably, there are no direct contacts between neighboring capsomers; instead, interactions between them are mediated by the "triplexes" at the sites of local 3-fold symmetry. This observation discloses the mechanism whereby the triplex proteins, VP19c and VP23, play their essential roles in capsid morphogenesis. In the mature capsid, density extends continuously between neighboring capsomers in the inner "floor" layer. In contrast, there are large gaps in the corresponding region of the procapsid, implying that formation of the floor involves extensive remodeling. Inside the procapsid shell is the hollow spherical scaffold, whose radial density profile indicates that the major scaffold protein, pre-VP22a, is a long molecule (> 24 nm) composed of three domains. Since no evidence of icosahedral symmetry is detected in the scaffold, we infer that (unless higher resolution is required) the scaffold may not be an icosahedral shell but may instead be a protein micelle with a preferred radius of curvature.

Cloning of the human cytomegalovirus genome as endonuclease XbaI fragments.

Restriction enzyme XbaI DNA fragments that represent 99% of the sequences from the long and short unique as well as the repeat sequences of the human cytomegalovirus (CMV) genome have been cloned into bacterial plasmid pACYC184. The viral DNA sequences associated with the recombinant plasmids were analyzed by restriction mapping and by hybridization to fragments of authentic viral DNA. The relationship of the cloned viral DNA fragments to the XbaI physical map of the viral genome is demonstrated. Even though large recombinant plasmids ranging from approx. 39 to 1.8 kb were isolated, most if not all of the viral DNA fragments were stable during propagation in Escherichia coli HB101.

Pseudorabies virus as a live virus vector for expression of foreign genes.

The cDNA coding for human tissue plasminogen activator (tPA) was cloned downstream from the promoter for pseudorabies virus (PRV) glycoprotein and flanked by downstream PRV DNA. After co-transfection with PRV DNA, this plasmid recombined to insert the tPA cDNA into the viral genome. In cells infected by this recombinant virus, tPA was detected by immunoprecipitation analysis and by enzymatic activity. Since it has a wide host range but does not infect humans, PRV is a possible vaccine vector for genes from animal pathogens.

Effects of GABA and various allosteric ligands on TBPS binding to cloned rat GABA(A) receptor subtypes.

1. [35S]t-butylbicyclophosphorothionate (TBPS) is a high affinity ligand for the picrotoxin site of GABA(A) receptors. Here we examined TBPS binding to the cloned receptors made of alpha 1, alpha 3 or alpha 6 in combination with beta 2 or beta 2 and gamma 2 subunits, in the presence of GABA and several allosteric ligands (diazepam, methyl 6,7-dimethoxy-4-methyl-beta-carboline-3-carboxylate (DMCM), 3 alpha,21-dihydroxy-5 alpha-pregnan-20-one (5 alpha-THDOC), pentobarbitone and Zn). The cloned receptors were transiently expressed in SF-9 insect cells by infecting with recombinant baculoviruses. 2. In alpha beta subtypes, GABA at nanomolar concentrations enhanced TBPS binding but inhibited binding at micromolar concentrations. Half maximal GABA concentrations for enhancement or inhibition of TBPS binding were correlated with high and low affinity GABA binding sites, respectively, in individual subtypes. The maximal enhancement of binding also varied according to the alpha isoform (alpha 3 beta 2 >> alpha 1 beta 2). In alpha beta gamma subtypes, TBPS binding was unaffected by GABA at nanomolar concentrations, but was inhibited by GABA at micromolar concentrations. Addition of gamma 2 thus appeared to abolish conformational coupling between high affinity GABA sites and TBPS sites, and also altered low affinity GABA sites; in particular, the half maximal GABA concentration for inhibition of TBPS binding changed from > 100 (alpha 6 beta 2) to 1 microM (alpha 6 beta 2 gamma 2). 3. Allosteric ligands also altered TBPS binding to sensitive GABA(A) receptor subtypes. For instance,diazepam only in the alpha 1 beta2 gamma 2 and alpha 3 beta 2 gamma 2 subtypes, and 5 alpha-THDOC in all the subtypes enhanced TBPS binding in the absence of GABA, and intensified the inhibitory effect of GABA. Pentobarbitone exhibited only the latter effect in all the subtypes we examined.4. DMCM and Zn, inhibitors of GABA-induced Cl currents in alpha beta gamma and alpha beta subtypes, respectively,produced opposite effects to agonists, decreasing TBPS binding in the absence of GABA and attenuating(or eliminating in the case of Zn) the inhibitory effect of GABA on TBPS binding.5. These results show that GABA binding sites and their conformational coupling with TBPS sites are differentially affected by the alpha isoform (particularly alpha 6 as compared to alpha l or alpha 3) and by quaternary interactions involving the gamma 2 subunit. Moreover, changes in TBPS binding by allosteric ligands include not only direct (allosteric) effects on TBPS sites but also indirect effects via GABA sites, and are consistent with their known subtype selectivity and functionality from previous studies.

