The role of KDEL-tailed cysteine endopeptidases of Arabidopsis (AtCEP2 and AtCEP1) in root development

Plants encode a unique group of papain-type cysteine endopeptidases (CysEP) characterized by a C-terminal KDEL endoplasmic reticulum retention signal (KDEL-CysEP) and an unusually broad substrate specificity. The three Arabidopsis KDEL-CysEPs (AtCEP1, AtCEP2, and AtCEP3) are differentially expressed in vegetative and generative tissues undergoing programmed cell death (PCD). While KDEL-CysEPs have been shown to be implicated in the collapse of tissues during PCD, roles of these peptidases in processes other than PCD are unknown. Using mCherry-AtCEP2 and EGFP-AtCEP1 reporter proteins in wild type versus atcep2 or atcep1 mutant plants, we explored the participation of AtCEP in young root development. Loss of AtCEP2, but not AtCEP1 resulted in shorter primary roots due to a decrease in cell length in the lateral root (LR) cap, and impairs extension of primary root epidermis cells such as trichoblasts in the elongation zone. AtCEP2 was localized to root cap corpses adherent to epidermal cells in the rapid elongation zone. AtCEP1 and AtCEP2 are expressed in root epidermis cells that are separated for LR emergence. Loss of AtCEP1 or AtCEP2 caused delayed emergence of LR primordia. KDEL-CysEPs might be involved in developmental tissue remodeling by supporting cell wall elongation and cell separation.


Introduction
Plants encode a unique group of papain-type cysteine endopeptidases (CysEP) characterized by a C-terminal KDEL endoplasmic reticulum (ER) retention signal (KDEL-CysEP) with RcCysEP from castor bean (Ricinus communis) as the founding member [1][2][3]. KDEL-CysEPs are not found in mammals or fungi, but are ubiquitous in spermatophytes; the number of PLOS

Arabidopsis mutant plants
Homozygous ko mutant plants were obtained for cep1 (SAIL_158_B06, [26]) and for cep2 (SALK_079519; T-DNA insertion in the second exon) by segregation analysis and genotyping. We performed three reciprocal back crossings in order to remove T-DNA insertions elsewhere in the genome. Transcription analysis confirmed homozygous cep1 ko [26] and cep2 ko mutant plants (S1 Fig). During back-crossing of the cep2 mutant allele, we recovered homozygous cep2 ko mutant plants. However, even by consecutive back crossing we were not able to recover Mendelian segregation of the cep2 mutant and WT alleles: No homozygous WT plants resulted from the back crosses, indicating a secondary T-DNA insertion which could not be removed. We refrained therefore from using the cep2 mutant allele for further crosses and modifications such as cep1 cep2 double mutant generation or transformation with reporter constructs. Since no second cep2 insertion line was available, we used two independent cep2-RNAi lines in order to confirm the observed cep2 ko mutant phenotype. cep1 mutant plants behaved like WT in the context of our research concerning primary root elongation (see Results). We used the cep2-RNAi construct to transform cep1 mutant plants in order to analyze cep2 knock down (kd) mutants in the cep1 background. Silencing of CEP2 was achieved using pHANNIBAL and the binary vector pART27 to accept the NotI fragment from pHAN-NIBAL (CSIRO Plant Industry, Canberra ACT 2601, Australia), and the Agrobacterium tumefaciens strain GV3010::pMP90. A representative CEP2 region in the 3'UTR comprising 134 bp was amplified from TAMU-BAC T29H11 as BamHI/XhoI-fragment and as ClaI/KpnI-fragment, respectively, using the primers 5'-AAC GGA TCC TCG AGA GAC TTT AAG TCA TTG AAA ACT G-3' and 5'-TGG ATC GAT GGT ACC TGC TGG CCA ATA TTA CAA GGA G-3'. The resulting PCR products were cloned into pHANNIBAL as BamHI-ClaI-fragment and as XhoI-KpnI-fragment, and the final plasmid was sequenced. The NotIfragment was cloned into pART27, and the final plasmid was transformed into A. tumefaciens by electroporation. Flowers from homozygous cep1 ko mutant plants were transformed by floral dipping [29] with A. tumefaciens harboring the CEP2-silencing plasmid and in parallel with A. tumefaciens harboring the CEP2-reporter including mCherry without the mature CEP2 protein (P CEP2 ::pre-pro-3xHA-mCherry-KDEL, [6]) resulting in plants silencing CEP2. Primers used for qRT-PCR were: ACT8 qRT forward: TGAGACCTTTAATTCTCCAGCTATG; ACT8 qRT reverse: CCAGAGTCCAACACAATACCG; CEP2 qRT forward: GCTGTTGCAAACCAACCT G; CEP2 qRT reverse: TTCCACAAGATCCCGTAAACA. qRT-PCR was carried out as described [26]. Two