Repression of microRNA 160 results in retarded seed integument growth and smaller final seed size in cotton

Nin Liu, Shuchng Wu, Zhonghu Li, Anm Qdir Khn, Hiyn Hu,Xinlong Zhng, Lili Tu,

aNational Key Laboratory of Crop Genetic Improvement,Huazhong Agricultural University,Wuhan 430070,Hubei,China

bKey Laboratory of Biology and Genetic Improvement of Oil Crops,Ministry of Agriculture and Rural Affairs,Oil Crops Research Institute of the Chinese Academy of Agricultural Sciences,Wuhan 430062,Hubei,China

ABSTRACT microRNA 160 (miR160), targeting auxin response factors (ARFs), plays many roles in plant development.We investigated the role of the miR160/ARF axis in regulation of cotton seed size.Suppressing miR160 activity,specifically in the seed coat,led to smaller seeds and less fiber production owing to attenuated growth of the maternal integument. Scanning electron microscopy and histology showed that expansion of cells in the integument was retarded in miR160-suppressed lines. Four GhARF genes were targeted by miR160 and were upregulated in miR160-suppressed lines, indicating that a miR160/ARF axis is present in cotton. Five genes (Ghir_A05G003740, Scaffold1878G000010, Ghir_D09G024980,Ghir_A11G010730,and Ghir_A05G041590),associated with reduced seed development were downregulated in miR160-suppressed lines. Our results suggest that the miR160/ARF axis controls maternal integument growth to influence seed size by directly or indirectly regulating seed development-associated genes.

1. Introduction

The seed is the major harvested product for food and other crops, so that seed size or weight is an important yield determinant. In cotton, the fiber is the main economic product. Given that fiber cells develop from the cotton seed epidermis, seed development strongly influences fiber growth, yield, and quality [1]. Cotton seeds are also the world sixth-largest source of vegetable oil [2]. In nature, seed size varies among plant species and is linked to evolutionary fitness. Large-seeded plants accumulate abundant nutrients for seedling establishment and are better able to tolerate stress,whereas small-seeded plants do well in dispersing and colonizing[3].The mechanisms that control seed size are thus of interest to both agriculture and biology.

A seed consists of a triploid endosperm, a diploid embryo,and a seed coat derived from maternal integument. The coordinated growth of the three components determines final seed size. Mutant screening showed that a VQ motif protein gene HAI-KU1,a leucine-rich repeat receptor kinase gene IKU2,and the WRKY transcript factor gene MINI-SEED 3 function in the same signaling pathway to regulate endosperm cellularization and seed size [4,5]. Epigenetic regulation also controls endosperm growth. For example, a group of polycomb proteins formed a complex that suppressed gene expression via histone methylation [6]. The embryonic transcription factors AP2 and bHLH, as well as phytohormone signaling pathways, also regulated embryo cell proliferation to influence seed size [7-9]. In such cases, zygotic genotype determines final seed size. However, maternal integuments,which supply the cavity for embryo and endosperm growth,impose a physical upper limit on embryo enlargement and control final seed size.Several QTL for kernel weight and size influenced seed coat development by G-protein signaling,mitogen-activated protein kinase signaling, and the ubiquitin-proteasome pathways in rice and Arabidopsis [10].The transcript factors TTG2 and GRF4 as well as cytochrome P450 family members KLU and EOD3 also regulated integument growth and further influenced final seed size[11-14].

miRNAs are noncoding RNAs 21 nt to 24 nt in length.Plant miRNAs are involved in regulation of morphogenesis, development, stress responses, and nutrient uptake [15-20]. A few miRNAs including OsmiR396 and OsmiR397 play important roles in regulation of seed development [12,21].

The present study investigated miR160 function in seed development.miR160,which targets ARF transcripts,is likely to be involved in auxin signaling transduction, auxin-cytokinin interaction,and/or jasmonic acid homeostasis to influence root development,floral organ morphogenesis,and callus formation[22-26].Our results show that miR160 also regulates the growth of integuments to influence seed size and fiber production in cotton.At least four ARFs were identified as targets of miR160 in the cotton seed. Histological analysis and qRT-PCR analysis suggested that the miR160/ARF axis regulates cell expansion to control cotton seed coat growth by regulating expression of GhSus1, orthologs of AtGIF, OsGL7, AtCYP78A6, and AtEXP8(Ghir_A05G003740, Scaffold1878G000010, Ghir_D09G024980,Ghir_A11G010730,and Ghir_A05G041590).

