Genomic organization, complex splicing pattern and expression of a ...

ã

Oncogene (2001) 20, 5930 ± 5939 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc

Genomic organization, complex splicing pattern and expression of a human septin gene on chromosome 17q25.3 Michael A McIlhatton1, James F Burrows1, Paul G Donaghy1, Severine Chanduloy1, Patrick G Johnston1 and SE Hilary Russell*,1

ONCOGENOMICS

1

Department of Oncology, The Cancer Centre, Queen's University Belfast, Belfast BT9 7AB

The Ov/Br septin gene, which is also a fusion partner of MLL in acute myeloid leukaemia, is a member of a family of novel GTP binding proteins that have been implicated in cytokinesis and exocytosis. In this study, we describe the genomic and transcriptional organization of this gene, detailing seventeen exons distributed over 240 kb of sequence. Extensive database analyses identi®ed orthologous rodent cDNAs that corresponded to new, unidenti®ed 5' splice variants of the Ov/Br septin gene, increasing the total number of such variants to six. We report that splicing events, occurring at noncanonical sites within the body of the 3' terminal exon, remove either 1801 bp or 1849 bp of non-coding sequence and facilitate access to a secondary open reading frame of 44 amino acids maintained near the end of the 3' UTR. These events constitute a novel coding arrangement and represent the ®rst report of such a design being implemented by a eukaryotic gene. The various Ov/Br proteins either di€er minimally at their amino and carboxy termini or are equivalent to truncated versions of larger isoforms. Northern analysis with an Ov/Br septin 3' UTR probe reveals three transcripts of 4.4, 4 and 3 kb, the latter being restricted to a sub-set of the tissues tested. Investigation of the identi®ed Ov/Br septin isoforms by RT ± PCR con®rms a complex transcriptional pattern, with several isoforms showing tissue-speci®c distribution. To date, none of the other human septins have demonstrated such transcriptional complexity. Oncogene (2001) 20, 5930 ± 5939. Keywords: septin gene; genomic organisation; complex splicing Introduction Loss of heterozygosity (LOH) analyses of sporadic ovarian and breast tumours have demonstrated high rates of loss from chromosome 17q25 (Kalikin et al., 1996; Russell et al., 1990). As a result of ®ne deletion mapping and positional cloning studies, a candidate

*Correspondence: SEH Russell, Department of Oncology, The Cancer Centre, Queen's University Belfast, U Floor, Belfast City Hospital, Belfast BT9 7AB; E-mail: [email protected] Received 3 May 2001; revised 20 June 2001; accepted 26 June 2001

septin gene has been identi®ed at 17q25.3 (Kalikin et al., 2000; Russell et al., 2000). Three recent reports on acute myeloid leukaemia have also identi®ed this gene, and another member of the septin family, CDCrell, as fusion partners of MLL in t(11;17)(q23;q25) or t(11;22)(q23;q11.2) translocations (Megonigal et al., 1998; Osaka et al., 1999; Taki et al., 1999). The septin family of novel GTP-binding proteins were ®rst discovered through genetic studies on Saccharomyces cerevisiae (Field and Kellogg, 1999; Longtine et al., 1996; Trimble, 1999). A group of four genes, CDC3, CDC10, CDC11 and CDC12, were identi®ed by a screen for temperature-sensitive mutants that demonstrated defects in regulation of the cell division cycle (Hartwell, 1971). Cells with mutations in any one of these genes were unable to complete cytokinesis, resulting in the formation of multi-nucleate cells with elongated buds. Septins have also been reported in a number of other organisms including, Drosophila (Fares et al., 1995; Neufeld and Rubin, 1994) and mammals (Hsu et al., 1998; Kinoshita et al., 1997; Xie et al., 1999), where they have been detected at the cleavage furrow of cells undergoing mitosis, again suggesting a role in cytokinesis. Collectively the septins represent a family of genes that demonstrate 526% sequence identity over their entire protein length (Longtine et al., 1996). Although they vary greatly over their respective amino and carboxy termini, they all have a conserved core region that contains a P-loop nucleotide-binding consensus sequence, a characteristic signature of GTP-binding and hydrolysis (Saraste et al., 1990). Most septins have a coiled-coil domain within their carboxy terminal region but there are exceptions, including S. cerevisiae Cdc10p, C. albicans CaCdc10p and S. pombe Spn2p (Longtine et al., 1996). This domain participates in protein ± protein interactions and may mediate homoor heterotypic contacts between septins. The yeast septins Cdc3p, Cdc10p, Cdc11 and Cdc12p appear to assemble into ®laments which constitute a protein sca€old important in recruiting and tethering other proteins to speci®c sites on the plasma membrane (Chant et al., 1995; DeMarini et al., 1997; Epp and Chant, 1997). It has been reported that mammalian (Hsu et al., 1998) and Drosophila (Field et al., 1996) septins also form ®laments, suggesting that this feature may be important for their function. Septins were

Septin gene, complex genomic structure MA McIlhatton et al

originally identi®ed due to their proposed role in cytokinesis. However, mammalian septins have been detected at high levels in brain, a predominantly nonmitotic tissue (Beites et al., 1999; Hsu et al., 1998; Xie et al., 1999), and levels of the mouse Nedd5 protein are increased in growth-arrested Siha cells (Kinoshita et al., 1997). These observations indicate that the septins may be involved in other cellular processes and recent reports have implicated them in vesicle tracking and exocytosis (Hsu et al., 1998; Beites et al., 1999). In our previous report on this new member of the septin family, analysis of over 205 kb of genomic sequence revealed seven exons of the gene (1a, 1b, 3, 4, 5, 6, 7) and identi®ed two transcripts designated Ov/Br septin a and b (Russell et al., 2000). In this paper, we extend our preliminary ®ndings and describe the genomic and transcriptional organization of this gene. We reveal a complex pattern of alternative splicing that generates multiple transcripts and describe a novel arrangement used to create additional diversity within the carboxy termini of the various isoforms.

