LEADING ARTICLE a New Recurrent and Specific Cryptic Translocation, T

A new recurrent and speciﬁc cryptic translocation, t(5;14)(q35;q32), is associated with expression of the Hox11L2 gene in T acute lymphoblastic leukemia OA Bernard1, M Busson-LeConiat1, P Ballerini2, M Mauchauffe´ 1, V Della Valle1, R Monni1, F Nguyen Khac1, T Mercher1, V Penard-Lacronique1, P Pasturaud3, L Gressin3, R Heilig4, M-T Daniel5, M Lessard6 and R Berger1

1U434 INSERM-CEPH and SD401 No. 434 CNRS, Paris; 2Service d’He´matologie Biologique, Hoˆpital Trousseau, Paris; 3CEPH, Paris; 4Ge´noscope, Evry; 5Laboratoire Central d’He´matologie, Hoˆpital Saint-Louis, Paris; and 6Laboratoire d’He´matologie, Hoˆpitaux Universitaires de Strasbourg, France

FISH identified a cryptic t(5;14)(q35;q32) in T acute lym- (Tables 1 and 2). As controls, 10 children with B cell lineage phoblastic leukemia (ALL), whereas it was not observed in ALL were also examined. Patients 6 to 8, identified as carrying B ALL samples. This translocation is present in five out of 23 (22%) children and adolescents with T ALL tested. the t(5;14) in the Laboratoire d’He´matologie (Strasbourg, RanBP17, a gene coding for a member of the importin ␤ pro- France), are reported elsewhere (He´lias et al., submitted for tein family, and Hox11Like2, an orphan homeobox gene publication). were mapped close to the chromosome 5 breakpoints and CTIP2, which is highly expressed during normal T cell differentiation, was localized in the vicinity of the chromosome Cytogenetic studies 14 breakpoints. The Hox11L2 gene was found to be tran- scriptionally activated as a result of the translocation, prob- ably under the influence of CTIP2 transcriptional regulation Cytogenetic studies were performed in patients examined in elements. These data establish the t(5;14)(q35;q32) as a Hoˆpital Saint-Louis on bone marrow and/or blood cells after major abnormality, and Hox11 family member activation as short-term culture for 17 and 24 h. In addition, PHA- an important pathway in T ALL leukemogenesis. Leukemia stimulated blood samples were also examined after 72 h in (2001) 15, 1495–1504. vitro culture in three patients. RHG bandingtechniques were Keywords: T ALL; FISH; t(5;14); CTIP2; Hox11L2 applied in every case, includingthe three patients from Stras- bourg. Karyotypes are summarized in Tables 1–3. Introduction Fluorescence in situ hybridization (FISH) T cell acute lymphoblastic leukemia (T ALL) is associated with a normal karyotype in 25 to 40% of patients, a higher percent- FISH analyses were carried out with the usual techniques.5 In 1–4 age than in B cell lineage ALL. Fluorescence in situ addition to whole chromosomes 5 and 14 paintingprobes hybridization (FISH) techniques have improved the detection (from INSERM U301 and Appligene Oncor, respectively), of subtle chromosome abnormalities, allowingthe identifi- YACs (CEPH library) located to 5q34-q35 were used as probes 5 cation of the B ALL specific t(12;21)(p13;q22), but spectral in FISH studies on metaphases of leukemic samples. BAC karyotype analysis of T ALL samples did not uncover new clones, chosen after consultation of the human chromosome 6 recurrent chromosomal abnormalities. Since abnormalities of 5 map, were then used in FISH experiments in order to refine chromosome 14, most of them affectingthe TCR genes,are the localization of the breakpoints on chromosome 5 common in T ALL, we searched for abnormalities of this chro- (Figure 1a). To locate the breakpoints on chromosome 14 a mosome in patients with T ALL usingFISH techniques. We series of probes coveringthe IGH locus to 14q32 (BAC now report the identification and characterization of a pre- 158A2, gift from S Romana (Hoˆpital Necker, Paris, France), viously undescribed recurrent chromosomal translocation. and Cos a1+a2, gift of European Concerted Action, Marseille, France), the TCL1 locus (BACs 1090N10 and 164H13) and the AKT1 locus (BAC 940A3) was used first. Other BAC clones Materials and methods were selected from available chromosome 14 maps or isolated upon PCR screeningof the CEPH library and used to Patients locate the breakpoints on chromosome 14 (Figure 1b). To delineate more precisely the localization of the rearrange- Thirty patients with T ALL, previously cytogenetically exam- ments, selected BAC clones were hybridized to metaphases ined in Hoˆpital Saint-Louis, Paris, were selected on the basis of patients with t(5;14). of cell pellet availability and prepared for cytogenetic studies. There were 16 children between 3 and 14 years of age, seven adolescents (15 and 17 years) and seven adults (18–37 years) Molecular studies

Enough frozen material was available for extensive molecular studies for patients 3 and 4. RNA was extracted usingcesium chloride method or Rnable reagent (Eurobio, Les Ullis, France) Correspondence: R Berger, U434 INSERM-CEPH, 27 rue Juliette and analyzed accordingto standard protocols. PCR with pri- Dodu, 75010 Paris, France; Fax: 331 53 72 51 92 mers A+B and C+D (see below) were performed to isolated Received 19 June 2001; accepted 30 June 2001 two probes correspondingto the 3 ′ untranslated region of Hox1 1-like gene activation in T ALL OA Bernard et al 1496 Table 1 Karyotypes of 25 patients with T cell ALL

