• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Apr 15, 1997; 94(8): 4223–4228.
Plant Biology

Fertilization-independent seed development in Arabidopsis thaliana


We report mutants in Arabidopsis thaliana (fertilization-independent seed: fis) in which certain processes of seed development are uncoupled from the double fertilization event that occurs after pollination. These mutants were isolated as ethyl methanesulfonate-induced pseudo-revertants of the pistillata phenotype. Although the pistillata (pi) mutant has short siliques devoid of seed, the fis mutants in the pi background have long siliques containing developing seeds, even though the flowers remain free of pollen. The three fis mutations map to loci on three different chromosomes. In fis1 and fis2 seeds, the autonomous endosperm nuclei are diploid and the endosperm develops to the point of cellularization; the partially developed seeds then atrophy. In these two mutants, proembryos are formed in a low proportion of seeds and do not develop beyond the globular stage. When FIS/fis plants are pollinated by pollen from FIS/FIS plants, ≈50% of the resulting seeds contain fully developed embryos; these seeds germinate and form viable seedlings (FIS/FIS). The other 50% of seeds shrivel and do not germinate; they contain embryos arrested at the torpedo stage (FIS/fis). In normal sexual reproduction, the products of the FIS genes are likely to play important regulatory roles in the development of seed after fertilization.

Keywords: ovule, apomixis

Arabidopsis seed, like the seed of other angiosperms, is a product of double fertilization, in which one of the two sperm cells fertilizes the haploid egg cell, giving rise to a diploid embryo, and the other sperm cell fertilizes the polar nuclei in the central cell, giving rise to the triploid endosperm (1). As the embryo and the endosperm develop, the ovule enlarges into a seed; the maternal tissues of the inner and outer integuments surrounding the embryo sac form the seed coat.

We have proposed a strategy for identifying genes that uncouple components of seed development from the fertilization process (2). Inactivation of genes that normally repress seed development may lead to precocious seed development without fertilization. In many apomictic plants, seed development does occur without fertilization or with only partial fertilization (3). In autonomous apomixis, seed development occurs without pollination and thus without fertilization of either the egg cell or the polar cell. In pseudogamous apomixis, pollination is required; in some cases, the pollination event results in fertilization of the polar cell but not of the egg cell. Apomixis has been described in a close relative of Arabidopsis, Arabis holboellii (3), and some genetic data support a one or two gene control of apomixis (3), so we reasoned that a mutational approach in Arabidopsis might detect mutants displaying some components of apomixis.

For the isolation of these mutants, we used stamenless pistillata (pi) (4). If pi plants are not pollinated, the siliques remain short; they only elongate when seed is formed. We identified mutants in which the siliques elongated without pollination (5), and recently Ohad et al (6) described a mutant that forms endosperm without fertilization.

In the present paper, we describe fis1 (fertilization-independent seed), fis2, and fis3, whose developmental and genetic characterization indicates that these genes normally have a controlling role in seed development after pollination and double fertilization.


Mutagenesis and Mutant Identification.

Heterozygous PI/pi seeds were generated by pollinating a female pi/pi with pollen from a PI/PI plant. For each mutagenesis, 2 g of F1 seeds was mutagenized as described (7). Each pot of the M1 plants was harvested separately by collecting at least 10 mature siliques from each plant to ensure that sufficient seeds were obtained from each M1 plant. In the M2 population, one–fourth of the progeny plants were pi/pi. PI/pi and PI/PI plants were identified by the presence of petals and stamens and were removed. Putative mutants were detected, in the pi/pi population, on the basis of elongation of siliques without formation of stamens.

Mapping of the fis Genes by Their Linkage to Morphological Markers.

FIS/fis heterozygous plants in a phenotypically self-fertile background were used as pollen donors in a cross to a multiply marked tester line, W100F (a male-fertile derivative of line W100) (8). In the segregating F2 population, recombination frequency was estimated as the proportion of FIS/fis plants in the total of FIS/fis plus FIS/FIS plants homozygous for each marker gene. Recombination frequency was converted to map distance (9).


