Search

Based on its homology to the estrogen receptor and its roles in osteoblast and chondrocyte differentiation, the orphan nuclear receptor estrogen-related receptor α (ERRα (ESRRA)) is an intriguing therapeutic target for osteoporosis and other bone diseases. The objective of this study was to better characterize the molecular mechanisms by which ERRα modulates osteoblastogenesis. Experiments from multiple systems demonstrated that ERRα modulates Wnt signaling, a crucial pathway for proper regulation of bone development. This was validated using a Wnt-luciferase reporter, where ERRα showed co-activator-dependent (peroxisome proliferator-activated receptor gamma co-activator 1α, PGC-1α) stimulatory effects. Interestingly, knockdown ofERRαexpression also enhanced WNT signaling. In combination, these data indicated that ERRα could serve to either activate or repress Wnt signaling depending on the presence or absence of its co-activator PGC-1α. The observed Wnt pathway modulation was cell intrinsic and did not alter β-catenin nuclear translocation but was dependent on DNA binding of ERRα. We also found that expression of active ERRα correlated with Wnt pathway effects on osteoblastic differentiation in two cell types, consistent with a role for ERRα in modulating the Wnt pathway. In conclusion, this work identifies ERRα, in conjunction with co-activators such as PGC-1α, as a new regulator of the Wnt-signaling pathway during osteoblast differentiation, through a cell-intrinsic mechanism not affecting β-catenin nuclear translocation.

A series of (hetero)arylpyrimidines agonists of the Wnt-beta-catenin cellular messaging system have been prepared. These compounds show activity in U2OS cells transfected with Wnt-3a, TCF-luciferase, Dkk-1 and tk-Renilla. Selected compounds show minimal GSK-3beta inhibition indicating that the Wnt-beta-catenin agonism activity most likely comes from interaction at Wnt-3a/Dkk-1. Two examples 1 and 25 show in vivo osteogenic activity in a mouse calvaria model. One example 1 is shown to activate non-phosphorylated beta-catenin formation in bone.

A high-throughput screening campaign to discover small molecule leads for the treatment of bone disorders concluded with the discovery of a compound with a 2-aminopyrimidine template that targeted the Wnt beta-catenin cellular messaging system. Hit-to-lead in vitro optimization for target activity and molecular properties led to the discovery of (1-(4-(naphthalen-2-yl)pyrimidin-2-yl)piperidin-4-yl)methanamine (5, WAY-262611). Compound 5 has excellent pharmacokinetic properties and showed a dose dependent increase in the trabecular bone formation rate in ovariectomized rats following oral administration.

Parathyroid hormone (PTH) activates multiple signaling pathways following binding to the PTH1 receptor in osteoblasts. Previous work revealed a discrepancy between cAMP stimulation and CRE reporter activation of truncated PTH peptides, suggesting that additional signaling pathways contribute to activation of the CRE. Using a CRE-Luciferase reporter containing multiple copies of the CRE stably transfected into the osteoblastic cell line Saos-2, we tested the ability of modulators of alternative pathways to activate the CRE or block the PTH-induced activation of the CRE. Activators of non-cyclic AMP pathways, that is, EGF (Akt, MAPK, JAK/STAT pathways); thapsigargin (intracellular calcium pathway); phorbol myristate acetate (protein kinase C, PKC pathway) induced minor increases in CRE-luciferase activity alone but induced dramatic synergistic effects in combination with PTH. The protein kinase A (PKA) inhibitor H-89 (10 µM) almost completely blocked PTH-induced activation of the CRE-reporter. Adenylate cyclase inhibitors SQ 22536 and DDA had profound and time-dependent biphasic effects on the CRE response. The MAPK inhibitor PD 98059 partially inhibited basal and PTH-induced CRE activity to the same degree, while the PKC inhibitor bisindolylmaleimide (BIS) had variable effects. The calmodulin kinase II inhibitor KN-93 had no significant effect on the response to PTH. We conclude that non-cAMP pathways (EGF pathway, calcium pathway, PKC pathway) converge on, and have synergistic effects on, the response of a CRE reporter to PTH. J. Cell. Biochem. 106: 887–895, 2009. © 2009 Wiley-Liss, Inc.

