Infectivity and insertional mutagenesis of endogenous retrovirus in autoimmune NZB and B/W mice

Abstract

Murine leukaemia virus has been suggested to contribute to both autoimmune disease and leukaemia in the NZB mouse and in the (NZB × NZW) F1 (abbreviated B/W) mouse. However, with apparently only xenotropic but no ecotropic virus constitutively expressed in these mice, few mechanisms could explain the aetiology of either disease in either mouse strain. Because pseudotyped and/or inducible ecotropic virus may play a role, we surveyed the ability of murine leukaemia virus in NZB, NZW and B/W mice to infect and form a provirus. From the spleen of NZB mice, we isolated circular cDNA of xenotropic and polytropic virus, which indicates ongoing infection by these viruses. From a B/W lymphoma, we isolated and determined the complete sequence of a putative ecotropic NZW virus. From B/W mice, we recovered de novo endogenous retroviral integration sites (tags) from the hyperproliferating cells of the spleen and the peritoneum. The tagged genes seemed to be selected to aid cellular proliferation, as several of them are known cancer genes. The insertions are consistent with the idea that endogenous retrovirus contributes to B-cell hyperproliferation and progression to lymphoma in B/W mice.

Introduction

NZB and (NZB × NZW) F1 (abbreviated B/W) mice suffer not only from autoimmune disease (haemolytic anaemia and lupus, respectively), but also from hyperproliferation of lymphocytes and, eventually, from lymphoma and leukaemia. Male B/W mice succumb to autoimmune disease later than female B/W mice, and thus live long enough to acquire lymphoma. Female B/W mice that have been cured of lupus (Wofsy & Seaman, 1987) do the same. Interestingly, the number of different B-cell clones found in the peritoneum of B/W mice decreases with age, and mono- or biclonality is common by 6 months (Tarlinton et al., 1988). This seems to be driven by antigen (Tarlinton et al., 1988), but may also be helped by tumorigenic mutations.

NZB is a mouse strain with two striking virologic characteristics: high titres of xenotropic murine leukaemia virus (XMV), and high synthesis rate of the envelope protein gp70. To link the virologic, immunological and tumorigenic characteristics of NZB, investigators hypothesized that XMV constitutes an ‘autoimmune virus’ (Levy, 1974). Although this retroviral hypothesis of autoimmune disease in NZB mice generated considerable enthusiasm when XMV was discovered, it was never proved and was later dismissed. This was mainly due to experiments with crosses of the NZB and the SWR mouse strains, in which autoimmunity in the progeny did not depend on the presence of replication-competent virus (Datta et al., 1978a). Meanwhile, endogenous retroelements are known to be adjuvants in the immune response to T-cell-independent type 2 antigens (Zeng et al., 2014), and they are implicated in monogenic (Beck-Engeser et al., 2011; Stetson et al., 2008) and polygenic (‘spontaneous’) (Beck-Engeser et al., 2009) autoimmune disease. Also, importantly, in the NZB × SWR mice, virus titres and high-level expression of NZB-like serum gp70 did correlate with lymphoma incidence (Datta et al., 1978b); but because the prototypic NZB virus is xenotropic, there seemed to be a limited number of effects it could have in lymphomagenesis.

While proliferation may be helped by binding of retroviral envelope protein to Toll-like receptor (TLR)-4 (Rassa et al., 2002), tumorigenesis via insertional mutagenesis by retrovirus requires infection. Generally, four types of retroviruses are distinguished by their host range, which is dictated by the sequence of the envelope (env) glycoprotein gp70: ecotropic (infectious in the mouse; making up less than 10 % of the endogenous retroviruses); xenotropic (XMV; infectious in species other than the mouse); and polytropic (PMV; infectious in the mouse and other species), including modified polytropic (MPMV; polytropic with a deletion in gp70, which may or may not affect host range) (Stoye & Coffin, 1987). The sequences of retroviral origin make up almost 10 % of the mouse genome. The genomic sites of XMVs, PMVs and MPMVs are defined in common inbred strains (Frankel et al., 1990; Jern et al., 2007) and in NZB mice (Frankel et al., 1992). Although few of the endogenous viruses are infectious, and fewer still are replication competent, a large variety of viruses can be produced in individual mice through somatic recombination. The viral diversity could presumably also lead to pseudotyping and phenotypic mixing. Pseudotyping and recombinant viruses may mediate insertional mutagenesis in the NZB mice, which constitutively only express xenotropic but no ecotropic virus.

The germ line-encoded virus sequences vary between mouse strains. Because the genomes of the NZB and NZW mice have not yet been sequenced, the complement of endogenous retroviruses in these genomes is only partially known. The NZB mouse can express two distinct groups of replication-competent xenotropic viruses: the NZB-X1 which is spontaneously produced, and the NZB-X2 which is induced upon treatment of fibroblasts with iododeoxyuridine (Elder et al., 1980). In addition to these xenotropic viruses, the NZW mouse produces an inducible ecotropic virus (Chattopadhyay et al., 1980; Stephenson et al., 1975); however, neither type has been sequenced.

To be able to characterize insertional mutagenesis in these mouse models, we describe here the virus types that are produced by NZB, NZW and B/W mice. We sequenced the gene fragments encoding the env proteins recovered from Mus dunni cells infected with plasma of the various mouse strains and identified three types of xenotropic virus, plus an additional variant xenotropic virus. Because we also recovered polytropic virus from all mice, we investigated whether pseudotyped virus might be infectious to NZB cells. This is indeed the case, and allows for the possibility that insertional mutagenesis generates the lymphomas of NZB mice late in life.

