HUA ENHANCER1 Mediates Ovule Development

Ovules are female reproductive organs of angiosperms, containing sporophytic integuments and gametophytic embryo sacs. After fertilization, embryo sacs develop into embryos and endosperm whereas integuments into seed coat. Ovule development is regulated by transcription factors (TF) whose expression is often controlled by microRNAs. Mutations of Arabidopsis DICER-LIKE 1 (DCL1), a microRNA processing protein, caused defective ovule development and reduced female fertility. However, it was not clear whether other microRNA processing proteins participate in this process and how defective ovule development influenced female fertility. We report that mutations of HUA ENHANCER1 (HEN1) and HYPONASTIC LEAVES 1 (HYL1) interfered with integument growth. The sporophytic defect caused abnormal embryo sac development and inability of mutant ovules to attract pollen tubes, leading to reduced female fertility. We show that the role of HEN1 in integument growth is cell-autonomous. Although AUXIN RESPONSE FACTOR 6 (ARF6) and ARF8 were ectopically expressed in mutant ovules, consistent with the reduction of microRNA167 in hen1, introducing arf6;arf8 did not suppress ovule defects of hen1, suggesting the involvement of more microRNAs in this process. Results presented indicate that the microRNA processing machinery is critical for ovule development and seed production through multiple microRNAs and their targets.


DNA Manipulation
The artificial miRNA construct targeting HEN1 (amiR-HEN1) was designed with the primers ZP7781/ZP7782/ZP7783/ZP7784 using WMD3-Designer. The amiR-HEN1 was cloned into pROKII-GFP to generate Pro LAT 52 :amiR-HEN1. Later, Pro Lat52 was replaced by Pro INO to generate Pro INO :amiR-HEN1. The RNAi-HEN1 fragment (2297bp to 2599bp of HEN1 coding sequence) was amplified with the primer pair ZP6753/ZP6754. The resultant PCR products were sub-cloned into the RNAi vector pTCK303 (Guo et al., 2010) to obtain the Pro UBQ10 :RNAi-HEN1 construct. Later, Pro UBQ10 was replaced by Pro INO to generate Pro INO :RNAi-HEN1. Pro HEN 1 was cloned into pENTR/SD/D-TOPO (Invitrogen) with the primer pair ZP5140/ZP5173, including a 1847 bp sequence upstream of HEN1 start codon. The entry vector was used in a LR reaction with the destination vector pMD163 (Curtis and Grossniklaus, 2003) to generate Pro HEN 1 :GUS. All primers are listed in Supplementary Table S1.

RNA in situ Hybridization
RNA in situ hybridization was performed as previously described . In brief, the emasculate pistils were fixed in 4% Paraformaldehyde solution (aladdin) at 4 • C overnight. Then the fixed tissues were embedded in Paraplast (Sigma-Aldrich) after dehydration and were then sectioned at 8 µm. RNA probes of ARF6 and ARF8 were amplified with the primer pairs ZP8093/8094 and ZP8095/8096, respectively. The sense and antisense probes were modified in vitro with digoxigenin-UTP by SP6 or T7 RNA polymerases (Roche), respectively. Sections were hybridized with 1.5 ng/µL probes at 42 • C overnight in a hybridization solution that contained formamide. Hybridization signals were detected by antidigoxigenin antibody (Anti-Digoxigenin-Ap Fab fragments; Roche). The samples were observed using an Olympus BX53 microscope. All primers are listed in Supplementary Table S1.

Phenotype Analysis
Pollen tube in vivo growth by histochemical GUS staining of Pro LAT 52 :GUS-pollinated pistils and aniline blue staining were performed as described . Whole-mount ovule clearing and CLSM of ovules were performed as described (Wang et al., 2016;Liu et al., 2019). Flowers at stage 12 were emasculated and left to grow for 12-16 h before pollination assays.

Fluorescence Microscopy
Lysotracker red staining was used to show cell silhouettes as described (Wang et al., 2016). CLSM of fluorescence materials was performed with a LSM880 (Zeiss) with the excitation and emission wavelengths set to 488 nm/505-550 nm for YFP and GFP signals and 561 nm/600 nm for RFP signals, respectively.

