Oncogene (1998) 17, 2799 ± 2803 ã 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc SHORT REPORT Identi®cation of the LMO4 gene encoding an interaction partner of the LIM-binding protein LDB1/NLI1: a candidate for displacement by LMO proteins in T cell acute leukaemia

G Grutz1, A Forster1 and TH Rabbitts*,1

1MRC Laboratory of Molecular Biology, Division of Protein and Nucleic Acid Chemistry, Hills Road, Cambridge CB2 2QH, UK

The T cell oncogenes LMO1 and LMO2 are activated Molecular studies of the LMO proteins has shown by distinct chromosomal translocations in childhood T that the LIM domain is a protein interaction domain cell acute leukaemias. Transgenic mouse models of this (Boehm et al., 1990; Schmeichel and Beckerle, 1994; disease demonstrate that enforced expression of Lmo1 Wadman et al., 1994). The LIM domain can bind to a and Lmo2 cause T cell leukaemias with long latency and diverse spectrum of protein motifs, including homo- that Lmo2 expression leads to an inhibition of the T cell dimerization (Sanchez-Garcia et al., 1995), heterodi- di€erentiation programme, prior to overt disease. These merization with the basic-helix ± loop ± helix (bHLH) functions appear to be partly mediated by interaction of protein TAL1/SCL (Valge-Archer et al., 1994; Wad- LMO1 or LMO2 with the LIM-binding protein LDB1/ man et al., 1994), with the GATA-1 DNA-binding NLI1. We have now identi®ed a new member of the Lmo protein (Osada et al., 1995; Wadman et al., 1997) or family, designated Lmo4, via its interaction with Ldb1. with the LIM-binding protein LDB1/NLI1 (Agulnick Lmo4 is widely expressed in mouse tissues, including et al., 1996; Bach et al., 1997; Jurata et al., 1996; adult thymus (mainly CD4, CD8-double positive T cells) Wadman et al., 1997). These interactions imply a role and embryonic thymus (mainly CD4, CD8-double for the LMO proteins in control of gene expression by negative T cells). These characteristics imply that modulation of formation of multimeric protein Ldb1-Lmo4 interaction may function in the T cell transcription complexes (Wadman et al., 1997). The developmental programme and that enforced expression role of Lmo2 in di€erentiation further emphasises the of LMO1 or LMO2 by chromosomal translocations or proposed function in the control of gene expression. transgenesis may displace Lmo4 from this complex and Null mutations of the Lmo2 gene in mice result in a thereby in¯uence T cell di€erentiation prior to T cell failure of yolk sac erythropoiesis (Warren et al., 1994) tumour occurrence. and in adult de®nitive haematopoiesis (Yamada et al., 1998). A role in these cell types seems to be carried out Keywords: LIM domains; T cell; leukaemia; chromo- by the formation of a multimeric DNA-binding somal translocations; protein interactions complex including Lmo2, Tal1, E47, GATA-1 and Ldb1 (Wadman et al., 1997) which can bind to a bipartite DNA site, comprising a E-box (a site for E47- Tal1 heterodimer) and a GATA motif (a site GATA- Proteins that carry zinc-binding LIM domains are 1). important in development, in transcription and in The three members of the LMO family are related in diseases (Dawid et al., 1998), including cancer sequence and two (LMO1 and LMO2) are involved in development (Sanchez-Garcia and Rabbitts, 1993). T cell tumorigenesis after chromosomal translocations. The LMO family of genes, which encode LIM-only Attempts to emulate the consequences of these proteins, was ®rst discovered by association of LMO1 chromosomal translocations have been made using (previously called RBTN1 or TTG1) with the speci®c transgenic mice with enforced expression of Lmo1 chromosomal translocation t(11;14)(p15;q11) in human (Aplan et al., 1997; Fisch et al., 1992; McGuire et al., T