Is Change in Ovary Carbon Status a Cause or a Consequence of Maize Ovary Abortion in Water Deficit during Flowering?^,

The first molecular events leading to maize ovary abortion involved genes affecting expansive growth rather than sugar metabolism, so the latter might be a consequence rather than causal.

Abstract

Flower or grain abortion causes large yield losses under water deficit. In maize (Zea mays), it is often attributed to a carbon limitation via the disruption of sucrose cleavage by cell wall invertases in developing ovaries. We have tested this hypothesis versus another linked to the expansive growth of ovaries and silks. We have measured, in silks and ovaries of well-watered or moderately droughted plants, the transcript abundances of genes involved in either tissue expansion or sugar metabolism, together with the concentrations and amounts of sugars, and with the activities of major enzymes of carbon metabolism. Photosynthesis and indicators of sugar export, measured during water deprivation, suggested sugar export maintained by the leaf. The first molecular changes occurred in silks rather than in ovaries and involved genes affecting expansive growth rather than sugar metabolism. Changes in the concentrations and amounts of sugars and in the activities of enzymes of sugar metabolism occurred in apical ovaries that eventually aborted, but probably after the switch to abortion of these ovaries. Hence, we propose that, under moderate water deficits corresponding to most European drought scenarios, changes in carbon metabolism during flowering time are a consequence rather than a cause of the beginning of ovary abortion. A carbon-driven ovary abortion may occur later in the cycle in the case of carbon shortage or under very severe water deficits. These findings support the view that, until the end of silking, expansive growth of reproductive organs is the primary event leading to abortion, rather than a disruption of carbon metabolism.

Water deficit (WD) largely decreases yield, with a maximum effect during flowering time in several species (Lilley and Fukai, 1994; Saini and Westgate, 2000; Rapoport et al., 2012). In maize (Zea mays), ovary development is highly drought sensitive (Boyer and Westgate, 2004) while pollen viability is not (Herrero and Johnson, 1981; Schoper et al., 1986, 1987). A disruption of carbon metabolism in ovaries has been suggested to be the main cause of abortion, based on a series of experiments showing that Suc feeding can partly reverse the effect of WD on abortion (Boyle et al., 1991; Zinselmeier et al., 1995a, 1999; McLaughlin and Boyer, 2004). These experiments had a common protocol in which a WD was imposed for 6 d following silk emergence (SE) and caused a drastic reduction in photosynthesis and starch content in ovaries (Zinselmeier et al., 1999). The Suc flux to ovaries decreased to near zero but was partly restored upon Suc feeding (Mäkelä et al., 2005). Enzyme activities and gene expression of cell wall invertases increased 5 to 8 d after silking in ovaries of well-watered (WW) plants, whereas they remained low under WD (Zinselmeier et al., 1995b, 1999; McLaughlin and Boyer, 2004). This led several authors to consider cell wall invertases, in particular INCW2, as a causal link between WD, sugar availability to ovaries, and ovary abortion (Boyer and McLaughlin, 2007; Ruan et al., 2012). We use the term abortion here in its broad sense: arrest of development of an organ (flower, ovary, grain, etc.). Therefore, it is not restricted to the failure of zygotic development after fertilization but also includes the developmental arrest of ovary (or floret) occurring earlier.

Several arguments have led us to question the generality of the link between carbon metabolism and ovary abortion under WD.

(1) A decreased photosynthesis at flowering time has a lesser effect on ovary abortion if caused by low light rather than by WD (Schussler and Westgate, 1991; Hiyane et al., 2010), whereas drought-caused abortion can occur without appreciable depletion of ovary sugar content (Schussler and Westgate, 1995; Andersen et al., 2002). This suggests a direct effect of low water potential, independent of assimilate supply (Schussler and Westgate, 1991).

(2) Several studies indicate that carbon availability to growing organs is increased by moderate WDs in different species because growth processes are more affected than carbon assimilation (Hummel et al., 2010; Muller et al., 2011; Pantin et al., 2013). The dependency of ovary abortion upon carbon supply, therefore, might be lower under moderate drought scenarios representative of European field conditions (Harrison et al., 2014) than under the severe drought stress occurring in the studies mentioned by Boyer and McLaughlin (2007) and Ruan et al. (2012).

(3) We show in a companion article that the causal link between WD at flowering time and ovary abortion involves the growth arrest of silks 2 to 3 d after first SE (Oury et al., 2016). In this case, abortion concerns the youngest apical ovaries whose silks did not emerge 2 d before silk arrest, irrespective of ovary growth rate (Oury et al., 2016). This suggests that the carbon supply to ovaries would not be involved directly, and one could expect that the earliest molecular events associated with abortion should occur in silks rather than in ovaries. Furthermore, the changes in ovary sugar content and enzyme activities, measured in several studies 5 to 8 d after SE (Zinselmeier et al., 1995b; Andersen et al., 2002), would occur when ovaries were already engaged in the abortion process for several days.

We have tested with a molecular approach the hypothesis proposed in a companion article (Oury et al., 2016) that ovary abortion under WD is caused by silk growth arrest 2 to 3 d after first visible silk and not by carbon availability. For that, we have measured the transcript abundances of genes involved in tissue expansion or in sugar metabolism, together with the concentrations and amounts of sugars and enzyme activities in silks and ovaries at several phenological stages encompassing SE. The resulting analysis supports the view that the first molecular events associated with WD in reproductive organs occur in silks rather than in ovaries and involve genes affecting expansive growth rather than sugar metabolism. Hence, we propose that changes in the transcript amounts and activities of enzymes involved in ovary sugar metabolism, and changes in the concentrations and amounts of sugars 5 d after SE, could be a consequence rather than a cause of the beginning of ovary abortion.