Fertilized and unfertilized siliques were dissected under a dissecting microscope. For optical microscopy, tissue was prepared according to Craig and Miller (10). For scanning electron microscopy, dissected ovaries were attached with colloidal graphite to a copper stub, frozen under vacuum, and examined according to Craig and Beaton (11).

Measurement of Ploidy of the Endosperm Nuclei.

To measure the DNA content of endosperm nuclei of fis1 and fis2 mutants and pollinated homozygous pi/pi plants, 2-μm sections were stained with 0.5 μg/ml 4′, 6-diamidino-2-phenylindole (12) in water for 4 minutes and mounted in water. All observations were made with a Nikon Optiphot fluorescence microscope using a ×40 objective (Nippon Kogaku, Tokyo). At least 50 nuclei were measured for each mutant. Images were collected and analyzed using the Image 1 processing system (Universal Imaging, Media, PA). The Total Intensity Function was used to measure the fluorescence intensity of each endosperm nucleus.


Mutants in Which Siliques Elongate Without Pollination.

In a screen of 50,000 pi/pi M2 plants, six putative mutants with elongated siliques but no reversion of the floral pistillata phenotype were identified. These mutants contained seed of size and external morphology comparable to the seeds developing after pollination of the pi/pi plants (Fig. (Fig.1).1). We have analyzed three of these fertilization-independent seed mutants (fis1, fis2, and fis3).

Figure 1
Pollination-independent silique and seed development in a FIS/fis mutant plant. (A) Autonomous silique and seed development of a pi/pi FIS3/fis3 plant; unpollinated (B) and pollinated (C) silique from a pi/pi FIS/FIS ...

Genetic Transmission of the fis Mutations.

Plants with the fis phenotype originally were isolated in a male-sterile pi/pi background so that the fis trait could not be transmitted via pollen and had to be transmitted via the female gamete. When pi/pi FIS1/fis1, pi/pi FIS2/fis2, and pi/pi FIS3/fis3 plants were pollinated with pollen from a PI/PI FIS/FIS plant, the ratios of the FIS/fis-to-FIS/FIS plants were, respectively, 3:200, 1:151, and 0:140, rather than the expected ratio of 1:1. Heterozygous FIS/fis plants were identified by the 1:1 ratio of normal and embryo-arrested seeds. No egg transmission of the fis3 locus was detected, so the pi/pi FIS3/fis3 mutant plant was screened for a rare phenotypic reversion of the pi phenotype, which produced some pollen. One flower of ≈500 screened in the mutant plant was found to have some pollen; the pollen from this plant was used to transfer the mutant fis3 trait to progeny. These results indicate that the fis traits have a low frequency of transmission via the female gamete.

In contrast, the pollen transmission of the fis trait was unaffected by the fis mutation. Pollen from FIS/fis PI/pi heterozygotes of each of the three fis mutants was used to pollinate a FIS/FIS pi/pi female line. Among the resulting F1 plants, FIS1/fis1 and FIS1/FIS1 genotypes segregated 63:66, FIS2/fis2 and FIS2/FIS2 segregated 59:59, and FIS3/fis3 and FIS3/FIS3 segregated 82:83, indicating normal pollen transmission of the fis trait in each of the three fis mutants. Together these results indicate that the fis mutations are female gametophytic in action.

To obtain fis/fis homozygous plants, progeny of FIS/fis PI/PI plants were screened for plants in which all seeds in a silique were embryo-arrested. One of the 450 plants for the FIS1 locus and 2 of the 1621 plants for FIS2 were homozygous fis1/fis1 and fis2/fis2. No plants homozygous for fis3 were found among 2000 plants tested.

The FIS Genes Map to Different Chromosomes.

