Searching for tagged male sterile mutants of Arabidopsis

Julie A. Glover (1), Katherina C. Blomer (2), Leigh B. Farrell (3), Abdul M. Chaudhury (4), & Elizabeth S. Dennis (4*)

1 & 4. CSIRO Division of Plant Industry,
GPO Box 1600,
Canberra ACT 2601 & Cooperative Research Centre for Plant Science,
PO Box 475,
Canberra ACT 2601 & Division of Biochemistry and Molecular Biology,
Australian National University,
Canberra ACT 0200,
Australia.

2. Present address: Peter MacCallum Cancer Institute,
Research Unit,
St Andrews Pl,
East Melbourne VIC 3002,
AUSTRALIA

3. Present address: Gene Shears Pty. Ltd.
PO Box 1238,
Neutral Bay NSW 2089,
Australia.

* Author for correspondence

Abstract

Although many male sterile mutants have been identified in Arabidopsis, few genes responsible have been cloned. In order to facilitate cloning of a male sterility gene, 23 of Feldmann's T-DNA-generated, reduced fertility lines were screened to identify a tagged male sterile mutation. Male sterile mutants were identified, as well as mutants which were both male and female sterile. Segregation of the kanamycin marker gene in the progeny of 15 of these lines was studied. 40% had functional T-DNAs inserted at a single locus, the remainder segregating for two or more functional T-DNA inserts. Linkage between T-DNA inserts and mutant phenotype was tested for six lines. Mutations in three of these lines were not linked to a T-DNA insert, three lines were identified where the mutation was segregating with a T-DNA insert.

Introduction

Fertile pollen is the culmination of physiological, biochemical and morphological processes requiring the coordinated temporal and spatial expression of a large number of gametophytic and sporophytic genes. These genes interact to control the formation of the stamens, differentiation of the sporogenous cells and meiosis, the post-meiotic development of free microspores, microspore mitosis, pollen differentiation and anthesis. Although mutants which have defects in one or more of these processes have been characterised in many different plant species (Kaul, 1988), very few of the corresponding genes have been isolated. Isolation of these genes, and characterisation of their function, expression and interaction will provide important understanding of the process of pollen formation. Male sterility is an important tool in plant breeding as it prevents self-pollination, which is known to cause inbreeding depression. Male sterility has been exploited in the production of hybrid seed as crosses between inbred plant lines often result in progeny with increased yield, enhanced disease resistance and better ability to endure environmental stresses, a general phenomenon known as hybrid vigour.

Many male sterile mutants of Arabidopsis have been characterised (Van der Veen and Wirtz, 1968, Goldberg et al, 1993, Preuss et al, 1993, Dawson et al, 1993, Chaudhury et al, 1992, Chaudhury et al, 1994, Regan and Moffatt, 1990, Hülskamp et al, 1995, Peirson et al, 1996, He et al, 1996) but only two of the genes responsible have been identified. One of these mutants lacks the adenine phosphoribosyl transferase enzyme which converts adenine to AMP, showing the importance of the purine salvage pathway during microspore development (Moffatt and Somerville, 1988). Another Arabidopsis male sterile mutant in which the mutated gene has been identified is ms2 which was generated by tagging with a maize transposable element (Aarts et al, 1993). The MS2 gene shares a short stretch of homology with a wheat mitochondrial gene. This may be significant as the mitochondrial genome is involved in most cases of cytoplasmic male sterility (Vedel et al, 1994). However ms2 is a nuclear encoded mutation, so the significance of this homology has yet to be clearly understood.

Recently a large population of T-DNA transformed plants has been produced by Feldmann and colleagues using a procedure known as seed transformation (reviewed in Forsthoefel et al, 1992). They screened the progeny of these transformants for visible differences in phenotype and identified many mutants, including plants with reduced fertility. These reduced fertility lines continued to flower later than wild-type plants and produced few (if any) seed (Feldmann, 1991). Among these mutants are male sterile mutants where the mutated gene is tagged by a T-DNA insert (Peirson et al, 1996, He et al, 1996). We obtained 23 of these reduced fertility mutants as part of our quest for identifying genes involved in pollen production in higher plants. We report here the results of screening these lines for T-DNA tagged male sterile mutants.

