The more and smaller cells mutants of Arabidopsis thaliana identify novel roles for SQUAMOSA PROMOTER BINDING PROTEIN-LIKE genes in the control of heteroblasty

Determination of leaf size and shape is a crucial process that influences the appearance of shoots. Plants have the ability to make leaves of amazingly diverse sizes and shapes, not only between different species but also within a species. The size and shape of leaves change dramatically in response to environmental conditions, such as light quality and quantity, daylength,nutrition and water availability (Ferjani et al., 2008). In addition to these external cues, internal signals that arise at certain developmental stages have a role in the regulation of leaf size and shape. In most species, some leaf traits change as a plant passes through developmental phases, such as embryogenic, juvenile vegetative, adult vegetative and reproductive phases(Tsukaya and Uchimiya, 1997; Kerstetter and Poethig, 1998; Tsukaya et al., 2000). This phenomenon is called phase change, or heteroblasty. In Arabidopsis thaliana (Arabidopsis, hereafter), lamina size, leaf length/width ratio, petiole length, serration number, and production of abaxial trichomes change in relation to heteroblasty. Cell number and cell size in leaves might also change with heteroblasty, as an increase in cell number and a decrease in cell size in higher node leaves is observed in various species (Ashby, 1948; Granier and Tardieu, 1998; Cnops et al., 2004; Cookson et al., 2007). However, reduced water availability caused by water deprivation to leaves at lower nodes, or diffusive inhibitory signals from lower leaves were previously thought to account for the reduction in cell size in leaves on higher nodes(Ashby, 1948); whether this phenomenon is physiological or genetically controlled remains to be answered.

Recent molecular genetic studies indicate that miRNAs and trans-acting siRNAs regulate heteroblasty. miR172 in maize promotes the juvenile-to-adult phase change by repressing its target, Glossy15, an APETALA2-like gene required for juvenile leaf traits(Lauter et al., 2005). By contrast, miR156 inhibits this phase change by repressing its target,SBP-box-containing genes, when overexpressed in a Corngrass1 mutant(Chuck et al., 2007). Also, in Arabidopsis, constitutive overexpression of miR156 severely inhibits the progression of heteroblasty. Moreover, overexpression of the miR156-insensitive form of the Arabidopsis SBP-box gene SQUAMOSA PROMOTER BINDING PROTEIN-LIKE 3 (SPL3), SPL4 or SPL5 accelerates the juvenile-to-adult phase change(Wu and Poethig, 2006). In addition, involvement of tasiR-ARF and its target ETTIN(ETT)/AUXIN RESPONSE FACTOR 3 (ARF3) and ARF4 in the regulation of heteroblastic traits, such as abaxial trichome production, is also proposed(Peragine et al., 2004; Hunter et al., 2006). However,how these genes regulate leaf size and/or shape is still unknown.

Leaf size and shape regulation is dependent on the control of cell proliferation and cell expansion (Tsukaya,2006), and spatial and temporal regulation of these two processes is crucial for leaf development. In eudicot species, cell proliferation first occurs throughout the leaf primordium and is gradually restricted to the proximal part (Donnelly et al.,1999). Arrest of cell cycle and subsequent post-mitotic cell expansion occur from the distal to the proximal part of a young leaf. The regulatory mechanisms of cell proliferation and cell expansion have been investigated in many studies. An interesting topic is how these two cellular processes are coordinated in the context of leaf development. The first example of such coordination was in barley leaves irradiated with γrays, which show no cell division during development but develop significantly larger cells (Haber, 1962). Recent studies using various mutant or transgenic plants have provided several lines of evidence for the presence of organ-level coordination of cell proliferation and cell expansion (Tsukaya,2002; Beemster et al.,2003; Tsukaya,2003; Horiguchi et al.,2006a; Ferjani et al.,2007). When cell number is reduced by a mutation in a cell-proliferation-promoting gene, the size of individual cells often increases. This phenomenon, called compensation syndrome, is observed in several Arabidopsis mutants. However, overexpression of such a gene[for example, ANGUSTIFOLIA 3 (AN3)/GRF-INTERACTING FACTOR 1 (GIF1) or AINTEGUMENTA (ANT)] results in an increase in cell number, but cell size does not change(Mizukami and Fischer, 2000; Kim and Kende, 2004; Horiguchi et al., 2005). These observations indicate that cell number and cell size in a leaf might not be determined by a simple trade-off between cell proliferation and cell expansion. The precise mechanism coordinating these two processes is unknown.

