Long-Range Effects of Retroviral Insertion on c-myb: Overexpression May Be Obscured by Silencing during Tumor Growth In Vitro

Abstract

The c-myb oncogene is a frequent target for retroviral activation in hemopoietic tumors of avian and mammalian species. While insertions can target the gene directly, numerous clusters of retroviral insertion sites have been identified which map close to c-myb and outside the transcription unit in T-lymphomas (Ahi-1, fit-1, and Mis-2) and monocytic and myeloid leukemias (Mml1, Mml2, Mml3, and Epi-1). Previous analyses showed no consistent effect of these insertions on c-myb expression, raising the possibility that other nearby genes were the true targets. In contrast, our analysis of four cell lines established from lymphomas bearing insertions at fit-1 (fti-1) (feline leukemia virus) and Ahi-1 (Moloney murine leukemia virus) shows that these display higher expression levels of c-myb RNA and protein compared to a panel of phenotypically similar cell lines lacking such insertions. An interesting feature of the cell lines with long-range c-myb insertions was that each also carried an activated Myc allele. The potential for oncogenic synergy between Myb and Myc in T-cell lymphoma was confirmed in transgenic mice overexpressing alleles of both genes in the T-cell compartment, lending further credence to the case for c-myb as the major target for long-range activation. In contrast, mapping and analysis of c-myb neighboring genes (HBS1 and FLJ20069) showed that the expression of these genes did not correlate well with the presence of proviral insertions. A possible explanation for the paradoxical behavior of c-myb was provided by one of the murine T-lymphoma lines bearing an insertion at Ahi-1 (p/m16i) that reproducibly down-regulated c-myb RNA and protein to very low levels or undetectable levels on prolonged culture. Our observations implicate c-myb as a key target of upstream and downstream retroviral insertions. However, overexpression may become dispensable during outgrowth in vitro, and perhaps during tumor progression in vivo, providing a potential rationale for the previously observed discordance between retroviral insertion and c-myb expression levels.

The c-myb proto-oncogene was first described as the cellular progenitor of v-myb, the transforming element of the avian retroviruses avian myeloblastosis virus and E26, which cause acute myeloblastic leukemia and erythroblastosis, respectively (38). c-Myb is a nuclear transcription factor of 75 kDa which plays a central role in differentiation, proliferation, and apoptosis (49). The normal c-Myb protein is highly expressed in immature hemopoietic cells, and its expression decreases dramatically during differentiation (20). c-myb is also a frequent target for insertional activation in myeloid leukemias and T- and B-cell lymphomas in chickens and mice (38, 50). A large number of Myb target genes have been identified by promoter binding and transactivation assays (30). Although the critical targets are unknown, potentially relevant mediators of its oncogenic effects include c-myc (12, 36, 51) and c-kit (34) and the antiapoptotic gene bcl-2 (15, 41).

In addition to retroviral insertions within the c-myb gene itself, a number of common clustered retroviral insertion sites have been mapped close to c-myb (25 to 200 kb) but outside the transcription unit. The fit-1 (fti-1) locus was originally identified in the cat as a common site of insertion for feline leukemia virus (FeLV) in thymic lymphomas induced by FeLV-myc recombinant viruses (44, 45). A conserved unique sequence from this locus was subsequently shown to map 100 kb upstream of c-myb in human and mouse genomes (2). The common murine leukemia virus (MuLV) insertion loci Ahi-1 and Mis-2 map approximately 35 and 160 kb downstream of c-myb and were identified in Abelson helper virus-induced pre-B lymphomas and Moloney MuLV (MMLV)-induced T-cell lymphomas, respectively (21, 32, 46). The Mml1, Mml2, and Mml3 loci were identified as common integration sites in MuLV-induced promonocytic leukemia and were mapped 25, 51, and 70 kb upstream of c-myb (17, 23). Finally, the Epi1 locus is a recently described common insertion site for somatically acquired retroviral integrations in acute myeloid leukemias of BXH-2 mice and was mapped 30 to 40 kb downstream of c-myb (5), close to, or coincident with, the Ahi-1 site.

Previous analyses of a number of these tumor series revealed no clear correlation with the level of c-myb expression (5, 17, 21, 23, 45, 46), raising the possibility that these insertions target nearby genes rather than c-myb itself. However, with a few exceptions, quantitative analysis of c-myb expression has been limited by the lack of established tumor cell lines bearing insertions at these loci. The lack of correlation might therefore be accounted for by tumor heterogeneity and/or postmortem RNA degradation. In this study we have identified two murine cell lines (p/m16i and G1-500/44i) rearranged at Ahi-1 and two feline cell lines (FT-1 and FTG) rearranged at fit-1, which have allowed careful analysis of the long-range effects on c-myb expression.

Our study supports the hypothesis that c-myb is a key target for transcriptional activation by these long-range insertions and provides a potential rationale for previous conflicting observations.