separate mother plants were transformed and gave rise to cep1/cep2 double ko/kd mutants 2.x or 3.x. Eleven individual homozygous lines of cep1/cep2 double ko/kd mutants (cep1+cep2-RNAi) were screened by qRT-PCR for CEP2 expression (S2 Fig). Line 2.21 with CEP2-regulation at 0.5% and line 3.14 with CEP2-regulation at 2% of WT were chosen for further analysis. Expression of the reporter protein pro-3xHA-mCherry-KDEL in young seedlings was below detection level using Confocal Laser Scanning Microscopy (CLSM). CEP1 expression in primary root cells was monitored by CLSM, using 10 days old cep1 ko mutant seedlings containing the CEP1 reporter with the CEP1 targeting sequences fused to a three-fold hemagglutinin-tag, the green fluorescent protein EGFP and the mature CEP1 protein under control of the endogenous CEP1 promoter (P CEP1 ::pre-pro-3xHA-EGFP-At-CEP1-KDEL, [26]). CEP2 expression was monitored by CLSM using 10 days old WT seedlings containing the CEP2 reporter with the CEP2 targeting sequences fused to a three-fold hemagglutinin-tag, the mCherry fluorescent protein and the mature CEP2 protein under control of the endogenous CEP2 promoter (P CEP2 ::pre-pro-3xHA-mCherry-AtCEP2-KDEL, [6]).

Whole mount immunolocalization
The main procedure was done as described [30]. Seedlings were grown for 10 days on Gamborg medium. Their roots were fixed in formaldehyde (4% in 1×PBS [140 mM NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , pH 7.4] and 0.1% Tween) for 1h, followed by several washes with distilled water and 1×PBS. Washed roots were placed on a slide to dry at RT. The samples were subsequently rehydrated in 1×PBS followed by incubation with Driselase (2% in 1×PBS) in a humid chamber at 37˚C for 45 min. After repeated washing with 1×PBS, samples were incubated in 4% IGEPAL (octylphenoxypolyethoxyethanol) and 10% DMSO in 1×PBS in a humid chamber at 37˚C for 1 h, followed by several washes in distilled water and 1×PBS, then incubated for 30 min at RT with a blocking solution (2% BSA in 1×PBS and 0.01% Tween), and finally with rabbit antibody to the AtCEP2 II peptide, that is the 15 amino acids of AtCEP2 directly adjacent to the C-terminal "KDEL" ER retention signal (C-IKLSSSNPTPK DGDV, [4]) (1:60; Eurogentec, Belgium) overnight at 4˚C. The preparation was washed with blocking solution, and subsequently incubated with cyanine CY2-coupled secondary anti-rabbit antibody (1:60; Dianova) at 37˚C for 4 h, and finally washed with blocking solution and covered with water for microscopy with a Leica SP8 and a 40× objective using a laser intensity of 20%. The cyanine CY2-coupled secondary antibody was excited at 488 nm and emission was measured at 500-550 nm. Autofluorescence of the nuclei was recorded by excitation at 405 nm and emission at 420-475 nm.

LR formation as synchronized by a 90˚gravitropic stimulus ("root bending assay")
The procedure was done as described [31].

Confocal laser scanning microscopy
CLSM was carried out as described ( [26], Fluoview FV 1000, Olympus, Japan) with excitation at 488 nm and emission at 503-550 nm for EGFP, excitation at 561 nm and emission at 570-630 nm for mCherry, excitation at 405 nm and emission at 420-450 nm for calcofluor and excitation at 488 nm and emission at 560-640 nm for propidium iodide. Single pictures or stacks of pictures with 0.5 μm increments at higher resolution and 2.5 μm increments at lower resolution were made. A Leica confocal laser scanning microscope (Leica SP8, Germany) was used for mCherry-CEP2 in epidermal cells to be separated for LR emergence. 0 WT plants, but longer than cep2 ko mutants ( Fig 1A). Interestingly, the residual CEP2 activity resulted in primary roots significantly longer than the cep2 ko mutants and shorter than the Col-0 WT plants (p<0.001, Fig 1B). The primary root length of 5-, 7-and 11-days old Arabidopsis seedlings in cep2 ko mutants was 58-63% of Col-0 WT, whereas the cep1+cep2-RNAi mutant with 0.5% CEP2 activity was 67-71% of Col-0 WT primary root length and the cep1 +cep2-RNAi mutant with 2% CEP2 activity was 73-75% of Col-0 WT ( Fig 1C). The primary root length in cep1 single ko mutant plants exhibited no significant difference compared to WT (p>0.05, Fig 2). Besides the root length depending on the CEP2 activity, we observed the rosette size to be smaller comparing the entire plant of cep2 mutants with WT. We concentrated our analysis, however, on the roots as a good model for cell elongation and directed growth.