Methods

Plant Lines

T-DNA transformed lines, ecotype Wassilewskija (Ws), were kindly provided by Kenneth Feldmann, University of Arizona. These lines were generated by seed transformation (Feldmann and Marks, 1987) and categorised as containing reduced fertility mutants (Feldmann, 1991).

Kanamycin selection

Seeds were surface-sterilised with 33:66:1 bleach/sterile H2O/10% SDS for 15 min and rinsed several times in double-distilled H2O before plating onto MS growth medium (Murashige and Skoog, 1962) containing 3% sucrose, 0.4% (w/v) agar and 50 (g/ml kanamycin (kan). In order to overcome dormancy plated seeds were kept in the dark at 4° C for at least 24 hours and then grown at 22 °C in 16h light/8h dark cycle with an irradiance of 150 µmol quanta.m-2.s-1 PAR. After 7-10 days plants were scored for kan resistance. For some mutant lines kanR plants were then transferred to MS plates and grown a further 10 days and then transferred to soil and grown to maturity and the fertility phenotype scored.

Phenotype

T3 seed from Feldmann's mutant lines were grown on soil, selfed and T4 seed collected from individual plants. These T4 families were grown at 22 °C under continuous fluorescent illumination in soil mix and self-sterile plants were easily detected in a population of self-pollinating plants because of difference in silique length. Female fertility was then tested by applying pollen from a wild-type fertile plant to the flowers of a self-sterile plant; the mutant was scored as female fertile/male sterile if outcrossing produced viable seed. Mutant plants from some lines were checked for the presence of pollen under a light microscope. If pollen was detected, these plants were used as pollen donors to determine pollen viability.

Southern analysis

Mutant lines which were shown to contain few inserts were subjected to southern analysis. Genomic DNA was isolated from plants essentially as described by Dean et al, 1992, except that purification on a CsCl cushion was added (Taylor and Powell, 1982). 2 - 5 µg of DNA was digested with 60 units of the restriction endonuclease EcoRI in a 100 µl reaction volume containing 1 mM spermidine for up to 18 h at 37 °C. Restricted fragments were separated on an 0.7% agarose gel and transferred onto Hybond N+ (Amersham) nylon filter and fixed in 0.4M NaOH for 20 min. The right border T-DNA probe corresponded to the HindIII fragment 23 from pTiC58 containing the nopaline synthase (NOS) gene (Figure 1). The pBR322 plasmid was also used to detect the presence of the T-DNA. Probes were radiolabelled using random oligo priming (Amersham). Hybridisations and washes were done at 65°C either as in Schmidt et al (1992) or Church and Gilbert (1984).

Results

Phenotypic Analysis

Since we completed these experiments, Feldmann has made these lines publicly available through the Arabidopsis Biological Resource Centre (ABRC) at Ohio State and the Nottingham Arabidopsis Stock Centre (NASC). The ABRC has published limited characterisations of the phenotypes of these line numbers in their Seed and DNA Stock List (February 1995). For most of the lines our results matched their data (Table 1).

Table 1: Phenotype of the Reduced Fertility Lines Examined (text version of table 1)

MUTANT LINE	ABRC PHENOTYPE	OUR PHENOTYPE
178	Male sterile	Male sterile/ pollenless
243	Male sterile	Female sterile or male/female sterile
547	Male sterile	Partially male sterile
783	Partially fertile, small anthers reduced pollen	Female sterile or male/female sterile
871	Male sterile	No phenotype detected
932	Male sterile/flower mutant	Partially sterile/dwarfed
1097	Male sterile	Pollenless
1180	Male sterile	Male sterile
1569	Male sterile	Partially sterile/dwarfed/bushy
1728	Reduced fertility mutant	Male sterile/pollenless
1746	Male sterile	No phenotype detected
1885	Male sterile	Male sterile/pollenless
1926	Male sterile	Male sterile - dehiscence mutant
2379	Male sterile/root mutant	Pollenless
2415	Male sterile	Female sterile or female/male sterile
2522	Male sterile	Male sterile/partially female sterile
2836	Partially male sterile	Partially male sterile
3181	Male sterile	No phenotype detected
3529	Male sterile	Partially male sterile
3914	Male sterile/ yellow green	Pollenless/partially female sterile
4416	Anther-affected/embryo defective and size mutant	Not determined
4791	Anther -affected	Not determined
4838	Male sterile	Partially male sterile

We could not detect any mutant plants segregating in three of the lines 871, 1746, and 3181, even though at least 30 plants from each line were examined. The ABRC Stock centre reports that these lines have low frequency of mutations and our data support this.

