Tissue-specific replication of Friend and Moloney murine leukemia viruses in infected mice.

We have studied the replication of ecotropic murine leukemia viruses (MuLV) in the spleens and thymuses of mice infected with the lymphocytic leukemia-inducing virus Moloney MuLV (M-MuLV), with the erythroleukemia-inducing virus Friend MuLV (F-MuLV), or with in vitro-constructed recombinants between these viruses in which the long terminal repeat (LTR) sequences have been exchanged. At 1 week after infection both the parents and the LTR recombinants replicated predominantly in the spleens with only low levels of replication in the thymus. At 2 weeks after infection, the patterns of replication in the spleens and thymuses were strongly influenced by the type of LTR. Viruses containing the M-MuLV LTR exhibited a remarkable elevation in thymus titers which frequently exceeded the spleen titers, whereas viruses containing the F-MuLV LTR replicated predominantly in the spleen. In older preleukemic mice (5 to 8 weeks of age) the structural genes of M-MuLV or F-MuLV predominantly influenced the patterns of replication. Viruses containing the structural genes of M-MuLV replicated efficiently in both the spleen and thymus, whereas viruses containing the structural genes of F-MuLV replicated predominantly in the spleen. In leukemic mice infected with the recombinant containing F-MuLV structural genes and the M-MuLV LTR, high levels of virus replication were observed in splenic tumors but not in thymic tumors. This phenotypic difference suggested that tumors of the spleen and thymus may have originated by the independent transformation of different cell types. Quantification of polytropic MulVs in late-preleukemic mice infected with each of the ecotropic MuLVs indicated that the level of polytropic MuLV replication closely paralleled the level of replication of the ecotropic MuLVs in all instances. These studies indicated that determinants of tissue tropism are contained in both the LTR and structural gene sequences of F-MuLV and M-MuLV and that high levels of ecotropic or polytropic MuLV replication, per se, are not sufficient for leukemia induction. Our results further suggested that leukemia induction requires a high level of virus replication in the target organ only transiently during an early preleukemic stage of disease.     

Distinct segments within the enhancer region collaborate to specify the type of leukemia induced by nondefective Friend and Moloney viruses.

The nondefective Moloney and Friend murine leukemia viruses induce T-cell lymphomas and erythroleukemias, respectively, after being injected into newborn NFS mice. In previous studies, we showed that the distinct disease specificities of the two viruses could be switched by exchanging a small segment, about 200 nucleotides in length, encompassing their enhancer regions. This segment included the direct repeat sequence and an adjacent GC-rich region of about 20 nucleotides defined in studies of Moloney murine sarcoma virus enhancer-promoter function (L. A. Laimins, P. Gruss, R. Pozzatti, and G. Khoury, J. Virol. 49:183-189, 1984). The direct repeats of Friend and Moloney viruses are identical in a central core sequence of 32 nucleotides but have sequence differences on either side of this core as well as in their GC-rich segments. To determine whether disease specificity resides in part or in all of the direct repeat and GC-rich region, we constructed recombinants between Friend and Moloney viruses within this segment and tested them for their disease-inducing phenotypes. We found that disease specificity, in particular the ability of Friend virus sequence to confer erythroleukemogenicity on Moloney virus, is encoded throughout the region in at least three separable segments: the 5' and 3' halves of the direct repeat and the GC-rich segment. When just one of these segments (either both 5' halves of the direct repeat, both 3' halves, or just the GC-rich segment) from Friend virus was substituted into a Moloney virus genome, it conferred only a negligible or low incidence of erythroleukemia (less than or equal to 5% to between 10 and 15%). Any two segments together were considerably more potent (35 to 95% erythroleukemia), with the most effective pair being the two halves of the direct repeat. Individual segments and pairs of segments were considerably more potent determinants when they were matched with a genome of the same origin. Thus, although sequences outside the enhancer region are minor determinants of disease specificity when the enhancer is derived entirely from either Friend or Moloney virus, they can play a significant role when the enhancer is of mixed origin. Some recombinant enhancers conferred a long latent period of disease induction. This was particularly striking when the 5' halves of each copy of the direct repeat sequence were derived from Moloney virus and the 3' halves were derived from Friend virus.(ABSTRACT TRUNCATED AT 400 WORDS)     

Mapping the viral sequences conferring leukemogenicity and disease specificity in Moloney and amphotropic murine leukemia viruses.

