Main

FOXP1 is a member of the forkhead box family of transcription factors that have diverse functions in different cell and tissue types. FOXP1 regulates Rag1 and Rag2 expression and is essential for B-cell development.1 Apart from this, the function of FOXP1 is largely unknown. Nonetheless, FOXP1 function is unlikely to be confined to its role in B-cell development. Gene expression profiling studies show high expression of FOXP1 mRNA in activated B cells and activated B-cell-like diffuse large B-cell lymphomas.2 FOXP1 protein expression was seen in 40–60% of diffuse large B-cell lymphomas3, 4 and strong expression was significantly associated with inferior overall survival.4, 5 Similarly, FOXP1 expression was found in 29% of extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue (MALT) and strong expression was associated with poor disease-free survival and transformation to diffuse large B-cell lymphoma.6 These findings suggest a role for FOXP1 expression in lymphoma pathogenesis.

In support of the above notion, FOXP1 genetic abnormalities were also described in lymphoma. Streubel et al7 first showed that FOXP1 was involved in t(3;14)(p13;q32) in MALT lymphoma. FOXP1 translocation was present in 10% of MALT lymphomas, in those from the ocular adnexa (4/20=20%), thyroid (3/6=50%) and skin (2/20=10%), but not in those from the salivary gland, stomach and lung.7 The translocation was subsequently found in one case of MALT lymphoma of the stomach,8 seven cases of diffuse large B-cell lymphoma of the stomach, thyroid and lymph nodes and also in two cases of B-cell non-Hodgkin lymphoma unclassified.8, 9, 10 In addition, FOXP1 gene amplification was reported in one case of diffuse large B-cell lymphoma in one study,8 but not in others,7, 9, 10, 11 The discrepancies in the incidence of FOXP1 genetic abnormalities remain to be further investigated in a large series.

Of the 19 cases of FOXP1 translocation-positive lymphoma reported,7, 8, 9, 10 FOXP1 breakpoints were characterized in 1 gastric MALT lymphoma and 1 gastric diffuse large B-cell lymphoma, both with t(3;14)/FOXP1-IGH. In the gastric MALT lymphoma, the FOXP1 breakpoint was 38 kb upstream of the first noncoding exon 1 and the remaining FOXP1 sequence was joined to immunoglobulin heavy chain gene (IGH) joining segment 6.7 Thus, FOXP1 is overexpressed most likely as a result of strong transcriptional activity of the IGH enhancer. In contrast, in the gastric diffuse large B-cell lymphoma, the FOXP1 breakpoint was in intron 2 (according to the Ensemble Genome Database), thus causing separation from its normal promoter. The remaining FOXP1 sequence was joined to IGH-Cα1 but in an opposite transcriptional orientation.9 Despite this, FOXP1 protein was strongly expressed, suggesting expression of variant FOXP1 transcripts under the control of an alternative promoter.9 Although little is known about the transcriptional regulation of FOXP1, the different FOXP1 breakpoints described are expected to have significantly different effects on FOXP1 transcriptional control. However, the number of FOXP1 translocation cases investigated for FOXP1 breakpoints is small and characterization of breakpoints in further cases is needed to understand how the translocation impacts on FOXP1 transcriptional regulation. Such knowledge is critical not only for understanding the role of FOXP1 translocation in lymphoma pathogenesis, but also for studies of FOXP1 transcriptional regulation in normal and malignant B cells lacking FOXP1 genetic abnormalities. We have screened a large series of MALT lymphomas of various sites, and diffuse large B-cell lymphomas from nodal and extranodal sites for FOXP1 translocation and mapped the FOXP1 breakpoints in the 16 translocation-positive cases identified.

Materials and methods

Materials

A total of 595 cases of lymphoma were studied and their histological diagnoses were reviewed. These included 321 cases of MALT lymphoma, 59 cases of MALT lymphoma with a diffuse large B-cell lymphoma and 215 cases of diffuse large B-cell lymphoma; and their tissue origins are summarized in Table 1. Formalin-fixed paraffin-embedded tissues were retrieved from each case. The use of such archival tissues for research was approved by the local ethics committees of the authors’ institutions, where required.

