Protein phosphorylation is a key molecular switch used to transmit information in biological signalling networks. The output of these signalling circuits is governed by the counteracting activities of protein kinases and phosphatases that determine the direction of the switch. Whereas many kinases have been functionally characterized, it has been difficult to ascribe precise cellular roles to plant phosphatases, which are encoded by enlarged gene families that may provide a high degree of genetic redundancy. In this work we have analysed the role in planta of catalytic subunits of protein phosphatase 2A (PP2A), a family encoded by five genes in Arabidopsis. Our results indicate that the two members of subfamily II, PP2A-C3 and PP2A-C4, have redundant functions in controlling embryo patterning and root development, processes that depend on auxin fluxes. Moreover, polarity of the auxin efflux carrier PIN1 and auxin distribution, determined with the DR5pro:GFP proxy, are affected by mutations in PP2A-C3 and PP2A-C4. Previous characterization of mutants in putative PP2A regulatory subunits had established a link between this class of phosphatases and PIN dephosphorylation and subcellular distribution. Building on those findings, the results presented here suggest that PP2A-C3 and PP2A-C4 catalyse this reaction and contribute critically to the establishment of auxin gradients for proper plant development.
Reversible protein phosphorylation is a versatile mechanism utilized in both prokaryotic and eukaryotic transduction pathways to connect external and/or internal signals to a given cell response. The balance between the opposing kinase and phosphatase activities in any given condition determines the level of phosphorylation of the target protein, thereby altering its activity, subcellular location and/or stability and hence the output of the switch. Protein phosphatase 2A (PP2A) is one of the most abundant serine/threonine phosphatases in eukaryotes (Cohen et al., 1989; MacKintosh and Cohen, 1989; Rojo et al., 1998; Shi, 2009), exhibiting a high degree of conservation in terms of both sequence and functional properties (Moorhead et al., 2009). It is a heterotrimeric enzyme consisting of a catalytic subunit (PP2A-C) whose activity, substrate specificity and localization are regulated by the binding of A (PP2A-A) and B (PP2A-B) regulatory subunits (Mumby, 2007). The current notion of how PP2A activity is regulated is based mostly on results obtained in metazoan systems, which suggest that B subunits confer target specificity upon binding to the essentially promiscuous core dimer formed by PP2A-C bound to PP2A-A (Virshup and Shenolikar, 2009; Slupe et al., 2011). The subtle differences that distinct B subunits may bestow on the PP2A holoenzyme would thus account for the versatility of this enzyme, and its manifold involvement in disparate cellular processes (Sents et al., 2012).
It appears that the function of PP2A may have expanded and diversified in plants, in as much as plant genomes contain a larger number of PP2A genes than animals. Whereas there is only one PP2A-C gene in Drosophila and two in mammals (Virshup and Shenolikar, 2009), the Arabidopsis genome codes for five PP2A-C genes, which can further be grouped into subfamily I (PP2A-C1, -C2 and -C5) and subfamily II (PP2A-C3 and -C4) based on their sequence conservation (Casamayor et al., 1994). In addition, Arabidopsis codes for three A subunits and 17 B, B′ and B″ subunits (Farkas et al., 2007), theoretically accounting for 255 different heterotrimer combinations that could eventually perform distinct functions. Moreover, the presence of these enlarged gene families in plants may confer a high degree of genetic redundancy. This is particularly significant in the case of the PP2A-C subunits, which share over 95% sequence identity within a given subfamily, and 80% between the two subfamilies. Indeed, until now, few studies have reported anomalies upon disruption of single genes of this family. A pp2a-c2 knockout mutant was shown to be hypersensitive to ABA (Pernas et al., 2007), whereas a pp2a-c5 knockout mutant showed a slightly reduced response to brassinosteroids (Tang et al., 2011). In addition, virus-induced gene silencing (VIGS) of PP2A-C subfamily I in tobacco induced the expression of pathogenesis-related genes, activated localized cell death and increased resistance to a virulent strain of Pseudomonas syringae (He et al., 2004). Induced cell death, ABA-related phenotypes and altered responses to pathogens and to brassinosteroids have also been described for mutants in PP2A-A and B′ subunits (Kwak et al., 2002; Tang et al., 2011; Trotta et al., 2011), suggesting that these regulatory subunits may form holoenzymes with the aforementioned subfamily I catalytic subunits to control these responses.