Mutants in two of the lines, 932 and 1569, had altered overall plant morphology in addition to fertility abnormalities. Many of the mutants that were scored in Feldmann's screen as reduced fertility were also identified in other mutant screens. For example, in addition to being isolated in the fertility screen, 2379 was scored as short root in the agar screen, 932 was scored as a flower mutant, and 4416 was scored as a embryo defective and a size variant. These may represent either pleiotropic phenotype due to mutations of the same gene or a number of independent mutations in the same line. As only small numbers of these plants were examined - it was difficult to differentiate between these two possibilities.

Many of the lines showed partial male sterility phenotypes (547, 2836, 3529, and 4838), which were not analysed further as identification and characterisation of these mutants was very difficult.

We showed some lines, 243, 783 and 2415, to have reduced female fertility, as applying pollen from a wild-type plant did not induce full seed set. As the reverse test wasn't performed, that is using the mutant as a pollen donor on wild-type flowers, the possibility still exists that mutant plants from these lines could be male sterile as well. This seems likely as the ABRC seed list reports these lines to be male sterile. In addition, we identified two mutant lines, 2522 and 3914, in which plants, as well as being male sterile, showed reduced female fertility. The remainder of the lines studied were found to be male sterile.

Scoring for kanamycin resistance

Figure 1 shows the T-DNA construct that was used to transform these lines. It contains, within the T-DNA borders, the neomycin phosphotransferase II (nptII) gene from Escherichia coli which detoxifies kanamycin by phosphorylation (Bevan et al, 1983). Therefore the presence of a T-DNA insert can be detected, in most cases, by scoring plants for kanamycin resistance. The nptII gene segregates as a dominant marker, that is plants containing only one copy of the gene express resistance levels indistinguishable from plants homozygous for the marker gene.

Figure 1:Structure of T-DNA in Ti plasmid 3850:1003
Diagram of the T-DNA region. AmpR, E. coli beta-lactamase gene; ORI, E. coli origin of replication; NPT1, kanamycin resistance gene with a bacterial promoter; NPT2, kanamycin gene with a dual plant promoter. Size of EcoRI restriction fragments are shown in kilobases. Region of homology to probe used to hybridise to filters in Figures 2,3 &4 is shown.

Over 35% of T2 plants produced by the seed transformation method contained more than one functional insert per transformed line (Feldmann et al, 1994). In order to determine the number of functional T-DNA inserts in the reduced fertility lines, seed was collected from at least 8 individual T3 plants from each line and 30 - 200 of these seeds (T4 families) plated onto media containing kanamycin. Approximately 10 days after germination, kanamycin resistant (kanR) plants could, in most cases, be easily differentiated from kanamycin sensitive plants (kanS) plants. The ratio of kanR to kanS plants in each family was determined, and (chi-squared calculated according to Mendelian segregation ratios for a dominant gene (P>0.05). Based on these tests, plants were designated as having one or more T-DNA insertions conferring kanR (Table 2).

Table 2: Number of T-DNA inserts as estimated from kanamycin segregation ratios of T4 progeny of the reduced fertility lines.(text version of table 2)

Line Number No insert 1 insert 2 inserts 3 or more inserts or non- segregating Other*

Ratio kanR:kanS All kanS 3:1 Between 3:1 and 15:1 63:1 or >63:1

178 19 2 1

243
3 1 4 2

547 3 7 1 4 1

1097 3 12 2 2 5

1180 9 4

1885 2 5 2 1 12

1926 17 9

2379 1 5 5
15

2415 2 1 1 7 6

2522 2 9

2836 6 2

3181 2 8

3529
3 3 5

3914 6 9

4416
4
1 3
* More plants need to be screened so that chi-squared analyses can differentiate between categories.