We also determined the complete sequence of a putative ecotropic NZW virus, which we isolated from a lymphoma derived from a B/W mouse. Therefore, we investigated whether insertional mutagenesis plays a role in hyperproliferation and lymphoma formation in B/W mice. Indeed, we found new retroviral integration events in endogenously activated splenic B-cells, in peritoneal B-cells of B/W mice and in the lymphoma from which we had isolated the ecotropic virus mentioned above. This supports the view that murine leukaemia virus (MLV) contributes to B lymphoid hyperplasia in these mice.

Results

MLV gp70 envelope protein subunit expression on lymphocytes

Retroviral env gp70 is encoded as part of a complete retroviral genome(s) (Lerner et al., 1976) and expressed on the cell membrane of lymphoid and epithelial cells of all mouse strains. However, mere expression of a viral protein does not necessarily imply the presence of infectious or replication-competent viral particles. Serum gp70, for example, is produced by the liver of all mouse strains (Hara et al., 1981) as an acute-phase protein (Hara et al., 1982). However, most of the gp70 that is expressed on lymphocytes and macrophages does indeed come from complete viral particles. This can be concluded from the increase in gp70 expression in TLR-7- and RAG-deficient mice (Young et al., 2012; Yu et al., 2012), an effect most easily explained by the spread of replication-competent virus.

To compare viral expression in the NZB, NZW and B/W mice with that in AKR and C57BL/6 mice, we stained cells from brain, thymus, spleen and peritoneum of various mouse strains with the 83A25 antibody, which has broad specificity to gp70. In C57BL/6 mice, no gp70 is expressed in the brain, but is expressed in thymus, peritoneum and, to a smaller extent, spleen (Fig. 1a, b). gp70 expression is much higher in the AKR mouse, which expresses the prototypic leukaemia virus Akv. In AKR, virtually all thymocytes and most splenocytes express MLV, whereas only a fraction of the peritoneal cells do so (Fig. 1a, b). NZB, NZW and B/W mice show gp70 expression patterns similar to those in the AKR mouse (Fig. 1a, b).

Fig. 1.

Expression of MLV gp70 in cells from various mouse strains. Flow cytometry histograms and dot plots of cells from C57BL/6, AKR, NZB, NZW and B/W mice, and an NYC lymphoma. (a) Flow cytometry histograms of gp70 expression in brain, thymus, spleen and peritoneum. y-axis, cell number; x-axis, fluorescence intensity on a logarithmic scale, in arbitrary units. Filled curve, first plus second antibody; empty curve, second antibody only. (b) Quantitative data from flow cytometry as in (a), representing individual mice. y-axis, mean fluorescence intensity; x-axis, mouse strains.

Viruses in plasma of NZB, NZW and B/W mice

Our flow data showed that gp70 is expressed in great abundance on thymus, spleen and peritoneal cells. To characterize infectious or replication-competent viruses, we infected M. dunni cells with the plasma of NZB, NZW and B/W mice. Apart from an endogenous MLV, which is expressed only when induced (Bonham et al., 1997) (and the sequence of which is known), M. dunni cells express no other retroviral sequences and can be infected by xenotropic, polytropic and ecotropic viruses. We isolated DNA from these cells infected with plasma, and amplified and sequenced proviral DNA for the proline-rich region (PRR) of the gp70 env protein of proviral MLVs. This region varies according to virus type, but the flanking amino acid sequences are invariant, so that two (to account for the codon variation) common PCR primer pairs cover the complete spectrum of MLV types. One primer pair covers the ecotropic Akv-type viruses (Akv primers), and the other mainly covers the xenotropic and polytropic viruses (NZB primers). As expected, Akv primers did not amplify any DNA from the supernatant of M. dunni cells infected with plasma from NZB, NZW and B/W mice, whereas the NZB primers yielded a 280 bp band. This may represent xenotropic and polytropic viruses (Fig. 2a). Owing to a 27 bp deletion in the PRR (Stoye & Coffin, 1987), the amplicon of MPMV runs slightly below that of the other viruses. We detected sequence-confirmed MPMV in the DNA from cells infected with plasma from young NOD mice (Fig. 2a), but not in the plasma of NZB, NZW and B/W mice. This indicates that these latter mice do not produce viral particles with genomes encoding env of MPMV.

Fig. 2.

PCR amplicons of MLV gp70 sequences from M. dunni cells infected with supernatants from cell lines and from plasma from various mice. (a) Agarose gels of the 280 bp PCR amplicons derived from the diagnostic PRR of MLV, using primers detecting xenotropic and polytropic, but not ecotropic MLV. AKR, C57BL/6, NZB, NZW and NOD denote DNA from M. dunni cells infected with plasma from the respective mice of different ages. NZB9-1, amplicon from isolated plasmid NZB9-1, which encodes inducible xenotropic virus XMV-1i; LNC, M. dunni infected with LNC lymphoma culture supernatant; NYC, NYC culture supernatant; M. dunni A, DNA from M. dunni cells (mock) infected with M. dunni supernatant; M. dunni B, DNA from non-infected M. dunni cells; M. dunni 9-1, M. dunni cells stably transfected with plasmid NZB9-1. (b) Sequence chromatograms of some of the PRR amplicons shown in (a).

We then sequenced the PRR amplicons without subcloning them, which allowed us to estimate the heterogeneity of viral genomes in the plasma. Fig. 2(b) shows the chromatograms of a segment in which the xenotropic sequences differ most from the polytropic sequences. Virus produced from M. dunni cells containing the proviral xenotropic NZB-X2 genome (NZB9-1) yielded a ‘clean’ chromatogram, whereas the chromatograms from NZB, NZW and B/W mice showed the dominant xenotropic sequence superimposed on polytropic viral sequences. In the B/W mouse, xenotropic and polytropic viruses are present at the same ratio (Fig. 2b). Owing to the additional and predominant expression of an Akv-type virus (see below), the chromatogram of the B/W-derived NYC lymphoma is much more complex (Fig. 2b). To identify the various viruses that infected the M. dunni cells, we sequenced bacterial subclones of amplicons covering the complete env gene region. The infectious viruses that were recovered as proviruses are listed in Table 1.