2. Materials and methods

2.1. Plant materials and RNA isolation

The cotton (Gossypium hirsutum) cultivar YZ1 was used as transgenic receptor and wild type. Non-transgenic and transgenic lines were planted in the experimental field and greenhouse under standard farming management at Huazhong Agricultural University, Wuhan, Hubei, China.Bolls were tagged on the day of anthesis and harvested at 0 and 10 days post-anthesis (DPA). The ovules and fibers were excised and then rapidly immersed in liquid nitrogen and stored at −80 °C. Total RNA was extracted from the collected tissues using a thiocyanate method[27].

2.2. Plasmid construction and genetic transformation

To suppress miR160 function specifically in the seed coat,we constructed a miR160 target mimic vector driven by a promoter of the MADS box gene floral binding protein 7(FBP7). The FBP7 promoter was used to replace the CaMV 35S promoter in the vector pGWB402 to drive expression specifically in the seed coat at −2 to 10 DPA.The genomic sequence of IPS1 from Arabidopsis (Col-0), containing a mini-ORF and miR399 complementarity motif was cloned. The latter motif was replaced with the miR160 complementarity motif by PCR as previously described[28].IPS1 with the new motif was then cloned under the FBP7 promoter(Fig.S1).A genomic sequence containing a miR160b precursor from Arabidopsis was also cloned and ligated downstream of the FBP7 promoter in pGWB402 to overexpress miR160.The oligonucleotide primers used for constructing the plasmids are listed in Table S1.

Agrobacterium tumefaciens (GV3101) with the construct was used to transform hypocotyls of YZ1. Infected hypocotyls were induced to produce regenerative shoots following Jin et al.[29].

2.3.Southern and northern blotting and qRT-PCR

Southern blotting was performed following Li et al. [30]. An NPTII fragment was used as the probe. The primers used are listed in Table S1.For Northern blotting,total RNA extraction,electrophoresis, membrane transfer, probe labeling, and hybridization were performed following Liu et al.[31].

To quantify gene expression,total RNA(3 μg)was reversetranscribed to cDNA using SuperScript II reverse transcriptase(Invitrogen, Carlsbad, USA). A 7500 real-time system (Applied Biosystems,Waltham,USA)was used to perform quantitative real-time PCR. The ubiquitin GhUBQ7 (Ghir_A11G011460) was used as endogenous reference control and the relative expression levels were calculated using the modified 2−ΔΔCTmethod [32]. The nucleotide sequences of genes and their homologs are presented in Fig.S6.

2.4.Scanning electron microscopy and histology

Fibers on ovules from wild-type, miR160-suppressed lines,and null plants (segregating nontransgenic plants derived from transgenic plants) were visualized by scanning electron microscopy using a JSM-6390/LV instrument (JEOL, Tokyo,Japan).

Ovules collected from flowers at 0 DPA and 10 DPA were fixed in FAA solution for 24 h.They were then dehydrated in a graded ethanol series (50%, 70%, 85%, 95%, and 100%, v/v),stained with eosin-Y, and embedded in paraffin. They were then cut into 10-μm sections and stained with Fast Green dye.Images of paraffin sections were made by light microscopy(Zeiss,Oberkochen,Germany).

2.5.Traits of seed and fiber measurement

Non-transgenic and transgenic lines were evaluated in twoyear trials based on a randomized block design. Mature bolls from the middle part of the cotton plant were hand-harvested at the same time from three replicate plots. More than 150 seeds from 12 bolls were measured for seed index (seed weight in grams per 100 seeds)and lint index(fiber weight in grams per 100 seeds).At least 10 g of mature fiber was used to determine fiber quality by HFT9000 instrument (Premier,Coimbatore,India).