To further investigate the regulation of cell proliferation and cell expansion, we isolated a number of mutants with altered cell number, size or both (Horiguchi et al., 2006a; Horiguchi et al., 2006b; Ferjani et al., 2007; Fujikura et al., 2007). Here,we report a new class of mutants named more and smaller cells(msc) that have increased cell number and decreased cell size, the opposite phenotype to compensation syndrome. These mutants also show accelerated heteroblasty. Analysis of leaves at various nodes demonstrated that adult leaves had an increased cell number and decreased cell size compared with those in juvenile leaves. This indicates that heteroblasty plays an important role in the regulation of cell number and size. Cloning of MSC genes and subsequent molecular and genetic analyses demonstrated that miR156 and its target SPL genes are involved in the regulation of heteroblastic change of cell number and size, whereas another group of heteroblasty-related genes (RDR6, SGS3, ZIP, ARF3 and ARF4) might not be involved.

The wild-type accession used in this study was Columbia-0 (Col). The msc1-D (newly isolated), msc2 (formerly line number 2025)and msc3 (formerly 2058) mutants were isolated from a T-DNA mutagenized population (Horiguchi et al.,2006b). Before analyses, all of the mutants were backcrossed to the wild-type Col at least three times. We also used an3-4(Horiguchi et al., 2005), ant-1 (Mizukami and Fischer,2000), fugu1, fugu2-1, fugu3-D, fugu4-D, fugu5-1(Ferjani et al., 2007), psd-1, psd-6 (Hunter et al.,2003a; Li and Chen,2003), rdr6-11, sgs3-11(Peragine et al., 2004), zip-1 (Hunter et al.,2003b) and axr1-3(Estelle and Somerville, 1987)mutants and the T-DNA insertion mutants arf3-2 (CS24604)(Okushima et al., 2005), arf4-2 (SALK_070506C) (Alonso et al., 2003) and sqn-5 (SALK_033511)(Prunet et al., 2008). The sqn-1 (Berardini et al.,2001) mutant and the transgenic plants that overexpress miR156, SPL3m, SPL3Δ, SPL4Δ or SPL5Δ were gifts from G. Wu and R. S. Poethig(Wu and Poethig, 2006). All of these mutants and transgenic plants were in the Col background. Plants were grown on rock wool at 22°C under a 16 hour light/8 hour dark photoperiod at a light intensity of approximately 40 μmol m^–2s^–1.

For histological analysis of cells, first, third or fifth leaves of 30-day-old seedlings were collected. Collected leaves were fixed in formalin/acetic acid/alcohol and cleared using chloral solution, as described by Tsuge et al. (Tsuge et al.,1996). Whole leaves and cells were observed as previously described (Fujikura et al.,2007). Because expansion of first, third and fifth leaves was already completed at that stage, cell area was uniform in all parts of the leaf (see Fig. S1 in the supplementary material). Areas of 20-30 cells were measured for each leaf and averaged. Mean ± s.d. of average cell areas from six individual plants are indicated in the figures. To calculate the total cell number in leaves, we measured cell density of observed images of cells, and multiplied the cell density by the area of the same leaf. For analysis of abaxial trichomes, plants were grown until the first few flowers opened.

The msc loci were genetically mapped using various genetic markers according to the sequence information available in The ArabidopsisInformation Resource (TAIR) database(http://www.arabidopsis.org/).

