Fine-scale dissection of the subdomains of polarity protein BASL in stomatal asymmetric cell division

Highlight Analysis and manipulation of BASL subdomains reveal that this polarity protein integrates multiple regulatory inputs to promote difference during stomatal lineage asymmetric cell division.


Introduction
Asymmetric cell division (ACD) plays a major role during the development of multicellular organisms (Horvitz and Herskowitz, 1992). Stem cells, in particular, divide asymmetrically to renew themselves while simultaneously generating cells that will differentiate. Plant development features continual post-embryonic generation of tissues and organs from a limited number of spatially restricted stem cells (De Smet and Beeckman, 2011). The molecular mechanisms underpinning plant ACDs have been a subject of active investigation in the past decades, but much of the work focused on fate asymmetry in daughter cells, rather than the initiation of polarity and asymmetry in the mother (Petricka et al., 2009;De Smet and Beeckman, 2011;Kajala et al., 2014).
A good system in which to investigate asymmetry is the stomatal lineage of Arabidopsis. Here, protodermal cells in the epidermis undergo ACDs to maintain the self-renewing precursors of the stomatal guard cells (Lau and Bergmann, 2012). Guard cells are used to modulate the opening and closing of stomata so that plants can acquire CO 2 from the atmosphere while minimizing water loss. During stomatal development, the basic helix-loop-helix (bHLH) transcription factor SPEECHLESS (SPCH) together with its partner bHLHs, the SCREAM factors (SCRM/ICE1 and SCRM2), functions as a master regulator of the ACDs that initiate and maintain the lineage (MacAlister et al., 2007;Kanaoka et al., 2008). SPCH controls the expression of a wide range of downstream genes to specify stomata lineage cell fate, cell polarity, and cell-to-cell communication (Lau et al., 2014).
Among SPCH targets are the polarly localized proteins BREAKING OF ASYMMETRY IN THE STOMATAL LINEAGE (BASL) (Dong et al., 2009) andPOLAR (Pillitteri et al., 2011). The localization of these proteins is intriguing, but their roles as regulators of division asymmetry are much less understood, in part because the plant proteins show no similarity to the intrinsic polarity proteins (such as the PARs) conserved among animals (Knoblich, 2008;Inaba and Yamashita, 2012). In fact, BASL and POLAR encode plant-specific proteins of unknown biochemical functions. Recent studies, however, suggest that BASL functions as a scaffold protein by interacting with the Mitogen-activated protein kinase (MAPK) kinase kinase YODA (MAPKKK YDA) and the MAPKs MPK3 and MPK6, pivotal negative regulators of stomatal development (Bergmann et al., 2004;Wang et al., 2007). BASL ensures that MAPK activity is preferentially inherited by the larger daughter of an ACD; because SPCH is negatively regulated by MAPK activity, the differential MAPK signalling in two daughter cells leads to preservation of SPCH stem cell activity in one daughter but not the other (Zhang et al., 2015).
Within an asymmetrically dividing cell, BASL is localized to two distinct subcellular compartments, the plasma membrane (PM) and the nucleus (Dong et al., 2009). BASL's subcellular localization is linked to cell behaviour, and the current model, based on observations of many BASLexpressing stomatal lineage cells, is that cells expressing BASL dually in the nucleus and at the PM undergo another ACD, whereas cells expressing nuclear-only or peripheralonly BASL differentiate into stomatal guard cells or epidermal pavement cells, respectively (Dong et al., 2009). Therefore, proportioning BASL between the nucleus and the PM has phenotypic consequences and must be strictly controlled. In this study, we focused on intrinsic features of the BASL protein that lead to its localization. By detailed characterization of its subdomains and by manipulating their subcellular localization, we address several questions about this intriguing protein: (i) Which sequences in BASL regulate its nucleocytoplasmic trafficking? (ii) What motifs in BASL are critical for its ability to be polarized? and (iii) What domains might mediate protein-protein interactions critical for BASL function? Our results provide new insights into the basic properties of the novel polarity protein BASL and enlighten future research directions towards understanding the mechanisms for cell polarity and ACD in plants.