Patient Sex/age (years) Karyotype

99 078 M/11 46,XY,del(9)(p21),del(11)(q13)[8]/46,XY[2] 97199 F/6 46,XX[12] 95172 M/37 46,XY[20] 9549 M/12 46,XY,t(11;14)(p13;q11)[15] 94 256 M/23 46,XY[7] 98 089 M/16 46,XY[14] 98099 M/14 46,XY[27] 97185 M/9.5 46,XY,t(10;13)(p12;q14)[2]/46,XY[15] 9641 M/17 46,XY,del(6)(q15q24)[7]/46,XY[9] 961 F/24 47,XX,+13[2]/46,XX[11] 99099 M/10.5 46,XY[20] 4875 M/25 46,XY,del(7)(p13)[2]/46,XY[7] 7165 M/19 46,XY[6] M/21 (relapse) 46,XY,del(6)(q21)[7]/46,XY[20] 94 290 M/11 46,XY,add(1)(q44),del(7)(q31),add(8)(q24),+10[23] 99071 M/3 46,XY,del(6)(q21),del(9)(p12)[8] 4682 F/5 46,XX[[13] 3529 F/16 88–90 (mar)[24]/46,XX[1] 5958 F/25 46,XX,del(1)(p33),del(5)(q13),del(11)(q13)[31]/46,XX[1] 5817 M/10 46,XY[8] 2691 M/6 45,XY,−6,−14,del(2)(p11p21),+r[43]/46,XY[1] 5949 M/3 46,XY,del(6)(q24),+mar[10]/46,XY[1] M/6.5 relapse 46,XY[15] 5852 M/9 46,XY[21] 98064 F/5 46,XX,del(9)(p21),+add(9)(p24)[9]/46,XX[6] 2008 F/15 46,XX,t(5;10)(q35,q21)[37] 99103 M/12 46,XY,del(5)(q14q31),del(11)(q13),add(14)(q32),del(15)(q24)[6]/46,XY′[10]

Table 2 Hematologic data of 8 patients with ALL and t(5;14)

Patient Sex/age (y) Bone marrow % Blood leukocytes % Blasts Hemoglobin Platelets FAB blasts ×109/l g/dl ×109/l

1a M/13 94 30 55 12.3 56 L2 1b M/15 86 12.9 13 234 2 M/18.5 78 27 64 15.2 158 L2 3 M/16 100 367 94 4.9 35 L1 4 M/9 90 187 93 6.6 54 L2 5 F/15 91 12 43 9.8 268 L2 6c M/7 85 38.8 10 13.1 77 7c M/6 91 77.9 84 9.9 146 8c M/5 80 51 85 4.4 16

aRelapse bSecond relapse cPatients 6, 7, and 8 are numbered 1, 2, and 3 in the article by He´lias et al (submitted for publication).

Table 3 Cytogenetic data of ﬁve patients with T-ALL and t(5;14)

Patient Materiala Karyotype

1 954310b Md/B17 46,XY[10/34] 97142c M 24 47,XY,+8,add(14)(q32)[5]/46,XY[16] 2 99088 M24,B24 46,XY[12],−46,XY[36] 3 8464 M17-B24,48 46,XY[3],−46,XY[5] 4 9638 B17 46,XY[11] 5 97184 B24/B72+PHA 46,XX[13]/46,XX[8]

aMd, M17, M24 B17, B24, B48: bone marrow direct examination (Md). M and blood (B) cell cultures for 17, 24 and 48 h. bRelapse. cSecond relapse. The karyotypes of patients 6–8 are shown in the article by He´lias et al (submitted for publication).

Leukemia Hox1 1-like gene activation in T ALL OA Bernard et al 1497

Figure 1 FISH analysis of the t(5;14) translocation. (a) Map of the chromosome 5 BAC clones used in this study. The map is derived from data obtained in Genebank (http://www.ncbi.nlm.nih.gov/) and from chromosome 5 maps (http://www-hgc.lbl.gov/). The name of the BACs and the accession number of the corresponding sequences are indicated. The arrowheads point to the BAC clones that give split signals on indicated patients’ metaphases. Location and orientation of the RanBP17 and Hox11L2 genes are shown, but not to scale. Note that data are derived from unﬁnished BAC sequences and overlaps between those sequences are not to scale. Other BAC clones were located usingBAC ends sequence data. (b) Map of the chromosome 14 BAC clones used in this study. The map is derived from data obtained in Genebank and from Genoscope (http://www.genoscope.cns.fr). The name of the BACs and the accession number of the corresponding sequences are indicated. The arrowheads point to the BAC clones that give split signals on indicated patients’ metaphases. Location and orientation of the human CTIP2 gene is indicated but is not drawn to scale.