A single point mutation (G to T) in the low-density lipoprotein receptor related protein 5 (LRP5) gene results in a glycine to valine amino acid change (G171V) and is responsible for an autosomal dominant high bone mass trait (HBM) in two independent kindreds. LRP5 acts as a co-receptor to Wnts with Frizzled family members and transduces Wnt-canonical signals which can be antagonized by LRP5 ligand, Dickkopf 1 (Dkk1). In the presence of Wnt1, LRP5 or the HBM variant (LRP5-G171V) induces beta-catenin nuclear translocation and activates T cell factor (TCF)-luciferase reporter activity. HBM variant suppresses Dkk1 function and this results in reduced inhibition of TCF activity as compared to that with LRP5. Structural analysis of LRP5 revealed that the HBM mutation lies in the 4th blade of the first beta-propeller domain. To elucidate the functional significance and consequence of the LRP5-G171V mutation in vitro, we took a structure-based approach to design 15 specific LRP5 point mutations. These included (a) substitutions at the G171 in blade 4, (b) mutations in blades 2-6 of beta-propeller 1, and (c) mutations in beta-propellers 2, 3 and 4. Here we show that substitutions of glycine at 171 to K, F, I and Q also resulted in HBM-like activity in the presence of Wnt1 and Dkk1. This indicates the importance of the G171 site rather than the effect of specific amino acid modification to LRP5 receptor function. Interestingly, G171 equivalent residue mutations in other blades of beta-propeller 1 (A65V, S127V, L200V, A214V and M282V) resulted in LRP5-G171V-like block of Dkk1 function. However G171V type mutations in other beta-propellers of LRP5 did not result in resistance to Dkk1 function. These results indicate the importance of LRP5 beta-propeller 1 for Dkk1 function and Wnt signaling. These data and additional comparative structural analysis of the LRP5 family member LDLR suggest a potential functional role of the first beta-propeller domain through intramolecular interaction with other domains of LRP5 wherein Dkk1 can bind. Such studies may also lead to a better understanding of the mechanisms underlying the reduced function of Dkk1-like inhibitory ligands of LRP5 with HBM-like mutations and its relationship to increased bone density phenotypes.

There is considerable evidence implicating the cAMP-signaling pathway in the anabolic action of PTH; and to date, all PTH and PTHrp peptides that stimulate cyclic AMP are active in animal models of osteogenesis. We have tested two C-terminally truncated peptides, PTH(1-29) and a modified PTH(1-21) (MPTH(1-21)), in in vitro and in vivo assays of PTH action. Each of the C-terminally truncated peptides was of low nanomolar potency in assays of receptor binding and cAMP stimulation. However, when we tested these peptides for functional response in Saos-2 cells stably transfected with a cyclic AMP response element (CRE) reporter, the C-terminally truncated peptides were two to four times less potent than would be expected from their binding and cAMP-stimulating properties. Furthermore, PTH(1-29), although active, was approximately 20-fold less potent than PTH(1-34) in a rat model of osteogenesis while MPTH(1-21) was inactive. The relative lack of activity of these peptides in vivo suggests that while activation of the cAMP pathway may be important for the anabolic effect of PTH fragments, it is not, of itself, predictive of their in vivo activity.

Human adenovirus vectors containing intact or largely deleted E3 region were used to construct adenovirus-hepatitis B recombinant viruses (Ad-HepB) and shown to produce substantial amount of recombinant protein, hepatitis B surface antigen (HBsAg), in tissue culture. Previously we showed that these viruses were able to elicit good anti-HBs antibodies in a dog model. In the present study, the Ad-HepB viruses were evaluated for replication and immunogenicity in chimpanzees which sustain permissive infection by human adenoviruses. Recombinants containing entire E3 region showed better replication pattern than their E3 deleted counterparts as evidenced by longer duration and high titers of virus shedding. The effect of E3 region was also seen in the antibody titers against HBsAg in that the E3 containing viruses showed better response than the E3 deleted viruses. The importance of E3 region for the development of adenovirus vectored vaccines is further discussed.

Recombinant adenovirus (Ad)-human immunodeficiency virus (HIV) vaccines expressing HIVIIIB Env and Gag proteins were evaluated for immunogenicity in chimpanzees following intranasal administration. When Ad7-, Ad4-, and Ad5-vectored vaccines were administered sequentially at 0, 24, and 52 weeks, respectively, to three chimpanzees, the inoculations resulted in limited virus replication in the nasopharynx, but extensive Ad-HIV replication occurred in the intestine. High-titered IgG serum antibody responses to Env and Gag that were nonneutralizing were induced following booster administration of Ad4-HIV recombinant viruses. Following the Ad5-HIV booster, low levels of neutralizing antibodies as well as V3 loop antibodies were induced in all three chimpanzees that persisted for several months. Administration of a gp160 subunit vaccine (baculovirus derived) in SAF-m 24 weeks later boosted broadly neutralizing serum antibodies that peaked within 1 month of the injection. Two additional subunit boosters 19 and 37 weeks later were progressively less effective at stimulating serum neutralizing antibody responses. Substantial local immune responses were induced in nasal, vaginal, and salivary secretions following the third Ad-HIV intranasal immunization. These responses were further boosted with the gp160 subunit vaccine, which also stimulated production of rectal antibodies. The predominant responses in all secretions tested were of the IgG isotype, although some IgA responses were also detected. Strong blastogenic responses to HIV recombinant Env and Gag proteins were induced after each immunization.

