Functional Identification of LZTS1 As a Candidate Prostate Tumor

Functional identi®cation of LZTS1 as a candidate prostate tumor suppressor gene on human chromosome 8p22

Yofre Cabeza-Arvelaiz1, Jorge L Sepulveda2, Russell M Lebovitz2, Timothy C Thompson3 and A Craig Chinault*,1

1Department of Molecular and Human Genetics, Baylor College of Medicine, Houston, Texas, TX 77030, USA; 2Department of Pathology, Baylor College of Medicine, Houston, Texas, TX 77030, USA; 3Department of Urology, Baylor College of Medicine, Houston, Texas, TX 77030, USA

Deletions in the 8p21-22 region of the human genome are responsiveness of tumor cells to growth controls among the most common genetic alterations in prostate (Harris et al., 1969). Subsequently, the technique of carcinomas. Several studies in dierent tumor tissues, microcell fusion has been widely used to achieve including prostate, indicate that there are probably reversion of the malignant phenotype of certain tumor multiple tumor suppressor genes (TSGs) present in this cells by introducing single normal human chromo- region. To identify candidate TSGs on 8p22 a YAC somes or subchromosomal fragments (Goyette et al., contig spanning this region was assembled and YAC 1992; Ichikawa et al., 1994; Murakami et al., 1996). clones retro®tted with a selectable marker (neo) were Even tumors that have arisen through multiple genetic transferred into rat prostate AT6.2 cells. Two over- alterations on dierent chromosomes have been lapping YAC clones showed greatly reduced colony- reverted to a less tumorigenic phenotype, indicating forming eciency, indicating they may carry a TSG. that introduction of a single TSG may be sucient to Two BAC clones encompassing the overlapping region suppress tumor growth. More recently, functional also appeared to exert suppressive eects on the growth complementation using YAC clones has been used of AT6.2 cells. Database searches for genes mapped to successfully to further localize putative TSGs to the critical region identi®ed a gene known as FEZ1 individual YACs, which are more amenable to (LZTS1) as a potential candidate suppressor gene. positional cloning approaches (Koreth et al., 1999; Subsequent experiments showed that over-expression of Murakami et al., 1998; Reddy et al., 2000). LZTS1 cDNA inhibited stable colony-forming ecien- Loss of heterozygosity (LOH) and homozygous cies of AT6.2, HEK-293 and LNCaP cells. In contrast, deletions at the human chromosome 8p region are LZTS1-transfected Rat-1 and RM1 cells were growth- among the most common genetic alterations in stimulated. Database searches also identi®ed additional prostate adenocarcinomas and other cancer tumors isoforms of the LZTS1 mRNA, as well as LZTS1 (Bova et al., 1996; Farrington et al., 1996; Kagan et al., protein domains reminiscent of those found in transcrip- 1995). These studies suggest that at least two dierent tion factors. Together these data suggest that the TSGs reside in the 8p21-22 region. Transfer into cancer LZTS1 gene is involved in the regulation of cell growth cells has provided further evidence that this chromo- and its loss of function may contribute to the somal region harbors genes that suppress tumorigeni- development of prostatic carcinomas, as well as other city (Ichikawa et al., 1994; Murakami et al., 1996). cancers. Oncogene (2001) 20, 4169 ± 4179. Positional cloning methods have led to the isolation of at least three genes localized to 8p21-22, PDGFRL Keywords: prostate cancer; tumor suppressor; human (also known as PRLTS; Fujiwara et al., 1995), NKX3.1 chromosome 8p; FEZ1; LZTS1 (Bhatia-Gaur et al., 1999) and N33 (MacGrogan et al., 1996), that have been postulated as candidate cancer genes. More recently, another gene from 8p22, referred Introduction to as FEZ1 (F37/Esophageal cancer-related gene- coding leucine-zipper motif), has been isolated and One of the most intensely studied areas of cancer found to be altered in several tumors including research in recent years has been that of tumor esophageal, breast and prostate (Ishii et al., 1999). suppressor genes (Anderson and Stanbridge, 1993). The HUGO Human Gene Nomenclature committee Fusion of normal and malignant cells aorded the ®rst has now designated this locus as the LZTS1 (leucine evidence that the normal genome may restore the zipper, putative tumor suppressor 1) gene. We have used YAC-based and BAC-based transfection approaches to narrow down the regions carrying genes that decrease stable colony-forming eciency of *Correspondence: AC Chinault; E-mail: [email protected] Received 28 November 2000; revised 9 April 2001; accepted 12 tumor cells or aect their growth properties. Using this April 2001 approach a putative TSG locus was mapped to Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4170 individual YAC and BAC clones in the 8p22 region. using overgo probes designed from sequences of a set Subsequently, database searches with STS and EST of 10 STS and EST markers that were previously markers identi®ed LZTS1 as a candidate TSG located mapped into this region; at least 20 BACs that mapped inside the critical region. Subsequent analysis of the to this region were identi®ed. The three BACs shown eects of over-expression of LZTS1 on colony in Figure 1c were selected to study eects on growth formation and cell growth suggests that LZTS1 is, in properties of AT6.2 cells. Because most BAC libraries fact, the gene responsible for the observed growth have been constructed in vectors that do not carry a inhibitory eects on rat AT6.2 cells and two human mammalian selectable marker, transfection and selec- cell lines. In contrast, this gene stimulated growth of tion of stable clones typically requires time-consuming some other cell lines studied. clone modi®cation. Instead, we used a simpler approach to rapidly identify BACs that exert suppressive eects on AT6.2 cells. BACs were co-transfected with a GFP reporter BAC of similar size into the Results AT6.2 cells using a method that allows the ecient delivery of large DNA molecules into cells through the YACs and BACs from 8p22 reduce colony-forming use of psoralen-inactivated adenoviral particles and efficiency of AT6.2 cells PEI (Baker and Cotten, 1997). The cell growth Using chromosome 8p STS markers, we constructed an suppressive eects and/or cytotoxic eects of these extensive contig of more than 80 YACs encompassing BACs on AT6.2 cells were assessed by simply the whole 8p22 region, as described in Materials and monitoring the decrease in GFP+ cells for a period methods. An overlapping set of nine YACs, indicated of 5 days post-transfection, as previously described in Figure 1a, was retro®tted with a neo selectable (Zeng et al., 1997). Representative ®elds observed on marker using the pRAN4 vector and these clones were days 1 and 5 following BAC transfection are shown in proven to be correctly targeted by gel electrophoresis Figure 1d. Two of the BACs (CTI-B 353G1 and CTI-B and PCR analyses (data not shown). Seven of these 316F10) appeared to exert eects on AT6.2 cells that YAC clones were successfully used for transfection by were similar to those observed with the two over- fusion to AT6.2 rat prostate tumor cells. After lapping YACs. These results allowed us to further selection in G418, numerous discrete colonies were minimize the region containing the putative TSG to observed in some of the plates from each of the YAC approximately 160 Kb. Together the results from the transfections. Generally, the frequency of G418r transfection experiments with YACs and BACs from colonies in these plates was similar to that observed the 8p22 critical region indicated that this region by others (Pachnis et al., 1990). However, a substantial harbors a gene that negatively regulates AT6.2 cell dierence between two overlapping YACs (911G7 and growth. 739B2) and the other ®ve YACs was observed in the number of colonies on the plates. This observation Identification of LZTS1 as a potential TSG in the YAC prompted us to assess the frequency of introduction of overlap region YAC sequences into AT6.2 cells by categorizing the plates according to the number and size of the G418r To identify the gene(s) responsible for the cell growth colonies in each plate. The results, shown in Figure 1b, inhibitory eect, we compiled sequencing and mapping indicated that the plates with very few small colonies data from several sources and constructed an inte- or no colonies comprised more than 90% of the total grated map of this minimal 8p22 subregion, as shown number of plates for the 911G7 YAC fusion clones and in Figure 2a. The locations of each marker as well as around 65% for the 739B2 YAC fusion clones. In fact, the YAC and BAC clones of interest within this as shown in Figure 1b, the characteristics of the sequence stretch are indicated, approximately to scale, colonies observed after transfection with these two in this map. Four genes were identi®ed after perform- YACs resembled those observed using YAC 204E7, ing recursive BLAST analysis searches of the GenBank which contains the TP53 gene. With this clone, which non-redundant database with the markers that mapped was used as a TSG control, a majority of the plates not to the BAC inserts: the lipoprotein lipase (LPL) gene, only showed reduction in the number of colonies the ATPase, H+ transporting, beta polypeptide, produced, but, also, the colonies that did form were isoform 2 (ATP6B2) gene, the solute carrier family 18 signi®cantly smaller (530 cells per colony). In (vesicular monoamine) (SLC18A1 or VMAT) gene and contrast, plates with a relatively large number of the F37/Esophageal cancer related gene-coding leucine colonies comprised more than 80% of the total number zipper motif (FEZ1) gene, now known as LZTS1. of plates for the other YAC fusion clones. Most Based on the published pattern of expression and the surviving clones derived from colonies transformed function (when known) of these genes, all except with YAC 911G7 showed little increase in cell number LZTS1 were excluded from further study as potential after a week, whereas clones transformed with other TSGs. Even though the function of the LZTS1 gene is YACs showed at least a ®vefold expansion (not unknown, it was previously identi®ed as a candidate shown). TSG on the basis of its dierential expression in tumor Next, a BAC contig encompassing the overlapping and normal tissues in several types of cancers (Ishii et region of the 911G7 and 739B2 YACs was constructed al., 1999). It was isolated originally from a YAC

