LETTERS

© 2003 Nature Publishing Group http://www.nature.com/naturegenetics

LETTERS

Spatial proximity of translocation-prone gene loci in human lymphomas Jeffrey J Roix1, Philip G McQueen2, Peter J Munson2, Luis A Parada1 & Tom Misteli1

Cancer cells frequently have disease-specific chromosome rearrangements1–4. It is poorly understood why translocations between chromosomes recur at specific breakpoints in the genome. Here we provide evidence that higher-order spatial genome organization is a contributing factor in the formation of recurrent translocations. We show that MYC, BCL and immunoglobulin loci, which are recurrently translocated in various B-cell lymphomas, are preferentially positioned in close spatial proximity relative to each other in normal B cells. Loci in spatial proximity are non-randomly positioned towards the nuclear interior in normal B cells. This locus proximity is the consequence of higher-order genome structure rather than a property of individual genes. Our results suggest that the formation of specific translocations in human lymphomas, and perhaps other tissues, is determined in part by higher-order spatial organization of the genome. We sought to investigate how the spatial organization of gene loci in the cell nucleus contributes to their participation in balanced translocations. As model systems, we studied several disease-specific chromosome rearrangements reported in B-cell lymphomas. In Burkitt’s lymphoma, normal expression of the oncogene MYC (8q24) is disrupted by recurrent translocations with the genes encoding immunoglobulin heavy chain (IGH, 14q21), light chain λ (IGL, 22q11) or light chain κ (IGK, 2p11; refs. 5,6). In mantle, follicular and diffuse large cell lymphomas, the IGH locus is frequently found in translocations with the oncogenes CCND1 (formerly BCL1, 11p13), BCL2 (18q12) or BCL6 (3q27), respectively6–8. We first assessed the global nuclear organization of translocationprone genes by localizing them in the karyotypically normal B-lymphoblastic MC/CAR cell line using fluorescence in situ hybridization (FISH; Fig. 1a–c). We visualized the loci in several thousand cells using a semi-automated high-throughput image acquisition system and characterized the radial distribution of each locus in the nuclear volume by measuring the distance of each fluorescence signal from the nuclear center (Fig. 1d–f). Some loci (IGL and CCND1) frequently localize near the nuclear interior (Fig. 1a,d,g,h), whereas others (MYC, BCL6, IGK) are preferentially located near the periphery (Fig. 1c,f,g,h). The IGH locus is enriched in a nuclear sub-domain

about halfway between the nuclear center and the nuclear envelope (Fig. 1b,e,g). The preferred positioning of all loci is statistically distinct from a uniform random distribution as determined by the Kolmogorov-Smirnov test (Fig. 1g,h and Supplementary Methods and Supplementary Tables 1–6 online). We also observed non-random positioning for the non-translocating TGFBR2 locus, as has been reported for several other gene loci (Fig. 1h; Supplementary Table 1 online; refs. 9–14). We conclude that human gene loci frequently have non-random and gene-specific distribution patterns in the interphase nucleus. Formation of balanced translocations requires that two chromosomes come into physical contact15–17. The degree of physical proximity between two gene loci before the translocation event might thus increase the probability of illegitimate joining. To test whether spatial separation between loci in normal cells is a contributing factor in formation of translocations, we took advantage of the fact that in Burkitt’s lymphoma, MYC translocates at differential frequencies with IGH, IGL or IGK5,6. We measured the physical distance between MYC and its various translocation partners in karyotypically normal MC/CAR cells and compared their physical proximity with the clinically observed frequencies of translocation using three spatial parameters (Fig. 2 and Table 1; ref. 4). First, we measured the average physical distance between translocation partners. MYC was separated from its two most frequent translocation partners, IGH and IGL, by 40.7% and 41.0% of the nuclear diameter, respectively (Table 1), whereas its separation from its rare translocation partner IGK was 47.1% (Table 1). This last value was similar to that observed for a negative control locus, TGFBR2, which has never been reported to translocate with MYC (mean separation 49.4%; Table 1). These differences were statistically significant by the Kolmogorov-Smirnov test (Supplementary Table 2 online). Second, to characterize the spatial relationship of potential translocation partners, we analyzed how frequently loci pairs are in close nuclear proximity. Intact gene loci in living cells can move within a radius of ∼2 µm18,19 and we therefore defined close loci pairs as two gene loci separated by 30% or less of the nuclear diameter (∼4.5 µm in MC/CAR cells), also taking into account the possibility that free DNA ends may be highly mobile. Whereas 32.7% of all MYC:IGH pairs and 28.4% of MYC:IGL pairs were close, only 20.8% of MYC:IGK pairs and

