Control of seed development in Arabidopsis thaliana by atmospheric oxygen


  • A. Kuang,

    1. Department of Plant Pathology & Crop Physiology, Louisiana Agricultural Experiment Station, Louisiana State University Agricultural Centre, Baton Rouge, LA 70803, USA
    Search for more papers by this author
  • M. Crispi,

    1. Department of Plant Pathology & Crop Physiology, Louisiana Agricultural Experiment Station, Louisiana State University Agricultural Centre, Baton Rouge, LA 70803, USA
    Search for more papers by this author
  • M. E. Musgrave

    1. Department of Plant Pathology & Crop Physiology, Louisiana Agricultural Experiment Station, Louisiana State University Agricultural Centre, Baton Rouge, LA 70803, USA
    Search for more papers by this author

M. E. Musgrave
Fax: 504 3881415; e-mail:


Seed development is known to be inhibited completely when plants are grown in oxygen concentrations below 5·1 kPa, but apart from reports of decreased seed weight little is known about embryogenesis at subambient oxygen concentrations above this critical level. Arabidopsis thaliana (L.) Heynh. plants were grown full term under continuous light in premixed atmospheres with oxygen partial pressures of 2·5, 5·1, 10·1, 16·2 and 21·3 kPa O2, 0·035 kPa CO2 and the balance nitrogen. Seeds were harvested for germination tests and microscopy when siliques had yellowed. Seed germination was depressed in O2 treatments below 16·2 kPa, and seeds from plants grown in 2·5 kPa O2 did not germinate at all. Fewer than 25% of the seeds from plants grown in 5·1 kPa oxygen germinated and most of the seedlings appeared abnormal. Light and scanning electron microscopic observation of non-germinated seeds showed that these embryos had stopped growing at different developmental stages depending upon the prevailing oxygen level. Embryos stopped growing at the heart-shaped to linear cotyledon stage in 5·1 kPa O2, at around the curled cotyledon stage in 10·1 kPa O2, and at the premature stage in 16·2 kPa O2. Globular and heart-shaped embryos were observed in sectioned seeds from plants grown in 2·5 kPa O2. Tissue degeneration caused by cell autolysis and changes in cell structure were observed in cotyledons and radicles. Transmission electron microscopy of mature seeds showed that storage substances, such as protein bodies, were reduced in subambient oxygen treatments. The results demonstrate control of embryo development by oxygen in Arabidopsis.


In the 1970s, experiments with representative C3 and C4 plants were conducted to understand the consequences of long-term inhibition of the oxygenase activity of Rubisco for plant growth. As expected, a great stimulation of vegetative growth occurred for C3, but not for C4 species when plants were grown at 5·1 kPa O2. Quebedeaux & Hardy (1973) reported on the unforeseen outcome of these experiments when they described the complete inhibition of seed production in soybeans and sorghum grown at 5·1 kPa O2. Because pollen and embryo sac development was normal, they concluded that an O2-sensitive process affecting the transfer of photosynthate from leaves to developing seeds was responsible for this finding. Subsequently these observations were variously explained by inhibition of respiration in developing seeds (Gale 1974) or of phloem unloading from the seed coat (Thorne 1982) at very low partial pressures of O2.

The critical oxygen pressure for seed production was determined to be 5·1 kPa O2 for soybean and sorghum (Quebedeaux & Hardy 1976). No prior studies have described the seed development that occurs in the intervening oxygen tensions between ambient (21·3 kPa O2) and the critical oxygen pressure for seed production, except in terms of dry weight per seed. At our laboratory, we are interested in the effect of these intermediate oxygen pressures on seed development for several reasons. First, our series of experiments on reproductive development in Arabidopsis in microgravity (Kuang et al. 1995; Kuang, Musgrave & Matthews 1996a; Kuang, Xiao & Musgrave 1996b) have implicated the gaseous microenvironment that develops around plant parts in the absence of convective air movement as a cause of reproductive anomalies observed repeatedly in plants grown under spaceflight conditions (Musgrave Kuang & Matthews 1997a; Musgrave, Kuang & Porterfield 1997b). Secondly, NASA is interested in using plants to regenerate air, clean water and produce food as part of a life-support system for long-duration space missions. Because plant productivity is higher at low oxygen tensions and less oxygen would be required to maintain the growth environment, a low oxygen/low pressure growth scenario is being considered for this specialized application. However, more information is needed about how oxygen concentrations control reproductive development. Using the model plant Arabidopsis thaliana we have been able to conduct a detailed microscopy study of embryo development during plant growth in subambient oxygen.