Publisher Summary Osteoblasts arise from mesenchymal stem cells, and the differentiation of these progenitors to osteoblasts is controlled by many growth factors and hormones including Wnts and parathyroid hormone (PTH). The Wnt signaling pathway modulates many cell-fate decisions throughout development by controlling gene expression, cell behavior, adhesion and polarity. Aberrant regulation of the Wnt pathway results in inappropriate expression of target genes that leads to a variety of abnormalities and degenerative diseases and is implicated in several human cancers including osteosarcoma. Studies have indicated that the secreted Wnt antagonist dickkopf-1 inhibits osteoblastic activity, promotes osteoclastic function and plays a role in the pathophysiology of multiple myeloma. Another Wnt antagonist, secreted frizzled-related protein-2, also appears to be involved in this disease. Osteosarcoma is a rare, sporadic malignancy that is found in only 4–5 people per million. These tumors are malignant mesenchymal cancers that produce osteoid and bone, occasionally in a trabecular pattern. Unlike most cancers, osteosarcoma is predominantly a disease of young people that peaks in the second and third decades of life and appears to be associated with the adolescent growth spurt. This cancer is also more common in patients with Paget's disease, and some studies suggest an additional peak after the age of 50. Pre-clinical safety studies with PTH revealed a link between long-term treatment with the peptide and the development of bone neoplasias. The mechanisms for the induction of osteosarcoma by PTH are not well understood but may involve stimulation of osteoblastic cell proliferation and/or prevention of cellular apoptosis.

Dkk1 is a secreted antagonist of the LRP5-mediated Wnt signaling pathway that plays a pivotal role in bone biology. Because there are no well-documented LRP5-based assays of Dkk1 binding, we developed a cell-based assay of Dkk1/LRP5 binding using radioactive 125I-Dkk1. In contrast to LRP6, transfection of LRP5 alone into 293A cells resulted in a low level of specific binding that was unsuitable for routine assay. However, co-transfection of LRP5 with the chaperone protein MesD (which itself does not bind Dkk1) or Kremen-2 (a known Dkk1 receptor), or both, resulted in a marked enhancement of specific binding that was sufficient for evaluation of Dkk1 antagonists. LRP5 fragments comprising the third and fourth β-propellers plus the ligand binding domain, or the first β-propeller, each inhibited Dkk1 binding, with mean IC50s of 10 and 196 nM, respectively. The extracellular domain of Kremen-2 (“soluble Kremen”) was a weaker antagonist (mean IC50 806 nM). We also found that cells transfected with a high bone mass mutation LRP5(G171V) had a subtly reduced level of Dkk1 binding, compared to wild type LRP5-transfected cells, and no enhancement of binding by MesD. We conclude that (1) LRP5-transfected cells do not offer a suitable cell-based Dkk1 binding assay, unless co-transfected with either MesD, Kremen-2, or both; (2) soluble fragments of LRP5 containing either the third and fourth β-propellers plus the ligand binding domain, or the first β-propeller, antagonize Dkk1 binding; and (3) a high bone mass mutant LRP5(G171V), has subtly reduced Dkk1 binding, and, in contrast to LRP5, no enhancement of binding with MesD. J. Cell. Biochem. 108: 1066–1075, 2009. © 2009 Wiley-Liss, Inc.

Adenoviruses possess a combination of features that make them highly suitable as vectors for expression of heterologous genes. Non-conditional and non-defective adeno-vectors have been constructed to obtain high level expression of a number of foreign genes and some of them have been shown in animal models to exhibit excellent promise as vaccine candidates.

