Multiple sequence elements in the U3 region of the leukemogenic murine retrovirus SL3-2 contribute to cell-dependent gene expression

Change of tropism of SL3-2 murine leukemia virus, using random mutational libraries

SL3-2 is a polytropic murine leukemia virus with a limited species tropism. We cloned the envelope gene of this virus, inserted it into a bicistronic vector, and found that the envelope protein differs from other, similar envelope proteins that also utilize the polytropic receptor (Xpr1) in that it is severely impaired in mediating infection of human and mink cells. We found that two adjacent amino acid mutations (G212R and I213T), located in a previously functionally uncharacterized segment of the surface subunit, are responsible for the restricted tropism of the SL3-2 wild-type envelope. By selection from a two-codon library, several hydrophobic amino acids at these positions were found to enable the SL3-2 envelope to infect human TE 671 cells. In particular, an M212/V213 mutant had a titer at least 6 orders of magnitude higher than that of the wild-type envelope for human TE 671 cells and infected human, mink, and murine cells with equal efficiencies. Notably, these two amino acids are not found at homologous positions in known murine leukemia virus isolates. Functional analysis and library selection were done on the basis of sequence and tropism analyses of the SL3-2 envelope gene. Similar approaches may be valuable in the design and optimization of retroviral envelopes with altered tropisms for biotechnological purposes.

B-Cell lymphoma induction by akv murine leukemia viruses harboring one or both copies of the tandem repeat in the U3 enhancer

Akv is an endogenous, ecotropic murine leukemia virus (MuLV) of the AKR strain. It has served as a prototype nonpathogenic or weakly pathogenic reference virus for studies of closely related potent lymphomagenic viruses such as the T-lymphomagenic SL3-3. We here report that Akv and an Akv mutant (Akv1-99) with only one copy of the 99-bp transcriptional enhancer induce malignant lymphomas with nearly 100% incidence and mean latency periods of 12 months after injection into newborn NMRI mice. Molecular analysis of tumor DNA showed that the majority of the tumors were of the B-cell type. Sequence analysis of proviral transcriptional enhancers in DNA of B-cell lymphomas revealed conservation of the enhancer sequence, as well as a lack of sequence duplications of the Akv1-99 variant, while the repeat copy number in Akv was subject to fluctuations. In support of a B-cell specificity of the Akv enhancer, a murine plasmacytoma cell line was found to sustain three- to fivefold-higher transient transcriptional activity upon the Akv and Akv1-99 enhancers than upon the enhancer of the T-lymphomagenic SL3-3 MuLV. Thus, the overall picture is that Akv MuLV possesses a B- lymphomagenic potential and that the second copy of the 99-bp sequence seems to be of minor importance for this potential. However, in one animal the lymphomas induced by Akv1-99 were of the T-cell type. Among the 24 tumors analyzed only this one harbored a clonal proviral integration in the c-myc locus. This provirus had undergone a duplication of a 113-bp sequence of the enhancer region, partly overlapping with the 99-bp repeat of Akv, as well as a few single nucleotide alterations within and outside the repeats. Taken together with previous studies, our results suggest that T- versus B-lymphomagenic specificity of the enhancer is governed by more than one nucleotide difference and that alterations in binding sites for transcription factors of the AML1 and nuclear-factor-1 families may contribute to this specificity.

CBF, Myb, and Ets binding sites are important for activity of the core I element of the murine retrovirus SL3-3 in T lymphocytes

