1College of Marine Studies and College of Agriculture and Natural Resources, University of Delaware, 700 Pilottown Road, Lewes, DE 19958, USA
2Department of Agronomy, Iowa State University, Ames, IA 50011, USA
* To whom correspondence should be addressed. Fax: +1 302 645 4007. E-mail: firstname.lastname@example.org
Received 19 November 2003; Accepted 10 February 2004
Plant reproduction is sensitive to water deficits, especially during the early phases when development may cease irreversibly even though the parent remains alive. Grain numbers decrease because of several developmental changes, especially ovary abortion in maize (Zea mays L.) or pollen sterility in small grains. In maize, the water deficits inhibit photosynthesis, and the decrease in photosynthate flux to the developing organs appears to trigger abortion. Abscisic acid also increases in the parent and may play a role, perhaps by inhibiting photosynthesis through stomatal closure. Recent work indicates that invertase activity is inhibited and starch is diminished in the ovaries or affected pollen. Also, sucrose fed to the stems rescues many of the ovaries otherwise destined to abort. The feeding restores some of the ovary starch and invertase activity. These studies implicate invertase as a limiting enzyme step for grain yields during a water deficit, and transcript profiling with microarrays has identified genes that are up- or down-regulated during water deficit-induced abortion in maize. However, profiling studies to date have not reported changes in invertase or starch synthesizing genes in water-deficient ovaries, perhaps because there were too few sampling times. The ovary rescue with sucrose feeding indicates either that the changes identified in the profiling are of no consequence for inhibiting ovary development or that gene expression reverts to control levels when the sugar stream recovers. Careful documentation of tissue- and developmentally specific gene expression are needed to resolve these issues and link metabolic changes to the decreased sugar flux affecting the reproductive organs.
Key words: Abortion, abscisic acid, invertase, ovary, photosynthesis, pollen, starch, sterility, sucrose, sugar
Source: Journal of Experimental Botany 2004 55(407):2385-2394
Agriculture depends to a large extent on the success of plant reproduction. Not only is it necessary for the next crop, but reproductive products like grain, fruit, many vegetables, and nuts are the bulk of the world food supply. In the USA, more than 75% of the harvested acreage is devoted to such crops (Agricultural Statistics, 1999). Consequently, when world economies give attention every year to the success of crop production, they are focusing mostly on plant reproduction.
The success of reproduction is determined to a large extent by the environmental conditions prevailing during the growing season. Boyer (1982) surveyed the yields of eight crops in the US, six of which had valuable reproductive structures (maize [Zea mays L.], sorghum [Sorghum spp., probably bicolor or halepense L.], wheat [Triticum aestivum L.], barley [Hordeum vulgare L.], oats [Avena sativa L.], and soybean [Glycine max L.]). The six reproductive crops had average yields during agricultural production that were only about 18% of the record yield. Assuming that record yields measure the maximum biological potential for reproduction in agricultural conditions, a large fraction of the potential was not realized.
These losses vary somewhat from year to year in different locations around the world. In the US, the availability of water was a major contributor to the variation (Boyer, 1982). Drought affected crop productivity nearly as much as all the other environmental factors combined. Irrigation has been one solution but is becoming less so as global water demand increases. In a detailed review of hundreds of studies, Salter and Goode (1967) point out that the largest response to irrigation is during early reproductive development for most crops. The biggest return per unit of water can be obtained by irrigating at this time. But despite the susceptibility, Salter and Goode (1967) could find few reasons for it. Research on mechanisms of reproductive failure was rare.
Since the analysis of Salter and Goode (1967), increased attention has been paid to reproductive failure during water deficits, focusing mostly on grain crops. Saini and Westgate (2000) reviewed this literature and highlight the accumulating evidence for signals emanating from the parent plant, especially carbohydrate availability and metabolism, as well as hormones. In wheat, barley, and rice (Oryza sativa L.), abscisic acid (ABA) was implicated as a cause of pollen sterility, and in maize decreases in the sugar stream from losses in photosynthesis appeared important for development of the female inflorescence. Since the Saini and Westgate (2000) review, however, most new work has involved maize, which will be the focus of this review. For the other small grains, the reader is directed toward the Saini and Westgate (2000) review.
Saini and Westgate (2000) point out that early reproduction is highly phasic, with each phase showing susceptibility to water deficits. Meiosis, anthesis, pollen fertility, pollination, female fertility, and early zygote development are susceptible, and their failure diminishes the number of kernels that the plant produces. Later in development, water deficits tend to reduce kernel size rather than number, and size seems to be determined in large part by the available photosynthetic reserves that can be moved to the grain. For example, McPherson and Boyer (1977) worked with maize subjected to water deficits beginning 2 weeks after pollination and extending until grain maturity. This long-term deficit occurred during the linear phase of dry mass accumulation by the kernel. The kernel size was reduced with little effect on kernel numbers. Similar results were obtained in a field study by Jurgens et al. (1978).