1National Cancer Institute and 2Mathematical and Statistical Laboratory, Division of Computational Bioscience, Center for Information Technology; National Institutes of Health, Bethesda, Maryland 20892, USA. Correspondence should be addressed to T.M. ([email protected]).

NATURE GENETICS VOLUME 34 | NUMBER 3 | JULY 2003

287

LETTERS BCL2 was near IGH more often than the negative control TGFBR2 (P = 3.0 × 10–7; Fig. 2c and Supplementary Table 3 online). Taken together, these observations indicate that the physical distance between translocationprone loci in normal cells correlates with their incidence of translocation in several B-cell cancers. To further probe the genome topology of IGL (22q11) IGH (14q32) BCL6 (3q27) lymphoma translocations, we asked whether close loci pairs were distributed uniformly f e d within the nucleus or were localized to specific nuclear subregions. We first analyzed how the radial distribution of individual locus alleles involved in close pairs compared to alleles not found in close pairs. Whereas most copies of the MYC locus predominantly localized towards the nuclear periphery (Fig. 1a,b), the subpopulation of IGH BCL6 IGL alleles in close association with IGH loci was markedly more likely to be localized in the nuclear interior (mean 58.6% versus 70.0% g h of radius; P = 4.6 × 10–5; Fig. 3a and 25 CCND1 (11p13) IGL (22q11) 25 BCL2 (18q24) IGH (14q32) Supplementary Table 5 online). Similarly, BCL6 (3q27) IGK (2p11) 20 20 TGFBR2 (3p23) MYC (8q24) BCL6 loci close to IGH were enriched near Random distribution Random distribution the nuclear center compared with the distri15 15 bution of BCL6 loci in distant pairs (mean 10 10 62.3% versus 72.9% of radius; P = 8.4 × 10–5; Fig. 3b and Supplementary Table 5 online). 5 5 As an alternative way of determining 0 0 whether close loci localized to nuclear subre0 20 40 60 80 100 120 0 20 40 60 80 100 120 Distance from center (% nuclear radius) Distance from center (% nuclear radius) gions, we analyzed the radial positioning of close pairs (Fig. 3c,d). The average radial Figure 1 Non-random radial positioning of gene loci. (a–c) FISH detection of gene loci. position of close pairs of the highly recurrent (d–f) Computational renderings to visualize radial locus distribution. The distribution of experimentally MYC:IGH pairs (52.1% of nuclear radius) measured positions (red) is compared with an expected simulated uniform random distribution (green). and the MYC:IGL pairs (52.0%) was signifiGene-specific patterns of radial distributions can be preferentially enriched in the nuclear interior (d), in medial nuclear subvolumes (e) or at the nuclear periphery (f). (g,h) Radial positioning of cantly closer to the nuclear center compared translocation-prone loci compared with a uniform random locus distribution (green). with the close