In addition to those phenotypes, developmental alterations associated with defects in auxin distribution have been reported in plants with mutations in regulatory A (Rashotte et al., 2001; Michniewicz et al., 2007) and B″ subunits (Fisher et al., 1996; Camilleri et al., 2002). A causative explanation for these phenotypes has been recently proposed in which PP2As would counteract the activity of Pinoid and Pinoid-like protein kinases in regulating the phosphorylation status of the PIN auxin efflux carriers, thereby determining their polarity and thus the directionality of auxin fluxes in the plant (Michniewicz et al., 2007; Kleine-Vehn et al., 2009; Rahman et al., 2010). However, to unequivocally assign a function to PP2A enzymes, including the regulation of PIN phosphorylation status and localization, it is necessary to characterize mutants in which the catalytic activity has been eliminated. This issue is all the more relevant because a recent report showing that PP2A-A may interact with catalytic subunits from a related class of phosphatases (Dai et al., 2012) has cast doubt on the use of pp2a-a mutants to unequivocally infer the involvement of PP2A catalytic activity in a signalling network.
In an effort to functionally characterize the Arabidopsis PP2A enzymes, we have identified knockout mutants for all five catalytic subunits. Interestingly, analysis of plants with different combinations of pp2a-c3 and pp2a-c4 null alleles has revealed a central role of subfamily II in controlling auxin distribution and in directing plant development.
Functional analyses of PP2A-C genes
The genome of Arabidopsis thaliana codes for five catalytic subunits of protein phosphatase 2A (PP2A-C) displaying high sequence similarity. By quantitative RT-PCR we detected expression of all PP2A-C genes (Figure S1 in Supporting Information), suggesting that they all are functional. PP2A-C3 and -C5 exhibited essentially identical levels of expression in flowers, leaves and roots. In contrast, organ-specific differences in expression were observed for PP2A-C1, -C2 and -C4. In particular, expression of PP2A-C1 in aerial organs is up to two-fold higher than in roots while PP2A-C4 has a complementary pattern of expression, with higher levels in roots than in aerial organs, as previously reported (Casamayor et al., 1994).
To elucidate the role of PP2A-C genes, homozygous T-DNA insertion lines interrupting each PP2A-C gene (hereafter named c1–c5 for clarity) were identified. Insertion mutants did not accumulate the full size transcript of the corresponding PP2A-C gene, suggesting that they correspond to null mutations (shown for c3 and c4 in Figure S2). The c2 mutant was originally isolated in the Wassilewskija ecotype (Pernas et al., 2007), so for comparisons we introgressed it into the Col-0 ecotype shared by the other four mutants. The introgressed c2 plants showed ABA hypersensitivity (Figure S3), confirming the results obtained in the original Wassilewskija background (Pernas et al., 2007).
No obvious developmental alterations were observed in any of the soil-grown single mutants, which could be due to genetic redundancy, given the high degree of sequence identity among PP2A-C genes and the overlap in gene expression patterns. We reasoned that phenotypic consequences may only be revealed if multiple mutations were combined. In particular, we were interested in studying the function of subfamily II (PP2A-C3 and -C4) for which no previous information was available. We therefore crossed the c3 and c4 mutants and analysed the F2 generation. Segregating plants of significantly reduced size were observed (Figure 1a). Genotyping showed that these smaller plants were homozygous for the T-DNA insertions in either PP2A-C3 or -C4 and hemizygous for the other gene (c3 c4/+ or c3/+ c4). Moreover, c3/+ c4 plants were consistently smaller than those with a c3 c4/+ genotype. No c3c4 double mutant plants were identified among adult soil grown F2 plants, suggesting that the simultaneous disruption of these specific genes was deleterious. However, inspection of the F2 population soon after in vitro germination revealed the presence of an additional phenotypic class of dwarf plantlets with small, thickened and fused cotyledons, short and thickened hypocotyls, and severely stunted or absent roots (Figure 1b). These misshapen plants had a c3c4 genotype and did not develop further.
To obtain independent confirmation that the observed c3c4 phenotypes were due to the mutations in subfamily II PP2A-C genes, c3/+ c4 plants were transformed with a 7 kb Arabidopsis genomic DNA fragment containing the complete PP2A-C4 gene sequence. In the T2 progeny, adult plants with a c3c4 genotype and carrying at least one copy of the introduced genomic fragment were identified, indicating that the seedling growth arrest and malformation phenotype was complemented. Vegetative development of c3c4 plants homozygous for the introduced genomic fragment was indistinguishable from wild-type plants (Figure S4). These results confirm that the activity of subfamily II PP2A-C genes is required for normal plant development and for completing the life cycle of the plant, in functions that are not redundant with those of subfamily I members.