Table 1. Virus types determined from sequences of proviruses integrated in infected M. dunni cells.

M. dunni cells were infected with plasma from NZB mice and supernatant of the LNC lymphoma cell line from NZB; with plasma from NZW and B/W mice; and from supernatant of the NYC lymphoma derived from a B/W mouse. XMV-1c, Constitutively expressed Xeno-I (X1) from NZB; XMV-1i, inducible Xeno-I (X2) from NZB. PMV, Polytropic virus; MPMV, modified polytropic virus; NZW-1, MLV isolated from NYC lymphoma of B/W mice. Numbers in the table body represent the number of isolated gp70 sequences encoded by the respective virus type. –, Not found; ND, not done.

Mouse strain	Virus type
	XMV	XMV-1c	XMV-1i	XMV-2	XMV-3	XMV-4	XMV-5	PMV	MPMV	NZW-1
NZB 8 mo	11	–	11	0	0	0	0	1	0	0
NZB 14 mo	6	–	6	0	0	0	0	–	0	0
LNC	23	1	21	0	1	0	0	0	0	0
NZW 1.5 mo	8	–	7	0	0	0	1	4	0	0
NZW 6 mo	ND	ND	ND	ND	ND	ND	ND	ND	ND	ND
B/W 3 mo	7	5	2	0	0	0	0	4	0	0
B/W 12 mo	5	–	4	0	0	0	1	0	0	0
NYC	3	1	0	0	0	0	2	0	0	11

Viruses in NZB mice

The plasma of NZB contained xenotropic and polytropic but not ecotropic (Jenkins et al., 1982) virus (Table 1). We set out to differentiate the xenotropic viruses according to Lamont et al. (1991) and displayed in Baudino et al. (2008) into four subgroups, Xeno-I through Xeno-IV. However, the original grouping was done with viruses derived from an AKR cell line, and the sequences displayed in Baudino et al. (2008) were from three different strains: the sequences of Xeno-I (comprising viruses X-1 and X-2, which differ at the C terminus of gp70) are from NZB mice; Xeno-II, Xeno-III and Xeno-IV are from C57BL/6; and PMV and MPMV are from newborn HRS/J mice. Therefore, the sequences determined in our study differ from those in other publications. For example, unlike the gp70 sequence of the prototypic polytropic virus MX27 from HSR/J mice, which starts with an isoleucine codon (Stoye & Coffin, 1987), the gp70 sequence of the NZB mice starts with a methionine codon and thus has a perfect ORF (Gourley et al., 1994). For the purpose of this paper, we have kept the nomenclature as displayed in Baudino et al. (2008), but have denoted them with Arabic numerals rather than roman ones, i.e. xenotropic viruses XMV-1 through XMV-5. Because we felt that Xeno-I X-2 virus is easily confused with Xeno-II virus, we denoted Xeno-I X-2 as XMV-1i (for inducible). For the same reason, we call Xeno-I X-1, which is expressed constitutively, XMV-1c (for constitutive). The culture supernatant from the NZB-derived lymphoma LNC (Peng et al., 1994) expressed predominantly the inducible XMV-1i (Table 1; GenBank accession number KM282107), although XMV-1c is the main and constitutively expressed virus in the NZB mouse (Elder et al., 1980). Among 23 env sequences there was only one sequence each of XMV-1c and XMV-3 (Table 1; GenBank accession numbers KM282106 and KM282108, respectively). We found no XMV-3, XMV-4 or XMV-5 virus (described below; KM282109) in the plasma of NZB; however, the lymphoma LNC produced XMV-3 (Table 1).

Xenotropic MLVs can infect M. dunni cells, and these viruses probably helped the non-replication-competent polytropic viruses that we recovered. However, xenotropic virus cannot help polytropic virus infect NZB cells. Rather, xenotropic virus would need to be pseudotyped with polytropic env to obtain the enabling gp70 sequence. To test this hypothesis, we made use of a by-product of the retroviral replication cycle: a fraction of the linear retroviral cDNA molecule is covalently circularized by host factors – reactions that prevent subsequent integration of the retroelement DNA. In this process, ligation of the two ends forms a circle, in which the two LTRs that flank the linear cDNA are joined (the so-called 2-LTR circles) (Fig. 3a), or homologous recombination between the LTRs results in formation of a 1-LTR circle. While the circular forms are dead-end products, they are a reliable and reproducible measure of recent retroviral infection events. We thus prepared and sequenced the fragment containing the breakpoints (sites of ligation) of retroviral circular DNA from NZB spleen cells. The fact that we could isolate 2-LTR circles indicated that endogenous MLV can infect NZB cells. Because we identified LTR sequences from both xenotropic and polytropic virus (Fig. 3b), we conclude that xenotropic virus is pseudotyped or has recombined with polytropic virus. This would be similar to polytropic MuLV being pseudotyped within ecotropic virions, when cells were co-infected with both types of viruses (Rosenke et al., 2012).

Fig. 3.

(a) Schematic of circularization of late reverse transcription (RT) cDNA product of endogenous MLV. Empty and filled arrows show PCR primers used to amplify the part of circular DNA that contains the site where the two LTR ends of the linear sequence are joined to form a circle. (b) Upper panel, 2-LTR sequences amplified from NZB spleen cells. Sequences are from a 0.5 kb amplicon that contains the 2-LTR junction (X); boxed, inverted repeat sequences; shaded, nucleotide differences; dots indicate nucleotides missing, as compared to prototype sequence. Lower panel, frequencies of U5 sequences with nucleotide replacements as compared to the consensus sequence; dashes indicate nucleotide identity.