pairs of the non-translocating control MYC:TGFBR2 (68.3%; Fig. 3c and Table 1 and Supplementary Table 6 online). Likewise, the radial posi22.2% of MYC:TGFBR2 pairs were close (Table 1). These differences tioning of BCL1:IGH close pairs (45.4%) and BCL2:IGH close pairs were statistically significant; close MYC:IGH pairs occurred more fre- (54.5%) was enriched towards the nuclear center compared with close quently than close MYC:IGL (P = 0.0064) or MYC:IGK pairs (P = 4.71 pairs of the rarely translocating BCL6:IGH (59.0%) and the non× 10–5), and close MYC:IGK and MYC:TGFBR2 pairs occurred at sta- translocating TGFBR2:IGH (61.3%; Fig. 3d and Table 1 and tistically indistinguishable frequencies (Fig. 2c and Supplementary Supplementary Table 6 online). These data indicate that the spatial organization of close locus pairs in the nucleus is non-random and Table 3 online). As a final measure for the relative proximity of loci we determined that recurrently close pairs are frequently positioned towards the the number of cells containing at least one close pair. We found at least nuclear center. Finally, we considered whether close locus positioning is a property one close loci pair in 79% of cells for MYC:IGH pairs and 74% of cells for MYC:IGL pairs, but in only 61% of cells for MYC:IGK pairs and of individual locus pairs or is due to higher-order genome organiza63% of cells for MYC:TGFBR2 pairs (Table 1 and Supplementary tion. To this end, we compared the relative distances of loci involved Table 4 online). We obtained similar results with the threshold for in translocations to loci that map on the same chromosomes but are not involved in recurrent translocations (Fig. 4 and Supplementary proximity set to 15% of the nuclear diameter (data not shown). To test whether the close proximity of loci involved in Burkitt’s lym- Table 3 online). The separation between the IGH locus and the nonphoma is also valid for other lymphomas, we investigated the position translocating TGFBR2 locus on the short arm of chromosome 3 is of the IGH locus relative to CCND1, BCL2 and BCL6 in normal similar to that between IGH and BCL6 on the long arm of chromoMC/CAR cells, as these loci translocate at differential frequencies in some 3, which is frequently involved in translocations (Fig. 4a and lymphomas of diverse topography6–8 (Fig. 2d and Table 1). We again Supplementary Table 2 online). Likewise, proximity between SOX7 found that high frequencies of translocations positively correlated (8p23) and TEP1 (14q11.2), not involved in translocations, is equivawith greater spatial locus proximity (Table 1). BCL1:IGH loci pairs lent to the separation of MYC:IGH, which map to the same chromowere close more frequently than BCL6:IGH pairs (P = 5.8 × 10–10), and somes but undergo translocations with high frequency (Fig. 4b).