The bone marrow microenvironment provides a unique opportunity in vivo to assess the role of genes in bone remodeling. The objective of this study was to determine whether Runx2 expression is regulated by rhBMP-2 in vivo and to examine the effect of Runx2 overexpression on bone in vivo. In the in vivo calvaria model we used, rhBMP-2 induced Runx2 protein expression in periosteal cells while in vitro, adenovirus-mediated Runx2 overexpression induced mineralization in mesenchymal stem cells. A single injection of adeno-Runx2 directly into the bone marrow of the right femur in mature rats, and subsequent analysis after 3 weeks, showed a significant bone mineral density (BMD) increase (∼15%) as compared to the controls. The whole-femur mean BMD of the active virus-injected group was 0.193 (g/cm2) while that of the control virus-injected group was 0.175 (g/cm2) (P

Isogenic cell lines differing only in the expression of the protein of interest provide the ideal platform for cell-based screening. However, related natural lines differentially expressing the therapeutic target of choice are rare. Here the authors report a strategy for drug screening employing isogenic human cell lines in which the expression of the target protein is regulated by a gene-specific engineered zinc-finger protein (ZFP) transcription factor (TF). To demonstrate this approach, a ZFP TF activator of the human parathyroid hormone receptor 1 (PTHR1) gene was identified and introduced into HEK293 cells (negative for PTHR1). Following induction of ZFP TF expression, this cell line produced functional PTHR1 protein, resulting in a robust and ligand-specific cyclic adenosine monophosphate (cAMP) response. Reciprocally, the natural expression of PTHR1 observed in SAOS2 cells was dramatically reduced by the introduction of the appropriate PTHR1-specific ZFP TF repressor. Moreover, this ZFP-driven PTHR1 repression selectively eliminated the functional cAMP response invoked by known ligands of PTHR1. These data establish ZFP TF-generated isogenic lines as a general approach for the identification of therapeutic agents specific for the target gene of interest.

The Runx2 (CBFA1/AML3/PEBP2alphaA) transcription factor promotes lineage commitment and differentiation by activating bone phenotypic genes in postproliferative osteoblasts. However, the presence of Runx2 in actively dividing osteoprogenitor cells suggests that the protein may also participate in control of osteoblast growth. Here, we show that Runx2 is stringently regulated with respect to cell cycle entry and exit in osteoblasts. We addressed directly the contribution of Runx2 to bone cell proliferation using calvarial osteoblasts from wild-type and Runx2-deficient mice (i.e., Runx2(-/-) and Runx2(DeltaC/DeltaC)). Runx2(DeltaC/DeltaC) mice express a protein lacking the Runx2 COOH terminus, which integrates several cell proliferation-related signaling pathways (e.g., Smad, Yes/Src, mitogen-activated protein kinase, and retinoblastoma protein). Calvarial cells but not embryonic fibroblasts from Runx2(-/-) or Runx2(DeltaC/DeltaC) mutant mice exhibit increased cell growth rates as reflected by elevations of DNA synthesis and G(1)-S phase markers (e.g., cyclin E). Reintroduction of Runx2 into Runx2(-/-) calvarial cells by adenoviral delivery restores stringent cell growth control. Thus, Runx2 regulates normal osteoblast proliferation, and the COOH-terminal region is required for this biological function. We propose that Runx2 promotes osteoblast maturation at a key developmental transition by supporting exit from the cell cycle and activating genes that facilitate bone cell phenotype development.

A unique mutation in LRP5 is associated with high bone mass in man. Transgenic mice expressing this LRP5 mutation have a similar phenotype with high bone mass and enhanced strength. These results underscore the importance of LRP5 in skeletal regulation and suggest targets for therapies for bone disease. A mutation (G171V) in the low-density lipoprotein receptor related protein 5 (LRP5) has been associated with high bone mass (HBM) in two independent human kindreds. To validate the role of the mutation, several lines of transgenic mice were created expressing either the human LRP5 G171V substitution or the wildtype LRP5 gene in bone. Volumetric bone mineral density (vBMD) analysis by pQCT showed dramatic increases in both total vBMD (30-55%) and trabecular vBMD (103-250%) of the distal femoral metaphysis and increased cortical size of the femoral diaphysis in mutant G171V transgenics at 5, 9, 17, 26, and 52 weeks of age (p < 0.01 for all). In addition, high-resolution microcomputed tomography (microCT) analysis of the distal femorae and lumbar vertebrae revealed an increase (110-232%) in trabecular bone volume fraction caused by both increased trabecular number (41-74%) and increased trabecular thickness (34-46%; p < 0.01 for all) in the mutant G171V mice. The increased bone mass was associated with significant increases in vertebral compressive strength (80-140%) and the increased cortical size with significant increases in femoral bending strength (50-130%). There were no differences in osteoclast number at 17 weeks of age. However, compared with littermate controls, the mutant G171V transgenic mice showed an increase in actively mineralizing bone surface, enhanced alkaline phosphatase staining in osteoblasts, and a significant reduction in the number of TUNEL-positive osteoblasts and osteocytes. These results suggest that the increased bone mineral density in mutant G171V mice was caused by increased numbers of active osteoblasts, which could in part be because of their increased functional lifespan. While slight bone anabolic activity was observed from overexpression of the wildtype LRP5 gene, it is clear that the G171V mutation, rather than overexpression of the receptor itself, is primarily responsible for the dramatic HBM bone effects. Together, these findings establish the importance of this novel and unexpected role of a lipoprotein receptor in regulating bone mass and afford a new model to explore LRP5 and its recent association with Wnt signaling in bone biology.

