Constitutive Expresser of Pathogenesis Related Genes 1 Is Required for Pavement Cell Morphogenesis in Arabidopsis

For over 50 years, researchers have focused on the mechanisms underlying the important roles of the cytoskeleton in controlling the cell growth direction and cell expansion. In our study, we performed ethyl methane sulfonate mutagenesis on Col-0 background and identified two new CONSTITUTIVE EXPRESSER OF PATHOGENESIS RELATED GENES 1 (CPR1) alleles with pavement cell (PC) morphogenetic defects. Morphological characterizations showed that polar growth initiation and expansion of PCs are seriously suppressed in cpr1. Closer cytoskeleton investigation showed that the directional arrangement of microtubules (MTs) during PC development is defective and the cortical fine actin filaments cannot be aggregated effectively to form actin cable networks in cpr1 mutants. These results suggest that the abnormal PC morphogenesis in cpr1 is accompanying with the aberrant arrangement of cytoskeleton. Site-directed mutagenesis and knockout within the F-box-associated (FBA) domain, which is reported to be a motif for recognizing particular substrates of CPR1, proved that the FBA domain is indispensable for normal CPR1 regulation of the PC morphogenesis. Further genetic analysis indicated that the defects on PC morphogenesis of cpr1 depend on two lipase-like proteins, ENHANCED DISEASE SUSCEPTIBILITY 1 and PHYTOALEXIN DEFICIENT 4. Our results provide further insights into the relationship between the cytoskeleton and PC morphogenesis, and suggest that the cytoskeleton-mediated PC morphogenesis control might be tightly linked to plant defense responses.

Genetic mapping of the mutant gene cpr1-j594 and cpr1-j2928 (Columbia background) were backcrossed to Ler Arabidopsis plants to establish a mapping population. The F2 progenies with mutant phenotype were used for map-based cloning and DNA from homozygote individuals was isolated for molecular mapping. A set of simple sequence length polymorphism (SSLP) and cleaved amplified polymorphic sequence (CAPS) markers were used to map the mutant gene. The markers were obtained from the Arabidopsis mapping platform (AMP) [26].

PC morphometric analysis
PCs from apical sectors of mature leaves were used for quantitative analysis of PC area, circularity, lobe length, and lobe number [11,27]. The images were collected using a HITACHI S-3400N scanning electron microscope. The equation used for circularity calculation is as follows: 4π area/perimeter 2 [27]. For each genotype, approximately 100 highly expanded cells, from at least ten independent leaves were measured.

MT and MF visualization using confocal microscopy
Both cpr1-j594 and cpr1-j2928 were crossed to transgenic plants expressing GFP-TUA6 and GFP-FABD2, respectively. F2 plants with steady green fluorescent protein were used for MT and MF analysis. Images were collected using a confocal microscope (LeicaTCS SP8).

Developmental characterization of PCs
Approximately 0.5 cm long wild-type abaxial leaf surface was observed by scanning electron microscope (SEM). Based on the positions of cells along the long axis of the leaf, the development of PCs was divided into three stages. Stage I, cells were localized in the base of the leaf. Stage II, cells were localized around the midpoint of the leaf. Stage III, cells were localized in the tip of the leaf. The different developmental stages of PCs were identified in cpr1-j594 and cpr1-j2928 plants based on their leaf positions with respect to those in wild-type leaves [9]. The images were collected using a HITACHI S-3400N SEM.

Plasmid construction and generation of transgenic plants
Gateway technology was employed for most genetic manipulations. For the complementation test, 3.5 kb of the genomic sequence including the full CPR1 genomic sequence and its 1.9 kb upstream sequence was amplified, and cloned into the pBIB-BASTA-GWR-GFP vector by in vitro DNA recombination (see primers in supplementary material S1 Table). For the overexpression construct, the CDS sequence of CPR1, CPR1I247V, and CPR1ΔFBA were amplified, and introduced into the destination vector pB35GWF (see primers in supplementary material S1 Table). All the cloned sequences were confirmed by sequencing analysis.