Table 1 Incidences of FOXP1 translocation and numerical changes in lymphoma
Full size table

Fluorescence In Situ Hybridization

A series of 10 BAC clones mapping to various positions of the FOXP1 gene were used to form three break-apart (BAP) fluorescence in situ hybridization (FISH) assays to detect FOXP1 translocation and to map FOXP1 breakpoints (Table 2; Figure 1). The 3 BAC clones telomeric of the FOXP1 gene were combined with 3 BAC clones centromeric of the IGH locus (kindly provided by Professor Reiner Siebert) to form a FISH assay to detect FOXP1–IGH fusion. BAC DNA extraction and labeling were performed essentially as previously described.12 The specificity and efficacy of the probes were tested on metaphase slides and on formalin-fixed paraffin-embedded tissue sections of a MALT lymphoma with t(3;14)(p13;q32)/FOXP1-IGH.

Table 2 BAC clones used to generate interphase FISH assays for detection of FOXP1 translocation
Full size table
Figure 1
figure 1

Schematic representation of FOXP1 genomic structure and the BAC clones used for detection of FOXP1 translocation and mapping of FOXP1 breakpoints. Please refer to Table 2 for the full name and position of the BAC clones.

Full size image

Interphase FISH was performed as described previously.12 For each probe set, the mean plus three standard deviations of false positive signals in 100 nuclei from seven reactive tonsils was used as the cut-off value for the diagnosis of chromosomal aberrations.12 FOXP1 translocation-positive cases were further screened with IGH (Vysis/Abbott Laboratories Ltd, UK), IG-κ and IG-λ BAP probes (kindly provided by Professor Reiner Siebert) to identify potential translocation partners. All MALT lymphomas with FOXP1 translocation were investigated with a MALT1 BAP probe, whereas diffuse large B-cell lymphomas with FOXP1 translocation were investigated with BCL6, BCL2 and c-MYC BAP probes (Vysis/Abbott Laboratories Ltd).

Immunohistochemistry

FOXP1 immunohistochemistry was performed on formalin-fixed paraffin-embedded tissue sections using the JC12 mouse monoclonal antibody, which recognizes specifically an epitope in the C-terminus (within exons 18–21) of FOXP1. Various conditions including different antibody dilutions and antigen retrieval methods were tested and the optimized conditions are described below. Briefly, sections were pretreated in a microwave for 15 min in DAKO Low Target Retrieval Solution and then incubated with the JC12 antibody (1/150 dilution) for 45 min, followed by incubation with secondary antibody and visualization by the streptavidin-immunoperoxidase method. FOXP1 immunostaining was evaluated independently by three assessors (AG, CMB, MQD) and scored according to the percentage of positivity (<30%, 30–70%, >70% cells) and the intensity of immunostaining (weak, moderate, strong) as described previously.4, 6

CD10, Bcl-6 and MUM1 were similarly immunostained in diffuse large B-cell lymphomas with FOXP1 translocation and the staining was assessed semiquantitatively by visual estimation. Cases were considered positive if 30% or more of the tumor cells were stained. Diffuse large B-cell lymphomas were subclassified into the germinal center B-cell-like or non-germinal center B-cell-like subtypes as described previously.3

Statistical Analysis

The χ2-test was used to analyze the difference in FOXP1 staining between MALT lymphoma and diffuse large B-cell lymphoma.

Results

FOXP1 Genetic Aberrations in Lymphoma

In total, 16 of the 595 cases of lymphoma studied with FOXP1 BAP probe 1 showed evidence of a FOXP1 translocation. These included eight MALT lymphomas, three MALT lymphomas with a diffuse large B-cell lymphoma component and five diffuse large B-cell lymphomas (Table 1). In the three cases of MALT lymphoma with a diffuse large B-cell lymphoma component, FOXP1 translocation was detected in the MALT lymphoma but presence of the translocation in the diffuse large B-cell lymphoma component could not be investigated due to the absence of diffuse large B-cell lymphoma in the tissues available for research. All cases of MALT lymphoma and MALT lymphoma with a diffuse large B-cell lymphoma component harboring FOXP1 translocation were from the stomach, accounting for 4 and 6% of the corresponding lymphoma subtype of the stomach, respectively. The five translocation-positive diffuse large B-cell lymphomas were from the stomach (n=2, 3%), intestine (n=1, 5%), tonsil (n=1, 8%) and lymph node (n=1, 2%).

Gain of one or more extra copies of the FOXP1 gene was seen in 17% of MALT lymphomas, 12% of MALT lymphomas with a diffuse large B-cell lymphoma component and 33% of diffuse large B-cell lymphomas (Table 1). These included three cases (one thyroid MALT lymphoma with diffuse large B-cell lymphoma, one intestinal diffuse large B-cell lymphoma and one gastric diffuse large B-cell lymphoma) showing four or more extra copies of the FOXP1 gene. Gain of an extra copy of the FOXP1 gene, but not the translocated allele, was seen in 3 of the 16 FOXP1 translocation-positive cases (19%).