Plant materials and growth conditions
Arabidopsis thaliana seeds were stratified for 1-2 days in the dark at 4°C. Seedling were germinated and grown in Petri dishes containing 1% of Phytoblend (Caisson Laboratories) and 0.5× Murashige & Skoog basal salts (Caisson Laboratories) at 22°C with long-day cycles (16-h light/8-h dark). The ecotype Columbia (Col-0, Col) was used as the wild type (WT) for transgenic studies. Mutants used in this study, basl-2 (WiscDsLox264F02) and basl-3 (SAIL_547_F11), were obtained from the Arabidopsis Biological Resources Center at the Ohio State University. basl-3, but not basl-2, is sensitive to Hygromycin selection, so was used as a genetic background for pMDC43 and R4pGWB510 complementation assay.

Plasmid construction and plant transformation
The Gateway cloning technology (Invitrogen) was used for most DNA manipulations. The coding regions of BASL and specific subdomains were amplified by Phusion high-fidelity enzyme (New England BioLabs) with appropriate primers (published in Dong et al. (2009) and listed in Supplementary Table S1 at JXB online) and sub-cloned into the pENTR/D-TOPO vector (Invitrogen). To introduce point mutations, pENTR/D-BASL was used as a template and specific mutations were introduced by a QuickChange II XL site-Directed Mutagenesis Kit (Agilent Technologies). The binary vector pMDC43 (Curtis and Grossniklaus, 2003), in which the 35S promoter was replaced with the BASL promoter, was used for expressing GFP-tagged full-length BASL and BASL_N, _I, _I2, _C, _NI, and _IC domains. To generate myristoylated GFP-BASL, the GFP fragment in the pMDC43 (containing the BASL promoter driving GFP-BASL) was replaced with Myr-GFP. The other orientation of Myr-BASL-GFP was generated by initial cloning into pENTR/D-BASLpro::Myr-BASL and subdomains, followed by recombination into pMDC107 (Curtis and Grossniklaus, 2003). To express BASL_NLSd, _NESd, and _NLSd/NESd in plants, R4pGWB510 (Nakagawa et al., 2008) was used to recombine with pENTR/D-YFP-BASL variants and pDONR P4-P1r-BASLpro (Zhang et al., 2015). All BASL-related constructs were driven by the BASL promoter for consistent experimenting. Plant transformations were carried out with the Agrobacterium (strain GV3101)-mediated floral dipping method (Zhang et al., 2006). Subsequent selection of transgenic plants followed standard protocols.

RT-PCR
Total RNAs were extracted from 7-days-post-germination (dpg) seedlings using an RNeasy Plant Mini Kit (Qiagen). The first-strand cDNAs were generated from 3 µg of total RNAs in a total volume of 20 µL reaction using the SuperScript TM First-Strand Synthesis System (Invitrogen). PCR reactions were carried out using 0.5 µL of the reverse transcription resultant for a total volume of 20 µL. Gene-specific primers were used to evaluate the endogenous expression level of BASL and 18s rRNA (primers listed in Supplementary Table S1 at JXB online) for 35-cycle or 25-cycle PCR reactions, respectively. 10 µL of PCR product from each were loaded for DNA electrophoresis.

Quantification of stomatal development and patterning
The quantification protocol followed the previous publications (Dong et al., 2009;Zhang et al., 2015) with minor modifications. In general, more than 12 independent T2 transgenic lines were maintained for each transgene. For each transgenic event, two representative lines with stable YFP/GFP expression, and single insertions preferred, were established and pooled together for quantification analysis. High 'stomatal index' (number of stomata/total number of epidermal cells expressed as a percentage) describes stomatal over-production, and 'clustered stomata' (more than two stomata in direct contact) indicates the patterning defects [overriding the 'one cell-spacing rule' (Geisler et al., 2003)]. Complementation of the basl phenotype was scored in dissected cotyledons of 5-dpg seedlings stained with PI. Similar central regions in the adaxial cotyledons were examined and imaged with an EC Plan-Neofluar (20×/0.5) lens on a Carl Zeiss Axio Scope A1 fluorescence microscope equipped with a ProgRes MF CCD camera (Jenoptik). Typically, about 1500-2000 epidermal cells were counted for phenotyping. Stomatal cells were scored as one of the four categories: pavement cells, stomata, clustered stomata, and small dividing cells as in Dong et al. (2009). Two key phenotypes of basl mutants (high stomatal index and clustered stomata) were used to evaluate the efficacy of transgenes in complementation assays. The numbers were subjected to normal probability tests and the OriginLab analysis, followed by the Mann-Whitney test (data point < 30).