Accession Numbers
Arabidopsis Genome Initiative locus identifiers for the genes mentioned in this article are: AT4G20910 for HEN1; AT1G09700 for HYL1; AT3G22886 for miRNA167; AT1G30330 for ARF6; AT5G37020 for ARF8.

hen1-8 Shows Reduced Fertility Due to Sporophytic Female Defects
To determine what caused the reduced fertility in hen1-8 (Chen et al., 2002;Yu et al., 2005Yu et al., , 2010, we performed the following experiments. First, we observed white and wrinkled ovules dispersed among developing seeds in the maturing siliques of hen1-8 plants, but not in those of wild type or of hen1-8/ + (Figures 1A,B), indicating that the reduced fertility is sporophytic. Indeed, segregation ratio by reciprocal crosses indicated that both the male and female gametophytes of hen1-8 were transmitted normally (Supplementary Table S2). Pollen development of hen1-8 is also comparable to that of wild type (Supplementary Figure S1). By dissecting siliques from crosses between wild type and hen1-8, we observed reduced fertility only when hen1-8 was used as the female parent ( Figures 1A,B), Representative seed set. For some, two overlapping high-magnification images were taken for one silique and were then overlaid with Photoshop (Adobe Systems) to show the whole silique. (B) Quantitative analysis of seed set. Results are means ± standard deviation (SD, n = 15). Different numbers indicate significantly different groups (One-Way ANOVA, Tukey's multiple comparisons test, P < 0.05). Bars = 500 µm.
suggesting that the reduced fertility of hen1-8 was due to sporophytic female defects.

hen1-8 Is Defective in Sporophytic Control of Ovule Development
To determine the reason for sporophytic female defects that caused reduced fertility in hen1-8, we examined the morphology of mature ovules by scanning electron micrographs (SEMs) and whole-mount ovule clearing (Wang et al., 2016). At maturation, wild-type ovules showed a typical anatropy with the micropyle proximal to the funiculus (Figures 2A,B). An embryo sac was clearly seen in the mature ovule of wild type ( Figure 2E). By contrast, micropyle structure was not discernible in a portion of hen1-8 ovules (Figures 2C,R). Instead, a bulge, likely a deformed embryo sac, was exposed (Figures 2D,G,H). These results suggested that ovule development is compromised in hen1-8.
To determine at which stage the hen1-8 ovules started to be defective, we performed confocal laser scanning microscopy (CLSM) of developing ovules. At early stages, i.e., before the meiosis of megaspore mother cell (MMC), hen1-8 and wild type are comparable although HEN1 is expressed in ovules throughout development (Supplementary Figure S2). However, at stage 3-I when the outer integuments of wild type started rapid and asymmetric growth, extending above the inner integuments (Figure 2I), the growth of hen1-8 outer integuments was delayed, hardly reaching the length of the inner integuments ( Figure 2L). At this stage, functional megaspore (FM) was formed both in wild type and in hen1-8 (Figures 2I,L). In wild type, from stage 3-III to maturation, the outer integuments continued extended growth, finally enclosing the inner integuments ( Figure 2J,K). Every mature ovules of wild type contains an embryo sac with a central cell, an egg cell and two synergid cells (Figure 2O). By contrast, the outer integuments of hen1-8 failed to enclose the inner structure (Figures 2M,N). At maturation, these ovules contain embryo sacs with abnormal cellular structures such that only one nucleus was visible (Figures 2P,Q,S).
To provide further evidence that HEN1 is the causative gene for the observed ovule defects in hen1-8, we performed additional experiments. Because hen1-2 is an allelic HEN1 mutant in Landsberg errecta (Ler) that contains exactly the same site mutation as in hen1-8 (Yu et al., 2010), we first examined ovule development of hen1-2 by CLSM. Indeed, hen1-2 showed the same ovule defects as those of hen1-8 (Supplementary Figure S3). Next, we crossed hen1-8 and hen1-2 and examined ovules of the F1 progenies. Ovules of the F1 progenies from the cross showed exactly the same defects (Supplementary Figure S4). These results indicated defective outer integument growth affected embryo sac development when HEN1 is mutated.