Growth conditions

Arabidopsis thaliana (L.) Heynh var. ‘Columbia’ plants were grown on peat–vermiculite artificial soil medium supplemented with a slow-release fertilizer in Sun transparent bags (Sigma) under 250 μmol m–2 s–1 illumination with photosynthetically active radiation (PAR) at 25 °C. The bags were continuously purged (200 cm3 min–1) with air or premixed gases containing different oxygen levels (2·5, 5·1, 10·1, 16·2 kPa), 0·035 kPa CO2, and the balance N2 (Crispi, Porterfield & Musgrave 1996). Gas composition in the premixed tanks and in the bags was confirmed by periodic sampling and gas chromatography. Plants were harvested at 45 d when siliques had yellowed.

Seed germination test

Mature seeds from 45-d-old plants were air-dried. One hundred seeds from each oxygen concentration were placed on filter paper moistened with distilled water in Petri dishes under continuous fluorescent illumination. Germination (emergence of radicle and cotyledons from the seed coat) and seedling length (determined by measuring from the point of cotyledon attachment to the tip of the root) were scored on the 5th and 10th days after imbibition, respectively.

Preparation of materials for microscopy

Non-germinated seeds from the germination tests were dissected and their embryos were fixed for light microscopy and scanning electron microscopy (SEM). Dried seeds were also briefly imbibed (about 30 min) in distilled water until the seed coats were softened. Then the embryos were dissected and fixed for transmission electron microscopy (TEM) to view embryos early in the germination process. Fresh seeds were also collected and dissected from 35-d-old plants for microscopy observation, because dry seeds from 45-d-old plants in the 2·5 kPa O2 treatment were shrivelled and empty, despite the fact that pollen from this treatment was viable and germinated on stigmas (not shown).

Fixation and observation

Materials harvested as described above were fixed with 1% formaldehyde and 2·5% glutaraldehyde in 0·1 kmol m–3 phosphate buffer, pH 7·0, and postfixed in 1% osmium tetroxide (OsO4). For SEM, embryos of nongerminated seeds were dehydrated with ethanol, dried in CO2 using a Denton DCP-1 critical-point drying apparatus (Denton Vacuum, Inc., Cherry Hill, NJ, USA), and observed and photographed with a Cambridge S-260 scanning electron microscope (Cambridge Instruments, Ltd, Cambridge, UK). For light microscopy and TEM, materials were dehydrated in an ethanol series, and then infiltrated with and embedded in Spurr's resin (Spurr 1969). Embedded tissues were sectioned with a DuPont Sorvall microtome (DuPont, Co., Newton, CN, USA). Sections 1 μm thick were stained with 1% toluidine blue O (TBO), observed and photographed using a Nikon light microscope (Nikon, Inc., Melville, NY, USA). Thin sections were stained with uranyl acetate and lead citrate, and observed under a JEOL JEM 100CX transmission electron microscope (JEOL, Ltd., Tokyo, Japan). Fresh material was observed and photographed using a Nikon dissecting microscope.


By the 5th day after imbibition, more than 95% of the seeds harvested from plants grown in the 21·3 kPa O2 environment germinated. The percentage of seed germination declined with the atmospheric oxygen concentration, and seeds from the 2·5 kPa O2 environment did not germinate at all ( Fig. 1). Seedling vigour, as assessed by seedling length 10 d after imbibition, also diminished as the oxygen concentration in which the seeds had formed was reduced (Fig. 1). Seeds from plants grown at 5·1 kPa O2 needed longer to germinate and cotyledons were small, pale and barely expanded (Fig. 2a). Roots were stunted, although in some cases root hairs differentiated normally (Fig. 2a). Seeds from the 10·1 kPa O2 environment that did germinate (63%) produced seedlings that were smaller and weaker (Fig. 2b) when compared with those from plants grown in 16·2 and 21·3 kPa O2 environments (Figs 2c & d; Fig. 1).

Figure 1.

. Percentage germination and mean seedling length (± SE) of seeds harvested from Arabidopsis thaliana plants grown in transparent bags purged with gases containing different partial pressures of oxygen (2·5, 5·1, 10·1, 16·2 and 21·3 kPa).

Figure 2.

. Representative 10-d-old seedlings from seeds harvested from Arabidopsis thaliana plants grown in (A) 5·1 kPa O2; (B) 10·1 kPa O2; (C) 16·2 kPa O2; and (D) 21·3 kPa O2. Arrowheads, cotyledons; bar in (A), 422 μm; bars in (B), (C) and (D), 1·0 mm.