In the absence of an adequate small animal model for testing the efficacy of adenovirus-vectored respiratory syncytial virus (RSV) vaccines, a ferret model was established for this purpose. Recombinant adenovirus types 4, 5 and 7 expressing the RSV fusion glycoprotein (F), the attachment glycoprotein (G) or both F and G were constructed previously. These recombinants contain a deletion of a large portion of the E3 region of the respective adenovirus vector. In addition, an Ad7(E3+)F recombinant virus which contains an intact E3 region was constructed to assess whether E3 region functions might enhance vaccine immunogenicity. Evaluation of these viruses in the ferret model demonstrated that Ad4 and Ad5 recombinants, administered intranasally to ferrets, induce stronger seroresponses to RSV than do Ad7 recombinant viruses. Ad7(E3+)F did not show enhanced immunogenicity relative to E3-deleted recombinant viruses. However, measurement of RSV infectivity in nasal washes, following intranasal RSV challenge, showed that five different vaccination regimens, Ad7F/Ad4F, Ad7G/Ad4G, Ad7FG/Ad4FG, Ad4F/Ad7(E3+)F and Ad5F/Ad4F, protected ferrets from RSV infection in a dose-dependent manner.

High yielding adenovirus (Ad)-hepatitis B recombinant (Ad-Hep B) viruses were prepared by insertion of the hepatitis B surface antigen (HBsAg) gene into the early region 3 (E3 region) of Ad4 or Ad7 vectors containing intact or largely deleted E3 regions. Both E3-deleted and non-deleted recombinants produced about six- to eight-fold higher amounts of HBsAg than previously reported recombinants. These recombinant viruses were evaluated for immunogenicity in dogs which sustain abortive lung infections by Ad4 and Ad7. Recombinants containing E3 deletions elicited 10- to 12-fold stronger anti-HBs primary responses than previously evaluated low yield recombinants. Further immunizations with heterotypic Ad-Hep B recombinants induced substantial anti-HBs booster responses as well as anti-‘a’ epitope responses. In contrast, recombinant viruses containing intact E3 regions induced only weak or negligible anti-HBs responses, although such viruses induced strong antibody responses to the Ad vectors. The immunogenicity of high-yielding Ad recombinants correlated with temporal expression of HBsAg and thus the dog represents a valuable model for evaluation of immune responses to heterologous proteins that are expressed early and that do not require efficient DNA replication. Recombinants expressing HBsAg late in the infectious cycle require further testing in the fully permissive chimpanzee model. This study establishes that the E3-deleted high yield Ad4 and Ad7 recombinants represent promising live oral hepatitis B vaccine candidates.

Recombinant adenovirus type 4, 5, and 7 expressing the fusion glycoprotein (F) gene, the attachment glycoprotein (G) gene, or both F and G genes of respiratory syncytial virus (RSV) was constructed. Intratracheal immunization of dogs with Ad7F induced moderate titers of RSV-neutralizing antibodies. After booster immunization with Ad4F, the dogs developed high titers of RSV-specific antibody. Subsequently, three two-dose vaccination regimens, Ad4F/Ad5F, Ad7G/Ad4G, and Ad7FG/Ad4FG, were compared with Ad7F/Ad4F for immunogenicity and protective efficacy. The results indicated that Ad4F/Ad5F was equal or greater in immunogenicity to Ad7F/Ad4F, but Ad7G/Ad4G and Ad7FG/Ad4FG were less effective than Ad7F/Ad4F in inducing RSV-neutralizing antibody. All vaccination regimens completely protected the lungs of dogs from RSV infection. A chimpanzee was sequentially immunized orally with Ad7F, Ad4F, and Ad5F. A low-level antibody response to RSV was induced after the primary immunization, but no significant increases were observed after booster immunizations.

(1990). Helper Independent Recombinant Adenovirus Vectors: Expression of HIV env or HBV Surface Antigen. International Reviews of Immunology: Vol. 7, No. 1, pp. 67-77.