Viruses in NZW mice

As with NZB plasma, we recovered polytropic and xenotropic, but no ecotropic virus from NZW plasma. As in NZB, the env of the polytropic virus of NZW starts with a methionine codon and has a perfect ORF. In addition, we could distinguish an XMV variant, named XMV-5, by the HHPPP motif in place of the HPPPS, and the SLVDGAY motif instead of the NLVEGAY motif in the PRR. Also, the hypervariable region VRA of gp70 (GenBank accession number KM282109) was quite different from the other known XMVs.

In the NZW mouse strain an ecotropic virus can be induced (Stephenson et al., 1975; Chattopadhyay et al., 1980) However, this virus has not been well characterized. We cloned a virus from the NYC lymphoma (designated NZW-1) and compared its sequence (GenBank accession number KP087798.1) to that of the ecotropic AKV (Herr, 1984). Altogether there are 12 amino acid exchanges between the two viruses, but within the gp70 there are only three, at positions 33, 45 and 147 (counting from the start methionine; Table S1, available in the online Supplementary Material). If, in analogy to the Moloney MLV gp70, we consider the segment of amino acids 81 to 163 (nt 6026–6271 of the viral RNA genome) to encompass the hypervariable region A (Bae et al., 1997), then only the E in position 147 in NZW-1, instead of the G in AKV, might influence receptor binding. In the PRR of the gp70 (‘diagnostic region’, nt 6574-6842 in Table S1), there was no difference. We therefore consider the NZW-1 virus to be most likely ecotropic. At positions 109 and 110 of gag p30, the virus has amino acids QR, which indicates an N-tropism (DesGroseillers & Jolicoeur, 1983; Kozak & Chakraborti, 1996). NZB and NZW mice were typed Fv1ⁿ (Baliji et al., 2010) and, more specifically, Fv1^nr (W. P. Rowe and J. W. Hartley, cited in Steeves & Lilly, 1977), which restrict some but not all N-tropic viruses (Kozak, 1985; Steeves & Lilly, 1977). Therefore, once induced, NZW-1 may be able to spread in both NZW and B/W mice. However, if it is restricted (Stevens et al., 2004), it may need some mutational/recombinational event to modify the capsid protein.

De novo insertions in B/W spleen cells

In light of likely ongoing retroviral infections in the NZB mouse and also, presumably, in the NZW mouse, MLV may be partially responsible for the hyperproliferation of spleen cells in the B/W mouse. Ecotropic MLV of NZW, or recombinant/pseudotyped MLV from either NZB or NZW, may cause insertional mutagenesis, which would, in turn, drive increased cell division.

In tumours, at least the first transformative event (e.g. a retroviral insertion) is clonal, and therefore the causative mutation is likely to be present in stoichiometric amounts. However, it is not known whether hyperproliferation in the B/W spleen also tends toward monoclonality. Therefore, to recover de novo retroviral insertion sites, we investigated single activated cells. By fusing spleen cells to generate hybridomas, we (i) selected the activated (dividing) cells, since these cells fuse preferentially. Subsequently, we (ii) separated the bulk of activated cells into single cell clones. To identify endogenous MLV integration sites, we analysed 80 hybridomas from the spleens of two 7-month-old, sick, non-immunized B/W mice and found a total of six de novo insertions in five hybridomas (Table 2 and Fig. 4). In addition to the names of the genes, Table 2 lists the location of the insertions in or near the genes and the layout of the insertions. A separate column gives the number of additional insertions at a given locus in lymphomas recovered by us, or in lymphocytic/myeloid tumours reported by others (Akagi et al., 2004; Sokol et al., 2014; Uren et al., 2008). Because the insertions were absent from the BALB/c-generated fusion partner NS0/1, and from the genomes of both the NZB and NZW mice, the insertions likely arose de novo in the B/W spleen cells. However, in the NS0/1 fusion partner, we identified other integration sites, which contributed to the tumorigenesis of the myeloma-derived cells (Knittel et al., 2014). Therefore, we considered the possibility that by generating hybridomas, we were studying subclones of the fusion partner, rather than the spleen cells themselves. These subclones would contain de novo insertions, which, however, should not be recovered in the bulk culture of NS0/1. Of 105 subclones tested, none contained any of the six insertions listed in Table 2.

Table 2. De novo retroviral insertions in proliferating spleen cells (hybridomas), peritoneal cells and lymphoma of B/W mice; and a germline MLV LTR specific to the parental NZW mouse.

‘Location’ gives the chromosome number and the exact position of the insertion, as determined by the UCSC Genome Browser (mouse genome assembly mm9). ‘Layout’ denotes the position of the tag (‘T’) in relation to the gene (‘G’); the transcription orientation of both T and G is given; for example, if the orientation of a tag is from right to left, it is denoted by ‘T − ’; and if the orientation of a gene is from left to right, it is denoted by ‘G+’. In short, the arrangement of tag versus gene in this case will be T − G+. In addition to the tag and gene positions relative to each other, the location of the tag is specified as being in an intron or exon; if outside the transcriptional unit, the distance in bp from the transcription start site is given. ‘Additional inserts’ gives the number of insertions at a given gene, in lymphocytic/myeloid tumours from other studies; insertions up to 50 kb from the transcription start were counted.