Hox11L2. Anchored PCR was performed as described using Results oligodT priming and 5–59 and 5–60 for the first and second PCR, respectively. FISH analysis usingchromosome 14 paintingprobe was per- Bispecific RT-PCR experiments were performed starting formed in a first step on the 30 patients with T ALL who were from random-primed cDNA. The specificity of the amplified previously cytogenetically examined in our laboratory. In five fragments was checked by direct nucleotide sequencing. of them (Tables 2 and 4) a translocation t(5;14), undetected Primer used are as follows: 5–61: gcgaattcTGGTA with bandingtechnique alone, was observed. A chromosome GATCTGGGTGAAGATG; 5–35: CCAGTCAAACAGCA 5 paintingprobe was used to confirm the reciprocal t(5;14) in TGGTGT; 5–59 (AATGACCTTTCTGTTGGTTA); 5–60 these patients. The t(5;14) was not detected in 10 children (gcgaattcAAAACCACACGAGTGAACAC; nucleotides in low- with B type ALL. ercase letters were added for cloningpurposes); Hox11L2F: GCGCATCGGCCACCCCTACCAGA; Hox11L2R: CCGCTCC GCCTCCCGCTCCTC; A: AGGTGGGCGGCGGGCAGAGTCC; Breakpoint on chromosome 5 B: GCCGGGGGTCGCCGAGCATTAT; C: AGGCCGTCCCCA GGTCAAATCCAC; D: CAGTCCCGCAGCCCGCATAGAACG. YAC clones mapped to the longarm of chromosome 5 were Computer analysis was performed locally or at the NCBI used in FISH experiments on patient samples. Out of several site (http://www.ncbi.nlm.nih.gov). Accession numbers for the YACs tested, split signals (hybridization signals on normal 5, BAC clones appear on the Figures. Others are as follows: rearranged 5 and 14) were observed using three YACs probes, NM 018014 (human BCL11A/CITP1), NM 016707 (mouse 885A6, 874C6, and 912B5, in patient 1. The first two YACs Evi9), AF186018 (CTIP1), NM 022898 (human generated signals of apparently similar intensity on both BCL11B/CTIP2), NM 021399 (mouse CTIP2), AB043584 derivative chromosomes, whereas YAC 912B5 consistently (human Rit alpha), AB043551 (mouse Rit alpha), CAA08834 generated a smaller signal on the derivative 5 than on deriva- (human Hox11), NP 068701 (mouse Hox11) AAC23900 tive 14. Splittingof YAC 885A6 was observed in the five (chicken Hox11), AAG14453 (xenopus Hox11), NP 001525 already identified patients (1 to 5), but neither in the 25 other (human Hox11L1), Q61663 (mouse Hox11L1), XP 003705 T ALL patients, nor in the 10 B ALL samples of this series. (human Hox11L2), NP 064300 (mouse Hox11L2), Usingthis YAC, split signalwas also observed in patients 6 to AAC23901 (chicken Hox11L2), AAG14452 (xenopus 8. In patient 8, two rearranged chromosomes 5 were present Hox11L2). while normal chromosome 5 was lacking.

Leukemia Hox1 1-like gene activation in T ALL OA Bernard et al 1498 Table 4 Immunophenotype of ﬁve patients with T ALL and t(5;14)

Patient CD1a CD2 cCD3 CD3 CD4 CD5 CD7 CD8 CD10 CD34 TdT DR TCR␣␤ TCR␥␦

1a ++ −+++++ 2 ++++/−+++++ −− 3 ++ ++++++/− 4 +− ++/−+ + + + + 5 ++ −+++++ −−

aRelapse. Immunophenotype of patients 6–8 are reported in the article by He´lias et al (submitted for publication).

