The type-B response regulators ARR10, ARR12, and ARR18 specify the central cell in Arabidopsis

Abstract

In Arabidopsis thaliana, the female gametophyte consists of two synergid cells, an egg cell, a diploid central cell, and three antipodal cells. CYTOKININ-INDEPENDENT 1 (CKI1), a histidine kinase constitutively activating the cytokinin signaling pathway, specifies the central cell and restricts the egg cell. However, the mechanism regulating CKI1-dependent central cell specification is largely unknown. Here, we showed that the type-B ARABIDOPSIS RESPONSE REGULATORS10, 12, and 18 (ARR10/12/18) localize at the chalazal pole of the female gametophyte. Phenotypic analysis showed that the arr10 12 18 triple mutant is female sterile. We examined the expression patterns of embryo sac marker genes and found that the embryo sac of arr10 12 18 plants had lost central cell identity, a phenotype similar to that of the Arabidopsis cki1 mutant. Genetic analyses demonstrated that ARR10/12/18, CKI1, and ARABIDOPSIS HISTIDINE PHOSPHOTRANSFER PROTEIN2, 3, and 5 (AHP2/3/5) function in a common pathway to regulate female gametophyte development. In addition, constitutively activated ARR10/12/18 in the cki1 embryo sac partially restored the fertility of cki1. Results of transcriptomic analysis supported the conclusion that ARR10/12/18 and CKI1 function together to regulate the identity of the central cell. Our results demonstrated that ARR10/12/18 function downstream of CKI1–AHP2/3/5 as core factors to determine cell fate of the female gametophyte.

ARABIDOPSIS RESPONSE REGULATOR10, 12, and 18 act downstream of CYTOKININ-INDEPENDENT 1 and ARABIDOPSIS HISTIDINE PHOSPHOTRANSFER PROTEIN2, 3, and 5 to regulate cell fate in the Arabidopsis female gametophyte.

IN A NUTSHELL.

Background: The sexual reproduction process of angiosperms requires correct development of the female gametophyte, or embryo sac, which contains seven cells that will develop into the embryo and endosperm after double fertilization. In Arabidopsis, CYTOKININ-INDEPENDENT 1 (CKI1), a histidine kinase that can constitutively activate the cytokinin signaling pathway, plays an essential role in central cell specification, which is also dependent on ARABIDOPSIS HISTIDINE PHOSPHOTRANSFER PROTEIN 2/3/5 (AHP2/3/5).

Question: Although type-B ARABIDOPSIS RESPONSE REGULATORS (ARRs) usually function downstream of the cytokinin signaling pathway as core transcription factors to regulate various biological processes during plant growth, it is not known which ARRs play key roles downstream of the CKI1–AHP2/3/5 module to specify the central cell.

Findings: We identified three type-B ARRs, ARR10, ARR12, and ARR18, which were polarly localized at the chalazal pole of the embryo sac after stage FG4. Genetic analyses revealed that the arr10 12 18 female gamete was never transmitted to the progeny. Expression pattern analyses of embryo sac cell markers revealed that the central cell identity was lost and the egg cell and synergid cell identities extended to the chalazal end in the arr10 12 18 embryo sacs. Genetic analysis showed that CKI1, AHP2/3/5, and ARR10/12/18 function in a common pathway to specify the central cell, which was also supported by transcriptomic results. In addition, constitutively activated ARR10/12/18 partially rescued the central cell defects of cki1 and ahp2/3/5. We demonstrated that the CKI1–AHP2/3/5–ARR10/12/18 signaling module has crucial functions in determining cell fates of the female gametophyte.

Next steps: We do not know the direct downstream components of the CKI1–AHP2/3/5–ARR10/12/18 signaling module during female gametophyte development. Moreover, the mechanism controlling polar localization of CKI1 and ARR10/12/18 in the chalazal pole of the embryo sac is worthy of further investigation.

Introduction

The female gametophyte (embryo sac) of higher plants is generated in the maternal ovule, the place where double fertilization occurs. Most angiosperms, including Arabidopsis (Arabidopsis thaliana), generate a Polygonum-type embryo sac, whose development can be divided into megasporogenesis and megagametogenesis (Ma and Sundaresan, 2010; Drews and Koltunow, 2011; Erbasol Serbes et al., 2018). Megasporogenesis comprises three successive major events: (1) the archesporial cell within the early ovule located in the gynoecium differentiates to form a megaspore mother cell (MMC); (2) the MMC undergoes meiosis and gives rise to four haploid megaspores; and (3) three megaspores close to the micropylar end degenerate and the megaspore close to the chalazal end specializes into the functional megaspore (FM; Christensen et al., 1997; Yang et al., 2010). Megagametogenesis is initiated when the FM undergoes three rounds of mitosis, producing a syncytium containing eight nuclei, along with vacuole formation, rapid cell growth, and nucleus migration. Subsequently, the syncytium undergoes cellularization to form a mature embryo sac (Christensen et al., 1997; Nakajima, 2018).

Megagametogenesis was artificially divided into seven stages (FG1–FG7) based on the number of nuclei, the formation of vacuoles, and cellularization in the embryo sac as follows (Christensen et al., 1997; Drews and Koltunow, 2011). FG1, the FM is established in the nucellus; FG2, the first mitosis of FM is completed and the resulting two nuclei are located at the micropylar end and the chalazal end, respectively; FG3, a central vacuole is formed between the two nuclei; FG4, a second mitosis is completed and the micropylar and chalazal ends have two nuclei each; FG5, a third mitosis results in four nuclei located at each end of the embryo sac; FG6, a nucleus from each end, known as the polar nuclei, moves toward the middle of the syncytium, and the syncytium undergoes cellularization to form an embryo sac with the typical seven-cell and eight-nucleus structure; FG7, the two polar nuclei fuse, and a mature embryo sac is generated. It contains four types of cells: one egg cell, one central cell, two synergid cells, and three antipodal cells (Christensen et al., 1998; Nakajima, 2018; Hater et al., 2020). The synergid cells attract and receive pollen tubes (Higashiyama et al., 2001). The egg cell not only acts as a gamete that fuses with the sperm cell to form the zygote during double fertilization, but also can activate the sperm cell (Sprunck et al., 2012). The central cell functions as another gamete fusing with the sperm cell to form the endosperm, and it also helps to attract the pollen tube (Chen et al., 2007; Li et al., 2015). Therefore, the establishment of correct cell identity in the embryo sac ensures that double fertilization can proceed successfully to produce viable offspring. However, the mechanisms regulating the differentiation of seven cells containing the same genetic material in one syncytium into four types of cells, especially the identity of two female gametes, are largely unknown.

Some factors involved in the fate determination of embryo sac cells have been identified through genetic screening (Tekleyohans et al., 2017). For example, altered expression of auxin biosynthesis or signaling pathway members in the embryo sac resulted in the ectopic expression of egg cell markers (Panoli et al., 2015). In the yucca1 yucca2 (yuc1 2) double mutant, the synergid cells expressed the molecular marker of the egg cell (Panoli et al., 2015), and the ectopic expression of YUC genes in the embryo sac resulted in the expression of the egg cell marker gene at the chalazal end (Pagnussat et al., 2009). The mutant embryo sacs where AUXIN RESPONSE FACTOR (ARF) genes were downregulated using artificial microRNAs (amiRNAs) expressed markers of egg cell identity in the synergid cells (Pagnussat et al., 2009; Liu et al., 2018). In the altered meristem program1 (amp1) mutant, an additional egg cell with full functionality was formed at the expense of a synergid cell (Kong et al., 2015). The plant-specific RWP-RK domain (RKD) transcription factors were thought to be egg cell determinants (Kőszegi et al., 2011; Tedeschi et al., 2017). A recent study showed that the nuclear position in the embryo sac determines egg cell specification and that auxin distribution likely acts as a positional cue (Sun et al., 2021).

Some key factors were also reported to regulate fate determination of the other female gamete, the central cell. AGAMOUS-LIKE80 (AGL80), also known as FEMALE GAMETOPHYTE111 (FEM111), is a MADS-box transcription factor. The agl80 embryo sac had defective central cell development, and the expression of some central cell molecular markers such as DEMETER (DME) and DOWNREGULATED IN DIF1 46 (DD46) was lost (Portereiko et al., 2006). AGL80 can interact with the transcription repressor TOPLESS (TPL), inhibiting the central cell from acquiring synergid cell and antipodal cell identities (Zhang et al., 2020). AGL61, another MADS-box gene, is also required for specifying the central cell (Steffen et al., 2008). MYB64 and MYB119, two predicted R2R3-MYB transcription factors, play a key role to restrict the central cell identity. In the myb64 119 double mutant, the central cell identity extended to both the chalazal and micropylar ends of the embryo sac (Rabiger and Drews, 2013).

In addition, cytokinin signaling mediated by CYTOKININ-INDEPENDENT 1 (CKI1), a homologue of cytokinin receptors ARABIDOPSIS HISTIDINE KINASE 2/3/4 (AHK2/3/4), plays a crucial role in determining the cell fate of the central cell (Pischke et al., 2002; Hejátko et al., 2003; Yuan et al., 2016). The extracellular domain of CKI1 cannot bind cytokinin (Yamada et al., 2001), but the overexpression of CKI1 induces a typical cytokinin response (Kakimoto, 1996). In the cki1-9 mutant, no central cell and antipodal cell identities were detected, whereas the egg cell and synergid cell identities extended to the chalazal end (Yuan et al., 2016).

The cytokinin signaling pathway in Arabidopsis is similar to the two-component signaling pathway in bacteria, which is mediated by a multistep phosphorelay (Hwang and Sheen, 2001; Mira-Rodado, 2019). Key components of this pathway include AHKs, ARABIDOPSIS HISTIDINE PHOSPHOTRANSFER PROTEINS (AHPs), and ARABIDOPSIS RESPONSE REGULATORS (ARRs) (Hwang and Sheen, 2001; Hwang et al., 2012; Kieber and Schaller, 2014). Many cytokinin primary responses in Arabidopsis are mediated by type-B ARRs (Argyros et al., 2008; Ishida et al., 2008; Zubo and Schaller, 2020), a group of MYB transcription factors, through activating the expression of a series of response genes (Hwang et al., 2012; Zubo et al., 2017; Xie et al., 2018). For example, ARR1/12 activate the expression of SHORT HYPOCOTYL 2 (SHY2)/INDOLE-3-ACETIC ACID INDUCIBLE 3 (IAA3) and inhibit the auxin signaling pathway to promote root meristem cell differentiation (Ioio et al., 2008). ARR1/10/12 regulate the size of root meristem (Argyros et al., 2008). A series of recent studies have shown that ARR1/10/12 activate the expression of WUSCHEL (WUS) to regulate formation of the axillary meristem (Wang et al., 2017), regeneration of the shoot (Meng et al., 2017; Zubo et al., 2017; Liu et al., 2020), and maintenance of the apical meristem (Xie et al., 2018).