We determined which tissue or cell type might be responsible for the shortened primary root length in CEP2 loss of function mutants. The cells in the upper part of the lateral root (LR) cap (PCD Site I), which are the most distant to the LR cap cells near the columella root cap (PCD site II), elongate before they degrade attached to the root surface; disturbed LR cap cell elongation might influence the primary root length. Furthermore, the primary root length might be influenced through the cell length in the meristematic zone and in the elongation zone (for a general orientation see S3 Fig).
Primary roots of 7 days old seedlings were stained with propidium iodide in order to visualize the cell wall and to measure cell dimensions in the LR cap and the elongation zone. Cells approaching PCD could also be identified since their cell membranes become permeable, resulting in propidium iodide stained nuclei in addition to the cell wall. We measured the In general, the longest cells in the LR cap are found at PCD Site I (Fig 4; S3 Fig, asterisk). The longest LR cap cells (the 3-5 LR cap cells exhibiting max. cell length) of cep2 ko mutants (50-60 μm) were shorter than the longest LR cap WT cells (75-90 μm); the longest LR cap cells of the cep1+cep2-RNAi (60-75 μm) were also shorter than WT cells, but longer than in cep2 ko mutants (50-60 μm) (Fig 4A, double arrows). PCD of LR cap cells in cep2 ko mutants was similar to WT as indicated by the nuclei of these cells stained-in addition to the cell wallwith propidium iodide indicating perforated cell membranes and thus the beginning PCD ( Fig  4A, arrowheads). This observation was confirmed by statistical evaluation of the 3-5 longest cells in the LR cap (p<0.001; Fig 4B and 4C). The mean length of Col-0 WT LR cap cells was 83.9 μm, versus 62.4 μm (74% of WT) in cep2 ko mutants, and versus 67.1 μm (80% of WT) in the cep1+cep2-RNAi mutants with 0.5% CEP2 activity and 69.3 μm (83% of WT) with 2% CEP2 activity ( Fig 4B left and 4C). There was no statistically significant difference in the mean width of the longest LR cap cells between WT and cep2 mutant plants (p>0.05; Fig 4B right).
PCD in the LR cap was similar in WT plants, in cep2 ko mutants and in cep1+cep2-RNAi mutants, giving rise to normal development of the LR cap ( Fig 5). The lengths of the PCD Sites I and II in cep2 ko mutants and cep1+cep2-RNAi mutants were similar to WT plants ( Fig 5) and both PCD sites can be recognized as a distinct zone containing cells with propidium iodide stained nuclei and cell walls (Fig 5, arrowheads). However, the PCD Site I was closer to the root tip due to the reduced cell length of the LR cap in cep2 ko mutants and again to a lesser extent in cep1+cep2-RNAi mutants compared to WT plants (Fig 5). The general morphological organization of the Arabidopsis root tip is depicted in S3 Fig  the edges of the respective root cap layers that change over time. At about 5 days after germination, there are two PCD sites, PCD site II is formed by the oldest root cap layer in which PCD has already moved towards the columella, whereas PCD site I is formed by the next younger root cap layer which still covers the entire meristem, so that the PCD happens close to the elongation zone. As these sites are not fixed but move in dependency of the age of the root imaged [8], slight differences between the images in Fig  The mean length of the meristematic zone, which is the distance between the quiescent centre and the first elongated epidermis cells ( We concentrated our analysis of CEP2 involvement in cell elongation on young trichoblasts (the elongated cells developing root hairs) as an example for epidermis cells, since they are localized at the end of the rapid elongation zone and can easily and unequivocally be identified through their emerging protrusions of root hairs. We calculated the mean cell length and width of the three longest trichoblasts in the elongation zone at the level of the transition from the rapid elongation zone with rapid cell elongation to the late elongation zone, where cell elongation is no longer increasing. The transition between the rapid and late elongation zones can be recognized by the formation of the emerging protrusions of root hairs in trichoblasts (  longest trichoblasts of the cep2 ko mutants also was significantly smaller than in WT plants (p<0.001; Fig 7B right and 7C) with a significantly smaller total root width at the elongation zone as compared to WT that parallels the trichoblast cell width (p<0.01; Parts A and B in S4  Fig). Loss of CEP2 alone results in trichoblasts with a width of 87% as compared to WT and thus with 87% an equally reduced width of the elongation zone. Residual CEP2 activity in cep1/2 double ko/kd lines resulted in 88-89% smaller trichoblasts and elongation zone as compared to WT. There is no indication that CEP2 activity is restricted to trichoblasts. The elongation of other cells in the epidermis such as atrichoblasts, or cortex cells might also be impaired.