2.6. Identification of miR160-targeted ARFs

Cotton genome sequence was obtained from Wang et al.[33]in the CottonGen Database (https://www.cottongen.org/), and ARF10, ARF16, and ARF17 protein sequences were retrieved from TAIR 10 (http://www.arabidopsis.org/). BLASTP with default parameters was used to search for orthologs of ARF10, ARF16, and ARF17 in the cotton genome. A phylogenetic tree was constructed to further verify cotton orthologs of ARF10,ARF16,and ARF17 using the neighbor-joining algorithm in MEGA software (https://www.megasoftware.net). These cotton genes were used to predict miR160-targeted ARFs using psRNATarget(http://plantgrn.noble.org/psRNATarget/).

2.7. Cluster analysis of gene expression profiles

Gene expression data were obtained from the BioProject database (PRJNA248163). A heat map of gene expression was constructed with Genesis (http://genome.tugraz.at/). Hierarchical clustering of genes was performed using the unweighted pair-group average linkage algorithm.

2.8. RNA ligase-mediated rapid amplification of cDNA ends(RLM-RACE)

A GeneRacer kit (Invitrogen) was used to map the cleavage sites of target transcripts. Total RNA extracted from 0 DPA ovules was ligated to adapters and then transcribed to cDNA following Liu et al.[34].

2.9. Histochemical analysis of GUS activity

Ovules dissected from ovaries (0-5 DPA) were incubated in GUS staining buffer at 37°C for 4 h,then washed twice in 75% alcohol and imaged with a stereomicroscope (Leica Microsystems, Weztlar, Germany). The staining buffer consisted of 50 mmol L−1sodium phosphate buffer (pH 7.0),0.9 g L−15-bromo-4-chloro-3-indolylglucuronide, 20% (v/v)methanol and 100 mg L−1chloromycetin.

3. Results

3.1. Suppressing miR160 activity leads to smaller seeds in cotton

Previous studies [23,26,35,36] showed that miR160 functions in the regulation of the development of several plant organs. To investigate the role of miR160 in cotton seed development, we constructed a miR160 target mimic vector containing a FBP7 promoter (a seed coat specificity promoter, −2 to 10 DPA) to suppress miR160 activity [37]. Six transformants were obtained(Fig. S2). At least three miR160-suppressed lines (M46, M74, and M75)expressed the target mimic of miR160 in cotton ovules(Fig.1-A,B),and the developing and mature seeds of these lines were smaller and lighter than those of non-transgenic lines in twoyear field trials(Fig.1-C,Table 1).Line M41,which was detected with a lower target mimic transcript abundance in cotton ovules,showed little difference in seed size and weight compared to the non-transgenic (null) line. Suppression of miRNA activity by expression of the target mimic of miRNA indicated that suppression of miR160 affects cotton seed development.

Overexpression of the AtmiR160b precursor in cotton showed that miR160 abundance was not significantly higher in five independent FBP::AtmiR160b transgenic lines than in non-transgenic lines(Figs.S2 and S3).We accordingly focused on the miR160-suppressed lines to investigate miR160 function in cotton.

Fig.1-Overexpressing the target mimic of miR160 leads to smaller seed size.(A)RT-PCR analysis of target mimic expression.(B)qRT analysis of target mimic expression.Error bars indicate the standard deviation of four biological replicates.Bars with different letters differ significantly according to Tukey's range test at P ＜0.05.“#”indicates no expression.(C)Images of seeds at 20 DPA(upper panel)and mature seeds without fibers(under panel).WT,wild type(Gossypium hirsutum cv.YZ1).M41,M46,M74,and M75 are miR160-suppressed lines.Null,nontransgenic plants segregating from miR160-suppressed lines.

Table 1-Comparison of seed index,lint index and lint percentage between wild type,null,and miR160-suppressed lines.

Table 2-Comparison of fiber quality parameters among wild type,null, and miR160-suppressed lines.

Fig.2-Scanning electron microscope images of 1.5 DPA ovules.(A-D)Fiber cells of WT(A),null(B),M46(C),and M74(D)at 60×magnification.(E-H) Fiber cells of WT(E), null(F),M46(G), and M74(H)at 500×magnification.WT,wild type(Gossypium hirsutum cv.YZ1);M46 and M74,miR160-suppressed lines.Null,nontransgenic plant segregating from miR160-suppressed lines.

