CA2259635A1 - Hmgi proteins in tumors and obesity - Google Patents

Abstract

The invention pertains to HMGI genes and proteins, specifically to HMGI-C and HMGI-Y genes and proteins; and the use of these products in various methods.
The invention further relates to methods of treating obesity and tumors by reducing the activity of HMGI genes; to methods of making non-human transgenic mammals; to methods of screening for HMGI inhibitors; and to methods of detecting for protein and antibodies to HMGI genes.

Description

CA 022S963S l999-0l-06 HMGI PROTEINS IN TUMORS AND OBESITY

BACKGROUND OF THE INVENTION

This application is a continuation-in-part of United States patent application serial no. 08/852,666, filed 7 May 1997, which application is a continll~tion-in-part of United States patent application serial no. 08/679,529, filed 12 July 1996.
StPt~m~nt of Rights to IIl~w~Lio~ls Made Under Federally-Sponsored Re3~ Cll and Development Part of the work performed during development of this invention 25 utilized United States Government funds. The United States Government has certain rights in this invention: NIH grant no. GM38731, HD30498, and lK11CA01498.

3 o Field of the Invention The present invention pertains to a method for treating obesity in a m~mm~l which comprises reducing the biological activity of HMGI genes in the 3 5 m~mm~l . In another embo~limPnt, the invention pertains to a method for treating a tumor in a patient by reducing the biological activity of normal HMGI genes which comprises ~lmini.ctering to the patient a therape~ltic~lly effective amount of an SUBS 1 l 1 u l ~ ; l (RULE 26) CA 022~963~ 1999-01-06 WO 98/S0536 PCT/US97t21299 inhibitor compound active against normal HMGI-C or HMGI(Y) genes. In another embodiment, the invention pertains to a method of producing a transgenic non-human m~mm~l, the germ cells and somatic cells of which contain an inactivated HMGI gene sequence introduced into the m~mm~l, or an ancestor of the m~mm~l, 5 at an embryonic stage. In another embodiment, the invention pertains to a method for screening c~nr~ t~ compounds capable of inhibiting the biological activity of norrnal HMGI proteins, or a fragment thereof, which comprises the steps of (a) incubating a HMGI protein, or a fragment thereof, with a c~n~ te compound under conditions which promote optimal interaction; and (b) m~ ring the binding 0 affinity of the c~n~ te compound to the HMGI protein, or a fragment thereof;
and (c) dele~ from the binding affinity which c~mli~l~te compounds inhibit the biological activity of HMGI proteins, or a fragment thereof. In another embodiment, the invention pertains to a method for sc,ce~ g c~nrlid~te compoundscapable of inhibiting the biological activity of normal HMGI genes which 5 comprises the steps of (a) transfecting into a cell a DNA construct which contains a reporter gene under control of a normal HMGI protein-regulated promoter; (b) mini~tering to the cell a c~n(~ t~ compound; (c) measuring the levels of reporter gene expression; and (d) determining from the levels of reporter gene expressionwhich c~n~ tt~ cc.ll,~ounds inhibit the HMGI biological activity. In another 2 o embodiment, the invention pertains to a method for detecting normal HMGI
proteins as a diagnostic marker for a tumor using a probe that recognizes normalHMGI proteins, which comprises the steps of (a) cont~rting normal HMGI proteins from a sample from a patient with a probe which binds to HMGI proteins; and (b) analyzing for normal HMGI proteins by detecting levels of the probe bound to the25 normal HMGI pluteils, wherein the presence of normal HMGI proteins in the sample is positive for a tumor. In another embodiment, the invention pertains to a method for ~lPtecting antibodies to normal HMGI proteins using a probe that recognizes antibodies to HMGI normal proteins, which comprises the steps of (a) treating a sample from a patient with a probe which binds to antibodies to normal 30 HMGI pruLeills; and (b) analyzing for antibodies to HMGI proteins by detecting levels of the probe bound to the antibodies to HMGI proteins, wh~.eill the presence of antibodies to normal HMGI proteills in the sample is positive for a tumor. Inanother embodiment, the invention pertains to HMGI genes and proteins for use asa starting point to isolate duwns~ lll target genes regulated by the HMGI genes 3 5 and proteins.

SUB~i 1 l l U l ~ ; 1' (RULE 26) wo 98/50536 PCT/USg7/21Z99 D~s~ ion of the Background The disclosures referred to herein to illllstrat~ the background of the invention and to provide ~ ition~l detail with respect to its practice are incorporated herein by reference and, for convenience, are referenced in the following text and respectively grouped in the appended bibliography.

HMGI Proteins in A.lil og~ s and Me~ r~-ynle Differenti~ti~n Underst~nrling various genes and pathways underlying development of mlllticell~ r org~ni~m~ provide insights into the molecular basis of the highly regulated processes of cellular proliferation and difrclellliation. In turn, genetic 5 aberrations in control of cell growth lead to a variety of developmental abnorm~ es and, most prominently, cancer (Aaronson, 1991). To pursue itlentification of genes involved in these fi-~ m~nt~l biological processes, theviable pygmy mutation (MacArthur, 1944) was investi~t~(l because it gives rise to mice of small stature due to a disruption in overall growth and development of the 20 mouse. An insertional tr~n~genir mutant facilitated cloning of the locus (Xiang et al., 1990) and subsequently it was shown that expression of the HMGI-C gene was abrogated in three pygmy alleles (unpublished results).

HMGI-C belongs to the HMG (high mobility group) family of 2 5 DNA-binding proteins which are abundant, heterogeneous, non-histone components of chrolllalill (Cluscchrrll et al., 1994). HMG~roleills are divided into three distinct familiPs, the HMG box-cnnt~ining HMGl/2, the active chromatin associated HMG14/17 and the HMGI proteins (Grosschedl et al., 1994). At present, the last family consists of two genes, HMGI(Y) (Johnson et al., 1988;
3 o Frie~im~nn et al., 1993) which produces two pl~teh~s via alternative splicing (Johnson et al., 1989) and HMGI-C (Manfioletti et al., 1991; Patel et al., 1994).
A prnminpnt feature of HMGI proteins is the presence of DNA-binding ~lo~n~in~
which bind to the narrow minor groove of A-T rich DNA (Reeves and Nissen, 1990) and are therefore referred to as A-T hooks. Recently, valuable insights have 35 been gained into their mPrll~ni.~m and role in L~ seli~Lion (Thanos and ~ni~tic, 1992; Du et al., 1993). The HMGI proteins have no llal scliptional activity per se SUB~ I l 1 U 1 k; ~ ; l (RULE 26) ..

WO 98/50536 PCT~US97/21299 (Wolffe, 1994), but through protein-protein and protein-DNA interactions olg~ni~e the framework of the nucleoproteill-DNA llallseli~tional complex. This frameworkis ~n~inr~l by their ability to change the conforrnation of DNA and these proleil~s are therefore termed archi~rctl~ral factors (Wolffe, 1994). In the well-studied case 5 of HMGI(Y) and the h~ rel.~il B promoter, HMGI(Y) stim~ tPs binding of NF-kB
and ATF-2 to applo~liate sequences and alters the DNA structure which allows thetwo factors to interact with each other and pre~ ably with the basal ll~sclil.tion m~rhinPry (Thanos and Maniatis, 1992; Du et al., 1993).

0 A number of studies have revealed an association between increased expression levels of HMGI proleh~s and transformation (Giancotti et al., 1987, 1989, 1993). For example, in ch~mir~lly, virally or spontaneously derived tumors, appreciable expression of HMGI-C was found in contrast to no detectable expression in normal tissues or untransformed cells (Giancotti et al., 1989). A
recent study has demonstrated a more direct role for HMGI-C in transformation (Berlingieri et al., 1995). Cells infected with oncogenic retroviruses failed toexhibit various phenotypic markers of transformation if HMGI-C protein SyllllRSiS
was specifically inhibited.

DNA probes adjacent to HMGI-C were mapped to the distal portion of mouse chromosome 10 in a region syntenic to the long arm of human chromosomP 12 including and distal to band ql3 (Justice et al., 1990). This genomic region is under intensive investigation because it is the location of consistent rearrangements in a number of neoplasms, mainly of mesenchymal origin (Schoenberg Fejzo et al., 1995). Lipomas, tumors mainly composed of mature fat cells, are one of the most common m~senrl ymal neoplasms that occur in hllm~n~ (Sreel~nt~i~h et al., 1991). Approxirnately 50~ of lipomas are characterized by cytogenetic rearrangements and the predominant alteration is a presumably b~l~need translocation involving 12ql4-15 with a large variety of chromosomal pa~ s including 1,2,3,4,5,6,7,10,11,13,15,17,21, and X
(Sreel~nt~i~h et al., 1991; Fletcher et al., 1993). This variability in reciprocal translocations along with duplications, inversions, and deletions of 12ql4-15 inthese tumors, strongly inrlir~Ps a primary role of a gene on chromosome 12 in lipomas. Furtherrnore, this gene may play a key role in normal dirr~ ion of primitive mPsenrllyme as not only lipomas, but also uterine leiomyomas (smooth muscle tumors), lipoleiomyomas (smooth muscle and adipose components), and SUB~ U l ~ ; l' (RULE 26) . . . _ , . .

WO 98/50~36 PCT/US97/21299 pulmonary chondroid hamartomas (primitive mPsenrhyme, smooth muscle, adipose, and mature cartilage components) are all clonal proliferations that arecharacterized by rearrangements of 12ql4-15 (Schoenberg Fejzo et al., 1995).
Interestingly, breakpoints in a lipoma, a pulmonary chondroid hamartoma and 5 uterine leiomyomata have been shown to map within a single YAC (Schoenberg Fejzo et al., 1995).

HMGI Fl~t~illS in l~rnm~ n Growth and DevPlo~nPnt 0 The first step in the molecular definition of the pygmy mutation was made possible by the isolation of a transgenic insertional mouse mutant at the locus, pgTgN40ACha (Xiang et al., 1990). A 0.5kb ApaI-ApaI single copy genomic sequence 2kb from the site of transgene insertion was identifiPd (Xiang et al., 1990) and used to initiate a bi-directional chromosome walk on normal mousegenomic DNA. The analysis of seven overlapping clones spanning 91kb deline~te~l a 56kb comrnon deletion between two informative ... "~ , pg and pgTgN4oAcha (Figure 8a).

The common area of disruption was investigated further for 2 o candidate transcription units. The technique of exon amplification (Buckler et al., 1991) was employed to identify putative exons and clones 803 and 5B, in the sameorientation, produced spliced products (Figure 8b). Their seq~lenre was determined (Ausubel et al., 1988) and a comparison to DNA sequence ~t~b~cec (GenBank and EMBL) revealed 100% homology to a previously identified gene, HMGI-C (Manfioletti et al., 1991) (Figure 8c). The HMGI members have been ~signPd multiple functions (Manfioletti et al., 1991) and recently, have been shown to play a critical role in regulation of gene ~ es~ion as archit~ctllral factors by inducing DNA conformational changes in the formation of the three-~1imPn~ional transcription complex (Thanos & ~ni~ti~, 199~; Du, W. et al., 3 o 1993)-Subsequently, the genomic structure of HMGI-C revealed that the gene contains five exons and spans a region of approximately llOkb (Figure 8d).
Single copy sequences from the l90kb cloned pygmy locus, sull~ui~ding and 35 including the HMGI-C gene (Figure 8d), were used as probes on Southern blots cont~inin~ DNA isolated from the two informative alleles (Xiang et al., 1990).

SUBSTITUTE S~k;~; l (RULE 26) . .

CA 022~963~ 1999-01-06 6 PCT/[~S97/21299 The genomic area enromp~cinE HMGI-C is completely deleted in the l~dl~sgellic insertional mutant pgTgN40ACha (A/A), whereas in the spontaneous mutant pg, the 5 ' sequences and the first two exons are absent (Figure 8d).

Mi~ of Disrupted Hl~GI Proteins in ~llm~-l Tumors Cancer arises from aberrations in the genetic m~çll~ni~m.c' that control growth and dirr~,lG,l~iation and ongoing elucidation of these m,orh~ni~m~
continues to improve the underst~n~ling of ~ n development and its various 10 abnorm~litiPs. Increasingly, ~cc~m~ ting experimental evidence points towardstranscriptional deregulation as one of the pivotal events in neoplasia. Many of the known transforming retroviral oncogenes, such as v-myc, v-fos and v-n~yb, are homologs of m~mm~ n transcription factors which are normally involved in proliferation and difrele.lliation control. Genes that encode for such transcription lS factors are frequently affected by the som~tir~lly acquired genetic changes which arise stoch~tir~lly over a lifetime of an o~ isll,. These alterations, which caneither activate expression of the relevant genes or disrupt them to create novelfusion pro~ins, affect llal~scriplion nelwulh~ and initiate cancer.

2 o One of the llalls~ tion factors whose disruption was shown to result in tumorigenesis is HMGI-C, which has attracted considerable attention for two reasons. First, a series of elegant experiments ~etnon.~trated that HMGI(Y) is involved in transcriptional regulation and is required for virus induction of the human i~ ,.r~loll-B gene expression. These obserations were incorporated into a novel model in which activation of gene expression is initi~tecl by a higher order transcription enhallcel complex. This functional nucleoprotein entity termed enh~nreosome is formed when several distinct lldnscli~Lion factors assemble on DNA in a stereospecific manner. Combinatorial m~ch~ni~m~ of the enhanceosome formation enable the cell to achieve high specificity of gene activation in response to multiple biological stimuli. As an esse~ti~l component of the e~h~nreosome, HMGI(Y) promotes the assembly of this three-~imP~.~ional structure through both protein-protein and protein-DNA interactions. The latter activity is mr~i~ted through the HMGI DNA-binding domains.

The function of HMGI-C, the other known mtonnher of the HMGI
family, in growth and development control is better understood at the biological SUB~ 1 l 1 U l ~; ~iH I1 ~;1 (RULE 26) CA 022~963~ 1999-01-06 leveh In hllm~n.~, rearrangements of HMGI-C were linked to the pathogenesis of several distinct types of solid tumors. Rearrange~.lk~ of the chromosomal band 12ql3-15, con~i.ctenfly found in a wide variety of benign mesenchymal neoplasms,disrupt HMGI-C and g~neldte novel chimeric transcripts. In the vast majority of 5 the analyzed tumors, these ll~nscli~ts consist of the HMGI-C DNA-binding domains fused to ectopic sequences provided by the translocation partner.

In the mouse, HMGI-C inactivation produced a dramatic disruption of both pre- and postnatal growth, reslllting in the pygmy phenotype. Pygmy mice0 exhibit signifirant growth retardation which is first appalel" in midgestation and becomes even more pronounced after birth. Adult animals are proportionally builtand viable but exhibit a 60% weight reduction colll~a,ed to their wildtype litterrnates. A detailed phenotypic analysis of the pygmy mouse revealed that the weight reduction in most of the tissues is comm~n~urate with the overall decrease 5 in body weight. Most illlere~ gly, HMGI-C inactivation does not affect the growth hormone-insuline-like growth factor endocr*ne pathway, suggesting that HMGl-C functions in a previously unknown growth regulatory mPch~ni~m The molecular basis of the pygmy mutation is not well understood.
20 High levels of the HMGI ~rolei"s are not required for cell growth per se and elevated HMGI ~Ayles~ion appear to be associated with the biological state of the cell more directly than with its high proliferation rate. Upon transformation with oncogenic retroviruses, expression of HMGI-C and HMGI(Y) in epithelial cells is dr;tm~tir~lly increased even though the proliferative capacity of the infected cells 2 5 remains lln~ffectecl . Furthermore, analysis of a llal~,ro.l~,ed cell line which retained its dirrele..li~te~ phenotype revealed that levels of the HMGI expression were signific~ntly lower than in cell lines which lost the* dirfelenLiaLion markers as a result of transformation. Other studies demonstrated that HMGI-C is expressed in less dirrtlel,liated mesenrllymal cells but is no longer present in their termin~lly 30 dirrcl~ rcl coullL~l~alL~. In combination, these results in-lic~t~ that the function of the HMGI proteins may be to m~int~in the un~lifr~,~,..li~te~l cellular state.
The diverse set of m~senr~ymal neoplasms in which HMGI-C is frequently disrupted by translocations of 12ql3-15 inrl~ es lipomas, uterine 35 leio"lyol"a, pulmonary hamartoma and pleomorphic adenom~c of salivary gland.
Another cytogenetic subgroup which can be ;~leltti~t~d in this set of tumors is SUB~ 1 l 1 U 1~ (RULE 26) .. . .. ..

CA 022~963~ 1999-01-06 wo 98/50536 PCT/US97/21299 .

charart~ri7P~ by rearrangements at 6p21-23. Intriguingly, HMGI(Y) has previously been localized to this chromosomal area.

Tr~ oc~ion Br~kl~Q~ c U~lr~ of the HMGI-C Gene in Uterine Lei~-llyu-llata Uterine leiomyomata, also known as fibroids, are the most cornmon pelvic tumors in women. Systematic histologic e~min~tion of hysterectomy specimens has shown a prevalence as high as 77% for these tumors in women of 0 reproductive age. Although benign, uterine leiomyomata constitute a major health problem as they are associated with abnormal uterine bleeding, pelvic pain, urinary incollthle.lce, spontaneous abortion, premature delivery, and infertility.
Symptomatic fibroids are the leading indication for hysterectomy, accounting for27% of the estim~tPd 680,000 procedures performed annually in the United States.
Several different consistent chromosomal rearrange.ne.ll~ have been ~ Pntifier~ in uterine leiomyomata, and they suggest involvement of a critical gene on chromosome 12 in the pathobiology. A translocation involving chromosomes 12 and 14, t(12; 14)(ql4-15;q23-24), le~rcse..l~ one of the most common rearr~ngemPnts, although trisomy 12, inversions and duplications of 12ql4-ql5, and translocations of 12ql4-ql5 with chromosomes other than 14 are not uncommon. The breakpoint in 12ql4-ql5 in uterine leiomyomata is in an intriguing chromosomal region because it is also the location of consistent rearrangements in other benign solid tumors, including lipomas and pleomorphic a(l~nnm~c of the salivary gland. Rearran~ell-e~ of 12ql3-15 have been reported in pulmonary chondroid hamartoma, en~lom~trial polyps, epithelial breast tumors,hern~nEiopericytoma, and an aggressive angiomyxoma. These tumors have the common proL,cllies of being mPsenrhyme-derived and benign. Therefore, it has been hypothesized that a single gene involved in mPsenrl yme dirre~c~.liation and 3 o growth could be responsible for these multiple tumor types.

H.R. Asher et al. (1995) reported that HMGI-C, an architectural factor that functions in l,al.scliplional regulation, is disrupted by rearrangement at the 12ql4-15 chromosomal breakpoint in lipomas and suggests a role for HMGI-C
3 5 in adipogenesis and mPsenrllyme dirr~ liation.

SUB~ 1 l 1 U l ~; ~iH h'l~; l (RULE 26) CA 022~963~ 1999-01-06 WO 9~/50536 PCT/US97/21299 g X. Zhou et al., (1995) shows that the pygmy pheno~yl.e arises from the inactivation of HMGI-C which function as architect~lral factors in the nuclear scaffold and are critical in the assembly of stereospeci~lc transcriptional complexes.

A.C. Finlay et al. (1951) discloses an antibiotic obtained from culture filtra~es of an Actinomycete, Streptomyces netropsis, isolated from a soil sample and ac~ign~ the name Netropsin.

A. DiMarco et al. (1962) disclose the physicorh~ l properties of, and the results obtained against some experimental tumors with, the antibiotic of the netropsin group, distamycin A. Distamycin A is reported to exhibit a strong irlhibition on ascites tumors [Ehrlich and sarcoma 180 (S180) ] and delays the growth of solid tumors (Ehrlich ca,~;inc,lna, S180, Walker carcinoma, and Oberling-Guerin-Guerin myeloma). Distamycin A is also reported to decrease the mitotic index of the Ehrlich ascites tumor and induces mitotic damages of tumor cells.

M.L. Kopka et al. (1985) discloses that X-ray analysis of the complex of netropsin with the B-DNA ~odec~m~r of sequenre C-G-C-G-A-A-T-T-~rC-G-C-G reveals that the antitllmnr antibiotic binds within the minor groove by displacing the water molecules of the spine of hydration. Nello~sin amide NH is reported to furnish hydrogen bonds to bridge DNA adenine N-3 and thymine 0-2 atoms occurring on ~ cent base pairs and opposite helix strands, exactly as withthe spine of hydration.
R. Reeves et al. (1990) discloses the ~lom~in.~ of the ~ n high mobility group (HMG)I chromosomal proteins nPcess~.~ and sufficient for binding to the narrow minor groove of stretches of A T-rich DNA. The three highly conserved regions within each of the known HMG-I proteins is reported to 3 o be closely related to the consensus seque~e T-P-K-R-P-R-C-R-P-K-K and that a synthetic oligopeptide corresponding to this consensus "binding domain'l (BD) sequ~rlce specifically binds to substrate DNA in a manner similar to the intact HMG-I proteins. Molecular Corey-Pauling-Koltun model building and computer ~imul~tions employing energy minimi7~tion programs to predict structure are 3 5 reported to suggest that the conse.,~us BD peptide has a secondary structure similar to the antitumor and antiviral drugs n~llup~ and distamycin, and to the dye SUBS 1 l 1 U 1~ ; l (~ULE 26) CA 022~963~ 1999-01-06 Hoechst 33258 and that in vitro these ligands, which also pl~felclltially bind to A
T-rich DNA, have been demonstrated to effectively compete with both the BD
peptide and the HMG-I proteins for DNA binding. The BD peptide is also reported to contain novel structural features such as a Asx bend or "hook" at its 5 amino-terminal end and laterally projecting cationic Arg/Lys side chains or "bristles" which may contribute to the binding plul)e.Lies of the HMGI proteins.The predicted BD peptide structure, referred to as the "A T-hook," represents a DNA-binding motif capable of binding to the minor groove of stretches of A T
base pairs.