CEP2 controls the growth of primary roots by influencing epidermal cell elongation. We therefore tried to localize the expression of CEP2.

Pro-CEP2 resides in the ER of root cap cells prepared for PCD and is still present in root cap corpses
Calcofluor-white was used to visualize the cell walls of the roots of young seedlings transformed with the reporter construct P CEP2 ::pre-pro-3xHA-mCherry-AtCEP2-KDEL in the Col-0 AtCEP2 and AtCEP1 in root development WT background (Fig 8). Within the transition zone, there were isodiametric, non-elongated epidermal cells from the meristematic zone (Fig 8 left, arrowheads) and elongating epidermal cells (Fig 8 left, arrows) indicating the beginning of the elongation zone. There were also small, elongated cells (Fig 8 left, asterisk) representing collapsing cells from the upper end of the LR cap at PCD site I in the process of degrading while still attached to epidermal cells. Typically, the width of root cap cells is half the width of the underlying epidermis cells. Pro-3xHA-mCherry-AtCEP2-KDEL was seen in LR cap cells prepared for PCD at the upper end of the LR cap and accumulated in the cytoplasm surrounding the vacuole (Fig 8, middle and right); typically, pro-CEP accumulate within the ER and ER derived organelles [6,26]. After death of LR cap cells, mCherry signal appears to localize to the degrading remains of LR cap cells (Fig  8, middle): dotted red signals on the left side of the root within the transition zone and on the right side of the root left to the word "Epidermis" and "Lateral". The faint dotted and weak red signals within epidermal cells (Fig 8, middle: upper part) could be interpreted as unspecific back ground signals of pro-3xHA-mCherry-AtCEP2-KDEL. However, even if we do find CEP2 expression in the epidermal cells of the meristematic zone, the transition zone, or the elongation zone, it would be much lower than in the root cap.

CEP2 was immuno-localized to root cap corpses sticking to epidermis cells in the rapid elongation zone
Maturation of pro-CEP2 to the enzymatically active protein by acidification results in autocatalytic cleavage of the pro-peptide and the KDEL-motif plus cleavage between mCherry and the mature CEP2 protein [6]. As a consequence, the mature CEP2 protein cannot be detected by CLSM. For that reason, whole mount immunolocalization with anti-peptide antibodies recognizing the 15 amino acids directly upstream of the C-terminal KDEL-motif (C-IKLSSSNPTP KDGDV; [6]) was used to localize mature CEP2 to cells at the upper end of the LR cap prepared for PCD (PCD site I) just below the transition zone and possibly to epidermal cells at the beginning elongation zone and the rapid elongation zone in the root tip of young seedlings (Fig  9). LR cap cells are very elongated with one small nucleus and are much longer than the cells beneath in the meristematic zone (Fig 9A and 9D; see 9E and 9F, asterisk). The distinct punctate CEP2 signals were localized exclusively in corpses of LR cap cells (Fig 9A and 9D arrows; 9E and 9F). The epidermal cells beneath were short (Fig 9E, double arrows) with their nucleus in the centre of the cell; they exhibited no CEP2 signal. As soon as the LR cap cells collapsed due to PCD, the length of the epidermal cells increased (Fig 9B, double arrows) and their nuclei are localized to the inner surface of the epidermal cell (Fig 9B and 9C). Both observations identify the transition between the end of the LR cap and the beginning of the elongation zone, the rapid elongation zone [32]. Within the transition zone, the distinct punctate CEP2 signals are localized exclusively to root cap corpses still attached to epidermal cell walls (Fig 9A and 9D arrow heads, 9B and 9C). This observation corresponds to the pro-3xHA-mCherry-CEP2-K-DEL localization in the degrading remains of the dead LR cap cells (Fig 8, middle).