Oncogene Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4171 (859A7) that overlaps with YACs 911G7 and 739B2. Effects of LZTS1 over-expression on AT6.2 and Its presence in the latter clones, as well as in the BAC HEK-293 cells clones that suppressed AT6.2 growth, was con®rmed by exon ampli®cation analyses, as shown in Figure 2b. To examine whether the LZTS1 gene could be LZTS1 was therefore selected for further studies. responsible for the reduction in cell growth observed

Figure 1 (a) YAC contig encompassing the human 8p22 chromosomal region. Nine YACs span the 8p22 region frequently deleted in prostate tumors located between markers D8S549 and D8S258. All clones contain markers indicated on the arbitrary scale at the top; approximate sizes of the YACs in kilobases are shown in parentheses. The orientation of the contig relative to the location of the telomere (TEL) and centromere (CEN) on 8p is indicated. (b) Eect of 8p22 YACs on the colony-forming eciency of AT6.2 cells following transfection. The percentage of plates with dierent numbers of G418r colonies is plotted for fusion clones that have been transfected successfully with seven of the neo-retro®tted 8p22 YACs from a. A fusion clone transfected with YAC 204E7 containing TP53 was included in this study as a positive cell growth suppressing gene control. (c) Physical mapping of the minimal overlap region of YACs 911G7 and 739B2. Three BACs that span the region of overlap located between markers WI-5962 and StSG30884 are shown. Approximate minimal sizes of the BACs in kilobases are shown in parentheses. (d) Growth suppressive eects of 8p22 BACs on AT6.2 cells. Each of the BACs shown above was delivered into AT6.2 cells along with a BAC of similar size containing the reporter green ¯uorescent protein (GFP) and transfectants were monitored by ¯uorescent microscopy on days 1 and 5. Representative ®elds from each transfectant plate are shown with the 8p22 BAC that was introduced indicated on top

Oncogene Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4172 following introduction of YACs and BACs into the AT6.2 cells, the complete cDNA and a truncated version (LZTS1-5') were cloned into the mammalian expression vector pcDNA-3.1. These constructs, together with the pEGFPN1 reporter vector, were transfected into the rat prostate cancer cell line AT6.2 and E1-transformed HEK-293 cells. After selection in antibiotic for 3 weeks, a large number of colonies was observed in plates containing cells transfected with the control construct LZTS1-5' plus the reporter vector, or with the reporter vector alone. In contrast, very few colonies were observed in plates transfected with the complete LZTS1 gene (Figure 3a,b). To further verify these results, cell proliferation rates of AT6.2 transfectants were determined. Colo- nies from each transfectant were collected and seeded in 96-well or 24-well plates at a density of 103 or 26103 cells/well without G418. After cells were allowed to attach for 16 h, growth rates were determined by counting cells in triplicate wells every 24 h for 4 days. The results indicated that the cell growth rate in culture was dramatically decreased in the LZTS1 transfectants compared to the controls (Figure 3c,d). Cell viability was usually 495% in the control wells, but much lower in the LZTS1 transfectant wells, although the latter could not be measured accurately because of extensive cell lysis (data not shown). Because few colonies of the LZTS1- transfected HEK-293 cells grew beyond 50 cells per colony, proliferation assays on these cells were not performed.

