The Gene ANTHER DEHISCENCE REPRESSOR (ADR) Controls Male Fertility by Suppressing the ROS Accumulation and Anther Cell Wall Thickening in Arabidopsis

Male sterility in plants is caused by various stimuli such as hormone changes, stress, cytoplasmic alterations and nuclear gene mutations. The gene ANTHER DEHISCENCE REPRESSOR (ADR), which is involved in regulating male sterility in Arabidopsis, was functionally analyzed in this study. In ADR::GUS flowers, strong GUS activity was detected in the anthers of young flower buds but was low in mature flowers. ADR + GFP fusion proteins, which can be modified by N-myristoylation, were targeted to peroxisomes. Ectopic expression of ADR in transgenic Arabidopsis plants resulted in male sterility due to anther indehiscence. The defect in anther dehiscence in 35S::ADR flowers is due to the reduction of ROS accumulation, alteration of the secondary thickening in the anther endothecium and suppression of the expression of NST1 and NST2, which are required for anther dehiscence through regulation of secondary wall thickening in anther endothecial cells. This defect could be rescued by external application of hydrogen peroxide (H2O2). These results demonstrated that ADR must be N-myristoylated and targeted to the peroxisome during the early stages of flower development to negatively regulate anther dehiscence by suppressing ROS accumulation and NST1/NST2 expression.


Introduction
Anther dehiscence is an important process in mature stamens. In this process, lignin accumulation in the endothecium of anthers enables secondary wall thickening. Subsequently, septum and stomium lysis completes dehiscence 1-6 .
The expansion of the endothecium provides an internal directed force for anther dehiscence, causing breakdown of the stomium. Then, desiccation of the epidermis causes differential shrinkage of thickened and unthickened parts of the cell wall, resulting in an outwardly bending force that leads to the retraction of the anther wall and the complete opening of the stomium 7-9 . Lignin is the major compound involved in secondary wall thickening in anthers, and its polymerization is dependent on hydrogen peroxide (H 2 O 2 ) levels 10 which is the major ROS form in plant cells and the substrate for peroxidase in catalyzing lignin polymerization [11][12][13][14] . The final polymerization steps of lignin biosynthesis occur after the activation of monolignols to free radicals, which is mediated by peroxidase + H 2 O 2 and/or oxidase (laccase) + O 2 , followed by non-enzymatic coupling of monolignol radicals to form the polymer lignin [11][12][13][14][15] . Interestingly, H 2 O 2 also plays an important role as a signal in activating transcription of lignin biosynthesis enzyme such as peroxidases 16 . It has been reported that catalases (CATs) are the other major H 2 O 2 -scavenging enzymes which are located in peroxisomes 17 . In a previous study, the NAC transcription factors NAC SECONDARY WALL THICKENING PROMOTING FACTOR1 (NST1) and NST2 were reported to regulate lignin synthesis of anther secondary walls 18 . Double mutations in NST1 and NST2 caused sterility due to anther indehiscence.
Although it is well established that the H 2 O 2 level in the anther endothecium is strongly correlated with the lignin polymerization and anther dehiscence, the mechanisms regulate H 2 O 2 level in the anther still remain to be investigated. In cells, the H 2 O 2 can be accumulated in organelles such as peroxisomes, which are single membrane-bound organelles with diverse metabolic functions 19, 20 . In association with ROS production, peroxisome targeted PAO (polyamine oxidase) has been reported to regulate pollen tube elongation 21 . A mutation in the peroxisomal membrane protein DAYU impairs pollen maturation and germination 22 . However, the association between anther dehiscence and peroxisomes remains unclear. Therefore, it is interesting to explore the mechanisms or factors that regulate or affect H 2 O 2 accumulation in peroxisomes.
To explore the association between H 2 O 2 -regulated anther dehiscence and peroxisomes, a novel ANTHER DEHISCENCE REPRESSOR (ADR) gene (At4g13540), which is potentially N-myristoylated and targeted to peroxisomes, was cloned from Arabidopsis and analyzed. The ADR protein contains a predicted conserved recognition site (GGSTSKD) for N-myristoylase 23 at the N-terminus of the protein. It has been reported that the N-terminal octapeptide of ADR can be myristoylated by AtNMTs (N-myristoyltransferase) 23 . ADR also contains a binding site for a peroxisomal targeting signal (PTS) in the middle of the protein, which indicates that ADR is likely targeted to the peroxisomes. The PTS binding site is critical for protein targeting and binding to peroxisomal matrix proteins (Pex5 and Pex7) or peroxisomal membrane proteins (Pex19) and thus allows the peroxisome entry or peroxisomal membrane association 24 . N-myristoylation involves the addition of the saturated C:14 fatty acid myristate to the N-terminus of proteins and affects the membrane binding properties of proteins 23, 25 . A mutation in the myristoylation domain did not interfere with peroxisomal targeting but disrupted the membrane association of proteins because it prevented the addition of the myristoyl group that is also essential for membrane association. Based on these observations, proteins lacking a myristoyl group can still bind to Pex through the PTS binding site and target to the peroxisome, but they cannot stably associate with the peroxisomal membrane. In this study, we demonstrated that ADR proteins are likely modified by N-myristoylation and targeted to peroxisomes. We showed that ectopically expressing ADR causes male sterility of the flowers due to anther indehiscence.
We also found that ADR functions to reduce ROS accumulation and suppresses the expression of NST1 and NST2. Thus, we propose a model in which ADR is myristoylated and negatively regulates anther dehiscence by promoting ROS scavenging in the peroxisome, which affect lignin polymerization and stomium rupture in Arabidopsis.  identity and 78% similarity to the most closely related ADR-like protein, At3g23930 (Fig. S1). In their N-myristoylase sites, 95% of the amino acids are identical (Fig. S1).