Results The Ov/Br septin has a complicated genomic structure, distributed over at least 240 kb Since our initial report, recurrent database analyses have revealed additional splice variants of the Ov/Br septin gene. Orthologous septin isoforms were identi®ed in rat that di€ered signi®cantly from the Ov/Br septin a and b sequences in their 5' untranslated regions (UTR) and were used to establish the existence of Ov/Br septin g and d transcripts. Moreover, a further two Ov/Br septin isoforms, corresponding to septin e and septin z, have been identi®ed (Kalikin et al., 2000), bringing the present number of 5' septin splice variants to six. These analyses allowed the available genomic sequence to be extended both upstream and downstream of the original 205 kb interval and led to identi®cation and mapping of the positions of the 3' exons 8 ± 12 (Figure 1a). The complex genomic organization facilitates alternative splicing which generates multiple mRNAs that di€er at their 5' termini and manifest further diversity in their 3' UTR con®gurations through splicing events that occur within the body of exon 12. All possible combinations of these transcripts are represented in Figure 1b. Alternative splicing of the Ov/Br septin gene generates six mRNAs with different 5' termini A number of orthologous sequences have been identi®ed in rat, corresponding to various isoforms of the Ov/Br septin gene. The rat genes SLP (septin-like protein) A and B (accession numbers AF170253 and AF173899) and Eseptin long (L) and short (S) (accession numbers AF180525 and AF180526; Fung and Scheller, 1999) demonstrate approximately 75% nucleic acid sequence identity with the various Ov/Br

septin isoforms, and 89 ± 96% at the amino acid level. They are identical in sequence, apart from regions at the 5' end that correspond to alternative ®rst exons, and the 3' UTR of Eseptin S which is *680 bp shorter. One explanation for the latter observation is that the cDNA has been truncated during the cloning procedures and that the apparent deletion is a misleading artifact, rather than a 3' splice variant, especially since the Eseptin S 3' UTR sequence does not contain a polyadenylation signal. The Eseptin L sequence (3 kb) is contained within those of SLP A (3.9 kb) and SLP B (3.8 kb) and it is probably representative of an incomplete cDNA that does not extend upstream to the proper transcriptional start site. As such, it is impossible to determine which isoform it is derived from. SLP A can be considered as the rat orthologue of Ov/Br septin a and demonstrates 76% nucleotide sequence identity over the entire transcript, including exon 1a and the 3' UTR. In contrast, SLP B and Eseptin S are comparable to neither Ov/Br septin a nor septin b over the ®rst 5' exon. Both sequences were used to interrogate the available genomic entries in the database. The ®rst *310 bp of the SLP B sequence identi®ed an orthologous region within the original 205 kb genomic interval, now designated the Ov/Br septin 1g exon; similarly, the ®rst 50 bp of Eseptin S identi®ed the Ov/ Br septin 1d exon. Primer sets SE1gF/ABO14R (348 bp) and SE1dRaF/ABO04 (513 bp) were used to amplify PCR products of the correct size from peripheral blood lymphocyte (PBL) cDNA (Figure 3b), con®rming that exons 1g and 1d are transcribed and processed within authentic Ov/Br septin mRNAs. Full-length Ov/Br septin g and septin d cDNAs were ampli®ed by LR ± PCR using primer sets SE1gBF/ ABO24Ra (3832 bp) and SE1dRaF/ABO24Ra (2990 bp) and sequenced (Figure 2a and c). For completion of the study on the di€erent 5' variants of the Ov/Br septin gene, the 5' UTRs and open reading frames (ORF) of the a and b isoforms were ampli®ed from PBL cDNA. These sequences were also analysed by NIX and PIX. A panel of human cDNAs (Origene) was utilized for 5' RACE (rapid ampli®cation of cDNA ends) in an attempt to isolate further splice variants of the Ov/Br septin gene. Although new 5' exons were not obtained, the sequence of the septin 1e exon was extended upstream by 75 nucleotides. This increases the length of the 5' UTR but does not change the ORF of septin e, that is, there are no in-frame upstream methionine codons (Figure 2B).