Transcriptional enhancers within the long terminal repeats of murine leukemia viruses are major determinants of the pathogenic properties of these viruses. Mutations were introduced into the adjacent binding sites for three transcription factors within the enhancer of the T-cell-lymphomagenic virus SL3-3. The sites that were tested were, in 5'-to-3' order, a binding site for core binding factor (CBF) called core II, a binding site for c-Myb, a site that binds members of the Ets family of factors, and a second CBF binding site called core I. Mutation of each site individually reduced transcriptional activity in T lymphocytes. However, mutation of the Myb and core I binding sites had larger effects than mutation of the Ets or core II site. The relative effects on transcription in T cells paralleled the effects of the same mutations on viral lymphomagenicity, consistent with the idea that the role of these sequences in viral lymphomagenicity is indeed to regulate transcription in T cells. Mutations were also introduced simultaneously into multiple sites in the SL3-3 enhancer. The inhibitory effects of these mutations indicated that the transcription factor in T cells that recognizes the core I element of SL3-3, presumably CBF, needed to synergize with one or more factors bound at the upstream sites to function. This was tested further by generating a multimer construct that contained five tandem core I elements linked to a basal long terminal repeat promoter. This construct was inactive in T cells. However, transcriptional activity was detected with a multimer construct in which the transcription factor binding sites upstream of the core were also present. These results are consistent with the hypothesis that CBF requires heterologous transcription factors bound at nearby sites to function in T cells.

The role of proximal and distal sequence variations in the presentation of an immunodominant CTL epitope encoded by the ecotropic AK7 MuLV

An emv-14-derived, replication-competent ecotropic murine leukemia virus [MuLV], designated AK7, was previously cloned from the AKXL-5 recombinant inbred mouse strain and partially characterized. While genetically encoding for an envelope-derived immunodominant CTL epitope [KSPWFTTL] located in the transmembrane region of p15TM, this virus, unlike the emv-11-derived virus AKR623, fails to be efficiently recognized by AKR/Gross MuLV-specific cytotoxic T lymphocytes [CTL]. AK7 thus provides the opportunity to study the role of retroviral sequence variations that are located outside of the immunodominant epitope as a mechanism of escape from CTL-mediated immune surveillance. In an attempt to identify which region[s] of the AK7 genome could account for its ability to evade efficient recognition by AKR/Gross MuLV-specific CTL, we have constructed recombinant murine retroviruses. The direct influence of a sequence variation twelve amino acids N-terminal to KSPWFTTL was explored with the use of chimeric viruses and determined not to significantly impair the presentation of KSPWFTTL to AKR/Gross MuLV-specific CTL. The long terminal repeat [LTR] derived from the AK7 virus, which possesses only one copy of the 99-base pair transcriptional enhancer in the U3 region, in contrast to AKR623 that possesses two copies of the tandem direct repeat enhancers, was also analyzed for its influence on the presentation of KSPWFTTL. Interestingly, our data indicate that the enhancer region derived from AK7 negatively influences the presentation of KSPWFTTL in the context of a recombinant AKR623 virus.

E- vectors: development of novel self-inactivating and self-activating retroviral vectors for safer gene therapy

We have developed novel self-inactivating and self-activating retroviral vectors based on the previously observed high-frequency deletion of direct repeats. We constructed spleen necrosis virus (SNV)-based viral vectors that contained large direct repeats flanking the viral encapsidation sequence (E). A large proportion of the proviruses in the target cells had E and one copy of the direct repeat deleted. Direct repeats of 1,333 and 788 bp were deleted at frequencies of 93 and 85%, respectively. To achieve a 100% deletion efficiency in target cells after ex vivo infection and drug selection, we constructed a self-activating vector that simultaneously deleted E and reconstituted the neomycin phosphotransferase gene. Selection of the target cells for resistance to G418 (a neomycin analog) ensured that all integrated proviruses had E deleted. The proviruses with E deleted were mobilized by a replication-competent virus 267,000-fold less efficiently than proviruses with E. We named these self-inactivating vectors E- (E-minus) vectors. These vectors should increase the safety of retroviral vector-mediated gene therapy by preventing the spread of vector sequences to nontarget cells in the event of coinfection with helper virus. We propose that direct-repeat deletions occur during RNA-dependent DNA synthesis and suggest that template switches occur without a requirement for RNA breaks. The minimum template dissociation frequency was estimated as 8%/100 bp per replication cycle. These vectors demonstrate that large direct repeats and template-switching properties of reverse transcriptase can be utilized to delete any sequence or reconstitute genes during retroviral replication.