To further localize the breakpoint on chromosome 5, BACs sequences correspondingto the recently described from the chromosome 5q35 map (see Figure 1a) were used, CTIP2/hRIT/BCL11B gene (see Ref. 7 and Materials and eventually allowingthe location of the breakpoints within methods). Comparison of cDNA and genomic sequences sequences covered by BAC 45L16. This BAC generated split allowed us to sketch the structure of the human CTIP2 gene signals in five patients with t(5;14) (2, 4, 6 to 8). Additional (Figure 2a) which is similar to the one of CTIP1/Evi9 gene.7,8 BAC clones were chosen on the basis of available end CTIP2 possesses at least three exons, a large 3′ exon on BAC sequences (2008H22, 2248N14) or isolated from the CEPH 3104H21 and two exons lyingfurther telomeric on BAC BAC library (593F7). In patient 1, splittingof the juxta telo- 889B13. Importantly, CTIP2 is oriented 5′ telomere-3′ centromeric BAC 2249B15 was obvious, with signals on the two mere, and apparently ends at nucleotide 223219 of BAC chromosomes 5 and one 14. An extra signal was always 3104H21. The murine COUP TF-interactingprotein 2 (CTIP2) observed on 10p12 in control and patient metaphases. BAC and the related CTIP1 are Kru¨ppel-like zinc finger proteins 593F7 generated split signals in patients 1, 2, 4, 6 to 8. BAC expressed in the brain and have been isolated because of their 2248N14 was split in patient 4. Taken together, these results interaction with the chicken ovalbumin upstream promoter showed that the chromosome 5 breakpoints lie within a transcription factor (COUP-TF) subfamily of orphan nuclear restricted area of chromosome 5 and the majority of them are receptors.7 Those receptors are generally considered to repress clustered in the sequences between the ends of the BACs transcription and CTIP1 has been shown to potentiate the 45L16 and 2248N14, between introns 21 and 24 of the repression by ARP1/COUP-TFII in an acetylase independant RanBP17 gene (see below). Due to the shortage of material, fashion.7 Evi9 was isolated as activated by retroviral insertion two patients (3 and 5) could not be extensively analyzed by in murine leukemia.8,9 The human homologous genes FISH. BCL11A for CTIP1/Evi9 and BCL11B/hRit for CTIP2 have also been isolated (see Figure 2b for a schema of the human CTIP2 protein). Breakpoints on chromosome 14 To investigate the expression pattern of CTIP2 in humans, the insert of an EST (H09748), derived from the CTIP2 locus To delineate the interval containingthe breakpoints on 14q32, was used as a probe in Northern blot experiments. A strong molecular probes encompassingthe IGH, TCL1, and AKT1 signal of approximate size of 9 kb was observed in normal loci were used at first. The FISH experiments demonstrated tissues containingT lymphocytes (thymus, peripheral blood, that the chromosome 14 breakpoints were located between spleen) and in brain (Figure 2c). It is also detected in T ALL- the AKT1 (centromeric) and TCL1 (telomeric) loci, without derived samples after only 4 h exposure (Figure 2d). On RNA evidence of direct involvement of these two genes. In order extracted from two t(5;14) samples (patients 3 and 4) and two to narrow the localization of the breakpoint, BAC clones T ALL samples (T1 and T2) devoid of t(5;14), the probe reacted selected from the chromosome 14 map were used as FISH with one or two species dependingon the samples. The mol- probes on t(5;14) bearingcells. The results are summarized ecular basis of these two species remains to be determined. in Figure 1b. No single BAC was found to encompass all the A larger transcript is observed in the lane corresponding to breakpoints. BAC 68I8 generated split signal in patient 2, BAC patient 4 confirmingthe direct involvement of the CTIP2 gene, 1082A3 gave split signal in patients 6 and 8, while BACs suggested by the location of the chromosome 14 breakpoint 1127D7 and 3104H21 were split in patients 1 and 4, respect- within BAC 3104H21 (see Figure 1b). Taken together, our data ively. In patient 7, the breakpoint was located within BAC suggest that human CTIP2 is highly expressed during normal 2576L4 (Figure 1b). Thus, the chromosome 14 breakpoints of and pathological T lymphoid differentiation, but do not sup- the seven patients tested are interspersed in a 700 kb area of port disregulation of this gene as an important recurrent chromosome 14 sequences. consequence of the t(5;14). A similar approach was undertaken to study the region around the breakpoint on chromosome 5. Interestingly, an EST Molecular studies cluster was found to lie within the 2249B15 sequences. This cluster turned out to correspond to the 3′ end of a gene enco- In order to identify a candidate gene involved in the t(5;14), dinga protein closely related to the RanBP16 protein and we identified and mapped transcribed sequences with respect hereafter called RanBP17.10,11 Comparison between RanBP16 to the chromosome 14 genomic sequences surrounding the and RanBP17 mRNA and genomic sequences allowed us to breakpoints. Extensive blast analysis did not allow the identi- determine the structure of the RanBP17 gene (see Figure 3a fication of known genes or reliable EST clusters within this and Table 5). It turned out to be quite large, more than 200 kb chromosomal region. At the telomeric end of the breakpoint lyingover three BAC clones, and to encompass most of the region, clone 3104H21 was found to contain transcribed chromosome 5 breakpoints. Expression pattern was investi-