The water deficits studied by McPherson and Boyer (1977) caused low enough water potentials (w) to inhibit photosynthesis substantially, and scarcely any dry matter accumulated as a result. Despite the lack of dry matter accumulation in the parent, the kernels continued to fill for some time. The dry matter for the kernels came from the parental reserves. In recent work in rice, the importance of reserves was also demonstrated (Yang et al., 2001a, b) when low w imposed during the linear phase of grain filling caused little loss in yield. Large amounts of carbohydrate were moved from the stems to the grain that made up for the lack of current photosynthesis. As a result, there is often a relationship between the dry matter in the grain at the end of the season and that in the parent (McPherson and Boyer, 1977; Yang et al., 2001a, b). Sinclair (1990) collected data from many maize plots subjected to water deficits and also found a correlation between kernel yield and shoot dry matter accumulation during a water deficit.
For low w around the time of pollination, however, the yield/dry matter relationship fails because kernel numbers diminish. Westgate and Boyer (1985) found a result similar to that of McPherson and Boyer (1977) and Sinclair (1990), but also tested earlier phases around the time of pollination. The early phases were unable to access plant reserves sufficiently to maintain reproductive development when current photosynthesis was inhibited by limited water. And unlike the situation during grain filling, a larger pool of reserves in the plant as a whole did not help to maintain ovary growth (Schussler and Westgate, 1994). Kernel number decreased irreversibly, i.e. abortion occurred. These results indicate that the response to a water deficit early in development (i.e. during pollination) differs from the response later in development (i.e. during grain filling). Large agricultural losses can occur during both phases, but the irreversibility of the early events is especially damaging.
An important feature of this type of work is the continually changing water status of soil-grown plants. In contrast to marine environments, where the water supply is essentially infinite and salinity often varies in a moderate range, water in soil is finite and gradually depleted. The available water changes from day to day and even during the day. Metabolism continually changes, and the yield reflects these changes. Keeping plant conditions constant and reproducible becomes a challenge. The capacity to control the timing and intensity of plant water deficit is paramount for designing meaningful genomic and metabolomic studies. Changes in gene expression can occur within minutes in response to a change in environmental conditions (Seki et al., 2002).
The finite water supply is traceable to the particulate nature of soil. The pores between particles fill with water by capillarity, and surface tension in the water prevents the water from spreading to nearby unfilled pores. As a result, partial rewatering fills a few pores and wets only a local volume of soil. The roots in the wetted part are very wet while others remain dry in the rest of the soil. Maintaining a steady, uniform soil water deficit while the plant is removing water is impossible.
Refocusing on the plant suggests a way to maintain plant water status, however. By adding in the morning only the amount of water used the previous day, the added water wets a small soil volume and enters slowly through a few roots, preventing the plant w from declining during the day. The added water is depleted by the end of the day, and the soil water content returns to that of the previous night. Because transpiration is minimal at night, the plant water status continues to be constant through the night. The water addition is repeated at first light the next morning. With this procedure, plant w can be held essentially stable for weeks. Boyer and McPherson (1975), McPherson and Boyer (1977), Westgate and Boyer (1985), and Setter et al. (2001) used this approach.
The efficacy of the method can be tested by measuring w. When Boyer and McPherson (1975) conducted maize experiments in controlled environments with irradiances similar to full sun, w measured in the upper leaves were essentially stable throughout the day and night (JS Boyer and HG McPherson, unpublished data) and for weeks afterward (Boyer and McPherson, 1975; McPherson and Boyer, 1977). In subsequent field experiments, the results were more variable because of variation in climatic conditions (Jurgens et al., 1978).
The measurements have the additional benefit of thermodynamically based data with a physically defined reference for precisely repeating conditions in subsequent crops. Working in controlled environments, this ability to repeat conditions allows progressive probing of molecular mechanisms, and comparison of the results with those from other scientists. Because differences in w drive water through the plant and thus vary from place to place and time to time, the measurements need to include sufficient conditions to be informative. For example, knowledge of pollen and floret w was needed throughout the day in order to interpret the cross-pollination experiments of Westgate and Boyer (1986b).
Split root systems are an extension of this approach and involve growing part of the root system in one compartment supplied with water and the other in another compartment containing dehydrated soil. The split roots provide an experimental way to separate the direct effects of low tissue w from those that develop indirectly from root-derived plant growth regulators, for example. The plant w varies diurnally because water is continually available to the wet part of the root system, but the time spent at high w tends to be shorter each day than in a plant with water around the entire root system. Split root systems have been used to study the effect of drying soil on a number of plant functions, especially root signals sent to the shoots (Davies and Zhang, 1991; Dembinska et al., 1992).
