The process of floral development and reproduction in higher plants is a complex strategy involving a diverse range of gene interactions. Male sterile (ms) mutants have been reported in many species of higher plant as the result of both spontaneous and induced recessive mutations. The isolation of mutants defective at specific stages of the process of male gametogenesis has been used as a tool to investigate processes of anther and pollen development. Male sporogenesis is an area of major importance in plant development biology and is also of commercial interest for the control of crop fertility, for example, for F 1 hybrid seed production, to reduce labour costs to industry and for the production of low-environmental impact genetically engineered crops.

A number of genes that are specifically expressed in the anther or pollen have been isolated but very few that have been identified as having a key role in male sporogenesis have been cloned.

We now report the cloning of a male sterility gene MSI which is vital for normal pollen formation and which expresses a protein having a key role in regulating transcription during male sporogenesis and anther development. The transcriptional regulator gene, MALE STERILITY 1, and the expressed MS 1 protein are characterised by their DNA and amino acid sequences given in the accompanying sequence listings below.

The promoter region of the gene is also shown in the accompanying sequence listing and is itself separately useful as hereinafter indicated.

Following the promoter region, the full genomic sequence of MS1 is given, including exons and introns. The MSI gene may also be utilised in

the form of cDNA, the sequence of which is also given in the accompanying sequence listing.

The Arabidopsis MALE STERILITY 1 gene which is critical for anther and pollen development has been discovered, isolated and cloned. The gene in Arabidopsis, and other related species, encodes a 77kDa protein which has homology to known transcription factors. MSI is believed to play a key role in regulating transcription during specific stages of male sporogenesis and anther development. Coding sequences from the gene can be driven by an appropriate promoter to induce male sterility in plants. Further, the MM promoter can be used to drive male sterility DNA such as that coding for a nuclease, protease or glucanase.

Alternatively or in addition, male sterility can be achieved by disrupting the proper expression of the MS1 gene, for example by transcribing RNA which is antisense to the RNA normally transcribed from the MM gene.

It will be appreciated that the same MS1 or a highly similar gene can be isolated by similar methods from other plant species, coding for the identical or closely homologous polypeptide. The invention contemplates the use of such related genes for the purposes described hereinafter.

Accordingly, the invention provides a genomic DNA sequence of MS1 substantially as specified in Sequence Listing 1.

The invention also provides a cDNA sequence of MM as substantially specified in Sequence Listing 3.

The invention further provides a protein sequence of amino acids corresponding to the expressed MS1 protein substantially as specified in Sequence Listing 2.

The invention provides isolated DNA or cDNA having a promoter sequence substantially as specified in Sequence Listing 4.

The invention further provides an isolated DNA or cDNA capable of expressing MSI protein or mutants thereof.

The invention provides the use of isolated cDNA or DNA or protein substantially as specified in any one of Sequence Listings 1 to 4 to control or induce male sterility.

The invention also provides a genetically modified plant using the cDNA or DNA sequences as specified in any one of Sequence Listings 1,3 or 4.

EXPERIMENTAL PROCEDURES The MSI mutation was generated by EMS and irradiation mutagenesis of Arabidopsis thaliana Landsberg erecta (Ler) seed. Five alleles of MSI, using EMS (MS1. 2, MS1. 3 and MS1. 4), X-rays MS1. 5 (Dawson et al., 1993) and gamma-rays MS1. 6 (Kalantidis et al., 1994) were discovered.

All allelic lines exhibit the same phenotype. The MSI mutation is recessive and appears as wild type (wt) in the heterozygous state.

Homozygous MSI mutants fail to produce viable pollen, with pollen degeneration occurring soon after microspore release from the tetrads.

The early stages of pollen meiosis and tetrad formation occur normally, followed by callose dissolution and microspore release. Immediately after release, the microspores develop an unusually granular, vacuolated cytoplasm, the tapetum becomes abnormally vacuolated and cytoplasmic degeneration begins. The microspores then appear to adhere to each other, collapse and the cytoplasm degenerates. Young buds show characteristic agglutination of the immature pollen. Mature flowers undergo complete degeneration of the microspores and tapetal cells

leaving an empty locule. In all other aspects plant development the MSI mutant is completely normal; female fertility is unaffected and the plant can be maintained by crossing using wt pollen.

We have previously mapped the MM gene to the top-arm of chromosome 5 and identified a marker, NIT4 (Bartel and Fink, 1994), which co- segregates with MM (Thorlby et al., 1997). This region corresponds to the clone AB007651 (MWD9) which was sequenced (85Kb) as part of the Arabidopsis International Sequencing Project (Nakamura et al., 1997) and is available from the Arabidopsis Biological Resource Center (Ohio State University, OH 43210, USA).

Further CAPS markers were designed based on the AB007651 sequence and used to confirm the MM location. Primers CAPS12.5 (5'ATT GCA TAG GCG CAG TTT CAC'3 and 5'TTT TCC CCC AAA CGA AAT CCA C'3) and CAPS2 (5'CAA CGG GCC ATC CCT GCG AA'3 and 5'CGC AGC CGG AGA CTC GTC GAA'3) were used to amplify DNA from parental, Ler and Sn (5)-1, and recombinant lines (Thorlby et al., 1997).

Three recombinants between the g4560 marker and MM (0.9cM), and thirteen recombinants between MS1 and g4111 (0.9cM) were used (Thorlby et al., 1997). CAPS12.5 PCRs were conducted for 35 cycles of 94°C, 30s; 58°C, 45s; 72°C, 2min. CAPS2 long-PCR (Elongase, Boehringer Mannheim) was carried out for 35 cycles of 94°C, 30s; 68°C, 9min. When digested with Xbal the Sn (5)-1 CAPS12.5 product gave two bands of and 0.4kb, whilst Ler gave a single 1. 5kb band. Sall digest of the CAPS2 Sn (5)-lproduct gave a 5.6kb and a 1.055kb band, whilst the Ler product was uncut (7.7kb).