Leukemia Hox1 1-like gene activation in T ALL OA Bernard et al 1499

Figure 2 Molecular analysis of the human CTIP2 gene. (a) Partial structure of the human CTIP 2 gene on chromosome 14, as deduced from comparison between human cDNA and genomic sequences. Exon sequences are in upper case letters and intron sequences are in lower case letters. The putative translational initiator condon is underlined. Predicted exons 1 and 2 are on BAC 889B13 sequences, whereas the third exon lies on the overlap between BACs 3104H21 and 2348N10. (b) Structure of the predicted human CTIP2 protein. Black boxes indicate the location of the predicted C2H2 zinc ﬁngers. (c) Nothern blot analysis of human CTIP2 expression in normal tissues using a probe derived from the 3′ untranslated sequences of the gene. This probe reacts with abundant 9 kb RNA species in thymus, spleen and peripheral blood lymphocytes (PBL). A faint signal is also detectable at 5 kb which might correspond to additional RNA species. (d) Northern blot analysis of human CTIP2 expression in malignant cell lines and patient samples. Loading is 15 ␮gof total RNA for all samples except Meg01and HEL, for which it was 2 ␮gof polyA + RNA. A short exposure time allows detection of two CTIP2 RNA species (shown by arrows) in some, but not all samples. T1 and T2 are control T ALL samples. CTIP2 expression is easily detected in Jurkatt (T ALL cell line) RNA upon longer exposure time. A larger, abnormal transcript, detected in patient 4 RNA is indicated by an arrowhead. gated using a 1.1 kb RanBP17 cDNA probe (exons 1 to 8). As size of the abnormal RanBP17 transcript in this patient. These shown Figure 3b, the probe reacted with a 4 kb species in data indicate the low level of expression of a truncated RNA extracted from several human tissues. When investigated RanBP17 transcript containingintron 20 in RNA of patient 4. for in human leukemic samples, RanBP17 expression was The lack of RanBP17 abnormal transcripts in patient 3 observed in some samples such as megakaryocytic leukemic material suggests that disruption of this gene is not a crucial cell lines (see Figure 3c). and recurrent consequence of the t(5;14). Based on these structural data, the t(5;14) could result in Further analysis of the available chromosome 5 genomic the dissociation of the body of the gene from its normal 3′ sequences surroundingthe breakpoint indicated the location end. We used anchored PCR techniques to isolate potentially of the Hox11L2 gene only 10 kb downstream of the 3’ end abnormal 3′ end of RanBP17 transcripts from patient 4 RNA. of RanBP17 in a similar orientation on the chromosome. The Sequence analysis of cloned cDNA fragments revealed prema- predicted product of Hox11L2 is closely related to the Hox11 ture termination of RanBP17 transcripts, which ends in the protein, whose gene is known to be rearranged in T ALL asso- 20th intron of the gene. No fusion cDNA containing chromo- ciated translocations.12–16 To check for Hox11L2 expression, some 14 sequences was isolated. RT-PCR experiments were RT-PCR experiments were performed which would amplify a performed startingfrom patients 3 and 4 and from two T ALL 242 bp Hox11L2 cDNA fragment. As shown in Figure 4b, the control samples (T1, T2) usingprimers designedto detect expected fragment was observed in lanes corresponding to intron 20 containingmature mRNA. Such mRNA species patients, 2, 3, 4 bearingt(5;14), but not from control T ALL could easily be detected in patient 4, but not in patient 3 samples or negative controls. The same results could be material nor in control samples (data not shown). RanBP17 obtained startingwith material from patients 6, 7 and 8 (data expression was also examined by Northern blot analysis of not shown). A mixture of two fragments from the 3′ untrans- RNA extracted from these samples, usingas a probe a 1.8 kb lated region of Hox11L2 was used in Northern blotting experi- cDNA fragment corresponding to exons 1 to 18. As shown ments, the result of which is presented Figure 4c. Small tran- Figure 3c, transcription of RanBP17 was not detectable in scripts (size Ͻ2 kb) are observed in lanes correspondingto those samples. In patient 4, a low expression could be patients 3 and 4, but not in other lanes such as control leu- observed at a size of 2.5 kb, in agreement with the predicted kemic samples (T1 and T2) or Jurkatt, Meg01 and HEL cell