Reduced transmission efficiency of the c3c4 haplotype through male and female gametophytes
Genotyping the progeny from c3 c4/+ and c3/+ c4 plants showed significant deviations from a normal Mendelian segregation (Table 1). For instance, c3c4 plants represented only 5 and 3% of the descendants, respectively, instead of the 25% expected, and the ratio of c3 c4/+ and c3/+ c4 plants to single mutant plants was reduced by over two-fold from the expected ratio (0.9 and 0.6 instead of a ratio of 2). Considering that the viability of c3 c4/+ and c3/+ c4 plants seems to be unaffected, their reduced number is suggestive of lower transmission efficiency of the gametes with a c3c4 haplotype. Consistent with defects in the gametophytes, c3 c4/+ and c3/+ c4 plants had smaller siliques (Figure S5) containing a high number of unfertilized ovules (Table S1). c3 c4/+ siliques were on average 14% shorter, while c3/+ c4 siliques were 42% shorter than the wild-type ones. To test whether gametophyte fitness was affected, reciprocal crosses were performed between the wild type and either c3 c4/+ or c3/+ c4 plants. These crosses confirmed that c3c4 male and female gametes are viable but have reduced transmission efficiency, which ranges from 0.29 to 0.63 (Table 2). Interestingly, while male c3c4 gametes from c3/+ c4 and c3 c4/+ plants showed similar performances, transmission efficiency of the c3c4 female gamete was two-fold lower in the c3/+ c4 than in the c3 c4/+ plants, suggesting a sporophytic maternal effect of the c4 mutation on the viability of the double mutant female gametophyte.
Table 1. Segregation of the progeny from self-pollinated c3 c4/+ and c3/+ c4 plants
n, number of analysed individuals; P, P-value obtained when the observed values are compared with the expected Mendelian segregation using the χ2 test. P < 0.01 indicates significance in the difference from Mendelian segregation.
Percentage of total progeny. Genotypes were determined by PCR.
Plants with a c3 c4/+ or c3/+ c4 genotype, respectively.
Table 2. Transmission efficiency (TE) of the c3c4 genotype through male and female gametes
P, P-value obtained when the observed values were compared with the expected Mendelian segregation using the χ2 test. P < 0.01 indicates significance in the difference from Mendelian segregation; WT, wild type.
Number of analysed individuals. The transmitted gamete from the mutant plants was determined by PCR genotyping of the progeny.
c3 c4/+ female × WT male
WT female × c3 c4/+ male
c3/+ c4 female × WT male
WT female × c3/+ c4 male
PP2A-C3 and PP2A-C4 are involved in establishing auxin gradients
Double homozygous c3c4 plants exhibit severe patterning alterations, including a complete absence of roots (Figure 1b), which are strongly suggestive of defects in polar auxin transport (Liu et al., 1993; Shevell et al., 1994; Geldner et al., 2004). This would be consistent with the proposed role of PP2As in PIN dephosphorylation and trafficking (Michniewicz et al., 2007; Kleine-Vehn et al., 2009; Sukumar et al., 2009; Rahman et al., 2010). To gain further evidence for this hypothesis, we first analysed in roots of the different mutants the activity of the DR5 promoter as a proxy for in vivo auxin distribution (Sabatini et al., 1999; Friml et al., 2003; reviewed in Tanaka et al., 2006). As previously described (Xu et al., 2006), a strong DR5pro:GFP signal was observed in the quiescent centre (QC), in columella initials and in columella cells of wild-type roots (Figure 2a, g). Interestingly, DR5pro:GFP fluorescence distribution in both c4 and c3/+ c4 plants was reduced to a narrower domain, with fluorescence generally being weaker, and in the columella it was restricted to the two central files (Figure 2c, e, i, k). This reduced expression domain is indicative of a depletion of auxin from the root tip. The DR5pro:GFP fluorescence was also altered in c3 and c3 c4/+ plants but, remarkably, in an opposite way to c4 and c3/+ c4 mutants. There was an increase in the overall DR5pro:GFP fluorescence levels in c3 and c3 c4/+ root tips, suggesting a higher accumulation of auxin. Moreover, the domain of DR5pro:GFP expression was expanded, appearing in vasculature and cortex/endodermis initials, and in lateral root cap cells (Figures 2b, d, h, j). This expanded domain of DR5 promoter activity is identical to that observed at short times after stimulating auxin production in roots, before the internal capacitor re-establishes the correct auxin distribution (Grieneisen et al., 2007), supporting the hypothesis that it may result from increased auxin accumulation at the tip due to impairment of auxin transport. Importantly, in the single c3 and c4 mutants these changes in DR5pro:GFP fluorescence distribution are uncoupled from obvious morphological alterations, indicating that they are a direct consequence of the absence of PP2A-C3 and -C4 activity. In addition, DR5pro:GFP fluorescence in c3c4 seedlings is observed at irregular spots in the malformed cotyledons, whereas no fluorescence is detected at the position of the putative root primordium (Figure 2f, l), further supporting the absence of a proper root meristem in c3c4 seedlings. These results suggest that PP2A-C3 and PP2A-C4 play distinct, specific functions in the regulation of auxin distribution in the roots.