MATERIALS AND METHODS

Cell lines.

The p/mi cell lines (p/m9i, p/m11i, p/m16i, and p/m23i, etc.) were derived from MMLV-induced thymic lymphomas in p53-null CD2-Myc mice (3). The G1-500/44i cell line, bearing insertions at c-myc and Ahi-1, was derived from an MMLV-induced thymic lymphoma in a CD2-Runx2 transgenic mouse (7) (Table 1). Cell lines were established and maintained as described previously (3). The monocytic tumor cell lines, WII 1-6, 2-2, and 2-10, which expressed almost undetectable levels of c-myb RNA and protein, were derived from MMLV-induced monocytic tumors. The MMLV was transduced with c-myc, and the cells were maintained as described previously (23). The feline T-cell line FT-1 was established from a spontaneous thymic lymphosarcoma in an FeLV-positive cat (27), while the feline T-cell line FTG was derived from a tumor induced by the T3 isolate of FeLV and contains an FeLV-myc recombinant virus (31) (Table 1). The FeLV negative feline T-cell line 3201 was derived from a naturally occurring thymic lymphosarcoma (40). Finally, the FeLV positive lymphoid tumor cell line T3, established from a naturally occurring thymic lymphosarcoma, contains replication-competent FeLV helper and recombinant FeLV provirus with a v-myc oncogene (29).

TABLE 1.

c-myc status of murine and feline lymphoma cell lines rearranged at Ahi-1 and fit-1

Species	Cell line	Insertion near c-myba	Myc statusb	Otherc
Murine	G1-500/44i	Ahi-1	MMLV 5′	CD2-Runx2 Tgd
Murine	p/m16i	Ahi-1	CD2-MYC Tg	p53 null
Feline	FT-1	fit-1	FeLV 5′
Feline	FTG	fit-1	FeLV-myc

^aProviral insertions within close proximity of c-myb—at Ahi-1 (MMLV [murine]) or fit-1 (FeLV [feline]).

^bAltered Myc expression due to proviral insertion (G1-500/44i and FT-1), a CD2-MYC transgene (p/m16i), or a FeLV-myc recombinant virus (FTG).

^cAdditional genetic information.

^dTg, transgene.

Transgenic animals.

CD2-vMyb and CD2-MYC-ER mice have been described previously (1, 6). Heterozygous animals were crossed together to create F₁ offspring heterozygous for both transgenes in addition to littermate controls with single or no transgenes. Animals were monitored and sacrificed when presenting with signs of lymphoma development. Genotypes were determined by Southern blot analysis carried out on tail biopsy specimens as described previously (1, 6).

Nucleic acid extraction.

Total cellular RNA was extracted from cultured cells using RNAzol B (Biogenesis, Bournemouth, United Kingdom), based on the guanidinium thiocyanate-phenol-chloroform extraction method (8). High-molecular-weight DNA was isolated from cultured cells using the Nucleon BACC2 kit (Tepnel Life Sciences, Manchester, United Kingdom).

Mapping and hybridization analysis.

Restriction enzyme digestion of DNA and separation by agarose gel electrophoresis were performed as described previously (29). RNA samples were separated on 1% agarose gels containing 2.2 M formaldehyde, and both RNA and DNA were transferred to Hybond N membrane (Amersham Pharmacia) in 20× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). Blots were hybridized using Rapid-Hyb (Amersham Pharmacia Biotech) and washed and exposed to X-ray film as previously described (29). Typically, blots were washed at high stringency (three times for 20 min in 0.1× SSC-0.5% sodium dodecyl sulfate [SDS] or 0.5× SSC-0.5% SDS, 60°C).

Probes used included a 2.5-kb EcoRI murine c-myb cDNA fragment from the pT7β plasmid (kind gift of Roger Watson, Ludwig Institute, London, United Kingdom) (19), a 1.6-kb NheI-EcoRI murine HBS1 cDNA fragment from the pBKmRFS plasmid (kind gift of Olivier Jean-Jean, Laboratoire de Génétique Moléculaire, Paris, France), a 2.1-kb XhoI human FLJ20069 cDNA fragment from the pME18S-FL3 plasmid (generously provided by Hiroko Hata, University of Tokyo, Tokyo, Japan), a 436-bp EcoRI/XhoI cDNA fragment of the murine orthologue of FLJ20069 (murine image clone 515050; MRC UK HGMP Resource Centre, Cambridge, United Kingdom), from the pBluescript SK plasmid and a rat glyceraldehyde 3-phosphate dehydrogenase (GAPDH) probe, a 750-bp EcoRI fragment digested from the plasmid pGapdh. PCR probes specific for exon 1 and exon 15 of the murine and feline HBS1 genes, exon 9A of the murine and feline c-myb genes, and the murine hypoxanthine phosphoribosyltransferase (HPRT) gene, were derived using the primers illustrated in Table 2. The 329-bp feline fit-1 probe was generated by PCR as described previously (2, 44), while the murine fit-1 probe was isolated by PCR using the degenerate primers zoopr1 and zoopr2 (Table 2) and mouse PAC clone RPC121, 431-G7 as a DNA template (library RPC121, MRC UK HGMP Resource Centre). The 0.8-kb Ahi-1 probe was digested with PstI/HindIII from the plasmid p2-1, and the Mis-2 probe was digested from the plasmid PM02A with EcoRI/SacI (both generously provided by Paul Jolicoeur, Clinical Research Institute of Montreal, Montreal, Canada) (32, 46). Rearrangements of the T-cell receptor (TCR) β-chain gene and the immunoglobulin heavy-chain gene were determined using a 0.7-kb EcoRI/HindIII fragment of the Jβ2 gene, (18) and a 1.7-kb BamHI/EcoRI fragment of plasmid J11 (26), respectively.