Leukemia Hox1 1-like gene activation in T ALL OA Bernard et al 1500

Leukemia Hox1 1-like gene activation in T ALL OA Bernard et al 1501 Figure 3 Molecular analyses of the human RanBP17 gene. (a) Pro- Table 5 Relationships between RanBP17 exons and BACs used posed structure of the human RanBP17 gene on chromosome 5 as deduced from comparison between human cDNA sequences and BACs 455F18 212004 376I20 77B11 33L5 45L16 2249B15 chromosome 5 genomic sequences. Exon sequences are in upper case Exons letters and intronic sequences are in lower case letters. The putative translation initiator condon is underlined. (b) Northern blot analysis 1 ++ of RanBP17 expression in normal tissues usinga cDNA probe span- 2 ++ ningexons 1 to 8 of the gene.A 4 kb transcript is detected in some, 3 ++ but not all lanes and is shown by an arrow). (c) Northern blot analysis 4 ++ of RanBP17 expression in patient samples usinga cDNA probe span- 5 ++ ningexons 1 to 18 of the gene.A normally sized 4 kb transcript, indi- 6 ++ cated by an arrow, is detected in lanes correspondingto erythroid 7 ++ (HEL) and megakaryocytic cell lines (Meg01 and MO7E), but not in 8 ++ those correspondingto patient samples or Jurkatt. A faint smear of 9 ++ apparent size of 2.5 kb, is observed in the lane correspondingto 10 ++ patient 4 and is indicated by an asterisk. Difference in migration 11 ++ between Meg01 and HEL samples and the others is attributed to 12 + + polyA RNA loadinginstead of total RNA. (d) Nucleotide sequence 13 ++ of truncated RanBP17 transcript. Exon 20 sequences are underlined. 14 ++ Codingsequences appear in uppercase letters and non-coding 15 ++ sequences in lowercase letters. The underlined PolyA run is found in 16 ++ both cDNA fragments and intron 20 sequences. The methionine 17 ++ (amino acid 744 of the normal predicted RanBP17 protein), encoded 18 ++ by the last codon of exon 20 is indicated. 19 ++ 20 + 21 + + lines. Taken together our data suggest that the t(5;14) results 22 23 ++ in the ectopic activation of the Hox11L2 gene. 24 ++ 25 ++ 26 + Discussion 27 + 28 + Rearrangements of chromosome 5 involving various breakpoint localizations have been reported in various subtypes of + indicates the presence of a given exon within the BACs ALL.17 Translocation t(5;14)(q31;q32) which juxtaposes the sequence. immunoglobulin heavy chain and the IL3 genes, resulting in overexpression of the latter is amongthe first recurrent translocations involving5q described in B ALL. 18 The development moiety as a result of the t(5;14). Based on its high similarity of FISH techniques makes it easier to recognise rearrange- with RanBP16,10,11 RanBP17 encodes a member of the ments undetected by chromosome bandingtechniques alone, importin-␤ superfamily of nuclear transport receptors (or as exemplified by the t(12;21)(p13;q22) in childhood B ALL5 karyopherins), which are involved in the transport of proteins and the t(5;11)(q35;p15.5) in childhood acute myeloid leuke- through nuclear pores of the nucleocytoplasmic membrane mia (AML).19 (see Ref. 20 for review). It could not be unambiguously dem- T ALL is known to have a peculiar high frequency of ‘nor- onstrated that the rearrangement of RanBP17 is the important mal karyotype’. We used a chromosome 14 specific FISH result of the t(5;14). It is not constantly overexpressed, as a probe to analyze the status of this chromosome in patients normal or truncated species in the t(5;14) samples tested. It with T ALL. We thereby uncovered a new cryptic translo- cannot be fused to CTIP2 to result in a hybrid gene since both cation t(5;14)(q35;q32) present in both children and ado- genes are inversely oriented on the chromosomes and we did lescent T ALL samples, which appears to be frequent since it not find any evidence of inversion at the chromosome break- was found in five out of 30 patients (16.7%) and in five of 23 points. Because it is not detectably expressed in T lympho- (22%) in children and adolescents in our series. This translo- cytes, RanBP17 is unlikely to be inactivated by the translocation would be specific since it was not observed in adults cation. In fact, the expression of RanBP17 in some human with T ALL or in children with B ALL. Despite slight variations leukemic cell lines suggests that transcription of this gene dur- from one patient to another, the t(5;14) samples exhibited ingsome steps of hematopoietic differentiation could favor the immature cortical T cell markers (CD1a). At the molecular occurrence of the t(5;14) by openingthe chromatin structure cytogenetic level, most of the t(5;14)(q35;q32) breakpoints are in this region of chromosome 5. clustered on band 5q35 within a single BAC clone in five out The Hox11L2 gene which lies telomeric to RanBP17 on of six patients, whereas the chromosome 14 breakpoints are chromosome 5 seems to be the important gene affected by spread out alongseveral hundred kilobases on band 14q32 the translocation. Indeed, transcription of Hox11L2 could be and their localization by FISH was only possible by usingsev- demonstrated in the five patients with t(5;14) studied, but not eral BACs. The practical consequence is that paintingprobes in T ALL patients’ samples lackingthis translocation. Interest- or a combination of a YAC probe for chromosome 5 and ingly, the t(5;14)(q33;q11), also observed in T ALL,21 but paintingprobe for 14 are, to date, the most efficient tools to involvingthe TCR ␣/␦ locus on chromosome 14, has been detect the translocation. BAC probes would be used in a reported to affect the RanBP17 gene on chromosome 5.10 We second step to refine the chromosomal breakpoint localiz- did not investigate such samples in our series, but it would be ation. of interest to analyze Hox11L2 expression in these patients. On 5q35, the RanBP17 gene, which lies in a 5′ telomere- Hox 11L2, alongwith Hox11 and Hox11L1, belongsto a 3′ centromere orientation, is structurally altered within its 3′ three-member family of homeobox-encodinggeneswhich are

