Lip, a Human Gene Detected by Transfection of DNA From a Human Liposarcoma Encodes a Protein With Homology to Regulators of Small G Proteins

Purpose/Method. Transfection experiments have been used to identify activated oncogenes in a wide variety of tumour types. Here we describe the use of transfection experiments utilizing DNA from a human pleomorphic liposarcoma to identify a novel gene, designated lip which maps to chromosome 19. Results. lip was expressed in all sarcoma cell lines examined and a wide variety of normal tissues. Sequencing of cDNAs prepared from transcripts of the normal lip gene indicates that lip is predicted to encode a 966 amino acid protein with a region of homology to proteins such as vav, dbl, lbc and ect-2 which act as GDP–GTP exchange factors for the RAS superfamily of small GTP-binding proteins, and the N-terminal 830 amino acids are identical to the recently identified gene p115-RhoGEF, an exchange factor for RHOA. In transfectants, lip has undergone a rearrangement which results in C-terminal truncation of the predicted LIP protein. However, we failed to detect this alteration in the primary liposarcoma used in the original transfection experiments, or in other sarcoma specimens examined. Discussion. When considered together, these observations suggest that transforming lip sequences represent an alternatively spliced form of p115-RhoGEF that is activated for transformation by C-terminal truncation during transfection, and is not widely involved in sarcoma development.


Introduction
T ransfection o f D N A into N IH 3T 3 m ou se ® broblasts has been u sed to survey a w ide variety of hum an tum ou rs for the presen ce of transfo rming oncogenes. A ltho ugh the m ajority of the genes detected by this assay are activated ras genes, a n um ber of other genes including m et, 1 ret, trk, 3 m as 4 , dbl, 5 ra f, 6 hst, 7 v av, 8 ufo/axl, 9,10 ect± 2 11 an d lbc 12 have also been iden ti® ed. T hese gen es en code protein s of several fu nction al classes, inclu ding grow th factors, tran sm embrane receptors w ith tyrosin e kinase activity. n on-receptor serine± threon in e kinases and regulators of sm all G T P-b ind ing proteins, all of w hich are thou ght to play a role in intracellu lar signallin g pathw ays w hich regulate cellu lar proliferation.
In an attempt to identify oncogenes activated in hum an soft tissue tum ours, D N A from 29 sarcom as w as exam ined for the ability to transfo rm N IH3T 3 cells. 13 These studies identi® ed an activated k-ras gene in a leiom yosarcom a, and a novel activated gene follow ing transfection of D N A from a pleom orphic liposarcom a. Genom ic fragm ents of this novel gene, designated lip, were cloned by screening a genom ic library prepared using DN A from a lip secondary transfectant with a hum an alu-repeat probe. Repeat-free subclones of these genom ic clones have been used to dem onstrate that this gene m aps to chromosom e 19 and is expressed as a 3.0-kb transcript in prim ary and secondary lip transfectants. 13 cDN As corresponding to the norm al lip gene have now been cloned from a norm al ® broblast cDN A library and through sequencing analysis w ere found to encode a protein with regions of hom ology to proteins such as exchange factors for sm all GT Pbinding proteins. In addition the N -term inal 830 am ino acids are alm ost com pletely identical to the recently cloned p115-R hoGEF. 14

RAC E polym erase chain reaction analysis
C ytoplasm ic RN A was extracted from subcon¯uent cultures of M CF -7 cells. 15 59 RACE (rapid am-pli® cation of cD N A ends) was perform ed as recomm ended using the 59 RAC E System (G ibco BRL), w ith the exception that RN A w as reverse transcribed using random hexam er prim ers (Pharm acia). Polym erase chain reaction (PC R) prim ers were as follows: ® rst round of am pli® cation GAG TT T-G TC TC CAG CT CG, second round of am-pli® cation CT CAAAATC CTC ATC CT CAGC .

Preparation and screening of cD N A libraries
cD N A clones corresponding to the normal LIP gene w ere obtained by screening a random ly prim ed H T1080 cD NA library l6 and an oligo-dT-p rim ed M 426 hum an ® broblast cDN A library l7 C lones corresponding to the transfected gene were obtained from a prim ary transfectant cD N A library constructed as follows. RN A was extracted from sub-con¯uent cultures of transfectant cells, l8 and poly(A) 1 RN A selected using oligo-(dT )-cellulose (Poly(A) Quik Kit, Stratagene). T he cD N A library w as constructed in the lam bda ZAP II vector (Stratagene) using the ZAP II cD N A synthesis kit (Stratagene). All three libraries were screened using Biodyne hybridization m em branes (Pall-B iodyne) as previously described. l6

cD N A sequencing
Partially deleted subclones of the cDN A inserts w ere generated using exonuclease III and m ung bean nuclease (Stratagene), and these subclones sequenced by the dideoxy chain term ination m ethod 19 using the Sequenase Version 2.0 sequencing kit (United States Biochem icals).
Alternatively, prim ary transfectant poly(A) 1 RN A w as reverse transcribed using random hexam er prim ers (Pharmacia) and Superscript (BRL), and prim ers derived from the norm al lip sequence were