cDNA cloning, expression, and chromosomal localization of a human UDP-GalNAc:polypeptide, N-acetylgalactosaminyltransferase.

Oligonucleotide primers derived from the cDNA encoding a full-length bovine UDP-GalNAc:polypeptide, N-acetylgalactosaminyltransferase (GalNAc-transferase) [Homa, F. L., Hollander, T., Lehman, D. J., Thomsen, D. R., and Elhammer, A. P. (1993) J. Biol. Chem. 268, 12609-12616], were used for PCR to isolate sequences encoding a homologous enzyme from human salivary gland cDNA. Comparison of the human and bovine nucleotide sequences reveals 94.8% sequence identity in their coding regions and 87% identity in their 3-untranslated regions. The translation of the human GalNAc-transferase coding region predicts an amino acid sequence which is nearly identical (99.6%) to that of the bovine counterpart; there are five conservative and one non-conservative amino acid substitutions between the two enzymes. Expression of the bovine and human cDNAs in the insect cell line, Sf9, resulted in the synthesis of proteins which appeared identical on SDS-PAGE and which had similar enzymatic properties. Screening of a somatic cell human/rodent hybrid panel with a probe derived from the human GaLNAc-transferase cDNA sequence indicated that the human GalNAc-transferase gene is localized to chromosome 18.

Expression of three recombinant proteins using baculovirus vectors in 23 insect cell lines.

Recombinant Autographa california baculoviruses expressing genes for pseudorabies virus glycoprotein (gp50T), human plasminogen (HPg), and beta-galactosidase (beta-gal) were used to infect 23 cell lines or strains. The objectives were to compare amounts of recombinant proteins expressed in the cell lines, compare yields from clones and parent lines, investigate the effects of long-term culture in serum-free medium on production, and determine if some lines yield gp50T with different glycosylation patterns. For HPg, IZD-MB0503 had the highest yield and four other lines (IPLB-TN-R2, IPLB-SF-1254, IPLB-LdEIta, and CM-1) had levels above that of SF-9 cells. For gp50T, four lines (IPLB-HvT1, IPLB-SF21AE, IPLB-SF21AE-15, and IPLB-SF-1254) had higher amounts than SF-9 cells. Some lines yielded gp50T with molecular mass about 1000 daltons larger than that from SF-9 cells, which suggests increased oligosaccharide processing. Equally high levels of beta-gal were expressed in three lines (SF-9, IZD-MB0503, and BCIRL-PX2-HNV3). The major conclusion is that no single cell line produced highest yields for all three recombinant proteins. Four lines were cultured in serum-free medium for 31-34 passages and then infected with the three recombinant viruses. For most cell line-recombinant combinations, the yields in serum-free medium were equal to or better than those in serum-supplemented medium. Medium composition had a much stronger effect on foreign gene expression than on susceptibility of cells to wild-type virus.

Immune response in pigs to Aujeszky's disease viruses defective in glycoprotein g1 or gX.