288

c

b

% Occurrence

% Occurrence


a

VOLUME 34 | NUMBER 3 | JULY 2003 NATURE GENETICS

Figure 2 Spatial proximity of translocation-prone oncogene loci. (a,b) Two-color FISH to detect the relational positioning of loci frequently disrupted in lymphomas. (a) MYC (red), IGH (green); (b) MYC (red), TGFBR2 (green). (c,d) Cumulative frequencies of close pair separation as a fraction of all possible locus pairings. (c) In Burkitt’s lymphoma, locus breakpoints of highly recurrent translocations (MYC:IGH) are more frequently found in close pairs than are variant translocation partners (MYC:IGL or MYC:IGK) or a negative control (MYC:TGFBR2). (d) Spatial proximity is observed for translocations in diverse lymphomas, as frequent partners (CCND1:IGH, BCL2:IGH) are in recurrent proximity more often than rare exchange partners (BCL6:IGH, TGFBR2:IGH).

b

a

MYC:IGH

MYC:TGFBR2

c

d

40

40

CCND1:IGH MYC:IGH These data are consistent with reported close BCL2:IGH MYC:IGL MYC:IGK BCL6:IGH positioning of translocation-prone chromoMYC:TGFBR2 TGFBR2:IGH 30 30 somes16 and suggest that relational positioning of loci is largely determined by higher 20 20 order chromosome territory positioning. Here we report evidence that spatial 10 10 genome topology is a contributing factor in the formation of specific cancerous chromo0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 somal translocations. We show that physical Separation (% of nuclear diameter) Separation (% of nuclear diameter) proximity of potential translocation partners correlates with their observed clinical incidence of translocation. The correlation is not directly proportional because translocations resulting in more severe undergo translocations (Supplementary Fig. 1 online). The most phenotypes are probably overrepresented in the clinical data set. In likely mechanistic explanation for a proximity effect is that two close addition to spatial positioning, other factors, such as the repair capac- loci have a greater probability of making physical contact and thereby ity of a cell, clearly also contribute to translocation frequencies. engaging in illegitimate repair. The frequent localization of close pairs Furthermore, the role of spatial proximity might be partially abro- to an interior nuclear subvolume allows for the possibility that translogated by the ability of free DNA ends to explore a relatively large sub- cations preferentially occur in a subsection of the nucleus. volume of the nucleus18,19. Regardless, together with earlier reports on Preferential spatial organization of translocation-prone loci has positioning of translocation-prone chromosomes and loci in several implications for the genesis of cancers16. Changes in genome topology tissues12,14,16,20,21, these observations support the notion that proxim- during differentiation may favor certain stage-specific lymphomas conity effects have a global role in the formation of genome rearrange- taining distinct chromosome rearrangements. For example, in developments. The possibility that spatial proximity may also contribute to ing B cells, immunoglobulin genes localize towards the nuclear interior tissue specificity of translocations is supported by our observation that during receptor class switching, when primary translocations in mantle in comparison to lymphocytes, MYC and IGH loci are further sepa- lymphoma involving CCND1 and follicular lymphoma involving BCL2 rated from each other in normal fibroblasts where they do not are thought to occur22–24. Peripheral repositioning of immunoglobulin

Occurrence (% of all pairs)

Occurrence (% of all pairs)


LETTERS

Table 1 Proximity of translocating gene loci in normal lymphoblasts

Burkitt’s lymphoma MYC:IGH

Clinical frequencya (reported cases)

Mean separation (% nuclear diameter)

Pair frequencyb (% of all pairs)

Cell frequencyc (% of cells)

Average radial positiond (% nuclear radius)

52.1

123

40.7

32.3

79.2

MYC:IGL

11

41.0

28.4

74.0

52.0

MYC:IGK

4

47.1

20.8

63.3

60.1

MYC:TGFBR2

0

49.4

22.2

61.0

68.3

B-cell lymphoma CCND1:IGH

446 (84 CLL)

36.7

37.8

83.3

45.4

BCL2:IGH

863 (24 CLL)

39.8

32.3

77.6

54.5

BCL6:IGH

52 (0 CLL)

43.0

29.67

72.2

59.0

0 (0 CLL)

45.5

28.6

68.3

61.3

TGFBR2:IGH aCase

numbers of specific translocations in lymphoma were obtained from the Mitelman Online Database of Chromosome Aberrations in Cancer (ref. 4). CLL, chronic lymphocytic leukemia. bThe fraction of all possible locus pairs that are close. cThe percentage of cells that contain at least one close pair. dAverage position of the radially distributed mean coordinate centers of closely paired loci.


289

Figure 3 Close locus pairs are distributed nonrandomly. (a,b) Radial distribution of locus alleles involved in close pairs compared to locus alleles not in close pairs. MYC alleles (a) and BCL6 alleles (b) are found more often in the nuclear center when in proximity to IGH. (c,d) Radial distribution of close locus pairs. (c) In Burkitt’s lymphoma, the close pairs of loci involved in translocations are found more often in the nuclear center in comparison with the MYC:TGFBR2 close pairs (green) not involved in translocations. (d) An interior bias of frequent translocation partners is also observed for locus breakpoints in other B-cell lymphomas.