5931

Intra-Exonic splicing within exon 12 of the Ov/Br septin establishes additional transcriptional complexity by generating different 3' UTRs The sequence of Ov/Br septin a is comparable to that of the MLL septin-like fusion (MSF) gene (Osaka et al., 1999), except that the latter has a 3' UTR deletion of 1640 bp within exon 12. To determine if the MSF sequence was indicative of alternative splicing occurOncogene

Septin gene, complex genomic structure MA McIlhatton et al

5932

Figure 1 (a) Genomic organization of the Ov/Br septin gene showing exon/intron positions. Microsatellite markers used in the original LOH analyses are indicated by ;. N (NotI) and S (SalI) represent restriction sites employed in the mapping studies. The dashed lines between exons 1e and z and exon 2 signify that these distances are unknown. (b) The Ov/Br septin gene is alternatively spliced at the 5' and 3' termini to generate multiple transcripts that encode similar or identical proteins. 5' alternative exons are identi®ed by the di€erently shaded boxes. Intra-exonic splicing within exon 12 removes the intervening non-coding sequences and accesses an ORF situated near the end of the 3' UTR. Initiation and termination codons of respective transcripts are represented by ! and $

ring within this region, primer set ABO07/AE3 was used for RT ± PCR on PBL cDNA. If the MSF 3' UTR was authentic, we reasoned that two products of 2100 and 570 bp would be ampli®ed. However, products of 2100, 410 and 362 bp were consistently obtained, even when cDNAs from other tissues were used (Figure 3b). Sequence analysis of the 410 and 362 bp products indicated that they signify deletions of 1801 and 1849 bp within the 2073 bp of exon 12, and we have termed these splice variants 2 and 3 (SV2 and SV3), respectively. These events occur at non-consensus splice sites within exon 12 and facilitate access to a small downstream ORF located close to the end of the 3' UTR (Figures 1b and 2a). We have described this phenomenon as intra-exonic splicing in order to distinguish it from simple cryptic splicing. Although these events occur wholly within the body of exon 12, they also involve the utilisation of a second discrete downstream ORF, paradoxically maintained within the Oncogene

non-coding 3' UTR. The Ov/Br septin a-z transcripts may all be subjected to intra-exonic splicing within this exon, thus generating a potential 18 di€erent isoforms (Table 1). The Ov/Br septin a-z isoforms encode highly similar or identical proteins The various Ov/Br septin isoforms, produced through alternative splicing of 5' and 3' exons, encode proteins that di€er over small regions of their amino and/or carboxy termini (Figure 1b). Indeed, the Ov/Br septin a, g and e proteins are identical, except for 7, 18 and 25 amino acids respectively at their amino termini. (Figure 4). Ov/Br septin b and septin z encode identical proteins of 422 amino acids. The septin b, z and d proteins can all be considered as shorter, inframe versions of the septin g (or a or e) isoform (Figures 1b and 2a). PIX analyses revealed a number of interesting

Septin gene, complex genomic structure MA McIlhatton et al

5933

Figure 2 (a) Nucleotide and predicted amino acid sequence of Ov/Br septin g. Primer sequences, highlighted in white, are sense (?) or antisense (/); [S] signi®es it was used in sequencing studies. The underlined methionine codons within exons 2 and 3 correspond to the in-frame start sites of septin b/z and septin d proteins respectively. Exon/exon boundaries are represented by /. Sites of intra-exonic splicing within exon 12 are identi®ed by ;; excision of sequences between positions [2] (1991 bp) and [1] (3793 bp) or [3] (1943 bp) and [1] generate SV2 and SV3 transcripts (see text). The polyadenylation site is symbolized by ***. (b) Nucleotide and predicted amino acid sequence of exons 1e and 2. (c) Nucleic acid sequence of exon 1d

properties of the various isoforms (data not shown), including the following: (1) The Ov/Br septin is unique in that it has a much longer amino terminus than the other mammalian septins of at least 154 amino acids;

(2) The carboxy terminus is comparatively short and does not contain a predicted coiled-coil domain; (3) The equivalent regions of both SV2 and SV3, generated through intra-exonic splicing of exon 12, Oncogene

Septin gene, complex genomic structure MA McIlhatton et al

5934

Table 1

Figure 3 (a) Northern blot analysis of the Ov/Br septin gene in multiple human tissues. A 3' UTR probe was hybridised against a Clontech Human II blot (top panel) and then stripped and reprobed with actin (bottom panel) to verify equivalent loading. Three septin transcripts of around 4.4, 4 and 3 kb are apparent. (b) RT ± PCR survey of septin transcripts. Transcript-speci®c primer sets were used to amplify products representing the known septin isoforms from a panel of human cDNAs. The resultant composite pro®le indicates restricted distribution of certain transcripts among those tissues investigated and demonstrates that not all septin isoforms are ubiquitously expressed. Intraexonic spicing within exon 12 is observed in all of the tissues examined, although the relative amounts of SV2 and SV3 varied. (The third PCR product of 2100 bp has not been included on this ®gure.) PBL cDNA was used for the positive control in all cases, except that of Ov/Br e which was not detectable in blood. Instead it was ampli®ed from cDNA prepared from the ovarian epithelial cell line OAW42

also lack a coiled-coil domain and do not demonstrate notable identity with any other protein; and (4) None of the isoforms encode predicted signal sequences or possess transmembrane domains, suggesting that they are neither modi®ed by glycosylation nor associated with membranes through direct insertion. Expression of various Ov/Br septin isoforms is restricted in a tissue-specific pattern The expression of Ov/Br septin was investigated by Northern blot analysis using a series of human tissues. A 648 bp probe was ampli®ed with primer set Oncogene

Properties of Ov/Br septin mRNAs

Ov/Br septin

mRNA (bp)

ORF

Protein (aa)

mol. wt. (kDa)

a b g d e z a SV2 b SV2 g SV2 d SV2 e SV2 z SV2 a SV3 b SV3 g SV3 d SV3 e SV3 z SV3