RESULTS

Transient WDs during Flowering Time Affected Ovary Abortion, Silk and Ovary Growth, and Sugar Content in the Same Manner in Hybrid B73xUH007 and Inbred Line B73

WD was imposed on plants of the B73xUH007 hybrid (experiment 1) or of the B73 inbred line (experiment 2). In both experiments, a common treatment with moderate water deficit (WD1) was imposed for 10 d, from tassel emergence to 6 d after SE, with a soil water potential of −0.3 MPa (Fig. 1). Transpiration was reduced in WD1 to 74% and 67% and photosynthesis to 71% and 74% of their values in WW plants in experiments 1 and 2, respectively (Table I). Abortion occurred at the ear tip in both experiments (Fig. 2B), as it does in field studies (Supplemental Fig. S2A). WD had no significant effect on the number of initiated ovaries (Fig. 2A), so the reduction in grain number was not due to a lack of floret initiation but to ovary (before fertilization) or grain (after fertilization) abortion. Pollen sterility was not involved in abortion because hand pollination was performed every day until the end of SE with fresh pollen of WW plants.

Figure 1.

Time course of sampling dates (A and D), soil water potential (B and E), and daily transpiration (C and F) during the period of WD in hybrid B73xUH007 (left column; experiment 1) and line B73 (right column; experiment 2). HE, Husk emergence; SE+5d and SE+15d, 5 and 15 d after SE, respectively; TE, tassel emergence (beginning of WD). Green, WW; red, WD1. Error bars represent 95% confidence intervals (n ≥ 3).

Table I. Effect of soil WD on grain loss, photosynthesis, transpiration, date of silk growth arrest, fresh weight in silks and ovaries, and soluble sugar content in silks and ovaries at SE.

GC, Growth chamber; GH, greenhouse. WD1, WD2, and WD3 represent soil WDs imposed from tasseling to 5 d after SE. Different letters within a column indicate significant differences in a Kruskal-Wallis test (P < 0.05).

Experiment	Location	Genotype	Treatment	Soil Water Potential	Grain Loss	Photosynthesis	Transpiration	Date of Silk Growth Arrest	Silk Fresh Weight at SE	Ovary Fresh Weight at SE	Silk Soluble Sugar Content	Ovary Soluble Sugar Content
				MPa	% WW			d after SE	% WW
Experiment 1	GH	B73xUH007	WW	−0.08a	0 a	100 a	100 a	6.5 a	100 a	100 a	100 a	100 a
			WD1	−0.28b	49 b	71 b	74 b	3.2 b	72 b	92 a	113 a,b	130 b
Experiment 2	GC	B73	WW	−0.08a	0 a	100 a	100 a	5.7 a	100 a	100 a	100 a	100 a
			WD1	−0.30b	42 b	74 b	67 b	2.4 b,c	60 b	95 a	129 b	129 b
			WD2	−0.50c	64 c	59 c	49 c	1.9 c	33 c	104 a	153 c	128 b
			WD3	−0.60d	77 d	45 d	40 d	1.3 d	20 d	103 a	–	127 b

Figure 2.

A, Grain number per ear (colored bars) and number of initiated ovaries per ear (white bars) in experiments 1 and 2. Error bars represent se (n ≥ 3). Different letters indicate significant differences in a Kruskal-Wallis test (P < 0.05). B, Ears 15 d after SE in experiments 1 and 2. Bar = 1 cm. C, Grain number as a function of the number of emerged silks at the date of silk growth arrest. Green, WW; red, WD1. Each point represents the mean value in one experimental treatment from experiment 1 (hybrid B73xUH007; colored triangles), experiment 2 (line B73; circles), and Oury et al. (2016; four hybrids; gray triangles). Arrows indicate the WD1 treatments, in which the characterization of metabolite content and enzyme activities (experiment 1; red triangle) or of transcript levels (experiment 2; red circle) was performed.

Responses to soil WD shared common features in the experiments with either the hybrid B73xUH007 or the inbred line B73 (Table I). First, total silk fresh weight was already reduced significantly when the first silk emerged, while ovary fresh weight was not yet affected. Second, soluble sugars accumulated during WD in silks and ovaries, suggesting that expansive growth was more affected than carbon availability in both organs. Third, silk growth and SE stopped 1 to 3 d after first SE in WD plants versus 6 to 7 d in WW plants. Finally, WD caused losses of grain number by 36% to 77% depending on the severity of the WD (Table I), with 98% of ovary abortion related to the number of emerged silks at the date of silk growth arrest (Fig. 2C). Indeed, a common relationship was observed between final grain number and silk number on the day of silk growth arrest, suggesting that the switch to abortion in apical ovaries of WD plants was triggered 1 to 3 d after SE and was associated with silk growth arrest. Hence, abortion in WD plants concerned both florets with nonexposed silks and, to a lesser extent, ovaries whose silks emerged less than 2 d before silk growth arrest and, therefore, were in contact with pollen. This relationship applied indifferently to the hybrid in WW and WD1 treatments of experiment 1, to line B73 in the four treatments of experiment 2, and to four hybrids and three treatments analyzed by Oury et al. (2016).

As a consequence, common causes of abortion and, probably, common mechanisms operated in the whole data set in both the B73 line and the hybrid. All results presented hereafter involve the hybrid, except for the transcriptome analysis that was performed in the inbred line for a better matching of probes that were specific of the B73 line (experiment 2, WD1).