TABLE 2.

Primers used in PCR and RT-PCR to amplify probe fragments and to assess levels of FLJ20069 gene expression

Gene (species)	Primer
	5′		3′
	Name	Sequence	Name	Sequence
HBS1 Ex 1 (murine)	Ex1Upa	ATGAGAGAGGTACTGGGGGACG	Ex1DOWNa	CTTTTGAAGGCTGTCCTGC
HBS1 Ex15 (murine and feline)	Ex15UP	GGCCAGAATGCATTGGTAGAGC	Ex15DOWN	CTCAGTGACAACACCAGCAGC
HBS1 Ex1 (feline)	Ex1UPcat	TGCCTTGATCACATGAGAGAGG	Ex1DOWNcat	CTGCACTCTATCTTGTTCCAG
Ex9A c-myb (murine and feline)	Ex9af	TCATCATGGTGTGATCTCAGCA GTTTTG	Ex9ar	CTGCGAGGGAGAGAAGGGTAGGCTT
Fit-1 (murine)	zoopr1	GGTAAATTA/GTGTT/CCC/TGCT CATGC/TACTG	zoopr2	GGCAAAGGGAC/AAA/GGAT/CG/AGA ACATGAGTCAC
HPRT (murine)	HPRT5′	GGGGGCTATAAGTTCTTTGC	HPRT3′	TCCAACACTTCGAGAGGTCC
FLJ20069 (murine)	LHFLJUP	GAGGATGCTGCCATATTAGTGCG	LHFLJDOWN	AGGTGCTATGGATCTTCTCACG

All probe fragments were separated by agarose gel electrophoresis, purified with the QiaQuik gel extraction kit (Qiagen), and radiolabeled by random priming using [α-³²P]dCTP (3,000 Ci/mmol; Amersham Pharmacia) to specific activities of 2 × 10⁸cpm per μg of DNA.

Western blot analysis.

Levels of Myb protein in murine and feline cell lines were investigated by Western blot analysis (43), using an anti-Myb clone 1-1 mouse monoclonal antibody (05-175; Upstate Biotechnology, Lake Placid, N.Y.) and a horseradish peroxidase-conjugated anti-mouse secondary antibody (Sigma). Antibody complexes were visualized by enhanced chemiluminescence (Amersham Pharmacia). β-Actin expression was detected with a goat polyclonal antibody (sc1616; Santa Cruz Biotechnology) and a horseradish peroxidase-conjugated anti-goat secondary antibody (Sigma).

RT-PCR analysis.

Levels of the murine orthologue of FLJ20069 and HPRT transcript expression in murine cell lines were investigated by reverse transcription (RT)-PCR. First strand cDNA was generated from total cellular RNA using cloned murine reverse transcriptase (Amersham Biosciences). For the murine orthologue of FLJ20069, 2 μl of first-strand cDNA reaction mixture was placed in a standard 50-μl PCR mixture containing the primers LHFLJUP and LHFLJDOWN (Table 2), each at 20 μM. PCR was performed using Taq polymerase (Perkin-Elmer) and 30 cycles of 1 min of denaturation at 94°C, 1 min of annealing at 58°C, and 1 min of elongation at 72°C, with a preceding 5-min denaturation step at 94°C. HPRT transcripts were detected using the primers HPRT5′ and HPRT3′ (Table 2) under identical conditions except that 20 cycles of 1 min of denaturation at 94°C, 1 min of annealing at 60°C, and 1 min of elongation at 72°C, were performed.

RESULTS

High levels of c-myb expression in murine and feline cell lines with retroviral insertions at Ahi-1 and fit-1.

We have previously reported the chromosomal mapping of the fit-1 locus to 100 kb upstream of the c-myb gene (2), while the common MuLV insertion locus Ahi-1 is located approximately 35 kb downstream of c-myb (21, 32). Both loci lie within a cluster of common retroviral insertion sites implicated in B- and T-cell lymphoma (Ahi-1, fit-1, and Mis-2) and monocytic and myeloid leukemia (Mml1, Mml2, Mml3, and Epi1) (Fig. 1).