Inducible and cell type-specific expression of VL30 U3 subgroups correlate with their enhancer design

The murine VL30 elements constitute one family of retrotransposons represented in 100 to 200 copies that are dispersed among the mouse chromosomes. On the basis of sequence homology, we have subdivided mouse VL30 members into four distinct U3 subgroups. The use of subgroup-specific probes in Northern (RNA) blot analyses shows that individual VL30 U3 subgroups are expressed in a tissue-specific manner. We show by in situ hybridization of mouse skin treated with 12-O-tetradecanoylphorbol-13-acetate (TPA) that VL30 expression is induced in epidermal keratinocytes but not in dermal fibroblasts. Transient transfections of reporter gene plasmids together with in vitro binding analysis indicate that TPA-induced VL30 transcription specific for keratinocytes is mediated by two cooperating sequence motifs in juxtaposed position. One sequence motif is shown to constitutively bind CREB- and Jun-related proteins in both keratinocytes and fibroblasts, whereas the other is a target for TPA-induced c-Rel/p65(NF-kappa B)-binding activity specifically in keratinocytes. These binding sites are found to be conserved within U3 subgroups and individual U3 regions showing induced expression in TPA-treated mouse epidermis. These results together with a sequence comparison between different U3 subgroups indicate that cell type-specific activity of transcription factors known to regulate VL30 transcription and the presence or absence of their cognate binding sites within individual U3 regions determine inducible and cell type-specific VL30 expression. The variable VL30 U3 regions might thus be useful tools to study inducible and cell type-specific transcription in many different cell systems.

A correlation between dexamethasone inducibility and basal expression levels of retroviral vector proviruses

Identical transcription units inserted at different positions of mammalian chromosomes may vary widely in transcriptional activity. We have used a set of ten cell clones with random unselected single integrations of retroviral vectors to study such position effects. The vector used carries a neo gene driven by the Akv murine leukemia virus long terminal repeat that has only a weak promoter-enhancer activity in the target cell, the lymphoid cell line L691. Under transient expression conditions, the strength of the Akv promoter-enhancer in the L691 cells is increased by dexamethasone. In cell clones with single vector integrations, a correlation is observed between the non-induced expression levels and the degree of dexamethasone induction. The strongest relative induction is found for the integrated vectors with the lowest non-induced expression levels and approaches the inducibility under transient expression. These results indicate that expression levels are composed of distinct contributions from the integrated vector and from the site of integration and are best explained in terms of a model in which the sites of chromosomal integration exert variable positive enhancer effects upon vector transcription.

Contributions to transcriptional activity and to viral leukemogenicity made by sequences within and downstream of the MCF13 murine leukemia virus enhancer

We have identified nucleotide sequences that regulate transcription in both a cell-type-specific and general manner in the long terminal repeat of the MCF13 murine leukemia virus. Besides the enhancer element, we have observed that the region between the enhancer and promoter (DEN) has a profound effect on transcription in different cell types. This effect, however, was dependent on the copy number of enhancer repeats and was detectable in the presence of a single repeat. When two enhancer repeats were present, the effect of DEN on transcription was abrogated except in T cells. DEN also makes a significant contribution to the leukemogenic property of the MCF13 retrovirus. Its deletion from the MCF13 virus dramatically reduced the incidence of thymic lymphoma and increased the latency of disease in comparison with the wild-type virus. This effect was most marked when one rather than two enhancer repeats was present in the mutant viruses. We also observed that the removal of one repeat alone remarkably reduced leukemogenicity by the MCF13 virus. A newly identified protein-binding site (MLPal) located within DEN affects transcription only in T cells, and its deletion attenuates the ability of an MCF13 virus with a single enhancer repeat to induce thymic lymphoma. This observation suggests that the MLPal protein-binding site contributes to the effect of the DEN region on T-cell-specific transcription and viral leukemogenicity. This study identifies the importance of nonenhancer sequences in the long terminal repeat for the oncogenesis of the MCF13 retrovirus.