The Growth of All Reproductive Organs Was Affected, with an Irreversible Effect on Ovaries Located at the Ear Tip

Ovary fresh weight was maintained in WD1 plants at all positions of the ear until SE (Fig. 3, D and E). Over the following 5 d, it was affected by WD with similar effects in both basal and apical ovaries (Fig. 3, D and E). Ovary volume (linearly related to fresh weight; Supplemental Fig. S3) had no straightforward relation with abortion frequency (Supplemental Fig. S1). In contrast, silks and husks were already affected from SE onward (Fig. 3, A and F), and the peduncle was affected even earlier (Fig. 3B). The growth of all organs of WD1 plants resumed after rewatering, including basal ovaries that recovered rapidly, with the exception of apical ovaries that did not grow over 9 d after rewatering and, therefore, were irreversibly arrested.

Figure 3.

Fresh weight (FW) of husks (A), peduncle (B), cob (C), basal ovaries (D), apical ovaries (E), and silks (F) as a function of developmental stages in hybrid B73xUH007 in experiment 1. Green, WW; red, WD1. HE, Husk emergence; SE+5d and SE+15d, 5 and 15 d after SE, respectively. Error bars represent se (n ≥ 6 at HE, SE, and SE+5d; n ≥ 3 at SE+15d). Different letters indicate significant differences in a Kruskal-Wallis test (P < 0.05).

The Carbon Export from Source Leaves Was Maintained in WD1 Plants

No clear temporal tendency was observed for photosynthesis, metabolite contents, and enzyme activities in leaves of WD1 plants during the period from husk emergence to 5 d after SE, so results are presented in Figure 4 as means over the whole period (four sampling dates). Suc content in leaves and activities of Suc phosphate synthase (SPS; EC 2.4.1.14) and cytosolic Fru-1,6-bisphosphatase (FBPase; EC 3.1.3.11), both involved in Suc synthesis for export, were maintained in WD1 plants at the same level as in WW plants (Fig. 4, C–E). Hexose contents tended to be higher in WD1 than in WW leaves: differences were significant for Glc content (11.1 versus 2.5 µmol g⁻¹) but not for Fru content (7.4 versus 5.1 µmol g⁻¹; Fig. 4C). Leaf starch content did not differ significantly between WD1 and WW plants, with a tendency to a reduction by 20% at the end of the light period and by 30% at the end of the dark period (Fig. 4B). This suggests that starch might be metabolized more rapidly in WD1 plants, thereby contributing to maintain Suc and hexose concentrations. A higher mobilization of starch reserves in WD1 plants during the night, when plant water status partly recovered, could sustain Suc export to sink reproductive organs (Fig. 4B). Overall, these results show that Suc export from source leaves was probably not appreciably affected by WD. This, coupled with the unaffected ovary growth in WD1 plants until SE and lower sink strength of peduncles (lower growth rate), suggests that the flux of Suc to ovaries was maintained until SE.

Figure 4.

A, Net photosynthesis (Pn; µmol CO₂ m⁻² s⁻¹), in mean values from 3 d after the beginning of WD to rewatering time. B and C, Starch in μmol Glc equivalent g⁻¹ fresh weight (B) and soluble sugar content, Suc (green or red), Glc (white), and Fru (gray), in µmol g⁻¹ fresh weight (C). D and E, FBPase (D) and SPS (E) activities (nmol min⁻¹ mg⁻¹ protein) in hybrid B73xUH007 in experiment 1. Green, WW; red, WD1. Metabolite content and enzyme activities were measured at the end of the day (ED) and 10 to 11 h later at the end of the night (EN). Error bars represent se (n ≥ 4). Different letters indicate significant differences between the four modalities (period of the day × water treatment) in a Kruskal-Wallis test (P < 0.05).

The Changes in Transcript Abundance in Ovaries and Silks Involved Genes Related to Expansive Growth Rather Than to Carbon Metabolism

In silks sampled at SE, genes involved in tissue expansion were much more expressed in WW than in WD1 plants (Fig. 5A). This was the case for genes involved in cell wall mechanical properties that drive expansive growth (Cosgrove, 2005; Babu et al., 2013; Xiao et al., 2014), namely four expansins and pectinases (exopolygalacturonase). Conversely, the transcripts of genes involved in carbon metabolism showed small differences in expression between WW and WD1 plants (Fig. 5B). Only two genes (four probes) out of 67 (92 probes) representing major gene families involved in sugar metabolism showed small but significant differences in transcript amounts, namely one Suc synthase and one TPP. Because TPP expression is increased by carbon deprivation (Yadav et al., 2014), the decrease observed in WD1 plants (Fig. 5B) suggests that their silks did not lack carbon.

Figure 5.

Transcriptome analysis at SE in silks in inbred line B73 in experiment 2. A, Genes involved in expansive growth and/or in cell walls. B, Genes involved in carbon metabolism. Colors represent the ratio of expression between WW and WD1 plants (see chart in Fig. 6C): green, higher expression in WW; red, higher expression in WD1. The details corresponding to each probe, identified with the numbers at right, are presented in Supplemental Table S1 (carbon metabolism) and Supplemental Table S2 (expansive growth). The criterion for selecting probes was a significant difference coupled with a ratio greater than 2 between the water treatments. Probes were selected when the criterion was fulfilled in silks (A; expansive growth) or in either silks or ovaries at least at one sampling date (B; carbon metabolism), so the list of probes involved in carbon metabolism (Supplemental Table S1) was common for both organs. No significant differences (Student’s t test, P > 0.05; n ≥ 4) and differences less than 2-fold are represented as black rectangles. CWI, Cell wall invertase; FLA, fasciclin-like arabinogalactan protein; NI, neutral invertase; SuSy, Suc synthase; TPP, trehalose 6-phosphate phosphatase; TPS, trehalose 6-phosphate synthase; UI, uncharacterized invertase; VI, vacuolar invertase.