Line Number	No insert	1 insert	2 inserts	3 or more inserts or non- segregating	Other*
Ratio kanR:kanS	All kanS	3:1	Between 3:1 and 15:1	63:1 or >63:1
178	19	2	1
243		3	1	4	2
547	3	7	1	4	1
1097	3	12	2	2	5
1180	9	4
1885	2	5	2	1	12
1926	17	9
2379	1	5	5		15
2415	2	1	1	7	6
2522	2	9
2836	6	2
3181	2	8
3529		3	3	5
3914	6	9
4416		4		1	3

One of the mutant lines, 4791, was totally kanS in our hands. As this line was initially isolated due to its kanR, one explanation for this is that the nptII gene in this line has been silenced and is no longer conferring kanR. Another explanation is that, as we were working with T4 generation plants, the transgenic plants may have been lost during segregation through the generations.

In all the lines tested except one, kanR plants were easily distinguished from kanS plants. In this line, 1728, kan ratios were very difficult to determine as plants were only partially resistant to kan and were difficult to rescue from plates. Castle et al, 1993 showed that embryonic lethal mutants (also produced by Feldmann) that showed mottling on kan plates contained truncations near the 3 prime end of the nptII gene. This could also be an explanation for the partial resistance of this line.

From the kan segregation ratios of the T4 families it was possible to infer the number of T-DNAs integrated in the T3 plants. Most lines contained functional T-DNAs integrated into at least two sites in the genome (178, 243, 547, 1097, 1885, 2379, 2415, 3529 and 4416). For other lines the segregation pattern of either all kanS or three is to one kanR:kanS suggested the integration of a functional T-DNA at a single locus (1180, 1926, 2522, 2836, 3181, 3914), although the data is also consistent with multiple T-DNAs inserting at this single locus.

Cosegregation

The fertility mutants produced in Feldmann's transformation procedure could have resulted in a number of ways. A gene essential for plant sexual reproduction could have been disrupted by insertion of a functional or silent T-DNA insert within or proximal to this gene or else mutagenesis associated with the transformation process could have occurred. Mutations in the latter category would not necessarily linked to a T-DNA insertion.

A large number of unlinked mutations have been detected in T-DNA transformed plants (Koncz et al, 1992). In order to determine if any of the reduced fertility mutations were due to insertion of a T-DNA into a gene for male fertility, the segregation of the reduced fertility phenotype and T-DNA were followed. This T-DNA could either be functional and encoding for kanR, or not conferring kanR - where the T-DNA is inserted only partially or the nptII gene has been inactivated. Six mutant lines 178, 1180, 1885, 1926, 2379 and 2522 were chosen for further study on the basis of their phenotype and/or the number of T-DNA inserts.

Line 178

Most of the segregating T4 families from this line were kanS. Two were segregating for a single kanR marker, and one for two unlinked inserts. In order to determine whether the kanR was linked to the male sterility phenotype, a fertile wild-type plant was used as a pollen donor for a cross with a mutant plant from one of the families containing a single functional kanR marker gene. The F1 seeds were grown, seeds collected and F2 plants grown to seed. Seeds were collected from 104 individual F2 fertile plants. This seed was then used to score the F2 parent separately for kanR (by germinating on kanamycin plates) and for the sterility phenotype (by growing plants without selection in soil) to test for cosegregation. 66 of these F2 plants were shown to be heterozygote for the male sterility and heterozygous for the kan marker gene. 38 F2 plants were homozygous wild type plants, as well as being all kanS The results of this experiment show that among a large segregating population, no recombination events had occurred between the male sterility phenotype and the kanR marker. This suggests that the T-DNA tag encoding for kanR is very closely linked to the male sterility phenotype in this mutant line.

In order to detect silent inserts, DNA from pooled F3 plants representing each of the 104 fertile and heterozygote F2 plants was isolated along with DNA from 60 male sterile plants from the original T3 population of this line. This DNA was subjected to Southern analysis; the plant DNA was probed with the RB of the T-DNA. An example of the results is shown in Figure 2. All of the male sterile and heterozygous plants were kanR and had a 14kb EcoRI fragment hybridising to the right end of the T-DNA that was absent from all of the homozygous fertile plants. Another T-DNA insert was also present in this population, represented by a 7.2kb EcoRI fragment. This insert segregated independently of the mutation as it was absent in some male sterile plants and present in some homozygous fertile plants. Plants containing only this insert were kanS, indicating that this insert is not conferring kanR.