As negative controls, no background signal was detected in the cep2 ko mutant using the same antibody or in Col-0 WT plants using pre-immune serum (not shown). We detected specific CEP2 signals in whole mount immunolocalization within the differentiating protoxylem of the root tip vasculature (S5 Fig) indicating that the antibody could penetrate all cell layers. The antibody binds only within the stele and not in epidermis cells. In a differentiated root, we found CEP2 signals in epidermis cells that have to be moved for the emerging lateral root (see next paragraph). The loss of function mutant phenotype demonstrates that CEP2 activity is required for epidermal cell elongation as we could demonstrate it for the trichoblasts. The CEP2 activity as it is localized in root cap corpses might impact epidermal cell elongation.

CEP1 and CEP2 are expressed in the root epidermis cells at lateral root (LR) emergence sites
LRs originate from the pericycle cells situated in front of the xylem poles. Arabidopsis LR emergence requires the new primordia to break through the three overlying outer tissues that are each composed of one layer of endodermis, cortex, and epidermis (see S6 Fig; [33,34]). The involvement of CEP1 in the cell separation necessary for LR emergence was investigated in young seedlings expressing the CEP1 reporter protein (P CEP1 ::pre-pro-3xHA-EGFP-At-CEP1-KDEL; Fig 10) in vivo by CLSM. No pro-EGFP-CEP1 expression was observed from developmental stage I to VI (not shown). From developmental stage VII on, when the newly formed LR primordium reaches the epidermis (S6 Fig), pro-EGFP-CEP1 was detected in the epidermis cell (Fig 10A-10D, asterisk) that must separate to make room for the emerging LR primordium (Fig 10A-10D, arrowhead). The expression of pro-EGFP-CEP1 in this epidermal cell was observed at developmental stage VIII, when emergence of the LR primordium through the epidermis is complete (Fig 10E-10H) and at later developmental stages, when the LR primordium had emerged from the epidermis and is defined as an adult LR (Fig 10I-10M). This pro-EGFP-CEP1 expressing epidermis cell has a strong curvature probably due to the mechanical forces during cell separation in the course of tissue remodelling (Fig 10). Interestingly, in this case only one cell had to be moved, perhaps resulting in the observed strong curvature. In other cases, more than one cell had to be moved, depending on the position of the emerging LR. CEP1 expression was observed in the epidermal cell next to the epidermal cell with the strong curvature (Fig 10G) indicating that the epidermal cells in the direct neighbourhood are also involved in giving room for the emerging LR.
We also analysed young seedlings expressing the CEP2 reporter protein (P CEP2 ::pre-pro-3xHA-mCherry-AtCEP2-KDEL) in vivo by CLSM (Fig 11A-11C). Pro-mCherry-CEP2 was localized in root epidermal cells that separate to make room for the emerging LR primordium (Fig 11A-11C, asterisk). Pro-mCherry-CEP2 continued to be expressed after LR emergence, as seen for pro-EGFP-CEP1. Furthermore, CEP2 was immunolocalized to epidermis cells that have to be moved for the emerging LR (Fig 11D-11F). Similar to CEP1 expression, CEP2 expression was detected in the two cells that have to give room for the emerging LR and in the epidermis cells next to lateral root outgrow in order to make the cells in the direct vicinity more flexible. The CLSM used to generate the data displayed in Fig 11 (Leica SP8) is more sensitive as compared to the CLSM used for the data presented in Fig 10 (Olympus Fluoview FV 1000), allowing for the detection of pro-mCherry-CEP2 (Fig 11A-11C), and this detection even at stages I-VI of LR primordium development.

Loss of function of CEP1 or CEP2 retards lateral root (LR) emergence
LR formation was synchronized by a 90˚gravitropic stimulus ("bending assay", [31]), which is a 90˚rotation of 7-10 days old seedlings on an agar-plate inducing a synchronous formation of LRs at the bending site, in order to investigate if the loss of CEP function delayed LR emergence. Eight stages of LR primordium development are defined (S6 Fig). Arabidopsis roots originate from pericycle founder cells and undergo several rounds of anticlinal and periclinal divisions resulting in a dome-shaped primordium that penetrates the endodermis and cortex (stages I-VI) until it reaches the epidermis (stage VII). It eventually emerges from the parental AtCEP2 and AtCEP1 in root development root (stage VIII) becoming an adult LR. Since CEP1 and CEP2 were found in root epidermal cells, we performed the quantitative evaluation of the stages in LR primordium development 49 hours after "root bending", when most of the WT plant primordia had just reached the epidermis (stage VII) or emerged from the parental root (stage VIII), thus combining the evaluation of stages I-VI (Fig 12).