In ovaries sampled at SE, differences in transcript abundance were very small between WW and WD1 plants in both basal and apical regions of the ear, for genes involved in either carbon metabolism or expansive growth (Fig. 6, A and B). Only one gene (the Suc synthase SHRUNKEN1), out of 67 (92 probes) tested for their role in carbon metabolism, showed a small significant difference. In particular, the transcript abundances of cell wall invertases, neutral invertases, and vacuolar invertases were not affected significantly by WD at this stage (Supplemental Table S1). Five days later, a clear difference appeared between basal and apical ovaries. The strongest differences were observed for genes involved in water transfer, such as aquaporins, and in cell wall mechanical properties, such as expansins, XET, cellulase, and pectin esterase, which were significantly and considerably more expressed in WW than in WD1 plants in apical but not in basal ovaries. The same applied, to a lesser extent, to genes involved in carbon metabolism (Fig. 6B; Supplemental Table S1). Several enzymes involved in starch metabolism, such as one starch synthase, one isoamylase, and several Glc-6-P/phosphate translocators, were more expressed in apical ovaries of WW than of WD1 plants, suggesting that apical ovaries of WW plants were beginning starch accumulation while those in WD1 were blocked. Several genes involved in sugar sensing, such as hexokinases (Granot et al., 2013) or trehalose 6-phosphate synthase (Yadav et al., 2014), were more expressed in WD1 than in WW plants, suggesting that starch accumulation in apical ovaries of WD1 plants was not restricted by Suc availability. Finally, several invertases were overexpressed in apical ovaries of WW compared with WD1 plants, in particular one cell wall invertase (INCW1) and to lesser extent neutral invertases (Fig. 6B; Supplemental Table S1). Differences were not significant in basal ovaries, except for one vacuolar invertase (INV2).

Figure 6.

Transcriptome analysis for ovaries located at either basal or apical positions on the ear at SE and SE + 5 d (SE+5) in inbred line B73 in experiment 2. A, Genes involved in expansive growth and/or in cell walls, B, Genes involved in carbon metabolism. C, Color chart for the ratio of expression between WW and WD1 plants: green, higher expression in WW; red, higher expression in WD1. The details corresponding to each probe, identified with the numbers at right and selected with the criteria described in Figure 5, are presented as Supplemental Table S1 (carbon metabolism) and Supplemental Table S3 (expansive growth). Insignificant differences (Student’s t test, P > 0.05; n ≥ 3) and differences less than 2-fold are represented as black rectangles. AGP, Arabinogalactan protein; CWI, cell wall invertase; NI, neutral invertase; TPS, trehalose 6-phosphate synthase; UI, uncharacterized invertase; VI, vacuolar invertase; WAK, wall-associated kinase; XET, xyloglucan endotransglycosylase.

Taken together, these results do not indicate a clear disruption of carbon metabolism in either silks or ovaries at SE. Differences in transcripts of genes involved in carbon metabolism were only observed in apical ovaries 5 d after SE, a date by which the switch to abortion had already been triggered. Hence, observed differences in the expression of genes involved in starch metabolism and those of invertase in apical ovaries may be seen as a consequence of the switch to abortion that stopped ovary growth, starch accumulation, and the cleavage of Suc. These results are consistent with the time sequence leading to abortion in both the B73xUH007 hybrid and the B73 inbred line (Fig. 2; Table I; Oury et al., 2016).

The Concentrations of Sugars Were Maintained or Increased in Reproductive Organs of WD Plants

The concentrations of starch, Suc, Glc, and Fru were measured in basal and apical ovaries, basal and apical silks, and cobs (Fig. 7, A–D) at husk emergence, at SE, and 5 d after SE. The total osmolyte content, calculated from the osmotic potential, increased with WD plants in all studied organs (Fig. 7, E–H). This occurred in ovaries before any response of fresh weight accumulation, suggesting a role for osmotic adjustment in growth maintenance. The osmolyte content due to soluble sugars, calculated from soluble sugar concentration and organ water content, was the main component of total osmolyte content, especially in ovaries and cobs (Fig. 7, E–H).

Figure 7.

Sugar content per unit of fresh weight (FW; top row) and osmolyte content (bottom row) as a function of developmental stage in basal ovaries (A and E), apical ovaries (B and F), basal silks (C and G), and apical silks (D and H) in hybrid B73xUH007 in experiment 1. From the bottom up in each bar: Suc, Glc (Glu), Fru, and starch for sugars and Suc, Glc, Fru, and other osmolytes (Os) for osmolyte content. Other osmolytes correspond to the difference between the total osmolyte content, calculated from osmotic potential, and the accumulated osmolyte contents due to soluble sugars. Green gradient, WW; red gradient, WD1. HE, Husk emergence; SE+5d, 5 d after SE. Values are means of samples harvested at the end of the day and 10 to 11 h later at the end of the night. Error bars represent se (n ≥ 6). Different letters indicate significant differences in a Kruskal-Wallis test (P < 0.05).

Suc, Glc, and Fru concentrations were at least equal in reproductive organs of WD1 compared with WW plants, regardless of date and position on the ear (Fig. 7, A–D). WD significantly increased the Suc concentration in ovaries and silks at both basal and apical positions and in the cob on the first two sampling dates, when no difference in growth rate was observable yet (Figs. 3 and 7, A–D). Suc concentration was maintained in the peduncle, for which growth was reduced from the first sampling date onward (Fig. 3; Supplemental Fig. S4). Five days after SE, the Suc concentration continued to be equal or higher in reproductive organs of WD1 plants compared with WW plants. These trends also applied to Glc and Fru concentrations (Fig. 7, A–D) in this experiment and to soluble sugars in experiment 2 carried out on line B73 (Table I).

Starch content was not affected by WD, regardless of the considered organs, except at the first sampling date in basal ovaries and silks, in which growth was not reduced yet. The maintenance of starch storage in all reproductive organs during the whole period of WD suggests that carbon availability was not limiting.