Pollination at low w
In the grains, losses in kernel number were often ascribed to a lack of fertilization of the egg, resulting in an undeveloped ovule. Pollen was sterile in barley, wheat, and rice when low w occurred during microsporogenesis (Saini and Westgate, 2000). ABA applied to the shoot in various ways caused similar pollen sterility (Saini and Westgate, 2000). Because ABA concentrations increase when plants have low w, high ABA was at first thought to account for pollen sterility (Saini and Westgate, 2000). However, using split roots to change ABA concentrations without lowering w, Dembinska et al. (1992) found that elevated ABA alone did not induce low grain numbers in wheat. Later, Dorion et al. (1996) and Sheoran and Saini (1996) reported decreases in invertase activity and starch accumulation in the developing pollen of wheat and rice when w were low. These investigators ascribed pollen sterility to losses in invertase activity that resulted in an inability of the pollen to metabolize incoming sucrose to hexoses.
In maize, low kernel numbers were also occasionally ascribed to sterile pollen (Lonnquist and Jugenheimer, 1943). However, Westgate and Boyer (1986b) and Schoper et al. (1987) found little evidence for pollen sterility at low w. Instead, Schoper et al. (1987) reported a sensitivity of pollen to high temperatures. High temperatures often accompany low w in the field, and there was genotypic variation for this sensitivity. In the absence of high temperature, pollen that developed normally retained its viability when low w occurred (Westgate and Boyer, 1986b). Silks sometimes emerged too late to be pollinated (Moss and Downey, 1971; Edmeades et al., 1993; Herrero and Johnson, 1981). Bassetti and Westgate (1993) found that silks lost receptivity to pollen as they aged, and receptivity was lost earlier if the plants were subjected to a water limitation. Much emphasis has been placed on minimizing these asynchronous effects by genetic selection. Edmeades et al. (1993) selected for genotypes exhibiting precocious silk growth (i.e. appearing before pollen shed begins), and this early development allowed silks to remain within the pollen-shed interval despite the delaying effect of low w. Precocious silk emergence is evident in many modern maize hybrids and contributes significantly to their improved drought performance.
Ovary abortion at low w
Despite the improvements in synchronization of pollen shed and silk emergence, kernel numbers often continue to be low when w is low around the time of pollination (Westgate and Boyer, 1985). Westgate and Boyer (1986b) focused on this question by shortening the exposure to low w. Rather than subjecting the plants to low w from pollination until maturity, they shortened the exposure by rewatering the plants soon after pollination. As few as 3 d of low w (water withheld for 6 d) caused the same decrease in kernel numbers as did the prolonged exposure. The failure of kernels to develop was irreversible and indicated that abortion had occurred.
Reciprocal cross-pollinations between plants at high and low w revealed that pollen having w as low as –15 MPa remained viable and able to fertilize the egg (Westgate and Boyer, 1986b). The female florets were not viable at such low w and were unable to form kernels at w below about –1.2 MPa. Therefore, the abortion was controlled by the female inflorescence.
In general, the silks of female florets were wetter than pollen. The silks hydrated the pollen and promoted pollen tube growth. Fertilization was typically successful even at the lowest floret w, and each unformed kernel contained an undeveloped embryo and remnants of an endosperm (Westgate and Boyer, 1986b). Reproduction was thus successful except for the last phase, when the embryo and endosperm were developing. The ears had low kernel numbers at maturity (cf. Fig. 1A, C).
Rescuing ovaries from abortion at low w
Gengenbach (1977) discovered a medium for culturing isolated embryos of maize, and Boyle et al. (1991) fed it to the stems of plants while they were being exposed to abortion-inducing w. The idea was to test whether the medium could keep the embryos alive when photosynthesis was inhibited at low w. Stems were fed an amount containing the dry mass normally produced by photosynthesis because the entire plant was starved for photosynthetic products. This amount was fed each day to a different stem position. The plants were then rewatered and allowed to develop for the rest of the season. At maturity, the low w treatment had caused nearly complete abortion in the unfed plants, but 80% of the ovaries developed in the fed plants. The rescued kernels appeared normal at maturity. If water was fed to the stems without the ingredients of the Gengenbach (1977) medium, no kernels developed.
This rescue indicated that low w was not in itself lethal. Instead, abortion was initiated because the parent did not supply an ingredient in the medium, and the missing ingredient caused the abortion. It should be noted that the feeding kept the kernels alive at low w so that photosynthesis and normal kernel development could resume when water was resupplied. Most of the kernel development occurred after rewatering.
Boyle et al. (1991) and Zinselmeier et al. (1995a) conducted deletion and addition experiments to identify the active component(s). Plant growth regulators (auxin, cytokinin), amino acids, cofactors, and salts had no activity. Sucrose alone was the compound that rescued the ovaries from abortion. A typical result, shown in Fig. 1B, indicates that about 68% of the kernels were rescued by the daily feeding of sucrose to the stems in this particular experiment. The plants were subjected to low w, photosynthesis had been nearly completely inhibited at the low w, and enough sucrose was fed to replace the missing sucrose completely. Photosynthetic activity recovered fully when the parent was rewatered, and kernel development continued. Typically, the feeding resulted in 40–80% rescue of the ovaries that otherwise would have completely aborted.
