Ubiquitylation activates a peptidase that promotes cleavage and destabilization of its activating E3 ligases and diverse growth regulatory proteins to limit cell proliferation in Arabidopsis

The characteristic shapes and sizes of organs are established by cell proliferation patterns and final cell sizes, but the underlying molecular mechanisms coordinating these are poorly understood. Here we characterize a ubiquitin-activated peptidase called DA1 that limits the duration of cell proliferation during organ growth in Arabidopsis thaliana. The peptidase is activated by two RING E3 ligases, BB and DA2, which are subsequently cleaved by the activated peptidase and destabilized. In the case of BB, cleavage leads to destabilization by the RING E3 ligase PRT1 of the N-end rule pathway. DA1 peptidase activity also cleaves the de-ubiquitylase UBP15, which promotes cell proliferation, and the transcription factors TCP15 and TCP22, which promote cell proliferation proliferation and repress endoreduplication. We propose that DA1 peptidase activity regulates the duration of cell proliferation and the transition to endoreduplication and differentiation during organ formation in plants by coordinating the destabilization of regulatory proteins.


INTRODUCTION
The shapes and sizes of organs are established by mechanisms that orient cell proliferation and determine the final numbers and sizes of cells forming the organ.
Transplantation experiments showed that some animal organs have an intrinsic mechanism that determines their final size by controlling the duration of cell proliferation (Barry and Camargo 2013), which is controlled in part by the HIPPO/YAP pathway that limits cell proliferation and promotes apoptosis (Pan 2010).
However, the mechanisms coordinating cell proliferation and cell size during organ growth remain poorly understood (Johnston and Gallant 2002). Due to the simpler planar structures of their organs such as leaves and petals, and the absence of cell movement due to rigid cell walls, plants have some experimental advantages for studying organ growth (Green et al. 2010).
Leaf growth in plants is initiated at shoot meristems (reviewed by Sluis and Hake 2015). After specification of boundaries and growth axes, the leaf lamina grows in an initial period of cell division in which cell size is relatively constant, followed by a transition to endoreduplication associated with cell expansion and differentiation (Breuer et al. 2010;De Veylder et al. 2011). The transition from cell proliferation to cell expansion is spatially and temporarily regulated during leaf growth, and appears to progress from the tip to the base of the leaf as a cell division arrest front (Kazama et al. 2010), accompanied by shifts in gene expression patterns (Efroni et al. 2008;Andriankaja et al. 2012). A key question is how the transition from cell proliferation to cell expansion and differentiation is coordinated to generate a correctly sized organ.
The RING E3 ligases Big Brother (BB) (Disch et al. 2006a) and DA2 (Xia et al. 2013) limit the duration of cell proliferation during organ growth. Members of the DA1 family also limit cell proliferation (Li et al. 2008), and loss of function mutations in BB and 4 DA2 interact synergistically with the da1-1 allele of DA1 to increase organ and seed size in Arabidopsis (Li et al. 2008;Xia et al. 2013), suggesting one of their growth limiting activities is mediated by enhancing the growth-repressive activity of DA1 family members. Genetic analyses showed that DA1 reduced both the stability of UBP15 (Du et al., 2014), a de-ubiquitylation enzyme promoting cell proliferation (Lui et al., 2008), and reduced the stabilities of TEOSINTE BRANCED 1/CYCLOIDEA/PCF (TCP) 14 and TCP15 proteins (Peng et al., 2015) which repress endoreduplication by transcriptional control of RETINOBLASTOMA-RELATED1 (RBR1) and CYCLIN A2;3 (CYCA2;3) gene expression (Li et al., 2012).
Here we show that DA1 is an endopeptidase activated by multiple ubiquitylation mediated by the E3 ligases BB and DA2. In a feedback mechanism, DA1 then cleaves BB and DA2, leading to their destabilization. DA1-mediated cleavage of BB exposed a destabilizing N-terminal that was substrate for the N-end rule E3 ligase PROTEOLYSIS 1 (PRT1). This mechanism is predicted to transiently activate DA1 peptidase, which also cleaves of UBP15, TCP15 and the related TCP22, leading to their predicted inactivation and de-stabilization. DA1 peptidase may therefore contribute to the concerted transition from cell proliferation to endoreduplication and differentiation, limiting organ size.