The amounts of sugars and starch also were calculated on an organ basis (Supplemental Fig. S5). The effect of WD on metabolite amounts per organ closely followed the effect on concentrations on the first two sampling dates, because growth was marginally affected by WD for ovaries, silks, and cobs at this time. The WD induced an accumulation of Suc and a maintenance of hexose and starch amounts per organ until SE, indicating that the fluxes of photosynthates toward these organs were at least maintained during the early phase of WD. Five days after SE, the amounts per ovary and per silk were lower for all sugars in WD1 plants. This was due to the drought-associated reduction in growth at this stage. This reduction of growth was not accompanied by a change in the pattern of sugar concentrations in WD1 plants in ovaries and silks, regardless of their position on the ear. In particular, the pattern of contents of different sugars was similar in apical and basal ovaries, indicating a similar carbon status in spite of the putative switch to abortion in apical but not basal ovaries. This suggests that the decrease in sugar amounts per organ probably followed, rather than preceded, the reduction in expansive growth due to WD.

WD Caused Limited Changes in Enzyme Activities in Ovaries

The activity of SPS, involved in Suc synthesis, was at least maintained under WD in all reproductive organs until 5 d after SE (significant increases at the first and third sampling dates except in silks; Fig. 8). This maintained or enhanced SPS activity in WD1 plants is consistent with the increase in Suc content (Fig. 7, A–D). The high level of SPS activity in WD1 plants, therefore, maintained Suc availability for its cleavage by invertases and SuSy (EC 2.4.1.13).

Figure 8.

Time courses of enzymes activities in ovaries and silks in hybrid B73xUH007 in experiment 1. From top to bottom: SPS (A, F, K, and P), cell wall invertase (CWI; B, J, L, and Q), neutral invertase (NI; C, H, M, and R), vacuolar invertase (VI; D, I, N, and S), and SuSy (E, J, O, and T) activities in basal ovaries (A–E), apical ovaries (F–J), basal silks (K–O), and apical silks (P–T) at three sampling stages in WW plants (green solid lines and circles) or WD1 plants (red dashed lines and circles). Values are means of samples harvested at the end of the day and 10 to 11 h later at the end of night. Error bars represent se (n ≥ 3). Asterisks indicate significant differences in a Kruskal-Wallis test (P < 0.05).

Changes with time and between organs in vacuolar and neutral invertase (EC 3.2.1.26) activities (Fig. 8) paralleled those of growth in WD1 plants (Fig. 3). Activities were reduced significantly by WD1 at the first sampling date in peduncle and silks but only at the last sampling date in ovaries and cobs. Vacuolar invertase activity was correlated significantly with the relative growth rate of organ fresh weight at pollination (Fig. 9). Vacuolar and neutral invertase activities increased significantly in all reproductive organs during the dark period, known to be favorable for expansive growth processes (Supplemental Fig. S6).

Figure 9.

Relationship between vacuolar invertase activity (VI act.) and relative growth rate (RGR) of organs at SE in experiment 1 in hybrid B73xUH007. In the inset, the x axis is on a logarithmic scale. Green, WW; red WD1. Large circles, Basal ovaries; small circles, apical ovaries; diamonds, cob; squares, peduncle; triangles, silks; reverse triangles, leaf. Error bars represent 95% confidence intervals (n ≥ 3). The organ relative growth rates between husk emergence (t1) and SE (t2) were calculated from the average fresh weight measured at t1 (FW1) and t2 (FW2): RGR = log(FW2/FW1)/(t2 – t1).

DISCUSSION

A Limitation of Water-Induced Expansive Growth Rather a Carbon Limitation in Reproductive Organs under Moderate WD

Even if we cannot exclude that some responses in gene expression might be specific to the inbred line in relation to the hybrid, we have hypothesized that the risk of false positive or negative results would have been increased if we had used in the hybrid a chip designed for the inbred line. The time course of events (photosynthesis, sugar accumulation, and position-time model of silk and ovule growth) leading to abortion was similar in the inbred line and the hybrid and is consistent with our position-time model of abortion (Oury et al., 2016).

Reproductive organs were rapidly expanding during the studied period, as indicated by growth in fresh weight, high expression of genes involved in cell wall mechanical properties, and high hexose contents and acid invertase activities (both cell wall and vacuolar). The latter is involved in cell and tissue expansion in numerous studies (Morris and Arthur, 1984; Sturm and Tang, 1999; Tang et al., 1999; Kohorn et al., 2006; Wang et al., 2010; Ma et al., 2014). WD reduced expansive growth without altering the carbon status of ovaries and silks as measured on a fresh weight basis; in particular, it induced soluble sugar accumulation and starch synthesis in all reproductive organs, as did an osmotic stress due to salt (Henry et al., 2015). This is confirmed by results concerning the trehalose biosynthetic pathway in reproductive tissues of maize plants subjected to osmotic stress. Trehalose-6-phosphate (T6P) functions as a signaling intermediate for reporting the cellular Suc status (Lunn et al., 2006; Yadav et al., 2014), and its content is increased in maize ovaries during salt stress (Henry et al., 2015) and WD (Nuccio et al., 2015). The reduction of transcript abundance of TPP, the enzyme converting T6P to trehalose, observed in silks of WD1 plants in this study (Fig. 5B) suggests an increased level of T6P and, therefore, a favorable carbon status of silks in WD1 plants. This is consistent with studies indicating that moderate WDs induce a carbon satiation, because expansive growth of sink organs is more affected than photosynthesis (Hummel et al., 2010; Muller et al., 2011; Pantin et al., 2013). The facts that the ratio between total invertase and SuSy activities was lower in ovaries than in silks, especially at SE (Fig. 8), and that SuSy did not respond to WD reinforce the idea that the energy status was not limiting for silk growth. Indeed, the breakdown of Suc into hexose phosphates requires only one inorganic pyrophosphate when initiated via SuSy and two ATPs when initiated via invertase (Geigenberger, 2003). Conversely, the transcript amounts of genes involved in water movements (aquaporins) and in cell wall properties (expansins and XET) were more expressed in ovaries of WW than of WD1 plants. Hence, it can be considered that the growth in fresh weight was limited by expansive growth, involving water entry into cells and cell wall mechanical properties, rather than by carbon availability that is largely independent of expansive growth under WD (Hummel et al., 2010; Tardieu, 2012; Pantin et al., 2013; Tardieu et al., 2014).