hen1-8 Ovules Showed Reduced Pollen Tube Attraction
CLSM of Pro ES1 :NLS-YFP;hen1-8 in which a nucleustargeted YFP was driven by an embryo sac-specific promoter (Pagnussat et al., 2009), often showed one nucleus, sometimes no nucleus at all, in the embryo sac in contrast to the eight nuclei structure in wild-type ovules (Supplementary Figure S5), suggesting that embryo sac development was compromised due to sporophytic defects in hen1-8. This result is also consistent with those obtained by optical section of mature hen1-8 ovules (Figure 2).
Because the embryo sac within an ovule attracts pollen tubes for fertilization (Higashiyama and Yang, 2017), abnormal embryo sacs due to defective integument growth might be the reason for reduced female fertility in hen1-8 (Figure 1). To test this hypothesis, wild-type or hen1-8 pistils were emasculated, and hand-pollinated with Pro LAT 52 :GUS pollen and pollen tube attraction at 12 h after pollination (HAP) was examined by histochemical GUS staining. In contrast to wild type in which almost all ovules were targeted by a pollen tube, as indicated by a blue blob inside embryo sacs (Figures 3A,C), over half of hen1-8 ovules failed to attract a pollen tube (Figures 3B,D). By aniline blue staining of pistils at 48 HAP, we determined that most wild-type ovules were fertilized as indicated by size increase (Figure 3E). By contrast, around half of hen1-8 ovules were not targeted by pollen tubes and were not fertilized (Figures 3F,G). Therefore, we concluded that defective embryo sac development in hen1-8 resulted in its reduced female fertility.

hyl1-2 Mimicked Ovule Defects of hen1-8
Ovule defects of hen1-8 were likely due to compromised miRNA processing because HEN1 is critical for the processing of various miRNAs (Yu et al., 2005(Yu et al., , 2010Zhao et al., 2012) and sin1/dcl1-7 showed a similar phenotype (Robinson-Beers et al., 1992). To provide further evidence that the miRNA processing pathway was critical for ovule development, we also examined hyl1-2 (Vazquez et al., 2004), a null mutant of HYL1 whose severely reduced fertility was restored to the wild-type level by exogenous HYL1 (Lian et al., 2013). Female gametophytes of hyl1-2 transmitted comparably to those of wild type (Xiong et al., 2020), indicating that HYL1 is not required for the development of female gametophytes. However, the homozygous hyl1-2 showed a significantly reduced seed set due to sporophytic female defects (Figures 4A-E). By ovule whole-mount analysis (Figures 4F-G), optical sections (Figures 4H-I), and SEM analysis (Figures 4L-M), we demonstrated that hyl1-2 was defective in ovule development due to the growth arrest of outer integuments. Because of the defects, hyl1-2 ovules showed a reduced ability to attract pollen tubes compared with those of wild type (Figures 4J-K), leading to significantly reduced female fertility (Figures 4B,D,E).

HEN1 Functions in a Cell-Autonomous Way
hen1-8 shows vegetative growth retardation (Chen et al., 2002), which could have an impact in female fertility. To exclude the possibility that ovule developmental defect of hen1-8 was resulted from its reduced vegetative growth, we attempted to downregulate the expression of HEN1 specifically in outer integuments by using the outer integument-specific promoter Pro INO (Wang et al., 2016). A dozen of transgenic lines containing either Pro INO :amiR-HEN1 (artificial microRNA-HEN1) were generated. The transgenic plants were comparable to that of wild  type regarding the growth of vegetative tissues (Supplementary Figure S6), consistent with the use of outer-integument-specific promoter. However, seed set of the transgenic plants was compromised (Figures 5B,C,F). We examined two independent transgenic lines representing mildly or severely affected types by CLSM. In the line of Pro INO :amiR-HEN1 in which seed set was reduced by 50% (Figures 5B,F), half of the mature ovules showed abnormal number of nuclei in their embryo sacs while the integuments were morphologically indistinguishable from those of wild type (Figures 5G,H). In the line where there was hardly any seed set (Figures 5C,F), most mature ovules had no discernible outer integuments or nucleus structure in their embryo sacs (Figures 5I,J).
Because the specificity of Pro INO used to downregulating HEN1 (Wang et al., 2016) and the constitutive expression of HEN1 in ovules (Supplementary Figure S2), we could not verify the downregulation of HEN1 in the Pro INO :amiR-HEN1 transgenic plants by quantitative real-time PCRs (qPCRs). Instead, we performed two experiments to support that HEN1 mediates integument growth in a cell-autonomous way. First, we generated Pro 35S :amiR-HEN1 transgenic plants and examined the transcript abundance of HEN1 in transgenic seedlings by qPCRs. In randomly selected two Pro 35S :amiR-HEN1 independent lines, HEN1 abundance was significantly reduced compared to that in wild type (Supplementary Figure S7), suggesting that the amiR-HEN1 expression did reduce the mRNA level of HEN1. Consistently, the transgenic plants were shorter and smaller than wild-type plants (Supplementary Figure S7). Second, we used a RNA interference (RNAi) approach instead of amiR to downregulate HEN1 specifically in outer integuments. The Pro INO :HEN1-RNAi transgenic plants phenocopied Pro INO :amiR-HEN1 in reduced seed set (Figures 5D,E,F) and defective ovule development (Figures 5K-N). These results suggested that HEN1 mediates outer integument growth in a cell-autonomous way.