RT-PCR analysis of ADR transcripts and detection of ADR expression by analyzing ADR::GUS transgenic Arabidopsis plants
Reverse transcription PCR (RT-PCR) was performed to determine the relative transcript abundance of ADR at different developmental stages and in various 6 organs of Arabidopsis. ADR expression was not detected in early seedling development (Fig. 1A). The transcript level of ADR was strongly detected in flowers and weakly detected in the roots, stem and siliques, but ADR transcripts were absent in the leaves of mature plants (Fig. 1A). When the expression of ADR in flowers at different developmental stages was further analyzed, significantly higher expression of ADR was observed in early development stages (stages 8-11) than in late flower development stages (after stage 12; Fig. 1B).
To further analyze the expression pattern of the ADR gene in flowers, a construct (ADR::GUS) was generated and transformed into Arabidopsis; twelve independent ADR::GUS plants were obtained. GUS activity in the ADR::GUS flowers was strongly detected in sepals but was relatively weakly detected in petals and carpels during early and late flower development (Fig. 1C, E). In the stamen, GUS activity was strongly detected in anthers during early flower development stages (before stage 10; Fig. 1C, D), but its expression was almost undetectable in anthers during late developmental stages (Fig. 1E, F).

ADR needs to targeted to peroxisomes to perform its function
It has been shown that the N-terminus of ADR can be myristoylated by an in vitro myristoylation assay after the first eight residues of the N-terminal peptide sequence recognized by N-myristoyltransferase 23 . Because N-myristoylation is known to affect the membrane-binding properties of proteins 25 , an proteins mainly accumulated in organelle-like structures ( Fig. 2A, B) which may be the peroxisomes since ADR contains a binding site for a peroxisomal targeting signal (PTS) as described above (Fig. S1). To examine that the organelle-like structures were peroxisomes, a construct of CATALASE 3 (CAT3) fused to mORANGE2 (CAT3+mORG2) was co-transformed with ADR+GFP into Agrobacterium and infiltrated into the leaf epidermis of N. benthamiana. It has been shown that CAT3 could localize to the peroxisomes 27 (Fig. 2C, D).
Both ADR+GFP and CAT3+mORG2 were also detected in the nucleus ( Fig. 2A, C, E). Because it has been reported that the maximum protein size that can diffuse freely through the nuclear pore is larger than 60 kDa 28 , the diffusion of these two proteins (both are smaller than 60kD) into the nucleus may be due to the over-expression of high amounts of the proteins in these cells.
It has been reported that some proteins containing MTS (mitochondria targeting sequence) are targeted to the mitochondria after N-myristoylation 29 . To further confirm the localization of ADR is not in other organelles such as mitochondria, a construct of the mitochondria marker MT fused to mCherry (MT+RK) was co-transformed with ADR+GFP into Agrobacterium and infiltrated into the leaf epidermis of N. benthamiana. Confocal images showed that ADR+GFP (Fig. 3A, B, E, F) did not co-localize with MT+RK ( Fig. 3C, D, E, F), indicating that myristoylated ADR is not associated with mitochondria.  (Fig. 4A, B, C). The severity of the sterile phenotype in the 35S::ADR plants was correlated with the ADR expression level (Fig. 4C, D).