Recent results suggest that PP2A is responsible for dephosphorylating and controlling the localization of the PIN family of auxin efflux transporters (Michniewicz et al., 2007; Kleine-Vehn et al., 2009; Rahman et al., 2010), which are essential for establishing the auxin gradients that guide many aspects of plant development (Benkova et al., 2003). As PIN1 is expressed in the stele and is responsible for the bulk of basipetal auxin transport into the root tip (Friml et al., 2003), we introduced the PIN1pro:PIN1-GFP marker in the mutant plants and analysed its expression and localization, along with potential root patterning defects similar to those described for mutants with compromised polar auxin transport (Geldner et al., 2004; Michniewicz et al., 2007). Since the double homozygous c3c4 mutant does not generate a root meristem, our analyses focused on the c3 and c4 single and c3 c4/+ or c3/+ c4 enhanced mutants. For all tested parameters, we found a general pattern of increasing phenotypic severity c3 = c4 < c3 c4/+ < c3/+ c4, reminiscent of what is observed at the whole plant level (Figure 1a). Average PIN1-GFP signal strength was diminished most markedly in the c3/+ c4 genotype (Figure 3a, topmost panel). Patterning defects and abnormalities in PIN1-GFP polarity were analysed and sorted into phenotypical classes (none, slight, severe; Figure 3a, b). Moreover, the percentage of cells in the meristem area with a clearly basal PIN1-GFP signal was determined (Figure 3a). Defects in patterning were most frequently observed around the QC, with ectopic cell divisions and aberrant division planes often leading to an increased number of cells in the QC, and a less clearly defined niche of initial cells, with aberrations in both their numbers and positions (Figure 3b). Both single c3 and c4 mutant plants exhibited a higher percentage of aberrations than the wild type. However, defects were never severe. In contrast, c3 c4/+ roots showed increasingly severe phenotypes, and c3/+ c4 roots were almost exclusively aberrant of the slight or severe classes (Figure 3a middle panel). Importantly, the increased frequency and severity of patterning defects were accompanied by a failure to polarly localize PIN1-GFP to the basal plasma membrane (Figure 3a, b, arrowheads). In severe cases, PIN1-GFP was basically absent from the basal plasma membrane, consistent with the results reported in knockout plants for the PP2A-A regulatory subunits (Michniewicz et al., 2007). Notably, complete reversion of PIN1-GFP polarity from the basal to the apical cell side, which has been described as an occasional occurrence in RNA interference (RNAi)-induced knockdown of all three regulatory PP2A-A subunit genes (Michniewicz et al., 2007), was not observed in our experiments, indicating that presumably one copy of either C3 or C4 still suffices to counter AGC3 kinase-dependent PIN phosphorylation for targeting PINs to the apical cell side (Dhonukshe et al., 2010).
Taken together, our results indicate that the subfamily II of the catalytic PP2A subunits is involved in regulating plant development, presumably by modulating PIN polarity to establish auxin pattern-guiding gradients.