Because sucrose is the main translocation form for the carbon fixed in photosynthesis, this result indicated that abortion was controlled by sucrose in the sugar stream delivered from the leaves to the ovaries. Whether the change in sucrose delivery itself signals the pedicel and/or ovary to initiate the abortion process has not been resolved. But at the molecular level it is clear that these two tissues respond differently to water deficits. The pedicel tissue was more responsive in terms of gene expression with many stress tolerance proteins up-regulated. Genes associated with cell division and growth, however, apparently were unaffected (Zinselmeier et al., 2002; Yu and Setter, 2003). These studies confirm the importance of sucrose in the sugar stream and suggest a possible triggering effect of changes in sucrose delivery on molecular events in the pedicel or ovaries.
There is abundant evidence that sucrose is unloaded from the phloem in the pedicel and hydrolysed to glucose and fructose by acid invertases before being taken up by rapidly filling maize kernels (Shannon, 1968, 1972; Shannon and Dougherty, 1972; Felker and Shannon, 1980; ap Rees, 1984; Doehlert and Felker, 1987; Griffith et al., 1987; Porter et al., 1987; Miller and Chourey, 1992; Thomas et al., 1993; Xu et al., 1995; Cheng et al., 1996). This also seems to be true when ovaries alone are present, because Zinselmeier et al. (1995b, 1999) found high activities of acid invertases prior to fertilization. The invertases began to be active when the ovaries became susceptible to abortion.
The hexose products of the invertase reaction were either converted to starch and stored in the ovary walls or were converted to other constituents necessary for ovary development (Zinselmeier et al., 1999). When low w occurred, the starch disappeared from individual ovaries destined to abort (cf. Fig. 1D, F). Glucose followed suit and decreased as the starch disappeared (McLaughlin and Boyer, 2004). It appears that the starch supplied glucose to the developing ovaries when the sucrose stream diminished at low w, buffering against low glucose. But the starch is a small pool and quickly depleted. It would be sufficient to supply glucose for a night when photosynthetic products may be in short supply but could not make up for days of inhibited photosynthesis that occur at low w (McLaughlin and Boyer, 2004).
Schussler and Westgate (1991) reported that ovaries excised from plants having low w absorbed sucrose less rapidly than controls. This suggests that the uptake process may have been affected at low w. Because invertase hydrolyses sucrose to hexoses, its activity may play a part in uptake. The decreased invertase activity might then account for the slower sucrose uptake these workers observed at low w.
Schussler and Westgate (1994, 1995) further investigated whether stored reserves such as sugars and starch affected the tendency of ovaries to abort. Plants with high or low reserves were grown and subjected to low w. Regardless of reserve status, abortion occurred similarly in each treatment and the investigators concluded that the flux of photosynthate was more important than the reserve status of the parent plants for triggering abortion. It would have been interesting to determine the status of the starch in individual ovaries about to undergo abortion. It seems possible that ovary starch could be broken down despite high reserves in the rest of the parental tissues.
Recent reviews have proposed intriguing interactions between ABA and sugars that might regulate developmental processes in well-watered plants (Finkelstein and Gibson, 2002; León and Sheen, 2003). Setter et al. (2001) reported increasing sugar and ABA concentrations in maize florets during a water deficit. Reports that these compounds increase in concentration in florets of water deficient maize (Schussler and Westgate, 1995; Zinselmeier et al., 1995b, 1999; Setter et al., 2001) raise the possibility that such changes (or their interactions) might signal abortion processes to commence. The difficulty in assigning a ‘signal’ value to these observations is that it is unclear how much of the increases in sugar and ABA concentrations were attributable to lower water contents in the florets. Zinselmeier (1991) reported that a water deficit prior to pollination decreased ovary water content by about 3–4% (87% to 83% moisture) and was severe enough to cause abortion of nearly all pollinated florets. The observed change in concentrations therefore depends on whether water contents or photosynthate flux was more affected, which would be likely to vary between experiments.
Thus, it is difficult to establish a link between changes in concentrations of, say, ABA and sugars, when all other constituents have increased as well because of water loss. In support of this concept, when Wang et al. (2002) measured catabolic activity for ABA at low w, there was a large increase on a fresh-weight basis but less when expressed on a per-kernel basis (see Fig. 5 in Wang et al., 2002). Similarly, carbohydrate concentrations increased on a fresh-weight basis, but the accumulation of carbohydrate had actually decreased when expressed on a floret basis (see Figs 6 and 7 in Setter et al., 2001).