cell acute leukaemias, T-ALL (reviewed in Rabbitts, 1992) or Lmo2 in thymic T cells (Fisch et al., 1992; 1998; Rabbitts et al., 1998). The LMO1 cDNA was Larson et al., 1994, 1995, 1996; Neale et al., 1995). In used as a probe for two related family members, these model systems, clonal T cell tumours arise with a designated LMO2 and LMO3 (Boehm et al., 1991a; long latency suggesting that the transgene is necessary Foroni et al., 1992), and it was found that LMO2 was but not sucient for overt tumour development. In also associated with chromosomal translocations in particular with Lmo2 transgenic mice, the e€ect of the human T cell acute leukaemias, viz. (t(11;14)(p13;q11) transgene expression is an inhibition of the T cell and t(7;11)(q35;p13) (Boehm et al., 1991a; Royer- developmental programme leading to an accumulation Pokora et al., 1991). The unique feature of this family of immature CD4, CD8-negative T cells (Larson et al., of genes is that they encode small proteins comprising 1995, 1996; Neale et al., 1995). The role in controlling two LIM domains only, in which signi®cant sequence T cell development may be mediated through multi- similarity is found between the family members. meric complexes analogous to those found in normal haematopoietic cells, since a DNA-binding complex comprising Lmo2, Tal1, E47 and Ldb1 was found in T *Correspondence: TH Rabbitts cell lines made from transgenic mouse tumours (Grutz Received 2 October 1998; revised 22 October 1998; accepted 22 et al., 1998). Thus, as in the haematopoietic setting, October 1998 control of target gene expression may be mediated by Identification of Lmo4 as a partner of Ldb1 G Grutz et al 2800 Lmo2 protein facilitating formation, through its LIM domains, of distinct multimeric protein complexes. In the tumour setting, these complexes would presumably be present after enforced LMO2 expression either by a chromosomal translocation or by transgenesis, thereby a€ecting T cell di€erentiation. A similar molecular mechanism is possible for the action of LMO1 in T cell tumorigenesis since LMO1 protein interacts with both TAL1 (Valge-Archer et al., 1994) and with LDB1 (Valge-Archer et al., 1998) in T cell tumours with the chromosomal translocation t(11;14)(p15;q11). Alterna- tively, the elevation of LMO1 or LMO2 by enforced expression may result in aberrant complex formation which serves to sequester one (or more) component into the complex preventing its normal function (Rabbitts, 1998). To facilitate the test of either model, it is necessary to identify and characterize the components of the multimeric complexes. In LMO protein mediated tumorigenesis, a common component is the LDB1 protein. Therefore, we have assessed protein partners of the Ldb1 molecule in T-cells and have discovered a new LIM-only protein, designated Lmo4, as a binding partner of Ldb1. Lmo4 is widely expressed and is a candidate molecule which may be involved in DNA-binding complexes or which might be displaced from Ldb1 interaction by enforced expression of Lmo1 or Lmo2 proteins in T cells. An Ldb1 coding sequence was cloned in a yeast two- hybrid (Fields and Song, 1989) bait vector and a cDNA library made from 2114 RNA was screened for interacting partners (the 2114 cell line was established from a T cell tumour which arose in a CD2-Lmo2 Figure 1 (a) Nucleotide and derived protein translation of mouse Lmo4. A cDNA clone was isolated corresponding to mRNA transgenic mouse (Grutz et al., 1998) and is a CD4, encoding the Lmo4 protein by interaction with an Ldb1 bait in a CD8 double negative immature T cell). Several yeast two-hybrid screen (Fields and Song, 1989). Sequencing of di€erent clones were isolated in this screening. One the cDNA revealed an open reading