Further interphase FISH with MALT1 BAP probe showed that none of the FOXP1 translocation positive MALT lymphomas or MALT lymphomas with diffuse large B-cell lymphoma had a translocation involving the MALT1 locus. For diffuse large B-cell lymphoma with FOXP1 translocation, interphase FISH was performed with BCL2, c-MYC and BCL6 BAP probes, respectively, and only one case (no. 14) was found to harbor a BCL6 translocation.

Clinicopathological Features of Lymphomas with FOXP1 Translocation

There were neither apparently uniform clinical nor histological features associated with these FOXP1 translocation-positive cases (Table 3). All MALT lymphoma cases showed characteristic histology and immunophenotype. Follicular colonization and prominent plasma-cell differentiation were each seen in three cases, with one case displaying both histological features. One case of stage-I gastric MALT lymphoma (no. 2) was treated by Helicobacter pylori eradication but showed no response to this first-line treatment.14, 15 Interestingly, one case of stage-I MALT lymphoma with a diffuse large B-cell lymphoma component (no. 9) showed complete remission following H. pylori eradication. With the exception of one patient with gastric MALT lymphoma who died of postsurgical infection, all other cases of MALT lymphoma with or without diffuse large B-cell lymphoma showed favorable responses to treatment.

Table 3 Summary of cases with FOXP1 translocation
Full size table

Of the five cases of diffuse large B-cell lymphoma, two cases were of germinal center B-cell-like subtype and the remaining three were of non-germinal center B-cell-like subtype.3 The histological and clinical features of these cases are summarized in Table 3.

Identification of the Translocation Partner of FOXP1

Interphase FISH with an IGH BAP probe showed a break in the IGH locus in 11 of the 16 FOXP1 translocation-positive cases (Table 3). Further FISH with the FOXP1–IGH fusion probe confirmed IGH as the translocation partner in each of these 11 cases. Of the remaining five cases in which FOXP1 translocation was not associated with IGH, further interphase FISH with IG-κkappa and IG-λ BAP probes were performed in four cases where materials were available (Table 3), and the results were negative.

Mapping FOXP1 Breakpoints

To map the location of FOXP1 breakpoints, further interphase FISH was performed using FOXP1 BAP probes 2 and 3, and additionally with BAC clones C2/H15/P21. Three cases (nos. 1, 2 and 16) showed a split signal with BAP probe 2, indicating breakpoints between BAC clones O22 and M13, corresponding to the region spanning 18 kb upstream of exon 1 to 54 kb downstream of exon 2 (Figure 1). Although the precise positions of the breakpoints in these cases were unknown, the relative sizes of the exons and introns in this region suggest that these breakpoints most likely reside downstream of exon 1. Eleven cases (nos. 3–9 and 12–15) displayed a split signal with BAP probe 3, indicating breakpoints between H23 and C2, corresponding to the region spanning from intron 3 to 13 kb downstream of exon 5 (Figure 1). The remaining two cases (nos. 10 and 11) showed no evidence of split signals with BAP probes 2 and 3, but displayed split signals with BAC clones C2/H15/P21. Further interphase FISH localized the breakpoint to the region covered by H15, which spans 60 kb upstream of exon 7 to exon 10 (Figure 1). In view of the long introns 6–7 but short exons 8–10 and introns 8–9, it is likely that the breakpoints in these two cases are located upstream of exon 8.

There was no difference in the location of FOXP1 breakpoints between MALT lymphoma and diffuse large B-cell lymphoma, or between cases with and without IGH as the translocation partner.

Correlation of FOXP1 Protein Expression with Its Gene Abnormalities

FOXP1 protein expression was investigated by immunohistochemistry in 119 cases of MALT lymphoma, 3 cases of FOXP1 translocation positive MALT lymphomas with diffuse large B-cell lymphoma and 139 cases of diffuse large B-cell lymphoma, where tissues were available. Strong uniform FOXP1 staining was found in most FOXP1 translocation-positive cases of MALT lymphoma and diffuse large B-cell lymphoma (Tables 3 and 4; Figure 2). However, such strong homogenous FOXP1 staining was also frequently seen in FOXP1 translocation-negative lymphomas, irrespective of the presence of normal or increased copy number of the FOXP1 gene, being significantly more frequent in diffuse large B-cell lymphoma (46%) than in MALT lymphomas (25%; P<0.0005; Table 4).