Marker CAPS12.5 was found to map 0.14cM from MSI towards ttg and CAPS2 to co-segregate with MSI.

The AB007651 clone was analysed for potential open reading frames (ORFs) and primers were designed to amplify, by long-PCR, across this region using the Ler wt and allelic MSI mutants. Twenty-two primer pairs were designed to long-PCR across the AB007651 clone. Each PCR was designed to span a region of ~5-lOkb. Undigested fragments were size separated by agarose gel electrophoresis to look for deletions; samples were also digested with selected restriction enzymes, based on the AB007651 sequence, to identify potential polymorphisms associated with the MSI mutations. Given that two of the allelic lines were generated by irradiation, it was expected that these would contain deletions and that aberrations associated with the MM gene would be detected.

The 2870bp PCR product generated with primers 14/6 (14 : 5'GCC ATT TCA TCG ACC CTT GTG'3'; 6 : 5'CGT GTG TGT TTG CGT GGT TGG ACA'3) showed loss of the 883bp Rsal restriction site in the MSI. 1 and MS1. 6 mutants. This region was sequenced and shown to correspond to a G to A transition associated with a splice site in a potential ORF. To confirm that this was the MM gene, the region corresponding to this ORF was amplified and directly sequenced from Ler and all of the allelic lines.

Sequencing revealed aberrations in this gene in all MS1 alleles, resulting in splice defects or truncated polypeptides.

Three mutants, MS1. 1, MS1. 6 and MS1. 4 showed mutations associated with the splicing of the second intron. RT-PCR was conducted to confirm this. RNA was isolated from Arabidopsis flowers using the RNeasy kit (Qiagen). RT-PCR was conducted using oligodT primed first strand DNA s nthesis (ReVerie-iT Advanced Biotechnologies) and an RNA- specific primer spanning the first splice site (RT-1. L 5'CCA TTG CCA ATA TGT TGG TTG'3 and RT-1. R sCAG CCT CAA CTC CAT TCC TT'3). PCR was conducted using HotStar Taq DNA polymerase (Qiagen) for one cycle of 15 min at 94°C, followed by 45 cycles of 94°C, 30s;

55°C, 30s ; 72°C, 90s. The Ler wild type produced an RT-PCR product of 351bp whilst the splice mutants MS1. 1 and MS1. 6 gave products of 506bp, due to the lack of splicing caused by the G to A transition at the splice donor site. Direct fluorescent sequencing (ABI310, Perkin Elmer Biosystems) of all the allelic and Ler RT-PCR products showed the expected lack of splicing in the MS1. 1 and MS1. 6 mutants. In addition, the MS1. 4 mutant showed a lbp deletion from the spliced message, corresponding to the G to A transition at the splice acceptor site.

Since six independent MSI lines all show mutations in this gene and differences in RT-PCR products were seen in the putative splice mutants, we therefore conclude that this ORF corresponds to the MSI gene.

The MSI gene described and isolated as presented above can also be isolated by a number of methods known in the art. The clones can then be used to identify homologous clones by hybridisation to DNA or cDNA from other plant species.

Isolation of genomic MSI gene: For example: Isolation from genomic DNA or directly from the AB007561 (MWD9) clone may be achieved by the use of appropriate PCR primers which flank the MS1 gene (e. g. 35L: 5'TGACAGAAATTGACGGAACACAGA3'and 3247R: sTCCAGAGGAGGCAAGGATATGA3). These can be used to amplify by long PCR across the MSI gene.

Isolation of MS1 cDNA : cDNA can be isolated using the previously generated genomic PCR product to screen a floral cDNA library (available from the Arabidopsis Biological Resource Centre) by hybridisation using standard techniques. Alternatively PCR primers can be used to amplify the MSI cDNA from RNA isolated from floral tissues.

First strand cDNA synthesis can be conducted using oligodT primers,

followed by RT-PCR using specific primers to the MSI gene, e. g. L3: 5'GGGGTACCTCAAGATCAAACCAAATTTCTG3'Rcloning : 5'GCTCTAGAGAAGGTTTAATATAACATGAAT3'. These primers incorporate a Kpnl and Xbal restriction site, respectively, to permit directional cloning into an appropriate vector.

Isolation of promoter regions: Promoter regions in these sequences can be identified by PCR amplification of fragments from the clone and linking to a reporter construct, such as the beta-glucuronidase (GUS) reporter gene. For example, a series of constructs of the 5'upstream region of the transcription start site can be cloned into pBI101 (Clontech), conjugated into Agrobacterium tumefaciens (strain LBA4404) and transformed into wild type plants. Transformants can then be analysed for expression of the GUS-MSI transcript by histochemical localisation of GUS in fixed, stained and cleared anther sections and by fluorimetric analysis of GUS activity. By this approach the promoter region and any other sequences required for tissue specific expression of MSI can be confirmed. The promoter sequence can then be utilised by standard cloning methodologies for control of male fertility.

Plant Growth Conditions and Microscopy Plants were grown as described by Dawson et al. (1993a). Seeds were obtained from the Nottingham Arabidopsis Stock Centre (The University of Nottingham) and DNA stocks from the Arabidopsis Biological Resource Center (The Ohio State University).

Flower buds were selected from a range of developmental stages and fixed, sectioned and stained (Dawson et al., 1993a).