Genetic and physical interactions of DA1, BB and DA2.
We previously identified genetic interactions between the da1-1 allele of DA1 and genes encoding the RING E3 ligases BB (Li et al. 2008) andDA2 (Xia et al. 2013) that led to synergistic increases in seed and organ sizes. In this study we use the da1-1 enhancing allele of BB called eod1-2 (Li et al. 2008) and refer to the mutant version as bb-eod1-2 and the wild-type version of as BB. The da1-1 allele, an R358K 5 change in a highly conserved region, had a negative influence on the functions of DA1 and the close family member DAR1, but the basis of this was not known, which complicated interpretation of DA1 function. We therefore assessed phenotypes of a loss-of function T-DNA allele of DA1 (da1-ko1).
Measurements of petal and seed sizes using high-resolution scanning showed that the da1-ko1 T-DNA allele led to increased petal and seed sizes (Fig. 1A,B), and it also interacted genetically with the loss-of-function allele bb-eod1-2 and da2-1 in both petal size and seed area. This showed that DA1 can be studied independently of other DA1 family members. Both the da1-1 and bb-eod1-2 mutations increased the maximum growth rate, while the double mutant da1-1 bb-eod1-2 showed a further increased maximum growth rate and continued to growth for approximately 5 days longer than either single mutant (Fig. 1C). The time at maximum growth rates was slightly earlier in bb-eod1-2 than Col-0, in contrast to da1-1 and da1-1 bb-eod1-2, which showed a 3-day retardation of the time of maximum growth rate, and final leaf sizes showed a more than additive increase in the double mutant, as observed previously (Li et al. 2008). These data indicated that BB may influence leaf final size at earlier stages of growth than DA1. We previously demonstrated that DA1 and DA2 physically interact (Xia et al. 2013). Pull-down experiments showed that GST-tagged DA1 also interacted with HIS-tagged BB, but not with HIS-tagged BBR (Big Brother Related, At3g19910), a close homolog of BB (Breuninger and Lenhard 2012) (Fig.   1D). These in vitro interactions were verified by Agrobacterium-mediated coexpression of BB-GFP and Myc-tagged DA1 in Nicotiana benthamiana leaves. Myc-DA1 was only detected in a complex with BB-GFP, and not GFP (Fig. 1E).