4455 3937 3984 3026 3792 3950 2654 2136 2183 1226 1991 2149 2606 2088 2135 1178 1943 2101

813 ± 2519 733 ± 2001 309 ± 2048 83 ± 1090 96 ± 1856 746 ± 2014 813 ± 2597 733 ± 2079 309 ± 2126 83 ± 1168 96 ± 1934 746 ± 2092 813 ± 2549 733 ± 2031 309 ± 2078 83 ± 1120 96 ± 1886 746 ± 2044

568 422 579 335 586 422 594 448 605 361 612 448 578 432 589 345 596 432

63.6 47.5 64.6 38.5 65.4 47.5 66.7 50.6 67.8 41.6 68.5 50.6 64.8 48.7 65.8 39.7 66.6 48.7

ABO11F/ABO12R (see Figure 2a) from within the 3' UTR. Three main transcripts, of estimated sizes 4.4, 4 and 3 kb, were observed after hybridization of a 32PdCTP labelled PCR probe to the Clontech Human II Northern blot (Figure 3a). We propose that the 4.4 kb transcript, present in all of the tissues examined, corresponds to Ov/Br septin a. The 4 kb transcript also appears to be ubquitiously expressed but this product could represent transcription of either septin b, g, e or z isoforms, or combinations thereof. The 3 kb transcript, comparable in size to Ov/Br septin d, is restricted to a subset of tissues which includes PBL, spleen and thymus. To complement the above analysis, RT ± PCR was performed on a panel of six human cDNAs (Origene) using transcript-speci®c primer sets. Although this approach is only semi-quantitative, its sensitivity is useful for compiling a transcriptional pro®le of the di€erent septin isoforms. Overall, the results (Figure 3b) are in agreement with those presented in the Northern blot. Septins a and b were detected in the six cDNAs examined, even though only low levels of b were found in prostate and small intestine. Under the PCR conditions employed, septins g, d and e were detected in the testis, whereas septin z was not detected at all. Products signifying SV2 and SV3 were ampli®ed from all six tissues but the location of primer set ABO07/AE3 (exon 9/3' UTR) makes it impossible to identify which of the septin a ± z transcripts are being alternatively spliced within exon 12. Discussion Previously, we have reported that a new member of the septin gene family, involved in a t(11;17)(q23;q25) translocation in acute myeloid leukaemia (Osaka et al., 1999; Taki et al., 1999), also maps to a region of high LOH in ovarian and breast cancer (Russell et al., 2000). These ®ndings underline the importance of

Septin gene, complex genomic structure MA McIlhatton et al

5935

Figure 4 ClustalX alignment of the human septins: major di€erences are con®ned to the amino and carboxy termini. Only the complete Ov/Br septin e protein sequence has been included. The di€erent amino and carboxy termini contributed by the other isoforms are positioned with respect to this sequence. The in-frame start site of Ov/Br septin b/z is indicated by b above methionine 165 in the septin e sequence; that of Ov/Br septin d is similarly indicated at position 252. Predicted casein kinase II and protein kinase C sites in exons 1g and 1e respectively are underlined. The available sequences of KIAA0202 and SEP2 proteins are incomplete at their amino termini. Amino acids of the conserved P-loop are indicated by ****. GenBank accession numbers are as follows: C10H, Q16181; NEDD5, Q15019; Ov/Br septin e (MSF-A), AAF23374; BH5, O43236; SEP2 (KIAA0128), Q14141; CDCrel-1, O15251; KIAA0202, Q92599

determining the signi®cant characteristics of this gene, with respect to a better understanding of how it may contribute to tumorigenesis. In this report, we describe the genomic structure of the Ov/Br septin gene and delineate the many 5' and 3' alternative splicing events that are a hallmark of this gene. Alternative splicing creates additional diversity from existing genes and it has been estimated that 22 ± 33% of all human genes may display this property (Croft et al., 2000; Hanke et

al., 1999). There are potentially 18 transcripts generated from the Ov/Br septin gene, although it is likely that this ®gure will be revised upwards, as Northern analyses (Taki et al., 1999; unpublished data) indicate the existence of further uncharacterized mRNAs. Many genes and gene families exhibit alternative splicing at both consensus and cryptic splice sites (Goodison et al., 1999; Kilpatrick et al., 1999; Missler and Sudhof, 1998; Zhang et al., 1999). However, the Ov/Br septin gene is Oncogene