Isolation and sequencing of cDN A clones
The genomic clone MC15, isolated in our previous study, l3 was used to screen an oligo-dT primed M426 human ® broblast library, l7 and two clones of 3.0 kb (11A) and 2.6 kb (13A) were isolated. Complete bidirectional sequencing of clone 11A generated a sequence of 3023 bp including a 39-bp polyA tail. Although this clone corresponded in size to the 3.0 kb transcript detected by Northern analysis in a variety of human tumour cell lines (data not shown), the open reading fram e present in this cDNA extended to the 59 end of the sequence, raising the possibility that an additional 59 sequence existed. The same probe was therefore used to screen 100 000 clones from a randomly primed HT1080 cDNA library and three additional clones 12C, 5A and 3B were isolated. Of these, 3B contained a 2.6-kb insert, was primed at nucleotide 2542 with respect to 11A and possessed an alternative 59 end ( Fig. 1(a)). Subsequent 59 RAC E analysis, using RNA derived from a human breast cell line, was used to demonstrate that 18 bp of the 59 sequence was missing from the original clone 11A. This upstream sequence, GAGG CTTCGGTTCCG-GTG, did not, however, encode an upstream stop codon or an initiating methionine. In clone 11A, there are two methionines at the 59 terminus which conform closely to the Kozak consensus sequence initiating at methionine, 20 the ® rst 7 bp and the second 52 bp from the 59 end of this clone. Neither is preceded by an in-fram e stop codon. In clone 3B the methionine at position 7 is absent ( Fig. 2(a)). W e have therefore chosen the methionine at position 52 as the most likely translation start site since this is present in both 11A and 3B. If this is the case, lip encodes a single major open-reading fram e of 2898 nucleotides predicting a protein of 966 amino acids. The sequence and the putative open-reading fram e are shown in Fig.  2 lip expression lip was expressed as a 3-kb transcript in all cell lines examined. These included the ® brosarcoma cell lines HT1080 and Hs913T, the leiomyosarcoma lines SK-UT-1 and SK-LMS-1, the rhabdomyosarcoma cell lines A204, RM S and RD, the Ewings sarcoma cell line A673, the promyelocytic leukaemia cell line HL-60 and the carcinoma cell line A431. In addition, lip was shown to be expressed in a variety of tissues including tonsil, spleen, renal cortex, lung, prostate, endometrium and breast (data not shown). The predicted lip protein contains nucleotide exchange factor and PH dom ains T he predicted lip protein contains regions of m oderate sim ilarity to the previously described dbl hom ology (DH) dom ain seen in the transform ing oncogenes dbl, vav, ect-2 and tim; 21 the yeast cell cycle gene cdc24, the bcr gene 22 and a nucleotide exchange factor for ras, p140-R asG RF. 23 Using the F AST A search program me, the core region of this D H dom ain (am ino acids 416± 610) was m ost closely related to the lbc, vav, ect-2 and cdc24 genes, dem onstrating 33.8% , 25.9%, 25.5% and 24.3% identity respectively (Fig. 3(a)).
A second region of sim ilarity to the pleckstrin hom ology (PH) domain 24 has been found to span am ino acids 649 to 758. T his dom ain, ® rst identi® ed as an internal repeat in pleckstrin, the m ajor substrate of protein kinase C in platelets, 25 has been identi® ed in a num ber of other proteins including the products of the vav, dbl, rasgrf, bcr, lbc and cdc24 genes. lip shows 24% identity with the pleckstrin C-term inal PH dom ain. Although the identity is low, the fam ily m embers noted so far have exhibited only 21± 25% identity. All ® ve of the previously de® ned subdom ains can be identi® ed, and the m ost highly conserved residues which de® ne these subdom ains 25 are also conserved in lip ( Fig. 3(b)).
The C-term inal 200 am ino acids are relatively proline rich (16% ) and contain a num ber of sites which conform to the m inim al consensus sequence for SH3 dom ain binding sites, P± */p± X± P (P 5 proline, */p 5 usually hydrophobic/proline, X 5 not conserved). 26 T here is an additional potential SH 3 dom ain-binding sequence in the am ino  phorylation at residue 487 and a single potential site for m itogen-activated protein kinase (M APK) phosphorylation, P± X± S/T ± P, at residue 954 ( Fig. 2(b)).
LIP shares identity with p115-RHOGEF, an exchange factor for RH OA p115-R H OG EF w as identi® ed using RH OA as an af® nity-puri® ed ligand, and subsequently cloned term inal region. AASPGPSRPG L, which closely resem bles the SH3 dom ain binding sequences seen in the hum an dynam in protein, a m icrotubule binding protein which has G TPase activity and is thought to be involved in vesicle traf® cking. 27 A Prosite database m otif search reveals 13 consensus sequences for protein kinase C phosph oryation, 13 consensus sequences for casein kinase phospho rylation, a single consensus sequence for tyrosine phos- The proteins were aligned as previously described. 30 The conserved am ino acid residues which anchor the ® ve previously described subdom ains 24 are show n in capitals below the alignm ent, with additiona l conserved residues m arked with an asterisk.
from hum an fetal brain cDN A libraries. 14 . Comparison of the cDN A sequences obtained for lip and p115-RhoGEF shows alternative 59 untranslated sequences, sequence variations resulting in six isolated am ino acid substitutions in the predicted protein sequence and an alternative 39 end. The 59 sequence identi® ed for p115-RhoEF is a shortened version of that identi® ed in clone 3B derived from an HT 1080 cDN A library, and clone 3B extends this sequence by 26 bp. The alternative 59 sequence identi® ed in clone 11A and w ith 59 RACE analysis using RN A from a hum an breast cell line contains an in-fram e upstream m ethionine and m ay represent an alternative start site for translation initiation (Fig. 2b) W ith the exception of the six am ino acid substitu-