Quantification of BASL variant polarization
Quantification of protein polarity degree was described in Zhang et al. (2015) with slight modifications: 3-dpg seedlings were counterstained with PI before mounting and examined on the confocal microscope. A total of 30 GFP-positive cells from at least five cotyledons were selected for polarity quantification. Along the PI-stained cell periphery of the stomatal ACD cells, GFP intensity was measured from two equal-sized regions (covering about 25% of the perimeter) at the PM that expressed high or low GFP, respectively. Uneven protein distribution was measured by the ratio of the high GFP intensity value over the low intensity value. Typically, measurements from 30 cells were subjected to the Student's t test (data point ≥30) using OriginLab software.

BASL nuclear localization requires the nuclear localization signals, but export is nuclear export signal independent
Shortly after mitosis in early stomatal lineage cells, GFP-BASL expression is visible in the nucleus, and over time accumulates at the PM (Dong et al., 2009). We previously showed that a mutant form (phospho-deficient) of BASL protein unable to be exported from the nucleus fails to rescue the basl mutant (Zhang et al., 2015); therefore, we looked for BASL motifs that might mediate its partitioning between the nucleus and at the PM. In principle, cytoplasmically synthesized proteins are imported into the nucleus via the nuclear localization signals (NLSs) of the protein. We conducted NLS prediction using the cNLS mapper (Brameier et al., 2007) (http://nlsmapper.iab.keio.ac.jp/cgi-bin/NLS_Mapper_form.cgi) and found that BASL contains two putative monopartite NLSs (positioned at amino acids 51-60 and 61-68, shown in red in Supplementary Fig. S1A at JXB online). We constructed YFP-BASL_NLSd (Fig. 1A, missing both NLSs, amino acids 51-64 deleted) for a localization and complementation assay. In contrast to YFP-BASL, which recapitulated the localization pattern of the previously published GFP-BASL (Dong et al., 2009) and complemented the loss-of-function basl-2 mutant (Fig. 1B, C), YFP-BASL_NLSd did not show obvious nuclear enrichment, but instead was diffuse in the cytoplasm and polarized at the cell cortex (Fig. 1D). DAPI nucleic acid staining confirmed that YFP-BASL_NLSd was not retained in the nucleus (Fig. 1E). Interestingly, YFP-BASL_NLSd was sufficient to rescue basl-2 at the scale of our phenotypic measurements, including stomatal production (stomatal index) and patterning (clustered stomata) ( Fig. 1B versus Fig. 1D, quantification in Fig. 1F and see Supplementary Fig. S1B, C at JXB online). These data are consistent with our previous finding that BASL functioned primarily at the cortical polarity site (Dong et al., 2009).
Nucleocytoplasmic shuttling often requires a leucine-rich nuclear export signal (NES), which is recognized by the importin family to cross the nuclear envelope (la Cour et al., 2004). The NetNES predictor (http://www.cbs.dtu.dk/services/NetNES-1.1/) suggested the possible existence of a weak NES in BASL (see Supplementary Fig. S1 at JXB online, amino acid position 65-71, scored at 0.4-0.5). We deleted this region to generate YFP-BASL_NESd. Confocal examination showed that, even without the putative NES, BASL protein still maintained its typical localization and function in plants ( Fig. 1F, G). Simultaneous deletion of the NLSs and NES, YFP-BASL_NLS/NESd, produced a protein mostly mimicking YFP-BASL_NLSd in localization and function (Fig. 1F, H). Together, our results indicate that the NLSs do promote nuclear localization, but export of BASL does not rely on its putative NES. Further, nuclear retention of BASL is not absolutely required for its peripheral polarity or for its function in stomatal ACDs.