3.2.Early fiber elongation is attenuated in miR160-suppressed lines

Seed weight and lint index in miR160-suppressed lines were reduced in comparison with non-transgenic lines in two-year field trials (Table 1). Although lint percentage seemed to be higher in miR160-suppressed lines than in the null,the trend was not significantly different in 2015 (Table 1). The five quality indices for mature fiber quality showed that fiber length and uniformity index were clearly reduced in suppressed lines (Table 2). Under scanning electron microscopy,fibers in null and WT elongated like hairs,but fiber in M46 and M74 initiated only as bulbs,indicating that fiber elongation in M46 and M74 was severely retarded (Fig. 2). However, almost no difference in fiber initiation density between miR160-suppressed lines, null, and WT was observed (Fig. 2). Thus early fiber cell elongation, not initiation, was attenuated following miR160 suppression.

Fig.3-Cell size was reduced in transgenic lines.(A-D)Images of cross sections of ovule at 0 DPA from WT(A),Null(B),M46(C),and M74(D).(E-H)Images of parenchymal cells of ovule inner integument at 10 DPA from WT(E),Null(F),M46(G),and M74(H).e,epidermis;o,outer integument of ovule.Bars in (A-H)represent 50 μm.(L)Size of epidermal cells of outer seed coat from ovules at 0 DPA.(M)Size of parenchymal cells of inner seed coat from ovules at 10 DPA.M46 and M74 are miR160-suppressed lines.Null,nontransgenic plant segregating from miR160-suppressed lines.

3.3. Seed coat cell size is reduced in suppressed lines

In tissue sections of 0 DPA ovules, outer integument cells including epidermis were significantly smaller in miR160 suppressed lines than in null and WT (Fig. 3-A-D, L). Except for the outer integument, the size of parenchymal cells from inner integument was significantly decreased in 10 DPA ovules of miR160-suppressed lines (Fig. 3-E-H, M). Given that miR160 was suppressed specifically in seed coats of transgenic plants, these results suggested that miR160 is involved in integument cell expansion in cotton.

3.4.miR160 suppresses the abundance of target genes in early seed development

The miR160 family is one of the highly conserved miRNA families in the plant kingdom[38]and targets a subgroup of the ARF family(ARF10,ARF16,and ARF17)in Arabidopsis.In cotton,16 genes are orthologous to ARF10,ARF16,and ARF17(Fig.S4),of which 15 are predicted miR160 targets (Table S2). In cotton,miR160 is preferentially expressed in early ovule development.After the ovule was fertilized, miR160 abundance decreased dramatically (Fig. 4-A, B). In contrast, most candidate miR160 targets were more highly expressed in 10-DPA than in 0-DPA ovules(Fig.4-C).The contrasting expression patterns in miR160 and candidate targets suggested that miR160 suppresses its target expression in early seed development.To further test this hypothesis, we measured candidate target transcript abundance in miR160-suppressed lines. Given that allotetraploid cotton harbors homoeologous gene pairs, eight genes were selected to represent miR160-targeted GhARFs. Of these, four genes orthologous to AtARF10 and AtARF17 (Ghir_A03G004010,Ghir_D05G010340,Ghir_D05G023480,and Ghir_A06G004210)were expressed more highly in 10-DPA ovules and fibers of suppressed lines than in null lines(Fig.5).RLM-RACEs showed that more cleavage sites of candidate gene transcripts were in 9th to 11th of the complementarity with miR160,verifying that miR160 negatively regulated Ghir_A03G004010,Ghir_D05G010340, Ghir_D05G023480, and Ghir_A06G004210 expression in early cotton seed development(Fig.6).

Fig.4-Expression patterns of GhmiR160 and candidate targets.(A)qRT analysis of GhmiR160 expression level in cotton.Error bars indicate standard deviation of three biological replicates.Bars with different letters are significantly different by Tukey's range test at P ＜0.05. R,H,L, P,An,St,-3O,0O,3O,10O,10F,15F,20F,and 25F represent root,hypocotyl, leaf,petal,anther,stigma,and ovule at −3 DPA,ovule at 0 DPA,ovule at 3 DPA,ovule at 10 DPA,fiber at 10 DPA,fiber at 15 DPA,fiber at 20 DPA,and fiber at 25 DPA,respectively.(B)Northern blotting analysis of GhmiR160 expression level in several ovule-development stages.5S RNA was used as loading control.(C)Hierarchical clustering of expression patterns of candidate targets in several ovuledevelopment stages.