Molecular cloning of osteoma-inducing replication-competent murine leukemia viruses from the RFB osteoma virus stock

We report the molecular cloning of two replication-competent osteoma-inducing murine leukemia viruses from the RFB osteoma virus stock (M. P. Finkel, C. A. Reilly, Jr., B. O. Biskis, and I. L. Greco, p. 353-366, in C. H. G. Price and F. G. M. Ross, ed., Bone--Certain Aspects of Neoplasia, 1973). Like the original RFB osteoma virus stock, viruses derived from the molecular RFB clones induced multiple osteomas in mice of the CBA/Ca strain. The cloned RFB viruses were indistinguishable by restriction enzyme analysis and by nucleotide sequence analysis of their long-terminal-repeat regions and showed close relatedness to the Akv murine leukemia virus.

Analysis of the significance of two single-base-pair differences in the SL3-3 and Akv virus long terminal repeats

Two single-base-pair differences between the long terminal repeats (LTRs) of the T-lymphomagenic murine retrovirus SL3-3 and nonleukemogenic Akv virus were tested for effects on activity of the LTRs. Evidence was obtained from electrophoretic mobility shift assays for the presence of at least one factor in both T and non-T cells that bound to the region of the viral enhancers that contained the differences. However, no significant differences in activity in expression assays were detected when the two base-pair differences were exchanged between the two LTRs. Therefore, they do not contribute to the higher activity of the SL3-3 LTR in T-lymphoma cell lines.

Involvement of nuclear factor I-binding sites in control of Akv virus gene expression

The U3 region of Akv murine leukemia virus carries a 99-base-pair repeat that is associated with transcriptional enhancement in murine NIH 3T3 cells. Deletion analysis points to a critical function of a region within the repeat unit related to the recognition sequences for nuclear factor I proteins but distinct from the sites previously analyzed in related viruses. Nuclear proteins binding to the critical site were detected in NIH 3T3 cells and in mouse livers. A protein fraction binding to this site was purified from mouse livers by ion-exchange and DNA affinity chromatography and shown to have nuclear factor I properties. Mutations that caused a partial or complete reduction of the in vitro binding were introduced into an Akv long terminal repeat with one 99-base-pair repeat copy driving a reporter gene, and the expression activities of the mutants in NIH 3T3 cells were found to correspond to their in vitro binding activities. This correlation strongly supports the role of nuclear factor I proteins in Akv expression. Residual expression activity was, however, detected in mutants devoid of in vitro binding. This residual activity may relate to the presence of additional sequences with homology to nuclear factor I binding sites both within and outside the repeat region. The ability of these sites to bind crude and purified protein fractions with nuclear factor I activity was analyzed, and the role of the sites within and outside the repeat region for control of gene expression of Akv and related viruses is discussed.

Enhancer functions in U3 of Akv virus: a role for cooperativity of a tandem repeat unit and its flanking DNA sequences

cis-Acting transcriptional control elements in the U3 region of the murine retrovirus Akv were analyzed in mouse NIH 3T3 fibroblast cells by using a transient expression vector system based upon a complete long terminal repeat with linked flanking sequences. Deletion analysis pointed to the essential role of sequences within the 99-base-pair direct repeats, and a fragment encompassing the two repeats was found to possess orientation-independent enhancer activity when positioned either upstream or downstream of the transcriptional unit. Removal of one copy of the 99-base-pair repeat led to a reduction in activity of about 2.5-fold when located in an intact U3 environment but to reductions of up to 2 orders of magnitude when placed in other sequence contexts. Our studies of enhancer functions in the presence of one versus two copies of the tandem repeat point to duplicate functions of repeat sequences and sequences flanking the repeat region and emphasize the complex overall organization of this U3 region.