European patent EP727~87A1 (960821) (Bullerdiek et al.) discloses the Multi-tumor Aberrant Growth (MAG) gene having the nucleotide sequence of any one of the strands of any one of the members of the High Mobility Group protein genes or LIM protein genes.

SUl~MARY OF THE INVENTION

The present invention pertains to a method for treating obesity in a m~mm~l which co~ lises reducing the biological activity of HMGI genes in the m~mm~l In this embo~im~ t~ at least 10% of the biological activity of HMGI
genes is reduced, and preferably at least 50% of the biological activity of HMGIgenes is reduced. In one embodiment, the biological activity of HMGI-C genes is reduced, and in another embodiment, the biological activity of HMGI-(Y) genes isreduced. The m~,..,..~l is preferably leptin-deficient or leptin receptor-deficient.
The reduction in biological activity of HMGI genes may be achieved by inhibitingthe expression of HMGI genes, by ~dmini.ctering to the m~mm~l a the.i1l e~ir~llyeffective amount of an oligonucleotide which has a nucleotide sequence 3 o complr-Tnrr~t~ry to at least a portion of the mRNA of the HMGI gene, by inhibiting the DNA-binding activity of HMGI genes, by a~mini.ctering to the m~mm~l a the~ ir~lly effective amount of an inhibitor compound selected from the group con~ tin~ of ntL,o~sil~, distamycin A, or Hoechst 33258 (bisben~imi-le), or by inhibiting the protein-protein interactions of HMGI proteins. The m~mm~l may be 3 5 a human or a rodent. The biological activity of HMGI genes may also be subst~nti~lly reduced by breeding the m~mm~l with an inactivated HMGI gene SUB~ 1 l l U l ~ ; r (RULE 26) CA 022~963~ 1999-01-06 ~ WO 98/50536 PCT/US97/21299 sequence introduced into the m~mm~l, or an anceslor of the m~mm~l, at an embr~vonic stage. The inactivated HMGI gene sequence may be an inactivated HMGI-C gene sequence and may be the inactivated HMGI-C gene sequence set out in Figure lO~

In another embo~imPnt, the present invention pertains to a method for treating a tumor in a patient by reducing the biological activity of normal HMGI genes which co,ll~,ises ~lmini~tPring to the patient a therapeutic~lly effective amount of an inhibitor compound active against normal HMGI-C or 0 HMGI(Y) genes. In this embodiment, the biological activity of normal HMGI-C
genes may be reduced or the biological activity of normal HMGI-(Y) genes may be recluce~ The reduction in biological activity of normal HMGI genes may be achieved by inhibiting the expression of normal HMGI genes, by a~mini.ctering tothe patient a therapeutic~lly er~;Live amount of an oligonucleotide which has a 15 nucleotide sequence complenlpnt~y to at least a portion of the mRNA of the normal HMGI gene, by inhibiting the DNA-binding activity of normal HMGI
genes, or by ~lm;.~ g to the patient a therapeutir~lly effective amount of an inhibitor compound selecte(l from the group con.~i~ting of ll~kopsill, distamycin A, or Hoechst 33258 (bisbe~?;~ P)~ In one embotlimP~t the tumor is mesenchyme-2 o derived and benign and may be uterine leiomyomata, lipomas, pleomorphic adenomas of the salivary gland, pulmon~ry chondroid hamartoma, enrlometrial polyps, epithelial breast tumors, hPn ~ngiQpericytoma, or angiomyxoma, and is preferably uterine leiomyomata, lipomas, or pleomorphic adenomas of the salivarygland. In another embodiment, the tumor is a m~lign~nt tumor of epithelial origin 25 and may be a carcinoma of the lung, colon, breast, prostate~ thyroid gland, or skin.
The reduction in biological activity of normal HMGI genes may be achieved by inhibiting the protein-protein interactions of H M GI proteins.

In yet another embodirnent, the present invention pertains to a 3 o method of producing a transgenic non-human m~mm~l, the germ cells and somatic cells of which contain an inactivated H M GI gene sequence introduced into the m~mm~l, or an ancestor of the m~mm~l, at an embryonic stage. In this embo-limPI~t, the inactivated HMGI gene sequence may be an inactivated H M GI-C
gene sequence and may be the inactivated HMGI-C gene sequence set out in Figure 3 5 lO. Preferably, the genome of the m~mm~l does not encode for both the functionally active leptin gene and the functionally active HMGI genes.

SUB~ 111 U l ~ ; l (RULE 26) W O 98/50536 PCT~US97/21299 In yet another embodiment, the present invention pertains to a method for sele~n~l,g c~n~ te compounds capable of inhibiting the biological activity of normal HMGI l)roteins, or a fragment thereof. The method colllprises5 the steps of (a) incubating a HMGI protein, or a fragment thereof, with a c~nt~ t~
compound under conditions which promote optimal interaction; and (b) measuring the binding affinity of the c~nt~ te compound to the HMGI protein. or a fragmentthereof; and (c) determinin~ from the binding affinity which c~n~ te compounds inhibit the biological activity of HMGI proteins, or a fragment thereof. The 0 can~ te compound may inhibit the biological activity of normal HMGI proteil,s,or a fragment thereof, in an amount of at least 10%. The binding affinity may bemeasured using a scintillation proximity assay or a fluorescence polarization assay.

In yet another embodiment, the present invention pertains to a 5 method for scl~enillg c~n~ te compounds capable of inhibiting the biological activity of normal HMGI genes. The method comprises the steps of (a) transfecting into a cell a DNA construct which contains a l~,polLel gene under control of a normal HMGI protein-regulated promoter; (b) ~lmi-~is~ g to the cella c~n~ te compound; (c) measuring the levels of reporter gene expression; and 2 0 (d) ~letermining from the levels of reporter gene expression which c~n~ ate compounds inhibit the HMGI biological activity. In this embodiment, the c~nrli(l~te compound may inhibit the biological activity of normal HMGI genes inan amount of at least 10%.

In yet another embodiment, the present invention pertains to a method for cletPcting normal HMGI ~loteh~s as a diagnostic marker for a tumor using a probe that recognizes normal HMGI proleills. The method comprises the steps of (a) cont~ting normal HMGI proteins from a sample from a patient with a probe which binds to HMGl proteins; and (b) analyzing for normal HMGI proteins by ~letecting levels of the probe bound to the normal HMGI
proteil,s, wherein the presence of normal HMGI ~roleills in the sample is positive for a tumor. In this embodiment, normal HMGI-C proteins may be detected or normal HMGI(Y) proteins may be detect~cl. In one embodiment, the tumor is mPsPnrhyme-derived and benign and may be uterine leiomyomata, lipomas, pleomorphic ~enom~s of the salivary gland, pulmo~ry chondroid hamartoma, endometrial polyps, epithelial breast tumors, h~m~ngiopericytoma, SUB~ 1 l l IJ 1~ ; l (RU~E 26) CA 022~963~ 1999-01-06 PCT/US97nl299 ~ WO 98/50536 or angiomyxoma. In another embo~imPnt, the tumor is a m~lign~nt tumor of epithelial origin and may be a cdlchlo,na of the lung, colon, breast, prostate, thyroid gland, or skin. The probe may be an antibody, the sample may be a biopsy sample, a urine sample, a blood sample, a feces sample, or a saliva s sample, and the method may be a histological assay, biochemir~l assay, flow cytometry assay, Western blot assay, or solution assay. A positive and negative control sample may be treated according to the method to assess the level of normal HMGI l)lolehls in a tumor sample and a nontumor sample, respectively.

0 In yet another embodim~Pnt the present invention pertains to a method for ~etecting antibodies to normal HMGI ~)lolt:ins using a probe that recognizes antibodies to HMGI normal ~IOl~illS. The method complises the steps of (a) treating a sample from a patient with a probe which binds to antibodies to nor}nal HMGI proteins; and (b) analyzing for antibodies to HMGI pr~"eins by rlPtectin~ levels of the probe bound to the antibodies to HMGI proteins, whereinthe presence of antibodies to normal HMGI pl'OteillS in the sample is positive for a tumor. In this embo~limPnt7 antibodies to normal HMGI-C may be ~let~PctP~ or antibodies to normal HMGI(Y) may be detectPd The probe may be normal HMGI-C or HMGI(Y) proteins. In one emboflimPnt, the tumor is mPspnrllyme-derived and benign and may be uterine leiomyomata, lipomas, pleomorphic ~ennm~ of the salivary gland, pulmonary chondroid hamartoma, en~lom~otrial polyps, epithelial breast tumors, h~m~giopericytoma, or angiomyxoma. In another embo-lim~nt, the tumor is a m~lign~nt tumor of epithelial origin and maybe a carcinoma of the lung, colon, breast, prostate, thyroid gland, or skin. The2 5 sample may be a biopsy sample, a urine sample, a blood sample, a feces sample, or a saliva sample and the method may be a histological assay, bioc-hPmil~l assay, flow cytometry assay, Western blot assay, or solution assay.

In yet another embodiment, the present invention pertains to HMGI
genes and proteins for use as a starting point to isolate d-)wl~llealll target genes regulated by the HMGI genes and ~lolcills.

SUB~ l~l l U 1~; S~;~; 1 (RULE 26) . W 0 98/50536 PCTrUS97/21299 B~U~EF D E S C~LrPrrIO N O F TEnE ~IGInRU3S

Figures l(A~ and l(B) illustrate the genomic structure of the human 5 HMGI-C gene.

Figures 2(A) through 2(F) illustrate FISH mapping of HMGI-C
lambda clones to lipoma tumor metaphase chromosomes from three lipomas revealing rearrangement of HMGI-C in all three tumors.

Figure 3 illustrates RT-PCR amplification of HMGI-C chimeric transcripts.

Figure 4 illustrates rearrang~mellLs of 12ql5 in human lipomas which disrupt the HMGI-C gene and produce chimeric tlansc.i~

Figure 5 illustrates RT-PCR using primers located on either side of the fusion site beLw~ell HMGI-C and novel sequences.

2 o Figures 6(A) and 6(B) illustrate novel sequences fused to the DNA
binding-domains of HMGI-C which encode tl~lls~ ional regulatory domains.

Figure 7 illustrates the structure and domain or~ni7~tion of HMGI-C and the predicted fusion proteins.
Figures 8(A) through (D) illustrate the i-lentifir~tion and genomic characterization of the HMGI-C gene at the pygmy locus in normal and mutant alleles.

Figure 9 illustrates HMGI-C gene ~,iession of three alleles at the mouse pygmy locus.

Figures lO(A) through(C) illustrate targeted disruption of the HMGI-C gene.

SUB~ 1 l l U 1~ (RULE 26) ' W O 98/50536 PCTrUS97/21299 Figures 11(A) through (C) illustrate expression of HMGI-C in development and growth.

Figure 12 illusLLdl~s a Northern blot demonstrating that expression 5 of the HMGI genes in obesity is dr~m~tis~lly increased. RNA isolated from the adipose tissue of two month old mice was used for Northern blot analysis (wt, wildtype; od, obese; db, di~beti~).

Figure 13 is a photograph illll~trating the phenotypic effects of 0 HMGI inactivation in obese ~"~ . Genotypes of the various progeny are shown under the photographs. P~ hle~ y~ body weight of the leptin-deficient obese mice (+/+ ob/ob) is reduced from 80 gram to 25 gram (normal weight) following HMGI-C inactivation (pg/pg ob/ob).

Figure 14 is a bar graph illustrating that HMGI inhibition reverses the hyperphagia of obese mice. Figure 14 also shows the effects of genotype on food col~u~ ion. Daily food co.,sunl~tion is calculated as equal to Iweight of food at 0 hours] minus [(weight of food at 24 hours) plus (food wasted)].

Figure 15 is a graph illustrating that irlhibition of HMGI supresses tumorigenesis. Figure 15A shows that knockout mice developed tumors with a frequency ten times lower than in the control animals. Figure 15B shows that tumor multiplicity exhibited a 20-fold decrease following HMGI-C inhibition.
Solid squares refer to normal mice and solid triangles refer to mice without HMGI-C.

DETAILED DESCRIPTION OF THE ~VENTION

Aberrations in the genetic mPcll~ni.cm~ that control growth and proliferation have emerged as a primary event in carcinogenesis. The function ofHMGI-C and HMGI(Y), two embryonically expressed DNA-binding proteins, was investig~t~d because their ~ sslon is highly associated with tumor development.
3 5 Disruptions of either HMGI-C or HMGI(Y) in humans result in a diverse array of solid m~senrhymal tumors. Most prominent among these neoplasms are uterine SUB~ 1 l 1 U 1~ ; l (RULE 26) ~ . . ~ . . .

leiomyomata, the most common pelvic tumors in women and the in-lic~tiQn for over 200,000 hysterectomies annually in the United States. In tumors of m~mm~ry and thyroid glands as well as in prostate cancer, HMGI expression is highly correlated with tumor progression and metastasis, suggesting that these proteins can 5 be used for as progression markers for a variety of tumor types.

Further proof for the pivotal role of HMGI proteins in both normal and pathological growth was obtained in the mouse system. Homologous recombination was used to inactivate murine HMGI-C gene. Demons~lalillg the 0 importance of the HMGI genes in growth regulation, HMGI-C knockout mice exhibit ci~nifir~nt growth retardation (mutant mice are 60% smaller than their wild-type litterm~trs) with the reduction in most tissues commensurate with the overall decrease in the body weight. Even more importantly, these pygmy mice are highly resistant to ch~mic~lly in~ red skin cancer. Specifically, the frequency 5 of tumor development in the knockout mice is 40% of that in the control anim~lc and tumor multiplicity exhibits a 20-fold decrease. Independently, inhibition ofHMGI-C synthesis was shown to render thyroid epithelial cells intransigent to retroviral ~lal~ro~ tin~. At the molecular level, HMGI proteins function in transcriptional regulation by promoting coop~dtive binding of the transcription 20 factors to DNA. Deregulation of the do~ edll~ target genes can easily accountfor the important biological roles of the HMGI proteh~s as well as for the dramatic consequences of their in~lol,liate expression.

Lipomas are one of the most common mesenchymal neoplasms in 25 hllm~n.c. They are characterized by consistent cytogenetic aberrations involving chromosome 12 in bands ql4-15. Inl~ ingly, this region is also the site of rearrangement for other m~senrllymally derived tumors. The present invention demonstrates that HMGI-C, an arrhitrchlral factor that functions in transcriptional regulation, has been disrupted by rearrangement at the 12ql4-15 chromosomal 30 breakpoint in lipomas. Chimeric transcripts were isolated from two lipomas inwhich HMGI-C DNA-binding domains (A-T hook motifs) are fused to either a LIM or an acidic transactivation domain. These results identify the first gene rearranged in a benign neoplastic process that does not proceed to a n~ n~nry and suggest a role for HMGI-C in adipogenesis and lnesenrllyme dir~ ialion~

SUBSTITUTE ~ (RULE 26) . .

CA 022~963~ l999-0l-06 W O 98t~0536 PCTrJS97121299 HMGI-C is an attractive c~n~ t~ gene to be implicated in lipoma formation. This gene is required in lla~ro~ ation (Berlingieri et al., 1995) and is a transcriptional regulatory factor as are many genes i(lP~tifiPd at translocation breakpoints in a variety of tumors (Rabbitts, 1994). Secondly, disruption of 5 HMGI-C leads to mice of small stature which, most intriguingly, have disproportionately less body fat than normal litterm~tes (Benson and Chada, 1994).
Finally, mouse HMGI-C maps to a region syntenic to human 12ql4-15 which is the area most frequently rearranged in lipomas (Mandahl et al., 1988). Therefore, the human homolog of the mouse HMGI-C gene was cloned and its possible role in 1 o lipomas investig~t~d .

Growth is one of the filn-1~m~nt~l aspects in the development of an organism. Classical genetic studies have isolated four viable, spontaneous mousemllt~nt~ (Green, 1989) disrupted in growth, leading to dwarfism. Pygmy is unique5 among these ..-u~ because its phenotype cannot be explained by aberrations in the growth hormone-insulin-like growth factor endocrine pathway (Lin, 1993; Li, et al., 1990; Sinha et al., 1979; Nissley et al., 1980). The present invention shows that the pygmy phenotype arises from the i-~c~iv~Lion of HMGI-C and are criticalin the assembly of stereospecific l~allsclilJlional complexes (Tjian & Maniatis,20 1994). In addition, HMGI-C and the other HMGI family m~mber, HMGI(Y)( Johnson et al., 1988), were found to be e~les~ed predominantly during embryogenesis. The HMGI family are known to be regulated by cell cycle dependent phosphorylation which alters their DNA binding affinity (Reeves et al., 1991). Overall, these results demonstrate the important role of HMGI proteins in25 m~mm~ n growth and development.

Among the most promin~-nt characteristics cu~ y exhibited by cancer cells are karyotypic aberrations which disturb genes essential for the regulation of filn(l~mPnt~l cellular processes. A wide array of solid mesenchymal 30 tumors is characterized by recurrent rearrangements of chromosomal bands 12ql3-15 or 6p21-23. This study shows that HMGI expression is normally restricted to undirÇclr~ ted, rapidly dividing cells but is activated in dir~L."i~ted adipocytes following translocations of 12ql3-15 or 6p21-23 in human lipomas. The present invention shows that the molecular ~a~lw~y of tumor development is dictated by 3 5 the precise nature of HMGI disruption and that HMGI mise~l,res~ion in a dirr~lc.~ t~l cell is a pivotal event in benign tumorigenesis.

SUB~ 111 U 1~ l (RULE 26) CA 022~963~ 1999-01-06 WO 98/50536 PCT/US97/2l299 Uterine leiomyomata are the most common pelvic tumors in women and are the indication for more than 200,000 hysterectomies annually in the United States. Rea.lallgel.lent of chromosome 12 in bands ql4-qlS is characteristic of uterine leiomyomata and other benign mesenrhymal tumors, and a YAC spanning chromosome 12 translocation breakpoints was i~nti~lP~l in a uterine leiomyoma, pulmonary chondroid hamartoma, and lipoma. Recently, it was demonstrated that HMGI-C, an architectural factor mapping within the YAC, is disrupted in lipomas,res~llting in novel fusion transcripts. This study concerns the loc~li7~tion of 0 translocation breakpoints in seven uterine leiomyomata 10 to > 100 kb upstream of HMGI-C by use of fluorescence in situ hybridization. These findings suggest a different pathobiologic ~ chAni!i... in uterine leiomyomata from that in lipomas.
HMGI-C is the first gene i~lPntifie(l in chromosom~l rearrangements in uterine leiomyomata and has important implications for an undersr~n(linl~ of benign 15 mesenchymal proliferation and differentiation.

Recently, molecular tlicsection of this chromosomal region has su~ost~nti~tP~ this hypothesis. To identify a gene at the breakpoint on chromosome 12 in uterine leiomyomata, a high-density physical map of the t(12;14) breakpoint 20 region was constructed and i~lPntifi~tl a YAC, 981fll, that spans the translocation breakpoints in a uterine leiomyomata, pulmonary chondroid ha"la"oma and a lipoma. Further detailed characte,i~lion showed that the gene for HMGI-C, an architPchlral factor that is a non-histone component of chromatin, maps within 981fll and is disrupted in lipomas. HMGI-C is rearranged in lipomas with 25 chromosome 12 translocations, resulting in novel chimeric trans~ that fuse the DNA-binding A-T hook domains of HMGIC with potential transcriptional activation ~lo~in.c.