Effects of LZTS1 expression under different promoters on HEK-293, Rat-1 and RM1 cells The eects of LZTS1 over-expression were further examined with three cell lines using constructs where transcription was under the control of two dierent promoters. A pBabeHygro/LZTS1 construct was made by cloning the complete LZTS1 cDNA into the pBabeHygro expression vector (Morgenstern and Land, 1990), where expression is driven by the Figure 2 (a) Precise mapping of the location on human retroviral gag promoter and hygromycin resistance is chromosome 8p22 of the YAC and BAC clones that exhibited used as a selectable marker. Transfection of HEK-293 growth-suppressive activity. An ideogram of human chromosome 8 is shown, along with a STS marker map for progressively cells with either this construct or the pCMVNeo/ expanded subregions of &7 Mb and 2 Mb, respectively, contain- LZTS1 construct described earlier substantially re- ing the LZTS1 gene. The locations of the YAC and BAC clones duced colony-forming eciency (Figure 4a,b) com- indicated by vertical bars were inferred from STSs contained pared to the truncated control constructs. In contrast, within these clones. Approximate distances from the p-telomere are indicated in megabases (Mb), based on the scale of the the colony-forming eciencies of normal Rat-1 Human Genome Browser website (http://genome.ucsc.edu). The ®broblasts and mouse prostate RM1 cells were not approximate locations of four known genes (LPL, LZTS1, inhibited (Figure 4c,d and e,f, respectively). In fact, SLC18A1, and ATP6B2) that have been mapped to the minimal paradoxically, LZTS1 appears to stimulate cellular critical subregion encompassed by the BACs are shown. However, proliferation in these two cell lines, as indicated by the some gaps still exist between and within the two sequence contigs spanning this subregion, and thus the distance between some of considerably larger colonies observed in LZTS1 the markers and genes may be inexact. (b) LZTS1 exon PCR analysis on 8p22 YAC and BAC clones. PCR analysis was performed with DNA prepared from each of the following CTI- BAC (B) clones and YAC (Y) clones: 1, B353G1; 2, B316F10; 3, B353G12; 4, Y911G7; and 5, Y739B2. The primer sets (described in Materials and methods) amplify regions of 0.477, 0.897 and Lane M contains DNA molecular size markers (100 bp ladder); 0.690 Kb from the LZTS1 gene exons 1, 2, and 3, respectively. sizes are indicated in Kb to the right

Oncogene Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4173

Figure 3 Suppression of colony formation after transfection of LZTS1 into AT6.2 (a) and HEK-293 (b) cells. Following co- transfection with the indicated LZTS1 constructs and control plasmids (LZTS1=pCNeoLZTS1, LZTS1-5'=pCNeoLZTS1-5' and GFP=pEGFPN1) and G418 selection, colonies were counted. The number of colonies observed after transfection with the control plasmid pEGFPN1 was considered to equal 100%. For the other experiments, the relative numbers of colonies on each plate from two independent transfections are shown with bars representing the standard error. (c) Growth curves of stably-transfected AT6.2 cells. Growth rates in media without G418 selection were determined for three AT6.2 clones stably-transfected with either the wild- type LZTS1, the truncated LZTS1-5' control, or the pEGFPN1 control plasmid. Each curve represents daily cell counts for 4 days from triplicate wells and error bars represent the standard error. (d) Phase-contrast photomicrographs of representative ®elds from two of the transfectants from (c) to illustrate the observed cell densities at the days indicated

transfectant plates. The higher number of colonies in Different effects of LZTS1 over-expression on human these plates compared to control transfectant plates prostate cell lines could be due in part to this increased cell proliferation, which may have led to the formation of satellite To ascertain whether or not these apparently con¯ict- colonies. Actually, the large size of these colonies in ing eects of LZTS1 on cell growth correlated with the some plates made it dicult to discriminate between status of endogenous expression, we compared the the boundaries of the colonies after 3 weeks of eect of LZTS1 stable transfection on colony-forming selection. eciency in the human prostate cancer cell lines

Oncogene Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4174

Figure 4 Eect of over-expression of LZTS1 by viral promoters in dierent cell lines. The expression of LZTS1 mRNA in the pBABEHygro and pCDNA3.1 vectors is driven by the retroviral gag and the cytomegalovirus promoters, respectively. After co- transfection with the indicated LZTS1 constructs (pBH=pBABEHygro; pBHLZTS1=pBABEHygroLZTS1; pCNLZTS1- 5'=pCMVNeoLZTS1-5', and pCNLZTS1=pCMVNeoLZTS1), and reporter plasmid (pEGFPN1) the transfectants were selected in hygromycin or G418 and colonies were counted. (a), (c) and (e) show suppression of colony formation by LZTS1 over-expression with HEK-293 cells, Rat-1 cells and RM1 cells, respectively. The number of colonies observed in cultures transfected with the control plasmids pBH or pCNLZTS1-5' was set to 100% for normalization purposes. Results from two independent transfections

