We have taken an important first step to overcome this difficulty by developing a genetic method for the detection of individual T(12;15)-carrying cells in mice. Using established gene-targeting methods in embryonic stem cells, we generated iGFP^5′Myc gene-insertion mice that harbor in the near upstream flank of the Myc locus a cDNA gene that encodes the enhanced GFP protein under control of the mouse immunoglobulin heavy-chain (Igh) variable gene promoter, V_H (Figure 1, top). The GFP reporter has been inserted ~1.5 kb 5′ of Myc exon 1, to make it possible to specifically detect derivative 15, der(15), of the T(12;15) translocation. Unlike der(12), which contains the deregulated Myc gene that drives tumor development (Eisenman, 2001; Levens, 2003) and is subject to secondary changes in the course of plasmacytomagenesis, including those caused by class switch recombination on the rearranged Igh–Myc domain (Kovalchuk et al., 1997), der(15) is thought to be inconsequential for malignant cell transformation and not affected by secondary structural modification. Moreover, cytogenetic analysis of hundreds of mouse PCTs has shown that der(15) is rarely, if ever, lost in T(12;15)⁺ tumors. These features render der(15) a particularly attractive host for a translocation reporter gene, such as the V_H-GFP that we use here. Because the GFP gene insertion site in iGFP^5′Myc mice is just upstream of the known 5′ boundary of the translocation breakpoint cluster in Myc, the reporter gene will invariably be allocated to der(15), irrespective of the actual break site in Myc, which occurs in most cases in intron 1 or exon 1 of the gene and, less frequently, in its near 5′ flank.

According to the scheme presented in Figure 1, GFP remains silent in B cells of strain iGFP^5′Myc mice that have not undergone T(12;15) exchange because the V_H promoter of the GFP gene on chromosome 15 cannot be activated in trans by Igh enhancers on chromosome 12 (Figure 1, top). However, in B cells that do undergo T(12;15) exchange (Figure 1, bottom), GFP is expressed upon the juxtaposition of V_H–GFP to an Igh enhancer in cis. Assembly of the targeting vector, generation of iGFP^5′Myc ‘knock-in’ mice, and the development of genomic PCR methods for genotyping transgenic off-spring are described online in Supplementary Methods and illustrated in Supplementary Figure 1.

Reporter genes targeted to oncogenes may either promote or inhibit tumorigenesis—by changing the structure, accessibility or biological function of the modified genetic loci. To evaluate whether either is true in the case of strain iGFP^5′Myc, we performed a PCT induction study in these mice. To accelerate tumor development, we combined the iGFP^5′Myc reporter gene with a human BCL2 transgene that was shown in previous work to lead to T(12;15)-harboring plasma cell tumors with short onset and full penetrance (Silva et al., 2003). Figure 2 shows that upon induction of chronic peritoneal inflammation using pristane (Anderson and Potter, 1969), PCTs arose as rapidly (onset 50–120 days) and with the same incidence (100%) in double-transgenic iGFP^5′Myc/BCL2 mice as in single-transgenic BCL2 mice. Characteristic features of mouse PCTs that are detectable by histomorphological criteria, by the expression of certain cell-surface markers including CD138 and by monoclonal Ig production, were also indistinguishable in iGFP^5′Myc/BCL2 and BCL2 mice (results not shown).

To evaluate whether the proclivity of PCT to undergo T(12;15) translocation was comparable in iGFP^5′Myc-transgenic mice and mice lacking the reporter transgene, we employed genomic PCR to analyze 13 tumors from iGFP^5′Myc/BCL2 mice for the presence of chimeric Igh–Myc junction fragments (Kovalchuk et al., 2000b). The remaining six tumors included in Figure 2 were not available for molecular studies. Using the Myc and Igh primers depicted in Figure 1, we readily detected T(12;15)-typical junctions in 11 of 13 PCTs: der(12)-typical junctions were found in 10 tumors, and der(15)-typical junctions were seen in nine (Table 1, columns 3–4). The translocation status of two tumors remained undetermined, presumably because they contained an unusual T(12;15) exchange that was not detectable with the PCR methods used here, harbored a variant Myc translocation that relied on an immunoglobulin light-chain instead of the Igh locus, or did not harbor an Myc translocation at all. The detection of T(12;15) in 85% of iGFP^5′Myc-carrying PCT was consistent with the incidence of this translocation in PCTs from normal mice of strain C (Janz, 2006).