Cells	Gene	Location	Layout	Additional inserts
B/W hybridoma
12	Lyzl4	Chr9 : 121461423	T+G −  25538	0
3	Lsamp	Chr16 : 40063012	G+T −  Intron 1	1
3	Jarid1b	Chr1 : 136518121	G+T+ Intron 21	2
52	Rsbn1l	Chr5 : 20422593	G − T −  Intron 3	1
8	Sema4b	Chr7 : 87355314	G+T+ Intron 2	15
9	Lrba	Chr3 : 86328110	G+T −  Intron 27	7
B/W peritoneal cells
297	Isg20l2	Chr3 : 87736045	G+T+ Exon 2	0
247	Trio	Chr15 : 27663180	G − T+ Intron 10	12
243	Popdc3	Chr10 : 45040457	G+T+ 1814	0
4	Nsmaf	Chr4 : 6368259	G − T −  Intron 1	6
B/W lymphoma
NYC	Prex1	Chr2 : 166558404	G − T+ 19122	24
NZW mouse
Germ line		Chr8 : 118995834	T −	0

Fig. 4.

De novo endogenous MLV integration sites in splenic B-cells of B/W mice. Screen prints of a customized version of the UCSC Genome Browser (mouse genome assembly mm9). ‘YourSeq’, Genomic sequence tag recovered in a chimeric sequence containing retrovirus LTR. Numbers at the top show nucleotide position at the particular chromosome. Bright green (PicoSL3) bars indicate insertion sites from our large lymphoma screen; magenta, dark green and brown bars indicate those published by the Retroviral Tagged Cancer Gene Database (RTCGD) (Akagi et al., 2004) and Uren et al. (2008), respectively. Exons are represented by bars, introns by horizontal lines; arrows denote the direction of transcription. (a) Lyzl4; (b) Lsamp; (c) Jarid1b; (d) Rsbn1l; (e) Sema4b; and (f) Lrba locus in the mouse.

Fig. 4 shows a graphic representation of the proviral insertions recovered from the spleen cells (‘YourSeq’). Of these, Lyzl4, Rsbn1l and Lsamp are not known to be cancer genes. However, Lrba has seven insertions in other mouse tumours and thus can be considered a mouse cancer gene. In human patients, mutations in the LRBA gene cause a syndrome of autoimmunity, lymphoproliferation and humoral immune deficiency, by substantially increasing degradation of the immune response inhibitor CTLA-4 (Lo et al., 2015). Sema4b is a known cancer gene. Eleven integrations into the Sema4b locus have been published, and in our lymphoma screen, we found four additional insertions (Fig. 4). Of these insertions, five are 5′ to the transcription start site, four are in intron 1, five are in intron 2, and one is in intron 4. This pattern suggests constitutive expression of the gene. Jarid1b (also known as lysine [K]-specific demethylase 5B) is also a likely cancer gene. It demethylates Lys-4 of histone H3, thereby playing a central role in histone code. Lysine-specific demethylases are known tumour suppressors. We also found an insertion in Fbxl10 (also known as lysine (K)-specific demethylase 2B, human orthologue KDM2B) in a plasmacytoma (Knittel et al., 2014).

De novo insertions in B/W peritoneal cells

The number of different B-cell clones found in the peritoneum of B/W mice decreases with age, and mono- or biclonality of their membrane IgM is common by 6 months. Many clones from different mice show similar rearrangements at both the Ig heavy and light chain loci (Stall et al., 1988). This suggests that the monoclonal peritoneal cells are precursors of the lymphomas suffered by B/W mice. As we had done previously for spleen cell hybridomas, we used anchored PCR to recover proviral insertions in peritoneal cells. Fig. 5 represents these insertions, and Table 2 lists the four de novo MLV integrations recovered from 17 3-month-old B/W mice. All insertions disrupt genes (counting the insertion in Popdc3 as a promoter insertion), which indicates selection of the genetic event in this cell population.

Fig. 5.

De novo endogenous MLV integration sites in peritoneal B-cells of B/W mice. Screen prints of a customized version of the UCSC Genome Browser depicting (a) Popdc3; (b) Trio; (c) Nsmaf; and (d) Isg20l2 locus in the mouse. Layout as in Fig. 4; PicoAKV2, insertions recovered from our (different) large B lymphoma screen; the insertions recovered in this study are marked by ‘YourSeq’.

Two of the four genes, Trio and Nsmaf, were also targeted in tumours in other studies (Table 2). This shows that these genes are common integration sites, and thus can be considered oncogenes. Trio, or triple functional domain (PTPRF interacting) protein, is a rho family GDP/GTP exchange factor. Nsmaf, a.k.a. FAN, or neutral sphingomyelinase activation-associated factor, encodes a WD-repeat protein that binds the cytoplasmic sphingomyelinase activation domain of the 55 kDa TNF receptor. The protein is required for TNF-mediated activation of neutral sphingomyelinase, and may play a role in regulating TNF-induced cellular responses such as inflammation (Montfort et al., 2009). Nsmaf is expressed in B- and T-cells and lymphoid organs (UCSC Genome Browser). Unlike Trio and Nsmaf, there are no reports that Isg20l2 or Popdc3 may act as an oncogene.

Passenger integrations versus drivers of tumorigenesis

Because genes represent only 2 % of the genome and, by and large, MLV integration into the genome is random, integrations are less likely to occur in or near genes than in intervening sequences. However, there is some preference of pseudotyped MLV to integrate into reference sequence (RefSeq) genes in HeLa cells: 34.0 % of the integrations landed in RefSeq genes, while in a set of computer-simulated random integrations, only 22.4 % did (Wu et al., 2003). Although transcription start sites were preferred, the association of integration sites with gene activity was weak (Mitchell et al., 2004). In contrast to integrations in short-term cell culture, where one expects little selection, tumours ex vivo have undergone strong selection for cell growth. In mouse end-stage tumours, irrespective of the distance to transcription start sites, the majority of integrations colocalized with common epigenetic enhancer markers in murine lymphoid tissues (Akagi et al., 2004; Sokol et al., 2014; Uren et al., 2008). Of 6132 tags we analysed previously in T lymphomas induced by exogenous MLV, 46.4 % were in genes. Of 833 tags in B lymphomas, 53.2 % were in genes.