Cytokinin receptors are abundantly expressed at the chalazal end of the ovule during female gametophyte development, and some ahk2 3 4 ovules cannot generate the embryo sac, suggesting that cytokinin signaling in sporophyte regulates establishment of the FM (Kinoshita-Tsujimura and Kakimoto, 2011; Cheng et al., 2013). As mentioned above, CKI1 regulates the development of embryo sac by specifying cell identities at the chalazal pole and restricting micropylar cell fates (Yuan et al., 2016). In addition, genetic studies indicated that the functions of both AHK2/3/4 and CKI1 in regulating female gametophyte development depend on AHPs (Deng et al., 2010; Cheng et al., 2013). However, ARRs that function downstream of either AHK2/3/4 or CKI1 to specify the female gametophyte and central cell fate are currently largely unknown.

In this study, we identified three type-B ARRs (ARR10/12/18) that are expressed during embryo sac development. The arr10 12 18 triple mutant was sterile because the arr10 12 18 female gamete cannot be transmitted to the offspring. The egg cell and synergid cell identities were extended to the chalazal pole, whereas the central cell and antipodal cell identities were lost in arr10 12 18 embryo sacs. We showed that ARR10/12/18 function as key transcription factors downstream of CKI1–AHP2/3/5 to determine cell fate in the embryo sac.

Results

ARR10/12/18 redundantly regulate female gametophyte development in Arabidopsis

The Arabidopsis genome encodes 11 type-B ARRs (Supplemental Figure S1A). To investigate whether they play a role in reproduction, the expression of all 11 type-B ARRs in seedlings, rosette leaves, inflorescence, and ovaries at the twelfth flower stage (as determined according to Smyth et al. (1990) and Ferrándiz et al. (1999)) was examined by reverse transcription–polymerase chain reaction (RT–PCR; Supplemental Figure S1B). All of these type-B ARRs were expressed in the ovary, and ARR12 and ARR18 had a higher level of expression in the ovary than in other tissues (Supplemental Figure S1B). Phylogenetical analysis revealed that ARR10, ARR12, and ARR18 were closely related members of a sub-branch (Supplemental Figure S1A). In addition, ARR10 was also expressed in the ovary (Supplemental Figure S1B). These data suggested that ARR10/12/18 may redundantly regulate the development of female reproductive organs.

To test whether ARR10/12/18 are involved in reproductive development, corresponding T-DNA insertion mutants, namely arr10-5, arr10-6, arr12-1, arr12-3, and arr18-2, were obtained from ABRC (Figure 1A). RT–PCR expression analysis of ARR10, ARR12, and ARR18 in inflorescences of wild-type (WT), arr10-5, arr10-6, arr12-1, arr12-3, and arr18-2 showed that all single mutants of these genes were null mutants (Supplemental Figure S1C). Phenotypic analysis revealed that the single and double mutants of these genes exhibited normal fertilization and early seed development (Supplemental Figure S2, A–F). However, no triple mutant segregated from arr10-5 12-1 18-2/+, arr10-5 12-1/+ 18-2, and arr10-5/+ 12-1 18-2, suggesting either that the arr10-5 12-1 18-2 gametophyte is defective or the embryo or seedling is nonviable.

Figure 1.

Loss-of-function of ARR10/12/18 results in ovule abortion. A, Schematic diagrams of ARR10, ARR12, and ARR18. Triangles and arrows indicate the site and orientation of T-DNA insertions, respectively. UTR, exon, and intron are represented by grey boxes, black boxes, and black lines, respectively. B–E, Dissected siliques of WT (B), arr10-5 12-1 18-2/+ (C), arr10-5 12-1/+ 18-2 (D), and arr10-5/+ 12-1 18-2 (E). Abortive ovules are indicated by white arrows in (B–D). Bars = 1 mm. F, Statistical analysis of ovule abortion rate in WT and mutant plants. Twenty siliques from WT, arr10-5 12-1 18-2/+, arr10-5 12-1/+ 18-2, and arr10-5/+ 12-1 18-2 plants were observed. Each point represents the abortion percentage in a silique. The data shown are means ± sd (n = 20 siliques for each genotype). Different letters above the bars indicate significant differences using the Tukey–Kramer multiple comparison test (one-way analysis of variance [ANOVA], P < 0.05).

To investigate whether ovule development or embryogenesis was affected, the developing siliques of arr10-5 12-1 18-2/+, arr10-5 12-1/+ 18-2, and arr10-5/+ 12-1 18-2 mutants were then examined. Different from the WT, only 48.7%, 49.6%, and 50.6% of ovules were successfully fertilized and able to develop into seeds in arr10-5 12-1 18-2/+, arr10-5 12-1/+ 18-2, and arr10-5/+ 12-1 18-2 siliques, respectively (Figure 1, B–F). This suggested that the female gametophyte of arr10-5 12-1 18-2 is dysfunctional. Consistently, the results of reciprocal crosses between arr10-5 12-1 18-2/+ mutants and WT plants showed that about half of ovules could not develop into seeds when arr10-5 12-1 18-2/+ was used as the female parent (Supplemental Figure S3, A–E). Moreover, transmission efficiency analysis revealed that the arr10-5 12-1 18-2 female gamete was never transmitted to the progeny. However, the transmission rate of the arr10-5 12-1 18-2 male gamete was normal (Table 1). Finally, complementation analyses performed with genomic sequences of ARR10/12/18 fused with yellow fluorescent protein (YFP) and driven by their native promoters showed that homozygous gARR10-YFP, gARR12-YFP, and gARR18-YFP transgenes completely restored the fertility of arr10-5 12-1 18-2 (Supplemental Figure S3, F–J). This confirmed that ARR10/12/18 were responsible for the observed fertility defects of arr10-5 12-1 18-2/+, arr10-5 12-1/+ 18-2, and arr10-5/+ 12-1 18-2. All these data indicated that ARR10/12/18 redundantly regulate development of the female gametophyte.

Table 1.

Transmission efficiency (TE) of female and male gametes of arr10-5 12-1 18-2

Female	Male	Genotype of progeny (arr10-5/+ 12-1/+)		TEF	TEM	P-value
		arr18-2/+	ARR18
arr10-5 12-1 18-2/+	Col-0	0	332	0	NA	3.52524E−74
Col-0	arr10-5 12-1 18-2/+	150	153	NA	98.03%	0.863165918

TE was calculated as follows: TE = number of progeny with T-DNA insertion/number of WT progeny × 100%. P-values were determined using the Chi-squared test. The expected value for normal gamete transmission is 100%. TE_F, female gamete transmission efficiency; TE_M, male gamete transmission efficiency; NA, no applicable.

ARR10/12/18 are localized at the chalazal end of the embryo sac following the FG4 stage

To investigate when ARR10/12/18 function during embryo sac development, we used confocal laser scanning microscopy (CLSM) to examine expression patterns of these genes in the above-mentioned complementation lines. ARR-YFP signals were detected in the one-nuclear embryo sac (FG1; Figures 2, A, F, and K), and then in the subsequent stages of embryo sac development (FG2–FG7; Figure 2, B–E, G–J, and L–O). Intriguingly, from the four-nuclear embryo sac stage (FG4) to the mature embryo sac stage (FG7), the expression of ARR10/12/18 was detected only in the chalazal pole of the embryo sac (Figure 2, C–E, H–J, and M–O). Consistent with their functions as transcription factors, ARR10/12/18 are primarily located in the nucleus (Figures 2, A–O). The overlapping expression patterns of ARR10/12/18 in the developing embryo sac further support the notion that they regulate embryo sac development.

Figure 2.

Expression patterns of ARR10/12/18 during female gametophyte development. A–O, Expression patterns of proARR10:gARR10-YFP (A–E), proARR12:gARR12-YFP (F–J), and proARR18:gARR18-YFP (K–O) at female gametophyte development stages FG1–FG7. ARR-YFP signals locate in the nucleus of the embryo sac. Merged images of YFP and DIC channels are shown. At least 100 ovules at each stage were observed in more than three independent transgenic lines. Embryo sacs are indicated with dotted white lines. ACN, antipodal cell nucleus; PN, polar nucleus; CCN, central cell nucleus. Bars = 10 μm.

Loss-of-function of ARR10/12/18 results in a collapsed embryo sac

Our genetic data indicated that development of the embryo sac was defective in the arr10-5 12-1 18-2/+ mutant (Figure 1; Supplemental Figure S3). To determine what defects occur in the mutant embryo sac, we examined the development of embryo sac in WT and arr10-5 12-1 18-2/+ plants by CLSM. All ovules of the arr10-5 12-1 18-2/+ mutant at stages FG1–FG5 generated embryo sacs that were undistinguishable from those of WT plants (Figure 3, A–D, F–I, and K), suggesting that initiation of the ovule, formation and meiosis of the MMC, and specification and mitosis of the FM are not affected in the mutant ovules.

Figure 3.

Defective embryo sacs are produced in the arr10-5 12-1 18-2/+ mutant plants. A–E, Female gametophytes of WT plants at stages FG1 (A), FG3 (B), FG4 (C), FG5/6 (D), and FG7 (E) were observed by CLSM. F–J, Female gametophytes of arr10-5 12-1 18-2/+ at stages FG1 (F), FG3 (G), FG4 (H), FG5/6 (I), and FG7 (J) were observed by CLSM. K, Statistical analysis of ovules with normal female gametophyte (FG) at each stage in WT and arr10-5 12-1 18-2/+ plants. Each point represents the rate of normal FG in one pistil. Five pistils were analyzed for each genotype at the examined developmental stage. The data are means ± sd. *** indicates statistically significant differences compared to WT, as determined by two-tailed Student’s t tests (P < 0.001); ns, no significant difference. Each image is a superposition of three to ten optical cross sections. The pale cyan areas indicate the female gametophytes. ACN, antipodal cell nucleus; CCN, central cell nucleus; CN, chalazal nucleus; ECN, egg cell nucleus; MN, micropylar nucleus; SCN, synergid cell nucleus. To ensure that all ovules were at the mature stage FG7, ovules in flowers at flower stage 12 were emasculated for 12 h before observed (E and J). Bars = 10 μm.