Septin gene, complex genomic structure MA McIlhatton et al

5936

Oncogene

remarkable in that intra-exonic splicing within exon 12 accesses coding information maintained as a small ORF at the end of the non-coding 3' UTR. To the best of our knowledge, this is the ®rst report of such a complex strategy being implemented within the transcriptional repertoire of a eukaryotic gene. Furthermore, these processes may impact on the posttranscriptional regulation of the various Ov/Br septin isoforms in that cis-acting signals governing stability (Ross, 1996) and localization (Hazelrigg, 1998) are contained in the 3' UTRs of many mRNAs. The genomic structure of the Ov/Br septin gene details 17 exons, asymmetrically distributed over *240 kb of sequence (Figure 1a). Alternative splicing is used to shu‚e various combinations of eight 5' exons, six of which are unique to their respective mRNAs. This results in the generation of a number of mRNA isoforms that di€er dramatically in the nature and extent of their 5' UTR sequences but share the common exons 4 ± 11. The genomic organization of these exons, and the fact that speci®c isoforms appear to demonstrate tissue speci®city, suggests that multiple promoters probably feature in transcription of Ov/Br septin. However, further work is required to verify this prediction. The ORFs of septins a, g and e begin within their respective exon 1 (Figure 1b) and conceptual translation yields proteins that are very similar in size. The septin e and z isoforms have previously been reported as MSF A and B, respectively (Kalikin et al., 2000). Ov/Br septin b and z mRNAs are remarkable because, even though they di€er over the ®rst half of their 5' UTRs, both ORFs initiate at the same position within exon 3 and encode the same protein. Exon 3 embodies a further twist in the complex transcriptional arrangements of the Ov/Br gene in that its translational context is partially non-coding within septins b and z, where it comprises part of the 5' UTR, but wholly coding within septins a, g and e, where it contributes to the ORF. Exon 2 exhibits a comparable ¯exibility in its translational context (Figure 1b). All of the standard intron/exon boundaries within the Ov/Br septin gene are demarcated by conserved splice donor/acceptor (GT/AG) dinucleotides (Senapathy et al., 1990), unlike the splicing events that occur within exon 12. As indicated in Figure 2a, intervening non-coding segments of either 1801 bp (between positions [2] and [1]; SV2) or 1849 bp (between positions [3] and [1]; SV3) are removed through alternative splicing at cryptic sites which mediates the juxtaposition of the terminal 193 nucleotides of exon 12 with the remaining upstream sequences. This facilitates access to a small second ORF of 44 amino acids and generates a potential further twelve transcripts with altered 3' UTRs, corresponding to Ov/Br septin a ± z SV2 and SV3. These events have been reported by Taki et al. (1999) but were attributed to only the septin a isoform and, without the genomic structure, it was not recognized that the 3' splicing occurred wholly within this exon. The experimental design adopted in this study makes it impossible to distinguish which isoforms of the Ov/Br septin are produced by splicing within exon 12. LR ± PCR with

septin a ± z speci®c primers on di€erent tissue cDNAs was unsuccessful in demonstrating the presence of cognate SV2 and SV3 transcripts (data not shown). Obviously, it is easier to detect the SV2 and SV3 isoforms by RT ± PCR using an exon 9/3' UTR primer set that will sample all six mRNAs simultaneously (Figure 3b), rather than by transcript-speci®c LR ± PCR, a fact that helps to account for these puzzling results. It is possible that detection of these events within exon 12 re¯ects aberrant, `leaky' splicing. An alternative interpretation is that they are indicative of programmed splicing, albeit at non-canonical sites, within the context of authentically processed hnRNA. We favour the latter case for the following reasons: there are expressed sequence tags (ESTs) and related clones in the database that encompass the 3' UTRs of SV2 and SV3; the corresponding regions of orthologous rat (Fung and Scheller, 1999) and mouse (Sorensen et al., 2000) 3' UTRs are more conserved (*76% identity) than the intervening sequences removed by splicing (*62% identity) and also encode small ORFs, suggesting evolutionary pressure to maintain the organisation of this region. Furthermore, from the data presented in Figure 3b, it is likely that the deletion observed in the 3' UTR of MSF (Osaka et al., 1999) may be an artifact, again probably arising from cloning procedures. Alignment of the Ov/Br septin with known human septins demonstrates that the central core of this protein, containing the characteristic P-loop consensus sequence, is highly conserved and that major variations occur within their respective amino and carboxy termini (Figure 4). This pattern is mirrored in the similar or identical proteins encoded by the various Ov/ Br septin isoforms (Figure 1b). It is reasonable to assume that these regions contribute to the unique properties and roles of individual isoforms. Currently, there is little data to indicate the possible functions of the three di€erent carboxy termini, except that none of them contain the predicted coiled-coil domain present in most other septins (Longtine et al., 1996). However, analysis of the Ov/Br septin amino terminus has been more instructive. This region is extended by at least 154 amino acids over that of the other human septins and does not demonstrate notable identity to other proteins in the database except rat SLP A and B. Furthermore, the amino terminal regions speci®cally encoded by exon 1g and 1e contain a single casein kinase II (consensus: [S/T]-X-X-[D/E]) and protein kinase C (consensus: [S/ T]-X-[R/K]) site respectively, whereas that of exon 1a has no recognizable motifs (PIX analysis, data not shown). These exon 1-encoded phosphorylation sites present a potential mechanism for di€erentially regulating the activity of septin g and e but their relevance to the normal, in vivo physiological control of Ov/Br septin again needs to be demonstrated. Overall, there is a good correlation between the Northern and RT ± PCR data that indicates the Ov/Br septin is widely expressed. Both methods reveal expression of septin a in all of the tissues examined (Figure 3). It is clear that speci®c isoforms are restricted