j594 and j2928 show severe PC morphogenetic defects
To identify more components regulating PC morphogenesis, we performed an ethyl methane sulfonate mutagenesis in Col-0 background seeds and screened more than 3000 independent lines. Two mutants, j594 and j2928, were identified. Compared with wild-type plants, j594 and j2928 developed petite seedlings (Fig 1A), narrower leaves (Fig 1C), and shorter shoots ( Fig 1B) and silique clusters (Fig 1D). Closer scrutiny by SEM showed abnormal PC shape in j594 and j2928 (Fig 2A-2C). Quantitative PC analysis of WT, j594 and j2928 using the geometric analysis method [12,27,28] showed that PCs in j594 and j2928 were smaller and exhibited less and shorter lobes (Fig 2E, 2G and 2H). Circularity has been reported as a quantitative descriptor of PC shape complexity [27], thus the circularity of PC in WT, j594 and j2928 was assayed. The results showed that PCs of j594 and j2928 have larger values of circularity compared with wild type (Fig 2F). These results implied that the PCs in j594 and j2928 exhibit severe morphogenetic defects.
Lobe initiation and outgrowth are both inhibited during PC development in j594 and j2928 Previous reports have divided PC development into three stages [9]. To investigate the mechanisms of cell shape formation in j594 and j2928 mutants, we recorded PC shape during the different developmental stages in each genotype. During Stage I, j594 and j2928 PCs formed During Stage III, the multiple shallow lobes fully extended formed a complex shape in the wild-type cells. In contrast, in j594 and j2928 PCs, few newly formed polar growth sites were observed and polar extension in this direction was also greatly inhibited (Fig 3, bottom panels). All these results indicated that lobe initiation and outgrowth are both inhibited during PC development in j594 and j2928.
j594 and j2928 are two new alleles of CPR1 To identify the map position of the molecular lesions in these two mutants, map-based cloning was performed. After fine mapping, both of the mutations were located in a 90 kb region on chromosome 4 ( Fig 4A) and all the genes in this region were sequenced. Interestingly, we found that both of the mutants have a point mutation in the same gene At4g12560, which encodes the F-box protein CPR1 [20]. In j594, a G to A mutation at the nucleotide position 181 converted a glycine to an arginine. In j2928, a mutation located in the last base of its only intron, lead to the abnormal splice of CPR1 ( Fig 4B). To determine whether j594 and j2928 were loss-of-function mutants of CPR1, a line with a T-DNA insertion in the 3'-untranslated region of CPR1 (SALK_045148) was obtained and characterized. The SALK_045148 line exhibited similar phenotypes to j594 and j2928 (Fig 1 and Fig 2D). When j594 and j2928 were backcross to Col-0, their F1 progenies showed no obvious phenotype (data not shown), indicating that j594 and j2928 are recessive mutations of a single gene. They were also crossed to SALK_045148 plants and the F1 progenies from both crosses showed mutant phenotypes ( Fig  4C and 4D). These results supported that j594 and j2928 are allelic to CPR1. To further confirm this result, the native promoter of CPR1 fused to its full length gene was transformed into j594 and j2928, and all the transgenic lines rescued plant and PC morphogenetic mutant phenotypes of j594 and j2928 ( Fig 5). Based on these results, we concluded that the phenotype of j594 and j2928 is due to the loss function of CPR1. Therefore, we renamed them cpr1-j594 and cpr1-j2928, respectively.

The FBA domain is indispensable for normal CPR1 function
To investigate the function of the FBA domain in CPR1, two CPR1 overexpression constructs with mutations in the FBA domain were used. The first construct had a point mutation (35S:: CPR1I247V) and the second lacked the FBA domain (35S::CPR1ΔFBA). Both constructs were expressed under a cauliflower mosaic virus 35S (CaMV35S) promoter in Col-0 background. Transgenic lines of both 35S::CPR1I247V and 35S::CPR1ΔFBA exhibited similar phenotypes to those observed in cpr1-j594 and cpr1-j2928 (Fig 6A and 6B). Quantitative PC shape analysis showed that PCs in 35S::CPR1I247V and 35S::CPR1ΔFBA lines are small, have short and reduced lobes, and large values of circularity (Fig 6C-6F). However, plants overexpressing the normal CPR1 exhibited phenotypes similar to those observed in wild-type plants and PCs ( Fig  6A and 6B). Meanwhile, expression of exogenous CPR1 did not affect the transcription level of endogenous CPR1 (S1 Fig). Thus, we concluded that 35S::CPR1I247V and 35S::CPR1ΔFBA could imitate the phenotypes of cpr1 loss-of-function mutants, suggesting that the overexpression of these mutated CPR1 has impact on the normal function of endogenous CPR1.