BASL cortical polarization requires conserved motifs within the C-terminal domain
Perhaps the most striking property of BASL is its polarized cortical localization in asymmetrically dividing stomatal cells ( Fig. 2A, 90.3% GFP-positive cells show protein polar accumulation, n = 144 in 2-dpg seedlings). We previously demonstrated that the C-terminal two-thirds of BASL (amino acids 95-262, BASL_IC) was sufficient for cortical localization, for polarization, and for function ( Fig. 2B-D, 88.4% cells show GFP polar accumulation, n = 129), but that regions I or C alone were not (Dong et al., 2009). We further dissected BASL_IC, creating a fragment BASL_I2 (amino acids 135-216) that overlapped the junction between I and C (Fig. 2B). This fragment was not sufficient to polarize or rescue basl-2 (Fig. 2D, E). These data suggested that BASL polarization requires information from the regions beyond BASL_I but within BASL_IC, including amino acids 95-135 and/or 216-262 (Fig. 2B). Because our previous data showed BASL_NI (amino acids 1-180, containing 95-135) was not functional or polarized (see Supplementary Fig. S2A, B at JXB online), we focused our analysis on the carboxyl terminal region (amino acids 221-262, Fig. 2B).
Deletion of the C-terminal 41 amino acids (amino acids 222-262, GFP-BASL_d41) completely eliminated polarization and the protein accumulated instead in the nucleus and cytoplasm (Fig. 2F). Concomitantly, GFP-BASL_d41 was unable to complement basl-2 (Fig. 2D), indicating that the last 41 amino acids are indispensable. To identify critical motifs within this essential domain, we blasted the last 41 amino acids of BASL against the phytozome database (http://phytozome.jgi.doe.gov/pz/portal.html). This search revealed blocks of conservation with other plant proteins (Fig. 3A), particularly around two motifs: FAFPI/VL (FxFP) and CCR/KF (boxed in Fig. 3A). The FxFP motif can mediate interactions between MAPKs and their substrates (Jacobs et al., 1999) and CCR/KF motifs at the C-terminus can be palmitoylated to retain proteins at membranes (Tani and Hannun, 2007;Mattison et al., 2012). These were provocative motifs given the membrane association of BASL and the importance of MAPK regulation in stomatal development (Wang et al., 2007;Lampard et al., 2008;Lampard et al., 2009).
To test whether the FxFP motif (black box in Fig. 2B) was necessary or sufficient for BASL activity, we first extended BASL_d41 by nine amino acids so that the motif was were scored. This applies to other figures in which this format is used. Note the similar rescue effect of YFP-BASL_NLSd, _NESd, and _NLS/NESd in basl-2 compared to that of YFP-BASL. All BASL-related constructs were driven by the BASL promoter. (G-H) Confocal images of YFP-BASL_NESd (G) and YFP-BASL_NLS/NESd (H) (green) in basl-2 to show protein localization (green) and complementation (cells outlined with PI staining, magenta). BASL does not rely on the NES for nuclear export, but requires the NLS for nuclear import. Arrows indicate proteins polarized to cortical crescents. The insets show the YFP channel only. Scale bars = 25 µm.
The role of the CCR/KF motif is less clear. GFP-BASL_ d6 (missing this motif in the last six amino acids) was similar to WT in its ability to polarize and complement the mutant (Fig. 2H, D, polarity shown in 82.8% cells, n = 128). GFP-BASL_2C2S, with the final two cysteines replaced by serines (Fig. 3A), maintained polarization activity and rescued basl-2 mutants, comparable to the WT (Fig. 3D, quantification in Fig. 3C). Therefore, regarding BASL's localization at the PM, we have no evidence that direct palmitoylation of BASL drives membrane association.