The results presented above suggested that the GFP reporter gene in strain iGFP^5′Myc is a passive genomic passenger that neither affects PCT development nor diminishes the likelihood that the T(12;15) translocation will be generated. However, because the present sample of translocation-bearing tumors is small (n = 11), a follow-up study producing a larger data set is necessary before it can be unambiguously decided whether the GFP reporter does or does not cause bias with regard to translocation occurrence and tumor development. Including a side-by-side comparison of hemizygous transgenic iGFP^5’Myc mice (containing one targeted and one normal Myc locus) and their homozygous transgenic counterparts (lacking the normal Myc locus) in that study may be particularly informative.

To identify PCTs in which T(12;15) is associated with GFP expression, we performed FACS analysis of the 11 tumors that showed PCR-based evidence of translocation. We stained PCT-containing ascites specimens with antibody to CD138 (syndecan-1, marker of plasma cells) and B220 (CD45R, marker of B cells) to distinguish the malignant plasma cells from populations of normal cells including peritoneal B cells. We found that 4 of 11 (36%) tumors expressed high levels of GFP (Table 1, column 2). This is illustrated in Figure 3a, in which PCT2 serves as a GFP⁺ example, and PCT8 as a GFP⁻ example. In the case of PCT2, 24% of ascites cells were neoplastic plasma cells, as indicated by the co-expression of GFP and CD138 (GFP⁺CD138⁺) and a lack of B220 expression (B220⁻). The fraction of GFP⁺ tumor cells in the three other cases characterized by high levels of GFP (PCT1, 3, 4) ranged from 6 to 31% (Supplementary Figure 2). To confirm GFP expression in these cells by an independent method that additionally provides insight into the location and distribution of translocation-bearing cells in vivo, we performed immunofluorescence microscopy with antibodies to CD138 and B220, on serial tissue sections of peritoneal granulomas, the sites of PCT development. In all four cases (PCT1–4), GFP⁺CD138⁺ tumor cells that infiltrated the granulomatous tissue (Figure 3b, top and bottom rows) were readily distinguished from GFP⁻B220⁺ B lymphocytes (Figure 3b, center row). Normal CD138⁺ plasmablasts and plasma cells residing in spleen, gut villi and bone marrow were GFP-negative (not shown).

To further show that GFP expression in PCT1–4 was caused by the T(12;15) translocation, we performed fluorescence in situ hybridyzation (FISH) of Myc and Igh as described in Supplementary Methods. FISH detects distinct Myc and Igh signals in normal cells, but a co-localized or merged Igh–Myc signal in T(12;15)⁺ cells. In the case of PCT1, 11 of 15 (73%) tumor cells analyzed at the metaphase stage of the cell cycle showed merged FISH signals on der(12) (Figure 3c). In the case of PCT2, this was true for 10 of 10 (100%) tumor cells. As PCT3 and PCT4 contained less than the 10 metaphase cells commonly required for reliable cytogenetic diagnosis, we analyzed these tumors by interphase FISH. In the case of PCT3, co-localization of Myc and Igh signals was seen in 34 of 154 (22.1%) cells from an ascites specimen that contained 35% tumor cells and 65% normal (mostly inflammatory) cells (Supplementary Figure 3). The observed co-localization frequency was significantly elevated compared with that in normal splenic B lymphocytes: 1.7% (3/180 cells). Similar findings were obtained for PCT4 (not shown).