Figure 2: Representative DNA gel blot analysis of DNA from plants from Line 178. DNA was isolated from individual plants. The genotype of these plants was determined by scoring in the next generation for those plants that produced seed. Lanes are labelled as follows : MM - homozygous male fertile plants, Mm - heterozygous plants, and mm - homozygous male sterile plants. The DNA was cleaved with EcoRI, resolved on an agarose gel, transferred onto nylon, and hybridised to a radiolabelled probe derived from the right border of the T-DNA. Sizes (in kilobases) of the DNA fragments are shown on the left. DNA representing 210 individual F2 plants was analysed in this manner. The larger T-DNA hybridising EcoRI fragment segregated with the ms mutation in these plants.

Line 1180

The segregation of the kanR marker suggested integration of a functional T-DNA at a single locus. Southern hybridisations on individual mutant plants using T-DNA segments as probes (NOS and pBR322), revealed mutant plants without T-DNA inserts, suggesting this line to contain an unlinked mutation.

Line 1885

The segregation on kanamycin of plants from this line suggested that functional T-DNAs were integrated into at least two independent loci. Similar to line 1180 however, southern hybridisations on individual mutant plants using T-DNA segments as probes (NOS and pBR322) detected mutant plants without T-DNA inserts, showing that the mutation in this line was unlinked to a T-DNA insert.

Line 1926

Limited study of this mutant line suggests that it is a dehiscence related mutant as pollen is produced but not effectively released from the anther locule. Families from this mutant line segregated either 3:1 kanR: kanS, or 100% kanS, suggesting the T-DNA has integrated at a single locus. For initial linkage analysis - 11 kanR plants were transferred to soil and the phenotype scored. 7 of these were fertile and shown to be heterozygous for male sterility in the next generation, and the other 4 were male sterile. As this data pointed to the mutant phenotype being linked to the T-DNA - a mutant plant was crossed with wild-type and the F2 population subjected to southern analysis and scoring similar to that described for Line 178. 27 of these F2 plants were scored as male sterile and 79 as heterozygous plants. When DNA from these plants was hybridised with a probe corresponding to the right border region of the T-DNA, three T-DNA hybridising bands were detected in all 106 of these plants (Figure 3). These bands were absent in the 20 homozygous wild-type plants tested. Further analysis showed up to ten T-DNAs have integrated in a tandem and inverted repeat arrangement into a single locus in this line. The ms mutation was deemed to be linked to this T-DNA locus.

Figure 3: Representative DNA gel blot analysis of DNA from plants from Line 1926. DNA was isolated from individual plants. The genotype of these plants was determined by scoring in the next generation for those plants that produced seed. Lanes are labelled as follows : MM - homozygous male fertile plants, Mm - heterozygous plants, and mm - homozygous male sterile plants. The DNA was cleaved with EcoRI, resolved on an agarose gel, transferred onto nylon, and hybridised to a radiolabelled probe derived from the right border of the T-DNA. Sizes (in kilobases) of the DNA fragments are shown on the left. DNA representing 126 individual F2 plants was analysed in this manner. All three T-DNA hybridising bands segregated with the ms mutation in these plants.

Line 2379

Kan segregation of progeny from this line suggested that it contained at least 3 insertion sites of a functional kanR gene. When T4 families containing only one insert were grown non-selectively on soil and the phenotype determined, some families were segregating three fertile to one sterile plant in some of these lines, suggesting the mutation may be linked to a T-DNA insert (Feldmann, 1992). However, when Southern analysis was performed on individual sterile and fertile plants from this line, bands hybridising to the T-DNA were not present in all the male sterile plants (NOS and pBR322 probes were used for these experiments). These data show that this mutant is unlinked to a T-DNA insert.

Line 2522

Kan segregation suggested a functional T-DNA inserted at a single locus. Southern analysis of individual plants showed 26 sterile plants contained a single 23kb band that was also present in three out of four fertile plants (Figure 4). These three fertile plants were shown in the next generation to be heterozygous for the mutation and the one without the 23kb band was shown to be a wild type plant. These data are sufficient to indicate that this line may be a tagged mutation, because if the T-DNA and the mutation were segregating independently each mutant would only have a 75% chance of containing that 23kb band corresponding to the T-DNA (Feldmann, pers. comm.).