The MT cytoskeleton is defective during PC morphogenesis in cpr1 mutants
Microtubules play an important role in maintaining PC morphogenesis [8,29]. Thus, we wondered whether the microtubule cytoskeleton was changed in cpr1-j594 and cpr1-j2928. The MT cytoskeleton in Col-0 and two cpr1 mutants was detected using the labeled marker TUA6-GFP. MT arrangement formed simpler cortical MT networks with relatively single oriented arrays in the mature PCs of the two cpr1 mutants than in wild-type PCs (Fig 7A-7C). To further investigate the differences in MT arrangement among cpr1-j594, cpr1-j2928, and wild-type PCs during their developmental process, MT arrangement during the three PC developmental stages was visualized. At first, the arrangement of cortical MTs was random in wild-type, cpr1-j594, and cpr1-j2928 PCs (Stage I) (Fig 8A, 8D and 8G). After lobe initiation, transverse MTs were most prominent in the neck region in wild-type PCs (Stage II) (Fig 8B), whereas the MT bundles were arranged in parallel through the whole PCs. (Fig 8E and 8H). Finally, after the full lobe expansion, mature PCs were formed and the cortical MTs exhibited a complex network with randomly oriented arrays in wild-type PCs (Fig 8C). In cpr1-j594 and cpr1-j2928 mature PCs, only simple cortical MT networks were observed (Fig 8F and 8I). All these results suggest that the directional arrangement of MTs during PC development is defective in cpr1 mutants.

The arrangement of the actin cytoskeleton is disturbed in cpr1 mutants
The actin cytoskeleton is also associated with plant cell polar growth and is essential for proper cell morphogenesis [30]. Therefore, we investigated the actin cytoskeleton in cpr1. GFP-FABD2 was used as a molecular marker to label the actin cytoskeleton in the PC. In wild-type PCs, abundant microfilaments formed actin cables networks (Fig 7D), whereas cpr1-j594 and cpr1-j2928 PCs did not form obvious networks (Fig 7E and 7F). These results indicated that the distribution and arrangement of the actin cytoskeleton is disturbed in cpr1. To determine at which developmental stage during PC morphogenesis started the actin cytoskeleton disruption in cpr1, the actin cytoskeleton during the three PC developmental stages was observed. In wildtype PCs, diffuse microfilaments spread throughout the cells at first (Stage I) (Fig 9A). During lobe initiation and expansion, microfilaments began to aggregate in apparently expanding lobes and gradually disappeared in the neck region of the cortex (Stage II). Some thin actin cables were also formed during this stage ( Fig 9B). Finally, the cortical diffuse microfilaments disappeared completely, and networks of actin cables were formed (Stage III) (Fig 9C). Little difference between cpr1-j594, cpr1-j2928, and wild-type PCs was observed during Stages I and II (Fig 9D, 9E, 9G and 9H). However, during the transition from Stage II to III, the cortical diffuse microfilaments in cpr1-j594 and cpr1-j2928 PCs did not totally disappeared and did not form apparent MF networks (Fig 9F and 9I). These results suggest that the cortical fine actin filaments cannot be aggregated effectively to form actin cable networks in cpr1 mutants and this maybe impact on the function of actin cytoskeleton.
PC morphogenetic defects in cpr1 require functional PAD4 and EDS1, and are temperature dependent ROP-GTPase has been reported to be crucial in regulating cytoskeleton organization in PCs [11,29]. However, yeast two hybrid (Y2H) and bimolecular fluorescence complementation (BiFC) assays showed that CPR1 cannot interact with ROP2, ROP4, and ROP6 directly (data not shown). Genetic interaction analysis between CPR1 and ROP2, and CPR1 and ROP6 also showed the CPR1 is not involved in ROP-GTPase signaling pathway (S3 and S4 Figs.). It has been reported that CPR1 negatively regulates defense responses through PAD4 and EDS1, two key components in the plant immune response network [20]. To test whether CPR1 regulates PC morphogenesis through the PAD4/EDS1-mediated signaling pathway, we characterize the PCs in cpr1 pad4 and cpr1 eds1 double mutants. PCs in single pad4 and eds1 mutants did not exhibited an obvious phenotype (Fig 10B and 10C), but in cpr1 pad4 and cpr1 eds1 double mutants the cpr1-mediated PC morphogenetic defects were suppressed (Fig 10D-10I; Fig 11). As CPR1-mediated plant immune response is temperature sensitive [25]. We also characterized the PCs of cpr1 mutants grown at 28°C. The abnormal PC shapes of cpr1-j594 and cpr1-j2928 were rescued at 28°C (Fig 10K and 10L; Fig 11), indicating that high temperature can rescue the phenotype of PCs in cpr1 mutants. Taken together, these results suggest that the CPR1-controlled PC morphogenesis requires normal PAD4 and EDS1, and is temperature dependent.