FIG. 1.

(A) Northern analysis of c-myb and c-myb exon 9A expression in murine and feline cell lines rearranged at Ahi-1 and fit-1, respectively, and controls. Total RNA samples (20 μg) were separated on formaldehyde-agarose gels and transferred to Northern blots which were probed with radiolabeled murine c-myb and feline c-myb exon 9A cDNA fragments and a radiolabeled GAPDH fragment, to control for RNA loading and integrity. Lower-stringency washes (0.5 × SSC-0.5% SDS, 60°C) were performed for hybridization to feline RNA. A solid black line indicates cell lines rearranged at Ahi-1 and fit-1. (B) Diagram of clustered retroviral insertion sites close to the c-myb locus in the murine and feline genomes showing the relative locations of fit-1, Ahi-1, and other common insertion sites mapped to this domain. Horizontal arrows above the c-myb gene represent the orientation of the gene, and arrows with angled stems above the insertion sites represent the orientation of the integrated proviruses.

In further test of the hypothesis that c-myb transcription is affected by nearby insertions, steady-state levels of c-myb RNA were examined in the two murine T-lymphoma cell lines carrying insertions at Ahi-1 (p/m16i and G1-500/44i). As shown in Fig. 1A, the levels of c-myb RNA expression in these cell lines were high compared to those in a panel of T lymphoma cell lines of similar derivation but lacking rearrangement at Ahi-1. Two other cell lines, p/m9i and p/m11i, showed similarly elevated c-myb levels. The basis of this overexpression is as yet unexplained, as we found no evidence of proviral insertions or other gross rearrangements using a range of probes derived from the fit-1-Mml1-c-myb-Ahi-1-Epi1-mis-2 cluster. High levels of c-myb RNA were also seen in the two feline lymphoma cell lines with insertions at fit-1 (FT1 and FTG). The levels of c-myb observed here were at least two- to fourfold higher than those in six other feline lymphoid or lymphoma-derived cell lines lacking rearrangements at fit-1 (3201, T3, Mya1, F422, FL74, and FL4), two of which are included in the analysis shown in Fig. 1A (3201 and T3).

Previous studies have shown that c-myb transcripts are differentially spliced (9, 39) and that the murine gene encodes an alternative larger product (89 kDa) due to the use of the alternative exon 9A (11, 33). Intriguingly, we noted a greater differential between tumor cell lines with overexpressed Myb and their apparently normal counterparts when an exon 9A probe was used (Fig. 1A), although these transcripts were in all cases less abundant than the fully spliced mRNA.

Analysis of Myb protein levels showed a close correspondence with steady-state RNA levels. As shown in Fig. 2, the levels of Myb p75 were highest in the Ahi-1-rearranged murine cell lines p/m16i and G1-500/44i, although once again higher levels were noted in p/m11i and p/m9i relative to the other lines. Similarly, higher levels of Myb p75 were observed in the feline cell lines carrying insertions at fit-1 compared to their unrearranged counterparts. An 89-kDa protein, presumably encoded by the exon 9A variant transcripts, was also expressed at higher levels in the same lines, with a significant correspondence between RNA and protein levels.

FIG. 2.

Western blot analysis of Myb expression in murine and feline lymphoma cell lines. Twenty micrograms of protein aliquots was separated on SDS-polyacrylamide gels and subjected to Western blot analysis, using an anti-Myb mouse monoclonal antibody, which detects both the major (p75) and the minor (p89) exon 9A isoform of Myb. β-Actin expression was analyzed as a control for sample loading and integrity. Antibody complexes were visualized by enhanced chemiluminescence. A solid black line indicates the cell lines rearranged at Ahi-1 and fit-1.

Association of long-range insertions at c-myb with deregulated Myc: Myb and Myc collaborate in T-cell lymphomagenesis.

It was of interest to note that all four cell lines with retroviral insertions at sites close to c-myb displayed altered c-myc genes, albeit due to different mechanisms (Table 1). The murine G1-500/44i and feline FT-1 lines have c-myc rearrangements consistent with MMLV and FeLV insertion, respectively (7, 27). The p/m16i lymphoma line was derived from a CD2-MYC transgenic mouse (3), while the FT-G cell line was derived by infection with the T3 FeLV strain, which harbors an FeLV-myc recombinant virus (31).