which translated into a LIM- clone was sequenced (p209) and found to encode a only protein, designated Lmo4. A similar cDNA has recently been small protein of 165 amino-acids (Figure 1a). described (Kenny et al., 1998) and the sequence of the human counterpart (which is identical in translated product with the Examination of the protein sequence revealed that mouse sequence) was previously deposited in the Genbank this cDNA clone corresponded to a mRNA encoding a database by J Racevskis, A Dill and H Ruan (Accession new member of the LIM-only LMO family of proteins. number: U24576). The sequence shown extends from codon 1. A similar cDNA has recently been described (Kenny et The residues likely to be involved in zinc co-ordination are shown in bold type and underlined. The EcoRI site de®ning the 3' end of al., 1998) and the sequence of the human counterpart the Lmo4 cDNA probe is underlined. (b) The four known (which is identical in translated product with the mouse members of the LMO family of proteins are aligned with respect sequence reported here) was previously deposited in the to Lmo4. Sequence identity to Lmo4 is indicated by a point (.). Genbank database by J Racevskis, A Dill and H Ruan The sequences are broken into the four regions of the proteins. (Accession number: U24576). In keeping with other Homology in the LIM domains is high between the four proteins but minimal in the NH2- and C-terminal stretches; this is also members of the LMO family, this new member is true between the otherwise highly similar Lmo1 and LMO3. The designated Lmo4. residues likely to be involved in zinc co-ordination are underlined. Lmo4 has two LIM domains together with short Methods: Yeast two-hybrid screen (Fields and Song, 1989) was amino-terminal (22 residues) and carboxyl-terminal carried out essentially as described by the Stratagene Instruction Manual. An oligo-dT primed custom made cDNA library stretches (25 residues). An alignment of the Lmo4 (Stratagene) was prepared in the vector pAD-GAL4 (as a fusion protein with Lmo1, 2 and 3 illustrates a high degree of with the GAL4 activation domain) using poly(A)+-selected RNA homology within the LIM domains (Figure 1b) and the from the T cell line 2114, derived from a CD2-Lmo2 transgenic mouse tumour (Grutz et al., 1998). The vector gives undirectional same spacing between the ®rst and second LIM 7 domains. Signi®cant identities in amino-acid sequence insertion of cDNA clones and the initial size was 2610 clones. The bait for protein interactions was Ldb1 cloned in pBD-GAL4, are also found in the short linker between the LIM linking Ldb1 sequences to GAL4 DNA-binding domain. The domains of all four proteins. In addition to the Ldb1 bait cDNA was a full length Ldb1 coding region which invariant cysteine/histidine/aspartate residues which therefore includes the LIM-binding domain as well as all other bind zinc of the LIM domains, most of the regions of the protein. This clone was pre-tested by transfecting it alone into YRG-2 yeast to assay for auto-activation properties, of hydrophobic and charged residues, which were which none were found. Approximately 106 individual clones were previously noted as conserved features of LIM screened by co-transfection of the 2114-cDNA library and one, domains (Dawid et al., 1995; Sanchez-Garcia and full length Lmo4 cDNA clone (p209) was obtained which could be Rabbitts, 1994), are also conserved (Figure 1b). In isolated and re-introduced into yeast with the Ldb1 bait resulting the Drosophila genome there is a single homologue of in activation. p209 does not activate in the absence of Ldb1 interaction bait. The Lmo4 clone was sequenced on both strands the LMO genes (Boehm et al., 1990; Zhu et al., 1995) using an