Total RNA was extracted from leaves using TRIzol reagent (Invitrogen,Carlsbad, CA, USA) following the manufacturer's instructions. For quantitative RT-PCR, total RNA was treated with amplification grade DNase I (Invitrogen)before reverse transcription. Reverse transcription was performed with the oligo(dT)₂₀ primer using the SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen). Quantitative PCR was performed with Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA,USA). The UBIQUITIN10 (UBQ10) gene was used as an internal control. The primers used are listed in Table 1. At least three biologically independent samples were analyzed and triplicated reactions were performed with each sample. Reactions performed without reverse transcription did not result in any amplifications (data not shown). For 5′-RLM-RACE,total RNA was extracted from the first and second rosette leaves of 14-day-old plants and 5′-RLM-RACE was carried out using the GeneRacer Kit(Invitrogen). For small RNA blots, 5 μg of total RNA was separated using 8 M urea-denaturing polyacrylamide gels and electrically transferred to a Hybond-N⁺ membrane (GE Healthcare, Buckinghamshire, UK). Blots were hybridized with a [γ-³²P]ATP labeled miR156-complementary oligonucleotide probe 5′-GTGCTCACTCTCTTCTGTCA-3′ at 40°C in ULTRAhyb-oligo hybridization buffer (Ambion, Austin, TX, USA). A U6snRNA-complementary oligonucleotide probe 5′-TCATCCTTGCGCAGGGGCCA-3′ was used as a loading control.

From a collection of numerous Arabidopsis mutants with altered palisade mesophyll cell number or size, or both, in the first leaf(Horiguchi et al., 2006a; Horiguchi et al., 2006b), we selected three mutants that had increased cell number and decreased cell size(Fig. 1C,D). We named these more and smaller cells (msc) mutants. All mscmutants had larger first leaves than the wild type(Fig. 1B). msc1 had no obvious phenotype other than a larger leaf area and a slightly earlier flowering phenotype (Fig. 2E,reduced total rosette leaf number). msc2 showed a delay in the initiation of the first pair of rosette leaves, as if the shoot apical meristem were inactive for a few days (Fig. 1E). msc3 had a delay in leaf emergence and expansion,but the final leaf size was larger than in the wild type(Fig. 1F). msc2 and msc3 were inherited as recessive mutations and msc1 was semi-dominant, and was thus designated as msc1-D.

We cloned MSC1-MSC3 genes using a map-based approach. msc2 and msc3 were found to be new alleles of previously reported mutants showing accelerated heteroblasty. These findings suggest that msc1-D might have a genetic lesion in the heteroblasty-related gene;therefore, we first described MSC2 and MSC3 genes and then considered MSC1.

The mutation in MSC2 was mapped to the PSD gene(At1g72560) (Hunter et al.,2003a; Li and Chen,2003). msc2 has an 870 bp deletion spanning the seventh intron to the eleventh exon of PSD and an insertion of T-DNA in this region (Fig. 2A). msc2had a very similar phenotype to the psd mutant [meristem pause(Fig. 1E), pointed leaves (data not shown) and accelerated heteroblasty(Fig. 2E, see below)]. Therefore, we tested for the cell number and size of psd-1 and psd-6 alleles and found that they had increased cell number and decreased cell size, as found in msc2 (see Fig. S2 in the supplementary material). F1 plants from a cross between msc2 and psd-6 failed to complement their phenotypes (see Fig. S2 in the supplementary material). We concluded that msc2 was a new allele of psd (psd-16).

The mutation in MSC3 was mapped to the SQN gene(At2g15790) (Berardini et al.,2001). msc3 had a 127 bp deletion from the seventh intron through the eighth exon of SQN(Fig. 2A). msc3 and sqn mutants also shared other phenotypes [leaf size and shape(Fig. 1F), disturbed flower phyllotaxis (data not shown) and accelerated heteroblasty(Fig. 2E, see below)]. We then examined the number and size of cells in sqn-1 and sqn-5leaves and confirmed that they had similar phenotypes to that of msc3(see Fig. S2 in the supplementary material). F1 plants from the cross between msc3 and sqn-1 did not complement their phenotypes (see Fig. S2 in the supplementary material). Thus, we conclude that msc3 is a new allele of sqn (sqn-6).