Tethering BASL and subdomains to the plasma membrane
Deletions and site mutations are useful to assess the contribution of specific motifs to BASL localization and function, but the subsequent mislocalization often attenuates our ability to assay polarity at the PM. Artificially tethering the BASL protein or domains would explicitly demonstrate how parts of BASL might function at the PM. In addition, many models for polarization require protein movement on and off the membrane (Altschuler et al., 2008;Slaughter et al., 2009). By taking advantage of the myristoylation (Myr) modification that permanently attaches the target protein to the PM (Martin et al., 2011), we tested aspects of BASL polarization and function at the PM and its biological significance in the nucleus (summarized in Supplementary Table S2 at JXB online).
A Myr-targeted peptide sequence (Greenwood and Struhl, 1999) was fused to the N-termini of BASL and its variants. We found that the myristoylated full-length BASL did not show obvious nuclear expression and that both orientations, Myr-BASL-GFP and Myr-GFP-BASL, were somewhat polarly distributed along the PM (Fig. 4A, B, polarity quantification in Fig. 4C) and largely rescued basl-3 mutants (complementation quantified in Fig. 4D). basl-3 is indistinguishable from basl-2 in stomatal phenotypes (see Supplementary Fig. S2C-E at JXB online), but the basl-3 T-DNA confers an easier selection for the expression vectors used in this study, and therefore was equivalently used for complementation assays. These results suggest that the on-and-off membrane mechanism may not be necessary for BASL polarization and function. We then generated Myr-BASL_d41 and Myr-BASL_d32 and found both are localized to the PM, but no obvious polar distribution could be detected (Fig. 4E, F). Myr-BASL_d41, similar to BASL_ d41, did not complement the stomatal defects in basl-3. Myr-GFP-BASL_d32 was also not discernably polarized (in contrast to GFP-BASL_d32, Fig. 2G), but, interestingly, it partially rescued the basl-3 mutant (Fig. 4D). It is possible that the higher overall levels of membrane-associated Myr-BASL_d32 obscured the polarization of this construct, but the functional motif in BASL_d32, that is FxFP, was sufficient to contribute at least partially to protein activity.
We continued to utilize Myr as a tool to test BASL subdomains (Fig. 5A). The N-terminal domain (BASL_N) directs nuclear localization and cannot rescue basl-2 null mutants (Dong et al., 2009). When Myr-BASL_N was fused to GFP, it was found to be localized primarily in the cytoplasm (Fig. 5B) and a small number of cells even showed nuclear expression (arrowhead in Fig. 5B). Such localization might result from the antagonistic activity between the NLSs in BASL_N and the Myr modification. However, Myr-BASL_NI (with the addition of the I domain) mainly localized to the PM (Fig. 5C), indicating that BASL_I might promote the PM association. Previously we showed that BASL interacts with the MAPKKK YDA via three MAPK-docking domains (Zhang et al., 2015), but none of which resides in BASL_I (Fig. 6H). Therefore, we propose that BASL_I might interact with new partner/s to associate BASL with the PM.
Without the N domain, BASL_IC was mainly cytoplasmic but polarized efficiently at the cell cortex and it fully rescued basl mutants (see Supplementary Fig. S2B at JXB online) (Dong et al., 2009). But, once myristoylated, association of the BASL_IC with the PM lacked obvious polarization (Fig. 5D) and its ability to rescue the basl mutant was significantly impaired (Fig. 5E). The decreased complementation capability of Myr-BASL_IC relative to that of BASL_IC (Fig. 5F) might be explained by the greatly reduced polarization and/or the decreased partition in the cytoplasm, where the activating MPK3/6 are enriched (Zhang et al., 2015).
We also assayed other subdomains, namely, the cytoplasmic BASL_I and BASL_C (Dong et al., 2009). Our data showed that these Myr-versions were predominantly enriched at the PM (see Supplementary Fig. S2F, G at JXB online) and were not effective in complementing the basl phenotype (Fig. 5E).