To map the FIS loci, pollen from FIS/fis PI/PI plants was used to pollinate W100F, a male-fertile derivative of W100 that contains 10 morphological mutations distributed on the arms of the five Arabidopsis chromosomes (8). Among the F2 progeny of FIS/fis W100F/+, plants homozygous for the different recessive morphological mutations were scored for FIS/FIS (all seeds in the siliques were normal) and for FIS/fis (the siliques contained a mixture of fully developed and embryo-arrested seeds). With these data, fis1 and an at the top end of chromosome 1 had a recombination frequency of 0.92 ± 0.91, translating to a map distance of 0.93 cM (n = 109). Similarly, the fis2-py recombination frequency of 9.28 ± 1.56 (map distance of 10.26; n = 345) and the fis2-er recombination frequency of 13.07 ± 2.73 (map distance of 15.14; n = 153) positioned fis2 between er and py on chromosome 2. Recombination frequencies of 14.45 ± 2.67 (map distance of 17 cM; n = 173) between fis3 and chromosome 3 marker gl1 and of 19.47 ± 2.9 (map distance of 24.67 cM; n = 187) between fis3 and hy2 positioned fis3 between hy2 and gl1 in chromosome 3.

Characterization of the fis Mutants.

Fertilized ovules of pi/pi plants developed into seeds, but unfertilized ovules remained small and eventually shrivelled (Fig. (Fig.22D). In unpollinated heterozygotes of the fis mutants, one–third to one–half of the ovules in the elongated siliques were transformed into seed-like structures resembling normal, sexually produced seed in external morphology and size (Fig. (Fig.2;2; ref. 13). All sexually fertilized seeds turn green and mature after pollination whereas seeds from pollinated FIS/fis heterozygotes contained green (mature) and white (embryo-arrested seed) at a 1:1 ratio. The fis ovules were similar to FIS ovules in early stages of ovule development. Both inner and outer integuments and the nucellar tissues of the fis mutants were indistinguishable from those of FIS plants (Fig. (Fig.2).2).

Figure 2
Cryo-scanning electron microscopy micrographs of ovules and seeds of fis mutants and fertilized wild-type plants. Developing ovules [nucellar column (n) protruding from the inner integument (ii) and the outer integument (oi) as shown in B] ...

Endosperm and Embryo Development in the fis Mutants.

Embryo sac, embryo, and endosperm development in ovules from the fis mutants were compared with those of ovules of the congenic Ler FIS plants (Fig. (Fig.3).3). In pi/pi ovules, no embryo or endosperm cells were seen (Fig. (Fig.33A). Three days after pollination of the pi/pi plant with pollen from a PI/PI plant, the ovules contained an embryo and free nuclear endosperm cells, and each ovule had expanded to the size of the mature seed (Fig. (Fig.33B).

Figure 3
Endosperm and embryo development of wild-type and fis seeds. (A) Longitudinal section through an ovule of an unfertilized pistillata (pi/pi) mutant. No endosperm or embryo development is visible. (B) Longitudinal section through an ovule of a ...

Fig. Fig.33 C and D shows the autonomous development of fis1 mutant ovules from a FIS1/fis1 heterozygous plant. As shown in Fig. Fig.33C, cellularized endosperm cells have formed inside the ovule, and the ovule has developed to the same size as the pollinated ovule, 3 days after pollination (Fig. (Fig.33B). More rarely, the fis1 ovule contained embryo-like structures at the micropylar end as well as free nuclear endosperm cells (Fig. (Fig.33D). In the mutant ovules from a FIS2/fis2 heterozygous plant, the ovule development was equivalent to the development of pi/pi ovules 3 days after pollination, and cellularized endosperm cells occasionally were accompanied by an embryo-like structure at the micropylar end (Fig. (Fig.33E). However, in the mutant ovules of fis3, even though the ovule developed in size as much as the fis1 and fis2 mutant ovules, the endosperm developed only to the free nuclear stage and not to the cellularized stage (Fig. (Fig.33F). In the ovules of a fis1 homozygous mutant plant (Fig. (Fig.33 G and H), giant multicellular structures often were seen at the micropylar end (Fig. (Fig.33G), and the endosperm development was either free nuclear (Fig. (Fig.33H) or cellularized (Fig. (Fig.33G).