Localisation of the MS1 gene Primers CAPS12.5 (5'ATT GCA TAG GCG CAG TTT CAC3'and 5'TTT TCC CCC AAA CGA AAT CCA C3') and CAPS2 (5'CAA CGG GCC ATC CCT GCG AA'3 and 5'CGC AGC CGG AGA CTC GTC GAA'3) were used to amplify DNA from parental, Ler and Sn (5)-l, and recombinant lines (Thorlby et al., 1997). Three recombinants between the 16kb-4D5 BAC marker and MS1 (0.3cM), and thirteen recombinants between MSI and pCTT718 (0.9cM) were used (Thorlby et al., 1997). CAPS12.5 PCRs were conducted for 35 cycles of 94°C, 30s; 58°C, 45s; 72°C, 2min.

CAPS2 long-PCRs, using the Extensor DNA polymerase enzyme according to the manufacturers'recommendations (ABgene) were carried out for 35 cycles of 94°C, 30s; 68°C, 9min. The PCR products were screened using a range of different restriction enzymes (Sambrook et al., 1989) to identify polymorphisms between the parental lines, Ler and Sn (5)-l. When digested with Xbal the Sn (5)-1 CAPS12.5 product gave two bands of 1.1 and 0.4kb, whilst Ler gave a single 1. 5kb. Sall digest of the CAPS2 Sn (5)-1 product gave a 5.6kb and a l. lkb band, whilst the Ler product was uncut (6.7kb).

Twenty-one primer pairs (Table 1) were designed to amplify putative ORFs in the AB007651 clone that might correspond to MSI, or to long-PCR (Extensor DNA polymerase; ABgene) across defined regions of the AB007651 clone. Each long-PCR was designed to span a region of approximately 5-lOkb.

Undigested fragments were size separated by agarose gel electrophoresis (Sambrook et al., 1989) to look for deletions; samples were also digested with selected restriction enzymes, including Bgll, Drat, EcoRl, HaeIIL Hindlll, Rsal, Sacl, Sall, Smal, Styl, Xbal, based on the AB007651

sequence, to identify potential polymorphisms associated with the MSI mutations.

RNA Analyses RNA was isolated from Arabidopsis flowers using the RNeasy kit (Qiagen). RNA samples were DNase treated and then purified using RNeasy spin columns (Qiagen). RT-PCR was conducted using oligodT primed first strand cDNA (Reverse-TrM _ ABgene) and RNA-specific primers spanning the first splice site (RT-l. L 5'CCA TTG CCA ATA TGT TGG TTG3'and RT-1. R 5'CAG CCT CAA CTC CAT TCC TT3'; Fig 4A).

Amplification was conducted using lu. l of the cDNA template, or equivalent RT-control which lacked MMuLV reverse transcriptase, and HotStar Taq DNA polymerase (Qiagen) for one cycle of 15 min at 94°C, followed by 45 cycles of 94°C, 30s; 55°C, 30s; 72°C, 90s.

Analysis of MSI expression was conducted initially by northern hybridisations using radiolabelled MS1 probes and filters containing 25gag total RNA (Sambrook et al., 1989). However, no detectable signal was seen. RT-PCR, as previously described, was therefore conducted using RNA isolated from buds, open flowers, leaves and the inflorescence stem.

Control PCRs were conducted to check the integrity of the cDNA using Arabidopsis actin-2 primers (Act2F : 5'TGCTGACCGTATGAGCAAAG3'; Act2R : 5'CAGCATCATCACAAGCATCC3') which generate a 419bp fragment. Actin-2 RT-PCRs were conducted using lu, l of the cDNA template, or equivalent RT-control which lacked MMuLV reverse transcriptase. PCR's were carried out using HotStar Taq DNA polymerase (Qiagen) for one cycle of 15 min at 94°C, followed by 35 cycles of 94°C, 30s; 50°C, 30s; 72°C, 60s.

RACE-PCR RACE-PCR was conducted to obtain full-length 5'and 3'ends of the wt MS1 message. Total RNA (5, ug) was isolated as previously described and 5'and 3'RACE-PCR conducted using the GeneRacer kit (Invitrogen) according to the manufacturer's instructions. The mRNA was reverse-transcribed using the GeneRacerTM oligoTM primer (Invitrogen) and then diluted 2-fold with sterile water. Gene specific 5'(MSlRACE. R5': 5'GCC GTT GAG AGA GAG CAA GTG ACC Axa') and 3' (MSIRACE. L3': 5'TCG AGA CAA GAA GTG AGG GAT GCA GCT A3') primers were designed based on the wt Ler MS1 cDNA sequence. Amplification was carried out using Extensor DNA polymerase (ABgene) and 1p1 of the 2x diluted cDNA. Amplification was initially conducted at a high annealing temperature to exploit the high melting temperature of the GeneRacer primers and to allow specific amplification products to accumulate. The annealing temperature was then lowered to maximise exponential amplification of the desired gene-specific template. The conditions used were 1 cycle of 94°C, 2min; 5 cycles of 94°C, 30 s and 70°C, 3 min; 10 cycles of 94°C 30 s; 70°C, 30 s and 72°C, 3 min; 30 cycles of 94°C, 30s; 68°C, 30s and 72°C, 3 min; 1 cycle of 72°C for 10 min.

The 5'and 3'RACE PCRs gave defined bands, but with a background smear. Nested PCR was therefore conducted using zu J. l of the initial RACE-PCR product, the nested GeneRacerTM primers and nested Gene-specific primers (5'nested-652 s'AAA AAG CCG CCA TTG TTT CCT3'and 3'-nested-4991 5'AGA CGT CGT GGT GGA GTC AGT G3').