We mapped the msc1-D mutation at low resolution to the lower arm of chromosome 3, near the SNP marker SGCSNP7. Around the marker, the SPL15 (At3g57920) gene was found, which has an miR156 target site (Rhoades et al., 2002). Recently, SPL15 and its closest homolog SPL9 were reported to be involved in the regulation of heteroblasty(Schwarz et al., 2008). Based on this information, we sequenced the SPL15 gene in the msc1-D background. We found a C-to-T nucleotide substitution in the miR156 target site in the msc1-D mutant(Fig. 2A,C). Although this mutation does not cause amino acid substitution, it could lead to reduced efficiency of miR156-targeted mRNA cleavage. To test this, mRNA cleavage sites of SPL15 transcripts were identified by 5′-RLM-RACE. A distinct band of amplified products of an expected size was observed in the wild type (Fig. 2B). We found that most cleavage events took place at the ninth and tenth residues from the miRNA 5′ end(Fig. 2C) as previously reported (Wu and Poethig,2006). In msc1-D, however, the amount of amplified products was significantly reduced (Fig. 2B), and most of the cleavage events occurred outside of the miR156 complementary site (Fig. 2C). We also tested expression levels of SPL15 by quantitative RT-PCR. The expression levels in msc1-D were much higher than in the wild type (Fig. 2D). These results strongly suggest that the mutation in the miR156 target site of the SPL15 gene leads to reduced cleavage and increased accumulation of SPL15 mRNA in the mutant. In addition, msc1-D showed accelerated heteroblasty(Fig. 2E, see below). As SPL15 promotes phase change(Schwarz et al., 2008), the mutated SPL15 is highly likely to be responsible for the phenotypes of the msc1-D mutant. We hereafter designate the mutant spl15-1D/msc1-D.

psd and sqn mutants were previously reported to exhibit accelerated heteroblasty (Telfer et al.,1997; Berardini et al.,2001). In Arabidopsis, juvenile leaves have trichomes only on the adaxial side and adult leaves have trichomes on both adaxial and abaxial sides; thus, trichome presence is a typical marker of heteroblasty. To investigate whether spl15-1D/msc1-D exhibits accelerated heteroblasty, we counted rosette leaves without abaxial trichomes (juvenile leaves) or with abaxial trichomes (adult leaves) and cauline leaves(reproductive leaves). The wild type produced 6.3 juvenile leaves on average,whereas the spl15-1D/msc1-D mutant had 4.3, indicating that it shows accelerated heteroblasty (Fig. 2E). psd-16/msc2 and sqn-6/msc3 produced 4.2 and 4.6 juvenile leaves, respectively, confirming the accelerated heteroblasty(Fig. 2E).

Because msc1-msc3 mutants show accelerated heteroblasty, their leaves might have characteristics of those at higher nodes in the wild type. If this interpretation is correct, wild-type leaves should exhibit progressive changes in the number and size of cells during phase change; such changes have been observed mainly in epidermal cells in previous reports(Cnops et al., 2004; Cookson et al., 2007). To address this possibility, we determined the number and size of palisade mesophyll cells at various nodes in the wild type. Cell number increased and cell size decreased in leaves at higher nodes compared with those at lower nodes (Fig. 3A,B). In spl15-1D/msc1-D and sqn-6/msc3 mutants, cell number and cell size resembled those in leaves at higher nodes than the corresponding nodes of the wild type (Fig. 3A,B). However, cell number in the third and fifth leaves of psd-16/msc2 did not exceed that of the wild type (Fig. 3A,B). This might be because of a genetic lesion in the PSD gene, which encodes a tRNA export mediator exportin-t(Hunter et al., 2003a; Li and Chen, 2003) and could lead to reduced efficiency of protein synthesis, and thus to a decline in plant growth.