Pollination Induces Changes in the Ovules of the fis Mutants.

When the fis1/fis1 or the fis2/fis2 homozygous mutant plants were pollinated with pollen from a FIS/FIS plant, embryos developed further than they did in the unpollinated fis1/fis1 and fis2/fis2 plants (Figs. (Figs.33I and and4).4). Heterozygous FIS3/fis3 plants, pollinated with pollen from a FIS/FIS plant, produced normal seed and torpedo-stage, embryo-arrested seed in a ratio of 1:1.

Figure 4
Pollination-induced silique and embryo development in a FIS1/fis1 heterozygote and a fis1/fis1 homozygote. (A) Torpedo stage Gus-positive (blue) embryos resulting from the pollination of a pi/pi FIS/FIS homozygous plant ...

Homozygous fis1 and fis2 plants, as well as heterozygous fis3 plants, were pollinated with pollen from a FIS/FIS plant, homozygous for a 35S-Gus reporter gene. The resulting torpedo-stage embryos were stained to detect the product of the Gus gene. As shown in Fig. Fig.4,4, all of the embryos resulting from self-pollination of the FIS/FIS 35S-Gus/35S-Gus plant stained blue, as did the embryos resulting from a pollination of a pi/pi FIS/FIS plant with pollen from a 35S-Gus/35S-Gus plant. In contrast, when 35S-Gus pollen was used to pollinate fis1/fis1 (Fig. (Fig.4),4), fis2/fis2 homozygotes, or FIS3/fis3 heterozygotes (data not shown), the resulting torpedo stage embryos were either Gus-positive or Gus-negative, suggesting that both zygotic and maternal embryos were present. The presence of Gus sequences in the blue embryos and their absence in the white embryos has been confirmed by PCR using primers from the Gus genes.

Wild-Type and fis Mutant Seeds Are Similar in Seed Coat Development.

After fertilization, the outer integuments of the Arabidopsis wild-type ovule develop polygonal structures with a central elevation called the columella (13). These structures were not seen in unfertilized ovules that did not develop any mature seed characters before they atrophied (Fig. (Fig.22D). Although the fis seeds were not fertilized, they did form the columella in the outer integument cells (Fig. (Fig.22J), and they were indistinguishable from normal zygotic seeds before they shrivelled.

fis Endosperm Cells Are Diploid.

The ploidy of the endosperm cells from fis1 and fis2 mutants was determined by measuring the fluorescence intensity of nuclei in 4′, 6-diamidino-2-phenylindole-stained sections. The average brightness of endosperm nuclei was 92.9 ± 20.8 (n = 57) for fis1 and was 135.0 ± 27.2 (n = 51) for the pollinated Ler control. The background values were 31.5 ± 6.9 (n = 50) and 34.3 ± 6.5 (n = 50), respectively. In a separate experiment, the brightness of autonomous fis2 endosperm nuclei was found to be 79.4 ± 14.4 (n = 40), and that of wild-type control nuclei was 108 ± 23.1 (n = 42). The background values were 37.3 ± 7.8 and 35.5 ± 6.2, respectively. In both cases, the results were consistent with the autonomous endosperm being diploid in contrast to the triploid condition of the sexual endosperm nuclei.


Relationship of the FIS Alleles to the Recently Described fie Mutant.

The fis3 mutant described here and earlier by us (5) resembles fie, recently described by Ohad et al. (6). The phenotype of fis3, i.e., endosperm development to the free nuclear stage, lack of an embryo, and absence of female transmission, parallels that of fie. Like fis3, fie maps in chromosome 3 between hy2 and gl1, which suggests that fis3 and fie might be allelic. The absence of egg transmission and the gametophytic expression of the mutations preclude direct complementation testing between fis3 and fie, and it remains possible that they are different loci.

