The Arabidopsis thaliana F-box gene HAWAIIAN SKIRT is a new player in the microRNA pathway

In Arabidopsis, the F-box HAWAIIAN SKIRT (HWS) protein is important for organ growth. Loss of function of HWS exhibits pleiotropic phenotypes including sepal fusion. To dissect the HWS role, we EMS-mutagenized hws-1 seeds and screened for mutations that suppress hws-1 associated phenotypes. We identified shs-2 and shs-3 (suppressor of hws-2 and 3) mutants in which the sepal fusion phenotype of hws-1 was suppressed. shs-2 and shs-3 (renamed hst-23/hws-1 and hst-24/hws-1) carry transition mutations that result in premature terminations in the plant homolog of Exportin-5 HASTY (HST), known to be important in miRNA biogenesis, function and transport. Genetic crosses between hws-1 and mutant lines for genes in the miRNA pathway also suppress the phenotypes associated with HWS loss of function, corroborating epistatic relations between the miRNA pathway genes and HWS. In agreement with these data, accumulation of miRNA is modified in HWS loss or gain of function mutants. Our data propose HWS as a new player in the miRNA pathway, important for plant growth.


Introduction
Selective degradation of proteins is carried out via the ubiquitin-proteasome pathway which is fundamental for many cellular processes, including development, hormonal signalling, abiotic stress and immunity in plants [1,2]. The abundance of key brakes and/or accelerators that control these processes is regulated by the 26S proteasome using complex mechanisms to avoid destruction of crucial proteins and the release of partially degraded polypeptides [2,3]. E1, E2 and E3 enzymes sequentially attach the small soluble protein ubiquitin to the proteins PLOS  DEGRADING NUCLEASE (SDN) exonucleases [47]. HASTY (HST), the plant homolog of Exportin-5 (Exp5), is involved in biogenesis or stability of some miRNAs and in transporting a yet to be identified component in the miRNA pathway [48]. The guide miRNA strand is merged into ARGONAUTE (AGO) proteins which carry out the post transcriptional gene silencing reactions (PTGS) [48][49].
In animals, regulation of miRNA biogenesis occurs at multiple levels. It occurs at the transcriptional level, during processing by Drosha (in the nucleus) and Dicer (in the cytoplasm), as well as by RNA editing, RNA methylation, urydylation, adenylation, AGO loading, RNA decay and by non-canonical pathways for miRNA biogenesis [50][51]. Although a vast amount of information has emerged relating to the biogenesis of miRNAs in plants, the mechanisms that modulate miRNAs and their generators in the canonical pathway, and/or the presence of non-canonical pathways are yet to be elucidated.
Here, we describe the identification and mapping of two hws-1 suppressor mutants (hst-23 and hst -24) in which the hws-1 sepal fusion phenotype is suppressed. These mutants are new mutant alleles of HASTY known to be involved in biogenesis or stability of some miRNAs and transporting of an unidentified component in the miRNA pathway. We demonstrate that mutation of HST as well as mutations of other genes in the miRNA biogenesis pathway and function are able to suppress hws phenotypes and vice versa. In agreement with these findings, the levels of miR163 and miR164 mature miRNAs in floral tissues are modified in lines that exhibit a loss or gain of function for HWS. The data support the hypothesis that HWS is a previously unidentified regulator of the miRNA pathway.
The hws-1 EMS mutagenized seeds were generated, screened and confirmed to be true suppressors by using specific primers to detect hws-1 mutation (S1 Table).
To identify the chromosome containing the shs-2 mutation, an AFLP-based genome-wide mapping strategy [53] was used on a subset of 40 DNA samples. Further mapping with all samples was performed with InDels [54]. For fine mapping, an additional 600 F2 plants were used. Once the region was narrowed down to a 59.4 Kb, candidate genes in the region were identified and a 6.927 Kb region of the HST gene was sequenced. A similar genomic region was amplified from the shs-3/hws-1 line for sequencing. Allelism tests between shs-2/hws-1 and shs-3/hws-1 were carried out by reciprocal crossing between the mutants. Primers used for mapping and sequencing are summarized in S1 Table. Phenotypic analyses The sepals and petals from twenty-five flowers (from six plants) from Col-0, hws-1, hst-24/hws-1 and hst-24 in Col-0 were carefully dissected, counted and photographed. Mature siliques and leaves dissected from 22 day-old plants from these lines were also recorded. Siliques from individual mutants and crosses between hws-1, hws-2, ddl-2, se-1, hyl-1, dcl1-9, hen1-5, hst-1 and ago1-37, were recorded following the same procedure.