Oncogene Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4175 TSUPr1 and LNCaP. The latter cell line has been mapped to this region. The growth suppression shown by Northern blot analysis not to express the full potential of YACs was ®rst assessed after their transfer length LZTS1 mRNA (Ishii et al., 1999). The results of into AT6.2 cells. Two overlapping YACs that reduced these experiments are shown in Figure 5a. It is obvious the colony-forming eciency of the host cells to an that forced expression of the LZTS1 gene strongly extent that was comparable to that attained with a abrogated colony formation by the LNCaP cells, TP53 YAC were identi®ed. By combining two whereas in the TSUPr1 cells LZTS1 over-expression published methods (Baker and Cotten, 1997; Zeng et slightly increased colony-forming eciency compared al., 1997), we then screened BAC molecules covering to the control. However, when the endogenous LZTS1 the YAC overlap region by monitoring the division expression in these cell lines was checked by RT ± PCR (and disappearance) of living cells over time as the levels of expression of this gene in both cell lines indicated by the number of GFP+ cells. Two were comparable; in fact, the only noticeable dierence candidate BACs that appeared to contain the gene between these cell lines appeared to be in the with growth suppressive activity were identi®ed. Precise expression of the minor isoforms observed in the mapping of the location of the STS and EST markers TSUPr1 cells, as shown in Figure 5b. contained within these BACs (Figure 2a) led to the identi®cation of the LZTS1 gene, previously known as FEZ1 (Ishii et al., 1999), as potentially responsible for Discussion the observed growth inhibitory eects of these clones on AT6.2 cell. Ectopic expression of LZTS1 cDNA in LOH analyses of prostate tumors suggest that the AT6.2 cells resulted in inhibition of colony-forming 8p21 ± 22 region of the human genome harbors mul- eciency and a signi®cantly decreased growth rate. tiple TSGs. In this study, we report the identi®cation These results provide the ®rst direct functional evidence of LZTS1 as a cell growth suppressing gene from the that the LZTS1 gene is involved in the regulation of 8p22 region using a functional approach with YAC cell growth and may function as a bona ®de TSG. and BAC transfection assays to progressively narrow Transfection of LZTS1 into the HEK-293 and down the critical region and identify potential TSGs LNCaP cells further con®rmed this gene's capability to negatively regulate cell growth. In the adenovirus- transformed HEK-293 cell line, E1B inhibits p53 activity by complex formation. Disruption of the p53/ E1B complexes releases active p53, which induces cell cycle arrest (Hutton et al., 2000). Thus, it is conceivable that ectopic expression of LZTS1 in HEK-293 cells could disrupt the sequestration of p53 by E1B and lead to growth arrest and/or apoptosis. In fact, there were some observations suggesting that LZTS1 over-expression may promote apoptosis in certain cells. For example, we observed an increased number of cells with sub-diploid DNA content in ¯ow- cytometric analysis of LZTS1-transformed AT6.2 cells Figure 5 Colony-forming eciency analysis after transfection of (not shown). It also appeared that the decrease in the the LZTS1 gene into human prostate cancer cell lines TSUPr1 and LNCaP. (a) Following transfection of the indicated cell lines number of GFP+ cells transfected with LZTS1 with either the LZTS1 (pCNeoLZTS1), or control plasmid (Figure 1d) was likely due to extensive cell loss, rather (pCNeoKLC) and G418 selection, colonies were counted. The than segregation of the GFP vector during cell number of colonies observed in cultures transfected with the proliferation. control plasmid pCNeoKLC was considered to equal 100%. The relative number of colonies on each plate from two independent Paradoxically, LZTS1 signi®cantly increased the transfections are shown and error bars represent the standard colony-forming eciency of Rat-1 and RM1 cells and error. (b) RT ± PCR analysis of LZTS1 mRNA expression in to a lower extent that of TSUPr1 cells. This ambiguous human prostate cancer cell lines. An ethidium bromide stained gel eect could be the result of inactivation of the with ampli®ed fragments from RT ± PCR on total RNA from downstream response to p53 or LZTS1 in these human prostate cancer cell lines TSUPr1 and LNCaP is shown. Products were produced by speci®c primers ¯anking the exon 3 particular cells. For example, such an eect is seen in coding sequence of the LZTS1 mRNA. In both cell lines Rat-1 cells as a result of promoter hypermethylation expression of transcripts containing the complete exon which prevents expression of p21WAF1 (Allan et al., (0.695 Kb) were detected, as indicated by an arrow. Other bands 2000), an eector of p53 apoptotic activity. If p21WAF1 (indicated by arrowheads), which presumably represent minor isoforms, were detected in the TSUPr1 cells only. Lane M also mediates the LZTS1 eect, its absence in these contained the DNA molecular size markers used (100 bp ladder); cells may avert the LZTS1 growth inhibitory eect. sizes are indicated in Kb The (ras+myc)-transformed RM1 cells have been

are shown and bars represent the standard error. (b), (d) and (f) show phase-contrast photomicrographs of a representative ®eld, observed by visible or ¯uorescent microscopy, of representative transfectants from experiments with HEK-293 cells, Rat-1 cells and RM1 cells, respectively, to illustrate colony sizes and morphologies