Further evidence that Myb and Myc may act in concert to promote the development of T-cell lymphoma came from a study of MMLV accelerated tumors of CD2-vMyb mice, in which c-myc was identified as a target for proviral insertional activation (10). To confirm that these genes act in synergy, we examined the effect of combining the CD2-vMyb and CD2-MYC-ER transgenes on tumor development. Although the CD2-MYC-ER model can be induced using tamoxifen, low levels of background activity result in a tumor incidence of 20% by 12 months of age (6). The lower incidence in noninduced CD2-MYC-ER transgenics permits a more sensitive system for the analysis of oncogene cooperation. As shown in Fig. 3, the combination of MYC and Myb was synergistic, with double-transgenic animals developing lymphomas significantly faster than either CD2-vMyb animals or CD2-MYC-ER animals. Consistent with the CD2-directed enhancer element, tumors were confirmed by flow cytometry analysis to be of T-cell origin. In parallel experiments significant collaboration was also observed between a noninducible CD2-MYC model and CD2-vMyb transgenic mice (data not shown).

FIG. 3.

In vivo collaboration of MYC and Myb alleles in T-cell lymphomagenesis, represented by tumor-free survival of CD2-vMyb mice crossed with CD2-MYC-ER transgenic mice. Survival curves are shown for CD2-MYC-ER/CD2-vMyb double-transgenic mice (filled circles) (n = 20), CD2-MYC-ER mice (open circles) (n = 26), CD2-vMyb mice (closed triangles) (n = 23), and nontransgenic littermate controls (open triangles) (n = 26).

Analysis of genes closely linked to c-myb: conserved location of neighboring genes in human, murine, and feline genomes.

It is conceivable that the long-range insertions at c-myb are selected because of their effects on other neighboring genes as well as, or instead of, c-myb. It was therefore important to identify the genes concerned and discover whether they are deregulated by proviral insertion. Examination of the draft human genome sequence released in 2001 demonstrated that the human MYB gene is flanked by two genes lying in the opposite transcriptional orientation (Fig. 4A). The gene located 5′ to human MYB, HBS1, encodes a putative GTP binding protein and is the mammalian orthologue of a yeast gene product with a possible role in protein translation elongation (48). 3′ to MYB is a novel gene designated FLJ20069. While the function of this gene is unknown, homology searches classify the encoded protein in the WD-40 repeat super-family of proteins, which is implicated in a wide variety of growth regulatory functions (28).

FIG. 4.

(A) Diagrammatic structure of the c-myb locus in human, mouse, and cat genomes showing the location of neighboring genes. Retroviral insertion sites which have been physically mapped (feline genome) or located by BLAST searching of the draft genome sequence are also depicted. Horizontal single-headed arrows represent the orientations of the genes, and single-headed arrows with bent stems above the insertion sites represent the favored orientation of the integrated proviruses. The open boxes indicate annotated genes from the public domain human and murine sequences, with vertical lines representing the positions of individual exons. The feline genes are less well characterized and are shown in outline structure only. The AL136797 transcript is the longest corresponding transcript encompassing the annotated FLJ20069 gene, as shown in panel B. The symbol X within the feline genome represents a site where the restriction enzyme XhoI cuts. (B) Detailed diagram of the human FLJ20069 gene and its predicted gene product structure. The portion of the long AL136797 transcript corresponding to the annotated FLJ20069 gene is represented by a double-headed arrow. The orthologous murine sequences corresponding to this gene are currently annotated as two separate elements, Q9CVY1 and ENSMUSG00000037571 (“novel”) (grey box). AL136797 encompasses 25 exons, while the murine orthologues Q9CVY1 and ENSMUSG00000037571 contain 6 and 13 exons, respectively (top). Exons 3 to 7 of AL136797 (human) are homologous to exons 2 to 6 of murine Q9CVY1, while exons 11 to 22 of AL136797 are homologous with exons 1 to 13 of murine ENSMUSG00000037571. Diagonal lines drawn between the human and murine orthologues indicate regions of homology. The dotted line above exons 2 to 5 of murine ENSMUSG00000037571 indicates the area used to design RT-PCR primers for analysis of the expression of the murine orthologue of FLJ20069 (probe FLJ). The solid black line between Q9CVY1 and ENSMUSG00000037571 indicates an additional region of homology identified between the murine genome and exons 7 to 10 of human AL136797. Protein family motifs, including coiled-coil, WD-40, and SH3 domains, are depicted in the predicted amino acid structure of human AL136797 (bottom) and the relative positions of the amino acid matches to murine Q9CVY1 and ENSMUSG00000037571 are indicated by horizontal black lines.

In view of the close linkage of the fit-1 locus to HBS1 in the human genome, it was important to determine whether this feature is conserved in the feline genome where multiple FeLV insertions have been mapped (45). High-molecular-weight feline genomic DNA was subjected to Southern blot analysis using feline fit-1 and HBS1 exon 1 cDNA fragments as probes. This analysis showed that both probes hybridized to the same 30- to 40-kb XhoI fragment, confirming the close linkage and relative orientation of these loci in the human and cat genomes. Similarly, pulsed-field gel analysis (data not shown) of murine PAC clones with the murine HBS1 probe showed that this general organization and linkage map is retained in the rodent (Fig. 4A).