Of the ten insertions shown in Table 2, nine were within genes, defined as the unit spanning 2 kb upstream of the transcription start site to the end of the 3′UTR (i.e. 90.0 %, with central confidence limits of 58.7–97.7 %, at 0.95 confidence). Intriguingly, this percentage of insertions into a gene is even (statistically) higher than the 53.2 % in B lymphomas generated by exogenous MLV. Several mutations in Table 2 also represent common integration sites, as catalogued by the published retroviral tags and our own B and T lymphoma screens. Therefore, we interpret integrations into genes of the B/W cells here as an indication of selection and as drivers of tumorigenesis.

Insertion in a B lymphoma of the B/W mouse

The retroviral insertions in some of the quasimonoclonal populations of peritoneal B-cells in B/W mice suggested that lymphomas from the B/W mice may be driven by insertional mutagenesis. We thus investigated the NYC cell line, which was adapted from a tumour of a B/W mouse that was treated with anti-CD4 antibodies and thus did not develop glomerulonephritis (Wofsy & Seaman, 1987). Because it expresses low levels of the characteristic surface marker CD5 (Fig. 6a), NYC is likely derived from a peritoneal B1 cell. The growth of cells of this line depends on the presence of the antigen receptor, which can be co-precipitated with MLV protein p30 (Jäck et al., 1992). Furthermore, the 2-LTR circles present in these cells indicate continuous MLV infection (Beck-Engeser et al., 2009).

Fig. 6.

Proviral insertion and Prex1 expression in the NYC lymphoma from a B/W mouse. (a) Flow cytometry dot plots displaying expression of IgM, CD5 and gp70 in NYC cells and B/W peritoneal cells. y-axis, side scatter (SSC); x-axis, forward scatter (FSC) and various antibodies: 2nd ab, second antibody alone in series of staining for gp70; IgM, CD5, gp70, stained with antibody to μ chain, CD5 and MLV gp70, respectively. (b) Screen print of a customized version of the UCSC Genome Browser (mouse July 2007 [mm9] genome assembly) depicting the Prex1 locus in the mouse. Display as in Fig. 4. The insertion recovered in this study is marked by ‘YourSeq’. (c) qPCR of MLV (left) and Prex1 (right) mRNA expression from NYC and other cells. The primers complementary to the MLV mRNA hybridize to segments of the reverse transcriptase. NYC, B lymphoma of the B/W mouse; MOPC104E and MOPC315, plasmacytomas adapted to cell culture; P3-X63, Ag8.653 and NS0/1, subclones of culture-adapted MOPC21; F9 cells, mouse teratoma; peritoneal cells, cells recovered from peritoneal cavity lavage; spleen LPS, spleen B lymphocytes activated with LPS and interleukin-4. y-axis, fold change mRNA, normalized to the endogenous control 18S rRNA, and expression level in pancreas set to 1.

In this cell line, we identified an insertion into Prex1 (Fig. 6b). Prex1 is a bona fide proto-oncogene; various other studies list 23 MLV insertions that presumably cause constitutive (over)expression of the gene and, in some cases, truncation. Prex1 functions as a RAC guanine nucleotide exchange factor and is synergistically activated by phosphatidylinositol-3,4,5-triphosphate and the β, γ subunits of heterotrimeric G protein (Welch et al., 2002). According to the UCSC Genome Browser, the protein is strongly expressed in brain, trachea, thymus, bone marrow, and B- and T-cells. Prex1 is also overexpressed in prostate cancer (Knight-Krajewski et al., 2004) and breast cancer (Sosa et al., 2010). In addition, patients whose tumours express Prex1 are more likely to develop metastasis, when compared with those whose tumours do not express Prex1 (Sosa et al., 2010).

De novo integrations ought to increase expression of both MLV and the target gene. Taqman analysis of the steady-state mRNA levels confirmed that NYC expresses large amounts of MLV mRNA – 1 or 2 orders of magnitude more than plasmacytomas or LPS-activated spleen cells, which are known for their high-level expression (Moroni & Schumann, 1975) (Fig. 6c). Stimulation of splenic B-cells also increases expression of Prex1 by an order of magnitude, to the level of that in peritoneal cells (Fig. 6c). NYC cells express more than twice as much Prex1 mRNA than do peritoneal, LPS-activated spleen cells or testicular cells (Fig. 6c). This is likely because Prex1 overexpression in NYC cells drives their proliferation. However, this high expression level is not a general prerequisite for proliferation, as, for example, NIH3T3 cells do not express Prex1, and the strongly proliferating thymocytes express only a little (Fig. 6c).

Discussion

Proliferative disease by insertional mutagenesis due to endogenous retrovirus activity is an established fact and a major cause of spontaneous mouse tumours in general. Integration is potentially mutagenic on two counts: it may disrupt a vital region of the host genome, or bring powerful regulatory functions of the virus to bear on the expression of host genes. Because the prototypic NZB virus is xenotropic, there seemed to be a limited number of effects it could have in lymphomagenesis. However, as we showed here, xenotropic virus replicates in NZB cells – presumably pseudotyped by polytropic virus. Therefore, recombinant virus could be generated and spread. This could happen at a higher frequency in NZW mice, once the Akv-like (putative) ecotropic NZW-1 virus is induced. However, we did not find the NZW-1 virus in the plasma of NZW or B/W mice and thus do not know how it is induced. Because we cloned the NZW-1 virus from a B/W lymphoma, we also do not know whether the virus is recombinant or encoded in the NZW germ line.