Oncogene Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4176 activity and cell DNA content by ¯ow cytometry did not produce conclusive evidence implicating this protein in G1 arrest (not shown). However, these transient assays were complicated by the presence of a signi®cant number of untransfected cells in these transient assays. To circumvent these problems we are establishing the system for tetracycline-regulated expression of LZTS1 (Gossen and Bujard, 1992). Analyses of the LZTS1 gene and its protein sequence detected features that resemble those of Figure 6 The upper part shows the organization of the LZTS1 transcription factors, which make it tempting to gene and the location of predicted protein motifs. The coding and conjecture a possible role for LZTS1 in gene non-coding sequences (represented by ®lled and open boxes, transcription modulation. For example, the overall respectively) of the LZTS1 cDNA (GenBank accession structural topography of LZTS1 (Figure 6) is reminis- #AF123659) were aligned to the LZTS1 gene sequence (accession #AF123653). The location and relative size of exons and introns cent of the structure of the bZip transcription factors (shown as solid lines) are shown to scale. The locations of CREB/ATF and CREM (Courey and Tjian, 1988; predicted functional domains in the encoded protein, including a Foulkes and Sassone-Corsi, 1996; Habener, 1990; phosphorylation box (P-Box), leucine zippers (LZ), and gluta- Molina et al., 1993; Pirrotta et al., 1987). Although mine-rich (Q) domains, are indicated with arrows. In the middle we detected no recognizable DNA binding motifs in are shown schematic representations of dierent LZTS1 isoforms and aberrant transcripts in the GenBank human EST database. the LZTS1 protein, it should be noted that some At least seven of the 29 LZTS1 EST containing coding sequences transcription modulators do not bind to DNA directly, in the database were alternatively spliced at cryptic splice sites, but can still in¯uence transcription by complex and three maintained co-linearity with genomic DNA across stabilization (Tsai et al., 1987). intronic regions. The structure and location of these presumptive isoforms and aberrant transcripts are depicted as black or hatched Several transcription factor genes also achieve boxes (transcribed exon or intron sequences, respectively) and multitasking through alternative splicing (Chiu et al., dashed lines (spliced intronic sequences). The types (A, B, and C) 1989; Courey and Tjian, 1988; Foulkes and Sassone- and GenBank accession numbers are indicated on the left and Corsi, 1996; Lucibello et al., 1989). For example, sub- right of each EST, respectively. The bottom part of the ®gure stoichiometric levels of the repressor isoform S- shows an expanded representation of six dierent LZTS1 isoforms previously reported as aberrant transcripts by Ishii et CREMb strongly antagonize the activation by the al. (1999). The location and relative sizes of exons and introns are complete CREM protein (Laoide et al., 1993). The shown to scale; spliced regions are indicated by dashed lines. The LZTS1 gene also appears to produce multiple GenBank accession numbers are shown on the right. The scale at transcripts (Figure 6) that could potentially play a role the bottom of the diagram indicates the approximate positions of the nucleotide residues in the LZTS1 gene in its mechanism of action by antagonizing the activity of the major isoform. BLAST searches against the human EST database identi®ed seven isoforms (or aberrant transcripts) which we grouped into three types shown to be unresponsive to the growth-inhibitory (A, B, and C), according to the location of the eects of endogenous and exogenous wild-type p53, dierentially spliced region (Figure 6). Type A which indicates that the downstream response to p53 transcripts are apparently generated through selective has been inactivated (Lu et al., 1992). It would be of splicing at non-canonical splice sites ¯anking the interest to investigate the downstream activation of sequence stretch coding for the LZ1 motif within exon speci®c p53 target genes such as p21WAF in this cell line. 2. They maintain co-linearity with genomic DNA It is also likely that the LZTS1 eects in these cells are across the intronic region immediately after exon 2, partly counteracted by the presence of myc, which has and may represent a relatively abundant aberrant been shown to antagonize the eect of TP53 on transcript since three (17%) out of a total of 18 ESTs apoptosis and p21WAF1 transactivation in K562 leuke- that encompassed this region were detected. EST mia cells and LNCaP cells (Ceballos et al., 2000; transcript types B and C may represent segments of Mitchell and El-Deiry, 1999). The TSUPr1 cell line previously described isoforms (Ishii et al., 1999), which carries an oncogenic mutation in the ras (V12) gene. are shown to scale at the bottom of Figure 6 to The presence of activated ras in this and other prostate illustrate that they could generate the bands detected in cell lines has recently been shown to abrogate the up- the RT ± PCR analysis of TSUPr1 cells RNA. regulation of p21WAF1 transcription by TGF-b 1 (Park AF123658 and AF12357 would produce 0.650 and et al., 2000). Alternatively, similar functional dualism 0.617 Kb fragments, respectively, by ampli®cation with has been seen with several transcription factors and the exon 3 primers. Similarly, other as yet unchar- oncoproteins (Evan and Littlewood, 1998; Johnson, acterized, isoforms alternatively spliced at exon 3 could 2000; Tuck and Crawford, 1989). Ectopic expression of generate the other observed bands of 0.35 and 0.25 Kb. E2F, myc or E1A is usually accompanied by apoptosis, As shown in Figure 6, in most isoforms the selectively which overwhelms the proliferative potential of these spliced stretches contain the LZ and/or Q-rich motifs. genes (Evan and Littlewood, 1998). This suggests that the degree of neutralization of the Preliminary studies aimed at elucidating LZTS1 endogenously expressed levels of full-length LZTS1 mechanisms by measurement of p53 levels, cdk2 may be determined by the levels of expression of