DA1 is multiply ubiquitylated by BB and DA2
The interactions of DA1 with BB and DA2 suggested that DA1 might be a substrate of these RING E3 ligases, so we conducted in vitro ubiquitylation reactions using BB, 6 DA2 and BBR (BB-Related) E3 ligases. Fig. 2A shows that BB ubiquitylated DA1 in an E1-and E2-dependent reaction, as did DA2 (Fig. 2B), while BBR did not (Fig.   2C). The extent of DA1 ubiquitylation suggested that DA2 was more efficient at ubiquitylation than BB, and the sizes of ubiquitylated DA1 indicated that between 4-7 ubiquitin molecules may be conjugated to DA1. Mass spectrometric analyses of ubiquitylated DA1 prepared in vitro was used to identify peptides containing the characteristic di-glycine ubiquitylation signature of a lysine residue (KGG). Analysis of DA1 ubiquitylated by DA2 or BB identified seven ubiquitylated lysine residues in DA1, with 4 lysines in the C-terminal domain of DA1 (K381, K391, K475 and K591) consistently conjugated with ubiquitin (Supplemental Fig. S1). This number of ubiquitylation sites concurred with the patterns of ubiquitylation observed in Fig.   2A,B, suggesting that DA1 molecules are multiply ubiquitylated (Haglund et al. 2003;Komander and Rape 2012). Mutation of the consistently ubiquitylated lysines to arginine in DA1 (termed DA1(4K-4R)) did not reduce ubiquitylation by DA2 in vitro ( Fig. 2D), and mass spectrometric analyses showed ectopic ubiquitylation of other lysines across DA1 (Supplemental Fig. S2). Therefore, the DA1 ubiquitylation mechanism has a preference, but not specificity, for certain lysines. These patterns of ubiquitylation are shown in Fig. 2D. DA1 and four other family members have multiple UIMs that interact with ubiquitin (Li et al. 2008;Peng et al. 2015). UIMs are part of a larger class of Ubiquitin Binding Domains (UBD) formed from a single α helix that is often found in multiple arrays (Hicke et al. 2005;Husnjak and Dikic 2012). Tandem UIMs have been shown to bind K63-linked ubiquitin chains in the mammalian DNA repair protein RAP80 (Sato et al. 2009). To assess their function in DA1, the N-terminal region of DA1 containing mutated UIM1 and UIM2 was fused to GST and expressed in E. coli, and conserved Ala and Ser residues, predicted to be in the α helical domain of the UIMs (Kim et al. 7 2007) (Supplemental Fig. S2), were mutated to Gly in both UIMs of GST-UIM1+2 and in DA1. GST-UIM1+2 bound ubiquitin and mutation of UIM1 alone did not reduce binding of ubiquitin, while mutation of UIM2 abolished ubiquitin binding, confirming that the GST-UIM1+2 protein bound ubiquitin via its UIM motifs (Supplemental Fig.   S2). Fig. 2E showed UIM1+2 conferred BB-and DA2-dependent ubiquitylation on GST in vitro, with DA2 again facilitating higher levels of ubiquitylation. Fig. 2F showed that mutation of both UIM1 and UIM2 in GST-UIM1+2 strongly reduced in vitro ubiquitylation of GST-UIM1+2 by BB and DA2. The UIM1 and UIM2 mutations in DA1 also reduced its ubiquitylation in vitro (Fig. 2G), and DA1 with mutated UIMs did not complement the large petal size in the double mutant da1-ko dar1-1 by (Fig. 2H).
To detect ubiquitylation in vivo, DA1 was expressed from the constitutive 35S promoter as an N-terminal GFP fusion protein and purified from seedling tissues using a GFP-Trap. Characteristic patterns of DA1 ubiquitylation were detected on purified GFP-DA1 (Fig. 2I, right panel). Therefore DA1 is ubiquitylated by the E3 ligases BB and DA2 in vitro by a UIM1 and UIM2 -dependent mechanism, DA1 is ubiquitylated in vivo, and UIMs were required for DA1 function.

DA1 cleaves BB and DA2 with a ubiquitin-dependent peptidase activity
A time-course of BB-HIS incubated with purified FLAG-DA1 that had been ubiquitylated by BB, or incubated with non-ubiquitylated FLAG-DA1, showed that in the presence of ubiquitylated DA1 a HIS-tagged BB fragment of approximately 35 kDa was produced after 4 h incubation (arrowed in Figure 3A). When ubiquitylated FLAG-DA1 was incubated with DA2-HIS, a 25 kDa HIS-tagged DA2 cleavage product was also detected after 4 h incubation (arrowed in Fig. 3A). Similar experiments using FLAG-DA1 ubiquitylated by DA2 showed identical patterns of BB-HIS and DA2-HIS cleavage (Fig. 3B). BBR-HIS did not show a cleavage product in these conditions. Thus DA1, ubiquitylated by either BB or DA2, generated cleavage products from both BB and DA2 in vitro.

8
Examination of the conserved C-terminal region of DA1 revealed an extended sequence motif, HEMMHX 15 EE (Supplemental Fig. S3), which is a zinc aminopeptidase active site found in clan MA endopeptidases (Rawlings et al. 2012).
The HEMMH motif was mutated to AEMMA, removing the putative Zn-coordinating histidine residues, to form DA1(pep).  Figure 3E) was required to identify the 25 kDa DA2-FLAG cleavage product, which was not generated by co-expression with DA1(pep). Fig. 3F shows the mutation in DA1 abolishing DA1 peptidase activity did not complement the da1-ko1 dar1-1 large petal phenotype, establishing that DA1 peptidase activity is required for in vivo function. To detect DA1 peptidase activity in vivo, transgenic plants expressing a BB::gsGreen-BB fusion protein, and a RING-domain mutant version that was predicted to be more stable in vivo due to reduced auto-polyubiquitylation (Disch et al. 2006a) were generated. Analysis of GFP-Trap purified proteins (Fig. 3G, left panel) showed a cleavage product of the expected size generated from RINGmutant gsGreen protein in two independent transformants. Full-length wild-type gsGreen-BB was not detected, although low levels of an expected cleavage product were identified. For comparison, the same constructs, together with a non-cleavable form (AY-GG, see Fig. 5B) were expressed using the 35S promoter in protoplasts with DA1 (Fig. 3G, right panel). This showed the predicted DA1-mediated BB cleavage product, which was not generated in the AY-GG version of BB.