Auxin Distribution but Not the Asymmetric PIN1 Localization Was Compromised in hen1-8 Ovules
Auxin is a determinant factor in ovule development (Benkova et al., 2003;Bencivenga et al., 2012;Ceccato et al., 2013). Both auxin receptors and response factors are regulated by miRNAs whose processing depends on the DCL1-HEN1-HYL1 pathway (Navarro et al., 2006;Gutierrez et al., 2009;Ren et al., 2012;Zhao et al., 2012). Therefore, we examined auxin responses by introducing DR5:GFP (Ulmasov et al., 1997) into hen1-8 and examining GFP distribution. In wild type, GFP signals were detected only in the epidermal cell layer of the nucellus at stage 2-III when both outer and inner integuments were initiated ( Figure 6A) and at stage 3-I when outer integuments underwent rapid growth to establish ovule anatrophy (Figure 6B). At maturation, GFP signals were hardly visible in ovules except in the vascular tissues of the funiculus (Figure 6C). The GFP distribution of hen1-8 was similar to, albeit weaker than, that of wild type at early stages ( Figure 6D) and at maturation in morphologically normal hen1-8 ovules ( Figure 6F). However, at stage 3-I, auxin maximum was expanded from the nucellus to the developing female gametophytes of hen1-8 ovules (Figure 6E), suggesting a spatially disturbed auxin response.
Because PIN1 is the key auxin efflux carrier responsible for auxin distribution during ovule development (Ceccato et al., 2013), we also generated the PIN1:GFP;hen1-8 plants to examine its distribution. As reported previously (Ceccato et al., 2013;Wang et al., 2016), PIN1 was asymmetrically distributed at the epidermal cells of the nucellus during ovule development (Figures 6G-H) and restricted to the funiculus at maturation (Figure 6I). No difference of PIN1 distribution was observed between wild type and hen1-8 (Figures 6J-L), indicating that compromised auxin maximum in hen1-8 ovules was likely resulted from signaling rather than auxin transport.
Suppressing the Ectopic Expression of ARF6 and ARF8 by Introducing arf6;arf8 Did Not Rescue Ovule Defects of hen1-8 Mutations of DCL1 (Robinson-Beers et al., 1992), HEN1, and HYL1 resulted in defective ovule development, suggesting a role of miRNAs in this process. Among miRNAs whose accumulation relies on HEN1 (Ren et al., 2012;Zhao et al., 2012), miRNA167 was demonstrated a positive regulator for ovule development by suppressing the expression of ARF6 and ARF8 (Wu et al., 2006;Yao et al., 2019). Indeed, ovule development of the mir167 mutants was largely recovered by introducing either arf6 or arf8 (Yao et al., 2019).
To test the possibility that ectopic expression of ARF6 and ARF8 resulted in the arrest of integuments in hen1-8, we performed the following experiments. First, we examined the expression of ARF6 and ARF8 in hen1-8 ovules by RNA in situ hybridization. As reported previously (Wu et al., 2006), ARF6 (Figures 7A,B) and ARF8 (Figures 7E-F) were highly expressed in the funiculus during ovule development in wild type. By contrast, signals of either ARF6 (Figures 7C,D) or ARF8 (Figures 7G,H) were detected in whole ovules of hen1-8, indicating its ectopic expression. Second, by quantitative RT-PCR (qRT-PCR), we could verify that transcript abundance of both ARF6 and ARF8 was significantly increased in hen1-8 ovules (Figures 7I,J). Third, we introduced the mutants of ARF6 and ARF8, i.e., arf6-1 and arf8-3 respectively, into hen1-8 and analyzed the resultant hierarchy mutants. Introducing arf6-1 or arf8-3 alone into hen1-8 did not affect vegetative growth whereas the arf6-1;arf8-3;hen1-8 showed a severe growth retardation (Supplementary Figure S8). Close examination of mature ovules from different genotypes showed that either arf6-1, or arf8-3, or the arf6-1;arf8-3 double mutant could not restore ovule developmental defects of hen1-8 (Supplementary Figure S9), suggesting that ectopic expression of ARF6 and ARF8 was not the reason for the arrest of hen1-8 integuments.