In an experiment similar to Setter et al. (2001), but conducted under field conditions, Andersen et al. (2002) reported some ovaries aborted as low w developed, but aborting and developing ovaries displayed no difference in ABA concentration expressed on a dry-weight basis. Like ovary water content, the dry-matter content of the ovaries is highly dependent on the photosynthate flux delivered to them. Sugars and starch can comprise as much as 70% of the ovary dry weight (Zinselmeier et al., 1999). As the delivery of sugars diminishes, the dry weight decreases relative to that of well-watered plants. Consequently, changes in ABA concentration depend on whether ABA or dry weight is more affected by the water deficit.
Andersen et al. (2002) found that sucrose, hexose, and starch concentrations were depleted in some low w treatments, but not in others. These investigators did not determine these concentrations in individual ovaries and instead sampled ears having a mixture of aborting and normal ovaries. Thus, it was inevitable that variable numbers of normal ovaries were included in the samples. Andersen et al. (2002) could detect no relationship between ovary abortion and ovary sugar and starch concentrations, but agreed with Schussler and Westgate (1994, 1995) and Zinselmeier et al. (1999) that abortion was controlled by the photosynthate flux to the ovaries.
When Zinselmeier et al. (1999) fed sucrose to the stems of maize, the photosynthate flux increased at low w and the starch pool was maintained in the ovaries (Fig. 1E). The presence of starch thus indicated whether ovary demand for sucrose (and glucose) was being met. Because the sucrose feeding prevented much of the abortion (Fig. 1B), Zinselmeier et al. (1999) concluded that aborting ovaries were starving for sugars at low w, and re-establishing the sugar stream prevented the abortion.
Zinselmeier et al. (1999) found that ovary starch, while high in the plants fed sucrose at low w, remained less than in the controls. Invertase activity was also not fully maintained by sucrose feeding. The downstream intermediates leading to starch biosynthesis were depleted at low w and only partially replenished by the feeding. Glucose and fructose are the products of invertase, and glucose concentrations were only partially maintained with sucrose feeding (McLaughlin and Boyer, 2004). The sucrose feeding rescued 40–80% of the otherwise aborting ovaries, and Zinselmeier et al. (1999) attributed the lack of 100% recovery to the remaining low invertase activity and depletion of downstream pools of metabolites.
These results suggest that molecular events triggering the low invertase activity were not fully reversed by feeding sucrose and might be responsible for the incomplete recovery from abortion with fed sucrose and thus the key to abortion itself. Moreover, in leaves and roots of young maize, Pelleschi et al. (1997), Kim et al. (2000), and Trouverie et al. (2003) report that invertase activity increased at low w. This marked contrast with the ovaries implies that the signal to abort development might be lacking (or undetected) in leaves and roots, most of which survive at low w, and present in ovaries, which abort.
Given these differences, possible changes in gene expression become of interest. In the maize leaves and roots at low w, increased invertase activity was accompanied by increases in mRNA abundance of Ivr2, a soluble acid invertase gene (Kim et al., 2000; Trouverie et al., 2003). By contrast, in maize ovaries, low activities of soluble and cell wall-bound acid invertases (Zinselmeier et al., 1995b, 1999; Andersen et al., 2002) were accompanied by lower mRNA abundance for Ivr2 and Incw2 (Andersen et al., 2002). This correlation between enzyme activity and mRNA abundance is consistent with the contrasting development between leaves and ovaries at low w. If the lower invertase activity triggers abortion, these results indicate that regulation occurs at least in part at the transcriptional level.
Nevertheless, the rescue of the ovaries by feeding sugar to the parent plant suggests that these transcriptional changes are either inconsequential for abortion or are responding to the sucrose. Koch (1996) and Sheen et al. (1999) reviewed the genes known to be sugar-responsive, and some invertases are in that group. Soluble acid invertase Ivr2 was reported to be up-regulated in response to increasing sugar levels, for example. Although the sugar-responsiveness for Incw2 and Ivr2 was not measured in the studies of Kim et al. (2000), Trouverie et al. (2003), or Andersen et al. (2002), the latter authors observed that lower Incw2 and Ivr2 mRNA levels corresponded to higher sucrose concentrations in the ovaries. They suggested that the invertases might respond to the quality of the sugar pool rather than its concentration, or possibly to other undetermined regulatory factors at low w. By contrast, working with leaves, Kim et al. (2000) reported increased sugar concentrations when the message increased for Ivr2, and Trouverie et al. (2003) could increase the mRNA for this gene by feeding abscisic acid to cut leaves. In view of the problems of working with bulk ovaries described above, further work seems necessary before the regulatory significance of these changes is understood.