9
A Förster Resonance Energy Transfer (FRET) DA1 peptidase sensor was constructed using eGFP donor and mCherry acceptor pairs (van der Krogt et al. 2008) connected by BB to provide another measure of DA1 peptidase activity in vivo.
Cleavage of the fluorophore pair by DA1 would increase the fluorescence lifetime towards that of eGFP-BB compared to that of the intact sensor protein by impairing energy transfer between the fluorophores. The peptidase sensor and a control donor sensor were transfected into da1-ko1 dar1-1 root protoplasts and Fluorescence Lifetime Imaging (FLIM) was performed. Fig. 4A shows that the fluorescence lifetime (τ) of the GFP-BB donor control was approximately 2.48 ns, while that of an intact donor-acceptor pair was approximately 2.25 ns, demonstrating efficient FRET. When co-transfected with DA1, the fluorescent lifetime of the donor-acceptor pair increased to approximately 2.38 ns. Lifetime imaging of typical transfected protoplasts showed a generalized cellular localization of DA1-mediated cleavage. Fig. 4B showed the eGFP-BB-mCherry donor-acceptor pair was cleaved by DA1 peptidase at the expected site in transfected root protoplasts. Therefore, DA1 has a latent peptidase activity that is activated by multiple ubiquitylation mediated by its UIM 1+2 domain and the RING E3 ligases BB and DA2, and activated DA1 peptidase then specifically cleaves these two E3 ligases.

Identification of a DA1 peptidase cleavage site in BB
To define the potential functions of DA1-mediated cleavage, the DA1 cleavage site in BB was identified using Edman sequencing of purified cleaved BB-HIS.
Supplemental Fig. S4 shows neo-N terminal amino acid sequences that had a unique match to six amino acids in BB (Fig. 5A). This indicated a potential DA1 cleavage site within BB between A 60 and Y 61 , consistent with the sizes of BB and its ca. 35 kDa cleaved form (Fig. 3A). Two mutant forms of BB were made to assess this potential DA1 cleavage site: a four amino acid deletion surrounding the site 10 (ΔNAYK); and changing AY to GG (AY-GG) (Fig. 5B). These proteins were coexpressed in Arabidopsis da1-ko1 dar1-1 mesophyll protoplasts as C-terminal FLAG fusion proteins with HA-DA1 and HA-DA1(pep). Fig. 5B showed that the mutant BB-FLAG proteins were not cleaved by DA1, establishing that DA1 peptidase activity cleaved BB between A 60 and Y 61 . A cleaved form of BB called MY61-BB was also made with an initiator Met followed by Y 61 (Fig. 5B). MY61-BB was expressed using the 35S promoter in da1-ko1 bb-eod1-2 mutant Arabidopsis. Its lack of complementation of bb-eod1-2 (Fig. 5C) showed that DA1 peptidase-mediated cleavage reduced BB activity.