Using tethered BASL subdomains to assay for putative interaction platforms
At the cortical polarity site, BASL scaffolds YDA and MPK3/6 and there is mutual positive reinforcement of polarity through interactions among the constituent members (Zhang et al., 2015). Loss of YDA did not fully abolish BASL polarity (Zhang et al., 2015), however, suggesting that other partners are required to polarize BASL. While identification of such partners is beyond the scope of this paper, hints of their target sites on BASL can be revealed by misexpressing small domains and assaying for dominant negative effects. The rationale for this approach is that the overexpressed target site will outcompete endogenous BASL for interaction with the partner, leading to failure of BASL to polarize or function in generating asymmetry (for examples, see Roth et al., 1998;Zhou et al., 2015).
We introduced BASL promoter-driven GFP-tagged BASL subdomains, including N, I, I2, C, NI, and IC ( Fig. 6 and see Supplementary Fig. S3 at JXB online), into the WT Col plants. We found that they showed a similar localization pattern as they did in basl mutants. The subdomains containing the N fragment, that is, BASL_N and BASL_NI, showed obvious nuclear enrichment (Fig. 6A, B). BASL-IC was cytoplasmic and polarized at the cortex (Fig. 6C) and BASL_I, _ I2, and C were primarily localized in the cytoplasm (Fig. 6D, and see Supplementary Fig. S3A, B at JXB online).
We then examined stomatal formation and patterning in transgenic plants. Elevated expression of the full-length BASLpro::GFP-BASL (an extra copy in WT) did not generate obvious alterations (see Supplementary Fig. S3C, D at JXB online). Overexpression of the subdomains BASL_NI and BASL_I, however, induced overproduction and clustered stomatal lineage cells, resembling that of the basl mutants (Fig. 6B, D versus Fig. 1B; quantifications in Fig. 6E). To exclude possible transgenic silencing, we conducted reverse transcription (RT)-PCR to show that endogenous BASL was normally expressed (see Supplementary Fig. S4 at JXB online). These results suggested that BASL_I generated a dominant negative effect likely competing with the endogenous BASL to bind to functional partner/s. Because the interaction between BASL and YDA or MPK3/6 uses motifs outside of this region (Zhang et al., 2015) and we did not find a physical interaction between BASL_I and YDA or MPK3/6 in a yeast-two-hybrid assay (data not shown), we suspect that BASL_I may interact with a novel partner/s for its function.
Interestingly, combining the C-domain with BASL_I eliminated its dominant negative effect, producing stomata comparable to those of the WT (BASL_IC versus BASL_I, Fig. 6E). These data indicate that the BASL_C domain suppresses BASL_I's interference, likely by bringing in interacting partners to form a functional complex, which is manifested by polarity recovery of BASL_IC (Fig. 6C). Overexpressing other subdomains (BASL_N, BASL_I2, and BASL_C) did not produce obvious phenotypes in stomatal development (see Supplementary Fig. S3D at JXB online).
Because GFP-BASL_I was found in the cytoplasm and at the PM (Fig. 6D), to further dissect where BASL_I confers dominant negative effects we overexpressed Mry-BASL_I-GFP. Intriguingly, Mry-BASL_I produced a stronger dominant negative effect than BASL_I did (Fig. 6F, G), suggesting that it is at the PM that critical BASL_I /partner interactions take place and outcompete endogenous BASL (Fig. 6H). Such an effect is likely due to the tight PM association rather than the protein levels of Myr-BASL_I, given that BASL_I is generally expressed more abundantly in plants (Fig. 6D versus Fig. 6F). The dominant negative phenotype is specific to the BASL_I region in that neither Myr-BASL_N or Myr-BASL_C created dominant negative effects (Supplementary Fig. S3F, G, quantified in Supplementary Fig. S3D at JXB online).

The BASL N-terminus regulates protein accumulation
It was previously noted that the BASL promoter activity had a broader expression pattern than that of the protein fusion (Dong et al., 2009). The BASL protein was primarily found in the stomatal linage cells (see Supplementary  Interestingly, the overall expression pattern of BASL subdomains suggests that the N domain of BASL confers stomatal lineage-specific expression because neither BASL_N nor BASL_NI were found outside of the stomatal lineage (Fig. 6A, B). However, without the N domain, the subdomains of BASL_I, I2, C, or IC showed a much broader pattern-they were also expressed in the pavement cells (Fig. 6C, D; see Supplementary Fig. S3A, B at JXB online)-which was similar to that of the BASL promoter (see Supplementary Fig. S3E at JXB online). These data suggest that the down-regulation or degradation of BASL protein in pavement cells might be mediated by the BASL_N domain. The dominant negative effect was evident (white brackets highlighting clustered stomata), indicating that the PM pool of BASL_I is interfering, possibly by competing with the endogenous protein to bind to partner/s at the PM. (G) Box plots comparing the dominant negative effect of GFP-BASL_I and Myr-BASL_I-GFP (relative to GFP-BASL). Mann-Whitney test, *** P < 0.001; * P < 0.01). (H) Diagram to summarize the proposed functions of BASL subdomains. Blue colour indicates the new findings from this study. The minimal polarity complex at the PM may require the combined functions of a novel (currently unidentified) BASL_I-mediated protein-protein interaction and BASL_C-mediated interaction with the YDA MAPK module. BASL-N, through the NLSs, directs the nuclear localization and also contributes to stomatal lineage-specific expression of the BASL protein.