Amplification was for 1 cycle of 94°C 2 min; 25 cycles of 94°C, 30s; 56°C, 30s; 68°C, 90s and 1 cycle of 68°C for 10 min. Strong bands were obtained. The PCRs were then directly sequenced to identify the transcription start and polyadenylation sites.

Sequencing PCR products (genomic and RT-PCR products) were gel, or column, purified (Qiaquick PCR clean-up, Qiagen) and sequenced using big dye terminator kit (PE Biosystems). Primers were designed based on the sequence database information and the determined genomic sequence.

Sequencing products were analysed using an ABI310 Genetic analyser (PE Biosystems).

Promoter fusions A 2.9kb region upstream of the MSI gene was PCR amplified using primers-2985XhoI: 5'CAA TGA GAC CCT CTC TCA TCT TGC'and the reverse primer-31Xbal : 5'CGA ATC AGA AAT TTG GTT TGA TCT TG3'. The PCR product was cloned upstream of the uidA gene between the XhoI and Xbal site of the MOG402 based Binary vector MOGIAA2 : GUS replacing the IAA2 promoter to create MOG MSl : GUS. The MOG MSI : GUS construct was introduced into Agrobacterium tumifaciens strainC58 (pGV3850) by electroporation (Wen-Jun and Forde, 1989).

Transformation of Ler Arabidopsis was performed by the floral dip method (Clough and Bent, 1998). The transformed plants were selected on MS plates containing 50mg/1 kanamycin.

Beta-glucuronidase activity was visualised by staining whole inflorescences, leaves, and stem tissues overnight in X-Gluc solution (Willemsen et al., 1998). Tissues were then cleared in 95% (v/v) ethanol, incubated overnight in FAA fixative (ethanol 50% v/v), acetic acid 5.0% (v/v), formaldehyde 3.7% (v/v)) and embedded in Technovit 7100 resin (Kulzer Histo-Technik). Sections (5-15u, m) were then stained for 20 min in ruthenium red (0.05% (w ! v), pH9), mounted and examined.

The MSI gene has been identified in Arabidopsis thaliana however those skilled in the art would readily be able to identify the corresponding genes in other plant species. Such genes could be identified by hybridisations, restriction fragment length polymorphism (RFLP) and other methods known in the art.

Genes or other DNA sequences, either natural or engineered, would be identified by hybridisation under stringent conditions (for example 35°C to 65°C in a 0.9M salt solution) using the MSI gene or fragments of it.

DNA sequences from other plant species fall within this patent provided they satisfy the hybridisation criteria, or would do so except for degeneracy in the genetic code.

The MS1 gene lies on the reverse strand of clone AB07561 and was not previously identified from sequence annotation. The MSI gene shows no direct homology to any previously identified genes involved in floral development or male sporogenesis. However, limited homology was detected to an Arabidopsis hypothetical protein (AC006069) from chromosome 2 (Lin et al., 1999).

The MM gene encodes for an unstable protein containing 672 amino acids. Prediction of protein localization sites show a putative nuclear targeting sequence at the N terminal region. Motif analysis of the deduced MM protein shows significant homology to the PHD finger motif (Asaland et al., 1995). This is a relatively rarely reported motif which is strongly conserved in a number of homeodomain proteins. It is a cysteine-rich domain which has a regular spacing between cysteine and histidine residues implying a possible metal binding function. PHD finger motifs (C4HC3 zinc-finger like motifs) have been reported in a diverse range of organisms ranging from humans (HRX), yeast, Drosophila

(trithorax), C. elegans to plants including Arabidopsis, alfalfa and maize.

The homologous proteins have diverse functions, but all are involved in transcriptional regulation. This motif is thought to be specifically involved in chromatin-mediated gene regulation (Asaland et al., 1995).

Wild type plants were analysed for the expression patterns of the MSI gene. No signal could be detected by Northern hybridisations, but RT-PCR showed low levels of expression in flowers and extremely low signal from stem tissues (see earlier description of RT-PCR). No expression was seen in leaf tissues.

The protein motif homology, the presence of a nuclear targeting sequence and the predicted instability of the MS1 protein is supportive evidence of the role of this protein in trancriptional regulation. The MSI gene therefore seems likely to be a critical factor associated with the regulation of gene expression during the later stages of anther and pollen development. As such the MS1 gene provides a valuable tool to analyse the molecular processes of anther and pollen development and will also provide a means to facilitate the future control of crop fertility.

The invention will now be described with reference to the drawings, of which:- Figure la shows the phenotype of the MSI mutant; Figure lb shows that full seed set can be achieved by cross- fertilisation of sterile flowers with wild-type (wt) pollen; Figure lc shows an Ler wt anther, viable pollen is clearly seen stained purple;

Figure 2a shows MSI pollen mother cells (PMCs) in meiosis, development is seen in the wt with PMCs surrounded by a layer of callose; Figure 2b shows MS1 isolated microspores (m) after release from tetrad-at this stage they appear similar to wt; Figure 2c shows Ler wt anther after microsphere (m) release prior to tapetal degeneration. Tapetal cells (t) have a granular appearance. Pollen mitosis is thought to occur at this point; Figure 2d shows MSI anthers at an equivalent stage to (c) in the MS1 mutant-the microspheres (m) enlarge and develop a granular, vacuolated cytoplasm and the tapetum (t) becomes abnormally vacuolated; Figure 2e shows an MSI anther at a slightly later stage-the tapetal cells and microspores enlarge further and they cytoplasm degenerates (d); Figure 3 shows a molecular-genetic map of the MSI region. The map corresponds to the sequenced AB007651 (MWD9) clone. The MS1 gene lies in the. region 6.4-9.8kb region of this clone.