WT plants exhibited no LR primordia in developmental stages I-VI, 14% LR primordia at developmental stage VII, 78% LR primordia emerging from the parental roots at developmental stage VIII; and 8% adult LR.
cep1 ko or cep2 ko mutants exhibited 12% LR primordia in developmental stages I-VI; they showed significantly more LR primordia (33%) than WT plants at developmental stage VII; cep single loss of function mutants had significantly fewer (p<0.05) LR primordia (52-53%) emerging from the parental root than WT plants at developmental stage VIII. Finally, 49 hours after root bending cep2 single ko mutants had with 2% adult LR significantly fewer adult roots at developmental stage VIII than WT plants (p<0.05); cep1 ko (3% LR) exhibited no significant difference to WT plants or to cep2 single ko mutants. cep1/2 double ko/kd mutants (cep1+cep2-RNAi) mutants did not show an additive phenotype. They had 1-2% LR primordia in stages I-VI, which was more than WT plants, but less than the single cep ko plants (p<0.05). At developmental stage VII, cep1/2 double ko/kd mutants with 29% primordia had more LR primordia than WT plants, but the difference between cep single ko plants and cep1/2 double ko/kd mutants was not statistically significant (p<0.05). At developmental stage VII, the cep1/2 double ko/kd mutants with 66% LR primordia emerging from the parental roots exhibited a phenotype quantitatively between single cep ko and WT plants. Finally, cep1/2 double ko/kd mutants (4% adult LR) exhibited no significant difference to WT plants or to cep2 single ko mutants.
Single loss of function of CEP1 or CEP2 resulted in the strongest quantitative phenotype in retardation of LR emergence. Compared to WT plants, there was a clear retardation in LR emergence resulting from cep single loss of function. cep double loss of function did not show an enhanced retardation of LR emergence compared with cep single loss of function. We conclude that CEP1 and CEP2 contribute to the emergence of LR primordia from the parental root by acting within the root epidermal cells.

Discussion
The Arabidopsis genome encodes three KDEL-CysEPs-AtCEP1, AtCEP2, and AtCEP3. In this study, we investigated the role of CEP during root development using A. thaliana seedlings as a model system.
We have demonstrated that CEP2 is involved in elongation of the primary root. Loss of CEP2 results in shortened primary roots due to reduced cell length of epidermal cells such as trichoblasts. Atrichoblasts and cortex cells might show similar elongation phenotypes. We cannot exclude that CEP2 is involved in general cell elongation resulting in an overall smaller growth of cep2 mutant plants. Loss of CEP2 also leads to a decrease in LR cap cell length at the level of PCD site I. The low level of CEP2 in two independent cep1/2 double ko/kd lines is sufficient to cause a measurable shortening of LR cap cells and the epidermal cells at the starting, rapid elongation zone of primary roots. The length of the Arabidopsis primary root is entirely dependent on cell extension in the elongation zone [8,35]. Loss of CEP2 in these genotypes results in a decreased root diameter rather than root compression, which could have influenced root length [36][37][38]. Thus, the shorter LR cap likely has no influence on primary root length; the root elongation is not hindered by the LR cap. The shortened primary root phenotype due to the loss of CEP2 is more likely resulting from the shortened epidermal cells such as trichoblasts. Thus, CEP2 participates in elongating the primary roots. The loss of function mutant phenotype demonstrates that CEP2 activity is required for cell elongation, possibly due to its cell wall remodeling ability. In contrast, since loss of CEP1 does not result in shortened primary roots, CEP1 does not appear to be involved in primary root elongation.
