AMP1 and MP antagonistically regulate embryo and meristem development in Arabidopsis

Cell division is unequally distributed in the plant body. Growth regions with controlled patterns of dividing cells, termed apical meristems, elongate the plant axis and an extended meristem, the procambium, retains pluripotent cells for subsequent vascular differentiation. In many locations,proliferative activity is inversely correlated with the differentiation status of cells and their balance defines the sizes of meristematic regions. In some meristematic tissues, as for example in the procambium, dividing cells give rise to limited cell numbers in specialized tissues(Esau, 1965), whereas in apical meristems permanent stem cells give off daughter cells indefinitely(Weigel and Jürgens,2002).

The controls regulating the balance between proliferating and differentiating cells are only partially understood. Where amenable to genetic dissection, as in the shoot apical meristem (SAM), these controls seem to comprise antagonistic activities acting in specific zones (reviewed by Bäurle and Laux, 2003; Williams and Fletcher, 2005). Antagonistic activities might also control the size of other meristems. A mechanism related to that in the SAM has been proposed for the root meristem(Casamitjana-Martinez et al.,2003), and the formation of procambium in the leaf seems to occur in competition with mesophyll differentiation(Scarpella et al., 2004).

Mutations in the presumptive glutamate carboxypeptidase AMP1 are associated with diverse morphological abnormalities including supernumerary cotyledons, shortened plastochrons and a bushy appearance, and are further characterized by cytokinin overproduction and upregulation of CYCD3;1(Chaudhury et al., 1993; Chin-Atkins et al., 1996; Nogué et al., 2000a; Nogué et al., 2000b; Riou-Khamlichi et al., 1999). However, amp1 mutants are phenotypically distinct from both cytokinin or CYCD3;1-overproducing plants and it is unclear what primary defect could account for the various aspects of the amp1 phenotype. Despite a wealth of phenotypic data, AMP1 function has not been genetically linked to other genes in embryo or meristem patterning. The AMP1 product bears similarities to mammalian N-acetyl α-linked acidic dipeptidases(NAALADases) (Helliwell et al.,2001), but neither its organismal or cellular localization nor the molecular identity of its targets is known.

Auxin distribution patterns have been implicated in positioning of lateral organs in shoots and roots (Reinhardt et al., 2003; Benkova et al.,2003), the formation of vascular tissues(Aloni et al., 2003; Avsian-Kretchmer et al., 2002; Mattsson et al., 2003) and in the generation of the root stem cell niche(Aida et al., 2004). In all these positions, robust patterns of auxin accumulation were found associated with patterned cell fate specification, including positioning of meristematic activities. Other plant hormones, specifically cytokinins, are also essential for promoting cell division (Bishopp et al., 2006), but their distribution patterns have not been as precisely correlated to cellular responses. Auxin regulates gene expression through auxin response factors (ARFs) and their nuclear co-regulators of the Aux/IAA family (Guilfoyle and Hagen,2001; Liscum and Reed,2002). Although most ARF functions are still elusive, patterning functions involving organ initiation and growth have been assigned to some ARFs, including MONOPTEROS (MP/ARF5). Mutations in MP lead to the absence of an embryonic root, the formation of reduced vascular systems and flowerless shoots(Berleth and Jürgens,1993; Przemeck et al.,1996). Among other ARFs, ARF7/NON-PHOTOTROPIC HYPOCOTYL 4 and ARF19 are required for local cell proliferation in the pericycle to produce lateral roots(Okushima et al., 2005a; Wilmoth et al., 2005). By contrast, ARF2 has been shown to restrict the size of Arabidopsis ovules and seeds and to negatively regulate certain cell proliferation genes (Ellis et al.,2005; Okushima et al.,2005b; Schruff et al.,2005).

Here we identify amp1 as a first loss-of-function suppressor of an arf mutant and present evidence that AMP1 has a role in balancing and restricting the meristem-promoting activity of auxin signaling. We document that MP has an important role in promoting meristematic niches in diverse locations and that this activity is dispensable in the absence of a counteracting pathway involving AMP1.

Unless otherwise noted, seeds were plated and plants grown as previously described (Hardtke et al.,2004). Origin of transgenic lines: CycB1;1::CycB1;1-GUS(Donnelly et al., 1999), pCLV3::GUS (Brand et al.,2002), SNO-GFP(Cutler et al., 2000). UBI3::Lti6b-GFP was generated by M. Aida in the laboratory of B. Scheres (Utrecht University, Utrecht, The Netherlands) by fusion of membrane marker 29-1 (Cutler et al.,2000) to the potato UBI3 promoter (L22576).

Cleared whole-mount samples were prepared as described in Berleth and Jürgens (Berleth and Jürgens,1993). Detection of β-glucuronidase (GUS) activity was as in Scarpella et al. (Scarpella et al.,2004) with the following modifications to the concentration of potassium ferro- and ferricyanide and incubation times: 10 mM for 1 hour(pCLV3::GUS), 0.5 mM for 16 hours (MP::MP-GUS embryos), 2 mM for 2 hours (CycB1;1:CycB1;1-GUS), 5 mM for 2 hours(MP::MP-GUS seedlings) or 5 mM plus 1% Triton X-100 for 1 hour(MP::MP-GUS nuclear localization). Scanning electron microscopy and confocal laser scanning microscopy were performed as described(Douglas et al., 2002; Gazzarrini et al., 2004).