Obesity Ml~t~tionc of HMGI-C are re~ollsible for ovelg~ vlh of fat lipomas, tumors composed of mature fat cells (Ashar et al., 1995; Schoenmakers et al., 1995). Removal (inactivation, inhibition, etc.) of HMGI-C in normal mice results in ~nim~l.c with a 20-fold reduction in the amount of fat tissue (Zhou et al., 35 1995). Removal (inactivation, inhibition, etc.) of HMGI-C in leptin-deficientobese mice, which are a widely accepted model of human obesity, decreases the SUB~ 1 l l U 1~ ; l (RULE 26) CA 022~963~ 1999-01-06 amount of fat tissue in these animals and le~Lon,s their normal weight (from 80 gram to 25 gram). Even a partial inactivation of HMGI-C, such as may be produced by a drug, results in the decreased amount of fat and in a decreased weight of obese animals. Moreover, food intake is also (1iminichetl as a result of 5 HMGI-C inhibition. This last study dem~ ales the use of HMGI-C inhibition to regulate the amount of adipose tissue. Applicants also describe small molecule inhibitors and ~ ice!~e oligonucleotides which can be used to inhibit the biological activity of HMGI-C.

o Tumorigenesis The first class of tumors treatable by the present invention includes carcinomas, m~lign~nt tumors of epithelial origin, which are commonly referred to as cancer, and include carcinomas of the lung, colon, breast, prostate, thyroid 5 gland and skin. A number of papers describe a correlation between tumor development and the presence of HMGI proteins. Specifically, HMGI proteins are absent in normal adult cells but are always found in m~lign~ns tumors (carcinomas). This, however, is a correlative observation which does not teach anything about treating those tumors. Applicants have discovered that HMGI-C
20 inhibition will be an effective method of treating tumors. Specifically, applicants employed a chP~i~l skin carcinogenesis assay, a widely accepted model of tumorigenesis, which is applicable not only to skin carcinomas but also to carcinomas of lung, colon, breast, prostate and thyroid gland. Two sets of mice were used, one with normal amounts of HMGI-C and one without any HMGI-C, 25 and to their skins certain chP~ lc known to induce cancers were applied. While normal mice developed tumors as expected, mice without HMGI-C were resistant to tumorigenPsi.c. The same types of inhibitors that were described for treating for obesity can be used for treating cancer.

Berlingieri et al., 1995 demonstrates that inhibition of HMGI-C by ~nticenc~P. in vitro prevents cellular llan~rollllation. Transformation is a process very different from tumorigenesis (carcinogenesis; tumor development; tumor gro~vth, etc.). In studying transformation, one isolates cells, puts them in test tubes, subjects them to various stimuli (I~hPmir~l~, viral infections, irradiation etc.) and analyzes their ability to be transrolll,ed, i.e., exhibit characteristics dirrelent from those of normal cells. In studying tumorigen~Psic, one takes anim~lc and SUBST~TUTE ~H~ (RUT E 26) *rB

CA 022~963~ 1999-01-06 studies their ability to develop tumors. Therefore, Berlingieri et al., 1995 does not teach a method for SUylJ~cSSillg tumorigenesis.

The second class of tumors treatable by the present invention includes benign tumors of mPsenrhymal origin7 as opposed to m~lign~nt tumors of epithelial origin, including lipomas, uterine leiomyomas and other tumors (Ashar et al., 1995; Schoe-nm~kPrs et al., 1995). These tumors are benign and therefore are not cancers. However, one group in this class, uterine leiomyomas, present a cignifir~nt health problem and complications associated with them (pain, infertility, 0 etc.) result in 200,000 operations to remove the uterus annually in the U.S.
Tumors of this second class are different from the first class (cancers) because in the first class, tumors have increased amounts of normal HMGI-C while tumors of the second class develop due to HMGI-C mutations.

As set out above, there are two classes of tumors which HMGI
genes are responsible for, and hence which are treatable by the present invention:
(1) benign mesel1chyll.al tumors, and (2) m~lign~nt epithelial tumors. Tumors oftype (2), m~lign~nt epithelial tumors, collsliluLe over 99% of all human tumors.There are two m~ch~ni~m~ by which HMGI genes can promote tumorigenesis:
2 o (a) HMGI genes can be disrupted by chromosomal translocations producing fusion ploleins in which a major part of a normal HMGI protein is replaced by a heterologous sequence derived from the translocation partner ("mutant HMGI
genes mrrh~ni.~rn"); and (b) HMGI genes can simply be activated and appear in a cell where the genes would not normally be present, without mutation ("normal HMGI genes m~rh~ni.~m ") Since HMGI proteins function in embryogenesis, the proteins should not normally be present in an adult cell. Hence the presence of these proteins, even in theirnorrnal forrn, in the "wrongl' cell can result in tumor development.
Applicants have discovered that rnerh~nicm (a), the mutant HMGI
genes mPrh~ni~m, causes tumors of type (1), benign mesenrhymal tumors.
Applicants have also discovered that mPch~ni~m (b), the normal HMGI genes m~rh~ni~m, causes tumors of type (2), m~lign~nt epithelial tumors, and that 3 5 inhibition of normal HMGI proleil~s will ~u~,ess the growth of tumors of type (2).
Applicants have further discovered that tumors of type (1), benign mPsenrhymal SUBS 111 U 1~; S~EET (~ULE 26) tumors, can be caused by ml~ch~ni~m (b), the normal HMGI genes m-och~ni~m, and not only by rn~rl~ni~m (a), the mutant HMGI genes mech~ni.~m.

In general, a normal HMGI gene (protein) is a gene not disrupted by 5 any chromosomal aberration. The sequences of the human and mouse genes and ~ro~eills are well known and are published. Moreover, there exist variations of these sequences, i~ e., conse~ i\re amino acid substitutions, changes of the nucleotide sequence outside the open reading frame (the part that actually codes for the protein), which preserve the normal molecular strucnlre and function of the 0 HMGI proteins. These variations also fall within the scope of the patent.

RESIJLTS

HMGI I'~ol~ s in Adi~oL_.,esis and Mf~ n.e Difft.~ lior~
Genomic Isolation and Characterization of the Human HMGI-C Gene To obtain genomic clones of HMGI-C, DNA from yeast strains harboring YACs, yWPR383 and yWPR384 were subcloned into the lambda PIXII
20 vector. Rec~llce there is extensive conservation (96%) between mouse and human HMGI-C homologs (Patel et al., 1994), mouse HMGI-C cDNA fra~ tc encompassing all five exons were used as probes on lambda libraries and five clones were isolated (Figure lA). Restriction mapping of lambda clones followed by Southern blot analysis allowed id~ontifi~tion of various restriction fr~gment~
2 5 cont~ining cross-hybridizing sequences. These fragments were subcloned and nucleotide sequence analysis col,fi.--.rd published data (Patel et al., 1994). The first three exons each contain a DNA binding domain encoding the A-T hook motif that is characteristic of the HMGI family (Reeves and Nissen, l990) and exons 4 and 5 encode the acidic domain of the molecule (Manfioletti et al., 199l) (Figure 3 0 lB). Notably, a large intron (>25kb) between exons 3 and 4 ~e~ es the DNA
binding domains from the r~ in-~çr of the protein (Figure lB).

Fluorescence In Situ Hybridization of Lambda HMGI-C Exon Clones to Lipoma Metaphase Chromosomes SUB~ 111 U 1~ (RULE 26) CA 022~963~ 1999-01-06 Lambda clones from 5' and 3' ends of HMGI-C were used as probes for FISH to tumor m~t~ph~e chromosomes. In lipoma ST90-375 cnnt~inin~ a t(12;15)(ql5;q24) translocation, lambda clone H403 which contains the 5' end of the gene gave a hybridization signal on the der(12), thus mapping l~ro~ al to the 5 breakpoint. In contrast, lambda clone H4002 which contains a portion of the 3'end of the HMGI-C gene, gave a hybridization signal on the der(15) and thereforemaps distal to the breakpoint (Figure 2). This result is con~i~t~nt with a disruption of HMGI-C due to the t(12; 15) in this lipoma. Two other lipomas with translocations in 12qlS were studied, similarly. In ST93-724 Cnnt~ining a lo t(3;12)(q29;qlS), lambda clone H409 cont~ining the 5' end of HMGI-C hybridized to the der(12), while the 3' end clone H4002 hybridized to the der(3) (Figure 2).
In ST91-198 with a t(12;13)(ql4-22;q21-32), the 5' clone H403 mapped on the der(13) suggesting a position distal to the breakpoint. However, from the 3' end, no hybridization to either derivative chromosome was noted in 20/20 metaphases using lambda clone H4002 inrlir~ting that this portion of HMGI-C is deleted (Figure 2). Therefore, in this tumor, the translocation appears to be proximal to HMGI-C with the S ' end of the gene retained but the 3 ' end deleted. Regardless of the chromosomal m~ch~ni~m which may include a complex rearr~ngmPnt in ST91-198, HMGI-C is disrupted in three out of three lipomas analyzed.
~entific~tion of Chimeric Transcripts The molecular structure of the HMGI-C transcripts in the lipomas was next investigat~l Total mRNA was isolated (Chirgwin et al., 1979) from primary cell cultures of ST90-375 t(12;15) and ST93-724 t(3;12) and 3' RACE
performed (Frohman et al., 1988). The resulting products were analyzed by agarose gel electrophoresis and DNA fr~gmPntc of size 441 and 627 bp were obtained from RNA samples isolated from ST90-375 and ST93-724, respectively (Figure 3). These two DNA fr~ment~ were purified, subcloned and seqnenred 3 o In both cases, sequence analysis revealed an in frame fusion of novel seq~lPnres to HMGI-C. These sequences differed between the two lipomas, and imm~ tely followed exon 3 of HMGI-C (Figure 4).

The ~lesellce and specificity of chimeric transcripts in the two lipomas were cnnfirmed further by an independent RT-PCR. cDNA was prepared from lipoma RNA samples but primers from the novel sequences, instead of oligo-SUBS 1 l 1 U 1~ (RULE 26) .

~ ' WO 98/50536 PCT/US97/21299 dT, were used for the reverse ~ scli~tion reaction so that only RNA ll~nswi~Ls spanning the translocation would result in a PCR amplification product (Figure 5).
Products of the predicted size were observed only when primers derived from the novel sequences were used to reverse transcribe RNA isolated from the corresponding cell lines. No products were seen in lipoma RNA from ST90-375 or ST93-724 when primers 724 or 375 were used, lei~ye~Lively.

Finally, the chromosomal origin of the novel sequences was ~letertninP~i using DNA p,~a.~d from a monochromosomal rodent-human somatic cell hybrid panel. Specific primers were deci~nP~l for the two novel sequPnre~c obtained from the lipoma cDNAs. PCR pelÇol~llcd on genomic DNA from the somatic cell hybrids demonstrated that the novel sequer~e fused to HMGI-C in ST93-724, with a t(3;12), was located on chromosome 3 (Figure 6) and the novel sequence from ST90-375, with a t(12;15), mapped to chromosome 15 (Figure 6).
Novel Sequences Encode for Transcriptional Regulatory Domains A ~let~ilPd co~ uL~l analysis of the novel sequenres from the two ampliffed fusion transcripts demonstrated that they encode putative ~-~nscliplional regulatory cilm~inc Tncpection of the predicted protein sequPnre from ST93-724 revealed the presence of two t~n(1P-nly arrayed LIM dom~inc (Sanchez-Garcia and Rabbitts, 1993) separated by the ch~raGteristic 8-10 amino acids (Figure 6A).
These domains are 50-60 amino acid residue motifs which are rich in cysteine andhicti~ine and were hrst i~Pntifie~l in three proteins, lin-11, Isl-1 and mec-3 (Way and Chalfie, 1988; Freyd et al., 1990; Karlsson et al., 1990). The domain is org~ni7P~l into two adjacent zinc fingers separated by a two residue linker (Feuerstein et al., 1994) and members of the LIM family of proteins may contain one or more LIM domains (Sanchez-Garcia et al., 1993). Many of the LIM-cont~ining proteins are ll~ns.;li~lion factors (Sanr~Pz-Garcia et al., 1993) and their 3 0 activity is thought to be regulated by protein-protein interactions through the ability of LIM dom~inc to dimerize (Feuerstein et al., 1994).

Cc,m~e~ analysis of the novel seqUpnre from ST90-375 did not reveal any signifir~nt homology with known sequ~P~e~. Notably, the carboxy-tPrrnin~l end of the predicted protein is highly acidic (pI 4.6) and rich in serine and threonine residues. Such domains have been implicated in llal~scli~Jtional activation SUB~ 111 u 1 k; ~ ; l (RULE 26) and have been shown to stin~ te ha,~c,iplion from remote as well as proximal positions (Mitchell and Tijan, 1989; Seipel et al., 1992).

Therefore, the ple~ d domain ol~an;~tion of the wildtype 5 HMGI-C and the fusion proLcills can be schPn ~tir~lly depicted as shown in Figure 7. In both fusion ~rotei"s, the C-termin~l domain of the wildtype HMGI-C, which does not activate ll~ns.;li~lion (Thanos and Maniatis, 1992; X.Z. and K.C., unpublished data) is replaced by ~ictinrt, potential Lldllsclip~ion regulation domains. These newly acquired functional domains in combination with the A-T
0 hooks of HMGI-C would give rise to unique ~Olcills that may contribute to the pathobiology of lipomas.

HMGI Proteins in Malnm~ Growth and Development Previous studies (King, J., 1955) had established that the pygmy phenotype could be observed at birth. T~lefol~, RNA from wildtype mouse embryos was isolated (Chirgwin, J. et al., 1979) and Northern blot analysis revealed a l,~l~c,i~t of 4.1kb (Figure 9). As expected from the genomic analysis, no d~tect~hle HMGI-C expression was observed in the spontaneous and transgenic 20 insertional mouse ~u~ . Additionally, a third allele exists at the pygmy locus (Green, M.C., 1989), ln(10)17Rk, which carries an inversion of chromosome 10 and the distal breakpoint is within intron 3 of the HMGI-C gene (data not shown).
No HMGI-C expression was d~tected in homozygous embryonic In(10)17Rk RNA
(Figure 9). Q~l~ntit~tion by phosphorimager analysis revealed that heterozygous mice expressed HMGI-C at approximately 50% wildtype levels. Therefore, the wildtype allele in the h~lc.(lzygous mice does not il1(;,case its expression levels to compensate for the loss of the deleted allele. This is consistent with the pygmymutation being semi-domin~rlt because there is a mild phenotypic effect on heterozygous mice (80% the weight of wildtype mice) (Benson, K. & Chada, K., 3 o 1994). Furth~rmore, HMGI(Y), the only other known m~tnher of the HMGI genefamily (Gro~sch~ll, R. et al., 1994), retained the same levels of e~ ssion in the mutant and wildtype mice (Figure 9). Therefore, there is no compensation by HMGI(Y) for the lack of HMGI-C expression in pygmy mice.

3 5 The mutant alleles described above arise from major disruptions of genomic DNA which result in large deletio~ or a chromosomal inversion. To SUB~ 111 U 1~ (RULE 26) CA 022~963~ 1999-01-06 ~ Wo 98/50~36 PCT/USg7/21299 exclude the possibility that a gene other than HMGI-C may be responsible for thepygmy phenotype, a mouse null mutant of HMGI-C was produced by targeted disruption. Mouse embryonic stem (ES) cells were gc~ dlcd that had 3.0kb of the HMGI-C gene, encomp~c.cing exons 1 and 2, replaced with a neomycin-resi~t~nre gene (Figure lO(A)). Matings between mice hel~,ozygous for the mllt~tecl allele produced mice homozygous for the disrupted allele (Figure lO(B)) at the expectedMendelian frequency of approximately 25% (13/51). Tmmlmnblot analysis demonstrated an ~hsçnre of HMGI-C in protein extracts from homozygous embryos (Figure lO(C)). Homozygous HMGI-C-/- mice revealed the classical 0 features of the pygmy phenotype which include reduced birthweight, craniofacial defects (shortened head) and an adult body weight of approximately 40% (39.8 +/-2.9) of wildtype littermates (Benson, K. & Chada, K., 1994). Therefo-~, it can be concluded that absence of HMGI-C expression in mice causes the pygmy phenotype.
Previously, a restricted number of adult tissues were analysed (Manfioletti, G. et al., 1991) and established that the endogenous eA~Jrcssion of HMGI-C could not be detected. Hence, a more co~ ~he~ re panel of tissues were e~r~mine~ to investigate the temporal and tissue specific eA~,tssion pattern of HMGI-C. Within the sensitivity of Northern blot analysis, HMGI-C expression was not ~et~cte~l in 18 adult tissues (data not shown). However, expression of HMGI-C was observed during mouse embryogenesis (Figure ll(A)) as early as 10.5 days post coitum (dpc), but essentially disa~l~ea~d by 15.5dpc. ~2~Tn~rk~hly, the other family member, HMGI(Y), showed a similar endogenous expression pattern (Figure ll(A)) with expression readily observed in 10.5-16.5dpc mouse embryos. The predomin~nt expression of HMGI-C and HMGI(Y) during embryogenesis suggests this arcl-itec~ral factor family functions mainly in m~mm~ n development.

The analysis of HMGI-C e~l,c~ion was further extended by its loc~li7~tion in the normal developing mouse embryo. Expression was observed in the majority of tissues and organs during embryogenesis as exemplified by the 11.5dpc mouse embryo (Figure ll(B)). Noticeably, HMGI-C eAl)lcssion was not seen in the embryonic brain except in a small, localized region of the forebrain(Figure ll(B)). This ex~ression pattern coincides with previous studies which demonstrated that most tissues in pygmy mice were 40-50 % smaller as compared to SUB~ Lll u 1~; SH 1~:~; l (RULE 26) *rB

CA 022~963~ 1999-01-06 wildtype tissues and the only tissue of normal size was found to be the brain (Benson, K. & Chada, K., 1994).

To initiate studies on the eluci~tion of the role of HMG~-C in cell 5 growth, embryonic fibroblasts were cultured from homozygous and wildtype embryos. Strikingly, the number of pg/pg embryonic fibroblasts was four-fold less as conl~a~d to wildtype fibroblasts after four days in vitro (Figure 11(C)) and was not due to cell death. This data, as well as similar studies in other systems (Ram, T. et al., 1993; BPrlingi~ri, M.T. et al., 1995), is consistent with a role for HMGI-lo C in cell proliferation and suggests that HMGI-C filn~tinn~ in a cell ~ulonol~ous manner. Fu~ ."~ore, absence of HMGI-C eAp~sion in the pygmy mutant would then lead to a decrease in cell proliferation and causes the reduced size of all the tissues except for the brain.

5 Inhibition of HMGI-C Suppresses Tumorigenesis Numerous studies, especjally those with HMGI-C inactivation of LranSgelliC mice (Zhou et al., 1995), demonstrated that HMGI proteins play a central role in both normal and aberrant growth regulation. Berlingieri et al.
20 (1995) studied the possible involvement of HMGI protein in Lf~,lsrollllation and were able to show that in vitro retroviral transformation requires the presence of HMGI-C protein. However? none of the previous reports addressed the role of the HMGI protein family in tumor growth in the context of the whole organism.
Furthermore, previous studies failed to elucidate the possible role of the HMGI
2 5 proteins in tumorigenesis in vivo.

Generation of the HMGI-C knockout mice (Zhou et al., 1995) provided a physiological model in which to study the effect of the HMGI proteinsin tumorigenesis in a defined and controlled manner. In order to determine 30 whether inhibition of the biological activity of HMGI proteins can be used tosuppress tumorigenesis, susceptibility of the knockout mice to tumor growth was ex~mine~l by subjecting 2-month-old an~mals to a two-stage carcinogenesis protocol ili7ing DMBA and TPA. Twenty HMGI-C knockout mice and 20 wildtype controls were used in this experiment. The backs of the ~nim~ were shaved 48 3 5 hours before tumor initiation and mice were inhi~terl with a single topical application of 200 nmol of 7, 12dimethylbenz[a]a~ acel.e dissolved in 200 ul of SUB~ ; l (RULE 26) CA 02259635 l999-0l-06 acetone. Starting one week later, animals were treated twice a week for the next18 weeks with topical applications of 6 nmo1 (4ug) of TPA dissolved in 200ul of acetone. Forty-eight hours after the last application, the mice were sacrificed,whole dorsal skin was excised and tumors were counted.

A striking dir~l~nce between the rates of tumorigenesis of normal and HMGI-C knockout mice was imm~di~t~ly a~al~llL. Transgenic mice without HMGI protein are highly le~islall~ to ch~ornir~lly in-hlced skin cancer. Promin~ntly, the knockout mice developed tumors with a frequency ten times lower than in the 0 control animals, see Figure 15A. Just as importantly, tumor multiplicity exhibited a 20-fold decrease following HMGI-C inhibition, see Figure 15B.

These results conclusively demonstrated that inhibition of HMGI
biological activity in the context of the whole organism was able to suppress 15 tumorigenesis. Therefore, the above studies provide proof-of-principle evidence that inhibition of HMGI biological activity can be used to su~ ss tumorigenesis such as observed in cancer patients.

Mi~ of Disrupted HMGI P~lt~ s in ~llm~n Tumors Isolation and Analysis of the Aberrant HMGI Transcripts Rearrangements of HMGI-C in human tumors always preserve the DNA-binding domains of the protein and the DNA-binding activity of the HMGI
25 archit~ct lral factors is es~enti~l for the enh~n-~er activation. Moreover, seql~enre analysis demo~Ll~led that the DNA-binding domains are completely conserved between human HMGI-C and HMGI(Y) (Figure not shown). Therefore, HMGI
expression was investigat~ in human tumors with karyotypic abnorm~lities involving chromosomal band 6p21-23.
F.~t~hli~hm~nt of cell lines is frequently associated with ~ccl~m~ tion of mutations in vitro. To exclude such artifacts, RNA was isolated directly fromfrozen tumor samples. Lipomas ST92-24269 t(4;6) and ST88-08203 t(6;11) were karyotyped and total RNA was purified from frozen tissues by cesium chloride 3 5 centrif~lg~tion. Next, ~mplifi-~tion of the HMGI lr~nsc~ L~ was performed using 3' RACE protocol. Upon analysis of the resulting reactions by gel electrophoresis, SUB~ l l 1 U l ~ (RULE 26) . .

aberrant HMGI(Y) products were readily detect~le (Figure not shown). At the same time, HMGI-C expression was not detected in these tumors (Figure not shown).