Region E3 of adenovirus encodes about nine overlapping mRNAs (a to i) with different spliced structures and with two major RNA 3' end sites termed E3A and E3B. Synthesis of E3 mRNAs was examined by the nuclease-gel procedure at early and late stages of infection by wild-type virus (rec700) and by several E3 deletion mutants. Our results, together with those obtained by electron microscopy (L. T. Chow, T. R. Broker, and J. B. Lewis, J. Mol. Biol. 134:265-303, 1979), suggest that E3 may be differentially regulated at early and late stages at both the promoter and RNA-processing levels. Early after infection, the E3 promoter is used to make mainly mRNAs a and h. Late after infection, the E3 promoter appears to be shut off and the major late promoter is used to make mainly mRNAs d and e. The late L4 mRNA 3' end site is not used early even though early E3 pre-mRNAs transcribe through the L4 RNA 3' end site. The nucleotide 768-951 exon, which is the y leader on many L5 mRNAs, is very abundant late. (Nucleotide +1 is the major E3 transcription initiation site.) Early after infection, the 951 5' splice site is enhanced 5- to 10-fold in the dl712 (delta 1691 to 2122) group of mutants; late after infection, these mutants resemble the wild type. We speculate that the activity of the 951 5' splice site is regulated at early and late stages; it is suppressed early to permit synthesis of mRNA a, and it is activated late to permit synthesis of mRNAs d, e, and L5. With dl719 (delta 2173 to 2237), the 951----2157 splice is enhanced both early and late; we suggest that this deletion enhances the 2157 3' splice site.

Using nuclease gel analyses, the sites in the approximately 4000 nucleotide (nt) E3 transcription unit of adenovirus 5 (Ad5) that encode the 5' ends, 3' ends, and 5' and 3' splice sites of the approximately 10 E3 mRNAs were determined. Transcription initiation of all mRNAs occurs at two major (nt 1 and 8) and approximately two minor sites, situated 20-30 nt 3' to a TATA box. There are two major 3' end sites (nt 2227 and 3308), located approximately 20 nt downstream from ATTAAA and an AATAAA sequences, respectively. Thus, ATTAAA, as well as the usual AATAAA, apparently can function as a 3' end signal. There are two 5' splice sites (nt 372 and 923), both with GT at the intron boundary. There are four 3' splice sites (nt 766, 1817, 2201, and 2880), all with AG at the intron boundary. The nt 1817, 2201, and 2880 3' splice sites are located immediately upstream from open reading frames, such that splicing at the different sites allows synthesis of completely different proteins.

We described a simple method to introduce site-specific mutations into region E3 of adenovirus (Ad). Mutations are made in cloned Ad2 EcoRI-D (map position 76-83), then ligated between Ad5 EcoRI-A (map position 0-76) and EcoRI-B (map position 83-100) to complete the viral genome. We have used this method to isolate a viable virus mutant (dl702) that is relevant to the problems of translation initiation and gene organization in the E3 complex transcription unit. mRNA a in region E3 encodes an abundant glycoprotein termed gp19K. There are two AUGs in mRNA a that are 5' to AUG1204 which initiates gp19K. One of these, AUG1022, could initiate a 6.7K protein, although this protein has not been identified in infected cells. Mutant dl702 has a deletion such that the 6.7K gene is fused in-frame to the gp19K gene. We report that the 6.7K-gp19K fusion protein is synthesized both in dl702-infected cells and after cell free translation of infected cell RNA. The quantity of fusion protein made is much less than that of wild type gp19K. The sequence context of AUG1022 for 6.7K is AGUAUAUGA, and that of AUG1204 for gp19K is CCAAGAUGA. The consensus sequence of eukaryotic initiation codons is CCPuCCAUGG, with the Pu at -3 being important (M. Kozak, Nucleic Acids Res. 12, 857-872, 1984). Our results suggest that (i) AUG1022 can initiate translation in vivo and therefore the 6.7K protein probably is made in infected cells, (ii) that mRNA a is a dicistronic mRNA encoding the 6.7K and gp19K proteins, and (iii) that the initiation codon for 6.7K may be much less efficient than that for gp19K. Thus, the E3 genes may be organized such that the relative abundance of the 6.7K and gp19K proteins is controlled by the efficiency of their initiation codons in the same mRNA.