BB stability is dependent on it N-terminus and N-end rule function
DA1 cleavage products of DA2 were unstable, indicating that one function of DA1mediated cleavage may be to destabilize proteins (Fig. 3E). This was also observed for BB in cell-free degradation assays, in which MY61-BB was unstable compared to wild-type BB (Supplemental Fig. S5). To test the role of the neo-N terminus of BB on protein stability, -61BB proteins with different N-termini (Y, G, MY) were expressed using the ubiquitin fusion technique (UFT) (Bachmair et al. 1986). HA-tagged constructs were translationally co-expressed in a cell-free rabbit reticulocyte system, with or without MG132 proteasome inhibitor, and translation stopped by the addition of cycloheximide. Y61-BB was highly unstable, whereas G61-BB was stable (Fig.   5D). Interestingly, the artificial MY61-BB was also highly unstable in a proteasomeindependent mechanism. The neo-N terminal sequence of DA1-cleaved BB starts with YK, a potentially destabilizing sequence of a type II N-end rule degron (Varshavsky 2011). The N-end rule E3 ligase PROTEOLYSIS 1 (PRT1) mediates stability of model N-end rule substrates with such aromatic amino-terminal residues (Potuschak et al. 1998). To assess the potential role of PRT1 in N-end rule mediated degradation of BB, we tested the binding of PRT1 to 17-mer peptides representing 11 variants of the neo-N termini of BB on a backbone sequence of an N-end rule test substrate in SPOT assays (Synthetic Peptide arrays On membrane support Technique). Purified recombinant HIS-MBP:PRT1 protein was incubated with the SPOT array and binding visualized by Western blotting. Recombinant PRT1 had a preference for binding to the large aromatic acids tyrosine and phenylalanine, consistent with previously suggested specificity (Fig. 5E) (Potuschak et al. 1998;Stary 2003;Faden et al. 2016). To assess whether PRT1 had a role in DA1-mediated BB degradation, BB was expressed with an N-terminal ubiquitin fusion and a Cterminal luciferase fusion to reveal neo-N termini in Col-0 or prt1 mutant mesophyll protoplasts. BB-luciferase (LUC) activity was reduced in wild-type protoplasts with a neo-N terminal tyrosine, which was not seen in prt1 mutant protoplasts (Fig. 5F).
Neo-N terminal glycine BB-LUC levels were not altered in either Col-0 or prt1 mutant protoplasts. This indicated a strong dependence of Tyr-61BB stability on PRT1 activity. In planta evidence supporting the role of DA1 in reducing the growth inhibitory role of BB via N-end rule mediated degradation was shown by the suppression of growth reduction in a transgenic line over-expressing -RFP-BB from the 35S promoter by over-expression of DA1 (Fig. 5G). Western blots (Fig. 5H) confirmed that over-expression of DA1 from the 35S promoter reduced RFP-BB protein levels.

Functional analyses of DA1
We previously showed that the da1-1 allele of DA1 has a negative interfering phenotype with respect to the closely related family member DAR1 (Li et al. 2008).
The peptidase activity of the protein encoded by the da1-1 allele, called DA1(R358K), which has an arginine to a lysine residue altered in a highly conserved C terminal region (Supplemental Fig. S4) was assessed. This mutation did not influence ubiquitylation of FLAG-DA1(R358K) (Fig. 6A) and neither did it create a site for ectopic ubiquitylation of FLAG-DA1(R358K) as determined by mass spectrometric analysis (Supplemental Fig. S1C). The peptidase activity of ubiquitylated FLAG-DA1(R358K) was qualitatively assessed in vitro and in vivo (using HA-DA1(R358K)) by comparison to wild-type DA1 peptidase activity (Fig. 6A,B). Both assays showed that DA1(R358K) had lower peptidase activity compared to DA1, suggesting regions of the conserved C-terminal region are required for peptidase activity and that the da1-1 phenotype may be due to reduced peptidase activity. Fig. 6B also shows that DA1(4K-4R) which is ubiquitylated (Fig. 2E), had peptidase activity towards BB. This suggested that precise patterns of ubiquitylation are not required for activating DA1 latent peptidase activity. The chs3-2d mutation changes a Cys to a Tyr in the third pair of conserved CxxC/H motifs (Supplemental Fig. S3 and S6), suggesting it may alter a possible LIM-like structure. This mutation was introduced into DA1 to create DA1(C274Y) and its activities assessed. Fig. 6A shows that DA1(C274Y) was not ubiquitylated by BB, and had no peptidase activity towards BB in vitro and in vivo (Fig. 6B). This implicated the putative LIM-like domain in DA1 in UIM-mediated ubiquitylation and activation of DA1 peptidase activity.

DA1 peptidase activity cleaves TCP15, TCP22 and UBP15.
The increased levels of UBP15 (Du et al. 2014), and TCP14 and TCP15 proteins (Peng et al. 2015) observed in the da1-1 mutant suggested that DA1 activity may 13 reduce the stability of these proteins by peptidase-mediated cleavage. Fig. 7A showed that DA1 peptidase cleaved UBP15 close to its C-terminus when transiently expressed together in da1-ko1 dar1-1 protoplasts. The reduced signal in the western blot with the C-terminal FLAG fusion was due to the short FLAG-tagged protein running off the gel. TCP15 and the closely related TCP22 proteins were cleaved by DA1 in protoplasts (Fig. 7B), but we could not consistently detect TCP14 cleavage by DA1 or DAR1. These data show that DA1 peptidase activity can cleave UBP15, which promotes cell proliferation, and TCP15, which inhibits endoreduplication (Du et al. 2014;Peng et al. 2015) .