The Moloney murine leukemia virus (MuLV) is a highly leukemogenic virus. To map the leukemogenic potential of Moloney MuLV, we constructed chimeric viral DNA genomes in vitro between parental cloned infectious viral DNA from Moloney and amphotropic 4070-A MuLVs. Infectious chimeric MuLVs were recovered by microinjection of recombinant DNA into NIH/3T3 cells and tested for their leukemogenic potential by inoculation into NIH/Swiss newborn mice. Parental Moloney MuLV and amphotropic 4070-A MuLV induced thymic and nonthymic leukemia, respectively, when inoculated intrathymically. With chimeric MuLVs, we found that the primary determinant of leukemogenicity of Moloney and amphotropic MuLVs lies within the 1.5-kilobase-pair ClaI-PvuI long terminal repeat (LTR)-containing fragment. The presence of additional Moloney env-pol sequences with the Moloney LTR enhanced the leukemogenic potential of a chimeric MuLV significantly, indicating that these sequences were also involved in tumor development. Since parental viruses induced different forms of leukemia, we could also map the viral sequences conferring this disease specificity. We found that the 1.5-kilobase-pair ClaI-PvuI LTR-containing fragment of Moloney MuLV was necessary and sufficient for a chimeric MuLV to induce thymic leukemia. Similarly, the same LTR-containing fragment of amphotropic MuLV was necessary and sufficient for a chimeric MuLV to induce nonthymic leukemia. Therefore, our results suggest that specific sequences within this short LTR-containing fragment determine two important viral functions: the ability to transform cells in vivo (leukemic transformation) and the selection of a specific population of cells to be transformed (disease specificity).     

Nuclear factors that bind to the enhancer region of nondefective Friend murine leukemia virus.

Nondefective Friend murine leukemia virus (MuLV) causes erythroleukemia when injected into newborn NFS mice, while Moloney MuLV causes T-cell lymphoma. Exchange of the Friend virus enhancer region, a sequence of about 180 nucleotides including the direct repeat and a short 3'-adjacent segment, for the corresponding region in Moloney MuLV confers the ability to cause erythroid disease on Moloney MuLV. We have used the electrophoretic mobility shift assay and methylation interference analysis to identify cellular factors which bind to the Friend virus enhancer region and compared these with factors, previously identified, that bind to the Moloney virus direct repeat (N. A. Speck and D. Baltimore, Mol. Cell. Biol. 7:1101-1110, 1987). We identified five binding sites for sequence-specific DNA-binding proteins in the Friend virus enhancer region. While some binding sites are present in both the Moloney and Friend virus enhancers, both viruses contain unique sites not present in the other. Although none of the factors identified in this report which bind to these unique sites are present exclusively in T cells or erythroid cells, they bind to three regions of the enhancer shown by genetic analysis to encode disease specificity and thus are candidates to mediate the tissue-specific expression and distinct disease specificities encoded by these virus enhancer elements.     

Sequences in the U5-gag-pol region influence early and late pathogenic effects of Friend and Moloney murine leukemia viruses.

Friend replication-competent murine leukemia virus (F-MuLV), clone 57, induces a severe early hemolytic anemia and a later erythroleukemia after inoculation of newborn IRW or ICFW mice, whereas Moloney MuLV (M-MuLV) induces only lymphoid leukemia. We have shown previously that the attenuated hemolytic and erythroleukemogenic abilities of an F-MuLV variant, clone B3, were due mostly to changes in the env gene and long terminal repeat, respectively. For the present study, we derived two constructs exchanging env fragments of F-MuLV 57 and M-MuLV and compared them with two constructs described by Chatis et al. (J. Virol. 52:248-254, 1984) exchanging the U3 region of the long terminal repeat of the same parental viruses. When comparing the hemolytic effect of these constructs with those of the parent, we found that the U5-gag-pol region of F-MuLV was required for development of severe early hemolytic anemia and that, unlike the env of F-MuLV B3, the env of M-MuLV was fully competent in inducing severe early hemolytic anemia when associated with the F-MuLV U5-gag-pol and U3 regions. As expected, induction of erythroleukemia depended on the presence of the F-MuLV U3 region; however, the presence of both the U3 and U5-gag-pol regions of F-MuLV appeared to be synergistic and was associated with a more rapid appearance of erythroleukemia.     

Neurodegenerative disease induced by the wild mouse ecotropic retrovirus is markedly accelerated by long terminal repeat and gag-pol sequences from nondefective Friend murine leukemia virus.