Two Aujeszky's disease virus glycoprotein genes, gX and g1, have been used to produce deletion mutants which have then been developed into vaccines. These deletions then allow differentiation between pigs infected with wild type virus and those given the vaccine. It is not clear whether the glycoproteins encoded for by these genes are needed to induce a full protective immune response, in which case deletion mutants would suffer from lack of potency. To test this, commercially available Aujeszky's virus vaccines which lacked either gX or g1 were compared and isogenic constructs were made which differed only in the absence or presence of gX and, or, g1. These constructs and vaccines were used to vaccinate the natural host of Aujeszky's disease, the pig, and potency was measured using challenge with wild type virus. In all cases vaccines which lacked g1 performed significantly less well than those in which g1 was present, whereas deletions of gX had no significant effect on vaccine performance.

A vaccine strain of pseudorabies virus with deletions in the thymidine kinase and glycoprotein X genes.

A pseudorabies virus (PRV) mutant with deletions in genes for glycoprotein X (gX) and thymidine kinase, designated delta GX delta TK, was constructed and evaluated as a vaccine for protecting swine against PRV-induced mortality. Doses greater than or equal to 10(3) plaque-forming units (PFU) of this strain given to mice provided protection from challenge exposure with virulent PRV. Sera tested from mice inoculated with delta GX delta TK had high titers of neutralizing antibody to PRV, but reactivity in the same sera was not significantly different from that in sera from noninoculated mice (controls) when sera from both groups were evaluated by use of an ELISA with gX antigen produced in Escherichia coli. Compared with noninoculated pigs (controls), those given delta GX delta TK (greater than or equal to 10(2) PFU) were protected completely from lethal challenge exposure, without experiencing adverse effects on weight gain and with reduction of shedding of virulent challenge virus. Serotest results indicated that, although inoculated pigs responded with strong neutralizing antibody titers, the response of delta GX delta TK-inoculated pigs to gX, as determined by ELISA before challenge exposure, was not significantly greater than the ELISA values obtained from control pigs. The ELISA values from a group of pigs inoculated with a commercially available vaccine were significantly (P less than 0.05) higher than those of control pigs. The experimental vaccine, delta GX delta TK, was avirulent for mice, swine, and sheep, but was mildly virulent for calves (mortality, 1 of 12) and more virulent for dogs (mortality, 3 of 6) and cats (mortality, 2 of 6).(ABSTRACT TRUNCATED AT 250 WORDS)

Functional expression of GABAA chloride channels and benzodiazepine binding sites in baculovirus infected insect cells.

We have employed the baculovirus expression system for the production of insect cell membranes having GABA/benzodiazepine binding sites. Three recombinant baculoviruses each harboring a different GABAA receptor cDNA were introduced into insect cells by simultaneous infection. Infected cells expressed GABA responsive Cl- channels and benzodiazepine binding sites with the same pharmacological specificity as animal cells expressing these receptor subunits.

Expression of feline leukaemia virus gp85 and gag proteins and assembly into virus-like particles using the baculovirus expression vector system.

In order to test components of feline leukaemia virus (FeLV) as subunit vaccines, we have constructed recombinant baculoviruses that express the FeLV envelope glycoprotein gp85 [Autographa californica nuclear polyhedrosis virus (AcNPV)-gp85] and the structural protein, gag (AcNPVgag). The gag protein is expressed and shed into the medium of infected cells as particles which have a buoyant density on sucrose gradients and appearance by electron microscopy similar to those of authentic FeLV virions. The gag precursor protein within the particles is not fully processed and appears to be a result of partial cleavage of the gag polypeptide. Insect cells that are coinfected with AcNPVgag and AcNPVgp85 shed particles that contain both the gag protein and the gp85 glycoprotein.

DNA of human cytomegalovirus: size heterogeneity and defectiveness resulting from serial undiluted passage.

The majority of DNA molecules associated with plaque-purified, low-multiplicity-passaged human cytomegalovirus (Towne strain) had a molecular weight of approximately 150 x 10(6) and a molecular complexity of approximately 140 x 10(6). Serial high-multiplicity passage resulted in the production of defective cytomegalovirions. An accumulation of smaller DNA molecules packaged into virions was directly correlated with a decrease in infectivity and an increase in the particle-to-PFU ratio. The majority of defective DNA molecules had a molecular weight of approximately 100 x 10(6). In addition, there were some viral DNA molecules of approximately 60 x 10(6). Restriction enzyme analysis of defective cytomegalovirus DNA detected unique DNA fragments, suggesting a specific alteration in the nucleotide sequence. Some reasons for the generation of defective cytomegalovirus DNA are discussed.