DA1 is a latent peptidase activated by multiple ubiquitylation
The addition of ubiquitin molecules to substrate proteins is a common posttranslational modification with many regulatory roles (Hoeller and Dikic 2010;Haglund and Stenmark 2006). We showed that DA1 was ubiquitylated at four lysine residues (Figures 2A-2D, S2) by the E3 ligases BB and DA2, which genetically (Fig.   1A,B) and physically ( Figures 1D,E) interact with DA1. Similar levels of DA1 ubiquitylation were observed in vivo (Fig. 2I). Loss of function mutants of BB and DA2 synergistically increased the large organ phenotypes of a DA1 loss of function mutant (Fig. 1A,B) (Li et al. 2008;Xia et al. 2013). Therefore ubiquitylation mediated by these two E3 ligases may increase the growth-limiting activity of DA1. Quantitative growth measurements of leaves (Fig. 1C) showed that individually the da1-1 mutant and bb-eod1-2 mutants had distinctive effects on leaf growth; da1-1 had a delayed time of maximum growth rate, while bb-eod1-2 showed a slightly accelerated time of maximum growth, compared to Col-0. In combination the mutants showed an even greater maximum growth rate and delayed time of maximum growth rate, increasing 14 final organ size. BB and DA2 act independently to influence final organ size (Disch et al. 2006b), but they both also interact with DA1 (Li et al. 2008;Xia et al. 2013) (Fig.   1A,B), indicating that BB and DA2 may ubiquitylate different substrates, perhaps for proteasomal-mediated degradation. Quantitatively, loss of function mutations in BB and DA2 show their independent effects on growth are less than that of da1-1 (Fig.   1A,B).
Members of the DA1 family have a canonical MA clan zinc metallopeptidase active site domain in their conserved C terminal region (Supplemental Fig. S3) (Rawlings et al. 2012;Tholander et al. 2010) that was required for limiting growth (Fig. 3F). BBand DA2-mediated ubiquitylation activated DA1 peptidase activity in vitro and in transiently expressed protoplasts (Fig. 3A,E), establishing a biochemical foundation for their joint activities in growth control. Zinc metallopeptidases are maintained in an inactive form by a "cysteine switch" (Van Wart and Birkedal-Hansen 1990) that coordinates a cysteine residue with the zinc atom at the active site to block it.
Conformational changes release this and activate the peptidase.
Ubiquitylation of DA1 has the potential to trigger a conformational change that may release inhibition of peptidase activity. (Hoeller et al. 2006) showed that UBD-and UIM-mediated monoubiquitylation of endocytotic proteins including epsin led to a conformational change, mediated by intramolecular interactions between UBDs/UIMs and cis-ubiquitin, which regulated endocytosis. The binding of ubiquitin to DA1 UIMs was required for DA1 function in vivo (Fig. 2H), and the UIMs conferred similar patterns of ubiquitylation on the heterologous protein GST as seen for DA1 (Fig. 2G for DA1 and Fig. 2E,F for GST-UIM1+2). Related observations were seen in the monoubiquitylation of epsin (Oldham et al. 2002) through coupled monoubiquitylation (Woelk et al. 2006), where UIMs recruit the UIM-containing protein to the ubiquitylation machinery by direct interaction with ubiquitin coupled to ubiquitin donor proteins (Haglund and Stenmark 2006). Mutation of cysteine 274 in the C-terminal zinc finger loop of the LIM-like domain of DA1 abrogated both ubiquitylation and peptidase activity (Fig. 6A,B), suggesting a functional role for this ancient conserved LIM-like domain (Burroughs et al. 2011) (Supplemental Fig. S7) in peptidase activation. Analyses of conformational changes caused by DA1 ubiquitylation and their influence on peptidase activity are required to establish this potential mechanism.