Observation of non-germinated seeds under SEM showed that embryos in non-germinated seeds from the 5·1 kPa O2 environment stopped growing at the heart-shaped to linear cotyledon stages (Figs 3a & b). Embryos of non-germinated seeds from 10·1 and 16·2 kPa O2 environments were at the curled-cotyledon stage to premature stage (Figs 3c & d). Those embryos were morphologically normal although shrunken cells were present. The anatomy of non-germinated seeds confirmed that most embryos from the 5·1 kPa O2 environment stopped growing at the heart-shaped to linear cotyledon stage, although some development continued to a curled-cotyledon stage (Figs 4a & b). In those embryos, cells were highly vacuolated and tissue systems were disorganized. Embryos of non-germinated seeds from 10·1 and 16·2 kPa O2 environments had developed to a late curled-cotyledon stage or up to the premature stage. However, tissues either in cotyledons or radicles were partially degenerated as a result of cell autolysis (Figs 4c & d).

Figure 3.

. Scanning electrom microscope photographs of embryos from non-germinated seeds of Arabidopsis thaliana plants grown under different atmospheric oxygen levels. (A) and (B): embryos from 5·1 kPa O2— at the heart-shaped and linear cotyledon stages; (C) an embryo from 10·1 kPa O2, (D) an embryo from 16·2 kPa O2— embryos at the curled-cotyledon stage or at maturity. Shrunken cells (arrowheads) are visible in each embryo. co, cotyledon; rd, radicle; bar, 5 μm.

Figure 4.

. Light microscope photographs of embryos dissected from non-germinated seeds of Arabidopsis thaliana. Embryos were sectioned and stained with toluidine blue O. (A) and (B): embryos of seeds from a 5·1 kPa oxygen environment, showing anomalous vacuolated cells. (C) and (D): embryos of representative non-germinated seeds from 10·1 kPa O2 and 16·2 kPa O2, respectively. Arrows indicate portions of cotyledon and radicle where tissue degenerated. co, cotyledon; rd, radicle; bars, 100 μm.

Seeds from the 2·5 kPa O2 environment were empty after dehydration, so it was necessary to observe the progress of their development in fresh material. Globular and heart-shaped embryos were observed in fresh seeds harvested after 35 d. Among those embryos, the embryo proper was typically poorly developed, in that tissues such as protoderm and provascular tissue had failed to differentiate (Fig. 5a) compared with normally developed embryos (Fig. 5b) at the same developmental stage; the suspensor was unusually enlarged, while the cytoplasm in suspensor cells was scarce (Fig. 5a).

Figure 5.

. Light microscope photographs of globular-stage embryos from fresh Arabidopsis thaliana seeds sectioned and stained with toluidine blue O. (A): embryo development in a seed from a 2·5 kPa O2 environment, showing a poorly differentiated embryo proper and an unusually enlarged suspensor cell (arrow) compared with (B): an embryo at the same developmental stage from 21·3 kPa O2. ep, embryo proper; s, suspensor; arrowheads, protoderm. Bars, 50 μm.

Structural differences in embryo cells relating to concentrations of atmospheric oxygen were also observed at the ultrastructural level, when seeds had been randomly sampled and peeled after a 30 min imbibition. At 21·3 kPa O2, embryo cells contained large protein bodies that were full of materials except for small, irregular spaces from which globoids had probably been lost during tissue processing (Fig. 6a). At 16·2 kPa O2, cells were full of approximately globular protein bodies containing small spaces where globoids had been lost. The protein bodies were 40% smaller in diameter compared with those at the 21·3 kPa O2 treatment (Fig. 6b). Lipid bodies were compressed in homogenized masses that spread through entire cells. At 10·1 kPa O2, a few nascent protein bodies had formed but in most cases a thin protein layer had been deposited inside tonoplasts along membranes, while large lipid bodies were present in cells (Fig. 6c). At 5·1 kPa O2, cells were disorganized, with lipid droplets free in the cytoplasm (Fig. 6d).

Figure 6.

. Ultrastructure of cotyledon cells in Arabidopsis thaliana embryos dissected from seeds after brief imbibition. Seeds were harvested from different oxygen levels: (A), from 21·3 kPa O2; (B), from 16·3 kPa O2; (C), from 10·1 kPa O2; D, from 5·1 kPa O2. (A): large protein bodies (pb) are full of material except for small, irregular spaces (arrowheads) from which globoids were probably lost during tissue processing. (B): protein bodies (pb) are in globular shapes and lipid bodies (asterisks) are compressed in a homogeneous mass and spread across entire cells. (C): a few protein bodies have formed and protein is deposited in tonoplasts along the membrane (arrowheads). (D): cells are disorganized and lipid droplets are free in the cytoplasm. Bars, 0·5 μm.