Molecular markers were mapped using Ler/Sn (5)-1 recombinant lines amarkers co-segregate with MSI, bmarker maps 0. 14cM from MSI towards ttg ; Figure 4a shows the structure of MSI gene. Dark boxes represent exons, open boxes represent untranslated 5'and 3'sequences; MSI mutation sites a: MS1. 5 (This deleted), b: MS1. 1 and MS1. 6 (gag2 to

A), c: MS1. 4 (Gl036 to A), d: MS1. 3 (C13110 to T), e: MS1. 2 (Cl500 to T). RT1. L and RT1. R correspond to the RT-PCR primer positions; Figure 4b shows the MM peptide sequence: NTS = nuclear targeting sequence, LZ = leucine zipper, PHD = plant homeo domain motif. The numbers show the amino acid numbers of the region of the protein; Figure 4c shows the truncated proteins produced by the MSI mutant alleles, unfilled boxes represents the equivalent amino acid sequence to wt, dark boxes show novel amino acids; values represent the relative numbers of each; Figure 5a shows MM RT-PCRs of wt and MSI splice mutants.

Lanes 1-4 floral cDNA and 5-8 floral RNA controls; 1 and 5 = Ler, 2 and 6 = MS1. 2,3 and 7 = MS1. 1, 4 and 8 = MS1. 6, u = unspliced MM transcript (506bp); s= spliced MM transcript (351bp); Figure 5b shows RT-PCRs showing tissue specific expression of MM in Ler wt tissues. Expression is only seen in cDNA prepared from closed buds. Lanes 1 = leaf cDNA, 2 = stem cDNA, 3 = cDNA, 4 = open flower cDNA, 5= leaf RNA control, 6 = stem RNA control, 7 = bud RNA control, 8 = open flower RNA control ; Figure 5c shows control actin-2 RT-PCRs showing expression in all Ler wt tissues. Lanes 1 = leaf cDNA, 2 = stem cDNA, 3 = bud cDNA, 4 = open flow cDNA, 5 = leaf RNA control, 6 = stem RNA control, 7 = bud RNA control, 8 = open flow RNA control and m= marker;

Figure 6a shows whole mounts of entire Ler inflorescence showing specific GUS staining in closed buds, around the stage of microspore release (arrows); Figure 6b shows anther squashes of Ler buds showing GUS expression from the tapetal tissues (arrow); Figure 6c shows sections through Ler anthers showing GUS expression from the tapetum (t). GUS staining is also seen within the locule and the developing microspores (m). This may be due to low levels of expression within the gametophytic tissue, or to release of the GUS protein during tapetal degeneration and deposition on the microspores during pollen maturation; Figure 6d shows a section through Ler anther after complete tapetal breakdown. No GUS staining is seen at this stage. = 50 microns; Figure 7 shows the PHD-finger motif of MS1. Alignment of the MS1 protein and a number of other PHD-finger containing regulatory proteins (aligned by Pfam, Sanger Centre). HT3.1, HX1A ; YDBB; Acc. No. 10362Y, YK09; Acc. No. P36124, YJK5 ; HRX, Acc. No. Q03164. Consensus symbols are: #, strongly conserved hydrophobicity; $, semi-conserved hydrophobicity.

Phenotype of the MS1 mutant The MSl. 1 mutant was generated by EMS mutagenesis of Arabidopsis thaliana Landsberg erecta (Ler) seed (Van der Veen and Wirtz, 1968J : It is a recessive mutation and appears as wild type (wt) in the heterozygous

state. Homozygous MS1 mutants fail to produce viable pollen (Figl), with pollen degeneration occurring soon after microspore release from the tetrads. Young buds show characteristic agglutination of immature pollen which appears viable using FDA staining (Dawson et al. 1993a), but can not be induced to form pollen tubes (data not shown); gentle pressure on the buds at this stage result in the degenerating pollen being exuded in as a"sausage-shaped"mass. Complete degeneration of the immature pollen then follows (Fig 2E, F), so that no pollen viable, or inviable, is present in the mature locule (Fig 1D). The early stages of callose production, pollen meiosis and tetrad formation occur normally, followed by callose dissolution and microspore release (Fig 2A, B). Initiation of exine synthesis appears normal, however, immediately after release from the callose wall, the microspores develop an unusually granular, vacuolated cytoplasm (Fig 2D), as compared to Ler wt (Fig 2C). At this stage the tapetum becomes abnormally vacuolated and cytoplasmic degeneration begins. The microspores appear to adhere to each other, collapse and the cytoplasm degenerates (Fig 2E, F). In mature flowers the microspores and tapetal cells have undergone complete degeneration resulting in an empty locule. In all other aspects of plant development the MS1 mutant is completely normal (Fig 1); female fertility is unaffected and the mutation can be maintained by crossing using wt or heterozygous pollen.

We have isolated a series of five alleles of MSI, using EMS (MS1. 2, MS1. 3 and MS1. 4), X-rays (MS1. 5) (Dawson et al., 1993a) and gamma-rays (MSI. 6) (Kalantidis et al., 1994). All allelic lines exhibit the same phenotype, of complete male sterility regardless of growth under a range of different environmental growth conditions (data not shown).