Analysis of the lip gene in the prim ary liposarcom a and in transfectant cell lines
H ybridization of clone 11A to Southern blots of EcoRI-digested norm al genom ic D N A gave bands of 15 and 7 kb. T hese bands w ere present in tum our D N A from the original liposarcom a, but bands of 17 and 3.7 kb were observed using D N A derived from prim ary and secondary transfectant cell lines (data not shown). T his data indicated that the lip gene m ight be activated by rearrangement, but if this w ere the case then the rearrangement had taken place during the transfection process and was not present in the prim ary liposarcoma. To further de-lineate any role lip gene rearrangement m ight play in sarcom a developm ent, an additional 52 prim ary tum ours were evaluated by Southern analysis of EcoRI-digested tum our D N A. This group included 12 m alignant ® brous histiocytomas, 10 leiom yosarcom as, nine liposarcom as, ® ve m alignant peripheral nerve sheath tumours, four rhabdom yosarcom as, two synovial sarcorm as, one chondrosarcoma, one ® brosarcom a, one post-irradiation spindle cell sarcom a, one derm ato® brosarcom a protuberans, one ® bromatosis, two haem angiomas, one lipom a, one neurilem m om a and one clear cell carcinom a. In no case was any rearrangement detected.

A nalysis of the mechanism of lip activation
T o exam ine the m echanism of activation of lip, we prepared an oligo-dT prim ed cDN A library using RN A from the lip transfectant cell line. Screening 200 000 clones using clone 11A as a probe resulted in the identi® cation of 10 partial length cD N A clones (T1 to T10) that were characterized by D N A sequencing. Clone T 4 exhibited 88% D N A sequence identity and 98% protein sequence identity to the hum an lip sequence, and probably corresponds to a m ouse lip clone. Analysis of the rem aining clones revealed that they could be divided into three groups, each of w hich exhibited loss of the 39 lip sequences. In clones T 1, 5, 6, 9 and 10, lip was replaced 39 to nucleotide 2387 by a new sequence that in database searches show ed regions of hom ology to hum an alu repeat sequences. This rearrangem ent replaces 188 am ino acids at the carboxy term inus of the predicted lip protein w ith 15 novel am ino acids (Figs 1(b) and 2(c)). This result was consistent w ith Southern analysis data showing that probes prepared from 39 fragm ents of the norm al lip cD N A failed to detect human sequences in lip prim ary and secondary transfectants (results not shown). At position 1747 there was a G ® A base change converting an alanine in lip to a threonine in the transfectant lip clone at codon 566. In addition, T 5 which was the longest of all the clones, showed an A ® G base change at position 1481 substituting an arginine in lip for a histidine in the activated lip clone at codon 477. The presence of these alterations, which lie in the D H dom ain, w as con® rm ed by sequencing a PC R product generated from reverse-transcribed lip transfectant RN A, using prim ers¯anking this region. The second group of cDN As was represented by a single clone, T3, w hich also encoded the G ® A alteration at codon 1747. In T 3, lip sequence is interrupted im m ediately after nucleotide 2300 by 138 bp of novel sequence, followed by an 87-bp sequence corresponding to nucleotides 2301± 2387 of lip and then by 280 bp of additional novel sequence and a polyA tail. T he novel sequences in T 3 w ere unrelated to those in the ® rst group of clones discussed above. In T3, lip is truncated by 216 am ino acids as a stop codon is introduced im m ediately following the break-point (Figs 1(b) and 2(d)). T his alteration removes an invarian t tryptophan from the PH dom ain discussed earlier. Analysis of the junctions between lip sequences and novel sequences in clone T 3 suggests that the 138 bp of the novel sequence inserted between nucleotides 2300 and 2301 m ay be a retained intron ( Fig. 1(b)). The third group of cD N A clones represented by clones T2 and T 7 w ere identical to clone T3 except that their 39 ends were respectively 133 and 136 bp upstream from the start of the polyA tail in T 3.