Some evidence also supports a role for post-transcriptional regulation. A peptide in maize can act as an invertase inhibitor (Jaynes and Nelson, 1971; Bate et al., 2004) by binding non-competitively to acid invertase at low sucrose concentrations (about 1–2 mM in tobacco (Nicotiana tabacum L.) (Weil et al., 1994). When invertase is extracted without sucrose, it becomes susceptible to the inhibitor in vitro (Jaynes and Nelson, 1971; Pressey, 1967). It is conceivable that low sugar concentrations in vivo might have the same effect. Because the inhibitor is present in maize (Jaynes and Nelson, 1971; Bate et al., 2004), its action could prevent a close correlation between transcriptional activity and the in vivo activity of the enzyme in some tissues or treatments.
It may be important that low w also affects cell division. Setter and Flannigan (2001) found decreased transcript abundance for genes encoding enzymes for cell proliferation and endoreduplication in maize endosperm when the plants were subjected to low w a few days after pollination. DNA synthesis was markedly reduced, suggesting that cell division was inhibited. If abortion involves an irreversible block in ovary development, decreased cell division would be expected and might involve down-regulation of cell-cycle genes at the transcriptional level.
Zinselmeier et al. (2002) used cDNA microarrays to measure differential gene expression in maize ovaries and pedicel tissues when low w occurred around the time of pollination. More than 1500 genes representing 27 regulatory and metabolic pathways were investigated. Of these, 1–3% showed differential expression 4 d after moderate water stress was imposed, with some genes down-regulated and others up-regulated. A greater number of genes were differentially expressed as the stress intensified. Interestingly, more genes showed differential expression in the pedicel tissues than in the ovaries themselves. Zinselmeier et al. (2002) hypothesized that this difference might be due to a hydraulic isolation of the ovary from the rest of the plant, buffering it against large changes in w. However, measurements of ovary w in water-deficient plants indicate that ovary water status responds directly to changes in plant water status (Westgate and Boyer, 1986a; Westgate and Thomson Grant, 1989; Zinselmeier et al., 1999). Relatively small changes in ovary w can have a large impact on kernel set.
In a parallel study, Zinselmeier et al. (2002) observed a ‘coordinate down-regulation of genes associated with starch synthesis’ in ovaries on well-watered plants covered with 95% shade cloth during silk emergence. The decrease in gene expression was substantial and consistent with the dramatic decrease in ovary starch content and inhibition of ovary growth caused by the shade treatment. But in contrast to the shade results, the expression profiles at low w showed no evidence of altered gene expression for starch synthesis.
Yu and Setter (2003) conducted a similar study but withheld water from plants 5–9 d after pollination. Gene expression profiles from endosperm and pedicel tissues showed the water deficit affected 79 genes in the pedicel and 56 in the endosperm. Most were up-regulated in the pedicel but down-regulated in the endosperm. Curiously, there were no changes in expression of starch synthesis genes reported in either tissue. Like Zinselmeier et al. (2002), these authors attribute the differences between the tissues to the hydraulic isolation of the endosperm from the pedicel, but suggest in addition that they may be related to differential ABA concentrations.
Both of these microarray studies involved sampling tissues after the stresses were imposed or relieved. In one situation, this was after moderate or severe water deficit developed (Zinselmeier et al., 2002), and in the other, kernels were harvested during severe stress and 3 d after rewatering (Yu and Setter, 2003). In view of the transient expression of many genes and limited lifetime of their mRNA (Seki et al., 2002), these sampling times might have missed the critical changes in gene expression that respond to changes in the sugar stream and trigger abortion. A larger number of sampling times would help with this problem. For example, it might be expected that changes in genes associated with rapid growth would be affected before senescence genes at low w. Sampling at only one or two times would be likely to miss this difference.
As with the invertase activities and gene expression, these results with microarrays are difficult to reconcile with those of the sucrose feeding experiments. Because 40–80% of the ovaries could be rescued by feeding sucrose to the parent at low w, the changes in gene expression must have been inconsequential for the rescued ovaries or reversed by sucrose feeding. It should be noted that feeding sucrose to the stems did not change the w of the parent or the ovaries. The plants experienced the same low w whether they were fed or not. Therefore, any changes in gene expression caused by low w should have persisted in the fed plants, unless they responded to the ingredients being fed.
If the genes responded to the feeding, more genes would be considered sugar-responsive than are now known (Koch, 1996; Sheen et al., 1999). The differential patterns would represent massive metabolic responses to a lack of substrate at low w. These are not starvation responses because the up-regulated genes in the microarrays suggest that there were active transcriptional responses to low w. This argues against abortion as a passive starvation response, as proposed by Zinselmeier et al. (1999).
Many passive responses are observed at the enzyme level at low w (Todd, 1972; Kramer and Boyer, 1995) and are not necessarily caused directly by the low w. Rather, many appear to be caused by changed availability of substrates, cofactors, or other features of the chemical environment immediately surrounding the enzymes (Kramer and Boyer, 1995). The chemical environment could be altered by water loss from the cells (Potter and Boyer, 1973; Kaiser, 1987) or depleted because of decreased photosynthesis (Westgate and Boyer, 1985; Boyle et al., 1991), or compartmentalized differently because of distorted or ruptured membranes (Fellows and Boyer, 1976, 1978), or other local effects (Shaner and Boyer, 1976; Huang et al., 1975; Pankhurst and Sprent, 1975). As a consequence, no single factor can account for the changed activities, and the control of activity often involves alterations in the particular cell environment that each enzyme requires, in addition to any transcriptional and translational regulation.