Ectopic expression of ADR causes plant sterility due to anther indehiscence
When the 35S::ADR flowers (Fig. 4E) were examined, they were similar to wild-type flowers (Fig. 4F), opening normally and producing normal sepals, petals and carpels with fully developed stigmatic papillae. The anthers of the 35S::ADR flowers were indehiscent at all stages of flower development (Fig. 4E,   G). In contrast, wild-type anthers were dehiscent, and pollen released after stage 12 of flower development (Fig. 4F, H). The 35S::ADR flowers with severe phenotypes were sterile and unable to set seed due to the indehiscence of anthers throughout flower development (Fig. 4C). To further examine pollen viability, Alexander's stain, which can distinguish viable pollen from nonviable pollen 30 , was applied. Normal viability (dark blue staining), similar to that of wild-type pollen (Fig. 5A), was observed in the 35S::ADR pollen (Fig. 5B) Furthermore, the anthers in these ADR mutants were dehiscent normally ( Fig.   S2C, D) similar to that in wild-type plants. This finding indicates a possible functional redundancy between ADR and other unknown genes. Further analysis indicated that the expression of ADR in these T-DNA insertion mutants was significantly reduced (Fig. S2E).

Ectopic expression of ADR-Gly resulted in normal fertility
To confirm the myristoylated ADR was associated with the sterile phenotype were significantly down-regulated in the flowers of 35S::ADR plants (Fig. 6E).
This result indicates that altered anther dehiscence in 35S::ADR plants is correlated with the altered expression of NST1/2 that participate in the regulation of secondary wall thickening in anthers.

ROS accumulation was lower in 35S::ADR anthers than in wild-type anthers
It has been known that the lignification of the secondary wall thickening in  (Fig. 6Q). In contrast, silique elongation was not observed in H 2 O 2 -untreated mock flowers throughout the flower development ( Fig. 6O, P).
This result confirmed that the indehiscent anther phenotype in 35S::ADR flowers is due to the reduction of the ROS accumulation and can be complimented by exogenous H 2 O 2 .

Discussion
In this study, the ADR gene in Arabidopsis was functionally analyzed. Ectopic The N-terminus of ADR is presumed to be myristoylated, and this myristoylation process is critical for its membrane association 23 . It has been reported that myristoylation alone is not sufficient to anchor a protein stably to a membrane; the N-terminal basic residues contribute to protein membrane association via electrostatic interactions with acidic phospholipids 26 . The identification of several basic residues in the N-terminus and a binding site for a peroxisomal targeting signal (PTS) in the middle of ADR supports the idea that it has a role in controlling anther dehiscence as a membrane-associated protein in the peroxisome. Clearly, this hypothesis was supported by the results of the transient expression experiment in which the ADR protein were associated with peroxisomes. These results support the assumption that myristoylation of ADR and its presence in the peroxisomes are important for its function to prevent anther dehiscence during the early stages of floral development (Fig. 7).
It is interesting to explore the exact role of ADR in the negative regulation of  (Fig. 7). Following lignin polymerization, excess H 2 O 2 can diffuse into the cytoplasm and regulate the expression of nuclear genes (Fig. 7), a process called retrograde signaling 34-36 .
Thus In conclusion, we demonstrated that myristoylated ADR associates with the peroxisome and negatively regulates anther dehiscence by promoting ROS scavenging, ultimately affecting lignin polymerization and stomium rupture in Arabidopsis.

Plant materials and growth conditions
The T-DNA insertion mutants of ADR (SALK_072305) Arabidopsis seeds were   Supplementary Table S1.  Table S1. Mitochondria fusion binary plasmid (CD3-991) which contained mitochondria marker fused with mCherry fluorescent protein was obtained from the ABRC (clone name: MT-RK) 40 .

Plant transformation and transgenic plant analysis
Constructs in this study were introduced into Agrobacterium tumefaciens strain GV3101 and transformed into Arabidopsis plants using the floral dip method as described elsewhere 42 . Transformants that survived in the medium containing kanamycin (50 μg ml -1 ) were further verified by PCR and RT-PCR analyses.

Histochemical GUS assay
Histochemical staining was performed under the standard method described previously 43, 44 .

Alexander's staining
For pollen analysis, pollen grains were mounted with Alexander's stain as previously described 30 .