All data obtained were used to perform statistical analyses and to create graphics. Regression analyses and ANOVA using generalized linear models were performed using GenStat 17.1. Graphics were created using Microsoft Excel 2016 and annotated using Adobe Photoshop 7.0.1.

miRNA Northern blots
Mature miRNAs were detected using the protocol described by [55]: total RNA was isolated from a cluster of buds and young flowers (up to stage 12, [56]) from Col-0, hws-1, and Pro 35 : HWS lines using TRIzol reagent (Life Technologies). Ten μg of total RNA from each line were used for northern hybridisation. Antisense probes were constructed using mirVana™ miRNA Probe Construction kit (Ambion) and radio labelled with γATP 32 P. Sequence information of probes is included in S1 Table. Yeast two-hybrid assay ProQuest™ yeast Two-hybrid system (Invitrogen) was used to study protein-protein interaction. The full length HWS coding region was cloned into pDEST32 and used to screen a stamen-specific tissue cDNA library [57]. Positive clones for Histidine bigger than 1mm in diameter were isolated and subjected to X-gal filter assays following manufacturer's instructions (Invitrogen). Plasmid DNA was isolated from selected individual clones, and then sequenced to identify the corresponding genes. To confirm the interaction, X-gal assays were repeated with the isolated clones.

Results
The mutants shs-2 and shs-3 are novel alleles of HASTY and suppress the sepal fusion phenotype of hws-1 To identify the substrate for the F-box HAWAIIAN SKIRT protein from Arabidopsis, we performed a suppressor screen by EMS-mutagenizing the hws-1 mutant in a Columbia-0 (Col-0) background. Screening of 308 individuals from 43 M2 populations resulted in the identification of two suppressor lines shs-2/hws-1 (suppressor of hws-2) and shs-3/hws-1 (suppressor of hws-3) that displayed no sepal fusion, suggesting suppression of the hws-1 phenotype (Fig 1I,  1J, 1K, 1M, 1Q, 1R, 1S and 1U). Reciprocal crosses between shs-2/hws-1 and shs-3/hws-1 yielded F1 individuals that displayed the same phenotype as the parents and restored the sepal fusion phenotype of hws-1 (S1 Fig) demonstrating that these suppressor mutations are allelic. The suppressor shs-2/hws-1 (in Col-0) was crossed to hws-5 (ffo-1, Landsberg erecta, Ler background) to generate a mapping population. The F1 individuals from this cross showed the sepal fusion phenotype suggesting that the mutant is recessive. The F2 population was then used for gene mapping. The shs-2 mutation was located in a 59.4 Kb region at the top of chromosome 3 ( Fig 1Y). This region contains 19 genes, including At3g05040 (HASTY-HST), a gene known to be involved in the export of mature miRNA molecules from the nucleus to the cytoplasm [48][49]. Analyses of the genomic region containing the HST gene in shs-2/hws-1 identified two transition mutations at positions 4.587 Kb and 5.517 Kb downstream from the ATG in shs-2/hws-1 line, resulting in a silent (ATC! ATT~Ile) and a premature termination (CAG! TAG; Gln !amber stop codon), respectively. In the shs-3/hws-1 line a transition mutation was located 0.583 Kb downstream of the ATG, introducing an earlier termination (GTG!GTA; Val!amber stop codon; Fig 1Y). Consequently, the shs-2 and shs-3 mutants were renamed hst-23 and hst-24. These mutations generate truncated versions of HST of 924 and 57 amino acids respectively, compared to the wild type HST protein consisting of 1202 aa. The double mutants hst-23/hws-1 and hst-24/hws-1 were back-crossed with Col-0 to obtain hst-23 and hst-24 single mutants for subsequent analyses (Fig 1D, 1F, 1L, 1N, 1T and 1V). The F2 progenies displayed a segregation ratio 3:1 confirming that these are single, recessive nuclear mutations. The hst-23 allele displayed relatively more severe floral and vegetative phenotypes compared to hst-24 allele (Fig 1 and S1 Fig).
To confirm that mutation of HST is responsible for the suppression of hws phenotype, we crossed hws-1 with hst-1, an independent mutant that harbours a mutation in the HST coding region that generates a truncated protein of 521 amino acids with the last 18 aa differing from the wild type protein [58]. As shown in Fig 1G, 1O and 1W, flowers from F2 individuals displayed no sepal fusion, thus corroborating that mutation in HST is able to suppress the phenotype of hws-1. Taken together these data demonstrate that mutations in HST suppress the hws phenotype, thus suggesting a putative role of HWS function in miRNA transport pathway.