ABI Prism 377 Sequencer and sequence data analysed by and the sequence corresponding to this Dlmo gene has automated sequence handling programmes (Bon®eld et al., 1995) Identification of Lmo4 as a partner of Ldb1 G Grutz et al 2801 a similar retention of these important residues and also brain, where we observe Lmo4 expression in whole of the spacing between LIM domains. It would appear brain and in speci®c zones e.g. cerebellum (Table 1). therefore that the functioning of LMO proteins in These observations with Lmo4 expression in mouse protein-protein interaction requires both amino-acid tissues are markedly di€erent from the more restricted conservation for precise protein contacts and spatial patterns observed with the other Lmo family members, organization to facilitate docking and interaction of Lmo1, 2 and 3. protein interfaces of various protein partners. The Lmo1, Lmo2 and Lmo3 genes have a rather restricted pattern of expression in embryonic and adult Table 1 Expression of Lmo4 mRNA tissues (Boehm et al., 1991b; Foroni et al., 1992; McGuire et al., 1991) and precise roles in embryogen- A. Mouse Nervous sytem Somatic tissues esis. This is particularly pertinent to Lmo2 in which E12.5 head Lung (adult) homozygous null mutations cause embryonic lethality E13 head Heart (adult) due to a failure of yolk sac erythropoiesis and in E14 head Liver (adult) addition, there is a complete lack of haematopoiesis in E15 head Spleen (adult) E16 head E16.5 thymus adult mice in the absence of Lmo2 (Warren et al., N1 whole brain Thymus (N1) 1994; Yamada et al., 1998). Lmo1 and Lmo3 have a 4 wk telencephalon Thymus (4 wk) similar restricted pattern of expression, being almost 4 wk di/mesenchephalon Thymus (adult) only found in developing nervous system tissue and in 4 wk cerebellum adult brain (Boehm et al., 1991b; Foroni et al., 1992; Cell lines McGuire et al., 1991). The pattern of expression of T cell line B cells Lmo4 was therefore of interest because its binding to 2114 Baf/3 (pro-B-cell) Ldb1 suggested a possible role in T cell complexes. WEHI 231 (pre-B-cell) Accordingly, RNA prepared from a variety of mouse J558L (plasmacytoma) tissues and cell lines and some human T cell lines, was B. Human examined by Northern ®lter hybridization for Lmo4 T cell lines mRNA (Figure 2 and Table 1). The most signi®cant KOP-T1 t(11;14)(p13;q11) feature of Lmo4 expression is that it is very widespread RPMI8402 t(11;14)(p15;q11) and appears in all locations tested. These data are ALL-SIL t(10;14)(q24;q11) similar to those recently published (Kenny et al., 1998). Erythroid A 2.3 Kb Lmo4 mRNA is expressed in the 2114 cells, HEL from which the yeast two-hybrid cDNA library was The sites of Lmo4 expression were determined by Northern ®lter prepared, in adult thymus as well as in the pro-B-cell hybridization using 10 mg of total RNA (except in the case of the line Baf3 (Figure 2). We ®nd expression in RNA from CD2-Lmo2 transgenic T cell tumour line 2114, where 0.1 mg of + all tissues examined, including nervous system, adult poly(A) mRNA was used). Expression was observed in all tissues or cell lines analysed, including human T cell acute leukaemia cell lines somatic tissues such as lung and liver and lymphoid with chromosomal translocation a€ecting of LMO1 (RPMI8402) and tissues. Expression in nervous tissue is detectable from LMO2 (KOP-T1) genes as well as one with a chromosomal embryonic stage E12.5 through gestation into adult translocation a€ecting the HOX11 gene (ALL-SIL)