The anomalous HMGI(Y) cDNAs were further characterized by sequence analysis. In lipoma ST92-24269, the Ll~nsc~ enro~ed for the S' end of HMGI(Y) followed by a novel seq~erlre (Figure not shown). Comparison of this latter sequence to the Genbank d~t~ba~e revealed that it was derived from the 3'UTR of wild-type HMGI(Y). PCR analysis of the genomic DNA from tumor ST92-24269 determined the llallscli~t was produced by an internal deletion of both exonic and intronic sequences (unpublished data) which removed 922 bp from the wild-type HMGI(Y) cDNA (Figure not shown).

Seq~l~nrin~ of the aberrant transcript in lipoma ST88-08203 revealed a fully intact HMGI(Y) open reading frame . A ~let~ile~ molecular analysis demo~ dt~d that this ~.allsclil)t was produced by the removal of 923 bp of the wild-type sequence from exon 8 (Figure not shown). I~ Lingly, the rearrangement was limited to the 3'UTR of the gene, leaving the coding seq~le~reintact. The~,fole, the aberrant tlanscliyt~ isolated from the lipomas with rearrangements of 6p21 are produced by internal deletions within the HMGI(Y) gene. The fin-lin~ in both tumors were confirmed by an independent RT-PCR in which an HMGI(Y)-specific reverse primer rather than oligo-(dT) was used for reverse transcription and subsequent PCR (Figure not shown).

2 5 In lipoma ST92-24269, the predicted HMGI(Y) fusion protein consists of the first two DNA-binding dom~in~ of HMGI(Y) fused in frame to an ul~inlellupted open reading frame (ORF) encoding for 108 amino acid residues. A
~~et~ (l e~min~tion of the ORF revealed an llmls~ ly high content of proline (17%) which is indicative of a potential transcriptional regulatory domain (Figure not shown). Therefore, the overall structure of this HMGI(Y) fusion protein is rern~rk~bly similar to ploteins produced by disruptions at 12ql3-15 which juxtaposed DNA-binding domains of HMGI-C to ~L~Liv~ lldllscli~lional regulatory domains.

3 5 Translation of the HMGI(Y) aberrant transcript in the tumor ST88-08203 predicted a normal protein. In contrast, in previously described lipomas SUB~ I l 1 u l ~ ; l (RULE 26) CA 022~963~ 1999-01-06 wo 98/50536 PCTtUss7/212ss chirneric HMGI-C Llal sc~ encoded for novel fusion proteins whose formation was proposed to be ~Pcec.c~ry for lipoma development. To establish whether differences in the overall domain o.g~ ion of the HMGI(Y) and HMGI-C
fusion pfoLeills found in lipomas are due to the distinct propcllies of these two 5 genes, an additional tumor with HMGI-C rearrangement, lipoma ST91-198 t(12;15), was analyzed. RNA was isolated from the primary cell culture and 3' RACE used to amplify the HMGI-C chimeric transcript (unpublished data). The molecular analysis of this cDNA revealed that it preserved the first three exons of HMGI-C that encode for the HMGI DNA-binding domains. However, the 10 endogenous HMGI exons four and five were removed and replaced by a heterologous sequence (Figure not shown). Notably, an in-frame stop codon present in this sequence tennin~tPs translation of the chimeric tl~nscli~t afteradding only ten amino acid residues to the HMGI-C DNA-binding domains. The sequence of the novel peptide did not contain any distinguishing features and 5 revealed no .~i~nificant homology with known proteins. Chromosomal rearrangement in tumor ST91-198 therefore results in a Ll~ t~l protein that consists mainly of the HMGI-C AT-hooks. Accc)l-lingly, a simple truncation of either HMGI(~) or HMGI-C is sufficient to cause lipomas.

2 o Lipomas Can Bypass Expression of the Wild-type HMGI Allele Expression of the wildtype HMGI proteins is highly associated with transformation and can be det~ctP-~l in a wide variety of tumors. Moreover, inhibition of HMGI-C synthesis was shown to render several distinct cell types 2 5 intransigent to retroviral transformation, suggesting that HMGI expression is required for tumorige~si~. Appreciable levels of wild-type HMGI(Y) expression that were found in tumor ST88-08203 (Figures not shown~ are in agreement with this hypothesis. Su~ ish~gly, the non-,calr~,lged allele was not expressed in lipoma ST92-24269 (Pigures not shown) where an HMGI(Y) fusion protein was 3 0 i~lentifi~(l.

In lipomas with rearrange.l,ellLs of 12ql3-15, chimeric transcripts are produced by the juxtaposition of HMGI-C with the heterologous sequences and therefore cannot be readily ~mplifi~d in the same PCR reaction with the wildtype35 cDNA. To assess the expression of the wild-type HMGI-C in these tumors, the highly polymorphic microsatellite sequence located in the 5' UTR of HMGI-C was SUB~ l l 1 U 1~ ; l (RULE 26) ~ ~ W O 98/50536 PCT~US97/21299 employed. Oligonucleotide primers complimPnt~ry to the sequences fl~n1-ing this polypyrimidine tract were synth~si~Pd and used for RT-PCR (Figure not shown).
Again, expression from the non-rearranged allele was only observed in the tumor with a llullcated HMGI-C. No expression of the wildtype allele but not in the 5 lipoma ST93-724, in which HMGI-C DNA-binding dom~in~ were fused to LIM
domains, motifs that function in transcriptional regulation (Figure not shown).

Dirre~ t~d Adipocytes Express HMGI-C in Lipomas but not in Normal Fat 0 During development, the ~ ression of the HMGI proteins is tightly regulated. HMGI expression is found in the developing tissues and organs of the mouse embryo but es~Pnti~lly ~ s~re~rs by the end of h~ uLe~ e development and can no longer be found postnatally. To confirm that HMGI ~x~-cssion in lipomas is not a consequence of the endogenous HMGI expression by the adult adipose tissue, immllnocytoch~ try was pel~l.ned with an antibody raised against HMGI-C. In full agl.,emellt with llu~llel'uus previous finrljngc which demonstrated that HMGI ~loleins are not eA~r~ ssed by diffclc~ te~l cells or adult tissues, HMGI-C expression could not be ~3etected in the adult adipose tissue (Figure not shown). Futhermore, RT-PCR with primers specific for HMGI-C and 2 o HMGI(Y) col-f,~ d that HMGI genes are not expressed in normal fat (unpublished data). However, the majority of difr ,~ tecl adipocytes in these neoplasms stained positively for HMGI-C (Figure not shown). Overall, HMGI-C
expression was detected in 75% (22 out of 29) of tumors (unpublished data).

2 5 Tr~n~o~tinn Bre~kl~oint,s U~.~t~ of the HMGI-C Gene in Uterine L~;o.~ o-l,ala FISH analysis was performed on metaphase cells from uterine leiomyomata with chromosome 12 rearrangements (Table not shown) by use of clones from the 5' and 3' ends of HMGI-C (Figure not shown). In contrast to lipomas, where translocation occurs in frame following exon 3, both 5' and 3' clones hybridized only, on the rearranged chromosome not derived from chromosome 12, in addition to the normal 12 homolog, indicating that the entire sequence encoding HMGI-C maps distal to the translocation breakpoint. In two uterine leiomyomata with typical t(12,14) translocations (ST90-194 and ST93-738), breakpoints mapped within the same lambda clones approximately 10 kb u~ .

SUB~ l l I U 1~ (RULE 26) , CA 022~963~ 1999-01-06 of exon 1 of HMGI-C (Figures not shown). These breakpoints were verified by Southern blot analysis, a 3.3 kb probe from lambda clone H528 detect~d rearranged bands in both tumors (unpublished data). In ST92-224, a uterine leiomyoma with a variant translocation involving chromosome 1, the breakpoint mapped within this same region, int1i~ting that this site on chromosome 12 contains a critical region for l~_a~ gell-ent regardless of the chromosomal origin of the translocated material. FISH analysis of ST94-114, another uterine leiomyoma with a characteristic t(12;14) revealed a breakpoint approximately 100kb S' of HMGIC. In two other uterine leiomyomata (ST93-165 and ST89-171), breakpoints occurred more than 100 kb u~sll~,alll of HMGI-C as the most 5' lambda clone in the contig (H121) is translocated to the der(l4) chromosome in these tumors. ST89-171 contains two normal chromosome 12 homologs in addition to a der(l4)t(12;14); therefore, hybridization signals corresponding tothree copies of HMGI-C were det~cte~l (Figure not shown). Finally, another uterine leiomyoma (ST93-220) with an atypical cytogenetic rearrangement in whichthe involved segment of chromosome 12 appeared to be proximal in band ql3 was determined by FISH to have a deletion starting approximately 10 kb upstream of HMGI-C and ext~n~ling up to about 100 kb 5' of exon 1 of HMGI-C (Figure not shown).
DISCUSSION

HMGI Proteins in Adipo~.-csis and M~ r~ -e Diffe~ lialion In this study, chimeric ll~nsc-;pts were i~ ti~led from two lipoma which resulted from fusion of the 5' end of the HMGI-C gene to novel sequences derived from different chromosomes. Three DNA binding domains corlt~ining the A-T hook motifs of HMGI-C are linked in these transcripts to seq~lPr~es encodingpotential transcriptional regulatory domains. In the case of lipoma ST90-375, the 3 o novel domain is highly acidic and rich in serine and threonine residues resembling the typical activation domains found in transcription factors. In lipoma ST93-724, the novel protein contains two LIM domains, motifs that promote protein-protein interactions.

SUB~ 1 l 1 U 1~; S~;~; l (RULE 26) CA 022~963~ 1999-01-06 HMGI-C, Chimeric Transcripts and Lipomas The chromosomal region 12ql4-15 is hypoth~si7e(1 to contain an important gene involved in lipomas because it is the most commonly rearranged 5 site (Mandahl et al., 1988). Our study establishes that HMGI-C is the gene disrupted in lipomas with chromosome 12 rearran~nPntC. The large intron (greaterthan 25kb) between exons 3 and 4 (li.ctinr-tly sepalaLes the DNA-binding from the acidic domains of HMGI-C. This provides a subst~nti~l target for translocations so that the three A-T hook motifs remain intact and confer the DNA-binding o specificity of HMGI-C to the fusion ploleins.

HMGI-C is a 109 amino acid residue protein (Patel et al., 1994) that consists of three DNA-binding domains (A-T hooks) linked to the carboxy-terminalacidic domain which does not activate transcription (Thanos and Maniatis, 1992;
5 X.Z. and K.C., unpublished data). The two lipoma translocations result in a novel protein con~ining A-T hooks of HMGI-C at the amino-terrnin--c fused to transcriptional regulatory dom~in~ at the carboxy end. The other reported example for an A-T hook cont~ining gene implicated in tumorigenesis is MLL (Tkachuk et al., 1992; Gu et al., 1992). However, the presence of a putative second DNA-20 binding domain (Ma et al., 1993) derived from the MLL gene and retained in thefusion protein obscures the exact contribution of the A-T hooks to tumor pathogenesis (Rabbitts, 1994). In these lipomas, the only known HMGI-C
functional domains retained in the fusion proteins are the A-T hooks. These motifs would therefore be responsible for DNA binding specificity of the fusion proteins.
25 Although it is possible that simple truncation of HMGI-C is sufficient to cause lipomas, a number of studies have ~lete~ i"~ that both domains of fusion proteins are n~ces~ry for transforming activity (de Thé et al., 1991; Kamps et al., 1991;Pendergast et al., 1991; May et al., 1993). Therefore, as proposed for other fusion proteins, the heterologous sequen~e in the lipoma fusion proteills would alter 3 o the biological activity of wildtype HMGI-C and lead to deregulation of do~ ea target genes.

The above model readily explains how the fusion protein produced in lipoma ST90-375 may function. The novel sequence from chromosome 15 3 5 encodes for an acidic peptide rich in serine and threonine residues. These features have been observed in a number of Llansclip~ional activation domains (Mitchell and SUB~ I l l u ~ ~;~1 (RULE 26) .

CA 022.,963., 1 999 - 0 1 - 06 PCT/~JS97121299 Tijan, 1989) inrl~rling the carboxy-terminal domains of homeobox proteins (Hatano et al., 1991) and NF-kB (Schmitz and R~e~erle, 1991). So, the acquisition of a transactivation domain by the DNA-binding ~om~in~ of HMGI-C, which normally possesses a transcriptionally inactive acidic domain, can easily be reconciled with aberrant regulation of the HMGI-C target genes. In the case of the t(3;12) in ST93-724, the fusion protein must operate by a dirrclelll meçh~ni~m to deregulate the HMGl-C target genes. The novel seql)~nre from chromosome 3 encodes for two t~n~lPn-ly arranged LIM motifs. The LIM domain is conserved amongst highly diverged species and LIM proteins have been shown to have important developmental funrtio~ which include pattPrning (Cohen et al., 1992), cell fate decision (Freyd et al., 1990) and dirr~.c.,lialion (Way and Chalfie, 1988).
LIM domain proteins are capable of protein-protein interactions (Sadler et al., 1992) through dimerization m.orli~ted by the LIM domains (Feuerstein et al., 1994). Therefore, LIM-LIM interactions bciwccn the ST93-724 fusion product and other nuclear prolcills could recruit potential L.dilsc.iL,Lional regulators to DNA
sequences with a specificity ~irt~t~d by the HMGI-C A-T hook motifs.
Deregulation of HMGI-C target genes would then contribute to lipoma development. It is h~cle;,ling to note that the majority of nuclear ~lOlCillS capable of interacting with LIM domains are known to function as transcription factors.
These include several LIM-homeodomain ~loteins (S~n~h~7-GarcIa et al., 1993 and references within) as well as basic helix-loop-helix proteins shPan-1 (Gennan etal., 1992) and TAL1 (Valge-Archer et al., 1994). While ovclc~ ession of LIM
proleins has been implicated in T-cell lymphom~ (reviewed by Sanrh~z-Garcia and Rabbits, 1993), this is the first example of a LIM domain occurring in a fusion 2 5 product.

A great heterogeneity in chromosomal palln(,LS translocated with 12ql4-15 is found in karyotypically abnormal lipomas in~ir~tin~ that a large number of sequences in the genome can be fused to HMGI-C. The present data demonstrate that novel sequences linked to HMGI-C in two lipomas encode for distinct domains. This suggests that a number of ~ltern~tive domains can be placed dow"~ of the HMGI-C A T hooks and contribute to the pathobiology of lipomas. Il~ ingly, both novel sequences described in this study encode for transcriptional regulatory domains. Therefore, the choice of novel sequences in chimeric lla"sc,i~l~, in lipomas is p,csu~-,ably not arbitrary but does require the presence of regulatory transcriptional dc~ in~ ~tt~rll~d to the HMGI-C A-T hooks.

SUB~ 111 U 1~ ; l (RULE 26) ~ . . .. . . .. . .

CA 022~963~ 1999-01-06 ~ ' WO 98/50536 PCT/I~S97/21299 A similar situation has been observed in the llq23 acute lel-k~mi~ where the MLLgene is translocated with multiple chromosomal partners which mostly encode different types of Llal~c.i~tional regulatory domains (Prasad et al., 1994). This could be a general m~rh~ni~m for tumors where nonrandom rearrangenlellls of a 5 specific chromosomal region involve a variety of partners. Chimeric transcription factors that promote tumorigenesis would be produced by juxtaposing DNA
binding domain(s) contributed by the collsi~L~lltly rearranged locus to distinct ~ypes of transcriptional regulatory domains.

0 HMGI-C, Pygmy and Adipogenesis The above studies demonstrate that an altered HMGI-C protein is involved in the abnormal growth and development of fat cells reslllting in lipomas.
This leads to the possibility that HMGI-C may normally play a role in adipogenesis 5 and analysis of the pygmy mouse strongly sukst~nti~tPs this hypothesis. The mouse mutant, pygmy, was found to be a null mutation of HMGI-C due to deletions within the gene (unpublished results). The obvious phenotypic characteristics ofthe pygmy mouse are its small stature with most tissues reduced cornmP...c...at~with the overall decrease in weight of the mouse (40~ of wildtype). Il~ c~lhlgly, 20 one tissue disproportionately reduced in weight is body fat. The fat index, areliable in~lirator of total fat content relative to body weight (Rogers and Webb, 1980), is approximately eight times lower in pygmy than in their wild-type itt~rm~tlos (Benson and Chada, 1994). The function of HMGI-C in adipogenesis could be related to its role in cells undergoing dirrerelltialion. It is expressed in 25 less difr~rellLiated cells but no det~ct~hle levels are observed in their terminally dirr~ ted coullL~,~a,Ls (Vartainen et al., 1988; Giancotti et al., 1987).
Therefore, lack of HMGI-C expression, as found in the pygmy mouse, could affect the differentiation of preadipocytes into mature adipocytes, cells capable of lipid storage. This developmPnt~l abnormality would lead to a decrease in fat deposition 30 and the phenotype observed in the pygmy mouse. The role of HMGI-C in adipogenesis and metabolic disorders such as obesity is thus of considerable interest.

SUB~i 1 l 1 U l ~ (R~JLE 26) . .

CA 022~963~ 1999-01-06 HMGI-C Inactivation Results in Reversion of the Obese Phenotype The above observations suggested that HMGI genes play a pivotal role in the growth and development of fat tissue and that they may be involved in 5 obesity as well. To address this latter question, expression of HMGI genes in fat tissue obtained from normal and morbidly obese animals was determin~d Signific~ntly, while HMGI-C expression was absent from fat tissue of nbrmal mice, .~ignifir~nt expression was readily ~letect~ble in l~NA isolated from fat of obese and diabetic mutant mice, two widely accepted models of obesity (Figure 0 12). A similar result was obtained with HMGI(Y) gene, whose expression in obesity was dr~m~tic~lly elevated (Figure 12).

The above experiment demonstrated that expression of HMGI
proteins in obesity is increased 10 to 100 fold and that HMGI inactivation could be 5 used to regulate the amount of adipose tissue in vivo. Therefore, an attempt was undertaken to regulate obesity by inhibiting the biological activity of the HMGIproteins. The term "HMGI genes" or "HMGI proteins", as used herein refers tO
both HMGI-C or HMGI(Y) genes or ~loLehls, rc;s~e~ ely.

2 o A classical mouse mutant called obese was selected for this experiment. The obese phenotype has previously been well characterized and its most prominent characteristic is a pathological weight gain due to an excessive food intake (Green, 1989). Mice homozygous for the obese mutation acc~lm~ t~?
signifi~nt amounts of adipose tissue and reach the weight of 80 grams as opposedto 25 grams in normal animals. Recently, the gene responsible for the obese phenotype has been cloned and was found to produce a hormone secreted by the fatcells (Zhang et al., 1994). Leptin, as this protein was called, is thought to beinvolved in the reg~llation of appetite, and its absence in the obese mutant leads to overeating and obesity (Rink, 1994).
To revert the obese phenotype, HMGI-C gene was inactivated by homologous recombination (Zhou et al., 1995) and the resulting mice were bred with the obese mut~nt~. During the experiment, male and female mice were m~int~in~ under altern~ting 12-h light and dark periods and provided water and food ad libitum. Since both ob/ob and pg/pg animals are sterile (Green, 1989), crosses were carried out in two stages. First, a pg/+Xobl+ intercross was SUBSTITUTE S}~;~; l (RULE 26) CA 022~963~ l999-0l-06 ~ ' WO 98/50536 PCT/US97/21299 undertaken and the progeny from this cross were genotyped using Southern blotting and PCR amplification. To screen for obese mutation, DNA was isolated from the mouse tails by standard methods (Sambrook et al., 1989) and PCR amplified using sense primer S'-CATTCTGAGTTTGTCCAAGATGC-3' and antisense primer 5'-GGTCTGAGGCAGGGAGCAGC-3'. PCR conditions were as follows:
denaturation at 95~C for 2 mimltes and 30 cycles of amplification at 94~C for 30seconds, 58~C for 30 seconds and 72~C for 30 seconds, followed by a final extension for 10 mim~tes at 72~C. The res~lting PCR products were digested with DdeI and electrophoresed on 8% polyacrylamide gel. Under these conditions, 0 amplification of the wildtype allele yields 150 bp products which contains no DdeI
restriction sites. The ob mutation substitutes T for C in position 369,gen~"~atil.g a novel DdeI site. Therefore, digestion of the PCR product from mutant allele generates unique products of 106 and 44 bp. Genotyping of the HMGI-C knockout mice was carried out as described previously (Zhou at al., 1995). The double 15 heterozygous ~nim~l~ (pg/+ ob/+) thus identified were intercrossed again and the double homozygotes (ob/ob pg/pg) obtained from this second cross were further analyzed.