PP2A-C3 and PP2A-C4 are involved in establishing the apical-basal axis of polarity, and the root and shoot apical meristem during embryogenesis
The c3 c4/+ and, especially, c3/+ c4 plants display developmental changes in roots, such as altered patterning of the distal root apical meristem (Figures 2e and 3b), that are consistent with the defects observed in PIN1 polarity and auxin distribution (Ding and Friml, 2010). However, root growth and meristem organization are influenced by many other factors (Osmont et al., 2007) and thus we could not unequivocally conclude that the defects observed were due to alterations in auxin fluxes. In contrast, during embryogenesis there are highly stereotypical developmental decisions that are directly controlled by PIN-mediated polar auxin fluxes (Jenik et al., 2007) and are not readily affected by external factors. We therefore examined whether these developmental processes were also affected by c3 and c4 mutations. In c3 and c4 single mutants we did not observe changes in embryo patterning (Figure 4a, b). In contrast, in the progeny from c3 c4/+ and c3/+ c4 plants we observed embryos with patterning alterations that fell in two phenotypic classes: 44% of the cleared embryos from c3 c4/+ plants and 38% from c3/+ c4 plants displayed weak defects that were observable from the globular stage, while 9% of the embryos from c3 c4/+ plants and 6% from c3/+ c4 plants had strong patterning defects that were evident since the two-cell stage. Based on their occurrence and the correspondence with morphological alterations observed at the seedling stage, we concluded that these two phenotypic classes correspond to c3 c4/+ or c3/+ c4 mutants, and c3c4 mutants, respectively. The c3 c4/+ and c3/+ c4 embryos developed normally until the dermatogen stage. Subsequently, the hypophysis divided prematurely and with a shift in the plane of cell division, disrupting the initial establishment of the root apical meristem (Figure 4a, b). Patterning of the rest of the embryo was not so clearly affected, but the root meristem remained relatively disorganized even at late stages of embryogenesis (Figure 4b), reminiscent of the phenotypes later observed in the mature root meristem (Figure 3b, lower panel). In contrast, c3c4 embryos already had dramatic developmental alterations at earlier stages of embryogenesis (Figure 4c). Significantly, the c3c4 embryos did not develop distinctive embryo proper and suspensor structures, indicating that the apical–basal axis of polarity was not established. As a result, patterning was completely disrupted and the embryos did not develop any recognizable organ or tissue at early stages. Interestingly, at late stages of development, c3c4 embryos attained some elements of apical–basal polarity, such as the formation of two to four rudimentary cotyledons at the apical end of a presumptive embryo proper, and a disorganized suspensor-like structure at the basal end. However, growth and patterning were still severely disrupted. The suspensor-like structure had multiple cell files and the embryo proper did not develop morphologically recognizable root and shoot apical meristems and provasculature (Figure 4c, lower panel). These developmental changes can be directly attributed to alterations in auxin distribution, strongly supporting that PP2A-C3 and -C4 participate in establishing the pattern-guiding gradients of this hormone.
PIN1 polarity and auxin distribution is disturbed in c3c4 embryos
To directly test whether the developmental defects observed in c3c4 embryos were associated with changes in auxin distribution and PIN polarity, we analysed PIN1-GFP localization and DR5pro:GFP expression in this double mutant. We chose relatively late stages of development, when we could unequivocally identify and dissect the c3c4 embryos for these assays. In wild-type embryos at late globular stage, PIN1 exhibits a clear localization facing the incipient root pole (Figure 5a, upper panel, arrowhead), which becomes even more pronounced at the early heart stage (Figure 5a, middle panel, arrowhead), thus playing a central role in defining the root meristem (Friml et al., 2003). At the same time, a strong signal is observed in the incipient cotyledon primordia (Figure 5a, middle panel, asterisks). In contrast, c3c4 embryos showed a marked absence of basally polarized PIN1 at the incipient root pole (Figure 5b, c, arrowheads) and generally in the incipient provasculature. This is in line with the failure to establish a root pole. However, the PIN1-GFP signal in the protodermis of the upper embryo part, where PIN1 localizes to the apical cell sides, was not as severely compromised (Figure 5a–c, asterisks), implying that PP2A-C3 and -C4 function is not strictly required for apical PIN localization. This could also explain why cotyledon-like structures are still forming and some sort of bilateral symmetry is established at later stages of development, although the overall pattern is irregular. At very late stages, tetracotyledon phenotypes were frequently observed, with a relatively regular PIN1-GFP distribution in the protodermis, where PIN1 is apical, but a marked absence of PIN1-GFP signal in the provasculature (Figure 5d), where PIN is basally localized. In line with these results, DR5pro:GFP fluorescence was markedly reduced and irregular at the incipient root pole in the double mutant (Figure 5; compare e and f, arrowheads), while there was still a normal signal at the incipient cotyledon primordia (Figure 5e, f, asterisks). At very late developmental stages, DR5pro:GFP signal at the site where normally a root pole would be established was either absent or a broad, poorly defined signal at the embryo base was observed (Figure 5g, question marks).
These results strongly support that PP2A-C3 and -C4 are essential to control the PIN-dependent polar auxin fluxes that establish the apical–basal axis of polarity and the root and shoot apical meristems during embryogenesis. Bilateral symmetry and specification of cotyledon primordia, however, appear to be largely independent of PP2A-C3 and -C4 function, suggesting that these processes depend on apically localized PIN1 in the protodermis, and further corroborating the role of PP2A in basal, but not apical, PIN polarity. The late establishment of weak apical–basal polarity in c3c4 embryos could mean that other PP2A-C subunits have a minor role in this process at the end of embryogenesis, or, alternatively, that the physical connection between the basal cell and the maternal tissues eventually confers to the embryo a minimal degree of apical–basal polarity that is independent of PP2A activity in the embryo itself. However, the complete lack of a functional root in the double mutants indicates that PP2A-C3 and -C4 are essential for coordinated PIN1 polarization towards the basal cell side in provasculature cells, and in their absence any remaining auxin fluxes are not sufficient to establish a functional root pole.