3.5. Expression of seed and fiber development-associated genes is reduced in miR160-suppressed lines

The ARF family mediates auxin signal transduction [39]. Suppression of miR160-targeted ARFs by ectopic expression of miR160 led to auxin signaling hypersensitivity [23]. We crossed DR5::GUS transgenic plants with M74 and null plants. DR5::GUS expression monitors auxin response[31,40,41].0 DPA ovules from M74×DR5::GUS and Null×DR5::GUS progeny plants were treated with 5 μL IAA to assay physiological response to exogenous auxin.Compared to ovules treated with water, GUS expression increased in IAA-treated ovules from M74×DR5::GUS and Null×DR5::GUS (Fig. S5-E, F). However, GUS expression in 0-5 DPA ovules did not differ significantly between M74 × DR5::GUS and null × DR5::GUS (Fig. S5-A-D), indicating that the difference in auxin signal transduction between miR160-suppressed line and null was negligible.

We then focused on seed development-related genes to identify the mechanism by which suppression of miR160 reduces seed weight and fiber length.The GhSus gene plays an important role in cotton seed development, and suppression of GhSus expression in transgenic cotton reduces seed weight[42].In our miR160-suppressed lines (M46 and M47), GhSus transcript level was lower than in null (Fig. 7-A). Besides GhSus, other genes control seed size in Arabidopsis and rice [10], and orthologs of these genes in cotton were selected for expression analysis in 10 DPA ovules (Fig. 7-F). Three genes (Ghir_Scaffold1878G000010,Ghir_D09G024980, Ghir_A11G010730) which are orthologous to AtGIF, OsGL7 and AtCYP78A6 were downregulated in M46 and M74, compared to null (Fig. 7-B-D). The abundance of AtEXP8 ortholog transcript in 10 DPA fibers was lower in M46 and M74 than in null(Fig.7-E).AtEXP8 belongs to the expansin gene family,increases cell wall expansion, and promotes root and fiber cell elongation [43-45]. We accordingly speculated that reduced expression of seed and fiber development-associated genes in miR160-suppressed lines led to the observed decrease in seed weight and fiber length.

Fig.5-GhmiR160 targets were induced in transgenic lines.(A-D)qRT analysis of GhmiR160 target expression level.Error bars indicate standard deviation of four biological replicates.Bars with different letters are significantly different by Tukey's range test at P ＜0.05.10O,ovule at 10 DPA;10F,fiber at 10 DPA.M46 and M74 are miR160-suppressed lines.Null,nontransgenic plant segregating from miR160-suppressed lines.

4. Discussion

After miR160 was first identified in Arabidopsis, various authors [22-25,35,36,46-49] reported that the miR160/ARF axis functions in controlling leaf shape, floral organ development, root cap formation, adventitious root initiation,shoot regeneration in vitro, callus initiation, symbiotic nodule development, seed dormancy and leaf water loss.However, few reports have proposed a role of miR160 in modulating seed size and weight. The preferential expression of miR160 in cotton ovules before and at anthesis followed by its downregulation after anthesis suggested that miR160 plays a role in regulating early seed development(Fig. 4-A, B). To verify the role of miR160 in seed development, we tried to overexpress a target mimic of miR160 driven by the CaMV 35S promoter, but it was too difficult to get regenerated cotton seedlings.We accordingly used a FBP7 promoter to express the target mimic specifically in −2 to 10 DPA ovule outer integuments [37]. We obtained seven transformants, and the upregulated expression in M46 and M74 of four miR160-targeted genes in early development seed and fiber indicated that miR160 was suppressed in the transgenic lines (Figs. 1 and 5, S2). Two-year field-trial data showed that cotton seed size and weight were significantly decreased in miR160-suppressed lines (Table 1). Histology further showed that the size of cells from the seed coat was decreased in miR160-suppressed lines(Fig.3).We concluded that smaller cells lead to reduced seed-coat size, finally limiting embryo growth in suppressed lines. Scanning electron microscopy showed that early elongation of fiber cells from ovule epidermis was retarded in miR160-suppressed lines (Fig. 2). All of these results suggest that miR160 promotes cell expansion in cotton.

Fig.6- Identification of miR160 targets by RLM-RACE.Arrows indicate positions of target cleavage sites.Numbers next to arrows indicate cleavage frequency.