All of the PCTs included in Table 1 arose in mice that were hemizygous for the iGFP^5′Myc transgene, thus carrying one targeted Myc locus and one normal Myc locus. To show directly that the GFP⁺ tumors used the targeted Myc allele for genetic recombination with Igh, we refined our PCR analysis of der(15) by replacing the Myc primer used for detecting the underlying Igh–Myc junction with a GFP primer that annealed further upstream (~2 kb) of Myc (Figure 1, green arrow below the scheme of der[15]). We combined the GFP primer with a PCR primer that annealed to the J_H portion of Igh (black arrow) and were readily able, in all four tumors (PCTs 1–4), to elongate the clonotypic Igh–Myc junction previously generated using the J_H and Myc exon 1 primer pair (red arrow). The der(12)-specific GFP–Myc–J_H fragments are shown in Figure 3d, lanes 1–4.

These results showed that in four independent tumors (PCTs 1–4), GFP activation was caused by juxtaposition of the V_H-GFP gene to Igh enhancers following chromosomal T(12;15) translocation.

We predicted that, in contrast to the four GFP⁺ tumors described above, the GFP⁻ neoplasms had used the normal rather than the targeted Myc allele for translocation with Igh. Consistent with this possibility, we failed to detect GFP–Myc–J_H junctions in PCTs 6–8 (Figure 3d). In the case of PCTs 10 and 11, however, clonotytpic GFP–Myc–J_H junctions were obtained (Figure 3d). This suggested that the targeted Myc allele had recombined with Igh, yet GFP had not been activated. Lack of GFP expression could potentially be due to (1) epigenetic changes that diminish or extinguish GFP expression by silencing the Igh–Myc chromatin domain or (2) genetic changes such as translocation breakpoints disrupting the Eµ enhancer or—in keeping with the role of activation-induced cytidine deaminase in the origin of the T(12;15) translocation (Ramiro et al., 2004; Franco et al., 2006)—crippling activation-induced cytidine deaminase-mediated point mutations in the V_H promoter or the coding sequence of the GFP gene (Muramatsu et al., 2000).

As molecular analyses, including DNA sequencing of translocation breakpoint regions, have not supported any of these possibilities thus far, alternative explanations must be considered to account for the apparent absence of GFP expression in PCTs 10 and 11. The most likely of those postulates the co-existence of a frank GFP⁻ neoplasm and incipient GFP⁺ tumor cell clones in the same mouse. According to this explanation, the mice contained an overt CD138⁺ plasma cell tumor, designated PCT 10 or 11, that was readily detected by flow cytometry, yet lacked GFP expression either because it contained an unusual T(12;15) exchange that was not detectable with the PCR methods used here, harbored a variant Myc translocation that used an immunoglobulin light-chain instead of the Igh locus, or did not harbor an Myc translocation at all. Be this as it may, both mice additionally contained at least one T(12;15)⁺ cell clone that had used the GFP-targeted Myc allele for recombination with Igh (based on the PCR results shown in Figure 3d) and, hence, expressed GFP. However, due to the small size of the underlying cell clone, the GFP⁺ cells remained undetected by FACS. Consistent with this theory, clonal diversity of T(12;15)⁺ PCTs has been repeatedly documented in strain C mice undergoing peritoneal tumor induction (Janz et al., 1993; Müller et al., 1996, 1997a; Kovalchuk et al., 2000a, b, 2003). Furthermore, it is well established that PCR analysis of clonotypic Myc-Igh junctions detects T(12;15) translocation-bearing cells with much greater sensitivity than possible with FACS analysis of GFP expression in iGFP^5′Myc mice. PCR routinely detects one T(12;15)⁺ cell among 10⁵–10⁶ normal cells (Janz, 2006), whereas conventional flow cytometry detects one GFP⁺ cell among 10²–10³ normal cells. The large difference in the detection limit (two to three orders of magnitude) of scoring T(12;15)⁺ cells by PCR versus FACS analysis would thus explain the findings in PCT10/11-carrying mice without postulating that the GFP reporter had failed.