Sizes of inflorescence meristems were determined on images taken from above the meristem by measuring the distance from the centre of the youngest recognizable floral primordium to the centre of the furrow separating the fifth flower primordium from the meristem as described(Yu et al., 2000). SAM sizes were determined on cleared medium longitudinal images using ImageJ 1.33 software(http://rsb.info.nih.gov/ij/)as being the area formed by the dome of the meristem connected by a straight line between the cotyledon primordia.

A survey of the molecular lesions and phenotypic strengths of amp1mutations, including six new alleles, identified amp1-10 and amp1-13 as likely null alleles with no recognizable AMP1transcripts and amp1-1 as an allele with pronounced residual gene activity (see Fig. S1 and Table S1 in the supplementary material). As there are no apparent AMP1 paralogs in the Arabidopsis genome, the two allele-strength categories probably reflect partial and complete loss of NAALADase activity in the AMP1 pathway.

Cell numbers and cell division patterns in the early wild-type Arabidopsis embryo are almost invariable. A particularly reproducible feature of the Arabidopsis embryonic fate map is the restriction of the descendents of the apical and basal daughter cells of the zygote. Basal cell descendents form the suspensor, but, except for the central portion of the root meristem, they do not contribute to the seedling pattern(Scheres et al., 1994). In amp1 embryos, abnormal divisions of basal cell derivatives gave rise to additional cell tiers in the embryo proper and basal cell derivatives regularly contributed to large parts of the seedling, including the hypocotyl and cotyledons (Fig. 1I,J,P-S;frequency of extra tiers in amp1-13: 36/42, 42/54, 63/63, 96/96 at 4-, 8-, 16-cell and globular stages, respectively). Conversely, cells from the apical part of the globular embryo (framed cells in Fig. 1G,H,I) no longer contributed to cotyledons but became incorporated into an oversized SAM(Fig. 1Q,R,S). At lower frequency, abnormal divisions of basal derivatives led to the formation of a complete second embryo from the same zygote, which was reflected in the appearance of twin seedlings from single seeds in amp1 mutant lines(Fig. 1V,W and see Table S1B in the supplementary material). Except for the oversized shoot meristem and frequent supernumerary cotyledons, the architecture of amp1late-stage embryos is remarkably normal(Fig. 1Y)(Conway and Poethig, 1997). These features suggest that the mutant phenotype is primarily a consequence of the increased cell numbers in early pro-embryos.

In conclusion, the patterning defects in amp1 mutant embryos can be traced back to the failure of basal cell descendents to attain suspensor cell fate. Instead of displaying suspensor-specific differentiation features,some of those cells proliferate and either generate additional embryos or contribute to inappropriately large portions of the embryo proper.

The enlarged SAM is not solely a consequence of abnormal cell specification in the embryo. As shown in Fig. 2B,F,J and Table 1A, in amp1 mutants, SAMs continued to increase in diameter post-embryonically and were enlarged in the central stem cell regions, as visualized by expression of pCLV3::GUS. In addition,expression of pCLV3::GUS was generally extended towards the flanks of the SAM (Fig. 2F). In extreme cases, the pCLV3::GUS expression domain was five times wider than in wild type and became concentrated in concrete spots(Fig. 2M). These spots might be correlated with the formation of multiple SAMs(Fig. 2N), which we observed in all amp1 alleles, reminiscent of what has been described for the corona mutant (Green et al.,2005). By contrast, the sizes of the amp1 inflorescence and floral meristems were not markedly abnormal [mean inflorescence meristem diameter±s.e.m.: wild type, 52.6±0.7 mm (n=16); amp1-9, 51.4±1.2 mm (n=12)](Fig. 2V,W).

In hypocotyls and roots, pericycle cells retain proliferation competence,but only a portion of them will later progress through the cell cycle to produce lateral roots (Beeckman et al.,2001; Himanen et al.,2002). In amp1 mutants, the proportion of pericycle cells that were actually proliferating was greatly increased. Lateral roots were initiated very early (Fig. 2AA, Table 1C) and adventitious roots from hypocotyls were frequent (Fig. 2AE, Table 1B). Once established, amp1 mutant root meristems were not expanded in diameter (data not shown).

In summary, our results document that AMP1 restricts stem cell pool sizes in the SAM and keeps division-competent cells dormant in the pericycle.