HWS has a role in the miRNA pathway HST is the Arabidopsis orthologue of Exp-5 from mammals, a protein involved in small RNAs export from the nucleus to the cytoplasm [48]. We previously showed that overexpression of HWS (Pro 35 :HWS) leads to phenotypes resembling those of mutants in miRNA pathway. This knowledge together with the fact that the HWS loss of function phenotype is suppressed by mutation in HST, prompted us to address if the HWS plays a role in miRNA biogenesis and function.
F2 plants were isolated and the double mutants identified by PCR. The genetic interactions showed that all tested miRNA biogenesis and function pathway mutants, were able to suppress the sepal fusion phenotype in the hws-1 and hws-2 independent mutants (Fig 2A, 2B, 2C, 2D, 2E, 2F, 2G, 2H, 2I, 2J, 2K, 2L, 2M, 2N, 2O, 2P, 2Q, 2R, 2S, 2T and 2U) the hws-2 allele harbour two T-DNAs inserted in opposite directions 465 and 491 bp downstream the ATG of HWS [15]. Interestingly, the hws mutants were also able to suppress the phenotypes of these mutants in some instances. It is particularly noticeable that the hws mutant was able to suppress the delayed or arrested development from siliques of the mutants ddl-2 (Fig 2A, 2B and 2C), dcl1-9 ( Fig 2J, 2K and 2L) and hen1-5 (Fig 2M, 2N and 2O). It should be noted that in older plants, towards the end of the production of siliques, the reciprocal suppression of phenotypes between hws and the biogenesis pathways mutants was less apparent (data shown for hws-1/ ddl-2; Fig 2C). These data support the proposal that HWS is an important regulator in the miRNA pathway.
To further address this conclusion, we evaluated the levels of mature miRNAs from MIR163 and MIR164 in developing flower buds, up to stage12 [56]. Compared to the Col-0, significant over-accumulation of miR163 and miR164 was observed in the hws-1 mutant, while reduction was observed in the Pro 35 :HWS line. (Fig 3). These results support our hypothesis that HWS regulates levels of miRNAs in flowers, and likely in other tissues where the HWS gene is expressed.
The HWS protein contains an F-box and a Kelch-2 repeat in its C-terminus [15]. Fbox proteins are important elements of the E3 SCF complex (from SKP1, Culling and F-box) that catalyse the ubiquitination of proteins to be degraded by the proteasome [62]. It is therefore likely that HWS forms a part of an SCF complex and identifies for targeted degradation protein (s) that are in the miRNA pathway. We performed a yeast-two hybrid screen using a cDNA library generated from stamen tissue from Arabidopsis flowers. A total of 1,280,000 clones were screened. From these, 66 histidine positive colonies were isolated. X-gal assays showed that among the 66 histidine positive colonies, 56 were positive for X-gal. From the 56 X-gal positive clones, 55 contained Arabidopsis SKP1 protein; among which, 36 contained only SKP1; 10 contained both SKP1 and PRXR1 (a protein involved in catabolism of hydrogen peroxide), and 9 contained SKP1 and FLA3 (Fasciclin-like arabinogalactan protein 3 precursor). One of the clones contained only SKP4. However, independent X-gal assays could only confirm the interactions between HWS and SKP1 or SKP4, suggesting that the isolated clones may not interact directly with HWS or alternatively interaction of HWS with other proteins require the presence of SKP1 (S2 Fig). These results confirm that the F-box protein HWS is part of an SCF complex likely targeting for degradation protein(s) involved in the miRNA pathway.

hws-1 and hst mutants reveal epistatic interactions and independent roles of HWS and HST during plant development
Previously, it was reported that mutation of HST induces pleiotropic effects during plant development, which include curling of leaf blades, reduction of leaf numbers, faster production of abaxial trichomes, reduction of leaf, sepals and petals size, laterally expanded stigma, inflorescence phyllotaxy defects and reduced fertility [58,[63][64]. We show here that mutations in HST are able to suppress the sepal fusion of hws-1.
To understand the biological role of HWS-HST interaction and its role in nuclear export, we addressed if HWS also affects the phenotypic variations associated with hst mutants, we performed phenotypic analyses in simple and double mutant lines hws-1, hst-1 and hst-24/ hws-1, hst-23/hws-1. Indeed, a reciprocal complementation of hst phenotypes by mutating HWS was observed when analysing hst-23/hws-1 and hst-24/hws-1 double mutants. Mutation of HWS (hws-1) was able to suppress phenotypes associated with hst mutations, such as the curling up of the leaf blades, the reduction of leaf numbers, the reduction of silique dimensions and fertility, the reduction of the expansion of stigmas and the disorientation of petals (Figs 1Q, 1R, 1S, 1T, 1U, 1V, 1W, 1X, 4D and 4E and S1 Fig). These results are in agreement with the data above and corroborate that HWS acts in the miRNA pathway.