Oncogene Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4177 isoforms, which could also explain the lack of GFP marker gene were kindly provided by Dr M Cotten inhibitory eects by LZTS1 in TSUPr1 cells (Figure (IMP, Vienna, Austria). 5a,b). In summary, we have identi®ed a functional eect of YAC modification and spheroplast fusion to AT6.2 cells LZTS1 by forced expression of this gene in various cell A homologous recombination approach using the polyethy- lines that results in cell growth suppression and lene glycol (PEG) transformation method (Burgers and possibly apoptosis. The apparent contradictory role Percival, 1987) was used to retro®t 8p22 and control TP53 of LZTS1 in other cell lines could re¯ect the complex YACs with the pRAN4 vector (Markie et al., 1993) interactions between its dierent isoforms with positive containing the aminoglycoside phosphotransferase (neo) gene, and negative regulators of cell growth that determine which confers resistance to neomycin/G418. Prior to use, the ultimate outcome in terms of cell cycle arrest, cell YACs were characterized to determine if they had been growth or apoptosis in each cell type. These results successfully retro®tted with the neo gene by PCR analysis support the notion that since regulation of cell using the following primers: neo-F:ccagagtcccgctcagaagaactc proliferation is intimately connected to the control of and neo-R:cttgctcctgccgagaaagtatcc. To show that they apoptosis, both activities can be aected by the loss of retained the sequences of interest, Southern analysis of PFGE-resolved chromosomes was conducted with a neo TSG function and may be critical factors in prostate DNA probe and labeled total human genomic DNA or TP53 carcinogenesis. cDNA. The neo-retro®tted YACs were fused to AT6.2 cells using previously described protocols for spheroplast fusion (Markie et al., 1993; Pachnis et al., 1990). A total of 1 ± 7 8 Materials and methods 5610 AT6.2 cells was fused to a 20-fold excess (2 ± 10610 ) of spheroplasted yeast cells suspended in STC buer (1 M Cell lines and culture conditions sorbitol, 10 mM Tris-HCl pH 7.5 and 10 mM CaCl2). The fusion mix was resuspended in 4 ml of 50% PEG-4000 AT6.2, a highly metastatic, rat prostate cell line was provided solution (Boehringer Mannheim), 10 mM CaCl2,75mM by Dr J Isaacs (Johns Hopkins University, Baltimore, MD, HEPES, pH 7.4) and after 120 s at room temperature, USA). This cell line has previously been used in microcell- 20 ml of RPMI 1640 medium was added and the cells were mediated chromosome transfer to infer the presence of collected. Cells were resuspended in complete medium and prostate TSGs on human chromosome 8 (Ichikawa et al., plated over 20 100-mm dishes. Selection was performed in 1994). Rat-1, a spontaneously immortalized rat ®broblast cell complete medium plus 600 mg/ml of G418 (Life Technolo- line was obtained from American Type Culture Collection gies); discrete resistant colonies usually appeared after 14 ± 21 (ATCC, Rockville, MD, USA). RM1 is a mouse prostate cell days. Colony formation was appraised by categorizing the line transformed with myc and ras (Baley et al., 1995). The plates, according to the observed number and size of the human embryonic kidney cell line (HEK-293), transformed G418r colonies in each plate, into three groups: (1) Plates with adenovirus type 5 E1 genes, as well as the human with no colonies or very few colonies (0 ± 5), small in size; (2) prostate cancer cell lines LNCaP and TSUPr1 were obtained Plates with a moderate number of colonies (6 ± 50) of varying from ATCC. Cells were maintained in a 5% CO2 environ- size including medium size (30 ± 50 cells) and large size (450 ment at 378C, in complete medium RPMI 1640 or D-MEM cells); (3) Plates with more than 50 colonies of various sizes with 2 mML-glutamine (Life Technologies, Gaithersburg, including semicon¯uent plates. MD, USA) supplemented with 10% fetal bovine serum (Life Technologies) in the presence of 0.4% antibiotic-antimycotic mixture (Life Technologies). BAC preparation and transfection BAC DNA was isolated using a large scale preparation YAC and BAC identification and selection method (Kirschner and Stratakis, 1999). BAC transfection into AT6.2 cells used psoralen inactivated adenovirus YAC clones used in the present study were identi®ed by particles as DNA carrier and polyethyleneimine (PEI, Mol. PCR-based screening (Chinault, 1994) of human genomic wt. 2000, Fluka) as condensing agent. The transfection YAC libraries (Albertsen et al., 1990; Bellanne-Chantelot et complexes were prepared as previously described (Baker al., 1992). An extensive YAC contig was constructed for the and Cotten, 1997), except that 4 mg of each BAC and 2 mgof 8p22 region by combining veri®ed STS content data from the reporter BAC were used for each transfection and the literature (Bookstein et al., 1994; Chumakov et al., 1995) and amount of adenovirus particles used was reduced (26109 supplemental data ®les from the Whitehead Institute/MIT particles/reaction). Cells were kept in serum free medium for Center for Genome Research (htpp://www.genome.wi.mit.e- 6 h and then refed with complete medium. To monitor the du) with additional PCR-based library screening results. A number of GFP+ cells in each plate, at days 1 and 5 post- subset of overlapping YACs giving complete coverage of the transfection the transfectants were examined by ¯uorescent region of interest was chosen for further studies. The 204E7 microscopy using a Nikon Labophot-2 ¯uorescent micro- YAC used as a control contains the human TP53 gene from scope (Nikon Inc., Melville, NY, USA), and pictures were chromosome 17. A human BAC library (CTI-B; Research taken using a special ®lter set to visualize GFP (CHROMA Genetics, Huntsville, AL, USA) was screened by ®lter Technology Corp., Battleboro, VT, USA). hybridization with overgo probes. To verify that they contained the region of overlap of the two 8p22 YACs (911G7 and 739B2), clones identi®ed by positive ®lter LZTS1 RT ± PCR and exon amplification hybridization were screened by whole cell PCR for the For reverse transcription (RT)-PCR, 100 ng of normal following 8p22 markers: WI-5962, WI-4688, WI-5355, WI- human prostate or testis poly(A)+ RNA (Clontech, Palo 4873, WI-9031, LPL, D8S1715, D8S258, stSG10118, WI- Alto, CA, USA) or 2 mg of total RNA was used to synthesize 9078, and stSG30884. BAC reporter constructs containing a ®rst-strand cDNA using the SuperScriptTM preampli®cation