No trait involved in carbon metabolism could account for the abortion of apical ovaries, except the absence of increase in the activity and transcript abundance of cell wall invertases 5 d after SE in ovaries with WD. The latter result is consistent with previous studies (Zinselmeier et al., 1995b, 1999; McLaughlin and Boyer, 2004) suggesting that the differential behavior of cell wall invertases in WW and WD plants is the cause of abortion. However, both basal and apical ovaries of plants in WD showed this absence of increase, while only apical ovaries aborted in our study. Furthermore, our companion article (Oury et al., 2016) strongly suggests that the switch to abortion occurs 1 or 2 d after SE in apical ovaries of water-deficient plants. The differential behavior of cell wall invertases, which was not yet significant at SE, therefore would occur after the switch to abortion. Thus, we suggest that it is a consequence and not a cause of ongoing abortion in apical ovaries, consistent with the fact that the mutation miniature kernel, which drastically reduces invertase activity in ovaries and disturbs the storage of carbohydrates in the endosperm, does not cause grain abortion (Miller and Chourey, 1992; LeClere et al., 2010).

It was demonstrated recently that the targeted overexpression of a TPP to developing maize ears improves the allocation of carbon to the ear during a WD and reduces grain abortion (Nuccio et al., 2015). Because it signals carbon starvation although Suc availability was high, the transgene probably modified sugar utilization. One could hypothesize an increased use of hexoses to contribute to osmotic potential and, therefore, maintain expansive growth in ovaries and silks. The magnitude of yield improvement due to the transgene in field studies (Nuccio et al., 2015) could be explained by a delay of 1 d in the date of silk growth arrest in our model (Fig. 2C; Oury et al., 2016).

Vacuolar Invertase Activity as an Indicator of Expansive Growth Rate under WD

The experiment presented here shows an early decrease in silk growth in WD before any effect on ovary growth. Consistently, the effect of WD on vacuolar invertase activity was observed 5 d earlier in silks than in ovaries and was already visible at SE (Fig. 8, D and N). The activity of vacuolar invertase was higher at the end of the night than at the end of the day (Supplemental Fig. S6G); that is, it followed the time course of expansive growth (with maximum rates during the night; Turc et al., 2016) and not that of photosynthesis (with maximum rates during the day). These differential behaviors suggest that vacuolar invertases followed expansive growth in our experiments. High activities of vacuolar invertases are associated with rapid elongation in organs such as epicotyl of pea (Pisum sativum), roots of Arabidopsis (Arabidopsis thaliana), cotton (Gossypium hirsutum) fibers, or maize roots and silks (Morris and Arthur, 1984; Miller and Chourey, 1992; Wang et al., 2010), whereas decreasing the vacuolar invertase activity causes decreased organ and cell sizes (Kohorn et al., 2006).

Taken together, these arguments suggest that the time course of vacuolar invertase activity in ovaries and silks was not linked to changing Suc availability but to expansive growth, itself linked to the flux of water into growing cells and to cell wall mechanical properties (Cosgrove, 2005). The early effect of WD on vacuolar invertase activity in silks at SE, therefore, would reinforce the conclusion that silk growth was a major process involved in ovary abortion (Oury et al., 2016).

A Carbon-Independent Process Associated with the Sequential Emergence of Silk Cohorts, Followed by a Carbon-Dependent Cause of Abortion

The results presented here apparently contradict those of previous studies (Boyle et al., 1991; Zinselmeier et al., 1995b, 1999; McLaughlin and Boyer, 2004). In particular, the ovary carbon status differed, with a cessation of the Suc flux toward ovaries and starch depletion in ovaries in the previous studies, versus a maintained starch storage, maintained Suc and hexose contents in ovaries, and maintained Suc and starch contents in leaves in our study. We suggest that the mechanisms presented here and those in the above mentioned series of experiments coexist but at different periods and in different conditions.

In our study, the decrease in silk growth and in the expression of genes associated with silk expansive growth preceded any event affecting apical ovaries. This suggests a switch to abortion linked to silk expansive growth via the sequential emergence and pollination of silks of plants under WD, as proposed by Oury et al. (2016). The analyses of transcriptomes, of metabolite concentrations, and of enzyme activities strongly suggest a carbon-independent process. Consistent with the mechanism linked to sequential silk development, the distribution of abortion frequency on the ear had a clear base-apex pattern in our study as in most field studies (Fig. 2B; Supplemental Fig. S2A).

These developmental mechanisms were avoided in the previous studies, in which WD began after sequential SE and pollination occurred at a single date at the end of SE. Because the WD was stronger and occurred slightly after the period reported in this study, a carbon-related abortion occurred and was partly relieved by Suc feeding (Zinselmeier et al., 1999). Reductions in Suc flux to young grains and in the activities of enzymes linked to Suc cleavage were involved in the process of abortion reported in the previous studies but also were observed in our study 5 d after SE. It is noteworthy that the carbon-driven abortion led to a random distribution of abortion frequency on the ear in plants subjected to WD, with or without Suc feeding (McLaughlin and Boyer, 2004; Supplemental Fig. S2D). This distribution is clearly different from that observed in our study.