Figure 2 Northern ®lter hybridization of mouse tissue RNA with an Lmo4 probe. RNA was extracted from a variety of mouse tissues and was analysed by ®lter hybridization with Lmo4 cDNA. Tissue RNAs were made from mice at various ages and comparable hybridizations levels were evident. Methods: Total RNA was prepared from mouse tissues using the TRIzol extraction procedure (Gibco-BRL) and poly(A)+ RNA (2214) was prepared by oligo-dT cellulose chromatography. 10 mg samples of total RNA or 0.1 mg of poly(A)+ RNA were denatured by glyoxal treatment (Thomas, 1980) and analysed on 1.4% agarose gels followed by capillary transfer to nylon membranes. The membranes were hybridized with randomly labelled DNA (Feinberg and Vogelstein, 1983) as described at 658C in 36SSC, 0.1% SDS (LeFranc et al., 1986). The Lmo4 probe employed was a cDNA clone extending from 240 bp upstream of the initiation codon to an EcoRI site at codon 151. Autoradiography of the washed ®lter was for 16 h at 7708C with an intensifying screen Identification of Lmo4 as a partner of Ldb1 G Grutz et al 2802 Lmo4 is widely expressed in lymphoid cells. All the function as independent modules in di€erent ways. T and B cell lines tested (both mouse and human) The NH2-terminal parts of LMO1 and LMO2 have showed Lmo4 RNA (Table 1). In addition, whole been shown to mediate transcriptional transactivation spleen from adult mice and thymus express the mRNA. in reporter gene systems (Sanchez-Garcia et al., 1995) It is signi®cant that Lmo4 levels do not vary to any but the functions of these domains in the other LMO extent between embryonic thymus (E16.5), new born proteins have not yet been assessed. mouse thymus through to adult thymus. Thus it seems The expression of Lmo4 in developing and mature T that CD4, CD8-negative immature thymocytes (present cells is of potential interest to the mode of action of in embryonic thymus) express Lmo4 as well as CD4, LMO1 and LMO2 in T cell tumorigenesis. Interaction CD8-positive thymocytes. This is consistent with of Lmo4 and Ldb1 may play a role in the normal T previous RT ± PCR results (Kenny et al., 1998). cell di€erentiation programme and imbalances in this The discovery of the Lmo4 gene extends the LIM- interaction may in¯uence this programme. Thus in only LMO family yet further. This family has thus far LMO-mediated T cell tumorigenesis, an intriguing two members (LMO1 and LMO2) known to be possibility is that LMO4 is displaced from a complex involved in chromosomal translocations in human involving LDB1 by sequestration of the latter by tumours, both in T cell acute leukaemia. Despite LMO1 and LMO2 (after elevated levels of these sequence conservation, a profound di€erence exists in proteins following chromosomal translocation-en- the extent of the expression between Lmo4 and the forced expression) (Rabbitts, 1998). This model is other Lmo family members, suggesting di€erent distinct from the formation of aberrant DNA-binding functions in diverse cellular settings within the multimeric complexes, after enforced LMO expression, developing organism and/or within the adult animal. which might control target gene expression by direct The putative functions of Lmo4, like those of Lmo1, 2 DNA binding. However, LDB1 and LMO4 are and 3 (reviewed in Rabbitts, 1998; Rabbitts et al., common elements in T cells, further suggesting that 1998), will presumably be primarily mediated by the interaction of these molecules could be a target protein interaction through the LIM domains. How- event in tumorigenesis. Now that LMO4-LDB1 ever, it is interesting that, whilst the LIM domains of interaction has been uncovered, these models can be the LIM-only proteins are very similar, no similarity tested in appropriate animal and cell systems. can be seen in the relatively small stretches of protein outside the LIM domains. Even comparing LMO1 and LMO3, which have almost identical LIM domains, these N- and C-terminal stretches show little conserva- Acknowledgement tion. Thus these non-LIM domain segments may GenBank Accession number AF096996.