Septin gene, complex genomic structure MA McIlhatton et al

to certain tissues. Septin d provides the best example, being detected in spleen, thymus, testis and PBL on Northern blots and again in testis by RT ± PCR (Figure 3b). Although there may be di€erential changes in the levels of the 4 kb mRNA between tissues, the similar sizes of septin b, g, e and z transcripts make interpretation dicult. Data from other groups con®rm ubiquitous expression of the Ov/Br septin (Kalikin et al., 2000; Osaka et al., 1999; Taki et al., 1999), although there are some di€erences. We are the only group to detect a 4.4 kb transcript, and there are also two reports of a 1.7 kb transcript (Osaka et al., 1999; Taki et al., 1999). These observations may be due to the choice of probes for the Northern analyses (this study alone used a 3' UTR probe) or variability in hybridization conditions employed by the four groups. The expression pro®les of the other mammalian septins are not comparable with respect to either relative levels or tissue speci®city (Hsu et al., 1998; Mori et al., 1996; Nagase et al., 1996; Xie et al., 1999; Yagi et al., 1998) and suggest that particular septins, or groups of septins, may perform di€erent roles in various cellular contexts. This is evident from S. cerevisiae where the septins CDC3, CDC10, CDC11 and CDC12 are involved in cytokinesis but only CDC3 and CDC11, in conjunction with SPR3 and SPR28, play a role in sporulation (DeVirgilio et al., 1996; Fares et al., 1996). The prevailing interpretation of the role of septins in cytokinesis is that they form heteromeric ®laments that assemble into a sca€old that directs speci®c proteins to precise destinations at the cytoplasmic face of the plasma membrane (Bi et al., 1998; Chant et al., 1995; DeMarini et al., 1997; Epp and Chant, 1997). This model is exempli®ed by the recently characterized hierarchy of protein localizations that occur at the mother ± daughter bud neck during cytokinesis in S. cerevisiae (Longtine et al., 2000; Shulewitz et al., 1999). Defects in the septin components of the ®laments or in proteins that direct ®lament assembly and localization a€ect cytokinesis (Barral et al., 1999; Longtine et al., 1998, 2000). Another cellular process that may involve septins is vesicle tracking. Hsu et al. (1998) reported that a number of septins, including CDC10, Nedd5, H5 and KIAA0128/SEP2, co-immunoprecipitated with the sec6/8 complex from rat brain extracts. An unknown ®fth septin was later identi®ed as Eseptin, the rat ortholog of the Ov/Br septin gene (Fung and Scheller, 1999). Furthermore, it is proposed that the septin CDCrell may in¯uence exocytosis through interactions with syntaxin-docking sites on vesicles (Beites et al., 1999). The current hypothesis for septin function in exocytosis is that they also e€ect their roles in the context of heteromeric ®lament assemblies (Beites et al., 1999; Hsu et al., 1998). The many Ov/Br septin isoforms identi®ed in this study, plus the other mammalian septins, provide a combinatorial diversity for ®lament formation. Such diversity, in conjunction with their collective heterogeneous expression pro®les, may facilitate the participation of septins in a variety of processes beyond

cytokinesis and exocytosis, possibly in tissue-restricted and/or cell-cycle dependent contexts. Further investigation is required to identify these activities and dissect the complex networks of interactions between septins and other cellular components. Understanding these mechanisms will help to de®ne the role of the Ov/Br septin and other members of this family in the development of acute myeloid leukaemia and in epithelial tumours of the ovary and breast.

5937

Materials and methods Data analysis Basic sequence manipulations were performed with the Lasergene (DNAStar) suite of programs. More advanced analyses were conducted through the web-based portals NIX (http://www.hgmp.mrc.ac.uk/ Registered/ Webapp/nix/) and PIX (http://www. hgmp.mrc.ac.uk/ Registered/ Webapp/pix/) at the UK HGMP Resource Centre. They provide an interface to a number of programs for DNA, including GRAIL, Genemark, Hexon, Gene®nder, TSSW, Repeatmasker, and protein analysis, including PRINTS, PROSITE, COILS, Tmpred, Sigcleave, PSORT. NIX and PIX enable the user to perform multiple queries on a sequence simultaneously and display the collective outputs as a simpli®ed hyper-linked graphical representation. Separate BLAST (http://www.ncbi.nlm.nih.gov/ BLAST/; Altschul et al., 1990) and FASTA (http://www. ebi.ac.uk/ fasta3/; Pearson 1990) analyses were reiterated under di€erent parameters in order to identify homologous and related Ov/ Br septin sequences. In addition to NIX, splice sites were identi®ed by the Splice Site Prediction by Neural Network program (http://www. fruit¯y.org/ seq_tools/ splice.html) at the Berkeley Drosophila Genome Project. Long range PCR (LR ± PCR) Full-length Ov/Br septin g and d cDNAs were ampli®ed from PBL cDNA by LR ± PCR using the Expand Long Template PCR System (Roche Molecular Biochemicals) according to the manufacturer's directions. The sequences of the primer sets used, SE1gBF/ABO24Ra (Ov/Br septin g) and SE1dRaF/ ABO24Ra (Ov/Br septin d), are highlighted in Figure 2. Brie¯y, the ®nal reaction mix comprised 16 bu€er 1 containing 1.75 mM MgCl2, 350 mM each dNTP, 300 nM each primer, PBL cDNA (1 ml) and 0.65 ml Expand enzyme mix. Ampli®cation parameters were as follows: [948C for 2 min]; [948C, 20 s; 628C, 30 s; 688C, 4 min; 30 cycles]; [688C for 5 min]. Products were resolved on a 1% DNA agarose gel and visualized on a UV-transilluminator after staining with ethidium bromide (Sambrook et al., 1989). Full-length g and d cDNAs were gel puri®ed with the QIAEX II Agarose Gel Extraction kit (Qiagen). They were sequenced with a number of primers, highlighted in Figure 2, using ABI Prism BigDye Terminator Cycle Sequencing chemistry (PE Applied Biosystems). Reactions were resolved, analysed and edited by standard procedures and software on a 377 automated sequencer (PE Applied Biosystems). Northern blot analysis A multiple tissue Northern blot (Human II; Clontech) was probed with a 3' UTR fragment of the Ov/Br septin gene. Oncogene