a

b 25

30

MYC:IGH close MYC:IGH distant

% Occurrence

% Occurrence

20

15

10

BCL6:IGH close BCL6:IGH distant

25 20 15 10

5

5 0

0 0

20

40

60

80

100

0

120

Distance from center (% nuclear radius)

40

60

80

100

120

d 25

25

MYC:IGH MYC:IGL

20

% Occurrence

% Occurrence

20

Distance from center (% nuclear radius)

c MYC:IGK

CCND1:IGH BCL2:IGH

20

BCL6:IGH

alleles after class switching, however, might MYC:TGFBR2 TGFBR2:IGH 15 15 then favor translocations with other partners, such as BCL6, as observed in later-stage diffuse 10 10 large-cell lymphoma23,25,26. Our observations 5 5 are consistent with such a model in which 0 0 cell-type-specific genome topologies place 0 20 40 60 80 100 120 0 20 40 60 80 100 120 Distance from center (% nuclear radius) Distance from center (% nuclear radius) oncogene loci in distinct and differential proximities to potential translocation partners. Proximity effects inherent in a precancerous genome may be of etiological importance for cancerous METHODS transformation, as disruption of specific oncogenes is often associated Cell lines. We grew MC/CAR cells (ATTC) in RPMI 1640 medium supple–1 –1 with discrete clinical outcomes. Although our evidence suggests that mented with 20% fetal bovine serum, 100 U ml penicillin, 100 µg ml strepL -glutamine. MC/CAR cells are EBV transformed, diploid tomycin and 1 mM normal spatial organization of the genome contributes to acquisition of and karyotypically normal. cancerous genome aberrations, the molecular mechanisms that shape normal and cancerous genomic landscapes remain to be identified. FISH and quantitative measurements. We harvested unsynchronized cells by

a Occurrence (% of all pairs)

120

BCL6:IGH TGFBR2:IGH

100

TGFBR2 80

60

40

20

BCL6 IGH 0 0

20

40

60

80

100

120

Chr. 3 Chr. 14

Separation (% of nuclear diameter)

b 120

MYC:IGH SOX7:TEP1 Occurrence (% of all pairs)


LETTERS

100

SOX7 80

60

TEP1

40

MYC 20

IGH

Chr. 8 Chr. 14

0 0

20

40

60

80

100

120

Separation (% of nuclear diameter)

Figure 4 Relational topology of gene loci is determined by higher-order genome organization. Cumulative separation distribution of the translocating IGH:BCL6 (a) and IGH:MYC pairs (b) compared with the non-translocating IGH:TGFBR2 (a) and SOX7:TEP1 pairs (b) located on the same respective chromosomes.

290

centrifugation and fixed them in methanol:acetic acid (3:1). For hybridizations, we obtained genomic DNA probes from the following bacterial artificial chromosome clones in the RP11 sequencing library: 454M12 (IGH), 29I20 (IGL), 319A7 (IGK), 237F24 (MYC), 675B4 (CCND1), 88J4 (BCL2), 88P6 (BCL6), 11L6 (TGFBR2), 981G7 (SOX7) and 324B11 (TEP1). We chose clones using LocusLink and Genome Mapviewer. Detailed protocols for FISH and probe generation are given in ref. 16. We produced probes by nick translation using dUTP conjugated with digoxigenin, biotin or Spectrum Red (Vysis). We hybridized probe combinations to cells overnight in a moist chamber at 37 ºC and detected digoxigeninand biotin-labeled probes with antibody to digoxigenin conjugated with fluorescein isothiocyanate (Roche) and streptavidin conjugated to Texas Red (Molecular Probes), respectively. We used DAPI to stain nuclear DNA. We examined hybridization specimens with a Nikon Eclipse E800 microscope equipped with epi-fluorescence optics and a Photometrics MicroMax cooled CCD camera controlled by Metamorph 4.96. We obtained three-dimensional z-stacks for approximately 70 complete optical field views (60×) per experiment. We collected nuclei in these images that gave two distinct hybridization signals per probe into two-dimensional maximum intensity projection image collages. We then measured geometrical coordinate centers of loci and nuclei using manual intensity thresholds and an automated Metamorph analysis tool. We calculated nuclear radii (Rn) and diameters (Dn) from Rn = (A/pi)0.5, where A is the thresholded nuclear pixel area. We calculated radial locus position as ((L:C)/Rn × 100), where (L:C) is the distance from the center of an individual locus to the nuclear center. We normalized the absolute spatial separations between locus pair centers (L1:L2) as a fraction of nuclear diameter ((L1:L2)/Dn) to account for natural variations in nuclear size, which may influence relative positioning. We analyzed the radial position of close locus pairs after calculating the mean coordinate position ((XL1 + XL2)/2; (YL1 + YL2)/2) of loci separated by less than 30% of the nuclear diameter. We used Excel for calculations and graphs. To determine the radial positioning of loci pairs, we used the center of mass along the axis between the two loci. For all quantitative analyses, we used at least 500 cells each containing four loci and four loci pairs from at least three experiments.