The results presented here, therefore, suggest that two periods with different dependency on carbon supply coexist. (1) From tassel emergence to 1 to 2 d after SE in WD, processes related to expansive growth would be predominant, as shown in this study, and result in the tip abortion observed in many field studies. (2) A carbon-dependent period of sensitivity of the young grain would begin with starch storage in fertilized embryos 5 d after SE and finish when grain number becomes insensitive to WD, 20 d after SE (Claassen and Shaw, 1970; Grant et al., 1989). The first period had a major effect in our study, in which the WD was moderate during flowering time, as it is in most drought scenarios in Europe (Harrison et al., 2014). A severe WD at flowering time, causing a drastic reduction in photosynthesis, may have an appreciable effect on abortion rate via carbon limitation in the field, as it had in several studies in controlled environments (Boyle et al., 1991; Schussler and Westgate, 1991; Zinselmeier et al., 1995b, 1999; McLaughlin and Boyer, 2004). This case was estimated as 18% of drought scenarios in Europe by Harrison et al. (2014).

MATERIALS AND METHODS

Plant Material and Growth Conditions

Maize (Zea mays) plants (hybrid B73xUH007 in experiment 1 and inbred line B73 in experiment 2) were grown in a greenhouse in 9-L cylindrical plastic pots at a density of three plants m⁻². Three seeds per pot were sown in compost composed of clay and peat (30:70, v/v) enriched in minerals. Plants were thinned to one per pot 3 d after emergence. Each pot was placed on a scale that allowed quantifying plant transpiration and calculating soil water content every 15 min from pot weight corrected by estimated plant weight. The latter was estimated by regularly sampling plants. A water-release curve of the soil was obtained by measuring the soil water potential of soil samples with different water contents (WP4-T Dewpoint Meters; Decagon Devices), allowing calculation of the mean soil water potential in each soil column (Caldeira et al., 2014). All plants were maintained at a soil water potential above −0.11 MPa by daily irrigation, except for a short period around flowering time for plants in WD treatments (Fig. 1, A and D). At tassel emergence, recorded individually for each plant, water was withheld for 2 d until soil water content reached the desired soil water potential (Fig. 1, B and E; Table I). Pots were then watered individually with the necessary volume to maintain soil water potential at the targeted value (for details, see Caldeira et al., 2014). Three levels of soil WD were performed in experiment 2, with soil water potential targeted values of −0.3, −0.5, and −0.6 MPa in WD1, WD2, and WD3 treatments, respectively, while only treatment WD1 was applied in experiment 1 (Table I). WD plants were rewatered at the level of WW plants 6 d after SE. Mean temperature and vapor pressure from tassel emergence to 6 d after SE were 25.4°C and 1.7 kPa during the day and 21.8°C and 1.2 kPa during the night. Plants of line B73 were transferred to a growth chamber with similar climatic conditions during the period of WD in experiment 2.

Photosynthesis

Because photosynthesis measurements are often unreliable in a greenhouse because of rapid changes in light intensity, they were all performed on plants grown in the growth chamber under stable climatic conditions close to average values experienced in the greenhouse in experiment 1. Photosynthesis was followed from 3 d after tassel emergence until 5 d after SE on the ear leaf (leaf 10 or 11) with a CIRAS-2 portable gas-exchange system (PP Systems) with a 2.5-cm² leaf chamber and an air flow rate of 300 cm³ s⁻¹. Light intensities were similar in the growth chamber at leaf level and in the gas-exchange system (300 µmol m⁻² s⁻¹). Air temperature and vapor pressure deficit in the leaf chamber unit of the CIRAS-2 system also mimicked the environmental conditions in the growth chamber (i.e. 25°C and 1.5 kPa). Measurements were carried out for the whole photoperiod (16 h; n = 4 and 17 in WW and WD treatments, respectively).

Plant Sampling and Measurements

Ears were sampled at husk emergence, SE, 5 d after SE just before rewatering, and 15 d after SE. At the first three sampling dates, a subsample of plants was harvested at the end of the day (before irrigation) and another subsample was harvested at the end of the night. Leaf discs were sampled with a punch, frozen in liquid nitrogen, and stored at −60°C. Ears enclosed in husks were detached from the plant and dissected immediately. Fresh weights of peduncle, ear, and silks were measured immediately, before these organs were frozen in liquid nitrogen and stored at −60°C. Ears were sliced into six to seven sections, each of them including five ovary positions along the ear rows (positions 1–5, 6–10, etc., counted from the ear base). Ovaries and cob were separated in liquid nitrogen and weighed. Ovaries were counted to calculate fresh weight per ovary. Organs were then ground in liquid nitrogen. Fifteen days after SE, plants were sampled at the end of the day. Fresh weights of husks, peduncle, ear, and silks were measured immediately, ears were photographed, and ovary or grain number and dimensions were measured with the software ImageJ. Ovaries or grain volumes were estimated as described by Oury et al. (2016).

Metabolites

Metabolites were extracted as described by Hendriks et al. (2003). Suc, Glc, and Fru (Jelitto et al., 1992), malate and fumarate (Nunes-Nesi et al., 2007), total amino acids (Bantan-Polak et al., 2001), and Pro (adapted from Troll and Lindsley [1955]; see also Carillo and Gibon [2011]) were determined in the ethanolic supernatant. Starch (Hendriks et al., 2003) and total protein (Bradford, 1976) contents were determined on the pellet resuspended in 100 mm NaOH. Assays were prepared on 96-well microplates using Starlet pipetting robots (Hamilton), and absorbance was read at 340, 570, or 595 nm in MP96 microplate readers (SAFAS). Ovary metabolite content was measured in all ear sections described above. Sections 1 to 3 counted from the ear base were used as replicates of basal ovaries, while sections 4 and above were used as replicates of apical ovaries. Only soluble sugar content (Suc, Glc, and Fru) was assayed in experiment 2.