Mapping and Localisation of the MS1 gene

We have previously mapped the MM gene to 29.8 0.8 cM on the top-arm of chromosome 5 and identified a marker, NIT4 (Bartel and Fink, 1994), which cosegregates with MM (Thorlby et al., 1997). This region corresponds to the sequenced clone AB007651 (MWD9) (Nakamura et al., 1997). Two PCR primers, CAPS12.5 and CAPS2, were designed using the AB007651 sequence to generate polymorphisms to confirm the location of MSI. Recombinants that had been previously generated between the Ler line carrying the mutations MSlcombined with lu, or ttg, and the alternative ecotype Sn (5)-1 (Thorlby et al., 1997) were used for fine-mapping. Amplification and digestion with a range of restriction enzymes was carried out on DNA from the parental, Ler and Sn (5)-1 lines, and recombinants on either side of MM (Thorlby et al., 1997).

Thirteen recombinants on the ttg side of MSI were analysed. These had breakpoints on average every 0.07 cM between the pCTT718 marker, which maps 0.9 cM from MM, and MSI. Three lu recombinants with breakpoints on average every 0.1 cM between an RFLP marker derived from the 16kb NotI fragment from the BAC clone 4D5 (16kb-4D5 marker maps 0.3 cM from MSI towards lu) and MM (Thorlby et al., 1997) were also used.

The CAPS12.5 primers gave a 1. 5kb band from both the Ler and Sn (5)-1 lines, however a polymorphism corresponding to approximately position 64 kb in the AB007651 clone, was seen when the fragments were digested with XbaI. The Sn (5)-1 CAPS12.5 product gave two bands of 1.1 and 0.4 kb, whilst the Ler band (and Columbia; AB007651 sequence) lacked the XbaI restriction site and gave a single 1.5 kb band. Eleven of the ttg recombinants gave the Sn (5)-1 phenotype with the CAPS12.5 marker, whilst the remaining two ttg recombinants exhibited the Ler phenotype, indicating they had breakpoints beyond CAPS12.5 and close to MS1. As expected, in all three of the lu recombinants the CAPS12.5 marker

co-segregated with MM (Sn (5)-1 phenotype). The CAPS12.5 marker was therefore calculated as mapping 0.14 cM from MM, towards ttg (Fig 3).

The CAPS2 primers gave a 6.7 kb band from both Ler and the Sn (5)-1 lines. When digested with SalI the Sn (5)-lproduct (and Columbia; AB007651 sequence) gave a 5.641 kb and a 1.055 kb band, whilst the Ler product was uncut (6.696 kb). This polymorphism corresponded to position 39.425 kb in the AB007651 clone. Recombinants on both sides of MM showed co-segregation (Sn (5)-1 phenotype) of the CAPS2 marker with MM (Fig 3).

Three markers, NIT4,9B3 (Thorlby et al., 1997) and CAPS2, that co-segregate with MM have been identified and these markers map to 18.3 kb, 37.1 and 43.6 kb respectively, on the AB007651 clone (Fig 3).

Additionally a marker 16kb-BAC 4D5 was found to map 0.3cM away from MSI towards lu. This marker was generated from the 4D5 BAC that overlaps with the AB007651 clone. Therefore, it was hypothesised that MM would lie in the first part of the AB007651 clone. The CAPS12.5 marker maps 0.14 cM from MM, which equates to position 64 kb on the AB007651 clone, and delimits the ttg side of the mapped region. The MSI gene was therefore predicted to lie between 0-64kb on the AB007651 clone.

The AB007651 clone was analysed for ORFs (GeneFinder; ACeDB) in this region that might correspond to MM and twenty-seven putative ORFs were identified; no prediction of which was MSI could be made based on putative protein homologies. Primers (Table 1) were therefore designed to amplify, by long-PCR across the AB007651 clone and by targeting to specific ORFs within the clone, using the Ler wt and allelic MSI mutants.

Given that two of the allelic lines were generated by X-ray and gamma irradiation, it was expected that these would contain deletions and that

polymorphisms in the PCR products associated with the MS1 gene would be detected.

Polymorphisms associated with the MSI mutation were not seen with any of the PCR products, except for the 2870bp product generated with primers 14/68. This corresponded to the 6920-9789bp region of AB007651. Loss of an 883bp Rsal restriction site was seen with the 14/68 PCR product from the MS1. 1 and MS1. 6 mutants, when compared to Ler wt. This region was sequenced and shown to correspond to a G to A transition, which resulted in the loss of the Rsal restriction site (Fig 4A).

This mutation is associated with a splice site in a potential ORF to confirm that this was the MS1 gene, the region corresponding to this ORF was amplified and directly sequenced from Ler and all of the allelic lines.

Sequencing revealed aberrations in this gene in all of the MS1 alleles, resulting in splice defects or truncated polypeptides (Figs 4A, C and Sequence Listing 4).

RT-PCR Three alleles, MSI. 1, MS1. 6 and MS1. 4 showed mutations associated with the splicing of the second intron. RT-PCR was conducted to confirm this.

The forward RT-PCR primer was designed to span the first splice site and the reverse to lie in the third exon (Fig 4A). The Ler wild type produced the expected RT-PCR product of 351bp, whilst the splice mutants MS1. 1 and MS1. 6 gave products of 506bp (Fig 6A) due to the lack of splicing caused by the G to A transition at the splice donor site. Direct fluorescent sequencing (ABI310, Perkin Elmer Biosystems) of all the allelic and Ler RT-PCR products showed the expected lack of splicing in the MSI. 1 and MS1. 6 mutants. No difference in RT-PCR product size could be detected by electrophoresis for the MS1. 4 mutant, however sequencing of this

RT-PCR product showed a 1bp deletion from the spliced message, resulting in a frameshift mutation and a premature truncation of the MSI protein (Fig 4C). This corresponds to the predicted mutation, observed from genomic sequencing, of a G to A transition at the splice acceptor site (Fig 4A).