VOLUME 34 | NUMBER 3 | JULY 2003 NATURE GENETICS

LETTERS


Statistical analysis. To characterize global aspects of radial locus distribution and loci separations, we used the Kolmogorov-Smirnov test to compare the corresponding cumulative distributions27,28. The null hypothesis is that the experimentally measured distributions are produced by the same underlying distribution (see Supplementary Methods for details.) To characterize specific trends in radial distribution and loci separations, we conducted contingency table χ2 analysis27,29. To test differences in close pair occurrence, we constructed tables for the number of experimentally observed pairs that were separated by less than 30% of the nuclear diameter. The null hypothesis of this test is that the number of close pairs of two given types of loci is independent of the identities of the loci. URLs. LocusLink is available at http://www.ncbi.nih.gov/LocusLink/. Genome Mapviewer is available at http://www.ncbi.nlm.nih.gov/mapview. The Mitelman Online Database of Chromosome Aberrations in Cancer is available at http://cgap.nci.nih.gov/Chromosomes/Mitelman. Note: Supplementary information is available on the Nature Genetics website. COMPETING INTERESTS STATEMENT The authors declare that they have no competing financial interests. Received 18 March; accepted 12 May 2003 Published online 15 June 2003; doi:10.1038/ng1177 1. Rowley, J.D. The critical role of chromosome translocations in human leukemias. Annu. Rev. Genet. 32, 495–519 (1998). 2. Knudson, A.G. Two genetic hits (more or less) to cancer. Nat. Rev. Cancer 1, 157–162 (2001). 3. Elliott, B. & Jasin, M. Double-strand breaks and translocations in cancer. Cell Mol. Life Sci. 59, 373–385 (2002). 4. Mitelman, F. Recurrent chromosome aberrations in cancer. Mutat. Res. 462, 247–253 (2000). 5. Boxer, L.M. & Dang, C.V. Translocations involving c-myc and c-myc function. Oncogene 20, 5595–5610 (2001). 6. Harris, N.L. et al. New approaches to lymphoma diagnosis. Hematology 2001, 194–220 (2001). 7. Siebert, R., Rosenwald, A., Staudt, L.M. & Morris, S.W. Molecular features of B-cell lymphoma. Curr. Opin. Oncol. 13, 316–324 (2001). 8. Willis, T.G. & Dyer, M.J.S. The role of immunoglobulin translocations in the pathogenesis of B-cell malignancies. Blood 96, 808–822 (2000). 9. Bartova, E. et al. The influence of the cell cycle, differentiation and irradiation on the nuclear location of the abl, bcr and c-myc genes in human leukemic cells. Leuk. Res. 24, 233–241 (2000).