Table 4 Correlation of FOXP1 expression and its genetic abnormalities in MALT lymphoma and diffuse large B-cell lymphoma
Full size table
Figure 2
figure 2

Examples of FOXP1 FISH and FOXP1 immunohistochemistry. FOXP1 interphase FISH with break-apart probe (BAP) set 1 shows a split signal and presence of an extra copy of the gene in a case of gastric MALT lymphoma (a). A FOXP1 translocation positive gastric MALT lymphoma shows a diffuse infiltrate of centrocyte-like cells, prominent plasma-cell differentiation and lymphoepithelial lesions (indicated by arrow) (b). The neoplastic nature of plasma cells is indicated by IgM expression (c) and Igκ light chain restriction (d). FOXP1 immunohistochemistry shows strong nuclear staining in both neoplastic centrocyte-like cells and plasma cells (e). Examples of diffuse large B-cell lymphoma display strong (f), moderate (g) and weak (h) FOXP1 staining.

Full size image

Discussion

Unlike t(11;18)(q21;q21)/API2-MALT1, t(1;14)(p22;q32)/BCL10-IGH and t(14;18)(q32;q21)/IGH-MALT1 which are specifically associated with MALT lymphoma,12, 16, 17 FOXP1 translocation has been found in MALT lymphoma and diffuse large B-cell lymphoma of nodal and extranodal sites.7, 8, 9, 10 However, similar to these three translocations, FOXP1 translocation also occurs at variable frequencies in MALT lymphoma of different sites. Our results, based on screening a large series of MALT lymphomas of various sites, showed that FOXP1 translocation was associated with those of the stomach, but not other mucosal sites. In line with this, two of the five extranodal diffuse large B-cell lymphomas with FOXP1 translocation were also from the stomach. These findings are in keeping with the results of the recent studies by Wlodarska et al8 and Haralambieva et al.10 In addition, we also found rare cases of MALT lymphoma and diffuse large B-cell lymphoma showing 4 or more extra copies of the FOXP1 gene, suggesting the presence of chromosome polysomy or FOXP1 gene amplification.8 Nonetheless, the incidences of FOXP1 translocation or gene amplification in MALT lymphoma and diffuse large B-cell lymphoma are low.

Detailed breakpoint analysis of t(3;14)/FOXP1-IGH was performed in two cases in previous studies.7, 9 In one case, the FOXP1 breakpoint was 38 kb upstream of the first noncoding exon 1 with the remaining FOXP1 sequence joined to IGH joining segment 6,7 whereas in the other, the breakpoint was in intron 2 with the remaining FOXP1 sequence joined to IGH-Cα1 but in an opposite transcriptional orientation.9 Here we show, based on analysis of 16 cases with FOXP1 translocation, that the FOXP1 breakpoint was within the 5′ region of the gene in most, if not all, cases. In 11 cases, the breakpoint was mapped to the region spanning intron 3–13 kb downstream of exon 5. In three cases, the breakpoint was mapped to the region spanning 18 kb upstream of exon 1–54 kb downstream of exon 2, and the relative lengths of the exons and introns in this region suggest that the breakpoints in these cases are likely to be located downstream of exon 1. In the remaining two cases, the breakpoints were mapped to a region spanning from 60 kb upstream of exon 7 to exon 10, and the relative sizes of the introns and exons in this region suggest that the breakpoints are likely to be located upstream of exon 8. Significantly, these breakpoints fall within the coding sequence of full-length FOXP1, which begins in exon 6. Interestingly, Wlodarska et al8 reported a t(2;3)(q36;p13) involving FOXP1 and an unknown partner gene in a case of small B-cell non-Hodgkin lymphoma, in which the FOXP1 breakpoint was similarly mapped to the region spanning exon 5 to exon 7. Thus, most, if not all, of these FOXP1 breakpoints separate the downstream FOXP1 sequence from its normal promoter. Irrespective of these different breakpoints, immunohistochemistry with JC12, a monoclonal antibody to the C-terminus of FOXP1, demonstrated high-level expression of the protein in each of these translocation positive cases.