In contrast, fis1 and fis2 map to chromosome locations different to either fis3 or fie. Unlike fis3 and fie, fis1 and fis2 have a low level transmission via the female gamete; this property enabled us to construct fis1/fis1 and fis2/fis2 homozygous plants. It was not possible to construct a fis3/fis3 homozygote; the fie/fie homozygote has not been described. Unlike fis3 and fie, the autonomous endosperm development in fis1 and fis2 goes beyond the free nuclear stage to the cellularized stage. fis1 and fis2 show a low frequency development of embryo-like structures that progress to the globular stage. These data indicate that the fis1 and fis2 mutations induce processes that are not under the control of fis3 or fie. These processes include the transition of the endosperm from the free nuclear to the cellularized stage and autonomous formation of embryos. We conclude that Arabidopsis has at least three genes, FIS1, FIS2, and FIS3(FIE), that prevent seed development without fertilization. We have identified seven other independent fis mutants, so it is possible that there are more than three loci.

In the fis Mutants, Partial Development of Seed Can Occur Without Pollination.

In the fis mutants, a number of steps in seed development occur without pollination. These events include the autonomous development of diploid endosperm, a low frequency development of globular, embryo-like structures, and the partial development of ovules into seeds, indistinguishable from developing sexual seeds in size and external morphology. Most fis seeds do not develop beyond the endosperm cellularization stage before atrophying. These results indicate that a substantial program of activation of genes involved in seed development is induced in plants carrying the mutagen-generated alleles at the FIS loci.

Seed Development in the fis Mutants After Pollination.

When homozygous fis1/fis1 and fis2/fis2 plants or heterozygous FIS3/fis3 plants were pollinated, each of the fis seeds had an embryo arrested at the torpedo stage. When these plants were pollinated with pollen from a FIS/FIS plant homozygous for the Gus reporter gene driven by the 35S promoter, the resulting torpedo stages were a mixture of Gus-positive (zygotic) and Gus-negative (maternal) embryos. In the fie mutant, all postpollination embryos were zygotic (6). If fie and fis3 are allelic, this difference might indicate that fis3 is a stronger allele than fie.

The postpollination arrest of embryos in the fis mutants indicates that factor(s) required for further embryo development are lacking. One explanation might be the ploidy of the endosperm; the autonomous diploid endosperm of the mutant might not nourish the embryo adequately so that neither the zygotic nor the apparently maternal embryos develop.

A Model of Action of the FIS Genes.

Although a detailed model of the action of the FIS gene products must await further characterization of the mutants and their interactions, the action of the FIS genes can be considered in a formal model (Fig. (Fig.5).5). In our model, we have grouped stages of seed development (14) into four phases. In phase 1, at 0 h after fertilization (HAF), the ovule contains an 8-celled embryo sac with an egg, two synergids, a diploid central cell, and three antipodal cells; the silique is short. We propose that FIS1, FIS2, and FIS3 specify polypeptides that act as negative regulatory elements repressing the conversion of a phase 1 ovule into a phase 2 developing seed (36 HAF). The FIS1FIS2 complex blocks both embryo and endosperm development. However, its effect on embryo development is only partial; other genes (X) also must block embryo development. The FIS1FIS2FIS3 complex blocks endosperm development. We assume that seed development is initiated in wild type by a pollination-induced inactivation of the FIS1FIS2FIS3 complex. The phase 2 partially developed seed normally proceeds through phase 3 (72 HAF) when the embryo is at the heart/torpedo stage and the endosperm is cellularized to the mature phase 4 seed (120 HAF, upturned U stage embryo). Our data suggest that there must be other control genes (Y) that are inactivated by pollination to facilitate this transition. For the sake of simplicity, we are proposing a negative regulatory control; but these genes also could act as positive regulators.