The wild mouse ecotropic retrovirus (WM-E) induces a spongiform neurodegenerative disease in mice after a variable incubation period of 2 months to as long as 1 year. We isolated a molecular clone of WM-E (15-1) which was weakly neurovirulent (incidence, 8%) but was highly leukemogenic (incidence, 45%). Both lymphoid and granulocytic leukemias were observed, and these leukemias were often neuroinvasive. A chimeric virus was constructed containing the env and 3' pol sequences of 15-1 and long terminal repeat (LTR), gag, and 5' pol sequences from a clone of Friend murine leukemia virus (FB29). FB29 has been shown previously to replicate to high levels in the central nervous system (CNS) but is not itself neurovirulent. This finding was confirmed at the DNA level in the current study. Surprisingly, intraperitoneal inoculation of neonatal IRW mice with the chimeric virus (FrCasE) caused an accelerated neurodegenerative disease with an incubation period of only 16 days and was uniformly fatal by 23 days postinoculation. Introduction of the LTR of 15-1 into the FrCasE genome yielded a virus (FrCasEL) with a degree of neurovirulence intermediate between those of 15-1 and FrCasE. No differences were found in the levels of viremia or the relative levels of viral DNA in the spleens of mice inoculated with 15-1, FrCasE, or FrCasEL. However, the levels of viral DNA in the CNS correlated with the relative degrees of neurovirulence of the respective viruses (FrCasE greater than FrCasEL greater than 15-1). Thus, the env and 3' pol sequences of WM-E (15-1) were required for neurovirulence, but elements within the LTR and gag-pol regions of FB29 had a profound influence on the level of CNS infection and the rate of development of neurodegeneration.     

Sequences responsible for the distinctive hemolytic potentials of Friend and Moloney murine leukemia viruses are dispersed but confined to the psi-gag-PR region.

Friend and Moloney murine leukemia viruses (F- and M-MuLV) induce distinct diseases in hematopoietic tissues following inoculation of newborn mice of susceptible strains. F-MuLV induces erythroleukemia preceded by severe early hemolytic anemia; M-MuLV induces thymomas and only very mild hemolysis. The major viral determinant of severe early hemolytic anemia residues in the env gene, but sequences located outside this gene can modulate this effect. By means of genetic chimeras of F- and M-MuLV, we have found that although they are confined to the 5' portion of the env gene intron, sequences that determine the distinctive hemolytic potentials of F- and M-MuLV are widely distributed over a region spanning the RNA encapsidation domain, the gag gene, and the portion of the pol gene encoding the viral protease. Within this large region, two fragments of M-MuLV, a 1.3-kb region encoding the matrix, pp12, and capsid proteins and a 0.8-kb region encoding the nucleocapsid and the viral protease, were capable, individually, of partially attenuating the capacity of F-MuLV for induction of severe early hemolytic anemia. In association, these two fragments conferred complete attenuation. Moreover, a second pair of adjacent fragments within this large region appeared to behave cooperatively to confer complete attenuation; a 0.36-kb region roughly corresponding to the encapsidation domain, although not detectably altering hemolytic potential on its own, deepened the attenuation conferred by the adjacent 1.3-kb region. Whether capable of inducing severe early hemolytic anemia or not and despite different efficiencies of induction of recombinant polytropic viruses, all chimeric viruses retained the erythroleukemogenicity of the F-MuLV parent.     

Identification of a gag-encoded cytotoxic T-lymphocyte epitope from FBL-3 leukemia shared by Friend, Moloney, and Rauscher murine leukemia virus-induced tumors.

FBL-3 is a highly immunogenic murine leukemia of C57BL/6 origin induced by Friend murine leukemia virus (MuLV). Immunization of C57BL/6 mice with FBL-3 readily elicits CD8+ cytotoxic T lymphocytes (CTL) capable of lysing FBL-3 as well as syngeneic leukemias induced by Moloney and Rauscher MuLV. The aim of this current study was to identify the immunogenic epitope(s) recognized by the FBL-3-specific CD8+ CTL. A series of FBL-3-specific CD8+ CTL clones were generated from C57BL/6 mice immunized to FBL-3. The majority of CTL clones (32 of 38) were specific for F-MuLV gag-encoded antigen. By using a series of recombinant vaccinia viruses expressing full-length and truncated F-MuLV gag genes, the antigenic epitope recognized by the FBL-3 gag-specific CTL clones, as well as by bulk-cultured CTL from spleens of mice immune to FBL-3, was localized to the leader sequence of gPr80gag protein. The precise amino acid sequence of the CTL epitope in the leader sequence was identified as CCLCLTVFL (positions 85-93) by examining lysis of targets incubated with a series of synthetic leader sequence peptides. No evidence of other CTL epitopes in the gPr80gag or Pr65gag core virion structural polyproteins was found. The identity of CCLCLTVFL as the target peptide was validated by showing that immunization with the peptide elicited CTL that lysed FBL-3. The CTL elicited by the Gag peptide also specifically lysed syngeneic leukemia cells induced by Moloney and Rauscher MuLV (MBL-2 and RBL-5). The transmembrane peptide was shown to be the major gag-encoded antigenic epitope recognized by bulk-cultured CTL derived from C57BL/6 mice immunized to MBL-2 or RBL-5. Thus, the CTL epitope of FBL-3 is localized to the transmembrane anchor domain of the nonstructural Gag polyprotein and is shared by leukemia/lymphoma cell lines induced by Friend, Moloney, and Rauscher MuLV.     