Fig. 7 - Detection of expression levels of seed and fiber development-associated genes in miR160-suppressed lines. (A-D) qRT analysis in 10 DPA ovule. (E) qRT analysis in 10 DPA fiber. Error bars indicate standard deviation of four biological replicates.Bars with different letters are significantly different by Tukey's range test at P ＜ 0.05 level. (F) 10 DPA ovules with fiber. M46 and M74 are miR160-suppressed lines. Null, nontransgenic plant segregating from miR160-suppressed lines.

Given that miR160-targeted genes such as ARF10, ARF16,and ARF17 are members of the auxin response factor family,we hypothesized that auxin signal transduction would be disrupted in miR160-suppressed lines.Surprisingly,there was little difference in auxin signal transduction between miR160-suppressed lines and null(Fig.S5).Perhaps the histochemical stain is not sufficiently sensitive to detect tiny differences of auxin signal. A more reasonable explanation is that miR160-targeted ARFs are not members of the ARF family involved in auxin signal transduction. miR160-targeted ARFs may regulate other signaling pathways. For instance, AtARF10 and AtARF16 control the expression of ABI3, which is a key ABA signal regulator that inhibits seed germination in Arabidopsis[46].In a previous study[50],exogenous ABA inhibited growth of in vitro ovules. We suspect that activating an ABA signal may account for the retarded growth of ovules in miR160-suppressed lines.

G hSus and orthologs of AtCYP78A6, AtGIF, and OsGL7 were downregulated in 10 DPA seeds of miR160-suppressed lines(Fig.7-A-D).The Sus gene encodes an enzyme that reversibly converts sucrose into fructose and UDP-glucose, and participates in cell wall formation and starch accumulation in plants [51]. Stressinduced increase in ABA can reduce its expression[51].In cotton,suppression and overexpression of GhSus genes showed a positive correlation with enzyme activity and seed growth and fiber elongation[52,53].AtCYP78A6 encodes an enzyme belonging to the cytochrome P450 family. With the close relative AtCYP78A9, it redundantly regulates seed size by generating an unknown signal molecule [14]. Besides enzymes, transcript coactivators encoded by AtGIFs physically interact with GRF transcript factors to promote cell expansion, given that overexpressing OsGIF in rice increases kernel weight and size in rice[12,54,55]. OsGL7 encodes a TON1 RECRUIT MOTIF-containing protein similar to the LONGIFOLIA2 protein in Arabidopsis,which is able to bind microtubules and could affect microtubule arrays[56]. OsGL7 increased cell expansion specifically in the kernellength direction to control kernel size [57]. The ortholog of AtEXP8, which belongs to the EXPANSIN family, was also downregulated in 10 DPA fibers of miR160-suppressed lines(Fig.7-E). The EXPANSIN gene family is known to control cell wall expansion, and suppression of the family member GbEXPATR dramatically reduces cotton fiber length[45,58].The promoter of GbEXPATR was downregulated by exogenous ABA[59].

We conclude that miR160 affects seed coat growth and further limits potential seed size by regulating the size of integument cells. miR160-targeted ARFs may activate ABA signals to directly or indirectly affect transcript factors,signal molecule generation, cell wall formation, or cell wall extension to regulate cell expansion and final cell size.The complex regulatory network of miR160/ARFs axis in controlling cell expansion invites further study.

Declaration of competing interest

The authors declare no conflict of interest.

Acknowledgments

This work was supported by the National Transgenic Plant Research of China(2016ZX0800-00-004).We thank Professor Yan Pei(Southwest University,China)for providing the FBP7 promoter and Professor Jian Xu(Department of Biological Science,National University of Singapore)for providing the DR5 promoter.

Author contributions

Nian Liu and Lili Tu: conceived and designed the research;Nian Liu and Shuchang Wu:constructed plasmids,performed genetic transformation, and measured seed and fiber traits;Nian Liu, Zhonghua Li, and Anam Qadir Khan: performed Southern and Northern blotting and qRT-PCR; Nian Liu and Haiyan Hu: performed histochemical analysis of GUS activity and observed fiber elongation by scanning electron microscopy; Nian Liu and Lili Tu: wrote the manuscript; Lili Tu and Xianlong Zhang:revised the manuscript.

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2019.12.004.