Surprisingly, inactivation of HMGI-C produced a complete reversal 20 of obesity in the leptin-deficient mice (Figure 13). In the absence of HMGI-C, pg/pg ob/ob mice did not develop an excess of adipose tissue and their body weight stayed at the normal level of 25 grams as opposed to 80 grams in ~/+ ob/ob anim~l~. A similarly dramatic effect was observed in mice which were homozygous for ob mutation but heterozygous for HMGI-C inactivation (pg/+
2 s ob/ob). In these animals the amount of fat tissue was signific~ntly reduced and the body weight decreased from 80 to 65 grams, even though these animals preserved one of the two HMGI-C alleles intact and e~lessed 50% of the normal HMGI-C
levels. This result specifically proved that inhibition of HMGI biological activity can be used to regulate growth and development of adipose tissue in m~mm~l~
3 o since less than a 100% HMGI inhibition results in a reduction in the amount of fat tissue.

Just as importantly, this reduction in weight was accompani-~ by a decreased food intake in the previously obese animals (Figure 14). Inhibition of35 HMGI-C biological activity resulted in a decrease of food intake from 7.25 grams in +/+ ob/ob ~nim~ to 3.75 grams in pg/pg ob/ob mice (Figure 14). Therefore, SUB~i 1 l 1 U 1~ ; l (RULE 26) ., CA 022~963~ 1999-01-06 the effects of HMGI inhibition are not limited to growth and dirr~"ellLiation ofadipose tissue but also result in an almost two-fold decrease in daily food intake.

It is important to consider the effects of HMGI inhibition in another 5 mouse mutant called "diabetes" (db). This model of human obesity and diabetes,characterized by excessive food intake, increased body weight and elevation of blood sugar, results from an inactivating mutation in leptin receptor (Chen at al., 1996). Therefore, our ability to prevent the detrimental effects of leptin deficiency in obese mice via inactivation of HMGI genes in~lir~t~ that inhibition of HMGI
0 biological activity will be beneficial in various disturbances of leptin molecular pathway, e.g., mutations of leptin receptor and/or leptin resi~t~nre. Signific~ntly, resistance to normal or elevated levels of leptin may be an important factor in human obesity (Tartaglia et al., 1995).

In combination, these results conclusively demonstrate the role of HMGI genes in obesity and provide proof-of-principle evidence that inhibition ofthe HMGI biological activity can be used to control the growth and development of adipose tissue such as occurs in obesity. Ir~hibition of the HMGI biological activity can also be used to regulate the amount of carcass fat in farm anirnals if, for 2 o example, an animal lacking adipose tissue is desired.

The term "biological activity", as used herein, means the ability of HMGI proteins to regulate and promote growth and development of adipose tissue or the ability of the HMGI proteins to form transcriptional regulatory complexesand regulate transcription of other genes. Such inhibition is effected using theconventional means known in the art as described in greater detail in the following non-limhing examples.

Relevance of HMGI Family in Tumors with Rearr~n~emPntc of 12ql3-15 or 6p21 Of major importance is the frequent observation of chromosomal rearr~n~mPnt~ in bands 12ql3-15 in a large group of benign solid tumors. Most ~ro-,-i,u.,lly, these include uterine leiomyomata (Nilbert and Heim, 1990; Rein et al., 1991), and pleomorphic adenomas of the salivary gland (Sandros et al., 1990;
3 5 Bullerdiek et al., 1993). Rearr~n~em~ntc of 12ql3-15 have also been reported in pulmonary chondroid hamartomas (Dal Cin et al., 1993; Fletcher et al., 1995), SUB~ 11 l U l ~; S~;~; 1 (RULE 26) CA 022~963~ 1999-01-06 endometrial polyps (Vanni et al., 1993), epithelial breast tumors (Rohen et al.,1993), hemangiopericytoma (Mandahl et al., 1993a), chondromatous tumors (Mandahl et al., 1989, 1993b; Bridge et al., 1992; Hirabayashi et al., 1992), diffuse astrocytomas (Jenkins et al., 1989), parosteal lipoma (Bridge et al., 1995), 5 and a giant-cell turnor of the bone (Noguera et al., 1989) Many of these tumortypes are of mesenchymal origin and it has therefore been hypoth~ci7~d that a single gene associated with growth and mesenchyme may be responsible for these multiple neoplasms (Schoenberg Fe3zo et al., 1995). Several lines of evidence implicate HMGI-C as a strong c~n~ te for such a gene at 12ql4-15. First, 0 physical mapping studies have shown chromosomal breakpoints for three of thesebenign tumors (lipoma, pl~lmon~ry chondroid l~ a,lo~lla and uterine leiomyoma) to map within a single YAC (Schoenberg Fejzo et al., 1995). This study assigns HMGI-C to the translocation breakpoint in lipomas, and chromosomal breakpoints in five analyzed uterine leiomyomata as well as a pl-lm~n~ry chondroid hamartoma5 have been found to reside within 10-lOOkb of exon 1 of HMGI-C (unpublished results). Second, the role of HMGI-C in growth control is apparent because its disruption in the pygmy mouse leads to aberrant growth and development. Also, ithas been shown in vitro that HMGI-C is required for transfo~nation (Berlingieri et al., 1995). Finally, prelimin~ry studies reveal that e~pre~sion of HMGI-C during2 o mouse embryogenesis is restricted mainly to the m~senrhymal component of tissues and organs (unpublished results). Taken together, these data in-lic~te that HMGI-C
is highly likely to be the gene disturbed by 12ql4-15 rearrangements in a numberof tumors of m~senrhymal origin.

2 5 Nonrandom involvement of 6p21-23 has also been observed in lipomas (Sre~ nt~i~h et al., 1991), pulmonary chondroid hamartomas (Fletcher et al., 1992, 1995) and uterine leiomyomata (Nilbert et al., 1990~ e~ gly, HMGI(Y), the other member of the HMGI protein family with a similar structure as HMGI-C that includes the three DNA-binding domains, has been localized to 3 o 6p21 (Frie(lm~nn et al., 1993) This raises an intriguing possibility that HMGI(Y), a molecule closely related to, but distinct from HMGI-C, could also be associated with benign tumors of m~senrhymal origin.

In s~ y, a disruption of the HMGI-C gene resulting in chimeric ~.~nsc,i~l~ is a characteristic feature of lipomas. As adipocytes play a key role in lipid homrost~ and m~i"~ re of energy balance in vertebrates, an SUBS 1 l l U 1~; S~ET (RULE 26~

CA 022~963~ 1999-01-06 underst~n~lin~ of HMGI-C function in adipogenesis may lead to insights into obesity and other metabolic disorders. In addition, the obvious role of HMGI-C in normal growth demonstrated by the phenotype of the pygmy mouse and its loc~li7.~tiQn at or adjacent to the translocation breakpoints in lipoma, uterine5 leiomyoma and pulmonary chondroid hamartoma suggests its fim-l~mPn involvement in a variety of benign tumors.

HMGI Proteins in M~mm~ n Growth and Development 0 The current study demonstrates that the absence of HMGI-C causes growth retardation in pygmy mice. Although the precise molecular mrch~ni~m remains to be elucidated, the function of HMGI prol~ills in cell proliferation could be regulated during the cell cycle through alteration of their DNA binding ability via phosphorylation by the cell cycle-dependent p34cdc2 kinase (Reeves, R. et al., 1991). Inactivation of the HMGI-C gene would perturb the cell cycle in the developing embryo and the resulting disruption of growth would produce the pygmy phenotype. The identific~tion of the pygmy gene as HMGI-C provides novel insights into the control of m~mm~ n growth and development and a molecular clue.to investigate the biochemical nature of the African pygmy phenotype (Sinha, Y. et al., 1979) and a mllltit~l~le of growth hormone-resistant human dwarf syndromes (Benson, K. & Chada, K., 1994).

Mis~l..e~ion of Disrupted HMGI Proteins in ~lm~r~ Tumors HMGI(Y) and HMGI-C, two homologous but distinct members of the HMGI family of archh~cll~ral factors, have now been shown to be disrupted inidentical tumors. Rearrangements of HMGI-C, first reported in lipomas, were later described in other mesenchymally derived neoplasms with translocations of 12ql3-15. Similar to HMGI-C, disruptions of HMGI(Y) will presumably be also 3 o responsible for uterine leiomyoma, pl~lmon~ry hamartoma, pleomorphic adenomas of salivary gland and other mesPnrhymal tumors with recurrent aberrations at 6p21-23.

SUB~ 111 U 1~ (RULE 26) .

CA 022~963~ 1999-01-06 Rearrangements within HMGI Genes are Required for I,ipoma Development In combination with previous studies on HMGI(C) and HMGI-Y, it is now possible to glean novel insight into the molecular m~C~l~ni~m of tumor 5 formation in lipomas and, by extrapolation, in related solid rnes~rlrhymal neoplasms. HMGI-C does not behave as a classical transforming oncogene since overexpression of full-length HMGI-C cDNA does not result in tumorigenesis; On the other hand, in all twelve analyzed lipomas, chromosomal rearrangements have produced disruptions in translocated HMGI alleles. While expression of an HMGI
0 gene is n~ces.s~ry for tumorigenesis, activation of an intact HMGI allele in amesenchymal cell will not be suf~ nt to produce a tumor. Therefore, disruptions within HMGI genes and the aberrant structure of the res--lting cDNA are requiredfor lipoma development.

A variety of the HMGI chimeric transcripts can be found in lipomas.
The comparison of these aberrant cDNAs demol~Lr~tes that rearrangements can range from a simple internal deletion to protein truncation to juxtaposition of ll~ns~ lional regulatory dom~in~ to HMGI DNA-binding ~om~in~. An aberration common to these twelve lipomas is a deletion of or within highly conserved and Imllsu~lly large and 3'UTR of an HMGI gene. The best example is lipoma ST88-08203, where the aberrant Lla~ t codes for the wild-type HMGI(Y) and the deletion is limited to its 3' UTR. Since translocations of 12ql3-15 which disrupt 3'UTR of HMGI-C while preserving its ORF are also observed in leiomyoma and pleiomorphic adenoma of salivary gland, 3'UTRs of HMGI genes may contain z5 important regulatory sequenres that function in growth regulation and/or tumor suppression.

Notably, the aberrant L~allsc~il)t~ isolated from lipomas with rearrangements of 6p21-23 were generated by internal deletions within the 3 o translocated HMGI(Y) allele. This observation suggests that in lipomas and related benign mPse~-~hymal tumors, HMGI genes may contain internal deletions and other submicroscopic rearrangements un~etect~hle by cytogenetic techniques. It is likely therefore that the contribution of the HMGI genes to tumorigenesis is more significant than predicted by karyotypic analysis.

SUBSTITUTE SHEET (RULE 26) CA 022~963~ l999-0l-06 WO 98/50536 PCT/US9712l299 Misexpression of HMGI genes in a Dir~.e,lLiated Cell Results in Tumorigenesis To understand the biological function of the HMGI proteins, it is important to analyze their expression profiles during both norrnal and pathological 5 growth. Promin~ntly, high levels of the HMGI expression are observed during mouse embryonic development in midgestation but it ess~nti~lly dissal,peal~ closer to the end of prt;gnallcy. Subsequently, no HMGI expression can be detected in any of the adult tissues. Lipomas are composed of mature adipocytes which, like other terminally dirrel~l.liated cells, normally do not express HMGI proteins.
0 However, transcriptionally active HMGI alleles are consistently found in solid mlosçnrhymal tumors with rearrang~lllellls of 12ql3-15 and 6p21-23.
Rearrangements of 12ql3-15 or 6p21-23 activate an HMGI allele normally silent inadult cells and the res..lting misexpression of the HMGI protein in the context of a dirl~lcllliated mesenchymal cell is a crucial step in tumor development. A notable 5 feature of this mecll~ni~m stems from the observation that during mouse embryonigenesis, HMGI-C is expressed in the mPsenrhymal component of the developing organs and tissues (unpublished data). Tumorigenesis in this case results from the temporally h,applu~r;ate ek~lcssion in an adult cell of a gene that is normally expressed during prenatal development in an embryonic cell of the 2 o same lineage. This is remini~ce~t of observations in B-cell leuk~mi~c where rearrangements of 8q24 chromosomal area activate c-myc expression in a precursorcell of B-lineage and result in neoplasia. Unlike the HMGI family members, however, the endogenous expression of c-myc is not restricted to embryogenesis and its hlapplop..ate expression takes place at the same time in the life of the2 5 organism when it is normally expressed. Even more different is a situation in some of the T-cell acute Iymphoid leuk~mi~c where the cause of neoplasia is ectopic expression in T-cell precursors of HOX11, normally expressed in the embryonic liver.

3 o Distinct Molecular Pathways of Tumorigenesis Exist in Lipomas The molecular analysis of the lipomas described above yields valuable information about the ~AI)les~ion state of the non-le~ llged HMGI
alleles. Wildtype HMGI ~A~ ,s~ion, normally associated with tumorigenesis, was 3 5 readily detectable in lipomas ST88-08203 and ST91-198, where chromosomal rearrangements produced an apparently normal HMGI(Y) and a truncated HMGI-C

SUB~ U 1~; SHEET (RULE 26) CA 022~963~ 1999-01-06 proteins, respectively. In contrast, the non-rearranged HMGI allele was not expressed in tumors ST92-24269 and ST93-724, where the aberrant HMGI
transc~ were predicted to encode fusion prot~hls consisting of the HMGI DNA-binding domains fused to putative transcriptional regulatory domains.

The above fintiing~ indicate that there are at least two distinct molecular pathways by which tumorigenesis in lipomas can proceed.- When a chromosomal rearrangement produces a disrupted HMGI protein with no intrinsic transcriptional activity, tumor development is dependent upon subsequent activation 0 of the non-rearranged allele. However, the requirement for wildtype HMGI
expression can be circumvented when, as a result of a translocation, a transcriptional regulatory domain is juxtaposed to the HMGI AT-hooks. The unlikely alternative mech~ni~m, in which the non-leallailged allele is activated by the fusion protein through a positive HMGI regulatory m~ch~ni~m, would postulate5 that such autoregulatory function is inhibited in the presence of ~la~sc~i~lional regulatory domains. Therefore, we conclude that distinct rearrangements of a single gene can activate alternative molecular palllways of tumor pathogenesis.

Molecular analysis of HMGI rearrangements in multiple tumor 20 samples can now be combined with the ~Ayr~sion studies of both disrupted and non-rearranged alleles to produce a m~rh~ni~tir~lly coherent model of lipoma development (Figure not shown). Tumor development is initi~ted when the chromosomal rearrangement disrupt an HMGI allele and results in the HMGI
misexpression in a dirr~,~ellliated mesen-~.hymal cell. Deletion within 3' UTR is 25 probably the minim~l rearrangement nPcess~ry for tumor formation Subsequently, one of the alternative tumorigenic pathways is selected based on the precise nature of the HMGl disruption. In the simplest model, the requirement for HMGI
expression in tumorigenesis could be circumvented if HMGI DNA-binding domains are juxtaposed with a transcriptional regulatory domain (Pigure not shown). The 30 reduced number of events involved in tumor form~tion would readily explain the most frequently observed translocation in lipomas, t(3;12)(q29;qlS), since it fuses DNA-binding domains of HMGI-C with LIM domains, motifs that are thought to function in lldnscli~tional regulation.

SUB~ 1 l 1 U 1~ ; l (RULE 26) CA 022~963~ 1999-01-06 The HMGI Proteins Play Dirr.,leilt Roles in Tumors of Epithelial and M~senrllymal Origin Benign tumors, unlike their m~lign~nt cou~ ,al L~, are 5 characterized by a limited number of highly specific genetic alterations involving only a few chromosomal regions. It was proposed therefore that the molecular analysis of these neoplasms would identify genes of major importance for growth and proliferation. The above studies with HMGI(Y) and HMGI-C in lipomas demonstrate that misexpression of HMGI proteins plays a significant role in the 10 development of a diverse array of human solid tumors. Clinically, a prominentfeature of these benign mesenchymal tumors is the extremely low rate at which they convert to m~lign~nry. Indeed, uterine leiomyomas progress to become leiomyosarcomas in less than 0.01% of the cases while conversion of lipoma to lipo~a.collla is even less frequent. Therefore, mise~lcssion of HMGI proteins, 5 while acting to increase the growth rate of the mesenchymal cells, does not seem to predispose the overproliferating cell to m~ n~nt transformation and may even play a protective role.

The apparent inability of the HMGI-expressing benign m~senrhymal 20 tumors to undergo m~lign~nt conversion is in a stark contrast with the situation seen in the tumors of epithelial origin. In these latter neoplasms, cellular hyperproliferation provides starting population for clonal expansion which, in turn, is followed by a stepwise progression to m~lign~nry. Even more intriguingly, epithelial cells cannot be transformed by overe~plcssion of HMGI-C while 25 chromosomal rearrangements which could disrupt HMGI-C and HMGI(Y) are not found in tumors of the epithelial origin. Finally, in epithelial tumors activation of HMGI expression is associated with the advanced stages of carcinogenesis rather than with early hyperplasia. The asynchrony between the expression patterns of HMGI proteins in epithelial and mesenchymal cells as well as distinct phenotypes3 o of the relevant tumors indicate that in tissues of dirrt~llL embryonic lineage HMGI
plolei-ls l~lrOl~ imil~r functions.

One possible explanation for this ph~nt~m~rlon is provided by the fact that HMGI proteins normally function in the developing mesenchyma. The 3s role of HMGI proteins in m~senrhymal tumorigenesis may therefore be closely related to that during normal development, such as growth rate regulation. In the SUB~ 111 U 1~ ; l (RULE 26) CA 022~963~ l999-0l-06 epithelial tumors, the HMGI architectllral factors, expressed outside of their normal cellular milieu, may be recruited to take part in the transcriptional regulation of genes that are involved speci~r~lly in the ~mal stages of tumor progression, such as invasion and metastasis. Regardless of the molecular details, the ability of HMGI-5 C and HMGI(Y) to execute distinct functions during tumorigenesis in diverse celltypes provides a powerful testim- ny to the biological potency of the HMGI
proteins and accounts for the dramatic con~equenre~ of their disn~ption.

Tr~n~'~c~tinn Breakpoints UIJ~lre~ll of the HMGI-C Gene in Uterine 1 o LL;~.Ilyol~aLa Translocation breakpoints in uterine leiomyomata reported here are in stark contrast to those observed in lipomas and other benign mesenchymal tumors in which translocations are found within the coding region of HMGI-C.
Unlike the f1n~ingc in uterine leiomyomata rearrangements in lipomas consistently result in disruption of HMGI-C, whereby DNA-binding A-T hook domains are separated from the 3' region of the gene. Recallce HMGI-C has no transcriptionalactivation domain (unpublished data), the pathobiology of lipomas appears to result from juxtaposition of direct or indirect activation ~lom~inc with the DNA-binding A-T hook ~r)m~in.c, although an ~ltern~tive explanation of truncation of the protein cannot be ruled out at present.

These studies of uterine leiomyomata suggest a completely different molecular mlocll~ni.cm because the entire gene appears to be retained, suggesting that both the 5' DNA-binding domain and the 3 ' domain of unknown function are n~CeS,c~ry. The finding that chromosomal rearrangements were located 10 to > 100kb upstream of HMGI-C in seven uterine leiomyomata suggestc that breakpoints might disrupt regulatory elem~ontc and alter the normal expression of HMGI-C, analogous to Burkitt lymphoma, where translocations up to 100 kb upstream of 3 0 MYC result in aberrant e~ ession and neoplasia.

This "regulatory hypothesis" is supported by cytogenetic and FISH
results for the karyotypically variant uterine leiomyoma ST89-171. In this tumor, three copies of HMGIC were present, suggesting a dosage meeh~ni.cm for altered 3s expression levels. Additionally, loss of the der(12) chromosome in ST89-171 SUBSTITUTE SHEET (RULE 26) CA 022~963~ 1999-01-06 provides further evidence that the der(14) chromosome, to which HMGI-C maps, contains the critical sequence.

This observation of an i-ller~ ial deletion upstream of HMGI-C in 5 one uterine leiomyoma with a variant rearrangement of chromosome 12 is important for the cytogenetic and molecular intel~retation of rearrangements in uterine leiomyomata and other tumors. This finding implies that uterine leiomyomata with unusual cytogenetic rearrangements of chromosome 12, and possibly other mesenchymal neoplasms without microscopically ~etect~ble 0 chromosome 12 rearrangements, may have submicroscopic rearrangements of a critical region upstream of HMGI-C. Characterization of HMGI-C expression in uterine leiomyomata of all cytogenetic subgroups is now warranted for a more complete underst~n~ing of the pathobiologic rmPch~ni~m.

Furthermore, this hlle.L,lcl~tion of a mPcll~ni.cm for dysregulation of HMGI-C in uterine leiomyomata is s~lbst~nti~tPd by observation of a rearrangement in a fibroid involving chromosomes 8 and 12 in which the 3' UTR of HMGI-C is di~lup~ed. Such a rearrangement results similarly in retention of the entire coding region of HMGI-C, a finding previously noted in variant translocations in Burkitt 2 o lymphoma. However, this translocation breakpoint mapping in uterine leiomyomata and the deregulation model differ largely from that reported by others in which intragenic breakpoints were found for some fibroids perhaps reflecting the relatively limited number of tumors analyzed. Alternatively, although there are no data to support the existence of alternative 5' exons of HMGI-C or other uncharacteristic genes in the region, such possibilities, which might be affected by chromosomal rearrangPment and contribute to tumor biology, cannot be excluded.
Regardless, a mPch~ni.~m of dysregulation not involving a filsion transcript must be considered for tumors without intragenic rearrangements of HMGI-C because irrefutable data implicate HMGI-C as the critical gene in benign mes~pnrhyma 3 o tuInors with rearr~n~mPnts of 12ql4-15 .