Figure 5
Wild-type and fis seed development. Seed development of wild-type Arabidopsis and fis mutants are compared at developmental phases (14). Phase 1 shows ovules connected to the ovary wall by the funiculus; in the subsequent phases, only the developing seed ...

In the fis1, fis2, and fis3 mutants, the phase 1 ovule proceeds without pollination to phases analogous to phases 2 and 3 of wild type. The seed-like structures have free nuclear endosperm and an embryo (Fig. (Fig.55a, fis1 and fis2), free nuclear endosperm without an embryo (Fig. (Fig.55b, fis1, fis2, and fis3), or seed-like structures with cellularized endosperm (Fig. (Fig.55c, fis1 and fis2). Even with pollination, the fis seed develops only to an embryo-arrested heart/torpedo stage and does not become a mature seed with a U-shaped embryo. The conversion of the embryo-arrested seed to mature seed may be blocked by a gene Z that is not inactivated by pollination. Mutations in the hypothetical Y gene would convert the fis seeds into embryo-arrested seeds without pollination, and additional mutation of the Z gene would give rise to full normal seeds.

The FIS Genes Might Be Components of Apomictic Seed Development Seen in Many Angiosperms.

The gametophytic nature of the fis mutations indicates that their time of action is downstream from the point at which apomictic processes operate in plants that have either aposporic or diplosporic seed development (3).

The FIS genes are not likely to be the initial controlling genes for apomictic seed development but are likely to be inactivated so that full development of apomictic seeds can proceed. The gene(s) that regulate the initiation of apomixis might therefore be positive regulators of a number of FIS-like genes. In this model, an apomictic mutant would be one that is simultaneously impaired in a number of FIS functions. We predict that naturally occurring apomictic plants would have their endogenous FIS functions either mutated or down-regulated. Cloning of the FIS genes, currently underway in our laboratory, followed by the characterization of the FIS genes and their products in natural apomicts related to Arabidopsis, such as Arabis holboellii, will enable us to determine whether this is the case.


We thank Diana Budd, Manel Samarakoon, and Ping Hua He for technical assistance. This work was supported by a Rockefeller Foundation research grant.


hour(s) after fertilization


1. Reiser L, Fischer R L. Plant Cell. 1993;5:1291–1301. [PMC free article] [PubMed]
2. Peacock W J. Apomixis Newsletter. 1992;4:3–7.
3. Asker S E, Jerling L. Apomixis in Plants. Boca Raton, FL: CRC; 1992.
4. Bowman J L, Smyth D R, Meyerowitz E M. Plant Cell. 1989;1:37–52. [PMC free article] [PubMed]
5. Peacock J, Ming L, Craig S, Dennis E, Chaudhury A. Induced Mutations and Molecular Techniques for Crop Improvement. Vienna: International Atomic Energy Agency; 1995. pp. 117–125.
6. Ohad N, Margossian L, Hsu Y C, Williams C, Repeth P, Fischer R L. Proc Natl Acad Sci USA. 1996;93:5319–5324. [PMC free article] [PubMed]
7. Chaudhury A M, Letham D S, Craig S, Dennis E S. Plant J. 1993;4:907–916.
8. Koornneef M, Hanhart C J, Van Loonen-Martinet E P, Van der Veen J H. Arabidopsis Inf Serv. 1987;23:46–50.
9. Haldane J B S. J Genetics. 1919;8:299–309.
10. Craig S, Miller C. Cell Biol Int Rep. 1984;8:879–886. [PubMed]
11. Craig S, Beaton C D. J Microsc (Oxford) 1996;182:102–105.
12. Coleman A W, Goff L J. Stain Technol. 1985;60:145–154. [PubMed]
13. Mansfield S G. In: Arabidopsis: An Atlas of Morphology and Development. Bowman J, editor. New York: Springer; 1994. pp. 372–377.
14. Bowman J, Koornneef M. In: Arabidopsis: An Atlas of Morphology and Development. Bowman J, editor. New York: Springer; 1994. pp. 351–354.

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...