Expansion of gene families during genome evolution may result in novel functionalities, or generate increased genetic robustness if redundant activities are maintained. The results presented here suggest a mixed model for the expanded PP2A-C gene family in the plant lineage that has provided both robustness and specificity to signalling networks controlled by these phosphatases.
Our results demonstrate that redundancy is maintained within subfamily II, such that developmental alterations are revealed only after deleting three of the four gene copies. Moreover, together with previously reported results (He et al., 2004; Pernas et al., 2007; Tang et al., 2011), our data indicate a functional diversification between the subfamilies I and II. Genetic analyses suggest that subfamily I may be involved in stress responses and in ABA and brassinosteroid signalling (Pernas et al., 2007; Tang et al., 2011) while, according to the results we present here, subfamily II would be involved in plant patterning through its function in establishing auxin gradients. Indeed, triple c1c2c5 knockout mutant seedlings, although severely dwarfed, still exhibit clearly distinguishable roots and shoots (our unpublished results). Is there any sequence signature that may account for this divergence? Sequence identity between proteins of the two subfamilies is over 80%, and many of the key domains and residues, including those important for binding the regulatory B subunits, are conserved in all Arabidopsis PP2A-C proteins. Interestingly, however, there are differences between the two subfamilies in a domain that may be important for the interaction with the regulatory A subunits. In human PP2A-C, Asp280 makes four intermolecular hydrogen bonds with the A subunit and, moreover, that residue is not conserved in related phosphatases that do not bind to the A subunit (Xing et al., 2006). The equivalent Asp284 in the plant PP2A-C subunits is embedded in a region that is well conserved between PP2A-C3 and -C4, but shows extensive sequence divergence from PP2A-C1, -C2 and -C5, including the substitution of Asp284 to Glu284. This region may thus define specificity of binding of PP2A-C subfamilies I and II to PP2A-A subunits in Arabidopsis, and be an important determinant in defining the formation of the distinct PP2A holoenzymes with different specific activities. Now that distinct physiological roles of members from both subfamilies are being revealed (He et al., 2004; Pernas et al., 2007; Tang et al., 2011; this work), it will be possible to study what sequence elements explain the functional divergence, a question that could be answered through the analyses of protein chimeras.
Interestingly, some level of functional divergence is also observed within the subfamily II members. Analysis of the DR5pro:GFP molecular marker revealed that PP2A-C3 and -C4 have distinct functions in auxin distribution in the root. Considering the high sequence similarity between the proteins encoded by PP2A-C3 and -C4, it is likely that the specific activities of these two genes in roots reflect differences in expression patterns that may result in them targeting different PINs. Moreover, the pattern of DR5pro:GFP fluorescence in c4 and c3/+ c4 roots is similar to that observed in plants expressing a phosphomimic version of PIN1, which shifts PIN1 to the apical side of vasculature cells, depleting auxin from the root tip (Zhang et al., 2010). Thus, PP2A-C4 may be co-expressed with PIN1 in the vasculature, where it would dephosphorylate it and direct it to the basal side to achieve proper auxin transport to the tip. Intriguingly, the pattern of DR5pro:GFP fluorescence in c3 and c3 c4/+ plants resembles the distribution observed in a triple mutant in the PID, WAG1 and WAG2 genes that results from a shift to basal localization of PIN2 in the most distal cells of the epidermis and in the lateral root cap (Dhonukshe et al., 2010). These genes encode AGC3 kinases that are thought to be responsible for phosphorylating PINs (Michniewicz et al., 2007; Dhonukshe et al., 2010). How then does a mutation blocking PP2A activity result in changes in auxin distribution similar to those caused by mutations blocking its antagonistic kinases? We speculate it may do so by altering PIN localization in different cells in a way that ultimately produces a similar change in auxin distribution.
Recently, it has been proposed that PP2A-A regulatory subunits form a complex with the PP6 catalytic subunits FyPP1 and FyPP3 to dephosphorylate PIN proteins and direct auxin fluxes (Dai et al., 2012). Although binding to scaffold subunits is very specific for the highly conserved PP2A and PP6 families in yeast and humans (Stefansson et al., 2008), scaffold subunits in Arabidopsis might be more promiscuous in their binding. While experimental evidence of physical interaction between PP2A-A and -C subunits in plants is still missing, both subunits have been shown to interact with, and be involved in, the dephosphorylation of the brassinosteroid-responsive BZR1 transcription factor (Tang et al., 2011). Moreover, for deacetylation of tubulin in Arabidopsis, PP2A-A and -C subunits have been shown to interact with histone deacetylase HDA14 in protein complexes from which PP6 is excluded (Tran et al., 2012).