To obtain embryo sacs all at the same mature FG7 stage, WT and mutant flowers at the twelfth stage were emasculated 12 h before the ovules were fixed and cleared for observation. Different from the early stages, 51.7% of ovules in the arr10-5 12-1 18-2/+ mutant produced degenerated embryo sacs. These had strong autofluorescence when compared with the WT, and no obvious central vacuole was found in the mutant embryo sacs at the FG7 stage (Figure 3, E, J, and K). Similar to the CLSM results, observation of mutant ovules by differential interference contrast (DIC) microscopy revealed that WT-like embryo sacs were generated in the arr10-5 12-1 18-2/+, arr10-5 12-1/+ 18-2, and arr10-5/+ 12-1 18-2 mutants at the FG1–FG5 stages (Supplemental Figure S4, A–D, F–I, K–N, P–S, and U). About half of the embryo sacs in the arr10-5 12-1 18-2/+, arr10-5 12-1/+ 18-2, and arr10-5/+ 12-1 18-2 mutants were collapsed at the mature stage, and no distinct nucleolus was found in mutant embryo sacs (Supplemental Figure S4, E, J, O, T, and U). Taken together, these data demonstrated that no functional embryo sacs can be successfully formed when the functions of ARR10/12/18 are disrupted.

Embryo sac cell fate determination is impaired in arr10-5 12-1 18-2

In the mature WT embryo sac, four types of cells are specified along the micropylar–chalazal axis, including two synergid cells, one egg cell, one central cell, and three antipodal cells, all of which can be identified by specific molecular markers. Microscopic analysis showed that collapsed embryo sacs were generated in the arr10-5 12-1 18-2/+ ovules, although three rounds of successive mitosis proceeded in the mutant embryo sac (Figure 3J; Supplemental Figure S4, J, O, and T), suggesting that the embryo sac cells may not be appropriately specified.

To test this hypothesis, the arr10-5 12-1 18-2/+ mutants were transformed with embryo sac cell-specific markers driving the expression of nuclear-localized YFP (NLS-YFP), and homozygous transgenic lines were analyzed to determine cell identities in the mutant embryo sacs. In WT plants, the presence of only one egg cell in all examined ovules was indicated by the expression of proDD45:NLS-YFP, an egg cell-specific marker (Steffen et al., 2007; Figure 4, A and Q). In the arr10-5 12-1 18-2/+ mutants, about half of the examined embryo sacs (48.5%, Figure 4Q) had NLS-YFP signals in the egg cell, similar to WT egg cells (Figure 4A). However, in the remaining putative mutant embryo sacs, additional NLS-YFP signals that extended to the chalazal end were detected either in both the central cell and antipodal cells (Figure 4B), or in either one of them (Figure 4, C and D). Similar results were obtained when the expression pattern of EC1.1, a second egg cell marker (Sprunck et al., 2012), was analyzed in the embryo sac of arr10-5 12-1 18-2/+ plants (Supplemental Figure S5, A–E and O). These results demonstrated that the central cell and antipodal cells may possess the egg cell molecular property in defective arr10-5 12-1 18-2 embryo sacs. We further analyzed the expression patterns of proDD31:NLS-YFP, a synergid cell-specific marker (Steffen et al., 2007), in the embryo sac of arr10-5 12-1 18-2/+ plants. Different from WT plants, extended expression of DD31 was also detected in the central cell, but not in the antipodal cells (Figure 4, E, F, and R; Supplemental Figure S5, F–H and P).

Figure 4.

Female gametophytic cell specification is defective in arr10-5 12-1 18-2. A–D, Expression patterns of the egg cell marker proDD45:NLS-YFP in the embryo sac of WT and arr10-5 12-1 18-2/+ mutant plants. In WT ovules, only the egg cell nucleus was labeled (A). The ovules of arr10-5 12-1 18-2/+ plants exhibited expanded expression of proDD45:NLS-YFP in the polar nuclei and antipodal nuclei (B), the polar nuclei (C), and the antipodal cell nuclei (D). E and F, Expression patterns of the synergid cell marker proDD31:NLS-YFP in the embryo sac of WT and arr10-5 12-1 18-2/+. The expression of proDD31:NLS-YFP was only detected in the synergid cell nuclei of WT ovules (E). The marker was also detected in the polar nuclei of arr10-5 12-1 18-2/+ plants (F). G and H, Expression patterns of the central cell marker proDD22:NLS-YFP in the embryo sac of WT and arr10-5 12-1 18-2/+ plants. The expression of proDD22:NLS-YFP was discovered in the central cell nucleus of WT ovules (G), but was absent in the arr10-5 12-1 18-2 embryo sacs (H). I and J, Expression patterns of the antipodal cell marker proDD1:NLS-YFP in the embryo sac of WT and arr10-5 12-1 18-2/+ plants. The expression of proDD1:NLS-YFP was found in the antipodal cell nuclei of WT ovules (I), but was lost in the arr10-5 12-1 18-2 embryo sacs (J). K and L, Expression patterns of the central cell (CC) and synergid cell (SC) double marker proDD22:NLS-mCherry proDD31:NLS-YFP in the embryo sac of WT and arr10-5 12-1 18-2/+ plants. The SC and CC nuclei were labelled in the WT embryo sacs (K). In arr10-5 12-1 18-2 embryo sacs, the CC marker signal was absent and the central cell ectopically expressed the SC signal (L). M and N, Expression patterns of the central cell (CC) and egg cell (EC) double marker proDD22:NLS-mCherry proDD45:NLS-YFP in the embryo sac of WT and arr10-5 12-1 18-2/+ plants. The CC and EC nuclei were labeled in the WT embryo sac (M). In arr10-5 12-1 18-2 embryo sacs, the CC marker signal disappeared and all the chalazal end nuclei ectopically expressed the EC signal (N). O and P, Expression patterns of the synergid cell (SC) and egg cell (EC) double marker proDD31:NLS-mCherry proDD45:NLS-YFP in the embryo sac of WT and arr10-5 12-1 18-2/+ plants. The double marker labeled SC and EC nuclei in the WT embryo sac (O). In arr10-5 12-1 18-2/+ plants, the SC marker signal was ectopically expressed in the central cell, and the EC marker signal was ectopically expressed in the antipodal cells and the central cell (P). Q–W, Statistical analysis of ovules expressing proDD45:NLS-YFP (Q), proDD31:NLS-YFP (R), proDD22:NLS-YFP (S), proDD1:NLS-YFP (T), proDD22:NLS-mCherry proDD31:NLS-YFP (U), proDD22:NLS-mCherry proDD45:NLS-YFP (V), and proDD31:NLS-mCherry proDD45:NLS-YFP (W) in WT and arr10-5 12-1 18-2/+ plants. For each genotype, more than 180 ovules from ≥5 siliques of three transgenic plants were analyzed. The data are means ± sd. Asterisk indicates statistically significant differences compared to WT plants, as determined by two-tailed Student’s t tests (***P < 0.001). Embryo sacs were indicated by dotted white lines in A–J. ACN, antipodal cell nucleus; CCN, central cell nucleus; ECN, egg cell nucleus; PN, polar nucleus; SCN, synergid cell nucleus. Bars = 10 μm.

The ectopic expression of the egg cell- and synergid cell-specific markers in the central cell and antipodal cells suggested that the cells at the antipodal end may have lost their identities in the arr10-5 12-1 18-2 embryo sacs. To test this hypothesis, we further examined cell fates in the arr10-5 12-1 18-2 embryo sacs using the central cell- and antipodal cell-specific markers. The results showed that both proDD22:NLS-YFP (Steffen et al., 2007) and proFIS2:NLS-mCherry (Luo et al., 2000), two central cell markers, were specifically expressed in the central cell of all examined WT ovules (Figure 4, G and S; Supplemental Figure S5, I and Q). Intriguingly, half of examined ovules in arr10-5 12-1 18-2/+ plants lost the expression of proDD22:NLS-YFP and proFIS2:NLS-mCherry in the embryo sac (Figure 4, H and S; Supplemental Figure S5, J, K, and Q), suggesting that no central cell was successfully specified. Similarly, the expression of proDD1:NLS-YFP and proDD13:NLS-mCherry (Steffen et al., 2007), two antipodal cell markers, was not detected in about half of the embryo sacs in the arr10-5 12-1 18-2/+ plants (Figure 4, J and T; Supplemental Figure S5, L–N and R), suggesting that the antipodal cells in the defective arr10-5 12-1 18-2 embryo sacs had lost their identity.

To rule out the possibility that no central cell marker expression in the mutant embryo sac was due to altered viability of the mutant embryo sac at the FG7 stage, dual markers of the synergid cell, egg cell, and central cell were introduced into the WT and arr10-5 12-1 18-2/+ plants to analyze their expression patterns. Consistent with the above results, half of the ovules in arr10-5 12-1 18-2/+ with synergid cell–central cell and egg cell–central cell double markers had lost expression of the central cell marker, but still expressed the synergid cell or egg cell marker in the central cell (Figure 4, L, N, U, and V). The expression of the synergid cell–egg cell double marker in putative arr10-5 12-1 18-2 embryo sacs was extended towards the chalazal pole, a pattern similar to that of each marker alone (Figure 4, P and W). Taken together, the expression of chalazal cell markers was absent in the arr10-5 12-1 18-2 embryo sac, whereas the micropylar cell markers showed expanded expression domains in the arr10-5 12-1 18-2 embryo sac. Our results supported that ARR10/12/18 redundantly specify fates of the central cell and antipodal cells, and restrict fates of the egg cell and synergid cells at the micropylar end.

Loss-of-function of ARR10/12/18 resulted in defective embryo sacs similar to those in the cki1 mutant

To provide more evidence supporting the idea that ARR10/12/18 are responsible for the observed female defects, an independent triple mutant arr10-6 12-3 18-3/+ was created by knocking out ARR18 with the clustered regularly interspaced short palindromic repeats (CRISPR)–CRISPR-associated protein 9 (Cas9) technology in the double mutant arr10-6 12-3 (Supplemental Figure S6A). Sequencing results showed that part of the first exon was deleted and an extra sequence from the first intron was retained in the arr18-3 coding sequence (CDS), leading to premature termination of arr18 (Supplemental Figure S6B). Like arr10-5 12-1 18-2/+, half ovules of arr10-6 12-3 18-3/+ plants without Cas9 protein were sterile (Figure 5, D and H), and no homozygous plants were isolated in their progeny. Because the promoter driving Cas9 expression is proEC1.2, which can effectively function in the zygote to edit the target gene after transformation (Sprunck et al., 2012; Wang et al., 2015), homozygous arr10-6 12-3 18-3 seeds can be generated in arr10-6 12-3 18-3/+ plants containing Cas9 protein. Different from the WT and arr10-5 12-1 18-2/+ plants, homozygous arr10-6 12-3 18-3 plants were completely sterile with short siliques (Supplemental Figure S6E), and no ovules successfully developed into seeds in arr10-6 12-3 18-3 (Figure 5, C and H). Results from reciprocal crosses showed that no seeds were produced when the female arr10-6 12-3 18-3 was pollinated with the WT pollen, demonstrating that arr10-6 12-3 18-3 caused complete dysfunction of the female gametophyte (Supplemental Figure S6, I, J, and M). In addition, in the arr10-6 12-3 18-3 embryo sac, the expression of the egg cell-specific marker proDD45:NLS-YFP extended into the antipodal pole, including the central cell and the antipodal cells (Figure 5, M–P and S), whereas the expression of the central cell-specific marker proDD22:NLS-YFP was not detected (Figure 5, Q, R, and T). Consistently, arr10-6 12-3 18-3 plants generated collapsed and degenerated embryo sacs at the FG7 stage that were similar to those of the arr10-5 12-1 18-2 mutants (Figure 5, J and L). Taken together, these data confirmed that ARR10/12/18 regulate development of the embryo sac.