The FBA domain is indispensable for normal CPR1 function
The F-box protein is one of the major subunits within the SCF E3 ubiquitin ligase complex which recognizes the substrates for ubiquitination and proteasome-mediated degradation [31]. Many F-box proteins have an F-box-associated (FBA) motif at its C terminus which specifically recognizes the substrates. In our study, overexpression in Col-0 background of CPR1 with a point mutation in the FBA domain or without the FBA domain exhibited the phenotypes of the CPR1 loss-of-function (Fig 6). These results indicate that the FBA domain is essential for normal CPR1 function and the mutated CPR1 proteins affect the function of endogenous normal CPR1. It has also been reported that the F-box motif of CPR1 interacts with ARABIDOPSIS SKP1 HOMOLOGUE 1 (ASK1) or ARABIDOPSIS SKP1 HOMOLOGUE 2 (ASK2) to form SCF complexes [20]. Based on these results, we speculated that both CPR1 mutant proteins without functional FBA domains, can competitively combine with SKP1 and CULLIN proteins to form SCF complexes, but these complexes fail to recognize theirs substrates for degradation. Due to this competition with the endogenous CPR1, few functional SCF complexes will be formed, reducing the physiological activity and signal transduction in plants. Therefore, 35S:CPR1I247V and 35S:CPR1ΔFBA transgenic plants also showed cpr1 lossof-function phenotype.
CPR1 controls the cytoskeleton arrangement independent of the ROP-GTPase signaling ROP-GTPases work as a key switcher that mediate the interactions between the actin and microtubule cytoskeletons, controlling the high interdigitation of PCs [8,10,11,29,32]. In  [33][34][35], which are involved in polar cell growth and morphogenesis [36]. However, how F-box proteins influence the polar growth and morphogenesis of PCs in Arabidopsis is still unknown. ROP6, a small Rho GTPase involved in plant pathogen responses [37], exhibits similar PC morphogenesis phenotype to that of cpr1 when it is constitutively activated in plants [10], implying that CPR1 might take part in the degradation of the small Rho GTPases. However, both cpr1 rop6, cpr1 rop2 double mutants and cpr1rop6 rop2 triple mutant cannot rescue the phenotype of cpr1 (S3 and S4 Figs.). Protein interaction assays also proved that CPR1 cannot interact with ROP2, ROP4, and ROP6. Therefore, ROP-GTPases may not be CPR1 substrates. Comparison of the amino acid sequences of CPR1 and FBXL19 showed that both have an F-box domain within the N-terminus; however, FBXL19 possesses a unique leucine-rich repeat in the C-terminus, which has been proved to interact with RAC3 in animal cells [33]. Based on this analysis, we speculated that some F-box proteins with the unique leucine-rich repeat in the C-terminus domain may play a role in the ubiquitin-degradation of the small Rho GTPase during PC morphogenesis.
CPR1-mediated control of the cytoskeleton arrangement depends, at least in part, on the plant defense responses The cytoskeleton plays an active role in modulating the response to various types of antimicrobial defenses by blocking fungal pathogen penetration of the surrounding cells [38][39][40]. Studies on simple and complex cells, such as cylindrical hypocotyl cells and leaf epidermal PCs, respectively, have produced a mass of evidence to prove the importance of the cytoskeleton in control of cell shape [8,11,15]. In our study, the high temperature and the loss-of-function of PAD4 and EDS1 can completely rescue the PC morphogenetic defects in cpr1 mutants (Fig 10). In  [25]. It has also been reported that EDS1 and PAD4 have a higher steady expression level at 22°C than at 28°C [41]. Thus, high temperature and, the loss-of-function of PAD4 and EDS1 could both rescue the morphogenetic defects in the PCs of cpr1 mutants due to the inhibition to the PAD4/EDS1-mediated defense response. Taken together, these results suggest that the CPR1-mediated control of the cytoskeleton arrangement depends, at least in part, on the plant defense responses.