Septin gene, complex genomic structure MA McIlhatton et al

5938

Primer set ABO11F/ABO12R (Figure 2a) was used to amplify a 648 bp fragment that was puri®ed with the Wizard PCR Preps DNA puri®cation System (Promega) and labelled with 32P-dCTP using the Megaprime DNA Labelling System (Amersham Pharmacia Biotech). Hybridization and posthybridization wash conditions are as described in the user manual. The membrane was air-dried and exposed to x-ray ®lm (Fuji) for up to 48 h. Survey of Ov/Br septin transcripts by reverse transcription ± PCR (RT ± PCR) Expression of the known Ov/Br septin transcripts was also investigated in normal tissues using RT ± PCR. A panel of six human cDNAs (Origene) provided templates for transcriptspeci®c ampli®cation of the six septin isoforms. The following primer sets were used: Ov/Br a [ABO17F (ACT CCC GCC GCT GCT AAA TA)/ABO16R (Figure 2a)]; Ov/Br b [SE1bF (GGA GAG GGG GCT GGG GCA TAA GGA)/ABO14R]; Ov/Br g [SE1gF/ABO14R (Figure 2a)]; Ov/Br d [SE1dRaF (Figure 2c)/ABO14R]; Ov/Br e [SE1eF (Figure 2b)/ABO14R]; Ov/Br z [SE1fF (GCT GGA CTC TCT CGC TGA CTC)/ ABO14R]. Primer set ABO7/AE3 (Figure 2a) was included to demonstrate the alternative splicing within exon 12 that gives rise to SV2 and SV3 transcripts. The positive control provided with the cDNAs was a primer set to the housekeeping gene cyclophilin. Brie¯y, the ®nal reaction mix comprised 20 mM Tris-HCl (pH 8.4), 50 mM KCl, 1.5 mM MgCl2, 200 mM dNTP, 200 nM each primer, 2 ml tissue cDNA (4 ± 20 ng) and 1 u Taq DNA polymerase (Gibco BRL, Life Technologies). Ampli®cation parameters were as follows: [948C for 4 min]; [948C, 30 s; 558C, 1 min; 728C,

1 min; 40 cycles]; [728C for 5 min]. Products were analysed on a 1.5% DNA agarose gel (Sambrook et al., 1989). Determination of Ov/Br septin a and b sequences Primer sets A7F (TGC CTC TGT CAC TGC AGA CT)/ A5R (TGT GCT CCA AAC TGA AGT AGC ATT G) and A6F (CCT CTC TCC TGC CTC CTA TTT TTG G)/ ABO14R were used to amplify the 5' UTR of septin a from PBL cDNA and primer set ABO09F (GGA GGC GGA CCT TGA GGG AGA GT)/ABO14R was used for that of septin b. The shared coding region was ampli®ed with primer sets ABO15F/ABO02, ABO03/ABO06 and ABO07/ABO08 (Figure 2a). PCR products were puri®ed and sequenced as above.

Acknowledgements This work is supported by grants from The Cancer Research Campaign, WellBeing and The Ulster Cancer Foundation. Data deposition The sequences reported in this paper have been lodged with GenBank. Accession numbers Ov/Br septin g, [AJ312319]; Ov/Br septin d, [AJ312320]; Ov/Br septin a, [AJ312321]; Ov/Br septin b, [AJ312322]; Ov/Br septin e (exon 1e+2), [AJ312323].

References Altschul SF, Gish W, Miller W, Myers EW and Lipman DJ. (1990). J. Mol. Biol., 215, 403 ± 410. Barral Y, Parra M, Bidlingmaier S and Snyder M. (1999). Genes Develop., 13, 176 ± 187. Beites CL, Xie H, Bowser R and Trimble WS. (1999). Nat. Neurosci., 2, 434 ± 439. Bi E, Maddox P, Lew DJ, Salmon ED, McMillan JN, Yeh E and Pringle JR. (1998). J. Cell Biol., 142, 1301 ± 1312. Chant J, Mischke M, Mitchell E, Herskowitz I and Pringle JR. (1995). J. Cell Biol., 129, 767 ± 778. Croft L, Schandor€ S, Clark F, Burrage K, Arctander P and Mattick JS. (2000). Nat. Genet., 24, 340 ± 341. DeMarini DJ, Adams AEM, Fares H, DeVirgilio C, Valle G, Chuang JS and Pringle JR. (1997). J. Cell Biol., 139, 75 ± 93. DeVirgilio C, DeMarini DJ and Pringle JR. (1996). Microbiol., 142, 2897 ± 2905. Epp JA and Chant J. (1997). Curr. Biol., 7, 921 ± 929. Fares H, Goetsch L and Pringle JR. (1996). J. Cell Biol., 132, 399 ± 411. Fares HF, Peifer M and Pringle JR. (1995). Mol. Biol. Cell, 6, 1843 ± 1859. Field CM, Al-Awar O, Rosenblatt J, Wong ML, Alberts B and Mitchison TJ. (1996). J. Cell Biol., 133, 605 ± 616. Field CM and Kellogg D. (1999). Trends Cell Biol., 9, 387 ± 394. Fung ET and Scheller RH. (1999). FEBS Letters, 451, 203 ± 208. Goodison S, Urquidi V and Tarin D. (1999). J. Clin. Path:Mol. Path., 52, 189 ± 196. Oncogene