Pro-CEP2 accumulates within the protoplasm of the longest from the upper end of the LR cap prepared for PCD (PCD site I) at the level of the transition zone and is still present in root cap corpses sticking to epidermis cells in the rapid elongation zone. Furthermore, we used a specific antibody to a peptide in the mature protein of CEP2 which recognizes the mature protein [6]. The CEP2 protein was immunolocalized to root cap corpses sticking to the cell wall of elongating epidermal cells at the beginning of the rapid elongation zone; CEP2 immunolocalization to the apoplast and to LR cap cells below prepared for PCD cannot be excluded. In these cell types, CEP2 could be involved in loosening the neighbouring cell wall for extension. The pro-form of KDEL CysEP is readily converted to the matureted protein at pH below 6.5. In Arabidopsis, the apoplastic pH in the LR cap is 5.3-5.5 and in the elongation zone 4.8-4.9, respectively [39]. At this pH, the relative enzymatic activity of the maturated CEP2 can be measured unequivocally within one minute. If CEP2 would be transported or would diffuse to the cell wall as the enzymatically inactive pro-enzyme, it would be immediately transformed by autocatalytic cleavage at a pH below 6.5 to the enzymatically active protein [3,6]. Our data indicate the possibility that CEP2 activity released from root cap corpses and also from root We demonstrated expression of pro-EGFP-CEP1 and pro-mCherry-CEP2 at stage VII and VIII of LR primordium development in the protoplasm of those root epidermal cells that are separated to enable LR emergence. CEP1 and CEP2, respectively, were expressed only in the two epidermal cells surrounding the emerging lateral root and in the epidermal cells in their direct neighbourhood. Normally, both cells are curved [40]. In some cases, depending on the specific site of the emerging LR, bending of just one epidermal cell seems to be enough for separation thus giving room to the emerging root tip. CEP1 and CEP2 exhibit a high degree of fine tuning concerning their gene expression and protein activity in a specific cell depending on the developmental program. Pro-mCherry-CEP2 was also detected at stage I-VI. GUS expression driven from the CEP1-promoter was detected at stage I-VI [10], whereas expression of pro-EGFP-CEP1 at stage I-VI might be below detection level of the CLSM used. cep1 or cep2 single loss of function caused the strongest phenotype in retardation of LR emergence, whereas cep1 cep2 double loss of function did not exhibit an enhanced phenotype. We speculate that in case of total loss of CEP the plant might recruit other proteases that are present in the course of lateral root emergence [41].
In angiosperms, LRs are derived from the pericycle layer deep within the parent root tissues [33,34]. The Arabidopsis root has a simple structure composed of the stele surrounded by the three one-cell layers endodermis, cortex and epidermis with the pericycle forming the outermost layer of the stele. The LR primordium penetrates these three cell layers in a spatiotemporally controlled process. The coordination of LR formation and emergence is controlled by auxin, which regulates the properties of cells overlaying LR primordia [42]. NAC1, a member of the NAC family, is induced by auxin and mediates auxin signaling to promote LR development [43]. Several genes encoding cell wall remodeling enzymes including polygalacturonase (PG), xyloglucan:xyloglucosyltransferase (XTR6) and subtilisin-like protease (AIR3) are induced by auxin in front of the LR primordium and promote cell separation [44]. By studying de novo root organogenesis using leaf explants of Arabidopsis, it was shown that wounding not only triggers the auxin-mediated fate transition of regeneration competent cells, but also induces the NAC pathway for root tip emergence [45]. The NAC1 transcription factor was specifically expressed in response to wounding. Here, the NAC1 pathway functioned independently of auxin and regulated expression of CEP genes. NAC1 overexpression induced many genes during de novo root organogenesis, especially CEP1 and CEP2. Up-regulation of CEP might be related to the degradation of extensins, which would promote wound healing, and this might be a barrier for regenerated root tip emergence [45].
For the involvement of CEP2 in tissue remodelling involving cell wall extension, the matureted protein should act within the cell wall. The CEP2 immunolocalization appears to be inside cell corpses or even apoplastic. In any case, CEP2 can diffuse out of the corpses into the apoplast of expanding epidermal cells.
It remains to be elucidated how the pro-enzymes of KDEL-CysEP as they are stored in the protoplasm in ER-derived compartments, known as ricinosomes, KDEL-tailed protease-accumulating vesicles (KVs) and ER-bodies [3,[5][6][7], are transported to the cell wall. A Golgidependent secretion pathway is assumed for cell wall remodelling enzymes, such as pectin methylesterases and polygalacturonases (PG), and for xyloglucan-endoglucanases/hydrolases (XTH) involved in cell elongation [46][47][48][49]. Since KDEL-CysEP are not glycosylated, as known for their founding member RcCysEP [12] and since they exhibit the ER retention signal [6], a Golgi-independent pathway including membrane fusion is postulated. In germinating mung bean (Vigna mungo) cotyledons, their KDEL-CysEP (Sulfhydryl-endopeptidase, SH-EP) is accumulated in KV for mass transport to protein storage vacuoles (PSV) [5]. KV bud from the ER and transport the pro-SH-EP in a Golgi-independent pathway directly to PSV. After fusion of the two membranes, the pro-SH-EP is converted to the mature SH-EP due to the low pH within the PSV and participates in storage protein mobilization [50]. ER bodies are produced by plants of the Brassicales order, which includes Arabidopsis. Pathogenesis-related proteins (PR), which have antimicrobial activity, accumulated in ER-bodies at a steady-state level and were secreted to the apoplast in response to attack by pathogenic fungi. This suggests that ER bodies are storage sites for various PR proteins released in response to pathogenic attack (for review see [7,51]). As ricinosomes are storage sites for KDEL-CysEPs, a transport similar to KVs and ER bodies to the apoplast by membrane fusion and/or secretion can be anticipated.