The 5′ end of the FOXP1 gene might also be disrupted by other means during oncogenesis. In one study FOXP1 was one of the most frequent targets of retroviral integration in myeloblastosis-associated virus type-2-induced chicken nephroblastoma.18 Significantly, all the integration events were clustered around the second coding exon (corresponding to human FOXP1 exon 7).18 Typically, the retrovirus exerts its oncogenic potential by deregulating cellular gene expression or causing alterations in the cellular gene product, consequently changing its function.

The similar disruption of the 5′ sequence of the FOXP1 gene by various chromosome translocations in human lymphoma and by retroviral integrations in the chicken nephroblastoma model suggests that these different events may activate the oncogenic potential of FOXP1 through a common mechanism of deregulating FOXP1 transcriptional control. As mentioned above, most, if not all, of these FOXP1 breakpoints separate the downstream FOXP1 sequence from its normal promoter. Thus FOXP1 translocation could result in the loss of normal full-length FOXP1 mRNA expression, and expression of variant FOXP1 transcript species with alternate 5′ ends, most likely under the control of an alternative internal promoter. In support of this speculation, there are 14 probable alternative promoters for FOXP1 according to AceView (http://www.humangenes.org). In addition, at least in the case of FOXP1 translocation with IGH as the translocation partner, the probability of promoter substitution is unlikely due to the opposite transcriptional orientation of the FOXP1 and IGH genes as demonstrated in detail by Fenton et al.9 More importantly, we have recently identified several FOXP1 transcript variants utilizing novel 5′ noncoding exons including two variants using an alternative exon 6b and four variants using an alternative novel exon 7(a–c) in activated B-cell-like diffuse large B-cell lymphoma.19 Furthermore, high-level expression of smaller FOXP1 proteins was found in activated B-cell-like diffuse large B-cell lymphoma by western blotting.19 Among FOXP1 transcript variants identified, those using an alternative novel exon 7 contain the translation initiating site (two ATG) in exon 8, and thus encode N-terminally truncated proteins (lacking approximately the first 100 amino acids of the full-length FOXP1), which was consistent with the detection of smaller FOXP1 proteins by western blotting.19 Although little is known about the function of the human N-terminally truncated FOXP1 isoforms, the murine FOXP1D that lacks the N-terminal polyglutamine domain exhibits altered transregulatory properties.20 Thus, it is tempting to speculate that FOXP1 translocations, at least those with a breakpoint mapped to intron 6–7, may lead to expression of FOXP1 isoform(s) with a truncated or alternate N-terminus.

Among the 16 cases of lymphoma with FOXP1 translocation studied, there were no obviously uniform clinical or histological features associated with the translocation. All the MALT lymphomas with FOXP1 translocation showed classic histological and immunophenotypic features of MALT lymphoma. Clinically, all four cases of FOXP1 translocation-positive gastric MALT lymphoma with long-term follow up showed durable complete remissions following gastrectomy. Of the five cases of diffuse large B-cell lymphoma with FOXP1 translocation, the histological, immunophenotypic and clinical features were also not remarkable. The markers, including CD10, Bcl6 and MUM1, commonly used for subclassification of diffuse large B-cell lymphoma into germinal center B-cell-like and non-germinal center B-cell-like subtypes were heterogeneously expressed in these cases, as described previously.10

Apart from deregulation of FOXP1 expression by chromosome translocation, overexpression of the protein was also frequently seen in MALT lymphoma and diffuse large B-cell lymphoma lacking the translocation and this was independent of FOXP1 gene copy number change. Interestingly, strong uniform FOXP1 expression was seen more frequently in diffuse large B-cell lymphoma (46%) than in MALT lymphomas (25%; Table 4), as demonstrated previously.3, 4, 5, 6 These data, together with detection of FOXP1 translocation and gene amplification in these lymphomas suggest that FOXP1 deregulation may play an important role in their development. The findings that FOXP1 overexpression is independent of its gene translocation and copy number changes indicate the presence of other mechanisms that deregulate FOXP1 expression in lymphoma. As discussed above, we have recently identified several FOXP1 transcript variants utilizing novel 5′ noncoding exons and high-level expression of smaller FOXP1 protein isoforms in activated B-cell-like diffuse large B-cell lymphoma.19 Significantly, the expression of these transcript variants and the smaller protein isoforms can be induced in normal B cells following antigenic or mitogenic stimulation.19 Our localization of FOXP1 translocation breakpoints and the implications of these breakpoints for the transcriptional deregulation of FOXP1 in lymphoma provide timely and important insights to guide future investigations of the role of FOXP1 in both B-cell biology and lymphoma development.