Mutations in mp are associated with the absence of an embryonic root (Berleth and Jürgens,1993) and MP has been implicated in the generation of a stem cell niche in the root meristem (Aida et al., 2004). As shown in Fig. 2D,H,L, MP was found to promote stem cell formation not only in the root meristem, but also in the SAM. In wild type, the SAM is initiated during embryogenesis and produces the first two leaf primordia approximately at the time of germination(Laux et al., 1996). In mp mutants, the SAM was typically not visible at germination and SAMs were also smaller in mp seedlings(Fig. 2D,H,L, Table 1A). As previously reported, mutant inflorescence meristems are unable to produce normal numbers of flowers, floral meristems produce fewer floral organs(Fig. 2U,Y)(Przemeck et al., 1996) and mp mutants have incomplete vascular systems(Fig. 3A). This defect has been traced back to a reduced procambium, the meristematic precursor tissue of vascular strands (Przemeck et al.,1996). Finally, we found that mp mutants produce adventitious roots only after many weeks in culture, in sharp contrast to the enhanced production of adventitious roots in amp1 mutants(Table 1B). In summary, mp mutants are defective in the generation of appropriately sized meristems in various locations.

We assessed the possibility that MP and AMP1 activity antagonize each other in the control of meristematic activities by constructing amp1 mp double mutants of various allelic combinations. We found that amp1 suppresses the phenotype of mp and can even restore viability and fertility in an mp mutant background. Whereas rootless mp mutant seedlings were not viable under normal growth conditions, amp1 mp double mutants frequently formed hypocotyls and roots (Fig. 1Z, Fig. 2AB, Table 1D) and could be grown on soil (Fig. 2O-Q). In fact, amp1 mp double-mutant embryo development could be indistinguishable from amp1 embryogenesis (data not shown). Further, in contrast to the generally flower-defective and invariably sterile mp mutants,inflorescences of amp1 mp double mutants had abundant fertile flowers with partially restored numbers of floral organs(Fig. 2T,X, Table 1E). Mutations in AMP1 also increased the reduced cotyledon numbers in mpmutants (Table 1F) and we observed a restoration of adventitious root formation in amp1 mpdouble mutants (Table 1B). Finally, loss of AMP1 function significantly restored vascular tissue formation in the mp mutant background(Fig. 3A). Whereas the mature vascular system in cotyledons of mp mutants was typically restricted to a short midvein and occasional short side branches, cotyledon venation of amp1 mp double mutants comprised several lateral circular veins similar to wild-type cotyledons.

Suppression of mp by amp1 mutations was observed in a variety of allelic combinations. The resulting phenotypes support the notion of quantitative antagonistic activities, because the suppression of mp was generally weaker in combinations involving the weak amp1-1 allele (Fig. 2P,Q, Table 1D). In summary, our results indicate that MP and AMP1 genetically interact in the regulation of meristematic activity.

Our observations suggest a model in which auxin-derived positional information through MP carves out meristematic niches by locally overcoming a general differentiation-promoting activity involving AMP1 (Fig. 3C). This model is consistent with the overall meristem-promoting activity of MP contrasted by the differentiation-promoting activity of AMP1, but clearly involves more than the superimposition of antagonistic phenotypes. First, amp1 also suppresses mpwhere it does not confer an opposite phenotype. For example, amp1increases vascular complexity only in an mp, not in a wild-type,background and restores flower formation in mp mutant inflorescence meristems, whereas amp1 inflorescence meristems are phenotypically normal. Conversely, a mutation associated with enlarged meristems, such as clavata 2, does not necessarily affect mp meristem phenotypes (see Fig. S2 in the supplementary material). Therefore, interaction of AMP1 and MP is not simply a consequence of opposite controls of meristem sizes. Instead MP seems to overcome AMP1 activity in meristematic regions and therefore becomes(partially) irrelevant in the absence of AMP1 activity(Fig. 3C).

We provide further evidence for similarities between AMP1 and mammalian NAALADases, including intracellular location and tunicamycin sensitivity (see Figs S3, S4 in the supplementary material). Because targets and pathways of mammalian NAALADases have turned out to be extremely difficult to unravel(Zhou et al., 2005), molecular details of the AMP1 pathway and the nature of the interaction with MP might also not be immediately tractable. At this point, we can rule out three obvious possibilities for this interaction. First, as explained above, the interaction cannot be reduced to the superimposition of opposite controls. Second, MP does not seem to overcome AMP1 activity by transcriptional downregulation of AMP1, because AMP1transcript levels are unaffected in mp seedlings(Fig. 3D). Third, MP and AMP1 are unlikely to interact physically, as they are localized to different cellular compartments (see Fig. S4B,C in the supplementary material). This finding is also reflected in the absence of semidominant suppressive effects,which are frequently associated with direct interaction, in amp1 mpdouble mutants. Both gene products are, however, co-expressed in many locations and their respective pathways could therefore interact in those cells (see Fig. S5 in the supplementary material).

In summary, mutant phenotypes and expression patterns suggest that MP locally interferes with AMP1-promoted cell differentiation to maintain meristematic niches.

We thank John Celenza, Sean Cutler, Thomas Laux, Mitsuhiro Aida, Ikram Blilou and Ben Scheres for seeds, Enrico Scarpella and Annemarie Meijer for plasmid pC1300-35S-GFP and Enrico Scarpella for helpful comments on the manuscript. This work was supported by an NSERC Discovery Grant and a Premier's Research Excellence Award to T.B.