Subambient oxygen depressed seed quality in Arabidopsis by limiting embryo development. Complete failure of seed development occurred at 2·5 kPa O2. In soybean, sorghum (Quebedeaux & Hardy 1975), and wheat (Musgrave & Strain 1988), no seeds were produced at 5·1 kPa atmospheric O2, and in rice, seed production was greatly reduced after exposure of plants to 3·5 kPa O2 for 20 d after heading (Akita & Tanaka 1973; Akita 1980). Thus the critical O2 concentration for seed production may vary slightly depending on plant species.

The effect of atmospheric O2 on embryo development in Arabidopsis is summarized in Table 1. As the results indicate, oxygen concentration in the environment apparently determines how far embryo development can progress in Arabidopsis. Previous studies determined that the oxygen demand of developing embryos increases as those embryos grow. Quebedeaux & Hardy (1975) exposed soybean plants to a low O2 environment at different developmental stages and observed that the greatest effect of low O2 on seed production occurred during flowering to pod filling. Embryo respiration was studied in vitro in cotton, poppy and Zephyranthes, and the amount of oxygen depletion was found to depend on the developmental stage, with large increases at the later stages (Forman & Jensen 1965; Johri & Maheshwari 1965, 1966). Embryos of Arabidopsis stopped growing at different developmental stages depending upon oxygen concentrations, consistent with the larger amounts of oxygen necessary for seed development at later stages.

Table 1.  . Percentage of embryos at different developmental stages in non-germinated Arabidopsis thaliana seeds from different oxygen treatments, based on observations of 20 non-germinated embryos Thumbnail image of

Analysis of material developed under subambient oxygen concentrations showed unusual features of tissue degeneration caused by cell autolysis, and changes in cell structure. Particularly interesting at the ultrastructural level was the effect of oxygen concentration on the development of protein bodies; these are formed in storage tissue during seed maturation and degraded during seed germination, and are a major source of needed reserves during early germination (Murray 1984; Vigil et al. 1985). Protein bodies occupied a large volume of cotyledon cells in the embryos of mature seeds from 21·3 and 16·2 kPa O2 environments; however, only a few nascent protein bodies were observed in seeds from the 10·1 kPa O2 environment and none were distinguishable in embryos from 5·1 kPa oxygen. Mansfield & Briarty (1992) observed that protein bodies first appeared in cotyledon cells 144 h after pollination and continued to accumulate to form large protein bodies containing crystal globoids during seed maturation in Arabidopsis. Protein bodies observed in cotyledon cells of mature seeds from the 10·1 kPa O2 environment appeared similar to those described in developing seeds 144 h after pollination. Only about 60% of the seeds harvested from the 10·1 kPa O2 environment germinated and the seedlings were smaller and weaker than the seedlings from seeds that developed in 21·3 and 16·2 kPa O2 environments (Figs 1 & 2). Sinclair, Ward & Randall (1987) reported that soybean pods exposed to 10·1 kPa O2 produced significantly lower seed weight compared with pods exposed to ambient O2. The gene expression of storage protein was studied in wild-type and embryo-lethal mutants of Arabidopsis, and indicated that accumulation of seed storage proteins in embryos of Arabidopsis occurs at late developmental stages and is closely tied to morphogenetic changes that occur during embryo development of seeds (Heath et al. 1986; Pang, Pruitt & Meyerowitz 1988).

Tissues surrounding developing embryos are known to limit oxygen resupply rates, and in concert with the respiratory demands of developing embryos produce in ovulo oxygen concentrations that are lower than those prevailing outside the plant (Hess & Carman 1993; Musgrave et al. 1997b). In fact, in vitro embryo culture has been shown to proceed more efficiently at lower oxygen tensions than at ambient oxygen (Hess & Carman 1993). Coupled with our results, these observations point to a relatively tight control of embryo development by oxygen concentrations. Issues of seed quality will thus determine the range of non-ambient oxygen concentrations that may be considered for plant growth environments in future specialized applications associated with the space exploration programme.


We thank Ying Xiao for her contributions in microslide preparation and Dr Marc A. Cohn for helpful discussion. This study was supported by NASA (NAG10–0139; NGT-51097). Approved for publication by the Director of the Louisiana Agricultural Experiment Station as manuscript #97–38–0050.