These fin-ling~ are con.~ tent with acc~m~ ting evidence for a primary role of HMGI-C in normal growth and dirr~ tiation of a variety of tissues. Besides expression of fusion tl~nscli~t~ in lipomas and other benign 3 5 mesenchymal tumors and in tnesen~ymal collll)ol~ellls of tissues in the developing mouse embryo, expression of HMGI-C is found only in cells after they become SUB~ U l ~; S~;~; 1 (RULE 26) CA 02259635 l999-0l-06 transformed and has been found to be n~cess~ry, but not sufficient7 for ru~ tiQn These studies indicate that HMGI-C also may be deregulated through translocation in uterine leiomyomata without involvement of a fusion transcript.

The present invention is further illustrated by the following examples which are not intent~ed to limit the effective scope of the claims. All parts and percentages in the examples and throughout the specification and claims are by weight of the fimal composition unless otherwise sE~ecifiPd.

Examples HMGI Proteins in A~ o~ and M~i..rl~yl.,e Differçn~i~tinn The GenBank ~Cces~ion numbers for the novel sequences in the chimeric transcripts from ST90-375 and ST93-724 are U28131 and U28132, respectively.

Isolation of YACs at the Human Pygmy Locus Initially, conserved fr~gmPnt~ were isolated from the cloned, mouse pygmy locus (Xiang et al., 1990; K. Benson and K.C., unpublished observations) and were used as probes on a normal, human lambda genomic library (Sambrook et al., 1989). The cross-hybridizing clones were isolated and relevant homologous 2 5 fra~m~ntc were subcloned and sequenced. Specific oligonucleotide L~ e~s (sequen~e 5 ' -AGGGGACAACAAATGCCCACAGG and 5 '-CGTCACCAGGGACAGTTTCACTTGG) were synthP~i7ecl and used to screen a human total genomic YAC library by the PCR-based method (Green and Olson, 1990). Four positive clones of Saccharomyces cerevisiae cnnt~ining YACs 3 o yWPR383, yWPR384, yWPR385 and yWPR386 were isolated.

Construction and Screening of Phage Libraries High molecular weight DNA was isolated from yeast strains harboring YACs yWPR383 and yWPR384 (Guthrie and Fink, 1991), and partially digested with Sau3A. After partial fill-in of the Sau3A site, DNA was subcloned SUB~ 111 U l ~ (RULE 26) _ CA 022~963~ 1999-01-06 at the partially filled XhoI site of the predigested lambda FIXII vector (Stratagene, La Jolla, CA) and packaged in vitro (GIGAPACK II p~cl~gin~ extract, Stratagene). To select clones derived from the human YACs, 6000 plaques from each library were probed with total human genomic DNA and hybridizing plaques were spotted on plates inoculated with SRB(P2) cells in a gridded array. After incubating the plates at 39~C for 12 hours, plaques were transferred onto DURALON (Stratagene) membranes. These grids were used for identifying lambda clones that cont~in~d human HMGI-C exons by probing with mouse HMGI-C cDNA (unpublished results), using the same hybridization conditions as 0 detailed below for Southern analysis. Overlaps between contiguous clones and colinearity with the genome were confirm~cl by a combination of clone to clone and clone to genomic hybridizations along with restriction mapping.

Southern Blot Analysis 10-12 mg of human DNA was digested with the ~plopliate restriction enzymes, products resolved on 0.8% agarose gels and transferred ontoDURALON (Stratagene) membranes. Blots were treated with prehybridization solution (50% fonn~mi~lP, Sx SSC, 10x Denhardt's solution, 0.05M sodium phosphate pH 6.8, 0.001M EDTA, 0.01 mg/ml denatured salmon sperm DNA, and 0.2% SDS) for 2 hours at 42~C. Probes were added to the hybridization solution (50% form~mi-le, 5x SSC, lx Denhardt's solution, 0.02M sodium phosphate pH 6.8, 0.001M EDTA, 0.01 mg/ml dellalul~d salmon sperm DNA, 0.2% SDS and 10% dextran sulfate) and hybridization was performed for 16 hours at 42~C. Membranes were washed with 2x SSC, 0.001M EDTA, 0.5% SDS, 0.05% NaPPi and 0.01M sodium phosphate pH 6.8, at 65~C for 3 x 1 hour periods and exposed to X-ray film at -70~C with i~ sirying screens.

I~entifir~tiQn and Characterization of Chimeric Transcripts First strand cDNA was synth~si7~d in a 20 ml reaction using an anchored oligo-dT primer S'-GCAATACGACTCACTATAG(T)13 and Superscript II RT reverse llallscliylase (BRL, Gaithersburg, MD) according to the m~nllfactl~rer's protocol. Primers used in the first round of 3' RACE (Ausubel et 3 5 al., 1989) were an HMGI-C exon 1 sense primer 5'-CTTCAGCCCAGGGACAACC and an ~ en~e adapter primer 5'-SUBSi 1 l 1 U 1~ ; l (RULE 26) CA 022~963~ 1999-01-06 GCAATACGACTCACTATAG. One ml of first-strand cDNA was combined with 25 pmole of sense primer in a 50 ml reaction mixture (60 mM Tris-SO4 (pH 9.1 at 25~C); 18 mM (NH4)2S04; 2 mM MgSO4; each dNTP at 200 mM; 2.5 U of Taq DNA polymerase (BRL)), denatured for 2 minllt~~ at 94~C and subjected to 5 cycles of linear amplification (Rother, 1992) using the following conditions: 94~C, 30 seconds; 58~C, 20 seconds; 72~C1 1 minute 30 seconds. Ten pmole of antisense primer were then added and 25 cycles of exponential amplification wereperformed (94~C, 30 seconds; 56~C, 30 seconds; 72~C, 1 minute 30 seconds).
One ml of the PCR reaction was reamplified for 20 cycles with a nested HMGI-C
o sense primer spanning exon 1 and 2, 5'-GGAAGCAGCAGCAAGAACC as described above. Five ml of each reaction were analyzed on a 1.5% agarose gel.
Reverse l.anscIi~tion for the detection of chimeric transcripts using novel sequence-specific primers was p~Iro.I.led as above except primers 375 (5'-CTTCTTTCTCTGCCGCATCG) for ST90-375 and 724 (5'-GTGAGGATGATAGGCCTTCC) for ST93-724 were used. Subsequent PCR
conditions were an initial denaturation at 94~C for 2 ~ ules; 30 cycles at 94~C,30 seconds; 58~C, 30 seconds; 72~C, 1 minute, followed by a final extension for 10 mimltPs at 72~C.

2 o Chimeric transcripts amplified by 3 '-RACE and Rl'-PCR were isolated from the gel, blunt-end cloned by standard methods (Sambrook et al., 1989) into the pCR-Script vector (Stratagene) and sequenced using the Sequenase kit Version 2.0 (USB, Cleveland, OH).

2 5 Chromosomal T .oc~1i7~tion of Novel Seql~Pn~es The NIGMS monochromosomal somatic cell hybrid mapping panel #2 was obtained from the Coriell Cell Repositories (Coriell ~n~tin-te for Medical Research, Cam~en, NJ). Primers used were derived from the novel sequences of the chimeric ~Idnscli~ts and 500 ng of genomic DNA from each somatic cell line was used as a template for PCR amplification. For the novel sequence derived from the chimeric transcript obtained from lipoma ST90-375, the primers were 5'- CAGAAGCAGACCAGCAAACC and 5~ lcTcTGccGcATcG and from lipoma ST93-724, the primers were S'-CTCTGGAGCAGTGCAATGTG and 3 5 S'-GTGAGGATGATAGGCCTTCC. PCR con~itiol-~ for the ST93-724 novel sequence primers were 26 cycles of 94~C, 15 seconds; 64~C, 30 seconds; 72~C, 1 SUB~ 111 U 1 k~ 1 (RULE 26) *rB

CA 022~963~ 1999-01-06 minute. For ST90-375, the same conditions were used except that the ~nnP~ling temperature was 62.5~C. PCR products were analyzed on a 7% acrylamide gel.

Tumor C~ell Lines and Chromosome Preparations Lipoma specimens were obtained from patients at the time of surgery. Tumor culture, metaphase chromosom~- harvesting, slide preparation, andtrypsin-Giemsa banding were performed as described previously (Fletcher et al., 1991). Met~ph~es with rearrangements of chromosome 12 in band ql5 were 0 identified and corresponding cell pellets stored in fixative at -20~C were used to prepare slides for FISH. These slides were stored at room temperature for at least 10 days prior to hybridization.

Lambda clones shown in Figure 1 were mapped to lipoma tumor metaphase chromosomes from ST90-375 [46,XX,t(12;15)(ql5;q24)], ST91-198 [46,XX,t(12;13)(qlS;q21-32)], and ST93-724 [46,XX,t(3;12)(q29;qlS)]
Karyotypes for lipomas ST90-375 and ST91-198 have been reported previously (Fletcher et al., 1993).

2 o FISH with Lambda Clones Slides for FISH were prepared as leco"~",~n-le~l in the Hybridization Kit (Oncor, Gaithersburg, MD) except for denaturation at 68~C for 30 seconds.
Lambda probes were labeled with digoxigenin-11-dUTP (Boehringer Mannheim, 2 5 ~nrii~n~rolis, IN) using 1 mg of the appropriate lambda DNA using dNTPs obtained from Boehringer Mannheim and the DNase l/DNA polymerase I mix from the BioNick Labeling System (BRL). Labeling reactions were performed at 16~C for 2 hours. 500 ng of digoxigenin-labeled lambda probe was lyophili7Pd with S mg of Cot-l DNA (BRL) and resllspen~d in 20 ml deionized water. 2 ml 3 o of resuspended probe was added to 9 ml Hybrisol VI (Oncor). The lambda probe was denatured, hybridized to slides, and washed according to standard protocols (Oncor). Digoxigenin-labeled lambda clones were detecte(l using the fluorescein-labeled antidigoxigenin antibody (Oncor) according to the m~mlfactllrer~ s recomm~n~l~tions. Metaphase chromosomes were counterstained with 4,6-3 5 di~mitiinQ-2-phenylindole-dihydrochloride (DAPI) according to the protocolsupplied by Oncor. Hybridization was observed using a Zeiss Axioskop SUB~ 1 l 1 U 1~ ; l (RULE 26) CA 022~963~ 1999-01-06 WO 98/50536 PCT/US9~/21299 microscope and images captured with the (:~ytoVision ~m~ging System (Applied Tm~ging), Fig~res l(A) and l(B) illustrate the genomic structure of the human HMGI-C gene. Figure 1(A): 403, H409, H5003, H1001 and H4002 are genomic lambda FIXII clones (see Materials and Methods) that contain the five exons (El -E5) of the human HMGI-C gene. Figure l(B): Exons are denoted by boxes and introns by a line. Overlapping lambda clones were not obtained within intron 3 and this region is denoted with a dashed line. Sequences encoding potential 0 functional domains, AUG and UAG codons are shown in the exons. The A-T
hook motifs of the DNA-binding domains are shown as stippled areas and the solidregion (in E5) encodes for the acidic domain of unknown function. The Figure is not drawn to scale because of the large 5' and 3' UTRs.

Figures 2(A) through 2(F) illustrate FISH mapping of HMGI-C
lambda clones to lipoma tumor metaphase chromosomes from three lipomas revealing rearrangement of HMGI-C in all three tumors. The normal chromosome 12 homologs provide internal positive hybridization controls and are marked by yellow arrows in each metaphase, while derivative chromosomes are marked by red arrows. Lambda clones H403 and H409 from the 5' end of HMGI-C were used as FISH probes to lipoma met~rh~c~ chromosomes from Figure 2(A) ST90-375 and Figure 2(C) ST93-724, respectively. Note hybridization on the normal chromosome 12 and the der(12), demonstrating that these clones map pro~cimal to the breakpoint in both lipomas. In contrast, when H403 was hybridized to lipoma metaphase chromosomes from Figure 2(E) ST91-198, hybridization was observed on the der(13) showing a map position distal to the breakpoint in this tumor.
H4002 from the 3 ' end of HMGI-C was used as a FISH probe to lipoma metaphase chromosomes from Figure 2(B) ST90-375 and Figure 2(D) ST93-724; note hybridization on the normal chromosome 12 and the der(15) or der(3), respectively, i~ ting that these clones map distal to the breakpoint in both lipomas. However, FISH with H4002 from the 3' end of HMGI-C on Figure 2(F) ST91-198 revealed hybridization on the normal chromosome 12 only, suggesting this clone is deleted from either der (12) or der(15) in this tumor. Met~rh~ce spreads were counterstained with DAPI. Lipoma karyotypes are: ST90-375, t(12;15)(ql5;q24); ST93-724, t(3;12)(q29;ql5); ST91-198, t(12;13)(ql5;q21-32).

S~JBSTITUTE ~ (RULE 26) CA 0225963~ l999-0l-06 .

Figure 3 illustrates RT-PCR amplifi~tion of HMGI-C chimeric llallscripl~. 3' R~CE on RNA from lipomas ST90-375 (375) and ST93-724 (724) yield 441 bp and 672 bp products. Reverse llailsc~i~Lion was ~e,ro~ ed with an oligo-dT prirner linked to an adapter sequence and was followed by a nested PCR
with sense primers from exon 1 and sp~nnin~ exons 1 and 2. DLD-1 is a colorectal adenocarcinoma cell line that expresses wild-type HMGI-C (data not shown) but under these conditions, the predicted 3.1 kb wild-type message was not amplified. Products were analyzed on a 1.5% agarose gel. M are molecular weight markers in kilobases.

Figure 4 illustrates rearrang~ ellL~ of 12ql5 in human lipomas which disrupt the HMGI-C gene and produce chimeric transcripts. HMGI-C
denotes the nucleotide and amino acid sequence of the wildtype gene and the openbox sequence corresponds to the end of HMGI-C exon 3. t(3;12) and t(12;15) refer to the nucleotide and predicted amino acid sequences of the chimeric transcripts from the cloned cDNA products obtained by 3' RACE on RNA isolated from prirnary cell cultures of ST93-724, t(3;12), and ST90-375, t(l2;15), respectively. Chr. 3 and Chr. 15 refer to the novel sequences derived from chromosome 3 or 15 in t(3;12) and t(12;15) lipomas, r~spe~ ely. Only the sequences immPrli~tely adjacent to the fusion sites are shown.

Figure 5 illustrates RT-PCR using primers located on either side of the fusion site between HMGI-C and novel sequences. RNA refers to the lipoma source of total RNA. Primer 375 is an oli~onucleotide that is complementary to the novel sequence from the chimeric kailscli~t of lipoma ST90-375 and is located 8 nucleotides dowll~ a,ll of the fusion point. Primer 724 is a comple~nent~ry oligonucleotide to the novel sequence from the chimeric transcript of lipoma ST93-724 and is located 425 nucleotides downstream of the fusion point. Total RNA
from both lipoma primary cell cultures was reverse transcribed using either 375 or 3 o 724 primers and PCR amplified using HMGI-C sense primer (which spans exons 1 and 2) and the ~nti~Prl.~e primer used for reverse lldns~ lion. Expected productsizes are: 180 bp from ST90-375 cDNA with 375 primer and 597 bp from ST93-724 cDNA with 724 primer.

Figures 6(A) and 6(B) illustrate novel sequences fused to the DNA
binding-domains of HMGI-C which encode transcriptional regulatory domains.

SUBS 111 U ~ (RULE 26) *rB

CA 022~963~ l999-0l-06 Figure 6(A) illustrates a comparison of the novel chromosome 3 sequence from ST93-724 with the LIM domain-cont~ining proteins, zyxin (Sadler et al., 1992), apterous (ap) (Cohen et al., 1992), Lh2 (Xu et al., 1993), Linll (Freyd et al., 1990), ~BTN-l (McGuire et al., 1989). Amino acids that con~ti~lt~ the LIM
domain consensus are highlightPd. The amino acid spacing between the consensus residues is iTl~lir~ted by an x. In addition to the totally conserved cysteine, hi~titlin~ and aspartic acid residues (Sadler et al., 1992), LIM domains are characterized by the presence of an aromatic residue adjacent to the first hicti~int~
and a leucine located C-termin~l to the central HxxCxxCxxC cluster. The lo positions of these conserved residues are inrlir~t~d by arrows. Each LIM domain is designated 1, 2 or 3 depending on its position relative to the N-terminus. The uni~lle~ d sequence of the two LIM domains in the various proteins are shown and gaps were introduced to permit ~lignmPnt of the two LIM domains.
Figure 6(B) illustrates the potential transactivation acidic domain encoded by the sequence derived from chromosome l5 in ST90-375. Acidic residues are underlined and the amino acids, serine and threonine, are in bold type.

Figure 7 illustrates the structure and domain o~ ni7~tion of HMGI-C and the predicted fusion proteins. The vertical dashed line shows the location of junction sites in the chimeric products. DNA binding domains of HMGI-C (AT) are preserved in the fusion proteins but the C-t~rmin~l domain (stippled) is replaced by potential transcriptional regulatory ~lom~in~. LIM, LIM domain; (--), acidic domain; S,T, serine-threonine rich domain.

2 5 HMGI Proteins in Ma~ n Growth and Development Figures 8(A) through (D) illustrate the i~lentifit~ion and genomic chara~Le~ tion of the HMGI-C gene at the pygmy locus in normal and mutant alleles. Figure 8(A): Delineation of the overlapping deleted genomic regions at 3 o the pygmy locus in the spontaneous and transgenic insertional mouse mut~ntc. The open box above clone 3 positions the 0.5kb ApaI-ApaI fragment and the filled boxes ,e~rese,ll single copy sequences used as probes to analyze genomic DNA
isolated from mice of varying genotypes (Xiang, X. et al., 1990). Solid and dashed lines represent presence or absence of genomic seque~lres, respectively, in 35 the transgenic insertional mouse mutant pgTgN40ACha (A) and the spontaneous mutant pygmy (pg). Figure 8(B): Exon amplification from lambda clones 803 and SUB~ 1 U 1~ ;l (RULE26) CA 022~963~ l999-0l-06 5B. The primary PCR exon amplification products in both sense (+) and antisense (-) orientations from the lambda clones shown in Figure 8(A) were analyzed on a 5% polyacrylamide gel (Buckler, A. et al., 1991). The 379bp PCR product observed in the control pSPLl lane results from splicing between the HIV tat andb-globin vector sequences (Buckler, A. et al., 1991). Figure 8(C): Sequence of exons amplified from clone 803 and comparison to the HMGI-C gene. Figure 8(D): A series of overlapping phage clones ext~ ing approximately 190kb at the pygmy locus. The discontinuous region represents an unclonable 1 lkb fragment asest~ t~d from Southern blots of cleaved genomic DNA probed with single copy 0 sequences from the end of the clonable region. The position and number of the HMGI-C exons (not drawn to scale) are shown above the wildtype locus. Single copy sequences were isolated at the in-lic~t~d positions and are represented by filled boxes below the wildtype locus. Thick bars and blank regions represent the genomic sequences that are present or deleted in the two alleles.
Methods. The 0.5kb ApaI-ApaI fragment (Xiang, X. et al., 1990) was used as a probe to isolate clones 3 and 4 from an EMBL3 mouse genomic library (a kind gift of Dr. E. Lacy) and a YAC (902C0711) from a mouse YAC
library (Lehrach, H. et al., 1990). YAC 902CO711 was further subcloned into 2 o lambda FIX II (Ausubel, F. et al., 1988) and 86 clones that hybridized to radioactively-labeled mouse genomic DNA were picked and tldn~ ,d to new plates in a gridded array (Ausubel, F. et al., 1988). Lambda clones 802, 906, SB, 803 and 308 were isolated after the walk was initi~tP(l with the 0.5kb ApaI-ApaIfragment and accomplished by repeated hybridization to filters of the array.
2 5 Overlaps between the contig clones and colinearity with the genome were confirmlo~l by a combination of clone to clone and clone to genomic hybridizations along with restriction mapping. Exon amplification was performed (Exon Trapping System, Gibco BRL) after the genomic inserts from the lambda clones were removed by cleavage wi~ SalI, partially filled-in (Ausubel, F. et al., 1988) 3 o and subcloned into a partially filled-in BamH1 cleaved pSPL1 plasmid (Buckler, A.
et al., 1991). The DNA was electroporated into COS-7 cells at 180V and 960mF
in a Bio-Rad Gene Pulser. Cytoplasmic RNA was isolated after 2-3 days and RT-PCR performed using primers supplied by the m~nllf~rblrer~ The secondary PCR
amplification products (Buckler, A. et al" 1991) from clones 803 and 5B were subcloned into the plasmid vector, pAMP10 (Exon Trapping System, Gibco BRL) and sequenced using the Sequenase Version 2.0 sequencing kit (USB) (Ausubel, F.