However, mutation of HWS could not supress other phenotypes associated with the hst mutation. Sepals and petals from hst-24 were reduced in size compared to that of Col-0 andhws-1 (Fig 4B). Sepals and petals of double mutant hst-24/hws-1 were comparable in size to the ones from the hst-24 single mutant demonstrating that loss of function of HWS was not able to supress the reduced petal size phenotype associated with the hst mutation ( Fig 4B). This observation suggests that HST must perform other functions independently of HWS.
Phenotypic analyses of flower organ number in hst-24 mutant showed the characteristic four sepals and four petals (Fig 4C and Table 1). However, a statistically significant (p<0.0001) increase of sepals and petals number of 10% was observed in the double mutant hst-24/hws-1 (Figs 1E, 4A, 4B and 4C and Table 1). Interestingly, the increments were only observed in the first ten flowers of each plant analysed, the subsequent fifteen flowers analysed displayed floral organ number comparable to the wild type. Approximately 58% of the flowers had an increase of both sepals and petals within a single flower. Taken together these data suggest that HWS interacts with HST in the miRNA pathway to control some biological functions, but must also act in an independent pathway to control others.

Discussion
Although plenty of knowledge has been generated since the discovery of the first miRNAs in 1993 [65][66][67], the complexity of mechanisms regulating their biogenesis, expression and mode HWS, a miRNA pathway player of action is not fully elucidated. Here we demonstrate a role for HWS in the miRNA pathway. Our first line of evidence comes from the isolation of two new HST alleles, hst-23 and hst-24, from a screening of EMS hws-1 mutant suppressor lines. These alleles were able to suppress the sepal fusion phenotype from hws-1. HST has been implicated in the export of an unidentified component of the miRNA pathway, miRNA biogenesis or miRNA function [48]. Our second line of evidence comes from our genetic crosses between hws-1 or hws-2 and ddl-2, se-1, hyl-1, dcl1-9, hen1-5, hst-1 and ago1-37 mutants from known genes regulating the biogenesis and function of miRNAs, that show suppression of the sepal fusion from hws-1, demonstrating that HWS has a role in biogenesis, stability and/or function of miRNA in addition to their transport involving HST. Interestingly, there was a noticeable reciprocal suppression of phenotypes between the hws and ddl-2, dcl1-9 and hen1-5 mutants in floral development, fertility and flower morphology, suggesting epistatic interactions. Suppression of phenotypes towards the end of flower production was less apparent, suggesting that the regulatory mechanisms becomes altered in a spatiotemporal way, or that HWS is targeting for degradation a yet to be identified protein that regulates genes of the miRNA pathway in a spatiotemporal fashion upstream of the miRNA biogenesis process. Alternatively, a compensatory mechanism to regulate microRNA biogenesis could be present; in agreement with this hypothesis, it has been previously demonstrated that such mechanisms exist to compensate cell number and associated organ sizes defects in plants [68]. Our third line of evidence comes from our Northern blot analyses where differential accumulation of mature miR163 and miR164 in floral tissues in the hws-1 mutant and the Pro 35 :HWS line were observed, suggesting that during development a differential regulation of mature miRNAs is required, and this is achieved by a pathway implicating HWS. It is known that miR163 negatively regulates mRNA levels of PMXT1, a member of the S-adenosyl-Met dependent carboxyl methyltransferase family, to modulate seed germination, seedling de-etiolation and root architecture in response to light [69]. While miR164 negatively regulates mRNA levels of CUC1 and CUC2 genes to modulate boundary formation in flowers [14,[17][18]. Our Northern blot results provide further evidence for a role of HWS in miRNA pathway and suggest that the sepal fusion phenotype observed in hws-1 maybe due to the over accumulation of miR164 which in turn modulates mRNA levels of CUC1, and CUC2.