Figure 5.

arr10 12 18 mutants exhibit reproductive defects similar to cki1 mutants. A–G, Dissected siliques of WT (A), arr10-5 12-1 18-2/+ (B), arr10-6 12-3 18-3 (C), arr10-6 12-3 18-3/+ (D), cki1-9/+ (E), cki1-10 (F), and cki1-10/+ (G) plants. White arrow heads indicate aborted ovules. Bars = 1 mm. H, Statistical analysis of ovule abortion rates in WT and mutant plants. More than 460 seeds from 10 siliques were analyzed for each genotype. Each point represents the ovule abortion rate of a silique. The data are means ± sd. Different letters above the bars indicate significant differences using Tukey–Kramer multiple comparison tests (one-way ANOVA, P < 0.05). I–K, Embryo sacs of WT (I), arr10-6 12-3 18-3 (J), and cki1-10 (K) plants at the FG7 stage observed by CLSM. The pale cyan areas indicate the female gametophytes. Bars = 10 μm. L, Statistical analysis of DFG and NFG in WT, arr10-6 12-3 18-3, and cki1-10 plants. At least 220 ovules from six pistils were analyzed for each genotype. The data are means ± sd. *** indicates statistically significant differences compared to WT plants (two-tailed Student’s t tests, P < 0.001). M–P, Expression patterns of the egg cell marker proDD45:NLS-YFP in WT (M) and arr10-6 12-3 18-3 mutant plants (N–P). All ovules of the mutant exhibited expanded expression of proDD45:NLS-YFP in the central cell, and/or antipodal cells (N–P). Embryo sacs are indicated by dotted white lines. Bars = 10 μm. Q and R, Expression patterns of the central cell marker proDD22:NLS-YFP in WT (Q) and arr10-6 12-3 18-3 mutant plants (R). The expression of proDD22:NLS-YFP was lost in all arr10-6 12-3 18-3 embryo sacs (R). Embryo sacs are indicated with dotted white lines. Bars = 10 μm. S, Statistical analysis of egg cell (S) and central cell (T) marker expression in WT and arr10-6 12-3 18-3 plants. At least 208 ovules from >5 pistils were analyzed. The data are means ± sd. *** indicates statistically significant differences compared to WT plants (two-tailed Student’s t tests, P < 0.001). ACN, antipodal cell nucleus; CCN, central cell nucleus; ECN, egg cell nucleus; SCN, synergid cell nucleus.

Although CKI1 does not bind cytokinins (Yamada et al., 2001), it can induce typical cytokinin responses (Kakimoto, 1996). The loss of CKI1 function led to defective embryo sacs that were very similar to those of the arr10 12 18 mutants (Pischke et al., 2002; Hejátko et al., 2003; Deng et al., 2010; Rabiger and Drews, 2013; Yuan et al., 2016). Moreover, type-B ARRs usually function as transcription factors downstream of the cytokinin pathway (Zubo and Schaller, 2020). We hypothesized that CKI1 functions together with ARR10/12/18 to regulate embryo sac development. Therefore, we created a cki1-10 mutant using the CRISPR–Cas9 technology as described in “Materials and methods” section (Supplemental Figure S6A). Sequencing results showed that the cki1 CDS had a deletion between the first and second exons in the cki1-10 mutant and this resulted in the loss of eight amino acid residues, leading to disruption of the first transmembrane domain of cki1-10 protein (Supplemental Figure S6B). The homozygous cki1-10 mutant was sterile and produced defective embryo sacs similar to those of the cki1-9 and arr10 12 18 mutants (Figure 5, F, H, K, and L). Half of the ovules of cki1-10/+ plants lacking the Cas9 protein were sterile (Figure 5, G and H), and no homozygous cki1-10 mutant plants could be isolated in their progeny. Consistent with the previous report (Yuan et al., 2016), ectopic or absent expression of marker genes was detected in cki1-9 embryo sacs (Supplemental Figure S7). Similar to the arr10-5 12-1 18-2 mutants (Figure 4, A–D; Supplemental Figure S5, A–E), the expression of the egg cell-specific marker proDD45:NLS-YFP in the cki1-9 mutant extended into the antipodal pole, including the central cell and antipodal cells (Supplemental Figure S7, A–E and O). The expression of the synergid cell marker proDD31:NLS-YFP in cki1-9 mutants also extended into the central cell, as in the arr10-5 12-1 18-2 embryo sacs (Supplemental Figure S7, F–H and P). Similarly, no expression of proDD22:NLS-YFP, the central cell marker, and proDD1:NLS-YFP, the antipodal cell marker, was detected in the putative arr10-5 12-1 18-2 and cki1-9 embryo sacs (Figure 4, G–J, S and T; Supplemental Figures S5, I–N, Q and R and S7, I–N, Q and R). All these results suggested that ARR10/12/18 function together with CKI1 to ensure the correct cell fates in the embryo sac.

ARR10/12/18 function with CKI1 and AHP2/3/5 to regulate embryo sac development

AHP2/3/5 function downstream of CKI1 to regulate embryo sac development (Deng et al., 2010; Yuan et al., 2016; Liu et al., 2017). To investigate the genetic relationship between ARR10/12/18 and AHP2/3/5, we first created the ahp2-2 3 5-2/+ mutant as previously reported (Liu et al., 2017). About half of the ovules in ahp2-2 3 5-2/+ were abortive, a result similar to that seen in the arr10-5 12-1 18-2/+ and cki1-9/+ mutants (Figure 6, A–D and G). Further observations using CLSM revealed that the putative ahp2-2 3 5-2 embryo sacs at the FG7 stage were collapsed and degenerated, as in the arr10-5 12-1 18-2 and cik1-9 embryo sacs (Figure 6, H–J). These results suggested that ARR10/12/18 may function with CKI1 and AHP2/3/5 during embryo sac development.

Figure 6.

Phenotypic analysis of higher order mutants arr10-5 12-1 18-2/+ cki1-9/+ and arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+. A–F, Dissected siliques of WT (A), arr10-5 12-1 18-2/+ (B), cki1-9/+ (C), ahp2-2 3 5-2/+ (D), arr10-5 12-1 18-2/+ cki1-9/+ (E), and arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+ plants (F). White arrow heads indicate sterile ovules. Bars = 1 mm. G, Statistical analysis of aborted ovules in indicated genotypes. Each point represents the abortion rate of a silique (10 siliques for each genotype). The data are means ± sd. Different letters above the bars indicate significant differences using a Tukey–Kramer multiple comparison test (one-way ANOVA, P < 0.05). H and I, Mature ovules of all above mutants produced NFG (H) or DFG (I). The DFGs showed embryo sac defect similar to that of arr10-5 12-1 18-2 when observed by CLSM. The pale cyan areas indicate the female gametophytes. Bars = 10 μm. J, Statistical analysis of NFG and DFG in WT, arr10-5 12-1 18-2/+, cki1-9/+, ahp2-2 3 5-2/+, arr10-5 12-1 18-2/+ cki1-9/+, and arr10-5/+ 12-1 18-2 ahp2-2 3 5-2/+. Percentages of NFG and DFG in six pistils for each genotype were analyzed. The data are means ± sd. Different letters above the bars indicate significant differences using a Tukey–Kramer multiple comparison test (one-way ANOVA, P < 0.05). K–P, Expression patterns of proDD22:NLS-YFP (K–M) and proDD45:NLS-YFP proDD31:NLS-mCherry (N–P) in ovules of WT (K and N), arr10-5 12-1 18-2/+ cki1-9/+ (L and O), and arr10-5/+ 12-1 18-2 ahp2-2 3 5-2/+ plants (M and P). Embryo sacs are outlined with autofluorescence that was excited with a 561-nm laser line and collected at 570 nm to 620 nm. Bars = 10 μm. Q and R, Statistical analysis of proDD22:NLS-YFP (Q) and proDD45:NLS-YFP proDD31:NLS-mCherry (R) signals in ovules of WT, arr10-5 12-1 18-2/+ cki1-9/+ (arr cki1), and arr10-5/+ 12-1 18-2 ahp2-2 3 5-2/+ (arr ahp) plants. Each point represents the rate of ovule expressing proDD22:NLS-YFP (Q) or proDD45:NLS-YFP proDD31:NLS-mCherry (R) in a silique (six siliques for each genotype). The data are means ± sd. Asterisk indicates statistically significant differences when compared with WT plants (two-tailed Student’s t tests, ***P < 0.001). ACN, antipodal cell nucleus; CCN, central cell nucleus; ECN, egg cell nucleus; PN, polar nucleus; SCN, synergid cell nucleus.

We then generated arr10-5 12-1 18-2/+ cki1-9/+ and arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+ higher order mutants by crossing to examine whether ARR10/12/18 genetically function together with CKI1 and AHP2/3/5 to control embryo sac cell fates. Because about half of ovules in arr10-5 12-1 18-2/+ and cki1-9/+ produced similarly collapsed embryo sacs and were uncapable to develop into seeds, we expected the arr10-5 12-1 18-2/+ cki1-9/+ mutant would produce about 75% of ovules exhibiting similarly defective embryo sacs if ARR10/12/18 and CKI1 work in a common genetic pathway. As expected, the seed abortion rate of 10 arr10-5 12-1 18-2/+ cki1-9/+ siliques was 74.8% (Figure 6, E and G). Similarly, the seed abortion rate of 10 arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+ siliques was 73.6% (Figure 6, F and G). Moreover, like arr10-5 12-1 18-2/+, ahp2-2 3 5-2/+, and cki1-9/+, arr10-5 12-1 18-2/+ cki1-9/+ and arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+ produced two types of embryo sacs at the FG7 stage when observed by CLSM: normal female gametophytes similar to WT embryo sacs and collapsed, degenerated female gametophytes (DFGs; Figure 6, H–J). Consistent with the seed setting results, about half of the mature embryo sacs in arr10-5 12-1 18-2/+ (52%), cki1-9/+ (51.1%), and ahp2-2 3 5-2/+ (50.8%) were DFGs, whereas about 75% of mature embryo sacs in arr10-5 12-1 18-2/+ cki1-9/+ (75.9%) and arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+ (75.4%) were DFGs (Figure 6, H–J). These data further suggested that ARR10/12/18 function with CKI1 and AHP2/3/5 during embryo sac development.