DA1 cleavage destabilizes its activating E3 ligases BB and DA2, and cleavage of BB leads to targeting by the N-recognin PRT1.
The RING E3 ligases BB and DA2 activate DA1 peptidase by ubiquitylation, and are also cleaved by DA1 peptidase (Fig. 3A,B,E,G and Fig. 4). Once cleaved, DA2 appeared to be destabilized in transiently expressed protoplasts (Fig. 3E).
Identification of the DA1 cleavage site in BB (Fig. 5A,B) revealed Y61-BB at the neo-N terminus of cleaved BB. This neo-N terminus conferred proteasome-mediated degradation in a cell free system (Fig. 5D). This degradation depended on recognition of the neo-N terminus by the Arabidopsis E3 ligase PRT1 (Fig. 5E,F) (Potuschak et al. 1998;Stary 2003), an N-recognin catalyzing N-end rule mediated degradation (Varshavsky 2011) with a suggested preference for aromatic amino acid N-termini. Interestingly, the neo-N terminal MY61-BB, which was used to express a cleaved version of BB in planta, conferred strong proteasome-independent instability ( Fig. 5D), in a mechanism that is not yet clear. The lack of MY61-BB function in vivo ( Fig. 5C) supported the observation that DA1-mediated cleavage of BB leads to its loss of function in vivo. Over-expression of BB strongly reduced growth, as expected from its inhibitory role in growth (Disch et al. 2006a). The reversal of this inhibition by over-expression of DA1, which reduced RFP-BB levels (Fig. 5H) and reversed growth inhibition (Fig. 5G), is consistent with a mechanism involving DA1-mediated reduction of BB activity via peptidase-mediated cleavage and subsequent degradation by the N-end rule pathway. Such an activation-destruction mechanism mediated by BB, DA2 and DA1 may provide a way of tightly controlling peptidase activity. These types of mechanisms, often involving ubiquitylation and proteolytic degradation, drive uni-directional cellular processes, for example in cell cycle progression (Reed 2003).

DA1 peptidase activity also cleaves diverse growth regulators
We previously showed that TCP14 and TCP15 function downstream of DA1 and other family members in controlling organ size in Arabidopsis, and reduced function of DA1 family members led to increased TCP14 and TCP15 protein levels (Peng et al. 2015). Similarly, levels of UBP15 protein, which promotes cell proliferation (Lui et al., 2008) and also functions downstream of DA1, were increased in the da1-1 reduced function mutant (Du et al., 2014). We showed that TCP15 and the related TCP22, and UBP15, were cleaved by DA1 peptidase activity (Fig. 7A,B), but we could not reliably detect TCP14 cleavage by DA1 or DAR1. DA1-mediated cleavage of TCP15 and UBP15 is a plausible mechanism that accounts for these observed reduced protein levels, similar to DA1-mediated inactivation and destabilization of BB by peptidase cleavage. Taken together, these observations suggest a mechanism ( Fig. 7C) in which DA1 peptidase, activated transiently by BB or DA2, coordinates a "one-way" cessation of cell proliferation and the initiation of endoreduplication through the cleavage and potential inactivation of proteins that promote cell proliferation and inhibit endoreduplication.

17
Arabidopsis thaliana Columbia (Col-0) was the wild-type plant used. Plants were grown in growth-rooms at 20 o C with 16 h day/8 h dark cycles, using either soil or MS medium supplemented with 0.5% glucose. Petal and seed areas were imaged by high resolution scanning (3600dpi: Hewlett Packard Scanjet 4370) and analyzed using ImageJ software (http://rsbweb.nih.gov/ij/).

In vitro DA1-mediated Cleavage Assays
FLAG-DA1 was ubiquitylated in vitro using either DA2-HIS or BB-HIS as E3 ligases, purified using FLAG-magnetic beads, quantified, and 100 ng added to 100ng BB-HIS, DA2-HIS or BBR-HIS in a 30 µl reaction in 50mM Tris HCl pH 7.4, 5 mM MgCl 2 .
Reactions were carried out at 30 o C for 4 h and terminated by the addition of SDS sample buffer.  Expressed proteins were purified using GFP-trap and immunoblotted.   Transformation efficiency was measured using pUBC::GUS control plasmid.

Mass Spectrometry Analysis
Luciferase activities were normalized to GUS activity, and activity of Gly-61BB-HA-LUC was taken as 100%. Three independent transformation experiments were conducted. The significance of differences was calculated using Student's t -tests (two sites, uncoupled). * p-value ≤ 0.05.