Cross-species detection of the nearest 3′ gene to MYB (FLJ20069) by Southern blot hybridization proved more difficult due to sequence divergence, but the detection of positive fragments in pulsed-field gel analyses of PAC clones overlapping Ahi-1 and the 3′ end of c-myb clearly demonstrated the presence of related sequences (not shown). The relative divergence of this gene compared to c-myb is reflected by the fact that neither human nor mouse cDNA clones hybridized well to feline DNA, and this difficulty limited extension of the comparative genomic map.

The longest cDNA corresponding to the annotated FLJ20069 gene is actually significantly longer (AL136797), encompassing 25 exons and encoding a 1,196-amino-acid protein which includes several clearly defined protein family motifs, including coiled-coil, WD-40, and SH3 domains (Fig. 4B). It appears most likely that the FLJ20069 cDNA represents the 3′ fragment of an alternatively spliced isoform of this larger gene, as its 3′ end is created by an alternative splicing event after exon 19 of AL136797.

The emerging picture from the draft murine genome sequence released in 2002 allowed us to construct a more precise map of the region (Fig. 4). Comparison with the human genome sequence identifies sequences homologous to virtually all of the exons of AL136797. However, the murine transcript map is much less complete, and there is currently no analogous full-length cDNA in the databases. The annotated sequence shows two elements, Q9CVY1 and ENSMUSG00000037571 (“novel”) as separate genes, although these are clearly contained within the domain of homology to AL136797, with matches to exons 3 to 7 and exons 11 to 22 of AL136797, respectively, as shown in Fig. 4B. The predicted amino acid matches are 79 and 87%, respectively. Moreover, BLAST searching using the human gene sequence revealed an additional area of homology between exons 7 to 10 of human AL136797 and the murine genomic sequence between the two annotated genes (represented by a horizontal bar in Fig. 4B). Together, these data suggest that the murine orthologue of the FLJ20069/AL136797 gene is of similar size and organization.

The positions of the respective common insertion loci were also identified by BLAST searching against the murine genome sequence. The Ahi-1/Epi-1 cluster is located in an intragenic region 3′ to c-myb, while the Mis-2 site is within the downstream gene(s).

Expression of genes linked to c-myb: lack of correlation with long-range proviral insertions.

To test whether the murine HBS1 gene is affected by retroviral insertions, a full-length HBS1 cDNA probe was used in Northern blot analysis of total RNA harvested from various cell lines with rearrangements at Ahi-1 and representative control lines. To control for RNA loading and integrity, signals were also quantified and normalized for GAPDH expression levels (data not shown). As can be seen in Fig. 5A, mHBS1 RNA was detectable in all cases, although levels varied significantly. Although p/m16i early passage cells (16iE) had relatively high levels, the G1500/44i cell line did not express particularly high levels, despite the presence of an MMLV insertion in a similar position and orientation. Moreover, the relatively high levels of HBS1 expression in the p/m9i and p/m11i cell lines, which possess no detectable rearrangements at or near the c-myb locus, do not support the hypothesis that HBS1 is an alternative and more important target for proviral insertion at this site.

FIG. 5.

Northern analysis of HBS1 expression in murine (A) and feline (B) lymphoma cell lines, and RT-PCR analysis of the expression of the murine orthologue of FLJ20069 (C). Total RNA samples (20 μg) were separated on formaldehyde-agarose gels and transferred to Northern blots which were probed with radiolabeled full-length murine HBS1 (A) or feline HBS1 exon 1 (B) cDNA fragments and GAPDH to control for RNA loading and integrity. For RT-PCR, 10 μl of each reaction mixture was separated by gel electrophoresis, transferred to a Southern blot, and probed with a radiolabeled murine FLJ20069 cDNA fragment (C). To control for sample integrity, cDNAs were also analyzed by RT-PCR with HPRT primers and probed with a radiolabeled murine HPRT fragment. 16iE and 16iL represent early- and late-passage p/m16i cells, respectively. A solid black line indicates cell lines rearranged at Ahi-1 and fit-1.

The feline cell lines rearranged at fit-1 (FT-1 and FTG) were found to express levels of HBS1 RNA similar to or less than those of the controls (Fig. 5B). This observation further suggests that HBS1 expression is not the major target for activation by retroviral insertions at fit-1, despite the fact that this common insertion site is significantly closer to HBS1 than to c-myb (Fig. 4A).

The expression levels of the murine orthologue of the FLJ20069 gene were too low to allow detection by Northern blot hybridization. We therefore used a semiquantitative RT-PCR analysis to compare levels. The analysis shown in Fig. 5C revealed no clear correlation between the expression levels of the murine orthologue of FLJ20069 and the expression levels of c-myb or the presence of proviral insertions at Ahi-1. Once again, the G1500/44i cell line expressed only low levels of this gene, while the highest levels were noted in the p/m9i, p/m11i, and p/m23i lines, which have no known rearrangement affecting the c-myb locus.

c-myb expression is down-regulated during in vitro passage of the murine cell line p/m16i, despite retention of a rearranged Ahi-1 locus.