Isolation of paralysis-inducing murine leukemia viruses from Friend virus passaged in rats.

Four clones of murine leukemia viruses (PVC-111, PVC-211, PVC-321, and PVC-441) were isolated from a paralyzed Fischer rat which had been infected with rat-passaged Friend leukemia virus. PVC-211 and PVC-321 viruses induced hind leg paralysis in rats and killed them within 1 month, and PVC-441 did so within 2 months after infection, whereas PVC-111 did not within 4 months. PVC-321 and PVC-441 but not PVC-111 virus grew well in brain and spinal cord media. The viral antigens were found often in glia cells and rarely in neurons of the rats infected with each of these PVC viruses. All of the PVC viruses induced neuronal degeneration but neither inflammation nor leukemic infiltration in the spinal cord. The isolated viruses were all ecotropic and NB-tropic. Age dependency of the susceptibility of rats to paralysis induction was observed.     

UAG readthrough is not increased in vivo by Moloney murine leukemia virus infection.

Expression of the pol gene of the murine leukemia viruses is subject to translational control at the UAG termination codon of the upstream gene gag. Previous experiments have suggested that: i) Moloney murine leukemia virus infection induces a tRNA(Gln)iii) in an in vitro system using the tobacco mosaic virus as template, this tRNA is able to increase readthrough at the UAG codon [1]. Here we demonstrate that, in vivo, Moloney murine leukemia virus infection does not increase translational readthrough at either the tobacco mosaic virus or the Moloney murine leukemia virus UAG stop codons.     

Different genes control the susceptibility of mice to Moloney or Abelson murine leukemia viruses.

The susceptibility of mice to lymphoma induction by Moloney or Abelson murine leukemia virus has been compared in BALB/c, C57BL/6, and BALB/cXC57BL/6 recombinant inbred strains. BALB/c mice were found to be susceptible to lymphoma induction by either virus, and C57BL/6 mice were found to be relatively resistant to lymphoma induction by either virus. The genes that control these patterns of susceptibility to each virus are not the same because susceptibility to each virus segregated independently in CXB recombinant inbred strains. We also found, as reported by Cook (W. Cook, Proc. Natl. Acad. Sci. U.S.A. 79:2917-2921, 1982), when injected intrathymically that Abelson murine leukemia virus rapidly induced thymomas in weanling B6 mice. Examination of the cellular phenotypes of the tumors induced by Abelson murine leukemia virus or by Moloney murine leukemia virus indicated that different lymphocyte subpopulations were the targets for tumor induction by each virus.     

Sequence homology between Moloney murine sarcoma virus and Moloney leukemia virus RNA.

The Moloney murine sarcoma-leukemia virus [M-MSV (MuLV)], propagated at high multiplicity of infection (MOI), was demonstrated previously to contain a native genome mass of 4 X 10(6) daltons as contrasted to a mass of 7 X 10(6) daltons for Moloney murine leukemia virus (M-MuLV). The 4 X 10(6)-dalton classof RNA from M-MSV (MuLV) was examined for base sequence homology with DNA complementary to the 7 X 10(6)-dalton M-MuLV RNA genome. Approximately 86% of the M-MSV (MuLV) was protected from RNase digestion by hybridization, whereas 95% of M-MuLV was protected under identical conditions. These results indicate that the small RNA class of high-MOI M-MSV (MuLV) contains little (perhaps 10%) genetic information not present in M-MuLV. Virtually all of the 1.8 X 10(6)-dalton subunits of M-MSV (MuLV) RNA contained regions of poly(A) since 94% of the RNA bound to oligo(dT) cellulose in 0.5 M KCl. This suggests that the formation of the 1.8 X 10(6)-dalton subunits occurs before their packaging into virions and does not result from hydrolysis of intact 3.5 X 10(6)-dalton subunits by a virion-associated nuclease.     