Structural analysis of the major immediate early gene of human cytomegalovirus.

The most abundant species of human cytomegalovirus (Towne) immediate early polysome-associated RNA originates from a region of ca. 2.8 kilobases (0.739 to 0.755 map units) within the XbaI-E DNA fragment. These sequences code for a 1.95-kilobase mRNA and are referred to as immediate early coding region one (M. F. Stinski, D. R. Thomsen, R. M. Stenberg, and L. C. Goldstein, J. Virol. 46:1-14, 1983). We have utilized the nuclease mapping technique of Berk and Sharp (A. J. Berk and P. A. Sharp, Cell 12:721-732, 1977) to examine this gene in detail. Cloned fragments of human cytomegalovirus DNA, either labeled with 32P in vivo or end labeled in vitro at the 5' or 3' termini, were hybridized to immediate early polysome-associated RNA. The hybrids were treated with single-strand-specific nuclease and subjected to electrophoresis in either neutral or denaturing gels. The major transcript was shown to be a spliced molecule containing a 3' terminal exon of 1,341 nucleotides. Upstream of the major body of the mRNA are three small exon sequences of 185, 88, and 121 nucleotides. The sequence of the exons as well as the locations of the intron-exon splice junctions were determined. Based on the DNA sequence, the viral mRNA molecule has one open reading frame which begins within the second exon and extends for 491 amino acid residues. The predicted molecular weight of the polypeptide originating from this region was estimated to be 64,000. It is hypothesized that this viral gene codes for the major regulatory protein controlling transcription of the viral genome at early times. The properties of the viral gene and its protein product are discussed.

Synthesis of human cytomegalovirus-specified RNA and protein in interferon-treated cells at early times after infection.

In human fibroblast cells treated with interferon, cytomegalovirus-specified immediate early RNA was found associated with the polyribosomes at concentrations and size classes similar to the virus RNA found in non-treated cells, Interferon treatment inhibited the translation of the immediate early virus mRNA; the relative rate of virus-specified immediate early protein and antigen synthesis decreased with increasing concentrations of interferon. In addition, the relative amount of virus-specified RNA associated with the polyribosomes at early times after infection was significantly reduced by treatment of the cells with interferon. Inhibition of infectious virus production in interferon-treated cells was primarily due to inhibition of immediate early virus protein synthesis and secondarily to suppression of early virus RNA synthesis. The role of the virus-specified immediate early proteins in regulating subsequent virus gene expression is discussed.

Temporal regulation of human cytomegalovirus transcription at immediate early and early times after infection.

The immediate early transcripts of human cytomegalovirus originated from restricted regions of the viral genome. In contrast, transcription at early times was complementary to all regions of the viral genome that were fractionated by restriction endonuclease treatment followed by agarose gel electrophoresis. The viral genome was also extensively transcribed when 2 h of protein synthesis or longer was permitted after infection in permissive cells treated with an inhibitor of viral DNA replication or in nonpermissive cells of animal origin that permit little or no viral DNA replication. The size and in vitro translation products of the cytomegalovirus-specified mRNA's at immediate early and early times after infection were determined. Discrete size classes of virus-specified polyadenylated RNA accumulated on the polyribosomes of cells infected in the presence of an inhibitor of protein synthesis. When 2 or 24 h of protein synthesis occurred after infection, there were changes in the relative abundance of the virus-specified RNAs that accumulated on polyribosomes. Treatment of nonpermissive cells had little effect on the size classes of viral RNA found associated with the polyribosomes at early times after infection. These viral mRNA's were assumed to represent early viral gene expression. In vitro translation of the viral mRNA isolated from polyribosomes at immediate early and early times after infection identified many of the virus-specified gene products and demonstrated (i) a switch from immediate early to early viral gene expression and (ii) a prolonged phase of early viral gene expression. The data also indicated that the initiation of viral RNA synthesis does not depend on the formation of viral protein, but that de novo viral protein synthesis may influence the extent of transcription of the viral genome.