Since six independent MM lines all show mutations in this gene and differences in RT-PCR products were seen in the putative splice mutants, we therefore conclude that this ORF corresponds to the MM gene. The MS1 gene lies on the reverse strand of clone AB07561 (Figs 3 and 5) and was not previously identified from sequence annotation.

RACE-PCR was utilised to identify the transcription start site of the MS1 cDNA. This was shown to lie 103bp upstream of the translation start (Sequence Listing 4). Sequencing of the Ler wt MSI cDNA supported the computer predictions of a 2019bp translated message, containing three exons with a 73bp untranslated 3'region prior to the polyadenylation tail (Sequence Listing 4).

MS1 Expression Patterns Wild type plants were analysed for the expression patterns of the MM gene. No signal could be detected by northern hybridisations (data not shown). RT-PCR using 45 cycles showed reproducible MSI products with low levels of expression seen in closed flower buds, but with no expression seen in open flowers (Fig 5B). No expression was seen in leaves or stem tissues. Control RT-PCR reactions using Arabidopsis actin-2 specific primers showed similar levels of high expression in cDNA prepared from leaves, stem, buds and open flowers (Fig 5C).

MM : GUS promoter fusions showed similar results to those seen from the RTPCR expression studies. Levels of reporter gene expression were low as expected, given the low levels of expression as detected by RT-PCR.

The reporter gene expression patterns reflected the phenotypic observations of the mutant plants, with GUS staining seen around the stages of microspore release from the tetrads (Fig 6A, B, C). Expression was not seen in younger anthers that did not show clearly defined microspores, or in anthers once the tapetal tissue had degenerated (Fig 6A, D). Maximal expression was seen from the tapetal tissues (Fig 6B, C) although GUS staining was also apparent within the locule and the pollen grains (Fig 6C). GUS staining was evident in the tapetal tissues at a late stage where tapetal cell walls were not clearly defined due to the onset of tapetal breakdown. No GUS staining was see in older anthers where complete tapetal degeneration had occurred (Fig 6D). GUS staining was not seen in vegetative tissues.

MS1 Sequence Homology The MSI genomic sequence shows no direct homology to any previously identified genes involved in floral development or male sporogenesis.

However, limited homology was detected to an Arabidopsis hypothetical protein (AC006069; TREMBL: Q9ZUA9) from chromosome 2 (Lin et al., 1999) which has a PHD-finger motif. Homology (29% ; Expect value of 4e-84) is seen throughout the protein and extends beyond the PHD-finger motif. The Q9ZUA9 protein has been identified as part of the genomic sequencing effort and has to date no defined expression pattern or function.

The MS1 gene encodes for a potentially unstable protein (ProtParam) comprising of 672 amino acids. Prediction of protein localization sites show a putative nuclear targeting sequence (Fig 4B) at the N terminal

region. Motif analysis of the deduced MS1 protein shows significant homology to the PHDfinger motif ( (Asaland et a1., 1995) Fig 7). This relatively rarely reported motif is strongly conserved in a number of homeodomain proteins. It is cysteine-rich with regular spacing between cysteine and histidine residues, implying a possible metal binding function. PHD-finger motifs (C4HC3 zinc-finger like motifs) have been reported in a diverse range of organisms ranging from humans (HRX), yeast, Drosophila (trithorax) and C. elegans, to plants including Arabidopsis, alfalfa and maize. The homologous proteins have diverse functions, but all are involved in transcriptional regulation. This motif is thought to be specifically involved in chromatin-mediated gene regulation (Asaland et al., 1995).

The gene MM may be used to control fertility in a number of ways:- a) The MM gene or fragments of the MSI gene can be cloned into a plasmid for example pBluescript, pJIT60. Expression can either be controlled by the endogenous MM promoter or the CaMV35S Alternatively, or in addition, male sterility can be achieved by disrupting the proper expression of the MM gene, for example by transcribing RNA which is antisense to the RNA normally transcribed from the MSI gene. b) Antisense expression of the MS1 gene or part of the gene may also be used in a controlled expression using an inducible promoter, for example stress inducible, chemically inducible.

A construct similar to that described in (a) can be produced but using an inducible promoter such as described in (d).

c) Further, the MSI promoter can be used to drive male sterility DNA such as that coding for a nuclease, protease or glucanase.

The MM promoter ensures tissue specific expression in the plant. This can be utilised with other genes known to cause cytotoxic effects in plants. For example, nuclease, protease or glucanase, or more specifically barnase expression, as described in Patent EP-A-0 344 029 (Plant Genetic Systems) and has been published by Mariani et al. Nature 347: 737-741.

This approach is also described in US patent 5 955 653 which describes combining the barnase gene or the actinidin gene to the A6 callase promoter from Brassica napus. d) Inducible expression of MM gene in male sterile background to switch on fertility Inducible promoters The promoter region is removed from the MS1 gene and replaced with a promoter that only responds to a specific external stimulus. Thus, the gene will not be transcribed except in response to the external stimulus.

Since male sterility results if the MSI gene product is not produced plants which carry the inducible MS1 construct and not the wild type versions of the MM gene will be sterile until gene expression is switched on.

An example of a responsive promoter system that can be used to control gene expression is the glutathione-S-transferase (GST) system in maize.

The GST family of enzymes detoxify a number of compounds used as pre- emergent herbicides (Wiegand, et al, 1986). It has been discovered that treating maize seed with GSTs results in increased tolerance to the herbicides. This is brought about by a naturally occurring maize gene that responds to the GSTs and induces gene expression. This promoter for this

gene, which has already been identified and cloned, could be used to inducibly express the MSI artificial gene construct.