10. Bartova, E. et al. Nuclear topography of the c-myc gene in human leukemic cells. Gene 244, 1–11 (2000). 11. Kozubek, S. et al. 3D Structure of the human genome: order in randomness. Chromosoma 111, 321–331 (2002). 12. Neves, H., Ramos, C., da Silva, M.G., Parreira, A. & Parreira, L. The nuclear topography of ABL, BCR, PML, and RARα genes: evidence for gene proximity in specific phases of the cell cycle and stages of hematopoietic differentiation. Blood 93, 1197–1207 (1999). 13. Parreira, L. et al. The spatial distribution of human immunoglobulin genes within the nucleus: evidence for gene topography independent of cell type and transcriptional activity. Hum. Genet. 100, 588–594 (1997). 14. Lukasova, E. et al. Localisation and distance between ABL and BCR genes in interphase nuclei of bone marrow cells of control donors and patients with chronic myeloid leukaemia. Hum. Genet. 100, 525–535 (1997). 15. Sachs, R.K., Chen, A.M. & Brenner, D.J. Proximity effects in the production of chromosome aberrations by ionizing radiation. Int. J. Rad. Biol. 71, 1–19 (1997). 16. Parada, L., McQueen, P., Munson, P. & Misteli, T. Conservation of relative chromosome positioning in normal and cancer cells. Curr. Biol. 12, 1692–1697 (2002). 17. Parada, L. & Misteli, T. Chromosome positioning in the interphase nucleus. Trends Cell Biol. 12, 425–432 (2002). 18. Chubb, J.R., Boyle, S., Perry, P. & Bickmore, W.A. Chromatin motion is constrained by association with nuclear compartments in human cells. Curr. Biol. 12, 439–445 (2002). 19. Vazquez, J., Belmont, A.S. & Sedat, J.W. Multiple regimes of constrained chromosome motion are regulated in the interphase Drosophila nucleus. Curr. Biol. 11, 1227–1239 (2001). 20. Nikiforova, M.N. et al. Proximity of chromosomal loci that participate in radiationinduced rearrangements in human cells. Science 290, 138–141 (2000). 21. Haigis, K.M. & Dove, W.F. A Robertsonian translocation suppresses a somatic recombination pathway to loss of heterozygosity. Nat. Genet. 33, 33–39 (2003). 22. Kosak, S.T. et al. Subnuclear compartmentalization of immunoglobulin loci during lymphocyte development. Science 296, 158–162 (2002). 23. Zhou, J., Ermakova, O.V., Riblet, R., Birshtein, B.K. & Schildkraut, C.L. Replication and subnuclear location dynamics of the immunoglobulin heavy-chain locus in B-lineage cells. Mol. Cell. Biol. 22, 4876–4889 (2002). 24. Stamatopoulos, K. et al. Molecular insights into the immunopathogenesis of follicular lymphoma. Immunol. Today 21, 298–305 (2000). 25. Skok, J.A. et al. Nonequivalent nuclear location of immunoglobulin alleles in B lymphocytes. Nat. Immunol. 2, 848–54 (2001). 26. Yoshida, S. et al. Identification of heterologous translocation partner genes fused to the BCL6 gene in diffuse large B-cell lymphomas: 5′-RACE and LA–PCR analyses of biopsy samples. Oncogene 18, 7994–7999 (1999). 27. Chapter 14. Statistical description of data.in Numerical Recipes in C (eds. Press, W.H., Teuolskky, S.A., Vettering, W.T. and Flannery, B.P.) 609–655 (Cambridge University Press, Cambridge, 1992). 28. Stevens, M.A. Kolmogorov-Smernov Statistics. in Encyclopedia of Statistical Science vol. 4 (eds. Kotz, S. and Johnson, N.L.) 393–396 (Wiley and Sons, New York, 1982). 29. Fienberg, S.E. Contingency Tables. in Encyclopedia of Statistical Science vol. 2 (eds. Kotz, S. and Johnson, N.L.) 161–171 (Wiley and Sons, New York, 1982).

291