Our data point to the hypothesis that putative target proteins of HWS, act upstream of the miRNA biogenesis pathway, or affect miRNA stability or function, or a combination of all of these. The HWS protein holds an F-box and a Kelch-2 repeat in its C-terminus [15]. It is likely that the interaction between HWS and its targets involves the Kelch-2 repeat. In agreement with this proposal, in our yeast-two-hybrid experiments we were able to demonstrate that HWS interacts with ASK1 and ASK4, two proteins that are part of the SCF complex, supporting the idea that HWS role in the miRNA pathway may be by targeting proteins for degradation through the SCF complex. Although these targets remain to be identified, putative candidates could be PROTEIN PHOSPHATASE 4 (PP4), SUPPRESOR OF MEK1 (SMEK1) [42], REGULATOR OF CBF GENE EXPRESSION (RCF3) or C-TERMINAL DOMAIN PHOSPHATASE-LIKE1 AND 2 (CPL1 and CPL2), that are known to be involved in controlling the phosphorylation status of HYL-1 to promote miRNA biogenesis [43]. Alternatively, the CAP-BINDING PROTEINS 20 and 80 (CBP20 and CBP80, also known as ABH1), important proteins during the biogenesis of miRNAs and ta-siRNA biogenesis [70]. It has been demonstrated that ABH1 (CBP80) is also able to suppress the hws-1 sepal fusion phenotype [71]. Therefore, CPB20 and CBP80 are strong candidates for targeted degradation through HWS. In the literature, some redundancy and cross-talk between known pathways generating miRNAs, ta-siRNAs and siRNAs, and other pathways that remain to be discovered, has been reported [72]. The role of HWS in the regulatory events during ta-siRNAs and siRNAs biogenesis pathways, among others, remains to be elucidated. Testing interactions of these proteins will shed light of the putative role of HWS in controlling the phosphorylation status of key players in the miRNA pathway.
It has been suggested that the AUXIN SIGNALING F-BOX 2 (AFB2) gene is post-transcriptionally negatively regulated by miR393, and a regulatory mechanism where miRNAs prevent undesired expression of genes involved in miRNA production has been proposed [73]. An alternative to this suggestion comes from the finding of numerous siRNAs in the proximity of the MIR393 target site for the F-boxes TIR1, AFB2, and AFB3 genes [74]. [74] suggested that the regulation of their transcripts occurs via siRNAs rather than MIR393. Further experiments will establish if this regulatory mechanism holds true for HWS.
We revealed that the hws-1 is able to suppress the curling up of leaf blades, reduction of leaf numbers, reduction in leaf size, expansion of stigma, petal orientation, and reduced fertility phenotypes characteristic of hst mutants [58,[63][64]. However, HWS and HST seem to also have independent roles as the hws mutation could not supress some phenotypes associated with the hst knockout. Moreover, the double mutant hst/hws exhibited increased sepals and petal number in the first ten formed flowers, a phenotype not seen in the hst-24 or hws-1 single mutants. The underlying mechanisms of the increased number of sepals and petals in the double mutant remain unknown. It has been reported that HST affects bolting and floral maturation timing [63], but there are no reports of HST affecting floral organ numbers. These findings suggest epistatic interactions between HWS and HST to fine tune development in plants, in a spatiotemporal way, in addition to independent roles for HWS and HST in plant development.
Previous findings point to the fact that genes involved in the miRNA pathway must have other roles in addition to miRNA biogenesis, transport or function. For example, ddl mutants have more severe morphological phenotypes than these of the dcl1-9 mutants; but the miRNA levels are reduced in the dcl1-9 compared to the ddl mutants [35]. Moreover, it has been demonstrated that DDL regulates plant immunity by poly(ADP-ribosyl)ation (PARylation) of proteins; and regulates plant development via the miRNA biogenesis pathway [75]. Another example is illustrated by CBP20 and CBP 80. It has been demonstrated that in addition to their role in miRNA biogenesis these proteins also act during the formation of a heterodimeric complex that binds the 5' cap structure of a newly formed mRNA by Pol II, aid in the pre-miRNA splicing and act during polyadenylation and during the export of RNA out of the nucleus [70,[76][77][78][79][80]. Therefore, it is likely that both HWS and HST have additional roles to that of miRNA pathway.
Our data shed light on the complexity of mechanisms regulating miRNA pathway, and place HWS as a new regulator in this pathway. In support of our findings, [71] have proposed HWS as a regulator of miRNA function in their screening studies for negative regulators of MIR156 activity.
Due to the impact on development that HWS exerts, this research is relevant for identifying novel strategies to generate more productive and resilient crops. As support to this, recently we showed that a mutant from the ERECTA PANICLE3, the HWS rice orthologue gene in rice, has decreased photosynthesis due to reduced stomatal conductance and attenuated guard cell development [81]. Moreover, [82], demonstrated that Arabidopsis mutants and a knock down line of OsFBK1, a second HWS rice orthologue gene, germinate better and have root systems that are more robust on exposure to ABA than wild type, important for drought tolerance.  Table. Primers and probes used in this study. Marker, sequencing, screening, yeast-twohybrid primers and probes used in Northern blots are included. (DOCX)