A recent study reported that the mutations of HYL1, DCL1, or HEN1 caused a reduced number of pollen and megaspore mother cells (Oliver et al., 2017). Our results strongly suggested that fertility reduction in hen1-8 was due to abnormal ovule development (Figure 3). First, hen1-8 as the pollen donor to wild type resulted in a full seed set (Figure 1), indicating normal pollen function. Second, pollen development is comparable between hen1-8 and wild type, despite the relatively small size of hen1-8 anthers (Supplementary Figure S1). Third, the heterozygous hen1-8 mutant also produced full seed set (Figure 1), arguing against female gametophytic defects. Indeed, reciprocal crosses between wild type and the heterozygous hen1-8 indicated that both male and female transmission of hen1-8 are normal (Supplementary Table S1). The discrepancy between the previous report (Oliver et al., 2017) and ours might be due to the fact that hen1-8 and hen1-2 are weaker alleles of HEN1.
Despite that the hen1-8 plants showed sub-optical vegetative growth (Chen et al., 2002;Ren et al., 2012), we believe that HEN1 works in a cell-autonomous way to regulate the asymmetric growth of outer integuments. Downregulating HEN1 specifically in outer integuments was sufficient to cause ovule defect similar to, even more severe than, that of hen1-8 (Figure 5), without affecting vegetative growth (Supplementary Figure S6). Because the cell-specific feature of these transgenic lines, it is difficult, if possible, to examine the reduction of HEN1 transcript abundance. To make it additionally difficult, in the more severely affected RNAi or amiR lines, the growth of outer integuments was arrested very early on. But the two different constructs used to downregulating HEN1 in outer integuments gave the same results, strongly supporting a cell autonomous role of HEN1 in outer integuments.
The accumulation of miRNA167 was significantly reduced in hen1 mutants (Yu et al., 2010;Ren et al., 2012). Consistently, ARF6 and ARF8, major targets of miRNA167 (Nagpal et al., 2005;Wu et al., 2006;Yao et al., 2019;Zheng et al., 2019), were ectopically expressed in hen1-8 (Figure 7). However, introducing arf6 or arf8 did not suppress developmental defects of hen1-8 ovules (Supplementary Figure S9). The inability is unlikely to have caused by substantially compromised growth of the arf6-1;arf8-3;hen1-8 triple mutant since introducing either arf6-1 or arf8-3 didn't aggravate the growth of hen1-8 but yet was not able to rescue its defects (Supplementary Figure S8). A more likely possibility is that more miRNAs downstream of HEN1 play roles in this process. Auxin maximum was altered in developing ovules of hen1-8 such that DR5 signals were expanded to the developing female gametophytes in hen1-8 rather than restricted to the nucellus as in wild type (Figure 6). Because the sporophytic integuments affect FG development (Bencivenga et al., 2011;Wang et al., 2016;Liu et al., 2019), these results indicated that auxin signaling in integuments was compromised in hen1-8. Indeed, genes encoding auxin receptors are also targets of Funiculus of wild-type ovules whereas integuments of hen1-8 ovules show strong signals by the antisense probes. (I,J) Relative ARF6 (I) or ARF8 (J) abundance in mature ovules of wild type or hen1-8 by qRT-PCRs. GAPDH and TUBULIN2 were used as internal controls. Results shown are means ± SE (n = 3). Asterisks indicate significant difference (t-test, P < 0.01). Bars = 50 µm for whole views of pistils and 20 µm for close-ups of ovules.
miRNAs (Navarro et al., 2006;Gutierrez et al., 2009). A genomewide small RNA sequencing will be useful to identify miRNAs that are expressed in integuments and whose reduced levels result in the arrest of integument growth in mutants of the DCL1-HEN1-HYL1 pathway.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
S-JW and SC performed all the experiments with the assistance of R-MZ and C-YD. SL and YZ conceived and supervised the project and secured the funding. S-JW, SL, and YZ analyzed the data. YZ wrote the article with input from all authors.

FUNDING
This work was supported by Natural Science Foundation of China (31871422 and 31771558 to SL, 31970332 and 31625003 to YZ). YZ's laboratory is partially supported by Tai-Shan Scholar Program by Shandong Provincial Government.