To confirm this result, the central cell marker proDD22:NLS-YFP and the dual marker of synergid cell–egg cell (proDD31:NLS-mCherry, proDD45:NLS-YFP) were introduced into WT, arr10-5 12-1 18-2/+ cki1-9/+, and arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+. As shown above, about half of embryo sacs in the arr10-5 12-1 18-2/+ mutant expressed the proDD22:NLS-YFP marker (Figure 4, G, H, and S). However, only 20.2% of ovules in arr10-5 12-1 18-2/+ cki1-9/+ and 21.2% of ovules in arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+ expressed the central cell marker proDD22:NLS-YFP (Figure 6, K–M and Q). Similarly, there were 50.8% of ovules in arr10-5 12-1 18-2/+ that showed normal expression of the synergid cell–egg cell dual marker (Figure 4, O and W), whereas normal expression of the dual marker was detected in 18.9% of ovules in arr10-5 12-1 18-2/+ cki1-9/+ and 24.3% of ovules in arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+ (Figure 6, N–P and R). Taken together, these data demonstrated that CKI1, AHP2/3/5, and ARR10/12/18 function as crucial signaling components to determine cell fates in the embryo sac.

Activated ARR10/12/18 are sufficient for the egg cell and synergid cells to express the molecular marker of the central cell

CKI1 is required for the specification of the central cell and antipodal cells and ectopic CKI1 expression can confer central cell fate to the micropylar cells (Yuan et al., 2016; Figure 7, A–C). On the other hand, our results showed that ARR10/12/18 are involved in the CKI1-mediated pathway that specifies the central cell (Figures 5 and 6). We therefore proposed that the ectopic expression of activated ARR10/12/18 in the embryo sac was able to confer central cell fate to the micropylar cells. To test this hypothesis, constitutively activated ARR10/12/18 lacking the signal receiver domains (ΔDDK; Sakai et al., 2000, 2001) were expressed using an ANTIKEVORKIAN (AKV) promoter, which functions throughout female gametophyte development (Rotman et al., 2005; Yuan et al., 2016), in the embryo sac harboring the central cell marker proDD22:NLS-mCherry. When ARR10ΔDDK, ARR12ΔDDK, or ARR18ΔDDK were expressed, some embryo sacs (6.8%, 7.1%, and 11.2%) exhibited proDD22:NLS-mCherry signals in the micropylar cells (Figure 7, D–L; Supplemental Figure S8). These data indicated that activated ARR10/12/18 are sufficient for expressing the central cell molecular marker in the egg cell and synergid cells, possibly conferring central cell identity to the egg cell and synergid cells.

Figure 7.

Ectopic expression of activated ARR10/12/18 confers the central cell marker on micropylar cells. A–L, Ectopic expression of CKI1 (A–C), ARR10ΔDDK (D–F), ARR12ΔDDK (G–I), and ARR18ΔDDK (J–L) by an AKV promoter results in central cell marker expression at the micropylar end of the embryo sac. A, D, G, and J, Expression of gCKI1-YFP (A), ARR10ΔDDK-YFP (D), ARR12ΔDDK-YFP (G), and ARR18ΔDDK-YFP (J) driven by an AKV promoter in the WT embryo sac. B, E, H, and K, Co-expression of the central cell marker proDD22:NLS-mCherry with gCKI1-YFP (B), ARR10ΔDDK-YFP (E), ARR12ΔDDK-YFP (H), and ARR18ΔDDK -YFP (K) in the WT embryo sac. C, F, I, and L, Merged images of YFP and mCherry. A homozygous proDD22:NLS-mCherry line was transformed with the indicated constructs, and the T3 transgenic plants were observed. CCN, central cell nucleus; ECN, egg cell nucleus; SCN, synergid cell nucleus. Bars = 10 μm.

Activated ARR10/12/18 partially rescued the defects of cki1-9 and ahp2-2 3 5-2

To explore whether ARR10/12/18 function downstream of CKI1 to regulate cell fates in the embryo sac, we created and analyzed transgenic plants expressing activated ARR10/12/18 using a native CKI1 promoter (proCKI1) in cki1-9. The seed setting rate of cki1-9/+ was about 50% (Figure 8, B and O), and the seed setting rate of homozygous cki1-10 plants was 0% (Figure 8, C and O), suggesting that no homozygous cki1 seeds could be obtained due to embryo sac lethality. Surprisingly, the homozygous cki1-9 plants harboring homozygous activated ARR10/12/18 driven by proCKI1 showed a seed setting rate of 1.7%, 2.6%, and 2.7% (Figure 8, D–F and O), respectively. These data demonstrated that female gametes containing the cki1-9 allele in these transgenic lines can be transmitted to the progeny. In other words, some defective cki1-9 embryo sacs were rescued by activated ARR10/12/18.

Figure 8.

Activated ARR10/12/18 partially rescue the fertility of cki1. A–F, Developing seeds in siliques of the WT (A), cki1-9/+ (B), cki1-10 (C) cki1-9 proCKI1:ARR10ΔDDK-YFP (D), cki1-9 proCKI1:ARR12ΔDDK-YFP (E), and cki1-9 proCKI1:ARR18ΔDDK-YFP (F). A null cki1-10 mutant is completely sterile (C). Expression of activated ARR10 (D), ARR12 (E), and ARR18 (F) in a cki1-9 null mutant leads to fertile seeds (white asterisks). Bars = 1 mm. G–J, Embryo sacs of the cki1-9 null mutant expressing activated ARR10 (H), ARR12 (I), and ARR18 (J) produce a central cell. Null cki1-10 mutants cannot express the central cell marker proDD22:NLS-YFP (G). Embryo sacs are indicated with dotted white lines. Bars = 10 μm. K–N, DFG are generated in all mature ovules of null mutant cki1-10 (K). A few NFGs are generated in mature ovules of cki1-9 proCKI1:ARR10ΔDDK-YFP (L), cki1-9 proCKI1:ARR12ΔDDK-YFP (M), and cki1-9 proCKI1:ARR18ΔDDK-YFP (N) when observed by CLSM. The pale cyan areas indicate the female gametophytes. Bars = 10 μm. O, Statistical analysis of the seed setting rate of WT, cki1-9/+, cki1-10, cki1-9 proCKI1:ARR10ΔDDK-YFP (cki1 ARR10ΔDDK), cki1-9 proCKI1:ARR12ΔDDK-YFP (cki1 ARR12ΔDDK), and cki1-9 proCKI1:ARR18ΔDDK-YFP (cki1 ARR18ΔDDK) plants. Each point represents the seed setting rate of a silique. Ten siliques were analyzed for WT and cki1-9/+. For cki1-10, cki1 ARR10ΔDDK, cki1 ARR12ΔDDK, and cki1 ARR18ΔDDK, 26 siliques were analyzed. The data are means ± sd. Asterisk indicates a statistically significant difference when compared to cki1-10 (two-tailed Student’s t tests, ***P < 0.001). P and Q, Statistical analysis of ovules with a central cell signal (P) and a normal female gametophyte (Q). Each point represents the proportion of ovules with a central cell signal (P) or a normal female gametophyte (Q) in a pistil. Six pistils of each genotype were analyzed. The data are means ± sd. Double and triple asterisks represent statistically significant differences when compared to cki1-10 (two-tailed Student’s t tests, **P < 0.01, and ***P < 0.001). ACN, antipodal cell nucleus; CCN, central cell nucleus; ECN, egg cell nucleus; SCN, synergid cell nucleus.

The above results also suggested that the normal central cell identity of the cki1-9 mutant was successfully restored by the activated ARR10/12/18, a result supported by the following pieces of evidence. No central cell identity in the cki1-10 embryo sacs was indicated using the central cell marker proDD22:NLS-YFP (Figure 8, G and P). However, in homozygous cki1-9 plants rescued by ARR10ΔDDK, ARR12ΔDDK, and ARR18ΔDDK, we found that 6.2%, 5.8%, and 8.3%, respectively, of embryo sacs exhibited central cell identity (Figure 8, H–J and P). In addition, although most of the embryo sacs in homozygous cki1-9 plants rescued by ARR10ΔDDK, ARR12ΔDDK, and ARR18ΔDDK were collapsed and degenerated at the FG7 stage, similar to those in arr10-6 12-3 18-3 (Figure 5, J and L) and cki1-10 plants (Figure 5, K and L), 5.8%, 5.6%, and 7.9%, respectively, of embryo sacs showing WT-like structures (Figure 8, L–N and Q). These results were consistent with the observed seed setting rate in the rescued homozygous cki1-9 plants (Figure 8, D–F and O).

Moreover, we also introduced activated ARR10/12/18 driven by an ARR18 promoter into cki1-9/+ plants and analyzed their phenotypes. Similarly, some normal seeds (2.6%, 2.5%, and 2.7%) were also produced in homozygous cki1-9 mutants expressing activated ARR10/12/18 (Supplemental Figure S9, A–F and O). Accordingly, microscopy observations revealed that central cell identity (6.2%, 5.8%, and 7.8%) and normal embryo sac structures (6.6%, 6.5%, and 9.0%) were rescued in a few cki1-9 ovules expressing activated ARR10/12/18 (Supplemental Figure S9, G–N and P–Q).

We used a similar strategy to examine whether ARR10/12/18 function downstream of AHP2/3/5 by expressing activated ARR18 in ahp2-2 3 5-2/+. The results showed that normal embryo sacs were generated in 9.6% of ovules of homozygous ahp2-2 3 5-2 mutants, and the seed setting rate was 7.5% in homozygous ahp2-2 3 5-2 (Supplemental Figure S10). Taken together, our data showed that the activated ARR10/12/18 can partially rescue the embryo sac defects of cki1-9 and ahp2-2 3 5-2, supporting the conclusion that ARR10/12/18 all function downstream of the CKI1–AHP2/3/5 signaling module to specify the central cell during embryo sac development.

Identification of differentially expressed genes in arr10-6 12-3 18-3 and cki1-10 embryo sacs

All ovules generated by the homozygous cki1-10 and arr10-6 12-3 18-3 mutants showed the same defective central cell specification, providing an excellent opportunity to identify genes coregulated by CKI1 and ARR10/12/18. Therefore, ovaries at the twelfth flower stage having embryo sacs at the FG4–FG7 stages of arr10-6 12-3 18-3 and cki1-10 plants were harvested for RNA-sequencing (RNA-seq) analyses to identify the differentially expressed genes (DEGs) in embryo sacs of both mutants by comparison with the WT (Supplemental Figure S11). DEGs were defined by a two-fold difference in expression with a q ≤0.05 as a cutoff using the DESeq2 software (Love et al., 2014).