Oncogene Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4178 system (Life Technologies) with an oligo(dT) primer. For 100 mm plates, cells were transfected with 5 mgof Transcripts were ampli®ed by thermal cycling for 35 cycles plasmid DNA, using 15 ml lipofectamine reagent. For the using the ELONGASETM kit (Life Technologies), with the study of stable colony formation G-418 (600 mg/ml) or forward (F) and reverse (R) primer set (5'?3'): LZTS1F3 hygromycin (200 mg/ml) selection was started 48 h post- (gtgggaggtgtgtgccagaagtc) and LZTS1R3 (gatatcgccaggtcccc- transfection and continued for 2 weeks, changing the medium agac). Ampli®cation of LZTS1 exon 1 (0.48 Kb) or exon 2 every other day until colonies were counted. (0.9 Kb) from genomic DNA was performed using the Cell proliferation assays were performed with AT6.2 following forward (F) and reverse (R) primers (5'?3'): transfectants. Colonies from each transfectant were collected LZTS1F1: ccctcacggagccacgactgc, LZTS1R1: tcctgcgtttccaa- (LZTS1 transfectant colonies were pooled to have enough cccactt; LZTS1F2: gaaatgggctccgagaagggtg, LZTS1R2: cacg- cells to perform the assay) and cells were seeded in 96-well or ctattggccagcaccag. Ampli®cation of exon 3 (0.7 kb) was done 24-well plates at a density of 103 or 26103 cells/well without with the LZTS1F3 and R3 primers used above. G418 and allowed to attach for 16 h. Triplicate wells from each control and LZTS1-transfected cells were counted at this point to assess the plating eciency of each clone. Expression and control plasmids, transfections and cell Although slightly lower than the number originally plated the proliferation assays number of cells was quite similar in all of the wells counted. The pCNeoLZTS1 and pCNeoLZTS1-5' plasmids contain Cell viability was determined by Trypan blue exclusion to be the wild-type complete FEZ1(LZTS1) cDNA coding region 93 ± 95% in all wells. Triplicate wells from each clone were (obtained from Dr C Croce, Kimmel Institute for Cancer counted every 24 h for 4 days to determine the proliferation Research, Philadelphia, PA, USA) and a truncated LZTS1 rate of each transfectant. cDNA (missing exon 3, which encodes the leucine zipper domain), respectively, cloned into the pcDNA-3.1 vector Computer-assisted homology and motif searches and alignments (Invitrogen, San Diego, CA, USA). The pBHygroLZTS1 plasmid contains the complete LZTS1 cDNA cloned into the BLAST analysis searches were performed with sequences pBABEHygro expression vector (Morgenstern and Land, from the EST and STS markers (from the Unigene database: 1990). The expression of the inserted gene in the pcDNA-3.1 htpp://www.ncbi.nlm.gov/Unigene/) that map to the 911G7 and pBABE vectors is driven by the human cytomegalovirus YAC and the BAC CTIB-353G1 against the non-redundant (CMV) promoter and by the retroviral gag promoters, (nr) GenBank database (http://www.ncbi.nlm.nih.gov/Gen- respectively. The GFP reporter vector pEGFP-N1 was bank/) to identify sequenced genes, BACs, and ESTs mapped obtained from Clontech. The pCNKLC control plasmid to the critical region. The sequence of the LZTS1 gene contains the kinesin light chain cDNA from plasmid pLBL1 (derived from BAC RPCI-11 353K12 on contig accession (Cabeza-Arvelaiz et al., 1993), cloned into pcDNA3.1. #NT_008015 and gene AF123653) was used to perform Cells were transfected using 3 ml lipofectamine reagent BLASTn search analyses (Altschul et al., 1990) of the human (2 mg/ml) according to the manufacturer's instructions (Life EST database. Technologies) with the indicated amounts of endotoxin-free plasmid DNA, puri®ed with the Endofree kit from QIAGEN (Valencia, CA, USA). For all transient transfections, the cells were incubated in serum free medium for 6 h and then refed Acknowledgments complete medium. Cells were transiently co-transfected with We thank Dr D Markie for providing the vector pRAN4 0.5 mg of reporter pEGFP-N1 plasmid and 2 mg of the and for helpful suggestions, Dr M Cotten for providing the plasmid containing the complete LZTS1 cDNA or the GFP BAC reporter vectors and helpful suggestions, Drs C control plasmids in 35 mm plates. At 16 ± 24 h and on the Croce and H Ishii for the FEZ1 cDNA, Dr L Pastori for fourth day post-transfection cells were visualized with a the adenovirus and Dr D Geng for assistance during the Labophot-2 ¯uorescent microscope (Nikon Inc), and pictures YAC structural analysis. This study was supported by NIH were taken using a GFP ®lter set (CHROMA Technology). SPORE Grant P50-CA58204.

References

Albertsen H, Abderrahim H, Cann HM, Dausset J, Le Bellanne-Chantelot C, Lacroix B, Ougen P, Billault A, Paslier D and Cohen D. (1990). Proc. Natl. Acad. Sci. Beau®ls S, Bertrand S, Georges I, Glibert F, Gros I, USA, 87, 4256 ± 4260. Lucotte G, Susini L, Codani J-J, Gesnoulin P, Pook S, Allan LA, Duhig T, Read M and Fried M. (2000). Mol. Cell. Vaysseix G, Lu-Kuo J, Ried T, Ward D, Chumakov I, Le Biol., 20, 1291 ± 1298. Paslier D, Barillot E and Cohen D. (1992). Cell, 70, 1059 ± Altschul SF, Gish W, Miller W, Myers EW and Lipman DJ. 1068. (1990). J. Mol. Biol., 215, 403 ± 410. Bhatia-Gaur R, Donjacour AA, Sciavolino PJ, Kim M, Anderson MJ and Stanbridge EJ. (1993). FASEB J., 7, 826 ± Desai N, Young P, Norton CR, Gridley T, Cardi RD, 833. Cunha GR, Abate-Shen C and Shen MM. (1999). Genes Baker A and Cotten M. (1997). Nucleic Acids Res., 25, 1950 ± Dev., 13, 966 ± 977. 1956. Bookstein R, Levy A, MacGrogan D, Lewis TB, Weissen- Baley PA, Yoshida K, Qian W, Sehgal I and Thompson TC. bach J, O'Connell P and Leach RJ. (1994). Genomics, 24, (1995). J. Steroid Biochem. Mol. Biol., 52, 403 ± 413. 317 ± 323.