During this study it was noted that the murine lymphoma cell line p/m16i, rearranged at Ahi-1, did not behave consistently. On closer examination it was discovered that while early passages (passage 1 and 16iE) of p/m16i cells overexpressed c-myb RNA and protein, later passages (passage 4 and 16iL) had down-regulated c-myb to very low levels.

This phenomenon was intriguing in light of previous analyses of murine monocytic cell lines rearranged at Mml1, some of which do not express detectable levels of c-myb (17). In the analyses shown in Fig. 6, Myb levels were compared in p/m16i cells at various stages of in vitro passage. As can be seen in Fig. 6A, steady-state levels of c-myb RNA declined with passage of p/m16i cells to very low levels. Similarly, loss of protein expression was marked, declining to undetectable levels in late passages (Fig. 6B).

FIG. 6.

Down-regulation of c-myb with passage of the p/m16i cell line. The monocytic tumor cell lines WII 1-6, 2-2, and 2-10, infected with a MMLV transduced with c-myc and expressing almost undetectable levels of c-myb RNA and protein, were also included as negative controls. (A) Twenty micrograms of total RNA samples were separated on formaldehyde-agarose gels, transferred by the Northern blotting procedure, and probed with radiolabeled murine c-myb and GAPDH fragments. p/m16i passages one to four represent earliest- to latest-passage p/m16i cells. (B) Twenty micrograms of protein from each cell lysate was investigated by Western blot analysis, using an anti-Myb mouse monoclonal antibody and an anti-β-actin antibody. Antibody complexes were visualized by enhanced chemiluminescence. The solid black lines indicate cell lines rearranged at Ahi-1. (C and D) Southern blot analysis to assess the clonality of early- and late-passage p/m16i cells. Twenty micrograms of HindIII and EcoRV digested DNA from early- and late-passage p/m16i cells were separated by agarose gel electrophoresis, transferred to Southern blots, and probed with TCR Jβ2 (C) and Ahi-1 (D) fragments, respectively. Abbreviations: G, unrearranged germ line allele; R, rearranged allele of the clonal 16i cell line.

To rule out the possibility that a distinct transformed clone had overgrown the original cell population, the clonal identity of later passages of p/m16i cells was confirmed by Southern blot analysis. As can be seen in Fig. 6C, the unique TCR Jβ rearrangements (R1 and R2) that characterize this cell line were unaffected at late passage. A further possibility was that the cell clone was genetically unstable and lost the chromosome carrying the Ahi-1 insertion during passage. This possibility was excluded by Southern blot analysis with the Ahi-1 probe that showed the continuing presence of the rearranged allele (R) in late-passage cells (Fig. 6D). Furthermore, analysis of viral expression with a U3 long terminal repeat (LTR) probe showed that general LTR activity was not impaired with passage, as the levels of viral RNA were unchanged (not shown).

Taken together, these results suggest that the loss of expression of c-myb during passage of p/m16i cells is due to altered regulation of the locus itself rather than chromosomal loss or global repression of viral LTR function. An interesting feature of the expression of the genes flanking c-myb is that both of these are down-regulated in late-passage p/m16i cells (Fig. 5A and C), suggesting that the phenomenon of down-regulation of c-myb affects a wider chromatin domain surrounding the gene.

DISCUSSION

Our analysis of a series of murine and feline T-lymphoma cell lines rearranged at Ahi-1 (MMLV) and fit-1 (FeLV) demonstrated that these cell lines overexpress c-myb compared to phenotypically similar cell lines lacking detectable rearrangements at the c-myb locus. This contrasts with previous studies of this locus (5, 17, 21, 23, 45, 46) that failed to demonstrate a consistent effect on c-myb expression arising from nearby proviral insertions and thus raised the possibility that these insertions targeted nearby genes rather than c-myb itself. Our observation that c-myb expression is dispensable for sustained tumor cell proliferation and can be down-regulated in vitro provides a rationale for these discordant reports. However, we also found that some lymphoma lines with no detectable proviral insertion near c-myb also displayed high levels of expression. The basis of overexpression is unknown in these cases, but elevated c-myb expression has been attributed to other mechanisms, such as mutational inactivation of the transcriptional attenuator region (42). Previous failures to correlate long-range proviral insertions with c-myb deregulation could therefore be explained by these two confounding phenomena.