When compared with the WT, 474 DEGs in arr10-6 12-3 18-3 ovaries and 401 DEGs in cki1-10 ovaries were identified (Figure 9, A–C; Supplemental Data Set 1, A and B). Most DEGs represented downregulated genes in arr10-6 12-3 18-3 (415/474) and cki1-10 ovaries (353/401) (Figure 9, A and B; Supplemental Data Set 1, A and B), consistent with the functions of type-B ARRs as transcriptional activators (Sakai et al., 2000; Xie et al., 2018). In addition, 76.8% (364/474) of DEGs in arr10-6 12-3 18-3 ovaries and 90.8% (364/401) of DEGs in cki-10 ovaries were overlapped, with similar expression trends (Figure 9, C and D). Among the overlapped DEGs, 92 genes were gametophytically expressed (Yu et al., 2005), and 49 genes were identified with confirmed expression in the female gametophyte (Steffen et al., 2007). These results confirmed that ARR10/12/18 and CKI1 function in the same pathway to regulate embryo sac development.

Figure 9.

RNA-seq analysis of gene expression in ovules of the WT, arr10-6 12-3 18-3, and cki1-10. A and B, Volcano plot of the DEGs between WT and arr10-6 12-3 18-3 (A) or cki1-10 (B) plants. Red dots (right sides) represent the upregulated genes and dots on the left side represent the downregulated genes in ovaries of arr10-6 12-3 18-3 (A) or cki1-10 (B) plants using a fold change ≥2 or −2 or less, q ≤0.05. C, Venn diagram depicting the DEGs between arr10-6 12-3 18-3 and cki1-10 plants. A total of 364 genes were differentially expressed in both mutants when compared with the WT. D, The heat map represents relative expression levels of 364 overlapped DEGs in the ovary of the WT, arr10-6 12-3 18-3, and cki1-10 plants. Three independent RNA samples for each genotype were used for RNA-seq analyses. E, GO descriptor enrichment of the overlapped DEGs by the online tool DAVID. n, gene number in each GO term. P ≤0.05.

Gene ontology (GO) analysis of the 364 overlapped DEGs using the online tool Database for Annotation, Visualization, and Integrated Discovery (DAVID; Huang et al., 2009) found 10 functional annotation terms having P ≤0.001 (Figure 9E; Supplemental Data Set 1C). Surprisingly, the majority of the overlapped DEGs (191/364) belonged to extracellular components (GO: 0005576) that may be involved in signal transduction. When compared with the identified peptides involved in reproduction in a previous study (Huang et al., 2015), a total of 150 cysteine-rich peptide and 16 noncysteine-rich peptide genes were represented. In the biological process (BP) category, 60 genes may regulate the stress response (GO: 0050832, 0031640), and 15 genes may be involved in proteolysis (GO: 0006508). According to the molecular function annotation, five enriched GO terms were identified, including enzyme inhibitor activity (GO: 0004857), hydrolase activity (GO: 0004553), pectinesterase inhibitor activity (GO: 0046910), lyase activity (GO: 0016829), and carboxylic ester hydrolase activity (GO:0052689; Supplemental Data Set 1C). These data strongly suggested that the CKI1–ARR10/12/18 pathway plays a key role in processing a variety of peptide signaling molecules that are involved in specification of the embryo sac cells and the functions of these cells during subsequent fertilization.

In order to validate the RNA-seq data, RT–qPCR analyses of 13 selected DEGs were performed, including seven LOW-MOLECULAR WEIGHT CYSTEINE-RICH (LCR) family genes (LCR19, LCR44, LCR47, LCR50, LCR59, LCR63, and LCR64), one defensin-like (DEFL) family gene (AT5G42232), two plant invertase/pectin methylesterase inhibitor superfamily genes (AT1G50325 and AT1G09370), two pectin lyase genes (AT2G43860 and AT5G27530), and one carbohydrate-binding X8 domain gene (AT4G09464). All these genes were significantly downregulated in the arr10-6 12-3 18-3 and cki1-10 mutants compared with the WT (Supplemental Figure S12), a result consistent with our RNA-seq results (Supplemental Data Set 1, A and B). Furthermore, the expression patterns of eight genes (LCR44, LCR47, LCR50, LCR63, AT5G42232, AT1G50325, AT2G43860, and AT4G09464) in the embryo sac at the FG7 stage were investigated with NLS-YFP driven by their promoters. The results showed that all eight genes were expressed in the central cell (Figure 10, A, C, E, G, I, K, M, O, and Q), and three of them, LCR44, AT5G42232, and AT1G50325, were also expressed in the antipodal cells (Figure 10, A, I, and K). Consistent with our RNA-seq and RT–qPCR results, no expression of all eight genes was detected in the putative arr10-5 12-1 18-2 embryo sacs (Figure 10, B, D, F, H, J, L, N, P, and Q). In short, by comparing transcriptomes of the unique arr10-6 12-3 18-3 and cki1-10 mutants with the WT, a total of 364 DEGs were identified, and these were further confirmed by RT–qPCR and promoter analysis. Our results also suggest that the CKI–ARR10/12/18 signaling module is required for the activation of a variety of genes possibly involved in female gametophyte development and functions.

Figure 10.

Expression patterns of DEGs in mature embryo sac of the WT and arr10-5 12-1 18-2/+. A–P, Selected DEGs exhibited expression in the central cell of WT embryo sacs (A, C, E, G, I, K, M, and O). LCR44 (A), AT5G42232 (I), and AT1G50325 (K) were also expressed in the antipodal cells of the WT embryo sacs. No expression of these genes was detected in putative arr10-5 12-1 18-2 embryo sacs (B, D, F, H, J, L, N, and P). Promoters of the selected genes were used to express NLS-YFP. Embryo sacs are indicated by dotted white lines. ACN, antipodal cell nucleus; CCN, central cell nucleus. Bars = 10 μm. Q, Statistical analysis of embryo sacs with YFP signal in the WT and arr10-5 12-1 18-2/+ plants. Each point represents the proportion of embryo sac at the FG7 stage with YFP signal in a pistil. Six pistils were analyzed for each genotype. The data are means ± sd. Asterisk indicates a statistically significant difference when compared to WT plants (two-tailed Student’s t tests, ***P < 0.001).

Discussion

ARR10/12/18 are essential signaling components downstream of CKI1 and AHP2/3/5 that redundantly determine embryo sac cell fates in Arabidopsis

In plants, the cytokinin signal is transmitted through a phosphorelay to activate type-B ARRs (Hwang et al., 2012; Kieber and Schaller, 2014). The type-B ARRs then act as transcription factors to regulate the expression of the downstream target genes, and ultimately control plant growth and development (Hwang et al., 2012; Zubo et al., 2017; Xie et al., 2018). Genetic analyses showed that type-B ARRs, especially ARR1/10/12, redundantly function in many developmental processes, such as root development and apical meristem maintenance (Mason et al., 2005; Argyros et al., 2008; Ishida et al., 2008; Xie et al., 2018). The arr1 10 12 mutants exhibited seedling defects very similar to those seen in the ahk2 3 4 mutants (Ishida et al., 2008). Therefore, ARR1/10/12 were considered to be the core transcription factors mediating the cytokinin signaling pathway (Mason et al., 2005; Ishida et al., 2008; Zubo and Schaller, 2020). However, the functions of ARRs in reproductive development were rarely reported.

We found that ARR10/12/18 are closely related (Supplemental Figure S1A), and are specifically expressed during embryo sac development (Figure 2), supporting the notion that they play a role in regulating the development of female gametophytes. Consistently, results of microscopy studies, genetic analyses, and complementation experiments demonstrated that ARR10/12/18 redundantly control embryo sac development (Figures 1 and 3 and Table 1; Supplemental Figures S2–S4). Further expression pattern analyses of embryo sac marker genes in arr10 12 18 mutants revealed that ARR10/12/18 specify the central cell and antipodal cells, and also restrict the micropylar cell fates in the embryo sac (Figure 4; Supplemental Figure S5), a pattern similar to that of CKI1 (Yuan et al., 2016; Figure 5; Supplemental Figures S6 and S7).

Cytokinin can regulate the initiation of ovule primordium in plants, and only a few ovule primordia can be formed in ahk2 3 4 (Kinoshita-Tsujimura and Kakimoto, 2011; Cheng et al., 2013). Although the ahk2 3 4 mutant showed severe abortion or even sterility, and the chalaza-localized sporophytic cytokinin signal played an important role in specifying the FM (Cheng et al., 2013), genetic analyses of the ahk2/+ 3 4 and ahk2 3/+ 4 mutants showed that the ahk2 3 4 female gametophyte was normal, demonstrating that AHK2/3/4 in the female gametophyte likely do not regulate its development (Nishimura et al., 2004; Deng et al., 2010; Kinoshita-Tsujimura and Kakimoto, 2011; Cheng et al., 2013). Taken together, this indicates that, although CKI1 and AHK2/3/4 are closely related histidine kinases involved in cytokinin signaling, they play distinct roles in regulating female gametophyte development. This suggests that CKI1 employs different AHPs and ARRs as downstream signaling components.

It was reported that the functions of both AHK2/3/4 and CKI1 depended on AHPs. The ahp1 2 3 4 5 mutant exhibited defective specification of FM and embryo sac cells (Deng et al., 2010; Cheng et al., 2013). The ahp2 3 5 mutant exhibited altered expression of embryo sac marker genes, as did cki1, indicating that AHP2/3/5 are sufficient to transfer the CKI1 signaling in determining cell fates of the embryo sac (Liu et al., 2017). However, defective specification of the FM similar to ahk2 3 4 was not found in ahp2 3 5 (Liu et al., 2017). These data indicated that AHPs possess differentiated functions, and that different combinations of AHPs regulate distinct biological processes of female gametophyte development. Similar scenarios may exist for ARRs in regulation of female gametophyte initiation and development. It was reported that ARR1/2/10/12 acted downstream of both AHK2/3/4 and CKI1 to regulate the establishment of FM and the development of the female gametophyte, respectively (Cheng et al., 2013). However, ahk2 3 4 and cki1 plants exhibited a more severe abortion rate than did arr1 2 10 12, suggesting that other type-B ARRs are also involved in AHK2/3/4- and CKI1-regulated female gametophyte development.

This study demonstrated that ARR10/12/18 are the key signaling components downstream of CKI1 in determining the fates of embryo sac cells. Several lines of evidence supported this conclusion. First, the arr10-6 12-3 18-3 and cki1-10 mutants created using CRISPR–Cas9 technology displayed similar defects, such as complete sterility, collapsed embryo sacs, and altered expression of marker genes (Figure 5; Supplemental Figure S6). Second, the arr10-5 12-1 18-2/+ cki1-9/+ and arr10-5 12-1 18-2/+ ahp2-2 3 5-2/+ mutants generated about 75% of defective female gametophytes, a result expected when arr10-5 12-1 18-2/+ mutants were crossed with cki1-9/+ and ahp2-2 3 5-2/+ mutants (Figure 6). Third, similar to the ectopic expression of CKI1, the expression of activated ARR10/12/18 using an AKV promoter in the embryo sac conferred central cell identity to the micropylar cells (Figure 7; Supplemental Figure S8). Last but not the least, the expression of activated ARR10/12/18 using a CKI1 or an ARR18 promoter in cki1 embryo sacs resulted in homozygous cki1 seeds (Figure 8; Supplemental Figure S9). Taken together, this study identified ARR10/12/18 as crucial members that function directly downstream of the CKI1–AHP2/3/5 signaling module to regulate female gametophyte development.