Likewise for gene expression, some of the microarray results might develop from the physiological changes occurring at low w. Some genes might respond passively while others are actively regulated by promoters or other features of the genome. A critical experiment will be to determine the effect of sucrose feeding on gene expression at low w. Those genes responding to the feeding would be implicated in ovary rescue from abortion. The others responding to low w but not sucrose would probably be passive or not limiting to ovary development.
A central feature of abortion is the time lag between triggering events and the cessation of development apparent several days later. Because of this lag, studies usually require blind sampling of the ovaries, before abortion can be seen. The samples might include individuals that are not going to abort, causing measurements to have an underlying baseline of normal ovary development. Differences between treatments become small and variable depending on how many normal ovaries are included in the sample.
No early markers, physiological or molecular, are available for identifying aborting ovaries. The closest alternatives seem to be treatments that ensure virtually all the ovaries will abort or develop normally, as in Fig. 1. With this approach, the baseline of normal ovaries is small or non-existent in the abortion treatments, and the fate of the sampled ovaries can be assigned confidently. Constant, repeatable growth environments with high radiation are valuable tools for producing these treatments. Highly controlled glasshouses or field sites with stable and reproducible weather patterns are also useful.
Another feature of abortion is the reduced size of the ovaries. At low w, ovary growth is inhibited and fresh weight fails to accumulate. The fresh weight may even decline because of dehydration. Expression of data on a fresh-weight basis thus gives highly variable results. A similar argument applies to ovary dry weight. Changes in ovary constituents (sugars, ABA, mRNA, etc.) may be greater or less than changes in fresh or dry weights, and the concentrations will rise or fall depending on which factor changes the most. Most important, concentrations changing from altered fresh or dry weights will change similarly for all the other cell constituents, rendering their significance debatable.
There seems to be no totally satisfactory solution to this problem. The best that can be offered is to determine the content of constituents in the whole ovary rather than concentrations. By making the whole-ovary measurements before treatments are imposed and monitoring the change as the treatments progress, in vivo fluxes are revealed. Fluxes into or out of the ovary result from rates of enzymatic and physiological processes, and accumulation of the constituent indicates the difference between influx and efflux, i.e. whether production (phloem delivery, transcription activity, metabolite production) exceeds use (breakdown, metabolism, consumption in biosynthesis). In an ovary that increases in size, these principles still apply but the change in size also may account for some of the change in content. As a result, measuring contents before treatment (organ development is identical) and following the changes during the treatment (organ development differs) will indicate whether production exceeds use regardless of whether size increases or not.
The upward slope in Fig. 2 indicates that delivery exceeds use after the treatment is imposed. The effect of increased size cannot be determined from this approach, but delivery clearly exceeds use in any event. Conversely, if the constituent is consumed faster than it is delivered, there is a downward slope. Zero slope means the constituent is available at the same level as before the treatments were imposed. The advantage of this whole organ, time-based approach is that flux information for the organ is obtained from simple measurements of content regardless of changes in organ size, fresh weight, or dry weight. Because the fluxes indicate how rapidly enzymatic and metabolic processes are acting in vivo in whole metabolic systems, the information is readily interpretable for development.
It should be noted that these changes take place during early water limitation, which is of importance to agriculture. They often begin before visible symptoms appear and do not involve severe desiccation where other enzyme and gene mechanisms might come into play (Kramer and Boyer, 1995). It also is noteworthy that many enzyme and gene changes are set into motion, but the key will be finding those few changes that have an impact on plant performance, i.e. that are limiting.
It is remarkable that abortion of newly formed zygotes can be so complete while the parent plant can recover so fully. In maize, about half of the shoot dry weight is normally in the kernels at the end of the growing season. A water deficit at pollination can shrink this fraction to zero. The farmer is left with a reasonable vegetative crop but no grain. Important enzymes and genes must determine these two developmental fates, and revealing them is likely to require knowledge of whole plant physiology, biochemical regulation of the physiology, and the particular genes linking them together. With these genes identified, it may be possible to prevent the irreversible fate of the ovaries and pollen on plants subjected to low w.
Andersen MN, Asch F, Wu Y, Jensen CR, Naested H, Mogensen VO, Koch KE. 2002. Soluble invertase expression is an early target of drought stress during the critical, abortion-sensitive phase of young ovary development in maize. Plant Physiology 130, 591–604.