Enzyme Activities in Experiment 1

Aliquots of 20 mg fresh weight were extracted as described by Biais et al. (2014), and enzyme activities were measured according to Gibon et al. (2004, 2006, 2009), Sulpice et al. (2007), and Biais et al. (2014). Assays for neutral invertase and acid invertase were performed after desalting (PD Multitrap G-25; GE Healthcare). For the determination of cell wall invertase, pellets obtained after centrifugation of the extracts were washed three times by resuspension in 500 µL of extraction buffer without leupeptin, dithiothreitol, and Triton X-100 followed by centrifugation (15 min at 4,000g). Cell walls were then washed in 300 µL of extraction buffer containing 1 m NaCl, leupeptin, dithiothreitol, and Triton X-100 and vigorously shaken for 5 min in a Qiagen Tissue Lyser II. After centrifugation (15 min at 4,000g), cell wall invertase was measured in aliquots of the supernatant, of which 5 µL was assayed with the same protocol as for acid invertase.

Water Content and Osmotic Potential

Lyophilization was used to estimate the dry matter content of aliquots, and water content was calculated as the difference between fresh weight and dry weight. About 20 mg of frozen ground samples was put in a tube on ice until thawing. After centrifugation (15 min at 13,000 rpm), cell fluid was collected and osmotic potential was measured by psychrometry using a Roebling microosmometer (type 13). The total osmolyte content (mosmol L⁻¹) in each sample was calculated from its osmotic potential.

Transcript Abundance in Experiment 2

Ovaries were sampled as described above at SE and 5 d later in WW and WD1 plants of line B73 grown in a growth chamber under stable climatic conditions close to average values experienced in the greenhouse. Silks were only sampled on day 1 of SE, but not 5 d later, because of the risks of confusing effects of pollination at SE plus 5 d (presence of pollen RNA and silk senescence due to fertilization). Line B73 was chosen rather than the hybrid in order to guarantee a good match with the probes of the chip. Samples were sorted into silks, basal ovaries (positions 1–15 along the row), and apical ovaries (beyond position 15). Three samples per ovary position and four samples for silks (one plant each) were collected at each sampling time and flash frozen less than 30 s after sampling. Expression levels were estimated with the custom oligonucleotide microarray Maïs 45K developed by Biogemma (Roche NimbleGen). This chip displays a set of 45,000 probes designed using pseudomolecules and gene models (FilterGeneSet) provided by the Maize Genome Sequencing Project (www.maizesequence.org, version 4a53), allowing for each probe a corresponding maize gene identifier. One to three probes were used to represent a gene. A total of 2,563 random probes were used for technical and internal hybridization controls. Total RNA was isolated from the tissues with the RNeasy Plant Mini Kit (Qiagen) following the manufacturer’s protocol. After controlling the RNA integrity by capillary electrophoresis (Bioanalyzer; Agilent), complementary DNA was synthesized using the Superscript Double-Stranded cDNA Synthesis Kit (Invitrogen/Fisher Bioblock no. W34277). The complementary DNAs were then purified by a phenol/chloroform step and further labeled with fluorescent Cy3 (One-Color DNA Labeling Kit; Roche NimbleGen). After overnight hybridization (16 h at 42°C), the microarray slides were washed and scanned with an MS200 Roche NimbleGen scanner. The laser power settings were optimized to the autogain mode, which determines the optimal range of acquisition for each scan (laser power was fixed at 100%). Raw hybridization intensities were normalized across all arrays with RMA Express, in which the quantile normalization method was employed (Bolstad et al., 2003).

Data were imported into R (R Development Core Team, 2014). The log₂ of the transcript abundance was calculated for each probe. Differences between treatments of the transcript abundance corresponding to a given probe were considered as significant if (1) the P value was lower than 0.05 in a Student’s t test and (2) it differed by a factor greater than 2 between treatments. Identification of genes was performed on Gramene.org (http://pathway.iplantcollaborative.org/organism-summary?object=MAIZE; accessed May 2015). Invertase gene identification was completed with data from Kakumanu et al. (2012). Gene functions were sorted out by following the ontologies proposed in MapMan with the maize genome release Zm_B73_5b_FGS_cds_2012 (http://mapman.gabipd.org/web/guest/mapmanstore). Two categories were particularly analyzed and presented in Figures 5 and 6, namely genes involved in carbon metabolism (Granot et al., 2013; Yadav et al., 2014) and those involved in expansive growth, either via water transport or via cell wall mechanical properties (Cosgrove, 2005; Wolf et al., 2012; Babu et al., 2013; Xiao et al., 2014).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Ovary volume and abortion frequency as a function of position along the ear.

• Supplemental Figure S2. Spatial distributions of aborted ovaries of plants subjected to WD during flowering time.

• Supplemental Figure S3. Relationship between ovary volume and ovary fresh weight.

• Supplemental Figure S4. Sugar content per unit fresh weight and osmolyte content as a function of developmental stages in cob and peduncle.

• Supplemental Figure S5. Sugar amount per organ as a function of developmental stages.

• Supplemental Figure S6. Relationship between activities or contents measured at the end of the day and at the end of the night.

• Supplemental Figure S7. Metabolite contents and enzyme activities in leaves as a function of treatments.

• Supplemental Table S1. Transcriptome analysis: data for genes involved in carbon metabolism expressed in maize ovaries and silks.

• Supplemental Table S2. Transcriptome analysis: data for genes involved in expansive growth expressed in maize silks.

• Supplemental Table S3. Transcriptome analysis: data for genes involved in expansive growth expressed in maize ovaries.

Glossary

WD

water deficit

WD1

moderate water deficit

SE

silk emergence

WW

well-watered