Structure of O-glycosidically linked oligosaccharides synthesized by the insect cell line Sf9.

The O-glycosidically linked oligosaccharides on the pseudorabies virus (PRV) glycoprotein gp50 synthesized by three different cell lines were studied. The intact membrane protein (gp50) was expressed in Vero cells and in the insect cell line Sf9. In addition, a truncated, secreted form lacking the transmembrane and cytoplasmic domains (gp50T), was expressed in CHO and Sf9 cells. The protein, both in intact and truncated form, synthesized by the two mammalian cells contained only the disaccharide Gal beta 1-3GalNAc, either unsubstituted or substituted with one or two sialic acid residues. By contrast, the major O-linked structure on gp50 and gp50T synthesized by Sf9 cells was the monosaccharide GalNAc. The Sf9 cells also linked lower amounts of Gal beta 1-3GalNAc to gp50 (12%) and gp50T (26%). None of the structures synthesized by Sf9 cells contained sialic acid. Measurements of the two relevant glycosyltransferases revealed that while all three cell lines contain comparable levels of UDP-GalNAc:polypeptide, N-acetylgalactosaminyltransferase activity, there is a greater variation in the levels of UDP-Gal:N-acetylgalactosamine, beta 1-3 galactosyltransferase, with the Sf9 cells containing the lowest level.

Immunization of cotton rats with the human respiratory syncytial virus F glycoprotein produced using a baculovirus vector.

The F glycoprotein of respiratory syncytial virus in insect cells was produced using a baculovirus expression vector to examine its potential as a subunit vaccine. Two different forms of the F glycoprotein were expressed: the intact F (F) glycoprotein and the truncated (Ft) glycoprotein, in which the COOH-terminal anchor region was deleted. The F glycoprotein remained cell associated, whereas the Ft glycoprotein was secreted into the media of infected cells. In contrast to the processing of the F0 precursor into its F1 and F2 subunits that was observed in mammalian cells, a second cleavage site within the F1 subunit was recognized by the insect cell proteases and resulted in the formation of two F1 subunits. The baculovirus-expressed Ft glycoprotein induced neutralizing antibodies in cotton rats and protected vaccinated animals from challenge with respiratory syncytial virus.

Assembly of herpes simplex virus (HSV) intermediate capsids in insect cells infected with recombinant baculoviruses expressing HSV capsid proteins.

The capsid of herpes simplex virus type 1 (HSV-1) is composed of seven proteins, VP5, VP19C, VP21, VP22a, VP23, VP24, and VP26, which are the products of six HSV-1 genes. Recombinant baculoviruses were used to express the six capsid genes (UL18, UL19, UL26, UL26.5, UL35, and UL38) in insect cells. All constructs expressed the appropriate-size HSV proteins, and insect cells infected with a mixture of the six recombinant baculoviruses contained large numbers of HSV-like capsids. Capsids were purified by sucrose gradient centrifugation, and electron microscopy showed that the capsids made in Sf9 cells had the same size and appearance as authentic HSV B capsids. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis demonstrated that the protein composition of these capsids was nearly identical to that of B capsids isolated from HSV-infected Vero cells. Electron microscopy of thin sections clearly demonstrated that the capsids made in insect cells contained the inner electron-translucent core associated with HSV B capsids. In infections in which single capsid genes were left out, it was found that the UL18 (VP23), UL19 (VP5), UL38 (VP19C), and either the UL26 (VP21 and VP24) or the UL26.5 (VP22a) genes were required for assembly of 100-nm capsids. VP22a was shown to form the inner core of the B capsid, since in infections in which the UL26.5 gene was omitted the 100-nm capsids that formed lacked the inner core. The UL35 (VP26) gene was not required for assembly of 100-nm capsids, although assembly of B capsids was more efficient when it was present. These and other observations indicate that (i) the products of the UL18, UL19, UL35, and UL38 genes self-assemble into structures that form the outer surface (icosahedral shell) of the capsid, (ii) the products of the UL26 and/or UL26.5 genes are required (as scaffolds) for assembly of 100-nm capsids, and (iii) the interaction of the outer surface of the capsid with the scaffolding proteins requires the product of the UL18 gene (VP23).