CEP1 is involved in tapetal PCD and pollen development [20]. At stage 5 of tapetal development, CEP1 was concentrated in the ricinosome-like precursor protease vesicles of the tapetal cytoplasm. At early stage 6, CEP1 appeared in tapetal cell vacuoles with fusion of these protease vesicles and transformed into the mature protein before rupture of the vacuole. Subsequently, the CEP1 was released from the vacuoles into the tapetal cytosol. CEP1 was also released to the cell wall by the fusion of small vesicles with the cell membrane during late stage 6 and stage 7. At late stage 7, the CEP1 entered the callose wall and probably participated in callose wall degradation [20]. We infer that KDEL-CysEPs can be transported to the cell wall by fusion of ERderived vesicles with the plasma membrane. In the apoplast, the pH is acidic enough to activate the accumulated KDEL-CysEP for release to tissues undergoing collapse.
Interestingly, a number of cell wall proteins are upregulated in cep1 mutants. These include three extensin family genes and other cell wall biogenesis genes such as proline-rich protein 4 (PRP4), xyloglucan endotransglycosylase/hydrolase 3 (XTH3) and pectin methylesterase 44 (PME44), suggesting that CEP1 is important for tapetal cell wall organization and degradation [20]. The moderately glycosylated extensins belonging to the family of hydroxyproline-rich glycoproteins are important and abundant structural proteins in the primary cell wall [11,[52][53][54]. The Arabidopsis genome encodes 59 extensin genes. Discovery of the lethal rsh (root-, shoot, hypocotyl-defective) embryogenic Arabidopsis mutant corresponding to EXT3 showed that extensins are essential for cell plate formation, evidenced by the aberrant mutant wall phenotype and EXT3 immunocytochemical localization [11,52,55].
Extensins are ubiquitous in all cell walls including LR cap cells and epidermal as well as cortical cells from the root tip up to the differentiation / root hair zone. The amount of extensins in the respective root cell walls varies considerably. Extensins are found predominantly in epidermal and cortical cell walls of the starting, rapid elongation zone with an increase in cell length up to 300% within less than 3 hours and root upwards to the late elongation zone with cells having already reached their maximal length [32]; comparably small amounts of extensins are in the LR cap cell walls [35]. Remodeling enzymes loosening the rigid cell wall structure are decisive for cell elongation [49,56,57]. The development of the Arabidopsis LR cap comprises cell elongation [58]. The relationship between the amount of CEP and extensin should influence the degree of cell wall attack. CEP2 is localized in root cap corpses sticking to epidermal cell walls at the beginning rapid elongation zone and in LR cap cells that are elongating before they are shed (PCD site I) and is thus directly involved in cell wall modulation. In a differentiated root, we see CEP signals only in selected epidermis cells supporting the LR outgrowth and within the stele.
Out data set the stage for uncovering novel roles for KDEL-CysEP in cells that are not undergoing PCD but are characterized by extensive cell wall expansion or remodeling possibly due to their cleavage of extensins. The involvement of a protease in cell wall remodeling points to the importance of the cell wall protein moiety in this process. It will be interesting to obtain further insights into the underlying mechanisms such as the membrane fusion of ricinosomes with the plasma membrane in the course of root development similar to other ER-derived vesicles. Upregulation of cell proteins such as extensins is already demonstrated for cep1 mutants in tapetal cell wall organization. It will be interesting to see whether extensin cleavage and its regulation determines the impact of KDEL-CysEPs, with extensive extensin cleavage resulting in the total breakdown of the cell wall and ensuing cell collapse, while more limited extensin cleavage would enable cell wall remodeling in the context of cell elongation or cell separation.