Figure 4: DNA gel blot analysis of DNA from plants from Line 2522. DNA was isolated from individual plants. The genotype of these plants was determined by scoring in the next generation for those plants that produced seed. Lanes are labelled as follows : MM - homozygous male fertile plants, Mm - heterozygous plants, and mm - homozygous male sterile plants. The DNA was cleaved with EcoRI, resolved on an agarose gel, transferred onto nylon, and hybridised to a radiolabelled probe derived from the right border of the T-DNA. Sizes (in kilobases) of the DNA fragments are shown on the left. 26/26 ms plants contained the 23kb band, along with 3/4 fertile plants.

Discussion

Phenotype

As reported previously, the reduced fertility mutants identified in Feldmann's massive screen could potentially be male sterile, female sterile or both female and male sterile. As Arabidopsis is self-pollinating, simple tests can distinguish between these three categories of mutants. If pollen from a wild-type plant is applied to a sterile plant and fails to cause silique elongation and seed formation, this indicates a blockage in female fertility. If pollen from this mutant plant fails to fertilise wild-type ovules then this plant can be identified as both male and female sterile. In two lines, 2522 and 3914, mutant plants were identified that were affected in both male and female fertility. These mutants could be meiotic mutants (Kaul and Murthy, 1985), however further analysis needs to be done to show this conclusively, as processes other than meiosis may require a common gene, for example post-pollen mitosis and megaspore mitosis. As the number of plants studied was small there is also a chance that there are two closely linked mutations in these lines.

The most common phenotype encountered among these lines was male sterility. Male sterility can be categorised into many different types including structural male sterility, microspore and pollen development mutants, dehiscence mutants and pollen function mutants (Chaudhury, 1993). Out of five of the male sterile mutants we studied, four of them appeared to be microspore development mutants (178, 1180, 1885, 2379) and one a dehiscence mutant (1926) (data not shown). Although not reported here, 178 has a conditional phenotype resulting from a complex interplay between genetic and environmental factors (in press).

Screening for sterility in the T2 generation will only identify sporophytic mutants. Male gametophytic mutations will produce 50% non-viable pollen in the T1 generation, so the egg cells will be fertilised only by the 50% viable pollen. Even though this type of mutation will be transmitted through the female line, all the progeny of the T1 plants will set seed, and will not be detected by a screen of this kind.

Number of insertion sites

The testing of around 1000 transformed lines generated from Feldmann's seed transformation procedure showed 57% of the lines segregated for a single functional T-DNA insert (ratio 3 kanR:1 kanS), and 35% for two or more inserts (ratio greater than 3:1) with 9% having non-Mendelian ratios (Feldmann et al, 1994). As fewer inserts generally facilitate analysis (and cloning) of a tagged line, we estimated the number of T-DNA inserts in 15 of the lines by kan segregation ratios (Table 2). From the results of these experiments, 6/15 (40%) of the tested mutant lines were segregating for a functional T-DNA insert at a single locus, whilst the remainder showed segregation ratios indicative of two or more functional T-DNA inserts. These results are comparable to those of other researchers. In some of the lines with two or more inserts it is difficult to say with certainty just how many functional inserts are present. Hundreds more seeds would need to be screened from the non-segregating lines to determine whether they are homozygous for the T-DNA insert or contain 3 or more inserts.

If some families are homozygous for the kanR marker gene this would indicate that the sterility phenotype was not caused directly by that T-DNA insertion. The logic behind this conclusion is simple - if the T-DNA was linked to the mutation, all plants homozygous for the T-DNA insert would also be homozygote for the fertility mutation, thus would produce few, if any seed. Thus families which did not segregate for the T-DNA insert would not be detected. It is on the basis of this idea that line numbers 1180, 1926 and 2522 were closely examined for linkage of the sterility phenotype to the T-DNA insert, as no kanR:kanR plants were detected in these lines.

Unlinked mutations

Similar to other transformation procedures, seed transformation generates a large number of mutations that are not associated with a T-DNA tag. Feldmann and others have reported that, in some cases 20-30% of lines examined contained mutations that were unlinked to T-DNA insertions. The analysis of the ms mutants obtained from Feldmann therefore required extensive studies to determine linkage between the T-DNA insertion and the mutation. Feldmann reports that any distortion of segregation ratios of either the mutant phenotype or the selectable marker suggests that the mutant is unlinked. As the lines, 871, 1746 and 3181 showed no detectable phenotype in our hands, and the report from the ABRC confirms that these mutations were low in frequency, these data point to these mutations being unlinked to a functional T-DNA insert. Southern analyses showed that three other mutant lines, 2379, 1180 and 1885, were unlinked to either a functional or a silent T-DNA insert.