Polarity establishment and cell fate determination in the embryo sac

During megagametogenesis, the FM undergoes three successive nuclear divisions to generate a syncytium with eight nuclei, which then forms the mature female gametophyte following cellularization (Christensen et al., 1997; Yadegari and Drews, 2004; Ma and Sundaresan, 2010; Nakajima, 2018). In this process, the nuclei are arranged along the micropylar–chalazal axis with obvious polarity. The position of nucleus in the embryo sac affects cell fate during maturation (Sprunck and Gross-Hardt, 2011; Skinner and Sundaresan, 2018; Sun et al., 2021). For example, loss-of-function of maize (Zea mays) INDETERMINATE GAMETOPHYTE1 (IG1) led to the production of additional nuclei within the embryo sac, which differentiated into extra egg cells, synergid cells, and central cells according to their positions in the embryo sac (Guo et al., 2004). In the Arabidopsis mutants eostre and amp1, one synergid nucleus moved to the position of the egg cell, and acquired egg cell identity (Pagnussat et al., 2007; Kong et al., 2015). An auxin gradient was thought to provide important positional cues for embryo sac cell fate determination (Pagnussat et al., 2009; Sun et al., 2021).

CKI1-mediated signaling may be another molecular mechanism that determines the polarity of embryo sacs (Yuan et al., 2016; Erbasol Serbes et al., 2018; Hater et al., 2020). CKI1 is polarly located in the chalazal endoplasmic reticulum of the embryo sac at the FG4–FG7 stages (Yuan et al., 2016). In cki1 mutants, the markers of chalazal end cell attributes disappeared, and the markers of micropylar end cell attributes extended to the chalazal end (Supplemental Figure S7; Yuan et al., 2016), suggesting that CKI1 may be required for the establishment or maintenance of embryo sac polarity, further regulating cell fate differentiation. Similar to CKI1, ARR10/12/18 were also polarly distributed in the chalazal nuclei in embryo sacs at the FG4–FG7 stages (Figure 2). Consistently, cell attributes at the chalazal end were lost, and the molecular markers of the micropylar end cells extended to the chalazal end in arr10 12 18 mutants (Figures 4 and 5; Supplemental Figure S5). When activated ARR10/12/18 were introduced into the embryo sac, perhaps disrupting the established polarity of the embryo sac, micropylar end cell identities were replaced by a central cell identity (Figure 7, D–L; Supplemental Figure S8). Therefore, polarly localized ARR10/12/18 and CKI1 work together to establish or maintain the polarity of the female gametophyte, finally determining embryo sac cell fates.

However, it is not known how CKI1 and ARR10/12/18 are polarly localized in the syncytium embryo sac. Independent posttranscriptional regulation at each pole of the syncytial embryo sac may play a critical role in polar localization of CKI1 and ARR10/12/18. On the other hand, polar localization of CKI1 and ARR10/12/18 may be caused by protein degradation at the micropylar pole. Either posttranscriptional regulation or posttranslational regulation may largely depend on different regulatory domains at both poles of the embryo sac that may be formed after the central vacuole is generated. At the same time, the embryo sac is embedded in the ovule, whose polarity may make certain somatic factors to guide polar localization of some signaling molecules such as CKI1 and ARR10/12/18 by communication between the somatic tissue and the female gametophyte during embryo sac development.

Currently, it is not known what signals activate the CKI1-mediated signaling and what downstream factors activated by the CKI1–AHP2/3/5–ARR10/12/18 signaling contribute to the establishment or maintenance of embryo sac polarity. Understanding these biological processes will greatly help us to unravel the molecular mechanisms determining cell fates during female gametogenesis.

ARR10/12/18 play crucial roles in regulating functions of the embryo sac cells

Due to its unique developmental pattern and structure, the female gametophyte of flowering plants has become a model for studying cell division, cell polarity, cell fate, and cell signal communication (Sprunck and Gross-Hardt, 2011). In order to explore the molecular mechanisms regulating female gametophyte development, various methods have been used in the past twenty years to identify genes specifically expressed in the embryo sac. One efficient method was to use microarray analysis or high-throughput sequencing to discover DEGs in specific mutants lacking the embryo sac such as sporocyteless/nozzle (spl/nzz) and determinant infertile 1 (dif1; Bhatt et al., 1999; Schiefthaler et al., 1999; Yang et al., 1999, Yu et al., 2005; Jones-Rhoades et al., 2007; Steffen et al., 2007). However, it is difficult to identify candidate genes involved in female gametophyte development in mutants with sterile female gametophytes, such as cki1/+ and arr10 12 18/+, that cannot generate homozygous mutants, through the similar high-throughput strategies.

To obtain arr10/12/18 and cki1 embryo sacs, homozygous arr10-6 12-3 18-3 and cki1-10 mutants were created by CRISPR–Cas9 (Figure 5; Supplemental Figure S6). RNA-seq analysis with these unique resources identified 364 DEGs in arr10 12 18 and cki1 ovaries (Figure 9, A–D; Supplemental Data Set 1, A and B). Because cell fates in the chalazal pole of the arr10 12 18 and cki1 embryo sacs were impaired, we speculated that the identified DEGs may be specifically expressed in the central cell and/or antipodal cells. Consistently, expression pattern analyses of eight selected DEGs showed that they were indeed specifically expressed at the chalazal end in mature embryo sac (Figure 10). This novel strategy makes it possible to effectively identify genes involved in female gametophyte development in a high-throughput manner, and this may open an avenue to reveal molecular mechanisms regulating embryo sac cell fates.

Functional enrichment of the identified DEGs indicated that products of more than half of the genes (191/364) are localized in the extracellular region, and most of them are putative small, secreted peptides (Figure 9E; Supplemental Data Set 1, A–C). Small peptides in the female gametophyte play critical roles during fertilization. For example, the egg apparatus-secreted EGG APPARATUS 1 (ZmEA1) in maize, the synergid cell-secreted LUREs in Torenia fournieri and Arabidopsis, and XIUQIUs in Arabidopsis, guide pollen tubes into the embryo sac (Márton et al., 2005; Okuda et al., 2009; Takeuchi and Higashiyama, 2012; Zhong et al., 2019). The egg cell-produced polypeptides EC1.1–EC1.5 can activate sperm cells before gamete fusion (Sprunck et al., 2012). The central cell-derived peptide EMBRYO SURROUNDING FACTORS 1 (ESF1) regulates early embryo patterning (Costa et al., 2014). Most of the small peptide genes identified in this study are expressed in the central cell (Figure 10), suggesting that considerable signal communication exists between the central cell and other embryo sac cells. These interactive signal communications may play crucial roles to maintain the established polarity in the embryo sac, thus keeping the acquired cell identities.

As a haploid female gamete, the central cell expresses some genes involved in embryo sac development and fertilization, such as maturation of the embryo sac and pollen tube attraction (Chen et al., 2007; Wu et al., 2012; Li et al., 2015). A recent study found that the egg cell-expressed aspartic endopeptidases EGG CELL-SPECIFIC1 (ECS1) and ECS2 play a critical role to prevent polytubey by specifically cleaving LURE1 only after successful fertilization (Yu et al., 2021). Homologous aspartic endopeptidases, such as CONSTITUTIVE DISEASE RESISTANCE 1 (CDR1), AT2G28010, and AT2G28040, were also identified in the DEGs identified in this work (Supplemental Data Set 1C), suggesting that aspartic endopeptidases may also play functions to regulate putative peptide signaling in the central cell. During fertilization, de-esterified pectin in the filiform apparatus is finely guarded to prevent polytubey (Duan et al., 2020), which is stringently regulated by pectin methylesterases (PMEs) and PME inhibitors (PMEIs). Interestingly, a total of 2 PMEs and 14 PMEIs were also identified as DEGs in this study (Supplemental Data Set 1, A and C). One PMEI gene (AT1G50325) showed expression in the central cell in the WT, but not in the arr10-5 12-1 18-2 embryo sacs (Figure 10, K, L, and Q). These results suggest that PMEs and PMEIs may play important roles to finely regulate the levels of de-esterified pectin during cellularization of the embryo sac.

However, the functions of most of these identified DEGs are still unknown, and many are worthy of further investigation in the future. In addition, the direct target genes and possible interacting proteins of ARR10/12/18 need to be identified for fully dissecting the CKI1–AHP2/3/5–ARR10/12/18 signaling module during female gametophyte development. Results of these proposed studies will help us to further understand how the embryo sac cells are specified and finally achieve their capabilities.

Materials and methods

Plant materials and growth conditions

The WT and mutant Arabidopsis thaliana plants used in this study were ecotype Columbia-0 (Col-0). Seeds of arr10-5 (SALK_098604; Yokoyama et al., 2007), arr10-6 (WiscDsLox385H04), arr12-1 (SALK_054752; Mason et al., 2005), arr12-3 (SAIL_210_A09), arr18-2 (SALK_131233; Mason et al., 2005), cki1-9 (SALK_057881; Rabiger and Drews, 2013), ahp2-2 (SALK_019024; Deng et al., 2010), ahp3 (SALK_041384; Hutchison et al., 2006), and ahp5-2 (SALK_079857; Hutchison et al., 2006) were obtained from the Arabidopsis Biological Resource Center (ABRC). T-DNA insertions were verified by DNA sequencing. After vernalization at 4°C for 2 days, the seeds were planted directly in the soil (peat moss, Pindstrup, Denmark). Plants were grown in a greenhouse with a 16-h light (120 µmol s⁻¹ m⁻² light intensity provided by white LED lamps)/8-h dark cycle at 20°C –22°C. Arabidopsis transformation was performed by using the Agrobacterium-mediated floral dip method (Clough and Bent, 1998).

Plasmids

To clone the promoters and the genomic sequences of target genes, PCR products were amplified from Col-0 genomic DNA and cloned into the donor vector pDONR/Zeo using BP clonase (Invitrogen, CA, USA). The target sequences in the entry clones thus obtained were transferred into appropriate destination vectors by in vitro DNA recombination reactions using LR clonase (Invitrogen, CA, USA). All primers used in this study are listed in Supplemental Data Set 2.

To complement the arr10-5 12-1 18-2 mutant and analyze the expression patterns of ARR10/12/18, constructs proARR10:gARR10-YFP, proARR12:gARR12-YFP, and proARR18:gARR18-YFP were generated by cloning the genomic fragments, including the upstream regulatory sequences and the gene-encoding sequences without the stop codons, of ARR10/12/18 into the destination vector pBIB-BASTA-GWR-YFP (Gou et al., 2010).