Coordinated disintegration reactions mediated by Moloney murine leukemia virus integrase.

The protein-DNA and protein-protein interactions important for function of the integrase (IN) protein of Moloney murine leukemia virus (M-MuLV) were investigated by using a coordinated-disintegration assay. A panel of M-MuLV IN mutants and substrate alterations highlighted distinctions between the intermolecular and intramolecular reactions of coordinated disintegration. Mispairing of the crossbone single-strand region and altered long terminal repeat (LTR) positioning affected the intermolecular, but not the intramolecular, reactions of coordinated disintegration. Partial components of the crossbone substrate were coordinated by M-MuLV IN, indicating a reliance on both LTR and target DNA determinants for substrate assembly. The intramolecular reaction was dependent on the presence of either the HHCC domain or a crossbone LTR 5' single-stranded tail. An M-MuLV IN mutant without the HHCC domain (Ndelta105) catalyzed reduced levels of double disintegration but not single disintegration. A separately purified HHCC domain protein (Cdelta232) stimulated double disintegration mediated by Ndelta105, suggesting a role of the N-terminal HHCC domain in stable IN-IN and IN-DNA interactions. Significantly, crossbone substrates lacking the LTR 5' tails were not recognized by the fingerless Ndelta105 protein. Collectively, these data suggest similar roles of the HHCC domain and 5' LTR tail in substrate recognition and modulation of IN activity.     

Assembly and composition of intracellular particles formed by Moloney murine leukemia virus.

Assembly of type C retroviruses such as Moloney murine leukemia virus (M-MuLV) ordinarily occurs at the plasma membranes of infected cells and absolutely requires the particle core precursor protein, Pr65gag. Previously we have shown that Pr65gag is membrane associated and that at least a portion of intracellular Pr65gag protein appears to be routed to the plasma membrane by a vesicular transport pathway. Here we show that intracellular particle formation can occur in M-MuLV-infected cells. M-MuLV immature particles were observed by electron microscopy budding into and within rough endoplasmic reticulum, Golgi, and vacuolar compartments. Biochemical fractionation studies indicated that intracellular Pr65gag was present in nonionic detergent-resistant complexes of greater than 150S. Additionally, viral RNA and polymerase functions appeared to be associated with intracellular particles, as were Gag-beta-galactosidase fusion proteins which have the capacity to be incorporated into virions. Immature intracellular particles in postnuclear lysates could be proteolytically processed in vitro to mature forms, while extracellular immature M-MuLV particles remained immature as long as 10 h during incubations. The occurrence of M-MuLV-derived intracellular particles demonstrates that Pr65gag can associate with intracellular membranes and indicates that if a plasma membrane Pr65gag receptor exists, it also can be found in other membrane compartments. These results support the hypothesis that intracellular particles may serve as a virus reservoir during in vivo infections.     

Preleukemic hematopoietic hyperplasia induced by Moloney murine leukemia virus is an indirect consequence of viral infection.

We previously showed that neonatal mice inoculated with Moloney murine leukemia virus (M-MuLV) exhibit a preleukemic state characterized by splenomegaly and increased numbers of hematopoietic progenitors. An M-MuLV variant with greatly reduced leukemogenic potential, Mo+PyF101 M-MuLV, does not generally induce this preleukemic state. In order to investigate the mechanism involved in M-MuLV induction of preleukemic hyperplasia, we tested the CFU-mixed myeloid and erythroid (CFUmix) from M-MuLV- and Mo+PyF101 M-MuLV-inoculated mice for the presence of virus by antibody staining and for the release of infectious virus. The majority of CFUmix colonies from both M-MuLV- and Mo+PyF101 M-MuLV-inoculated mice contained infectious virus even though M-MuLV-inoculated mice showed elevated levels of CFUmix while the Mo+PyF101 M-MuLV-inoculated mice did not. This indicates that direct infection of hematopoietic progenitors was not sufficient to induce hyperplasia. Rather, hematopoietic hyperplasia may result indirectly from infection of some other cell type.     

Sequences responsible for erythroid and lymphoid leukemia in the long terminal repeats of Friend-mink cell focus-forming and Moloney murine leukemia viruses.