Several explanations have been postulated for the high percentage of unlinked mutations in seed transformation of Arabidopsis. One hypothesis suggests aborted T-DNA integration into the plant genome. T-DNA integration is thought to occur by illegitimate recombination involving a double stranded break and repair mechanism (Gheysen et al, 1991, Mayerhofer et al, 1991). If this process is interrupted, the T-DNA may be removed by a repair mechanism leaving single-base pair changes or deletions ('footprints') in the nuclear DNA (Koncz et al, 1992). Weigel et al (1992) sequenced an 'untagged' allele of LEAFY from Feldmann's population and found that it contained a single base change which may have been generated in this manner.

Another way in which untagged mutants could be produced is by the transformation process activating an endogenous transposon. Transposons have been found in Arabidopsis, including a superfamily of retrotransposons Ta1-Ta10 (Konieczny et al, 1991) and the transposon-like element Tat1 (Peleman et al, 1991) but no evidence has been obtained yet to show that these elements are still active. More recently, a mobile endogenous transposon, Tag1 was also discovered when researchers were screening a T-DNA-Ac transformed mutant population of Arabidopsis for chlorate resistance mutants (Tsay et al, 1993). Tag1 inserted into the CHL1 gene, and its high activity was shown by the high reversion of this mutation to chlorate sensitivity. One hypothesis was that transposition of the Tag1 element may have been stimulated by the integration of the T-DNA (Tsay et al, 1993). The presence of the Ac transposase may also be enhancing the Tag1 activity.

Several researchers have reported Agrobacterium sequences other than the T-DNA have been integrated into plant genomes. Martineau and colleagues analysed several hundred plants from different plant species and found that approximately 20-30% of these contained Agrobacterium DNA from beyond the T-DNA region stably integrated into the plant genome (Martineau et al, 1994). Similarly the T-DNA region in an embryo lethal mutant of Arabidopsis generated by seed transformation was found to contain a portion of the virulence gene region (Castle et al, 1993). The integration of these non-T-DNA bacterial sequences could be responsible for some of the 'untagged' mutants.

Linked mutations

In these experiments we identified three T-DNA transformed reduced fertility lines from Ken Feldmann's population where the mutant phenotype was genetically linked to a functional T-DNA insert. We identified Line 178 to be segregating for a male sterile mutation as well as at least two T-DNA insertion loci. By determining kan ratios and detecting the presence of T-DNA hybridising bands on southern blots in a population of segregating plants, we established that one of these T-DNAs was genetically linked to the male sterile mutation, suggest that the male sterile mutation is tagged by this T-DNA.

The 1926 line also segregated for a male sterile mutation that was closely linked to a T-DNA insertion locus. However southern analysis from this line suggests that there are multiple T-DNAs inserted at this locus.

Another mutant line, 2522, was also identified which contained a single T-DNA insertion linked to the sterility phenotype, established by southern analysis on individual sterile and fertile plants. Mutant plants in this line produced abnormal pollen and also showed reduced female fertility in our hands.

Further Work

We report here three mutants of the fertility pathway of Arabidopsis that are potentially tagged by T-DNA insertions. Although allelism tests have not been performed between these lines, the very different phenotypes of these mutants suggest three different genes are involved. It should be possible to clone the genes corresponding to these mutations using the T-DNA as a tag.

The work on mutant lines 1926 and 2522 is at a similar stage, plant DNA adjacent to the T-DNA has been rescued from both these lines, although the plant DNA from 1926 contains a repeated element - complicating further work. The work on 178 is the most advanced. Plasmid rescue was used to clone the T-DNA/plant junction and this fragment used to screen cDNA and genomic libraries. Candidate clones have been sequenced. Mutant plants have been transformed with the candidate clones to ascertain by complementation whether a gene essential for pollen production in Arabidopsis has been cloned.

Acknowledgments

JAG was supported by an Australian Postgraduate Award administered by the Australian National University, as well as a Graduate Assistantship from the Cooperative Research Centre for Plant Science. We wish to thank Bjrg Sherman, Ying Luo and Ian Watson for their excellent technical assistance.

References