SUBSTITUTE S~;~; l (RULE 26) CA 022~963~ 1999-01-06 et al., 1988). A 344bp fragment corresponding to the complete open reading frameof the HMGI-C gene (Mar~loletti, G. et al., 1991) was amplified from 12.5dpc mouse embryos (see text) using reverse transcription (RT) and PCR. Lambda clones cont~ining the HMGI-C exons were then isolated by hybridization of the 344bp radioactively-labeled fragment to the gridded array o~ lambda clones and subsequently conn~cted through chromosome walking. The RT-PCR conditions for isolation of the 344bp fragment consisted of first strand cDNA synthesis with primer 1 (5'-ATGAATTCCTAATCCTCCTCTGC-3'), followed by PCR
amplification with primers 1 and 2 (5'-ATGGATCCATGAGCGCACGCGGT-3').
0 PCR conditions were 94~C, 0.5 minute; 55~C, 0.5 minute; 72~C, 1 minute; for 30 cycles. The amplified product was confirmed by sequencing analysis (Ausubel, F.
et al., 1988).

Figure 9 illustrates HMGI-C gene expression of three alleles at the mouse pygmy locus. The wildtype allele is r~l r~s~llted by +, the transgenic allele pgTgN40ACha by A, the spontaneous mutant allele by pg and an allele at the pygmy locus which involves a paracentric inversion on chromosome 10 (In(10)17Rk) by Rk.

2 o Methods. The genotypes were established for mice in line A and the spontaneous mutant pg as previously described (Xiang, X. et al., 1990), while mice cont~ining the In(10)17Rk inversion were ~let~cted by a PCR-based RFLP
(unpublished results). RNA was isolated from 12.5dpc embryos and equal amounts (5mg) were analyzed by Northern blot hybridization (Ausubel, F. et al., 1988).
The probes were a 138bp nucleotide cDNA fragment encompassing exons 2 and 3 of the HMGI-C gene and a 340bp cDNA fragment cont~ining the complete coding sequence of the HMGI(Y) gene (Johnson, K. et al, 1988). The blot was subsequently hybridized to an oligonucleotide complementary to murine 28S
ribosomal RNA (Barbu, V. & Dautry, F., 1989) to ensure equal amounts of RNA
3 o were present in each lane and the results are shown in the lower panel.

Figures lO(A) through(C) illustrate targeted disruption of the HMGI-C gene. Figure lO(A): Targeting strategy. Endogenous HMGI-C gene (top), targeting vector (middle) and predicted mutant gene (bottom). The targeting vector was created by replacing the 3kb DNA fragment cont~inin~ exonl (E1) and exon2 (E2) with a PGK-neo cassette. The vector also includes a MCl-tk cassette at the 5' SUBSTITUTE ~ ; l (RULE 26) CA 022~963~ 1999-01-06 ~ WO 98/50536 PCT/US97/21299 end of the long homologous segment. B-, BamHI; Probe, a 4kb HincII fragment used to identify the disrupted allele. Figure lO(B): Southern blot analysis of mice from a heterozygous cross. DNA from tails of the mice was digested with BamHI
and hybridized to the external probe (see Figure lO(A~). The positions of the 5 bands corresponding to the wildtype allele (10.5kb) and the mutant allele (9.3kb) are inrlir~teA Figure lO(C): Western blot analysis of wildtype (+/+), heterozygous (+/-) and homozygous (-/-) 12.5dpc embryos with anti-GST-H~iIGI-C rabbit IgG.

o Methods. Genomic clones of the mouse H M GI-C gene were isolated from the mouse pygmy locus as described in Figure 8 legend. Lil~a~i~ed vector (lOmg) was electroporated into AB1 ES cells at 280V, SOOmF, and homologous recombination events enriched for by selection with G418 (350mg/ml) and 2rnM gangcyclovir (Syntex) on SNL76/7 feeder cells. Six targeted clones were obtained and three were injected into C57BL/6J blastocysts to generate chim~er~c. Chimaeric males were mated to C57BL/6J females, and heterozygous offspring h~lel~;rossed to produce subsequent generations. Southern blot analysis of the progeny from heterozygous crosses was pe.ro.llled as described (Ausubel, F. et al., 1988) Proteins were extracted from 12.5 dpc mouse embryos from a heterozygous cross with Iysis buffer cont~ining 50mM Tris-HCI (pH 7.5), 10~
glycerol, 5mM m~gn~sium acetate, 0.2mM EDTA, l.OmM PMSF, and 1% SDS.
lOmg of each sample was separated by 15% SDS-PAGE, transferred to a nylon membrane (Duralon, Stratagene) and HMGI-C was ~i~tected using rabbit IgG anti-mouse GST-HMGI-C, HRP-conjugated goat anti-rabbit IgG and ECL substrate 2 5 (Amersham).

Figures l l(A) through (C) illustrate ~lession of HMGI-C in development and growth. Figure ll(A): Temporal expression pattern of HMGI-C
and HMGI(Y) ~etermin~ by Northern blot analysis of RNA (5mg) isolated from 3 0 the head (H) and body (B) of mouse embryos whose ages in days post coitum are inAir~tPA at the top of the panel. No expression of HMGI-C was ~etected in placenta at any of these stages (data not shown). The probes are described in the legend of Figure 9. Figure ll(B): Spatial localization of HMGI-C tldllscli~Jt~ in 11.5dpc mouse embryos. Photomicrographs of 8 mm, a~ ent7 para~ggit~l sections through 11.5dpc mouse embryos hybridized with the ~ntice~e (A) or sense (B) strand of exon 2 and 3 of HMGI-C or stained histoch~mi~lly with SUB~ 111 U 1~; S~;~; l (RULE 26) h~em~toxylin and eosin (C). G, gut m~senr~lyme; H, heart; L, liver; Lb, limb bud; M, mandible; N, median nasal process; NE, neural epith~ m; O, otocyst.
Magnification: 25X. Figure ll(C): Growth of wildtype and pygmy embryonic fibroblasts. Fibroblasts derived from 13.5dpc embryos were seeded at a concentration of 1.7 x 103 cells per cm2 in DMEM cont~ining 10% fetal bovine serum. Cell number (ordinate) was determined on day 4. Small bars le~lcsell~
standard deviations of triplicate experiments. P < 0.001. The genotypes of embryos were determined as previously described (Xiang, X. et al, 1990) 0 Methods. For in si~u hybridization, CBA/J embryos (11.5dpc) werefixed in 4% paraformaldehyde, dehydrated and embedded in paraffin. Paraffin sections were depar~ffini7~d and hybridized with sense and ~nti~e~e riboprobes corresponding to exons 2 and 3 of HMGI-C as previously described (Duncan, M.
et al., 1992). Sections were stained with h~ toxylin and eosin according to standard procedures.

Translocation Bre~krQ;nt~ Upstream of the HMGI-C Gene in Uterine Lei~ yulllala 2 o Fluorescenre In Situ Hybridization (FISH) Slides for FISH were ~r-,~al~d as recommPn~1e~1 in the Hybridization Kit (Oncor, Gaithersburg, MD), except for denaturation at 68~C for 30 seconds.
HMGI-C clones were in the lambda FIXII vector (Stratagene, La Jolla, CA). They 2s were labeled with digoxigenin-1-dUTP (Boehringer ~q~nnht~im, In(li~n~rolis, IN) with 1 ~4g of the a~ro~liate lambda DNA, dNTPs from Boehringer Mannheim, and the DNasel/DNA polymerase mix from the BioNick Labeling System (BRL, Gaithersburg, MD). Labeling reactions were ~elrollned at 16~C for 2 hours. Five hundred nanograms of digoxigenin-labeled probe were lyophilized with 5 ~g of Cot-1 DNA (BRL) and resuspended in 20 ~4l of deionized water. Two microliters of resuspended probe were added to 9 ~l Hybrisol VI (Oncor). The probe was denatured, hybridized to slides, and washed according to standard protocols (Oncor). Digoxigenin-labeled lambda-clones were ~letected with fluorescein-labeled antidigoxigenin antibody (Oncor) according to the m~mlfactllrer' s s recomm~n~i~tinns~ and m~t~rh~ chromosomes were coulltu.~l~ined with 4,6-mil1ino-2-phenylindole-dihydrochloride (DAPI). Hybridization was observed SUBSTITUTE ~ (RULE 26) CA 022~963~ 1999-01-06 with a Zeiss Axioskop microscope, and images were captured with the CytoVision Tm~ging System (Applied Im~ging, Pittsburgh, PA) Inhibition of HMGI Biological Activity Using ~nti~çn~P Oligonucleotides.

~nti.cen~e oligonucleotides, in particular antisense oligonucleotides to the HMGI genes, can be used to inhibit HMGI biological activity. 'Such antisense oligonucleotides have a nucleotide seqllPn~e complement~ry to at least a portion of the mRNA transcript of the human HMGI genes and are hybridizable to 0 the mRNA transcript. Preferably, the oligonucleotide is at least a 15-mer. More preferably, the oligonucleotide is a 15- to 21-mer. While oligonucleotides having a sequence complem~nt~ry to any region of the human HMGI genes can be used, oligonucleotides complementary to a portion of the mRNA transcripts (i) including the translation initiation codon, and/or (ii) beginning with the second codon from the 5 ' end of the ll~nscli~ts, are particularly l)~ef~ ;d.

The following 15- through 21-mer oligonucleotides are complem~nt~ry to the human HMGI-C rnRNA transcript beginr~ing with the translation initiation codon:
5'-GCC CTC ACC GCG TGC GCT CAT-3' 5'-CC CTC ACC GCG TGC GCT CAT-3' 5'-C CTC ACC GCG TGC GCT CAT-3' 5'- CTC ACC GCG TGC GCT CAT-3' 2 5 5'-TC ACC GCG TGC GCT CAT-3' 5 '-C ACC GCG TGC GCT CAT-3 ' 5'- ACC GCG TGC GCT CAT-3' Similarly, the following 15- through 21-mer oligonucleotides are 3 o complern~nt~ry to the human HMGI(Y) mRNA l~ sclil)l beginning with the translation initiation codon:

SUBS 1 l l U 1~; SHEET (RULE 26) CA 022~963~ 1999-01-06 WO 98/~0536 PCT/I~S97/21299 5'-CTT CGA GCT CGA CTC ACT CAT-3 ' 5'-TT CGA GCT CGA CTC ACT CAT-3' 5'-T CGA GCT CGA CTC ACT CAT-3' 5'- CGA GCT CGA CTC ACT CAT-3' 5'-GA GCT CGA CTC ACT CAT-3 ' 5'-A GCT CGA CTC ACT CAT-3' 5'- GCT CGA CTC ACT CAT-3' Such oligonucleotides are most advantageously pre~cd by using 0 any of the commercially available, automated nucleic acid synthesizers such as the Applied Biosystems 380B DNA Synth~si7Pr~ Since the complete nucleotide sequences of DNAs complelnPnt~ry to HMGI transcripts are known, antisense oligonucleotides hybridizable with any portion of the rnRNA ll~nscri~t may be prepared by the oligonucleotide synthesis methods known to those skilled in the art.
For in vivo use, the ~nticence oligonucleotides may be combined with a conventional ph~rm~rel-tir~l carrier, such as ~ictill~d water, physiological saline, aqueous solution of dextrose and the like. In addition to a~lminictration with conventional carriers, the ~nti.c.on.ce oligonucleotides may be ~(lminictered by a 2 o variety of specialized oligonucleotide delivery tçchni~ es. For example, oligonucleotides can be encapsulated in nnil~m~llar liposomes or in reco~ ed Sendai virus envelopes.

For in vivo use, the antisense oligonucleotides may be a~minictçred 2 5 intravenously in a therape~lfic~lly effective amount suf~lcient to result inextracellular concentrations of 10 to 100 mg/ml. The precise dosage amount and the duration of ~minictration of the ~nticçncP oligonucleotide for the purposes of the present invention will depend upon exigencies of the m~-liral situation and the judgment of the physician carrying out the tre~nPnt in accordance with the 30 conventional practice among mr~ir~l or veterinary professionals. The effective amount of the ~nticçnce oligonucleotide will depend upon such factors as the age, weight and condition of the subject as well as the frequency of ~lminictration and the manner in which the subject responds to tre~tmP-nt Greater or lesser amountsof oligonucleotide may be ~lmi~ d, as required.

SUB~ .; l (RULE 26) CA 022~963~ 1999-01-06 In regulating the amount of carcass fat in farrn ~nim~c, the effective amount of the ~nticence oligonucleotide will depend upon such factors as the ageand weight of the animal and degree of reduction of the carcass fat desired and can be detP-minPd in accordance with conventional methods.

Inhibition of HMGI biological activity using small molecules.

As archit~ct~lral components of the enh~nreosome, a higher order transcription enhancer complex that forms when several distinct transcription factors assemble on DNA in a stereospecific manner, HMGI proteins function to regulate the expression of dowl~LIealll target genes. Disruption of the enhanceosome assembly, by interfering either with protein-DNA or protein-proteininteractions of HMGI proleills results in loss of transcriptional regulation. Small molecule drugs which hlLelr~lc with the function of HMGI proteins as architectural factors can therefore be used to regulate growth and development of adipose tissue.

One method for inhibiting HMGI biological activity can inhibit HMGI DNA-binding function by small molecule drugs which have the same DNA-binding specificity as HMGI proteins. Examples of such small molecules include 2 o netropsin, distamycin A and Hoechst 33258 (bisben7imi~lP), which are cornmercially available, for example, from Sigma. These molecules have been shown to compete with the HMGI proteins for binding to AT-rich DNA (Reeves and Nissen, 1990) suggesting that they possess a structure sirnilar to the HMGI
DNA-binding domains and will be able to inhibit HMGI biological function.
The aforementioned small molecules can be ~lTnini~tered orally, subcutaneously or intravenously to an Ol~;~lllSnl in which regulation of an amount of adipose tissue is needed in an amount sufflcient to result in inhibition in whole or in part of the biological activity of HMGI proteins. The precise dosage amount and the duration of ~-lmini.ctration of the HMGI inhibitor for the purposes of the present invention will depend upon exigencies of the medical situation and the judgment of the physician carrying out the tre~t-n~nt in accordance with the conventional practice among rnPdic~l or v~elh~al~ professionals. The effective amount of the inhibitor will depend upon such factors as the age, weight and condition of the subject as well as the frequency of a~minictration and the manner SUB~i 11 l LJ ~ ; l (RULE 26) CA 022~963~ 1999-01-06 WO 98/50536 PCT/US97t21299 in which the subject responds to tre~tm~nt. Greater or lesser amounts of the inhibitor may be admini~tered, as required.

Assays For Isolation of Small Molecules Which Inhibit Biological Activity of 5 HMGI Proteins Additional small molecule drugs which bind to HMGI pr~teins directly may be obtained by methods known to those skilled in the art. For example, HMGI protein or their fragments may be immobilized on scinti1l~ting 0 plates and a library of various radiolabeled compounds can be screened against the plate using high-throughput screening equipment available cornmercially from, for example, Hewlett-Packard. Binding of a compound to an immobilized HMGI
protein or its fragment will result in increased scintill~tiQn counts. Specific areas of HMGI plut~ins which present attractive targets are, for example~ HMGI DNA-5 binding domains with a cnn.~e~.c~ls sequence TPKRPRGRPKK (Reeves and Nissen,1990) or the sequence PRGRPKGSKNK, implicated in protein-protein interactions involving HMGI ~rolchls (Leger et al., 1995).

Altern~tively, a cell-based assay can be used to isolate small 20 molecules which bind to HMGI proteins or their fr~gJnPnt~. In this assay, a DNA
construct cont~ining a reporter gene such as luciferase gene under control of a HMGI-regulated promoter such as human interferon-l~ promoter (Thanos and Maniatis, 1992) is transfected into a cell line which expresses proteins required for induction of human int~relon-l~ gene, i.e., NF-kb, ATF-2 and an HMGI genes. A
25 library of various compounds is then screened using this cell-based assay andmolecules that inhibit HMGI biological activity are isolated based on their ability to decrease the e~rcssion of the reporter gene.

Throughout this application, various publications have been 3 o rcrerel1ced. The disclosures in these publications are incorporated herein by ~eferellce in order to more fully describe the state of the art.

Refe. ~llces Aaronson, S. A. (1991), Science 254, 1146-1152.

SUBS l l l U ~ ; l' (RULE 26) Asher, H. R., Fejzo M. S., Tk~h~nko, A., Zhou, X., Fletcher, J.
A., Weremowicz, S., Morton, C. C., and Chada, K., Cell 82 57-65 (1995) Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., 5 Seidman, J. G., Smith, J. A., and Struhl, K. (1989), Current Protocols in Molecular Biology (New York: Greene Publishing Associates and Wiley Interscience) .

Barbu, V. &Dautry, F. NucleicAcidsRes. 17, 7115 (1989).

Benson, K. and Chada, K. (1994), Genet. Res. 64, 27-33.

Berlingieri, M. T., Manfioletti, G., Santoro, M., Bandiera, A., Visconti, R., Giancotti, V., and Fusco, A. (1995), Mol. Cell. Biol. 15, 1545-1553.

Bridge, J. A., Persons, D. L., Neff, J. R., and Bhatia, P. (1992), Cancer Detect. Prev. 16, 215-219.

Bridge, J. A., DeBoer, J., Walker, C. W., and Neff, J. R. (1995), Genes Chrom. Cancer 12, 70-72.

Buckler, A. et al. Proc. natn. Acad. Sci. U.S.A. 88, 4005-4009 (1991).
Bullerdiek, J., Wobst, G., Meyer-Bolte, K., Chilla, R., Haubrich, J., Thode, B., and Barnitzke, S. (1993), Cancer Genet. Cytogenet. 65, 27-31.

Chen, H., Charlat, O., Tartaglia, L.A., Woolf, E.A., Weng, X., 30 Ellis, S.J., Lakey, N.D., Culpepper, J., Moore, K.J., Bleil~lL, R.E., Duyk, G.M., Tepper, R.I., and Molgell~L~ill, J.P. (1996). Cell 84, 491-495.

Chirgwin, J., Przybyla, A., MacDonald, R., and Rutter, W. J.
(1979), Biorh~ try 18, 5294-5299.

SUBSTITUTE ~ (RULE 26) ~ .

CA 022~963~ 1999-01-06 ~ WO 98/50536 PCTrUS97/21299 Cohen, B., McGuffin, M. E., Pfeifle, C., Segal, D., and Cohen, S.
M. (1992), Genes Dev. 6, 715-729.

Dal Cin, P., Kools, P., De Jonge, I., Moerrnan, P., Van de Ven, W., and Van den Berghe, H. (1993), Genes Chrom. Cancer 8, 131-133.

de Thé, H., Lavau, C., Marchio, A., Chomienne, C., Degos, L., and Dejean, A. (1991), Cell 66, 675-684.

0 DiMarco, A., Gaetani, M., Orezzi, P., Scotti, T., and Arcamone, F. (1962), Cancer Chemo. Reports 18,15-19.

Du, W., Thanos, D., and Maniatis, T. (1993), Cell 74, 887-898.

Duncan, M., DiCicco-Bloom, E.M., Xiang, X., Benezra, R. &
Chada, K. Dev. Biol. 154, 1-10 (1992).

Feuerstein, R., Wang, X., Song, D., Cooke, N. C., and Liebhaber, S. A. (1994), Proc. Natl. Acad. Sci. USA 91, 10655-10659.
Finlay, A.C., Hochstein F.A., Sobin, B.A., and Murphy, F.X.
(1951), J. Amer. Chem. Soc. 73, 341-343.

Fletcher, J. A., Kozakewich, H. P., Hoffer, F. A., Lage, J. M., Weidner, N., Tepper, R., Pinkus, G. S., Morton, C. C., and Corson, J. M.
(1991), N. Engl. J. Med. 324, 436-443.

Fletcher, J. A., Pinkus, G. S., Donovan, K., Naeem, T., Sugarbaker, D. J., Mentzer, S., Pinkus, J. L., and Longtine, J. (1992), Cancer Res. 52, 6224-6228.

Fletcher, J. A., Kozakewich, H. P., Schoenberg, M. L., and Morton, C. C. (1993), Genes, Chrom. Cancer 6, 24-29.

3 5 Fletcher, J. A., Longtine, J., Wallace, K., Mentzer, S. J., and Sugarbaker, D. J. (1995), Genes Chrom. Cancer 12, 210-223.

SUBSTITUTE ~ (RULE 26) CA 022~963~ l999-0l-06 W098/50536 PCTrUS97/21299 Freyd, G., Kim, S. K., and Horvitz, H. R. (1990), Nature 344, 876-879.

Frie~lm~nn, M., Holth, L. T., Zoghbi, H. Y., and Reeves, R.
(1993), Nucleic Acids Res. 21, 4259-4267.

Frohman, M. A., Dush, M. K., and Martin, G. R. (1988), Proc.
Natl. Acad. Sci. USA 85, 8998-9002.

German, M. S., Wang, J., Chadwick, R. B., and Rutter, W. J.
(1992), Genes Dev. 6, 2165-2176.

Giancotti, V., Pani, B., Andrea, P. D., Berlingieri, M. T., Di 5 Fiore, P. P., Fusco, A., Vecchio, G., Philp, R., Crane-Robinson, C., Nicolas, R.
H., Wright C. A., and Goodwin, G. H. (1987), EMBO J. 6, 1981-1987.