Leukemia Hox1 1-like gene activation in T ALL OA Bernard et al 1502

localized outside the four homeobox genes clusters.16 The are expressed essentially in neural tissues duringdevelopment HOX11 protein has been shown to regulate transcription and also in liver and pancreas of adult mice.26–30 The corre- through site-speciﬁc DNA binding and to possess transforming spondingknock-out models indicated a very speciﬁc role of properties in animal models.16,22–25 All three Hox11 family these proteins duringnormal embryogenesis.Hox11 is a cen- members exhibit a restricted pattern of expression, since they tral player in controllingspleen formation, perhaps through

Leukemia Hox1 1-like gene activation in T ALL OA Bernard et al 1503 Figure 4 Molecular analysis of the human Hox11L2 gene. (a) Par- abnormalities. In addition, it should be kept in mind that other tial map of the RanBP17 and Hox11L2 genes on chromosome 5. ‘cryptic’ rearrangements may await characterization. Finally, Exons appear as empty boxes. The transcribed sequences (hatched the description of this new cryptic translocation will lower the box) isolated in patient 4 and the cryptic polyadenylation site present frequency of the so-called ‘normal karyotypes’ in T ALL in intron 20 are shown. The extremities of some of the BAC clones samples. Molecular tools described here will now allow the used in this study are indicated (see also Table 5). The majority (but not all) of the chromosome 5 breakpoints is predicted to lie within a accurate estimation of the frequency of Hox11 gene family restricted area, between introns 20 and 24 of RanBP17, upstream of activation in lymphoid malignancies, its specificity and poten- the Hox11L2 sequences. (b) RT-PCR analysis of Hox11L2 expression tial prognostic significance. in malignant samples. A specific 242 bp fragment is amplified from patients with t(5;14), but not from two control samples (T1 and T2) or Hela. A human ARNT cDNA fragment was amplified separately to ensure the integrity of the RNA samples (data not shown). −, No tem- Acknowledgements plate. (c) Northern blot analysis of Hox11L2 expression in patient ′ samples. A probe correspondingto the predicted 3 untranslated This work was supported by INSERM and the Ligue Nationale region of Hox11L2 (see Materials and methods) was used to probe the same membrane as Figures 2d and 3c. Note that the normal Hox11L2 Contre le Cancer. Thomas Mercher and Richard Monni are transcript is not yet known. (d) Comparison of Hox11 family member recipients of fellowships from the Ministe`re de l’Education proteins. The conserved regions and the homeobox domain are under- Nationale. Florence Nguyen Khac is supported by the Acade´- lined (adapted from quoted reference). The conserved region C region mie Nationale de Me´decine. We thank W Vainschenker for encompass the pentapeptide potentially interactingwith the PBX providingus with the M70E cell line. proteins. regulating cell survival,31,32 and the two other genes appear References important for specific neuronal functions.33,34 Interspecies conservation of the primary sequences of these proteins is 1 Raimondi SC, Behm FG, Roberson PK, Pui CH, Rivera GK, Murphy observed outside the homeobox domain suggesting common SB, Williams DL. Cytogenetics of childhood T-cell leukemia. important biological functions (Figure 4d). It is noteworthy Blood 1988; 72: 1560–1566. 2 Berger R, Le Coniat M, Vecchione D, Derre J, Chen SJ. Cytogenetic that Hox11L2 appears more closely related to Hox11 than studies of 44 T-cell acute lymphoblastic leukemias. Cancer Genet Hox11L1. Because of their frequent and specific involvement Cytogenet 1990; 44: 69–75. by transcriptional activation in lymphoid leukemogenesis, the 3 Berger R. Cytogenetics in adult acute lymphoblastic leukemia. Rev Hox11L2 and Hox11 gene products deserve further investi- Clin Exp Hematol 1998; 5: 68–84. gation. 4 Heerema NA, Sather HN, Sensel MG, Kraft P, Nachman JB, Ste- Unexpectedly, extensive database searches did not allow inherz PG, Lange BJ, Hutchinson RS, Reaman GH, Trigg ME, Arthur DC, Gaynon PS, Uckun FM. Frequency and clinical sig- the identification of a gene, in the vicinity of the majority of nificance of cytogenetic abnormalities in pediatric T-lineage acute the chromosome 14 breakpoints. CTIP2, the closest gene, lies lymphoblastic leukemia: a report from the Children’s Cancer in a 5′ telomere-3′ centromere orientation, several hundred Group. J Clin Oncol 1998; 16: 1270–1278. kb telomeric to the area containingmost of the chromosome 5 Romana SP, Le Coniat M, Berger R. t(12;21): a new recurrent trans- 14 breakpoints. The CTIP2 gene encodes a member of the location in acute lymphoblastic leukemia. Genes Chromosomes zinc finger family of transcriptional regulators closely related Cancer 1994; 9: 186–191. 7 6 Rowley JD, Reshmi S, Carlson K, Roulston D. Spectral karyotype to CTIP1, which exhibits autonomous repressive properties. analysis of T-cell acute leukemia. Blood 1999; 93: 2038–2042. BCL11A, the human counterpart of CTIP1, is involved in a 7 Avram D, Fields A, Pretty On Top K, Nevrivy DJ, Ishmael JE, Leid t(2;14) translocation observed in B leukemia and lymphoma.35 M. Isolation of a novel family of C(2)H(2) zinc finger proteins Except in one instance, CTIP2 is not structurally altered as a implicated in transcriptional repression mediated by chicken oval- result of the t(5;14) in our patients, as judged from its presently bumin upstream promoter transcription factor (COUP-TF) orphan known structure. This gene is however expressed at very high nuclear receptors. J Biol Chem 2000; 275: 10315–10322. 8 Nakamura T, Yamazaki Y, Saiki Y, Moriyama M, Largaespada DA, level duringT lymphoid differentiation and this expression Jenkins NA, Copeland NG. Evi9 encodes a novel zinc finger pro- may favor the translocation process by openingthe chromatin tein that physically interacts with BCL6, a known human B-cell structure of this region of chromosome 14. CTIP2 is also proto-oncogene product. Mol Cell Biol 2000; 20: 3178–3186. expected to provide strongtranscriptional activator elements 9 Saiki Y, Yamazaki Y, Yoshida M, Katoh O, Nakamura T. Human which induces overexpression of Hox11L2, as a result of the EVI9, a homologue of the mouse myeloid leukemia gene, is t(5;14). It is likely that the observed high T cell expression of expressed in the hematopoietic progenitors and down-regulated duringmyeloid differentiation of HL60 cells. Genomics 2000; 70: CTIP2 within hematopoietic differentiation underlies the 387–391. association between t(5;14) and T ALL. To date, transcrip- 10 Koch P, Bohlmann I, Schafer M, Hansen-Hagge TE, Kiyoi H, Wilda tional activation, leadingto ectopic expression of a normal M, Hameister H, Bartram CR, Janssen JW. Identification of a novel product, appears to be the most frequent consequence of gen- putative Ran-bindingprotein and its close homologue. Biochem omic abnormalities in T ALL, whereas gene fusion leading to Biophys Res Commun 2000; 278: 241–249. expression of chimeric product apparently occurs less 11 Kutay U, Hartmann E, Treichel N, Calado A, Carmo-Fonseca M, Prehn S, Kraft R, Gorlich D, Bischoff FR. Identification of two frequently. novel RanGTP-bindingproteins belongingto the importin beta Characterization of ‘hidden’ rearrangements, whose detec- superfamily. J Biol Chem 2000; 275: 40163–40168. tion was dramatically improved usingFISH techniques, is a 12 Dube ID, Kamel-Reid S, Yuan CC, Lu M, Wu X, Corpus G, Rai- major goal of actual cytogenetics. As examplified by the TAL1 mondi SC, Crist WM, Carroll AJ, Minowada J et al. A novel human story, which was firstly isolated as rearranged by a translo- homeobox gene lies at the chromosome 10 breakpoint in lymph- cation involvingthe TCR ␣/␦ locus and later shown to be fre- oid neoplasias with chromosomal translocation t(10;14). Blood 1991; 78: 2996–3003. quently activated as a result of the inframicroscopic SIL-TAL1 13 Hatano M, Roberts CW, Minden M, Crist WM, Korsmeyer SJ. fusion (see Ref. 36 for review), analysis of rare aberrations can Deregulation of a homeobox gene, HOX11, by the t(10;14) in T be the first step toward the discovery of frequent molecular cell leukemia. Science 1991; 253: 79–82.