Cheng W-H, Taliercio EW, Chourey PS. 1996. The Miniature1 seed locus of maize encodes a cell wall invertase required for normal development of endosperm and maternal cells in the pedicel. The Plant Cell 8, 971–983.
Dorion S, Lalonde S, Saini HS. 1996. Induction of male sterility in wheat by meiotic-stage water deficit is preceded by a decline in invertase and changes in carbohydrate metabolism in anthers. Plant Physiology 111, 137–145.
Felker FC, Shannon JC. 1980. Movement of 14C-labelled assimilates into kernels of Zea mays L. An anatomical examination and microautoradiographic study of assimilate transfer. Plant Physiology 65, 864–870.
Griffith SM, Jones RJ, Brenner ML. 1987. In vitro sugar transport in Zea mays L. kernels. 1. Characteristics of sugar absorption and metabolism by developing maize endosperm. Plant Physiology 84, 467–471.
Kim J-Y, Mahé A, Brangeon J, Prioul J-L. 2000. A maize vacuolar invertase, IVR2, is induced by water stress: organ/tissue specificity and diurnal modulation of expression. Plant Physiology 124, 71–84.
Lalonde S, Beebe DU, Saini HS. 1997. Early signs of disruption of wheat anther development associated with the induction of male sterility by meiotic-stage water deficit. Sexual Plant Reproduction 10, 40–48.
Schoper JB, Lambert RJ, Vasilas BL, Westgate ME. 1987. Plant factors controlling seed set in maize: the influence of silk, pollen, and ear-leaf water status and tassel heat treatment at pollination. Plant Physiology 83, 121–125.
Seki M, Narusaka M, Ishida J, et al. 2002. Monitoring the expression profiles of 7000 Arabidopsis genes under drought, cold and high-salinity stresses using a full-length cDNA microarray. The Plant Journal 31, 279–292.
Setter TL, Flannigan BA. 2001. Water deficit inhibits cell division and expression of transcripts involved in cell proliferation and endoreduplication in maize endosperm. Journal of Experimental Botany 52, 1401–1408.
Sheoran IS, Saini HS. 1996. Drought-induced sterility in rice: changes in carbohydrate levels and enzyme activities associated with the inhibition of starch accumulation in pollen. Sexual Plant Reproduction 9, 1661–1669.
Trouverie J, Thévenot C, Rocher J-P, Sotta B, Prioul J-L. 2003. The role of abscisic acid in the response of a specific vacuolar invertase to water stress in the adult maize leaf. Journal of Experimental Botany 54, 2177–2186.
Yang J, Zhang J, Wang Z, Zhu Q. 2001a. Activities of starch hydrolytic enzymes and sucrose-phosphate synthase in the stems of rice subjected to water stress during grain filling. Journal of Experimental Botany 52, 2169–2179.
Zinselmeier C, Sun YJ, Helentjaris T, Beatty M, Yang S, Smith H, Habben J. 2002. The use of gene expression profiling to dissect the stress sensitivity of reproductive development in maize. Field Crops Research 75, 111–121.
Fig. 1. Ear development at maturity and ovary starch at pollination in maize subjected to low w for a few days around the time of pollination. Mature ears: (A) Controls from plants at high w, showing about 495±10 kernels per ear (SE for 9–11 ears). (B) Ears from plants at low w around the time of pollination, but having sucrose infused into the stems. These show about 340±15 kernels per ear. (C) Ears from plants at low w, producing 15±3 kernels per ear. Ears are somewhat crooked because ovaries were sampled in mid-ear around the time of pollination, which caused slight deformation at maturity. Sampled areas are on back of ear and not visible. For these experiments, w had decreased to about –1.45 MPa by the day of pollination, and net photosynthesis was near zero. The plants were rewatered 1 d later and kept hydrated until maturity (JE McLaughlin and JS Boyer, unpublished data). Ovary starch on the day of pollination: (D) Control ovary from plant at high w. Starch (region staining black) is located in the basal pedicel tissues around the phloem termini and in the ovary wall around the nucellus. There is no starch in the nucellus. Silk normally attached to the ovary apex has been removed. (E) Ovary from plant at low w around the time of pollination but having sucrose infused into the stems. Note that starch is present in the same location as in (D) but in lesser amount. (F) Ovary from plant at low w. Starch has nearly disappeared from pedicel and ovary wall. Bar indicates 1 mm. (Chemical analyses and detailed images are available in Zinselmeier et al., 1999, from which the starch images were adapted. ©American Society of Plant Biologists. Reprinted with permission.)
Fig. 2. Diagrammatic changes in ovary constituents with time, expressed on a whole ovary basis. By comparing contents before and after treatments are imposed, the change (slope) gives information about fluxes to the ovary. Upward slope indicates that production of the constituent (influx) exceeds use (efflux). Downward slope indicates production was less than use. Zero slope indicates that production equalled use. See text for further details.
Source: Journal of Experimental Botany 2004 55(407):2385-2394