Region E3 of the adenovirus encodes about ten overlapping mRNAs ( a to j ) with different splicing patterns and with two RNA 3′ end sites termed E3A and E3B. We have examined how deletions in 12 viable virus mutants affect differential RNA processing in E3. We assayed E3 mRNAs by the nuclease-gel and RNA blot procedures. Some deletions had no effect whereas others (e.g. deletion of a 3′ splice or the E3A 3′ end signal) had the anticipated effects on RNA processing. However, deletions in two regions had surprising effects. Deletions in one region (nucleotides 1691 to 2044) enhanced splicing at the upstream 951 5′ splice site and the downstream 2157 and/or 2880 3′ splice sites. Some of these deletions prevented RNA 3′ end formation at the downstream E3A site. Deletion in the other region (nucleotides 2173 to 2237) enhanced an upstream splice site (951 to 2157) such that almost all pre-mRNA was processed into mRNA f . We suggest that these two regions contain cis -acting signals that regulate differential RNA processing. We discuss the results in terms of RNA folding and scanning models for splicing, as well as models for differential RNA 3′ end formation at the E3A versus the E3B site.

The E3 complex transcription unit of adenovirus encodes overlapping mRNAs (a to i) with different exon structures. The major mRNAs are a (approximately 40% of the total) and c (approximately 15%), which are spliced once, and f (approximately 15%) and h (approximately 25%), which are spliced twice. mRNA a uses the upstream E3A polyadenylation site, and the other mRNAs use the downstream E3B polyadenylation site. We analyzed virus deletion mutants to identify sequences important in alternative pre-mRNA processing in region E3. Our main finding is that a 64-base-pair deletion in dl742 causes mainly mRNAs f and h to be formed. mRNAs a and c are barely made. dl742 does not delete either a splice site or a polyadenylation site. Thus, the sequences deleted must function in alternative pre-mRNA processing independently of the signals at the actual splice and polyadenylation sites. The lack of synthesis of mRNA a by dl742 does not appear to result from a defect in the E3A polyadenylation signal but rather from an increase in splicing activity which results in the synthesis of doubly spliced mRNAs f and h at the expense of singly spliced mRNAs a and c. This suggests, in the wild-type situation, that the frequency of use of the E3A versus the E3B polyadenylation site may be determined by the rate of splicing, as well as, presumably, the rate of cleavage-polyadenylation at the E3A site.

We mapped the location of the E3A RNA 3' end site in the E3 transcription unit of adenovirus 2. The procedure used was nuclease-gel analysis with 32P-labeled RNA probes. The poly(A) addition sites were microheterogeneous and were located approximately 17 to 29 nucleotides downstream from an ATTAAA sequence. To identify the sequences that make up the E3A RNA 3' end signal, we constructed five viable virus mutants with deletions in or near the E3A RNA 3' end site. The mutants were analyzed for E3A RNA 3' end formation in vivo. No effect was observed from a 47-base-pair (bp) deletion (dl716) or a 72-bp deletion (dl714) located 22 and 19 nucleotides, respectively, upstream of the ATTAAA. In contrast, E3A RNA 3' end formation was abolished by a 554-bp deletion (dl708) that removes both the ATTAAA and the poly(A) addition sites, a 124-bp deletion (dl713) that removes the ATTAAA but leaves the poly(A) addition sites, and a 65-bp deletion (dl719) that leaves the ATTAAA but removes the poly(A) addition sites. These results indicate that the ATTAAA, as well as downstream sequences, including the poly(A) addition sites, are required for E3A RNA 3' end formation.

Region E3 of adenovirus is a "complex" transcription unit: i.e. different mRNAs and proteins arise by differential RNA 3' end selection and differential splicing of the primary transcript. We are using viable virus mutants to understand the controls that dictate the specificity and efficiency of the RNA processing signals. We describe a novel class of deletion mutations that enhance a natural 5' splice site located approximately 740 nucleotides (nt) upstream. In particular, deletions within nt 1691-2044 in the E3 transcription unit result in a 5-fold enhancement of the 5' splice site at nt 951 (as reflected in steady-state mRNA). The effect is specific, because the deletions do not affect the 5' splice site at nt 372, and because deletions within nt 2044-2214 do not affect either the 951 or the 372 5' splice sites. As a consequence of the enhanced splicing at the 951 5' site, synthesis of the major E3 mRNA and the major E3 protein (gp19K) are dramatically reduced. At least one of the natural 3' splice sites, located at nt 2157, is the recipient of the enhanced splicing at the 951 5' splice site. We conclude that sequences located within nt 1691-2044 affect (probably in cis) splicing at the 951 5' splice site. We speculate that nt 1691-2044 includes a splicing control region which functions to suppress splicing at the 951 5' splice site.