Despite the high degree of homology (91%) between the nucleotide sequences of the Friend-mink cell focus-forming (MCF) and the Moloney murine leukemia virus (MuLV) genomic long terminal repeats (LTRs), the pathogenicities determined by the LTR sequences of the two viruses are quite different. Friend-MCF MuLV is an erythroid leukemia virus, and Moloney MuLV is a lymphoid leukemia virus. To map the LTR sequences responsible for the different disease specificities, we constructed nine viruses with LTRs recombinant between the Friend-MCF and Moloney MuLVs. Analysis of the leukemia induced with the recombinant viruses showed that a 195-base-pair nucleotide sequence, including a 75-base-pair nucleotide Moloney enhancer, is responsible for the tissue-specific leukemogenicity of Moloney MuLV. However, not only the enhancer but also its downstream sequences appear to be necessary. The Moloney virus enhancer and its downstream sequence exerted a dominant effect over that of the Friend-MCF virus, but the enhancer sequence alone did not. The results that three of the nine recombinant viruses induced both erythroid and lymphoid leukemias supported the hypothesis that multiple viral genetic determinants control both the ability to cause leukemia and the type of leukemia induced.     

Characterization of a progressive neurodegenerative disease induced by a temperature-sensitive Moloney murine leukemia virus infection.

A progressive neurodegenerative disease occurred following infection of mice with a temperature-sensitive (ts) isolate of Moloney (Mo) murine leukemia virus (MuLV), ts Mo BA-1 MuLV. This NB-tropic ecotropic MuLV, which was ts for a late function, induced a syndrome of tremor, weakness of the hind limbs, and spasticity following infection of several strains of laboratory neonatal mice, including NFS, C3H/He, CBA, SJL, and BALB/c. The latent period of 8 to 16 weeks was considerably longer than that observed for the acute paralytic diseases observed following neonatal infection with other ts Mo-MuLV, rat-passaged Friend MuLV, and some wild mouse ecotropic MuLVs. Spongiform pathology without inflammation and degeneration of neurons devoid of budding virions occurred in the cerebellar grey matter, brain stem, and upper spinal cord; but lower spinal cord anterior horn cells were less obviously affected than in other MuLV-associated neuroparalytic syndromes. ts Mo BA-1 MuLV differed from other ts Mo-MuLV mutants that are capable of inducing a neuroparalytic syndrome in that while infected nervous system tissue contained high levels of MuLV p30 and gp70, no evidence of precursor accumulation or abnormal processing of MuLV p30 or gp70 could be demonstrated. The localization of virus within the nervous system suggests that direct neuronal infection may not be the etiologic mechanism in this MuLV-induced neurodegenerative disease.     

Secondary structural features in the 70S RNAs of Moloney murine leukemia and Rous sarcoma viruses as observed by electron microscopy.

The secondary structural features in the 70S RNAs of the Prague strain of avian Rous sarcoma virus, subgroup A (PR-RSV-A), and Moloney murine leukemia virus (M-MuLV) were compared by electron microscopy. The PR-RSV-A genome contained two subunits joined by a linkage structure as in the genomes of M-MuLV and other mammalian retroviruses. In both viral genomes, a highly reproducible hairpin occurred at about 70 nucleotides from the 5' end of each subunit and contained 320 +/- 8 nucleotides. The stable point of linkage between the subunits in both viral genomes involved fewer than 50 nucleotides and occurred at 466 +/- 9 nucleotides from the 5' end. This places the linkage about 350 nucleotides further toward the 3' end of the subunit than the binding site of primer tRNA. Another structural feature common to both genomes was a loop in each subunit. In M-MuLV, the loop contained 3.9 +/- 0.10 kilobases (kb) and occurred at a distance of 2.2 +/- 0.05 kb from the 5' end. In PR-RSV-A, the loop was smaller (2.3 +/- 0.10 kb) and further (3.3 +/- 0.10 kb) from the 5' end. When M-MuLV RNA was heated to 70, 85, or 90 degrees C and cooled, the hairpin consistently reformed at the 5' end. No other structures typical of the native molecules reappeared. In RNA samples heated to 70 degrees C, a new loop reproducibly occurred near the 5' end of each subunit, but this loop was not found in samples heated to higher temperatures. Based on all of these findings, we conclude that the genome of PR-RSV-A shares several features with M-MuLV and other mammalian retroviruses and that the primer tRNA molecules are not involved in the linkage of the two subunits in either genome. We also conclude that the dimer linkage and the loops in subunits are typical of the native molecules and that their formation requires a special environment.