Giancotti, V., Buratti, E., Perissin, L., Zorzet, S., R~lm~in, A., Portella, G., Fusco, A., and Goodwin, G. H. (1989), Exp. Cell Res. 184, 538-545.

Giancotti, V., Bandiera, A., Ciani, L., Santoro, D., Crane-Robinson, C., Goodwin, G. H., Boiocchi, M., Dolcetti, R., and Casetta, B.
(1993), Eur. J. Biochem. 213, 825-832.

Green, E. D. and Olson, M. V. (1990), Proc. Natl. Acad. Sci. USA
87, 1213-1217.

Green, M.C. in Genetic Variants and Strains of the Laboratory M
3 o ouse (eds. Lyon, M. & Searle, A.) 12-403 (Oxford University Press, Oxford, 1989).

Grossch~cll, R., Giese, K. and Pagel, J. (1994), Trends Gen. 10, 94-100.

SUB~ 111 U l ~ ; l (RULE 26) CA 022~963~ l999-0l-06 Gu, Y., Nakamura, T., Alder, H., Prasad, R., C~n~ni, O., Cirnino, G., Croce, C. M., and Can~ni, E. (1992), Cell 71, 701-708.

Guthrie, C. and Fink, G. R. (1991), Meth. Enzymol. 194, 218-219.
Hatano, M., Roberts, C. W. M., Minden, M., Crist, W. M., and Korsmeyer, S. J. (1991), Science 253, 79-82.

Hirabayashi, Y., Yoshida, M. A., Ikeuchi, T., Ishida, T., Kojima, T., Higaki, S., M~hin~mi, R., and Tonomura, A. (1992), Cancer Genet.
Cytogenet. 60, 35-40.

Jenkins, R. B., Kimmell, D. W., Moertel, C. A., Schultz, C. A., Menezes, R. M., Scheihauer, B., Kelly, P. J., and Dewald, G. W. (1989), Cytogenet. Cell Genet. 51, 1019.

Johnson. K. R., Lehen, D. A., Elton, T., Barr, P., and Reeves, R.
(1988), J. Biol. Chem. 263, 18338-18342.

Johnson, K. R., Lehen, D. A., and Reeves, R. (1989), Mol. Cell.
Biol. 9, 2114-2123.

Justice, M. J., Siracusa, L. D., Gilbert, D. J., Heisterk~mr~, N., Groffen, J., Chada, K., Silan, C., Copeland, N. and Jenkins, N. A. (1990), 2 5 Genetics 125, 855-866.

Kamps, M. P., Look, A. T., and Baltirnore, D. (1991), Genes Dev 5,358-368.

Karlsson, O., Thor, S., Norbert, T., Ohlsson, H., and Edlund, T
(1990), Nature 344, 879-882.

King, J. Genetics 53, 487-497 (1955).

SUBSTITUTE SHEET (RULE 26) CA 022~963~ 1999-01-06 Kopka, Mary L., Yoon, Chun, Goodsell, David, Pjura, Philip, and Dickerson, Richard E., (1985), Proc. Natl. Acad. Sci. USA, Vol. 82, pp. 1376-1380.

Leger, H., Sock, E., Renner, K., Grummt, F., and Wegner, M.
(1995). Mol. Cell. Biol. 15, p. 3738-3747.

Lehrach, H. et al. in Genome Analysis Volume 1: Genetic and Physical Mapping (eds Davies, K. & Til~hm~n, S.) 39-81 (Cold Spring Harbor Laboratory Press, New York, 1990).

Li, S. et al. Nature 347, 528-533 (1990).

Lin, S-C. et al. Nature 364, 208-213 (1993).
Ma, Q., Alder, H., Nelson, K. K., Chatterjee, D., Gu, Y., Nakarnura, T., Can~ni, E., Croce, C. M., Siracusa, L. D., and Buchberg, A. M.
(1993), Proc. Natl. Acad. Sci. USA 90, 6350-6354.

MacArthur. J. (1944), Amer. Nat. 78, 142-157.

M~n~ hl~ N., Heim, S., ~rh~den, K., Rydholm, A., Willen, H., and Mitelman, F. (1988), Hum. Genet. 79, 203-208.

l~anrl~hl~ N., Heim, S., Arheden, K., Rydholm, A., Willen, H., and Mitelman, F. (1989), Cancer 65, 242-248.

Mandahl, N., Orndal, C., Heim, S., Willen, H., Rydholm, A., Bauer, H. C. F., and Mitelm~n, F. (1993a), Cancer 71, 3009-3013.
Mandahl, N., Willen, H., Rydholm, A., and Mitelman, F.
(1993b), Genes Chrom. Cancer 6, 121-123.

Manfioletti, G., Giancotti, V., Bandiera, A., Buratti, E., Sautiere, P., Cary, P., Crane-Robinson, C., Coles, B., and Goodwin, G. H. (1991), Nucleic Acids Res. 19, 6793-6797.

SUBSTITUTE ~ ; l (RULE 26) CA 022~963~ 1999-01-06 May, W. A., Gishizky, M. I., TPSSnjrL, S. L., Lunsford, L. B., Lewis, B. C., Delattre, O., 71lcm~n, J., Thomas, G., and Denny, T. D. (1993), Proc. Natl. Acad. Sci. USA 90, 5752-5756.

McGuire, E. A., Hockett, R. D., Pollock, K. M., Bartholdi, M. F., O'Brien, S. J., and Korsmeyer, S. J. (1989), Mol. Cell. Biol. 9, 2124-2132.

Mitrh~ll, P. J., and Tijan, R. (1989), Science 245, 371-378.

Nilbert, M., and Heim, S. (1990), Genes Chrom. Cancer 2, 3-13.

Nissley, S., Kn~7~1~, R. & Wolff, G. Horm. Metab. Res. 12, 158-164 (1980).
Noguera, R., Llombart-Bosch, A., Lopez-Gines, C., Carda, C., and Fern~n-lez, C. (1989), Virchows Arch. A. Pathol. Anat. Histopathol. 415, 377-382.

Patel, U. A., Bandeira, A., Manfioletti, G., Giancotti, V., Chau, K-Y., and Crane-Robinson, C. (1994), Biochem. Biophys. Res. comIn. 201, 63-70.

Pendergast, A. M., Muller, A. J., Havlik, M. H., Maru, V., and Witte, O. N. (1991), Cell 66, 161-171.

Prasad, R., Leshkowitz, D., Gu, Y., Alder, H., Nakamura, T., Saito, H., Huebner, K., Berger, R., Croce, C. M., and Can~ni, E. (1994), Proc.
Natl. Acad. Sci. USA 91, 8107-8111.
Rabbitts, T. H. (1994), Nature 372, 143-149.

Ram, T., Reeves, R. & Hosick, H.L. Cancer Res. 53, 2655-2660 (1993)-SUB~ l U ~ ;l (RULE 26) *rs . ~

Reeves, R., and Nissen, M. S. (1990), J. Biol. Chem. 265, 8573-8582.

Reeves, R., T ~ng~n, T.A. & Nissen, M.S. Proc. natn. Acad. Sci.
U.S.A. 88, 4005-4009 (1991).

Rein, M. S., Fri~lm~n, A. J., Barbieri, R. L., Pavelka, K., Fletcher, J. A., and Morton C. C. (1991), Obstet. Gynecol. 77, 923-926.

0 Rink T.J. (1994), In search of a satiety factor. Nature 372 p. 406-407.

Rogers, P., and Webb, G. P. (1980), British J. Nutrition 43, 83-86.

Rohen, C., Bonk, U., Staats, B., Rarnit7ke, S., and Bullerdiek, J.
(1993), Cancer Genet. Cytogenet. 69: 68-71.

Rother, R. P. (1992), Biotechniques 13, 524.

Sadler, I., Crawford, A. W., Michelsen, J. W., and Beckerle, M.
C. (1992), J. Cell Biol. 119, 1573-1587.

Saitoh, Y. & ~ ~ernmli, U.K. Cell 76, 609-622 (1994).

Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). Molecular cloning: A Laboratory Manual, 2nd edition (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press).

SfinrhP7-Garcia, I., Osada, H., Forster, A., and Rabbitts, T. H.
3 o (1993), EMBO J. 12, 4243-4250.

.S~nrhr7-Garcia, I., and Rabbitts, T. H. (1993), Sem. Cancer Biol.
4, 349-358.

Sandros, J., Stenman, G., and Mark, J. (1990), Cancer Genet.
Cytogenet. 44, 153-167.

SUBS l 11 U ~ (RULE 26) CA 022~963~ l999-0l-06 Schmit7., M. L., and Baeuerle, P. A. (1991), EMBO J. 10, 3805-3817.

Schoenberg Fejzo, M., Yoon, S-J, Montogomery, K., Rein, M. S., Weremowicz, S., Krauter, K. S., Dor nan, T. E., Fletcher, J. A., Mao, J., Moir, D. T., Kucherlapati, R. S. and Morton, C. C. (1995), Genomics 26, 265-271.

Seipel, K., Georgiev, O., and Sch~ffn~r, W. (1992), EMBO J. l1, 104961-4968.

Sinha, Y., Wolff, G., Baxter, S. & Domon, O. Proc. Soc. Expt.
Biol. Med. 162, 221-223 (1979).

Sre~ nt~i~h, C., Leong, S. P. L., Karakousis, C. P., McGee, D.
L., Rappaport, W. D., Villar, H. V., Neal, D., Fleming, S., Wankel, A., Herrington, P. N., Carmona, R., and Sandberg, A. (l99l), Cancer Res. 51, 422-433.

Tartaglia, L.A., Dembski, M., Weng, X., Culpepper, J., Devos, R., Richards, G.J., Camp~leld, L.A., Clark, F.T., Deeds, J., Muir, C., Sanker, S., Moriarty, A., Moore, K.J., Smutko, J., Mays, G.G., Woolf, E.A., Monroe, C.A., and Tepper, R.I. (1995). Cell 83, p.1263-1271.

Thanos, D., and Maniatis, T. (1992), Cell 71, 777-789.

Tjian, R. & Maniatis, T. Cell 77, 5-8 (1994).

Tk~rh~k, D. C., Kohler, S., and Cleary, M. L. (1992), Cell 71, 30691-700.

Valge-Archer, V E., Osada, H., Warren, A. J., Forster, A., Li, J., Baer, R., and Rabbitts, T. H. (1994), Proc. Natl. Acad. Sci. USA 91, 8617-86212.

SUB~ 1 l 1 U l ~ ; l (RULE 26) CA 022~963~ 1999-01-06 Vanni, R., Dal Cin, P., Marras, S., Moerrnan, P., Andria, M., Valdes, E., Deprest, J., and Van den Berghe, H. (1993), Cancer Genet.
Cytogenet. 68, 32-33.

Vartainen, E., Palvimo, J., Mahonen, A., Linnala-K~nkln-nP~, A., and Maenpaa P. H. (1988), FEBS Lett. 228, 45-48.

Way, J. C., and Chalfie, M. (1988), Cell 54, 5-16.

0 Wolffe, A. (1994), Science 264, 1100-1101.

Xiang, X., Benson K. F., and Chada, K. (1990), Science 247, 967-969.

Xu, Y., Ral-l~cs~re, M., Fisher, P., Rathbun, G., Oltz, E. M., Yancopoulos, G.
F. W. (1993), Proc. Natl. Acad. Sci. USA 90, 227-231.

Zhang, Y., Proenco, R., Maffel, M., Barone, M., Leopold, L., and FriP~lm~n, J.M. (1994) Position~l cloning of the mouse obese gene and its human homologue. Nature 372, p.425-431.

Zhou, X., Benson, K. F., Ashar, H. R., and Chada, K., Nature 376 771-774 (1995).

While the invention has been particularly described in terms of specific embodimPnt.c, those skilled in the art will understand in view of the present disclosure that numerous variations and mo~lifirations upon the invention are now enabled, which variations and mo~lifir-~tinns are not to be regarded as a departure from the spirit and scope of the invention. Accordingly, the invention is to be 3 o broadly construed and limited only by the scope and spirit of the following claims.

SUB~ 1 l l U 1~ ; r (RULE 26)

Claims (61)

We claim;

1. A method for treating obesity in a mammal which comprises reducing the biological activity of HMGI genes in the mammal.

2. The method according to claim 1, wherein at least 10% of the biological activity of HMGI genes is reduced.

3. The method according to claim 2, wherein at least 50% of the biological activity of HMGI genes is reduced.

4. The method according to claim 1, wherein the biological activity of HMGI-C genes is reduced.

5. The method according to claim 1, wherein the biological activity of HMGI-(Y) genes is reduced.

6. The method according to claim 1, wherein the mammal is leptin-deficient or leptin receptor-deficient.

7. The method according to claim 1, wherein the reduction in biological activity of HMGI genes is achieved by inhibiting the expression of HMGI genes.

8. The method according to claim 1, wherein the reduction in biological activity of HMGI genes is achieved by administering to the mammal a therapeutically effective amount of an oligonucleotide which has a nucleotide sequence complementary to at least a portion of the mRNA of the HMGI gene.

9. The method according to claim 1, wherein the reduction in biological activity of HMGI genes is achieved by inhibiting the DNA-binding activity of HMGI genes.

10. The method according to claim 9, wherein the inhibition of the DNA-binding activity of HMGI genes is achieved by administering to the mammal a therapeutically effective amount of an inhibitor compound selected from the group consisting of netropsin, distamycin A, or Hoechst 33258 (bisbenzimide).

11. The method according to claim 1, wherein the reduction in biological activity of HMGI genes is achieved by inhibiting the protein-protein interactions of HMGI proteins.

12. The method according to claim 1, wherein the mammal is a human.

13. The method according to claim 1, wherein the mammal is a rodent.

14. The method according to claim 13, wherein the biological activity of HMGI genes is substantially reduced by breeding the mammal with an inactivated HMGI gene sequence introduced into the mammal, or an ancestor of the mammal, at an embryonic stage.

15. The method according to claim 14, wherein the inactivated HMGI gene sequence is an inactivated HMGI-C gene sequence.

16. The method according to claim 15, wherein the inactivated HMGI-C gene sequence is set out in Figure 10.

17. A method for treating a tumor in a patient by reducing the biological activity of normal HMGI genes which comprises administering to the patient a therapeutically effective amount of an inhibitor compound active against normal HMGI-C or HMGI(Y) genes.

18. The method according to claim 17, wherein the biological activity of normal HMGI-C genes is reduced.

19. The method according to claim 17, wherein the biological activity of normal HMGI-(Y) genes is reduced.

20. The method according to claim 17, wherein the reduction in biological activity of normal HMGI genes is achieved by inhibiting the expression of normal HMGI genes.

21. The method according to claim 17, wherein the reduction in biological activity of normal HMGI genes is achieved by administering to the patient a therapeutically effective amount of an oligonucleotide which has a nucleotide sequence complementary to at least a portion of the mRNA of the normal HMGI gene.

22. The method according to claim 17, wherein the reduction in biological activity of normal HMGI genes is achieved by inhibiting the DNA-binding activity of normal HMGI genes.

23. The method according to claim 22, wherein the inhibition of the DNA-binding activity of normal HMGI genes is achieved by administering to the patient a therapeutically effective amount of an inhibitor compound selected from the group consisting of netropsin, distamycin A, or Hoechst 33258 (bisbenzimide).

24. The method according to claim 17, wherein the tumor is mesenchyme-derived and benign.

25. The method according to claim 24, wherein the tumor is uterine leiomyomata, lipomas, pleomorphic adenomas of the salivary gland, pulmonary chondroid hamartoma, endometrial polyps, epithelial breast tumors, hemangiopericytoma, or angiomyxoma.

26. The method according to claim 25, wherein the tumor is uterine leiomyomata, lipomas, or pleomorphic adenomas of the salivary gland.

27. The method according to claim 17, wherein the tumor is a malignant tumor of epithelial origin.

28. The method according to claim 27, wherein the tumor is a carcinoma of the lung, colon, breast, prostate, thyroid gland, or skin.

29. The method according to claim 17, wherein the reduction in biological activity of normal HMGI genes is achieved by inhibiting the protein-protein interactions of HMGI proteins.

30. A method of producing a transgenic non-human mammal, the germ cells and somatic cells of which contain an inactivated HMGI gene sequence introduced into the mammal, or an ancestor of the mammal, at an embryonic stage.

31. The mammal according to claim 30, wherein the inactivated HMGI gene sequence is an inactivated HMGI-C gene sequence.

32. The mammal according to claim 31, wherein the inactivated HMGI-C gene sequence is set out in Figure 10.

33. The mammal according to claim 30, wherein the genome of the mammal does not encode for both the functionally active leptin gene and the functionally active HMGI genes.

34. A method for screening candidate compounds capable of inhibiting the biological activity of normal HMGI proteins, or a fragment thereof, which comprises the steps of:
(a) incubating a HMGI protein, or a fragment thereof, with a candidate compound under conditions which promote optimal interaction; and (b) measuring the binding affinity of the candidate compound to the HMGI protein, or a fragment thereof; and (c) determining from the binding affinity which candidate compounds inhibit the biological activity of HMGI proteins, or a fragment thereof.

35. The method according to claim 34, wherein the candidate compound inhibits the biological activity of normal HMGI proteins, or a fragmentthereof, in an amount of at least 10%.

36. The method according to claim 34, wherein the binding affinity is measured using a scintillation proximity assay.

37. The method according to claim 34, wherein the binding affinity is measured using a fluorescence polarization assay.

38. A method for screening candidate compounds capable of inhibiting the biological activity of normal HMGI genes which comprises the steps of:
(a) transfecting into a cell a DNA construct which contains a reporter gene under control of a normal HMGI protein-regulated promoter;
(b) administering to the cell a candidate compound;
(c) measuring the levels of reporter gene expression; and (d) determining from the levels of reporter gene expression which candidate compounds inhibit the HMGI biological activity.

39. The method according to claim 38, wherein the candidate compound inhibits the biological activity of normal HMGI genes in an amount of at least 10%.

40. A method for detecting normal HMGI proteins as a diagnostic marker for a tumor using a probe that recognizes normal HMGI proteins, which comprises the steps of:
(a) contacting normal HMGI proteins from a sample from a patient with a probe which binds to HMGI proteins; and (b) analyzing for normal HMGI proteins by detecting levels of the probe bound to the normal HMGI proteins, wherein the presence of normal HMGI
proteins in the sample is positive for a tumor.

41. The method according to claim 40, wherein normal HMGI-C
proteins are detected.

42. The method according to claim 40, wherein normal HMGI(Y) proteins are detected.

43. The method according to claim 40, wherein the tumor is mesenchyme-derived and benign.

44. The method according to claim 43, wherein the tumor is uterine leiomyomata, lipomas, pleomorphic adenomas of the salivary gland, pulmonary chondroid hamartoma, endometrial polyps, epithelial breast tumors, hemangiopericytoma, or angiomyxoma.

45. The method according to claim 40, wherein the tumor is a malignant tumor of epithelial origin.

46. The method according to claim 45, wherein the tumor is a carcinoma of the lung, colon, breast, prostate, thyroid gland, or skin.

47. The method according to claim 40, wherein the probe is an antibody.

48. The method according to claim 40, wherein the sample is a biopsy sample, a urine sample, a blood sample, a feces sample, or a saliva sample.

49. The method according to claim 40, wherein the method is a histological assay, biochemical assay, flow cytometry assay, Western blot assay, or solution assay.

50. The method according to claim 40, wherein a positive and negative control sample are treated according to the method of claim 38 to assess the level of normal HMGI proteins in a tumor sample and a nontumor sample, respectively.

51. A method for detecting antibodies to normal HMGI proteins using a probe that recognizes antibodies to HMGI normal proteins, which comprises the steps of:
(a) treating a sample from a patient with a probe which binds to antibodies to normal HMGI proteins; and (b) analyzing for antibodies to HMGI proteins by detecting levels of the probe bound to the antibodies to HMGI proteins, wherein the presence of antibodies to normal HMGI proteins in the sample is positive for a tumor.

52. The method according to claim 51, wherein antibodies to normal HMGI-C are detected.

53. The method according to claim 51, wherein antibodies to normal HMGI(Y) are detected.

54. The method according to claim 51, wherein the probe is normal HMGI-C or HMGI(Y) proteins.

55. The method according to claim 51, wherein the tumor is mesenchyme-derived and benign.

56. The method according to claim 55, wherein the tumor is uterine leiomyomata, lipomas, pleomorphic adenomas of the salivary gland, pulmonary chondroid hamartoma, endometrial polyps, epithelial breast tumors, hemangiopericytoma, or angiomyxoma.

57. The method according to claim 51, wherein the tumor is a malignant tumor of epithelial origin.

58. The method according to claim 57, wherein the tumor is a carcinoma of the lung, colon, breast, prostate, thyroid gland, or skin.

59. The method according to claim 51, wherein the sample is a biopsy sample, a urine sample, a blood sample, a feces sample, or a saliva sample.

60. The method according to claim 51, wherein the method is a histological assay, biochemical assay, flow cytometry assay, Western blot assay, or solution assay.

61. HMGI genes and proteins for use as a starting point to isolate downstream target genes regulated by the HMGI genes and proteins.