The experimental insertion frequency (6 of 80) was not high enough to conclude that this mutagenesis contributes in an important way to hyperproliferation in the spleen. Perhaps the splinkerette method identifies only a small fraction of inserts, and we are underestimating the frequency. However, in 4 of 17 mice, the yield from quasimonoclonal cells from the peritoneum was better. We thus suggest that hyperproliferation is the initial step of lymphomagenesis in the peritoneum of these mice. There, the tendency toward monoclonality strongly points to a preleukaemic stage. The initiating step may be antigenic activation of a cell that will grow into a clone, followed by insertional mutagenesis mediated by MLV. MLV then activates a proto-oncogene (or deactivates a tumour suppressor) and causes hyperproliferation – more aggressive growth and larger cell numbers, from which new oncogenic events can be selected. This would lock the cell into continuous proliferation (Fig. S1a). The peritoneal cell-derived NYC lymphoma with its insertion in the Prex1 gene may be an example of this sequence of events. Alternatively, the retroviral insertion may happen earlier in ontogeny, such as in a pro-B-cell. However, in a proto-oncogene, the insertion may have no effect until the B-cell reaches the mature stage and is activated by antigen. In that case, many B-cells of different specificities would contain the first mutation, and these would only hyperproliferate after antigenic activation.

Mouse tumours characterized in B/W mice share at least three characteristics with human chronic lymphocytic leukaemia cells (Stall et al., 1988): they apparently are driven, in part, by (self) antigen (Dühren-von Minden et al., 2012; Jäck et al., 1992; Tarlinton et al., 1988), they express the surface marker CD5 (Tarlinton et al., 1988) and they have the particularly puzzling feature of an indolent phase of tumour evolution followed by a rapid diffuse expansion of a large B-cell lymphoma (Richter's syndrome). The underlying cause of this transition has been difficult to pinpoint. We suggest that, at least in the B/W mouse, the cause is an additional insertional mutagenesis event.

The ability of MLV to cause insertional mutagenesis in the B/W mice may also provide a mechanistic link between autoimmunity and cancer. Analogous to the B/W mice, which suffer from autoimmune disease and lymphadenopathy and eventually lymphoma (Stall et al., 1988), human systemic lupus erythematosus patients have a well-documented increased risk of non-Hodgkin's lymphoma (Bernatsky et al., 2005). As another example, BALB/c mice injected with mineral oil suffer from inflammation and develop both a lupus-like disease (Satoh et al., 2003) and plasmacytomas, the latter of which (in addition to translocations) are driven by de novo insertions of endogenous MLV (Knittel et al., 2014). The increased susceptibility to cancer in autoimmune humans and mice could be solely due to the large number of proliferating cells, which are the targets of mutagenesis. In that case, the tumour formation would be an epiphenomenon of the autoimmune disease with unknown aetiology. But we hypothesize that similarly to in lymphoma formation, in autoimmunity an immature B-cell may not be tolerized to self-antigen because the retroelement insertion blocks the relevant pathway (Fig. S1b). Because a single mutation in a stem cell may jeopardize the tolerization of a larger lymphocyte population, whereas cancer generally requires accumulation of several mutations within the same cell, autoimmune disease will precede cancer. For cancer, the genome must also allow repeated productive viral recombination to overcome the superinfection barrier – a requirement that is hindered by certain haplotypes (Datta et al., 1983).

In the autoimmune-prone mouse strain MRL/lpr, a connection between insertional mutation and autoimmunity is evident. The lpr mutation is caused by the insertion of a retrotransposon into the Fas gene, which results in aberrant splicing of the Fas message. This, in turn, results in proliferative disease, which accelerates the lupus-like disease (Drappa et al., 1993). Lack of tolerance also happens in non-autoimmune mice that (i) have a mutation leading to hyperreactive external calcium entry in B-cells (Yu et al., 2005), (ii) are overexpressing B-cell activators, including BAFF (Mackay et al., 1999), or (iii) are deficient (knockout) in inhibitors, including CD22 (Nitschke et al., 1997; O'Keefe et al., 1996; Otipoby et al., 1996; Sato et al., 1996), Lyn (Hibbs et al., 1995; Lamagna et al., 2013, 2014) and FcγRIIb (Bolland & Ravetch, 2000). Similarly, we suggest that insertional mutagenesis may accelerate lupus in B/W mice. However, unlike the germ line insertions introduced by the experimenter, or that in the MRL/lpr mouse, the insertions would be generated somatically, and when accumulating, they would lead to lymphoma.

Methods

Flow cytometry

The following antibodies were used: FITC-coupled goat anti-mouse IgM (μ chain specific; Southern Biotechnology), PE-coupled anti-CD5 (rat IgG_2a,κ; Pharmingen) and rat monoclonal antibody 83A25 specific to MLV gp70 (IgG_2a) plus APC-coupled goat anti-rat Ig (Pharmingen).

2-LTR sequences

Circular DNA was isolated (von Schwedler et al., 1990) and retroviral 2-LTR circle sequences were amplified. The following primers were used: ‘NZB primers A’: 5′-CAG TGT GGA GTT CTT ACC CTT-3′ (U3rev) and 5′-ATC CGA ATC GTG GTC TCG CTG-3′ (U5forw); and ‘NZB primers B’: 5′-TAC CCT TTT TAT AGG GCT GGG G-3′ (U3rev) and 5′-GCG CCA GTC CTC CGA TAG ACT G-3′ (R'c'forw). The 50 μl PCR mixture contained 2 μl of circular DNA, 0.2 μM of each primer, 200 μM dNTP, 1.5 mM MgCl₂ and 1 U of recombinant Taq DNA polymerase (Invitrogen) in 1 ×  PCR buffer (final concentration). Cycle conditions: 94 °C for 5 s; 35 cycles of 60 °C for 1 min and 72 °C for 1 min; and finally, 72 °C for 5 min. PCR-amplified products were gel purified using a Qiaquick gel extraction kit (Qiagen), cloned into pCR2.1 TOPO vector using a TOPO TA cloning kit (Invitrogen), and sequenced with M13 sequencing primers.