The results presented here suggest a functional specialization of PP2A catalytic subunits from subfamily II in the regulation of PIN protein polarity and hence of auxin fluxes and plant patterning. Putative orthologues of PIN proteins have been identified throughout the land plants, even in non-vascular ones like Physcomitrella (Paponov et al., 2005). Interestingly, putative subfamily I and II orthologues are also present in those organisms, raising the possibility that the specialized function of subfamily II in PIN dephosphorylation is ancient, and possibly pre-dates the colonization of the terrestrial habitat.
Arabidopsis thaliana (Heyn.) ecotype Columbia was used throughout. Plants were grown on soil in the greenhouse at 21°C under a 16-h light/8-h dark photoperiod. T-DNA insertions in each of the PP2A-C genes were identified in the TAIR and SIGnAL databases (http://signal.salk.edu; www.arabidopsis.org). The Salk-insertion lines, SALK_102599, SALK_035009, and SALK_139822, with T-DNA insertions in genes At1g59830 (PP2A-C1), At3g58500 (PP2A-C4) and At1g69960 (PP2A-C5), respectively, and the SAIL insertion line, SAIL_182_A02, which has a T-DNA insertion in gene At2g42500 (PP2A-C3), were obtained from the Arabidopsis Biological Resource Center (ABRC). The T-DNA insertion in the PP2A-C2 gene originally identified and characterized in the Wassilewskija ecotype (Pernas et al., 2007) was introgressed into the Columbia ecotype after six sequential crosses. Homozygous mutants were identified by genomic PCR using specific primers (Table S2). Arabidopsis transgenic lines expressing DR5pro:GFP and PIN1pro:PIN1-GFP have been described (Benkova et al., 2003; Friml et al., 2003). Expression of PP2A-C in the corresponding homozygous mutants was analysed by RT-PCR using specific primers (Table S3).
For quantitative RT-PCR analyses, plants were grown on soil in the growth chamber at 21°C under a 14-h light/10-h dark photoperiod. Leaves were harvested from rosettes of 3-week-old plants. Open, whole flowers were collected from 4-week-old plants. Roots were collected from 11-day-old seedlings germinated on 1× MS medium containing 1% (w/v) sucrose. Plant tissues were frozen in liquid nitrogen and stored at −80°C.
We isolated RNA from frozen tissues with TRIzol (Invitrogen, http://www.invitrogen.com). Traces of DNA were eliminated with TURBO DNA-free (Ambion, http://www.invitrogen.com). Two micrograms of RNA was used to make cDNA with the High-Capacity cDNA Archive Kit (Applied Biosystems, http://www.appliedbiosystems.com), according to the manufacturer's instructions. Quantitative RT-PCR was performed on an ABI Prism 7300 (Applied Biosystems), using the FastStart Universal Probe Master (ROX; Roche, http://roche-applied-science.com), and with sets of primers (Table S4) and Universal ProbeLibrary probes (Roche) designed online with ProbeFinder v.2.20 (Roche). Each pair of primers amplified a DNA fragment of the expected size from wild-type cDNA but failed to do so when cDNA from the corresponding T-DNA insertion mutant was used as a template.
For tissue-specific expression, three biological replicates were analysed in each case. CT values were obtained with the 7300 Systems SDS software version 1.3 (Applied Biosystems, http://www.appliedbiosystems.com). Relative fold expression changes were calculated by the comparative CT method: fold change is calculated as 2−ΔΔCT. The ΔCT values were calculated as the difference between the CT value of the specific gene and the CT value of EF1-α. ΔΔCT was the difference between ΔCT and the CT value of the calibrator. In Figure S1, the calibrator is the floral sample. The ΔΔCT values reported are averages of three independent trials (including technical and biological replicates).
To examine gametophytic transmission of the c3 or c4 alleles, reciprocal test crosses were performed between the wild type (Col-0) and enhanced mutant lines (c3 c4/+ and c3/+ c4). Seeds harvested from crosses were germinated and grown on soil, and genomic DNAs from the F1 progeny were analysed by PCR using the primer combination to detect the corresponding PP2A-C gene and the T-DNA insertion border. Transmission efficiency (TE) of the mutant alleles via each type of gamete (TE male and TE female) was calculated as described previously (Howden et al., 1998).
Microscopy and image analysis
For observations of embryo development, siliques in a series of developmental stages were opened longitudinally. The developing seeds were dissected and cleared in chloral hydrate (2 mg ml−1). Embryos were observed under the Leica DMR microscope with differential interference contrast (DIC) optics (http://www.leica.com). Photographs were taken with a Leica MPS-60 camera.