Hanke J, Brett D, Zastrow I, Aydin A, Delbruck S, Lehmann G, Luft F, Reich J and Bork P. (1999). Trends Genet., 15, 389 ± 390. Hartwell LH. (1971). Exp. Cell Res., 69, 265 ± 276. Hazelrigg T. (1998). 95, 451 ± 460. Hsu SC, Hazuka CD, Roth R, Foletti DL, Heuser J and Scheller RH. (1998). Neuron, 20, 1111 ± 1122. Kalikin LM, Qu X, Frank TS, Cadu€ RF, Svoboda SM, Law DJ and Petty EM. (1996). Genes Chrom. Can., 17, 64 ± 68. Kalikin LM, Sims HL and Petty EM. (2000). Genomics, 63, 165 ± 172. Kilpatrick GJ, Dautzenberg FM, Martin GR and Eglen RM. (1999). Trends Pharm. Sci., 20, 294 ± 301. Kinoshita M, Kumar S, Mizoguchi A, Ide C, Kinoshita A, Haraguchi T, Hiraoka Y and Noda M. (1997). Genes Develop., 11, 1535 ± 1547. Longtine MS, DeMarini DJ, Valencik ML, Al-Awar OS, Fares H, DeVirgilio C and Pringle JR. (1996). Curr. Opin. Cell Biol., 8, 106 ± 119. Longtine MS, Fares H and Pringle JR. (1998). J. Cell Biol., 143, 719 ± 736. Longtine MS, Theesfeld CL, McMillan JN, Weaver E, Pringle JR and Lew DJ. (2000). Mol. Cell Biol., 20, 4049 ± 4061. Megonigal MD, Rappaport EF, Jones DH, Williams TM, Lovett BD, Kelly KM, Lerou PH, Moulton T, Budarf ML and Felix CA. (1998). Proc. Natl. Acad. Sci. USA, 95, 6413 ± 6418. Missler M and Sudhof TC. (1998). Trends Genet., 14, 20 ± 26.

Septin gene, complex genomic structure MA McIlhatton et al

Mori T, Miura K, Fujiwara T, Shin S, Inazaw J and Nakamura Y. (1996). Cytogenet. Cell Genet., 73, 224 ± 227. Nagase T, Seki N, Ishikawa K, Ohira M, Kawabayashi Y, Ohara O, Tanaka A, Kotani H, Miyajima N and Nomura N. (1996). DNA Res., 3, 321 ± 329. Neufeld TP and Rubin GM. (1994). Cell, 77, 371 ± 379. Osaka M, Rowley JD and Zeleznik-Le NJ. (1999). Proc. Natl. Acad. Sci. USA, 96, 6428 ± 6433. Pearson WR. (1990). Methods Enzymol., 183, 63 ± 98. Ross J. (1996). Trends Genet., 12, 171 ± 175. Russell SEH, Hickey GI, Lowry WS, White P and Atkinson RJ. (1990). Oncogene, 5, 1581 ± 1583. Russell SEH, McIlhatton MA, Burrows JF, Donaghy PG, Chanduloy S, Petty EM, Kalikin LM, Church SW, McIlroy S, Harkin DP, Keilty GW, Cranston AN, Weissenbach J, Hickey I and Johnston PG. (2000). Cancer Res., 60, 4729 ± 4734. Sambrook J, Fritch EF and Maniatis T. (1989). Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Saraste M, Sibbald PR and Wittinghofer A. (1990). Trends Biochem. Sci., 15, 430 ± 434. Senapathy P, Shapiro MB and Harris NL. (1990). Methods Enzymol., 183, 252 ± 278. Shulewitz MJ, Inouye CJ and Thorner J. (1999). Mol. Cell Biol., 19, 7123 ± 7137. Sorensen AB, Lund AH, Ethelberg S, Copeland NG, Jenkins NA and Pedersen FS. (2000). J. Virol., 74, 2161 ± 2168. Taki T, Ohnishi H, Shinohara K, Sako M, Bessho F, Yanagisawa M and Hayashi Y. (1999). Cancer Res., 59, 4261 ± 4265. Trimble WS. (1999). J. Mem. Biol., 169, 75 ± 81. Xie H, Surka M, Howard J and Trimble WS. (1999). Cell Motil. Cyto., 43, 52 ± 62. Yagi M, Zieger B, Roth GJ and Ware J. (1998). Gene, 212, 229 ± 236. Zhang T, Xiong H, Kan LX, Zhang CK, Jiao XF, Fu G, Zhang QH, Lu L, Tong JH, Gu BW, Yu M, Liu JX, Licht J, Waxman S, Zelent A, Chen E and Chen SJ. (1999). Proc. Natl. Acad. Sci. USA, 96, 11422 ± 11427.

5939

Oncogene