D iscussion
H ere, w e describe the cloning and characterization of a gene detected in N IH 3T3 transfection experim ents using D N A from a hum an liposarcom a. This gene, lip, shares hom ology with the oncogenes dbl, vav, ect-2, tim and lbc, and the yeast cell cycle gene cdc24, encompassing both the D H and the PH dom ains. C D C24 functions as an exchange factor for the RH O-like protein C D C42 in budding yeast, 27 D BL is known to act as an exchange factor for C D C42Hs and RHO A, 28 EC T-2 binds RHO C and RAC1 11 and LBC acts as an exchange factor for RHO A, RHO B and RH OC . 29 M oreover, w ith the exception of six isolated am ino acid differences, lip is identical over its N -term inal 830 am ino acids (including D H and PH domains) to the recently identi® ed protein p115RhoGEF , which functions as an exchange factor for RHO A. 14 Guanine nucleotide exchange factors bind preferentially to the nucleotide-depleted state of G-pro teins, and by stim ulating the release of G D P they prom ote the subsequent binding of G TP, and hence G -protein activation. T he D H dom ain appears to function prim arily as an G D P± GT P exchange dom ain for m embers of the RHO fam ily. The PH dom ain 30 appears to be involved in a wide variety of m olecular interactions. T he C -term inal regions of several PH dom ains bind to the beta-gam m a subunits of heterotrim eric Gproteins, 31,32 and the N -term inal region to phosph oinositol-4,5-bisph osphate, 33 which im plies that the PH dom ain is im portant for m embrane localization. In all the exchange factors which possess both dom ains, the PH domain is located im m ediately C-term inal to the D H dom ain. T his suggests that the PH dom ain is im portant for the function of the exchange factor dom ain.
In transfectants, lip appears to be activated by C -term inal truncation, a rearrangem ent which appears to have taken place during the transfection procedure, and was not present in the prim ary liposarcom a. Moreover, exam ination of an additional 52 tumours by Southern analysis revealed no evidence in support of a role for lip rearrangement in sarcom a developm ent. Support for C -term inal truncation as the m echanism of lip activation com es from evidence that pll5RhoGEF is activated for transfo rm ation in the N IH3T 3 focus-form ing assay by both N -and C-term inal truncation. l4 M oreover, additional mem bers of this protein fam ily have also been show n to be activated by truncation. 8,11,34± 36 In general, these activating m utations app ear to involve the rem oval of putative regulatory C -term inal and/or N -terminal dom ains, but leave the D H and PH dom ains intact. 37 In the oncogene lfc, m utation of the conserved tryptophan residue in the PH domain abolishes transfo rm ing activity as does rem oval of only three am ino acid residues from the N -terminal region of its D H dom ain. D eletions w hich do not involve the D H or PH dom ains do not abolish transform ing activity. In addition, replacement of the PH dom ain with an alternative m embrane-localizing signal such as an isoprenylation or m yristoylation site restores transform ing activity. 37 T his supports the view that the PH dom ain is required for D H dom ain activity, which appears to be prom oted by localizing the exchange factor to the plasm a m embrane.
lip is predicted to encode a protein with a number of differences from p115RhoG EF. In addition to differences in six amino acid residues within the N-terminal 830 amino acids. the C-terminal 133 amino acids are unique to lip. W ithin this C-terminal region there are a number of proline-rich regions which conform to the minimal consensus sequence for SH3 domain binding sites. 26 SH3 domain-binding sites have been identi® ed in a number of proteins including, for example, the mammalian RAS exchange factor mSOS. 38 In SOS, these proline-rich sequences mediate binding to the SH3 domains of the adaptor protein GRB2, 39,40 an interaction which recruits SOS to the cell membrane, where it is pivotal in the signal transduction pathway from receptor tyrosine kinases to RA S proteins. LIP and p115RhoG EF may represent alternatively spliced forms of the same gene, and if the proline-rich sequences present in LIP serve a regulatory function, perhaps via a role as SH3 domain-binding sites, then this alternative splicing may represent a mechanism by which LIP/ p115RHOGEF function is regulated.

A cknowledgem ents
T he M 426 ® broblast cDN A library w as kindly provided by D r S. Aaronson. This work is supported by grants from the Cancer Research C am paign and the M edical Research C ouncil.