As stated above, a follow-up study producing a larger sample of T(12;15)⁺ tumors in iGFP^5′Myc mice is necessary before it can be determined with which frequency, if any, the GFP-targeted Myc locus undergoes T(12;15) exchange without activating the GFP reporter. Using to that end homozygous transgenic iGFP^5′Myc mice may be particularly productive, because in these mice only GFP-targeted Myc alleles are available for T(12;15) exchange.

The newly developed iGFP^5’Myc mice provide proof of principle that the targeted insertion of a transgenic reporter in the vicinity of an oncogene is a viable approach for detecting, enumerating and studying translocation-bearing cells in their normal tissue context. Combined with sensitive in vivo imaging techniques, such as two-photon microscopy (Germain et al., 2005; Wessels et al., 2007), the iGFP^5′Myc mice may afford the possibility to visualize the migration and trafficking of Igh–Myc translocation-carrying cells in vivo. Additional questions that can be readily addressed with the assistance of the GFP reporter include the abundance and tissue distribution of T(12;15)⁺ tumor cells, using fluorescence microscopy, immunohistochemistry, flow cytometry and whole-body fluorescence imaging. Still more interesting are questions that might be attacked in future studies on translocation-carrying B-lineage cells that have not yet completed malignant transformation. Figure 4, which depicts our most recent findings in tumor-free iGFP^5′Myc mice, suggests that the GFP reporter may have great potential in that regard. Thus, 8 to 9-month-old tumor-free iGFP^5′Myc mice hyper-immunized with the T-cell-dependent antigen, sheep red blood cells, showed a nearly threefold increase in the number of GFP-expressing B220⁺CD19⁺ splenocytes (0.12%) compared with age-matched non-immunized iGFP^5′Myc mice (0.041%) and normal C mice (0.0088%) used as background controls. The result implicates the germinal center reaction in the promotion of GFP⁺ cells and, hence, T(12;15)⁺ B-lymphocytes that can be operationally defined as pre-malignant PCT precursors.

Relatively simple questions to be asked initially in follow-up studies include whether ‘pre-malignant’ T(12;15)⁺ cells are scattered in lymphoid tissues or clustered in certain anatomical sites, such as germinal centers. A similarly trivial question is whether T(12;15)⁺ cells can also reside in non-lymphoid tissues; for example, the lamina propria of the gut, as suggested by PCR results obtained in our laboratory (Müller et al., 1997a, b). Somewhat more involved will be attempts to pinpoint the stage(s) of B-cell differentiation at which translocations take place: T(12;15) caused by V(D)J recombination may occur in immature B cells in the bone marrow, whereas translocations caused by aborted isotype switching and VDJ hypermutation may take place in mature B cells in secondary lymphoid tissues (Wang et al., 2009). A third set of studies will focus on the mechanism of translocation (Gostissa et al., 2009a); for example, crossing the iGFP^5′Myc reporter on genetic backgrounds that contain certain activation-induced cytidine deaminase (Cheng et al., 2009) or DNA double-strand break repair mutations (Nussenzweig et al., 1997; Ouyang et al., 1997; Difilippantonio et al., 2000) may provide important insights into the role of these pathways in the origin of T(12;15). Similarly, breeding the iGFP reporter onto certain cytokine including IL-6 transgenic (Suematsu et al., 1992) backgrounds may enhance our understanding of growth, differentiation and survival requirements of aberrant T(12;15)⁺ cell clones.

In conclusion, extension of the present reporter gene insertion approach to the pre-malignant state (Igh–Myc-bearing tumor precursors) and to other types of oncogene-activating or fusion-gene translocations (Mitelman et al., 2007) may lead to fundamental new insights into the nature, developmental stage and site of origin of tumor progenitors that are of great relevance for human cancer.

SuppData