A precedent for long-range activation via retroviral insertion is provided by the Gfi-1-Pal-1-Evi-5-Eis-1 cluster on mouse chromosome 5, where insertions at Gfi-1 or within 50 kb of its surrounding loci activate Gfi-1 expression in B- and T-cell lymphomas (35, 37, 52). Furthermore, long-range cis-activation has been observed at the c-myc gene, where insertions within a 300-kb domain, at the Mlvi-1 and Mlvi-4 loci, appear to activate c-myc expression (25). A further argument in favor of c-myb as the target of long-range effects of proviral insertions as distant as 100 kb upstream (fit-1) or 35 kb downstream is that these are frequently associated with activation of Myc (Table 1) (45), an oncogene which collaborates strongly with Myb in T-cell lymphomagenesis (this study and reference 10).

Perhaps the most intriguing observation of this study is the behavior of the murine cell line p/m16i, which carries a proviral insertion at Ahi-1. Although early passages of this cell line were found to overexpress c-myb RNA and protein, later passages had down-regulated c-myb to almost undetectable levels. Importantly, this phenomenon was repeatable, as Myb down-regulation was observed in two separate experiments involving prolonged culture of p/m16i cells and has been seen independently in a monocytic tumor cell line carrying an insertion at Mml1 (L. Wolff, unpublished data). It appears, therefore, that the growth advantage conferred by elevated c-myb levels can be lost during prolonged in vitro culture. It is possible that with time other genetic lesions accumulated that negated the requirement for c-myb. An important related question is whether loss of c-myb expression can also occur in vivo during the course of tumor development. An intriguing parallel may be drawn to a previous study on MMLV infection of BALB/c and NIH mice, in which c-myb rearrangements were detected in lymphoid tissues at 4 weeks postinfection but were absent from the mature T-cell lymphomas (4). This observation suggests that MMLV insertions directly into c-myb occur in vivo in early proliferative lesions but are not selected in the mature lymphoma cell. This could be due to a lack of requirement for Myb in the mature tumor cell or a positive selection against high Myb levels at this stage. The latter explanation could also account for the selection for apparently less potent activating insertions at the nearby clustered sites rather than in c-myb itself.

The expression of c-myb in murine lymphomas was paralleled to a limited extent by the nearest 5′ gene, HBS1, whose product may have a role in the regulation of protein translation (48). This did not appear to hold, however, for feline lymphomas with insertions at fit-1, despite the close proximity of this locus to HBS1. It is notable that the FeLV insertions that have been mapped at fit-1 have with one exception been oriented away from c-myb, in the expected direction for enhancer-mediated activation of this gene (24). Moreover, we did not find any examples of insertions that up-regulate HBS1 without affecting c-myb. Our observations are in accord with a recent study examining chromosomal aberrations in pancreatic cancer tissues (47) that identified a high-copy-number amplification on 6q and showed that both c-myb and HBS1 were within the amplified unit. However, HBS1 was found to be overexpressed only where it was amplified, in contrast to c-myb, which was found to be overexpressed in the majority of pancreatic cancer tissues and cell lines examined. This observation, and the finding that HBS1 was located in the breakpoint of the amplification in one sample, suggests that c-myb is the most likely target for the amplification at 6q24 and that HBS1 may simply be coamplified. It is also possible that retroviral insertions at this locus can alter the transcriptional activity of a number of genes within the general vicinity. Thus, alterations in HBS1 expression patterns would not be the result of selection events during tumorigenesis but would be the consequence of coordinated gene regulation at this locus. In this scenario long-range activation of c-myb would represent the oncogenic lesion, with HBS1 acting as a neutral marker.

Similar observations were made with respect to the nearest gene 3′ to c-myb, the murine orthologue of FLJ20069, whose expression was not well correlated with the presence of proviruses at Ahi-1/Epi-1. Once again, the favored transcriptional orientation of the proviruses at Ahi-1 (downstream to c-myb and in the same transcriptional orientation) points to c-myb as the more likely candidate for insertional activation (22). However, in view of the complex structure and emerging evidence of alternative splicing in the human equivalent of this gene, further studies are merited to address the oncogenic significance of this gene and to address the possibility of subtle regulatory effects arising from proviruses in the antisense orientation. The product of the AL136797/FLJ20069 gene is a WD-40 repeat protein, whose members include regulators of cell signaling, differentiation, and apoptosis (28). A “two-hit” model might account for the frequent selection of tumors with 3′ insertions that appear to be less potent with respect to Myb activation.

A key conclusion of this study is that care is required when screening for novel oncogenic targets by retroviral gene tagging (22). While transformed cell lines offer an enriched and renewable source of material to examine the consequences of insertions, it is clear that such lines can diverge from the original tumor due to the acquisition of additional lesions in vitro. Moreover, the ability of the tumor cell lines such as p/m16i to proliferate despite down-regulation of the rearranged c-myb allele contrasts with the behavior of other dominant oncogenes, such as Myc and Ras, which have been shown to be vital for the maintenance of the transformed state (13, 14). This phenomenon may have practical consequences for the efficacy of therapeutic strategies based on antisense targeting of Myb expression in hemopoietic tumors (16).