Native gene inactivation To permit sterility to be controlled by inducible expression of the MS1 gene, the endogenous wild type MS1 gene requires inactivation. This can be achieved by a number of methods known to the art. For example:- i) By homologous recombination (Yoder, and Kmic, 1991). Gene targeting by antisense gene expression can be utilised to inactivate the endogenous gene. Backerossing can be used to selectively cross the gene into the desired plant to replace the wild type gene; ii) Targeted mutation systems can be used to knock out the wild type gene e. g. utilising transposable element. Plants can thus be selected that are male sterile, the engineered inducible MS1 gene can then be transferred into this plant using appropriate well established methods, e. g. Agrobacterium transformation, biolistics, electroporation, and fertility restored by GST spraying.

Advantages of MS1 to control fertility:- i) Highly specific control of male fertility which does not effect other processes in the plant as visualised by the unaltered phenotype of other aspects of the MSI mutants; ii) Sterility is maintained over a range of environmental conditions and is not effected by abiotic stresses;

iii) Naturally occurring system of male sterility which is being harnessed to provide biotechnological control of fertility; iv) As a transcription factor MS1 may have a key controlling role for regulating other genes associated with pollen and anther development; therefore breakdown of the sterility to restoration of fertility is unlikely.

Isolation of the MS1 gene provides a means to artificially construct trangenes which will result in the control of fertility in plant species. The molecular weights quoted are putative based on the numbers of amino acids present, as deduced from the DNA sequence. The 77kD MS1 protein from Arabidopsis thaliana has 672 amino acids and the molecular weights given do not involve any glycosylation or post-translational events.

The introduction of the MS1 gene into selected host plants to confer male sterility can be achieved by the methods described above or by other methods known to those skilled in the art. For this purpose, reference can be made inter alia to US Patents 5 859 341 and 5 955 653, and 6 018 101, the disclosures of which are hereby incorporated by reference.

It will be appreciated that we seek protection not only for the DNA material of MSI specified in Sequence Listing 1, the cDNA material of Sequence Listing 3, plant gene promoter specified in Sequence Listing 4, isolated DNA and cDNA material as specified in Sequence Listings 1 and 3, and a protein having the sequence of Sequence Listing 2 (and their uses in controlling male sterility in plants), but we also seek protection for variants of the DNA, cDNA and protein which have an equivalent effect (and the use of such variants to control male sterility in plants). We also seek protection for plants, plant cells, plasmids, vectors, or plant seeds with male sterility controlled using the DNA, cDNA or protein sequences of Sequence Listings 1,2 and 3 (and equivalent variants of the aforementioned DNA, cDNA or protein).

APPENDIX 1 Table 1. Primers designed to amplify across the AB007651 (MWD9) clone.

Primer Forward primer Reverse primer Region of Size name sequence 5'-3'sequence 5'-3'AB007651 PCR amplified product (bp)(bp) 1 TTGGTGACTTTATTGGCAGC CTCGACGTTCTCCGCCGTG 313-5213 4900 TGAAGCG G 2 CAACGGGCCATCCCTGCGA CGCAGCCGGAGACTCGTCG 38372-45067 6696 A AA 3 CCACGGCGCCTGTGTTTCC TGAAAAAGCGTGTCATGGT 44500-53575 9076 C GATGCAA 4 TGGTTGTATGGACTTGGGG TCAGTGGCCAGTGGGCAGA 53001-62064 9064 TGATGGA ACAA 5 CCCACTGGCCACTGAGCTG TCGCCCGGAACAGGACGA 62050-69353 7304 GAGG cnr 6 TGGTGA=AfTGGCAGC-r TGTCCAACCACGCAAACAC 314-9789 9475 GAAGCG ACACG 7 TCTITCGCC'ITLTGGTGCAC CGCTTCAAACGTGCAATIT 15838-24194 8356 GACC TCGCC 8 TTCTCGCCGACACGCAGCT TCGGAAGGCCGCTTGTGAA 24023-35511 11488 TTTCT CTTTG 9 GTGAGGACAATCAGGGGC AGCAATCTAAGCGCCTTCG 35140-43786 8646 AGTGGC CCGAG 10 CTCCAAAGATGACCGTCGC CCCAACAGGACCAGCAAG 65962-76300 10338 GGAAA CAGAGG 11 GCAAAGCTCTTGGGAGGCC AGAAAATCCGTTTGCGGAA 73316-82790 9474 CTGAG GGCGA 13 TGCTCCATITCCGGTTTACC GTCGCTCTCCTCTTCCCCTG 3348-6451 2251 A A 14/6R GCCATTTCATCGACCCTTGT CGTGTGTGTTTGCGTGGTTG 6920-9789 2870 G GACA 15 ATCACCCGTTGCAAAGAAT TATCATTI'GTTGGCCGTGCG 11153-12583 1451 CAA T 16 TGACTCAAAAGGCAAGTG GCCGATITCTTACGCGCTCT 13684-16482 2799 AAGCAG T 17 GCGCTAGCGGTGGTGATTAC TTCACTTGCTCGTCGAAAC 22656-24213 1558 CG 18 CCCTAGCCGTTGGATGTTAC TCACGGTTGGGAACAGAG 27034-29444 2411 ACA 19/8R GGCCGTCTCACTCTCTCGCT CAAAGTTCACAAGCGGCCT 33417-35511 2095 A TCCGA 20 GCCTCAGTGTTCAAAGAAT GTATCGGTCTCGTTCGTGCG 37624-39540 1917 ATGGG 21 GCGCCTAGCAAAAACTGA TGTGTTGAGGATGAGAAGG 39375-40905 1531 CGA ATGG 22 CATCTCGTGGAAAACAGGA CCAGAATGCTTAGCACTTG 40526-42779 2254 TCG TAGTTGG