To indicate cell fates in the embryo sac, the upstream regulatory sequences of embryo sac cell type-specific markers, including DD1, DD13, DD22, DD31, DD45, EC1.1, and FERTILIZATION INDEPENDENT SEED2 (FIS2), were cloned into the Gateway destination vector pFYTAG (Zhang et al., 2005) to drive the expression of NLS-YFP. The β-glucuronidase (GUS) sequence in pBIB-HYG-GUS-GWR was replaced with an NLS-mCherry fragment digested using XbaI and SacI to create the destination vector pBIB-HYG-GWR-NLS-mCherry. Promoter sequences of DD13, DD22, DD31, EC1.1, and FIS2 were then cloned into pBIB-HYG-GWR-NLS-mCherry to create the expression constructs proDD13/DD22/DD31/EC1.1/FIS2:NLS-mCherry. To create the dual marker vectors, the HindIII/EcoR I fragment of the expression vector pHEE401 (Wang et al., 2015) was replaced with combined egg cell–synergid cell, synergid cell–central cell, or egg cell–central cell markers.

To drive the expression of CKI1 genomic sequence (gCKI1) and activated ARR10/12/18 (ARR10ΔDDK, ARR12ΔDDK, and ARR18ΔDDK) in the embryo sac, the CaMV 35S promoter in the destination vector pBIB-BASTA-35S-GWR-YFP was replaced with either an AKV promoter (Rotman et al., 2005), a CKI1 promoter, or an ARR18 promoter digested with HindIII and KpnI to create the destination vectors proAKV-GWR-YFP, proCKI1-GWR-YFP, and proARR18-GWR-YFP.

The pHEE401 binary vector was used to create the arr18-3 and cki1-10 mutants using the CRISPR–Cas9 technology as described in a previous report (Wang et al., 2015).

Transgenic plants

For complementation experiments, the constructs proARR10:gARR10-YFP, proARR12:gARR12-YFP, and proARR18:gARR18-YFP were introduced into arr10-5 12-1 18-2/+ plants. To examine embryo sac cell fates, the corresponding marker genes were transferred into Col-0, arr10-5 12-1 18-2/+, cki1-9/+, arr10-5 12-1 18-2/+ cki1-9/+, and arr10-5 12-1 18-2/+ ahp2-2 3 5-2 plants, and homozygous transgenic plants were identified for statistical analysis. To analyze embryo sac cell fates in arr10-6 12-3 18-3 plants, the arr18-3 vector was introduced into homozygous arr10-6 12-3 proDD22:NLS-YFP and arr10-6 12-3 proDD45:NLS-YFP plants.

Microscopy analysis

To count seed setting rates, dissected siliques were observed and photographed with a stereomicroscope (M165C, Leica). To observe the embryo sac using DIC microscopy, inflorescences of 6-week-old plants were fixed overnight in FAA (9-mL 50% ethanol [v/v], 0.5-mL acetic acid, and 0.5-mL formaldehyde), then cleared in 70% ethanol (v/v). Ovules were mounted in chloral hydrate glycerol solution (8-g chloral, 2-mL glycerol, and 1-mL water), and photographed by a Zeiss microscope (Axio Imager.Z2).

Ovule development was analyzed with CLSM as described in a previous study (Christensen et al., 1997). Developing inflorescences were fixed in 4% glutaraldehyde in 12.5-mM dimethyl arsenate (pH 6.9) under vacuum conditions for 30 min, then incubated overnight at room temperature. After dehydration in gradient ethanol series (30%, 50%, 70%, 80%, 90%, and 100%) for at least 30 min each concentration, the samples were cleared with benzoate and benzyl alcohol (v/v = 2:1) for 1 h, then the ovules were dissected and observed under a confocal laser scanning microscope (LSM880, Zeiss) with an excitation wavelength at 488 nm and the emission was collected at 500–550 nm.

To observe YFP and mCherry fluorescence proteins, appropriate ovules were dissected in 10% glycerol (v/v) and then analyzed by a confocal laser scanning microscope (A1R+ Ti2-E, Nikon). The YFP-fused proteins were excited with a 514-nm laser line and the emission was collected at 520–560 nm. The mCherry-fused proteins were excited with a 561-nm laser line and the emission was collected at 570–620 nm.

RNA-seq and data analysis

To identify the DEGs in the arr10-6 12-3 18-3 and cki1-10 mutant plants, ovaries of flowers at stage 12 were collected and three biological replicates (about 200 ovaries per sample) were prepared for each genotype. Total RNA was isolated using a RNeasy Plant Mini Kit (74903, Qiagen, Germany), and mRNA sequencing was conducted using a BGISEQ500 platform (BGI, Wuhan, China) (Fehlmann et al., 2016). Briefly, mRNA was enriched with oligo(dT) beads and fragmented for synthesizing double-stranded cDNA with random N6 primers. After ligation with the appropriate adaptors, the double-stranded cDNA was amplified using specific primers to create a library for sequencing. The raw data were filtered by the SOAPnuke software developed by BGI (https://github.com/BGI-flexlab/SOAPnuke). Then, the sequence reads were mapped to the Arabidopsis genome using the Bowtie2 software (Langmead and Salzberg, 2012). The expression level of each sample was normalized by reads per kilobase per million mapped reads (RPKM). DEGs were defined by a two-fold expression difference between the mutant and WT samples having a q ≤0.05 as the cutoff using the DESeq2 software (Love et al., 2014). Heatmap2 in the “gplots” package of R program (version 4.1.0) was used to construct the heat map of relative gene expression. For functional annotation of these DEGs, the DAVID (Huang et al., 2009) was used to classify these genes according to their putative or known functions.

RT–PCR analysis

Total RNA was isolated using the total RNA pure Plant Kit (DP437, Tiangen, China). To examine the expression levels of type-B ARRs, isolated total RNA from seedlings, leaves, inflorescences, and ovaries from WT plants was used to synthesize cDNA using Hifair^® Reverse Transcriptase (Yeasen, China). Then, the resulting cDNA was used as the template for PCR analysis. β-TUB3 was used as a reference gene. To validate the results of RNA-seq, cDNA was synthesized from total RNA of ovaries from arr10-6 12-3 18-3, cki1-10, and the WT plants using the Hifair ^® Reverse Transcriptase (Yeasen, China). For each sample, total RNA was extracted from about 200 ovaries. RT–qPCR reactions for selected candidate genes were performed using Hieff qPCR SYBR Green Master Mix (Yeasen, China) and the primers are listed in Supplemental Data Set 2. Relative expression was calculated according to the ΔΔCT method (Livak and Schmittgen, 2001). The experiments were repeated for three times using independent RNA samples. Data were normalized to the reference gene β-TUB3, and presented as mean ± sd of three biological replicates.

Statistical analysis

The significance of differences for pairwise comparisons was estimated by two-tailed Student’s t test in Microsoft Excel 2019. One-way Turkey’s analysis of variance analyses among various genotypes were conducted using Microsoft Excel 2019 with a Real Statistics Data Analysis Tool (www.real-statistics.com), followed by least significant difference post hoc test at P < 0.05. Statistical data are provided in Supplemental Data Set 3.

Data resources

The raw data set from RNA-seq has been deposited in the NCBI Sequencing Read Archive database. The accession number for the raw data of Col-0_1 reported in this article is SRR15969053; the accession number of Col-0_2 is SRR15969052; the accession number of Col-0_3 is SRR15969051; the accession number of arr_1 is SRR115969050; the accession number of arr_2 is SRR15969049; the accession number of arr_3 is SRR15969048; the accession number of cki1_1 is SRR15969047; the accession number of cki1_2 is SRR15969046; the accession number of cki1_3 is SRR15969045.

Accession numbers

The accession numbers of genes discussed in this article are: ARR10 (AT4G31920), ARR12 (AT2G25180), ARR18 (AT2G25180), CKI1 (AT2G47430), AHP2 (AT3G29350), AHP3 (AT5G39340), AHP5 (AT1G03430), EC1.1 (AT1G76750), DD45 (EC1.2, AT2G21740), DD22 (LCR80, At5g38330), DD31 (At1g47470), DD1 (At1g36340), FIS2 (AT2G35670), DD13 (At3g59260), AKV (AT4G05440), LCR44 (AT3G06985), LCR47 (AT3G42473), LCR50 (AT2G12465), and LCR63 (AT4G30067).

Supplemental Data

The following materials are available in the online version of this article.

Supplemental Figure S1. Phylogenetic and expression analyses of type-B response regulators.

Supplemental Figure S2. Dissected siliques of single and double mutants of ARR10, ARR12, and ARR18.

Supplemental Figure S3. ARR10/12/18 are responsible for the female defects.

Supplemental Figure S4. Female gametophyte development in arr10-5 12-1 18-2/+, arr10-5 12-1/+ 18-2, and arr10-5/+ 12-1 18-2 plants.

Supplemental Figure S5. The arr10-5 12-1 18-2 female gametophyte is defective in cell fate determination.

Supplemental Figure S6. The arr10-6 12-3 18-3 and cki1-10 mutants exhibit complete abortion due to maternal defects.

Supplemental Figure S7. The cki1-9 female gametophyte is defective in cell fate determination.

Supplemental Figure S8. Statistical analysis of ovules with ectopic expression of a central cell marker.

Supplemental Figure S9. Activated ARR10/12/18 partially rescue the fertility of cki1-9.

Supplemental Figure S10. Activated ARR18 partially rescues the fertility of ahp2-2 3 5-2.

Supplemental Figure S11. Embryo sac development in ovaries at flower stage 12 of the WT, arr10-6 12-3 18-3, and cki1-10 plants.

Supplemental Figure S12. Validation of RNA sequencing results by RT-qPCR.

Supplemental Data Set 1. A, DEGs in arr10-6 12-3 18-3 plants. B, DEGs in cki1-10 plants. C, GO analysis of the overlapped DEGs between arr10-6 12-3 18-3 and cki1-10 plants.

Supplemental Data Set 2. Primers used in this study.

Supplemental Data Set 3. Summary of statistical analyses.

Supplemental File 1. Alignments used to generate the phylogeny presented in Supplemental Figure S1A.

 

X.G. and M.Z. conceived the project, designed the experiments, and analyzed the data. X.G. and M.Z. wrote the manuscript. M.Z. performed most of the experiments. L.T. performed the complementation experiments. J.Z. and Z.W. analyzed the expression patterns of ARRs, and contributed to the higher-order mutants. R.L., H.T., C.H., Y.Z., and M.L. contributed to the generation and analysis of mutant plants. Z.W. and M.Z. analyzed the RNA-seq data. J.Y. and J.L. helped prepare the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plcell) is: Xiaoping Gou (gouxp@lzu.edu.cn).