Nuclear localization and potential degradation mechanisms for cell type-specific expression of BASL
The expression of BASL was initially seen in the nuclei of the protodermal cells in Arabidopsis seedlings (as early as 16 hours after germination), before cortical polarization was evident (Dong et al., 2009). Our previous studies suggested that BASL transiting through the nucleus might serve as a mechanism by which it could be phosphorylated and activated by MPK3/6 in the nucleus (Zhang et al., 2015). Here our data showed that, without the NLSs, BASL remains in the cytoplasm, but it is still able to be polarized and to function. We hypothesize that BASL can be phosphorylated by cytoplasmic MAPKs or other kinases in the cell [localization of MAPKs in plant cells reported in Samajova et al. (2014)]. Proteins smaller than 60 kD can diffuse out of the nucleus freely (Laba et al., 2014), which explains why BASL (below 30 kD) does not rely on the predicted NES for nuclear export. It is also possible that an unknown interacting protein transports BASL out of the nucleus in an NES-independent manner.
Although BASL's function within the nucleus is still mysterious, that BASL_N restricts BASL expression to the stomatal lineage cells ( Fig. 6 and see Supplementary Fig. S3 at JXB online) but BASL_IC is also found in pavement cells suggest that this compartment (nucleus) might be involved in a protein degradation process in pavement cells. Alternatively, motifs in BASL_N outside of the NLS could be targeted directly by the protein degradation system. Protein degradation is regulated by a highly conserved pathway, the ubiquitin-proteasome pathway (Callis and Vierstra, 2000). Preliminary analysis of the BASL protein sequence indeed suggested the possible existence of the ubiquitination sites with high confidence scores (data not shown). Future studies are required to elucidate whether ubiquitination and the proteasome degradation system mediate BASL protein abundance in the stomatal lineage cells, and which sites in BASL are ubiquitinated.

BASL_I and possible function at the cortical polarity site
Overexpression of Myr-BASL_I at the PM created a strong dominant negative effect in stomatal development, suggesting critical protein-protein interaction mediated by BASL_I at the PM. The dominant negative effect of BASL_I being released by the integration of BASL_C also indicates that BASL_I-mediated interacting protein/s contributes partially to cell polarity and asymmetric division, and that the minimal functional complex may require both BASL_I-and BASL-C-mediated interacting proteins (Fig. 6H). The discovery of the FxFP MAPK DEF-docking motif in BASL_C led to the recognition of two other MAPK D-docking motifs in BASL (Fig. 6H) and a link to MPK3/6, which can phosphorylate BASL and allow for its polarization at the cell cortex (Zhang et al., 2015). These MAPK docking motifs also seemed to mediate the interaction with the MAPKKK YDA, such that both YDA and MPK3/6 participate in the BASL polarity complex in the stomatal ACD cells (Zhang et al., 2015). However, what binds to BASL_I remains unknown. Thus, our studies urge the discovery of interacting partners of BASL_I at the PM and further investigation of how the BASL-YDA-MPK3/6 module may cross talk with BASL_I-interacting proteins to function together in stomatal ACD.
In summary, in this study we characterized subdomains of the BASL polarity protein in detail. Although BASL protein is not large in size (262 amino acids), our domain analyses revealed that it is subjected to multiple interesting regulatory mechanisms for its localization and function (Fig. 6H). Our results support PM localization of BASL as being critical for its function in stomatal lineage asymmetric cell division. These findings inspire us to search for new partners and to investigate multiple pathways that are associated with specific BASL subdomains in cell polarity and ACD in plants.

Supplementary material
Supplementary data are available at JXB online. Figure S1. The predicted NLS and NES in the BASL protein. Figure S2. Localization and function of BASL subdomains. Figure S3. Overexpression of BASL subdomains by the BASL promoter. Figure S4. Semi-quantitative RT-PCR for the endogenous expression of BASL in the overexpression lines of BASL subdomains. Table S1. Primers used in this study. Table S2. Localization and overexpression phenotype of BASL subdomain in the WT Col plants.