A total of 59 cytocidal (cyt) mutants were isolated from adenovirus 2 (Ad2) and Ad5. In contrast to the small plaques and adenovirus type of cytopathic effects produced by wild-type cyt+ viruses, the cyt mutants produced large plaques, and the cytopathic effect was characterized by marked cellular destruction. cyt mutants were transformation defective in established rat 3Y1 cells. cyt+ revertants and cyt+ intragenic recombinants recovered fully the transforming ability of wild-type viruses. Thus, the cyt gene is an oncogene responsible for the transforming function of Ad2 and Ad5. Genetic mapping in which we used three Ad5 deletion mutants (dl312, dl313, and dl314) as reference deletions located the cyt gene between the 3' ends of the dl314 deletion (nucleotide 1,679) and the dl313 deletion (nucleotide 3,625) in region E1B. Restriction endonuclease mapping of these recombinants suggested that the cyt gene encodes the region E1B 19,000-molecular-weight (175R) polypeptide (nucleotides 1,711 to 2,236). This was confirmed by DNA sequencing of eight different cyt mutants. One of these mutants has a single missense mutant, two mutants have double missense mutations, and five mutants have nonsense mutations. Except for one mutant, these point mutations are not located in any other known region E1B gene. We conclude that the cyt gene codes for the E1B 19,000-molecular-weight (175R) polypeptide, that this polypeptide is required for morphological transformation of rat 3Y1 cells, and that simple amino acid substitutions in the protein can be sufficient to produce the cyt phenotype.

Five viable virus mutants were constructed with deletions near a 3' splice site located at nucleotide 2157 in the E3 transcription unit of adenovirus 2. The mutants were examined for splicing activity at the 2157 3' splice site in vivo by nuclease-gel analysis of steady-state cytoplasmic mRNA. Splicing was not prevented by an exon deletion (dl719) that leaves 16 5'-proximal exon nucleotides intact or by intron deletions that leave 34 (dl717, dl712) or 18 (dl716) 3'-proximal intron nucleotides intact. The sequences deleted in one of these intron mutants (dl716) include the putative branchpoint site used in lariat formation during splicing. Thus, a surrogate branchpoint site apparently can be used for splicing. Another intron mutant (dl714) has a deletion that leaves 15 3'-proximal intron nucleotides intact; remarkably, this deletion virtually abolished splicing, even though the deletion is only 3 nucleotides closer to the splice site than is the deletion in dl716 which splices normally. The three nucleotides deleted in dl714 that are retained by dl716 are the sequence TGT. The TGT sequence is located on the 5' boundary of the pyrimidine-rich region upstream of the nucleotide 2157 3' splice site. Such pyrimidine-rich regions are ubiquitous at 3' splice sites. Most likely, the TGT is required for splicing at the nucleotide 2157 3' splice site. The TGT may be important because of its specific sequence or because it forms the 5' boundary of the pyrimidine-rich region.

Region E3 of adenovirus encodes about 10 overlapping mRNAs with different spliced structures. The mRNAs are 5' coterminal and form two major 3'-coterminal families termed E3A and E3B. As a group, the mRNAs have two 5' splice sites and four or five 3' splice sites. We previously described a novel class of virus mutants with deletions that enhance distant upstream and downstream 5' and 3' splice sites in region E3 (S. L. Deutscher, B. M. Bhat, M. H. Pursley, C. Cladaras, and W. S. M. Wold, Nucleic Acids Res. 13:5771-5788, 1985). We now report that two of these mutants, dl710 and dl712, are defective in RNA 3'-end formation at the E3A site. This result was surprising because the deletions in dl710 and dl712 are upstream of the putative signal for E3A RNA 3'-end formation. The explanation that we favor for this result is that the enhanced splicing activity in these mutants results in the splicing out of the E3A 3'-end site from the RNA precursor before the E3A 3' ends have a chance to form.

The E19 protein of adenoviruses is a transmembrane protein that abrogates the intracellular transport of class I antigens by forming complexes with them in the ER. We show here that the E19 protein is retained in the ER even in the absence of class I antigens. To define the region conferring residency in the ER, we examined two mutant forms of the viral protein. A 5 amino acid extension of the 15-membered cytoplasmic tail of the protein reduces its interaction with class I antigens but does not change its intracellular distribution. Shortening the tail to 7 amino acids also diminishes the affinity for class I antigens; however, this mutant E19 protein becomes transported to the cell surface. Thus, we concluded that a small stretch of amino acids exposed on the cytoplasmic side of the ER membrane is responsible for the retention of the E19 protein in the ER.