References

Agulnick AD, Taira M, Breen JJ, Tanaka T, Dawid IB and Larson R, Fisch P, Larson T, Lavenir I, Langford T, King G Westphal H. (1996). Nature, 384, 270 ± 272. and Rabbitts TH. (1994). Oncogene, 9, 3675 ± 3681. Aplan PD, Jones CA, Chervinsky DS, Zhao XF, Ellsworth Larson RC, Lavenir I, Larson TA, Baer R, Warren AJ, MK, Wu C, McGuire EA and Gross KW. (1997). EMBO Wadman I, Nottage K and Rabbitts TH. (1996). EMBO J., 16, 2408 ± 2419. J., 15, 1021 ± 1027. Bach I, Carriere C, Ostendor€ HP, Anderson B and Larson RC, Osada H, Larson TA, Lavenir I and Rabbitts Rosenfeld MG. (1997). Genes Dev., 11, 1370 ± 1380. TH. (1995). Oncogene, 11, 853 ± 862. Boehm T, Foroni L, Kaneko Y, Perutz MP and Rabbitts TH. LeFranc MP, Forster A, Baer R, Stinson MA and Rabbitts (1991a). Proc. Natl. Acad. Sci. USA, 88, 4367 ± 4371. TH. (1986). Cell, 45, 237 ± 246. Boehm T, Spillantini M-G, Sofroniew MV, Surani MA and McGuire EA, Davis AR and Korsmeyer SJ. (1991). Blood, Rabbitts TH. (1991b). Oncogene, 6, 695 ± 703. 77, 599 ± 606. Boehm T, Foroni L, Kennedy M and Rabbitts TH. (1990). McGuire EA, Rintoul CE, Sclar GM and Korsmeyer SJ. Oncogene, 5, 1103 ± 1105. (1992). Mol. Cell. Biol., 12, 4186 ± 4196. Bon®eld JK, Smith KF and Staden R. (1995). Nucl. Acids Neale GA, Rehg JE and Goorha RM. (1995). Blood, 86, Res., 23, 4992 ± 4999. 3060 ± 3071. Dawid IB, Breen JJ and Toyama R. (1998). Trends Genet., Osada H, Grutz G, Axelson H, Forster A and Rabbitts TH. 14, 156 ± 162. (1995). Proc. Natl. Acad. Sci. USA, 92, 9585 ± 9589. Dawid IB, Toyama R and Taira M. (1995). C.R. Acad. Sci. Rabbitts TH. (1998). Genes Devel., 12, 2651 ± 2657. Paris, Sciences de la vie/Life Sciences, 318, 295 ± 306. Rabbitts TH, Bucher K, Chung G, GruÈ tz GG, Warren AJ Feinberg AP and Vogelstein BA. (1983). Anal. Biochem., 132, and Yamada Y. (1998). Cancer Res., in press. 6 ± 13. Royer-Pokora B, Loos U and Ludwig W-D. (1991). Fields S and Song O. (1989). Nature, 340, 245 ± 246. Oncogene, 6, 1887 ± 1893. Fisch P, Boehm T, Lavenir I, Larson T, Arno J, Forster A Sanchez-Garcia I, Axelson H and Rabbitts TH. (1995). and Rabbitts TH. (1992). Oncogene, 7, 2389 ± 2397. Oncogene, 10, 1301 ± 1306. Foroni L, Boehm T, White L, Forster A, Sherrington P, Liao Sanchez-Garcia I and Rabbitts TH. (1993). Sem. Canc. Biol., XB, Brannan CI, Jenkins NA, Copeland NG and Rabbitts 4, 349 ± 358. TH. (1992). J. Mol. Biol., 226, 747 ± 761. Sanchez-Garcia I and Rabbitts TH. (1994). Trends Genet., Grutz G, Bucher K, Lavenir I, Larson R, Larson T and 10, 315 ± 320. Rabbitts TH. (1998). EMBO J., 17, 4594 ± 4605. Schmeichel KL and Beckerle MC. (1994). Cell, 79, 211 ± 219. Jurata LW, Kenny DA and Gill GN. (1996). Proc. Natl. Thomas PS. (1980). Proc. Natl. Acad. Sci. USA, 77, 5201 ± Acad. Sci. USA, 93, 11693 ± 11698. 5205. Kenny DA, Jurata LW, Saga and Gill GN. (1998). Proc. Valge-Archer V, Forster A and Rabbitts TH. (1998). Natl. Acad. Sci. USA, 95, 11257 ± 11262. Oncogene, in press. Identification of Lmo4 as a partner of Ldb1 G Grutz et al 2803 Valge-Archer VE, Osada H, Warren AJ, Forster A, Li J, Warren AJ, Colledge WH, Carlton MBL, Evans MJ, Smith Baer R and Rabbitts TH. (1994). Proc. Natl. Acad. Sci. AJH and Rabbitts TH. (1994). Cell, 78, 45 ± 58. USA, 91, 8617 ± 8621. Yamada Y, Warren AW, Dobson C, Forster A, Pannell R Wadman I, Li J, Bash RO, Forster A, Osada H, Rabbitts TH and Rabbitts TH. (1998). Proc. Natl. Acad. Sci. USA, 95, and Baer R. (1994). EMBO J., 13, 4831 ± 4839. 3890 ± 3895. Wadman IA, Osada H, Grutz GG, Agulnick AD, Westphal Zhu T-H, Bodem J, Keppel E, Paro R and Royer-Pokora B. H, Forster A and Rabbitts TH. (1997). EMBO J., 16, (1995). Oncogene, 11, 1283 ± 1290. 3145 ± 3157.