PCR amplification of ecotropic and polytropic PRR and full-length polytropic env

M. dunni cells were infected using plasma from mice or cell culture supernatant. Genomic DNA was isolated 10 days after infection using a DNA Blood and Tissue kit (Qiagen). The ecotropic proline-rich region (PRR) was amplified using primers specific for the ecotropic virus AKV gp70 280F CCGGGTCCCAATAGGGCCAAACCC and AKV gp70R GGACGGCAACCCCCTCGTAGTATG. The env was amplified using gp70 specific forward (876gp70F ACGCCGCTCACGTAAAAGCGG) and reverse (876gp70R GGCGTTACTGCAGCTAGCTTG). The PRR of the polytropic gp70 was sequenced using primers specific for the PRR domain gp70 280F CCGCGTCCCCATTGGGCCTAATCC and gp70 280R GGACGGCAACCCCTTCGTAGTAGG. There were no bands in the non-infected M. dunni cells.

Recovery of de novo MLV integration sites

De novo MLV integration sites were recovered using the splinkerette approach (Mikkers et al., 2002), described in detail in Knittel et al. (2014). Cellular DNA was restriction-digested with the enzymes Apo1, Nsp1 or BstY1 and ligated to a dsDNA oligonucleotide with a stable hairpin (splinkerette). Primers (given below) were such that they amplified a chimeric sequence containing part of the retroelement LTR and genomic sequences upstream of the integration site. Further validation of the de novo integration site included amplification of a chimeric fragment with a genomic 5′ primer and a 3′ LTR primer. Finally, because the databases provide the genome of the C57BL/6 mouse but not that of the NZB or NZW mouse, DNA fragments of the NZB and NZW mice spanning the retroelement insertion site were resequenced. Resequencing excluded the possibility that the retroelement insertions were in the germ line of these mouse strains.

The oligonucleotides for the splinkerettes were, for Apo1, 5′-CGAAGAGTAACCGTTGCTAGGAGAGACCGTGGCTGAATGAGACTGGTGTCGACACTAGTGG-3′ and 5′-AATTCCACTAGTGTCGACACCAGTCTCTGATTTTTTTTTTCAAAAAAA-3′; for Nsp1, 5′-CGAAGAGTAACCGTTGCTAGGAGAGACCGTGGCTGAATGAGACTGGTGTCGACACTAGTGGGATCCATG-3′ and 5′-GATCCCACTAGTGTCGACACCAGTCTCTAATTTTTTTTTTCAAAAAAA-3′; for BstY1, 5′-CGAAGAGTAACCGTTGCTAGGAGAGACCGTGGCTGAATGAGACTGGTGTCGACACTAGTGG-3′ and 5′-GATCCCACTAGTGTCGACACCAGTCTCTAATTTTTTTTTTCAAAAAAA-3′. The samples were then digested with EcoRV restriction nuclease to prevent amplification of internal viral fragments. The ligated fragments were amplified in three PCR rounds using nested primers. First primer set: forward primer binding to the splinkerette, CGAAGAGTAACCGTTGCTAGGAGAGACC; reverse primer specific for the viral LTR, CCTGTTTGGCCCTAGTCCCGGTAC. Cycle conditions: activation at 95 °C for 2 min; 27 cycles of 95 °C for 20 s, 60 °C for 30 s, 72 °C for 2 min 15 s; final elongation at 72 °C for 5 min. Second primer set: GTGGCTGAATGAGACTGGTGTCGAC (splinkerette) and CCAGCCTCTCTGTACTTCC (LTR), with PCR conditions being the same. Third-round adaptor binding primers contained the restriction sites and were therefore specific for each enzyme: Apo1, ATGAGACTGGTGTCGACACTAGTGGAATT(T/C); Nsp1, CTGGTGTCGACACTAGTGGGATCCATG(C/T); BstY1, GACTGGTGTCGACACTAGTGGGATC(C/T). The LTR primer was CTGAGAACATCAGCTCTGGTAC. Conditions were the same as in rounds one and two, except for an annealing temperature of 63 °C. For the second and third rounds of PCR, 2 μl of a 1 : 200 dilution of the product of the first PCR was used as a template.

Proviral insert validation

To validate the integration sites, PCR was run with genomic up- and downstream primers of the presumed integration site, as well as with a genomic forward primer and a proviral reverse primer. As the viral primer, either CTGAGAACATCAGCTCTGGTAC or CCACAGATATCCTGTTTGGCC was used.

Quantitative real-time reverse transcriptase (RT)-PCR

Total mRNA was isolated using an RNeasy MiniKit (Qiagen). After digestion with DNase I (Ambion), the mRNA was reverse-transcribed into single-stranded cDNA using a SuperScript III First Strand Synthesis System for RT-PCR (Invitrogen), followed by quantitative PCR with an Applied Biosystems 7300 real-time PCR cycler. A gene-specific forward and reverse primer (Applied Biosystems) and a gene-specific FAM-labelled MGB probe (Applied Biosystems) were used. MLV, MLVfor AAGCGGGTGGAAGACATCC, MLVrev AGCCCGCTCAAGAGGTTGT, MLV probe CCCCACCGTGCCCAACCCT (Lötscher et al., 2007); Prex1 (Applied Biosystems); mouse 18S rRNA (Applied Biosystems) was used as endogenous control. Quantitative real-time PCR data were analysed according to the comparative C _T method.

Supplementary Data

Supplementary Data