Oncogene Functional identification of LZTS1 as a candidate TSG Y Cabeza-Arvelaiz et al 4179 Bova GS, MacGrogan D, Levy A, Pin SS, Bookstein R and Ishii H, Baa R, Numata SI, Murakumo Y, Rattan S, Inoue Isaacs WB. (1996). Genomics, 35, 46 ± 54. H, Mori M, Fidanza V, Alder H and Croce CM. (1999). Burgers PM and Percival KJ. (1987). Anal. Biochem., 163, Proc. Natl. Acad. Sci. USA, 96, 3928 ± 3933. 391 ± 397. Johnson DG. (2000). Mol. Carcinog., 27, 151 ± 157. Cabeza-Arvelaiz Y, Shih LC, Hardman N, Asselbergs F, Kagan J, Stein J, Babaian RJ, Joe YS, Pisters LL, Glassman Bilbe G, Schmitz A, White B, Siciliano MJ and Lachman AB, von Eschenbach AC and Troncoso P. (1995). LB. (1993). DNA Cell Biol., 12, 881 ± 892. Oncogene, 11, 2121 ± 2126. Ceballos E, Delgado MD, Gutierrez P, Richard C, Muller D, Kirschner LS and Stratakis CA. (1999). Biotechniques, 27, Eilers M, Ehinger M, Gullberg U and Leon J. (2000). 72 ± 74. Oncogene, 19, 2194 ± 2204. Koreth J, Bakkenist CJ, Larin Z, Hunt NCA, James MR and Chinault AC. (1994). Current Protocols in Human Genetics, McGee JOD. (1999). Oncogene, 18, 1157 ± 1164. Vol. 1. Dracopoli, NC, Haines JL, Korf BR, Moir DJ, Laoide BM, Foulkes NS, Schlotter F and Sassone-Corsi P. Morton CC, Seidman CE, Seidman, JG and Smith DR (1993). EMBO J., 12, 1179 ± 1191. (eds). John Wiley and Sons, Inc.: New York, pp. 5.5.1 ± Lu X, Park SH, Thompson TC and Lane DP. (1992). Cell, 5.5.10. 70, 153 ± 161. Chiu R, Angel P and Karin M. (1989). Cell, 59, 979 ± 986. Lucibello FC, Lowag C, Neuberg M and Muller R. (1989). Chumakov IM, Rigault P, Le Gall I, Bellanne-Chantelot C, Cell, 59, 999 ± 1007. Billault A, Guillou S, Soularue P, Guasconi G, Poullier E, MacGrogan D, Levy A, Bova GS, Isaacs WB and Bookstein Gros I, Belova M, Sambucy JL, Susini L, Gervy P, Glibert R. (1996). Genomics, 35, 55 ± 65. F, Beau®ls S, Bui H, Massart C, De Tand MF, Dukasz F, Markie D, Ragoussis J, Senger G, Rowan A, Sansom D, Lecoulant S, Ougen P, Perrot V, Saumier M, Sovarito C, Trowsdale J, Sheer D and Bodmer WF. (1993). Somat. Bahouayila R, Cohen-Akenine A, Lacroix B, Lucotte G, Cell. Mol. Genet., 19, 161 ± 169. Sahbatou M, Schmit C, Sangouard M, Tubacher E, Dib C, Mitchell KO and El-Deiry WS. (1999). Cell Growth Dier., Faure S, Fizames C, Gyapay G, Millasseau P, Nguyen S, 10, 223 ± 230. Muselet D, Vignal A, Morissette J, Menninger J, Lieman Molina CA, Foulkes NS, Lalli E and Sassone-Corsi P. J,DesaiT,BanksA,Bray-WardP,WardD,HudsonT, (1993). Cell, 75, 875 ± 886. Gerety S, Foote S, Stein L, Page DC, Lander ES, Morgenstern JP and Land H. (1990). Nucleic Acids Res., 18, Weinssenbach J, le Paslier D and Cohen D. (1995). 1068. Nature, 377, 175 ± 297. Murakami Y, Nobukuni T, Tamura K, Maruyama T, Sekiya Courey AJ and Tjian R. (1988). Cell, 55, 887 ± 898. T, Arai Y, Gomyou H, Tanigami A, Ohki M, Cabin D, Evan G and Littlewood T. (1998). Science, 281, 1317 ± 1322. Frischmeyer P, Hunt P and Reeves RH. (1998). Proc. Natl. Farrington SM, Cunningham C, Boyle SM, Wyllie AH and Acad.Sci.USA,95, 8153 ± 8158. Dunlop MG. (1996). Oncogene, 12, 1803 ± 1808. Murakami YS, Albertsen H, Brothman AR, Leach RJ and Foulkes NS and Sassone-Corsi P. (1996). Biochim. Biophys. White RL. (1996). Cancer Res., 56, 2157 ± 2160. Acta, 1288, F101 ± F121. Pachnis V, Pevny L, Rothstein R and Costantini F. (1990). Fujiwara Y, Ohata H, Kuroki T, Koyama K, Tsuchiya E, Proc. Natl. Acad. Sci. USA, 87, 5109 ± 5113. Monden M and Nakamura Y. (1995). Oncogene, 10, 891 ± Park BJ, Park JI, Byun DS, Park JH and Chi SG. (2000). 895. Cancer Res., 60, 3031 ± 3038. GossenMandBujardH.(1992).Proc. Natl. Acad. Sci. USA, Pirrotta V, Manet E, Hardon E, Bickel SE and Benson M. 89, 5547 ± 5551. (1987). EMBO J., 6, 791 ± 799. Goyette MC, Cho K, Fasching CL, Levy DB, Kinzler KW, Reddy DE, Keck CL, Popescus N, Athwal RS and Kaur GP. Paraskeva C, Vogelstein B and Stanbridge EJ. (1992). (2000). Oncogene, 19, 217 ± 222. Mol. Cell Biol., 12, 1387 ± 1395. Tsai SY, Sagami I, Wang H, Tsai MJ and O'Malley BW. Habener JF. (1990). Mol. Endocrinol., 4, 1087 ± 1094. (1987). Cell, 50, 701 ± 709. Harris H, Miller OJ, Klein G, Worst P and Tachibana T. Tuck SP and Crawford L. (1989). Oncogene Res., 4, 81 ± 96. (1969). Nature, 223, 363 ± 368. Zeng YX, Somasundaram K and El-Deiry WS. (1997). Nat. Hutton FG, Turnell AS, Gallimore PH and Grand RJ. Genet., 15, 78 ± 82. (2000). Oncogene, 19, 452 ± 462. Ichikawa T, Nihei N, Suzuki H, Oshimura M, Emi M, Nakamura Y, Hayata I, Isaacs JT and Shimazaki J. (1994). Cancer Res., 54, 2299 ± 2302.

Oncogene