Leukemia Hox1 1-like gene activation in T ALL OA Bernard et al 1504 14 Kennedy MA, Gonzalez-Sarmiento R, Kees UR, Lampert F, Dear precursors with both primitive and definitive hematopoietic poten- N, Boehm T, Rabbitts TH. HOX11, a homeobox-containingT-cell tial. Blood 1998; 92: 877–887. oncogene on human chromosome 10q24. Proc Natl Acad Sci USA 26 Raju K, TangS, Dube ID, Kamel-Reid S, Bryce DM, Breitman ML. 1991; 88: 8900–8904. Characterization and developmental expression of Tlx-1, the 15 Lu M, GongZY, Shen WF, Ho AD. The tcl-3 proto-oncogenealt- murine homologof HOX11. Mech Dev 1993; 44: 51–64. ered by chromosomal translocation in T-cell leukemia codes for 27 Hatano M, Iitsuka Y, Yamamoto H, Dezawa M, Yusa S, Kohno Y, a homeobox protein. Embo J 1991; 10: 2905–2910. Tokuhisa T. Ncx, a Hox11 related gene, is expressed in a variety 16 Dear TN, Sanchez-Garcia I, Rabbitts TH. The HOX11 gene enco- of tissues derived from neural crest cells. Anat Embryol (Berl) des a DNA-bindingnuclear transcription factor belongingto a dis- 1997; 195: 419–425. tinct family of homeobox genes. Proc Natl Acad Sci USA 1993; 28 Logan C, Wingate RJ, McKay IJ, Lumsden A. Tlx-1 and Tlx-3 90: 4431–4435. homeobox gene expression in cranial sensory ganglia and hind- 17 Mitelman F, Johansson B, Mertens F. Catalog of Chromosome brain of the chick embryo: markers of patterned connectivity. J Aberrations in Cancers, Wiley-Liss, New York, 1998. Neurosci 1998; 18: 5389–5402. 18 Meeker TC, Hardy D, Willman C, Hogan T, Abrams J. Activation 29 Patterson KD, KriegPA. Hox11-family genesXHox11 and of the interleukin-3 gene by chromosome translocation in acute XHox11L2 in xenopus: XHox11L2 expression is restricted to a lymphocytic leukemia with eosinophilia. Blood 1990; 76: 285– subset of the primary sensory neurons. Dev Dyn 1999; 214: 289. 34–43. 19 Jaju RJ, Haas OA, Neat M, Harbott J, Saha V, Boultwood J, Brown 30 Uchiyama K, Otsuka R, Hanaoka K. CHox11L2, a Hox11 related JM, Pirc-Danoewinata H, Krings BW, Muller U, Morris SW, gene, is expressed in the peripheral nervous system and subpopul- Wainscoat JS, Kearney L. A new recurrent translocation, ation of the spinal cord duringchick development. Neurosci Lett t(5;11)(q35;p15.5), associated with del(5q) in childhood acute 1999; 273: 97–100. myeloid leukemia. The UK Cancer Cytogenetics Group (UKCCG). 31 Roberts CW, Shutter JR, Korsmeyer SJ. Hox11 controls the genesis Blood 1999; 94: 773–780. of the spleen. Nature 1994; 368: 747–749. 20 Gorlich D, Kutay U. Transport between the cell nucleus and the 32 Dear TN, Colledge WH, Carlton MB, Lavenir I, Larson T, Smith cytoplasm. Annu Rev Cell Dev Biol 1999; 15: 607–660. AJ, Warren AJ, Evans MJ, Sofroniew MV, Rabbitts TH. The Hox11 21 Whitlock JA, Raimondi SC, Harbott J, Morris SW, McCurley TL, gene is essential for cell survival during spleen development. Hansen-Hagge TE, Ludwig WD, Weimann G, Bartram CR. Development 1995; 121: 2909–2915. t(5;14)(q33–34;q11), a new recurringcytogeneticabnormality in 33 Hatano M, Aoki T, Dezawa M, Yusa S, Iitsuka Y, Koseki H, Tanig- childhood acute leukemia. Leukemia 1994; 8: 1539–1543. uchi M, Tokuhisa T. A novel pathogenesis of megacolon in 22 ZhangN, Shen W, Ho AD, Lu M. Three distinct domains in the Ncx/Hox11L.1 deficient mice. J Clin Invest 1997; 100: 795–801. HOX-11 homeobox oncoprotein are required for optimal trans- 34 Shirasawa S, Arata A, Onimaru H, Roth KA, Brown GA, Horning activation. Oncogene 1996; 13: 1781–1787. S, Arata S, Okumura K, Sasazuki T, Korsmeyer SJ. Rnx deficiency 23 Hawley RG, FongAZ, Reis MD, ZhangN, Lu M, Hawley TS. results in congenital central hypoventilation. Nat Genet 2000; 24: Transformingfunction of the HOX11/TCL3 homeobox gene. Can- 287–290. cer Res 1997; 57: 337–345. 35 Sonoki T, Satterwhite E, Willis TG, Siebert R, Nowak R, Arriola 24 Hough MR, Reis MD, Singaraja R, Bryce DM, Kamel-Reid S, Dard- EL, Liu H, Harder L, Gesk S, Steinemann D, Oscier DG, Schelgel- ick I, Breitman ML, Dube ID. A model for spontaneous B-lineage berger B, Tucker PW, Dyers MJS. The BCL11 gene family: deregu- lymphomas in IgHmu-HOX11 transgenic mice. Proc Natl Acad lated expression of BCL11A in lymphoma. Blood 2000; 96 Sci USA 1998; 95: 13853–13858. (Suppl. 1): 542a. 25 Keller G, Wall C, FongAZ, Hawley TS, Hawley RG. Over- 36 Begley CG, Green AR. The SCL gene: from case report to critical expression of HOX11 leads to the immortalization of embryonic hematopoietic regulator. Blood 1999; 93: 2760–2770.

Leukemia