For confocal microscopy of root tips, 5-day-old seedlings were mounted in liquid growth medium or water and cell boundaries were stained with propidium iodide at 5 μg ml−1. For assessment of the PIN1-GFP signal and patterning phenotypes in seedling roots, strictly median sections of a 250 μm × 250 μm field (1024 × 1024 pixels) covering the entire lower root meristem were obtained under constant imaging conditions. The average GFP signal strength was measured for the whole frame using the Fiji ImageJ package (Schindelin et al., 2012). Patterning defects and PIN polarity were visually classified in three arbitrary classes. For patterning defects, we observed the entire frame. Phenotypic class ‘none’ refers to micrographs with fewer than four visible patterning defects, such as cell division plane aberrations, changes in shape of individual cells, or ectopic divisions. Phenotypic class ‘slight’ refers to micrographs presenting four or more of the above-mentioned defects, but where the overall pattern, especially regarding cell sizes and tissue layers, was still maintained. The class ‘severe’ refers to micrographs where there were frequent cell division plane aberrations, ectopic divisions or changes in shape and size of groups of cells which alter the overall pattern, especially regarding cell sizes. For PIN1 polarity defects, we observed only stele cells within a maximal distance of 120 μm to the quiescent centre. We assessed polarity using the relation of signal strength at the basal membrane to signal strength at the lateral membrane in these cells. In control seedlings, the basal signal is visibly stronger than the lateral one, allowing for a rapid visual assessment of PIN1 polarity by counting the number of stele cells with clear basal PIN1 polarity. For each genotype, we observed micrographs of exactly the same dimensions. The classes were defined as follows: ‘none’ for images where at least 20 cells have clearly basal PIN1-GFP signal, ‘slight’ where between 19 and 10 cells have a clearly basal PIN1-GFP signal, and ‘severe’ where fewer than 10 cells exhibit clearly basal PIN1-GFP signal. Since cell number in the observed region is not exactly constant between individual roots and genotypes, we also counted for each root the number of cells with a clearly basal PIN1-GFP signal, and divided it by the total number of cells in the region of interest.
Per genotype, between 27 and 31 images from two independent experiments were analysed with no indication of significant variability between experiments. c3 c4/+ and c3/+ c4 seedlings were obtained from a segregating population, and their genotypes were verified by PCR after imaging. For GFP observations in embryos, embryos at the desired stages were isolated from dissected ovules and mounted in 5% glycerol. A Leica SP5 inverted confocal microscope was used with constant imaging conditions for all comparative analyses.
Construction of expression vectors and plant transformation
For the complementation test, the genomic PP2A-C4 region was amplified by PCR from Col-0 genomic DNA using specific primers (Table S5) with GATEWAY adapters (gw-At3g58500up/gw-At3g58500rev) that amplified from position −2209 (upstream of the translational start codon) to +1258 (downstream of the stop codon). The purified PCR product was cloned via a pDONR207 donor vector into a pGWB1 destination vector (Nakagawa et al., 2007).
The accuracy of the construct was confirmed by DNA sequencing. The transformation vector was introduced into Agrobacterium tumefaciens strain GV3101 (pMP90; Koncz and Schell, 1986) by heat-shock. Arabidopsis thaliana ecotype Col-0 or c3/+ c4 mutant were transformed with the construct using the floral dip method (Clough and Bent, 1998). The transformants were screened using appropriate antibiotics.
Sequence data from this work can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: At1g59830 (PP2A-C1), At1g104030 (PP2A-C2), At2g42500 (PP2A-C3), At3g58500 (PP2A-C4), At1g69960 (PP2A-C5), At4g05320 (UBQ10) and At5g60390 (EF1-alpha).
We thank J. Friml and E. Benkova for DR5pro:GFP and PIN1pro:PIN1-GFP reporter lines, and the Nottingham Arabidopsis Stock Centre and ABRC for materials. We also thank M. A. López Carrasco for the data on germination of pp2a-c2 mutants. We are grateful to Professor Rod Casey for helpful suggestions and critical review of the manuscript. This work was supported, in part, by the Spanish Ministry of Science and Innovation (BIO2005-08528 and BIO2008-03052 to JJSS; BIO2009-10784 to ER). TD was a recipient of a FPU Scholarship from the Spanish Ministry of Science and Innovation. MaS was a recipient of post-doctoral contracts from the CSIC I3P programme (co-financed by the European Social Funds) and MiS was a recipient of an EMBO long-term and a Marie Curie Intra-European fellowship.