The progression of heteroblasty in Arabidopsis, in a broad sense,is inhibited by tasiR-ARF-mediated repression of ETT/ARF3and ARF4, and by miR156-mediated repression of SPLgenes (Hunter et al., 2006; Wu and Poethig, 2006). We investigated these two pathways in relation to the heteroblastic regulation of cell number and size. Initially, we observed accelerated heteroblasty mutants rdr6, sgs3 and zip, in which tasiR-ARF expression was not detected (Hunter et al.,2006). As shown in Fig. 4A,B, rdr6 and sgs3 had a slightly increased cell number, although cell size was not significantly affected, except for the fifth leaf of sgs3. The zip mutant, by contrast, showed no alteration in cell number and size (Fig. 4A,B). Next, we investigated the arf3-2 and arf4-2 mutants, in which heteroblasty is retarded(Hunter et al., 2006), and found that they showed no change in cell number or size in any leaves compared to the wild type (Fig. 4C,D). These results suggest that tasiR-ARF-mediated regulation of ETT/ARF3 and ARF4 might not be required for the heteroblastic change in cell number and size. In the arf3-2 arf4-2double mutant, although cell number was reduced in all leaves tested, an increase in cell number in the leaves at higher nodes occurred in a similar manner to the wild type (Fig. 4C,D). Moreover, cell size did not change compared to the wild type.

We then examined miR156 constitutively overexpressing plants. In these plants, cell number and size in the first leaf were the same as in the wild type (Fig. 4E,F). However,the cell number increase and cell size reduction in leaves at higher nodes were considerably inhibited (Fig. 4E,F), indicating that heteroblastic change in cell number and size was severely inhibited. By contrast, transgenic plants constitutively overexpressing SPL3 with a mutation in the miR156 target site (35S:SPL3m) or SPL3 or 4 with a deletion of the miR156 target site (35S:SPL3Δ and 35S:SPL4Δ) (Wu and Poethig,2006) had a significantly increased cell number and decreased cell size, most remarkably in the first leaf(Fig. 4E,F). Those overexpressing SPL5 with a deletion of the miR156 target site (35S:SPL5Δ) showed no significant alteration in cell number but a considerable reduction in cell size(Fig. 4E,F). These results indicate that miR156 and its target SPL genes (SPL3subclass) might be involved in the regulation of heteroblastic change in cell number and size.

We further investigated the expression levels of miR156 or various SPL genes in msc1-msc3 mutants. miR156levels were partially reduced in psd-16/msc2 and sqn-6/msc3(Fig. 5A). By contrast, we found preferential upregulation of specific SPL genes among 10 SPL genes with the miR156 target sequence: SPL13and SPL15 were upregulated 2.6- and 3.6-fold, respectively, in psd-16/msc2, and SPL3 and SPL13 were upregulated 3.9- and 5.6-fold, respectively, in sqn-6/msc3(Fig. 5B). In spl15-1D/msc1-D, only SPL15 was markedly upregulated, as expected (Fig. 5B). These results suggest that altered cell number and size in msc mutants might have arisen from upregulation of a few SPL genes, although the relative importance of individual SPL genes might differ in the respective msc mutants. To test whether the upregulation of these SPL genes accounts for the phenotypes of psd-16/msc2 or sqn-6/msc3, we crossed them with 35S:miR156. The phenotypes of psd-16/msc2 and sqn-6/msc3 were effectively suppressed by miR156 overexpression (Fig. 6A,B), suggesting that their phenotypes are dependent on upregulated SPL genes.

Our finding that cell number and size change in relation to heteroblasty allowed us reconsider the phenotypes of two groups of other known mutants. The first group includes mutants exhibiting compensation syndrome and the second is an auxin resistant 1 (axr1) mutant.

As three msc mutants show accelerated phase change, one might assume that mutants exhibiting compensation syndrome that have a decreased cell number and increased cell size opposite to the msc mutants should show delayed phase change. To test this possibility, we investigated phase change in various compensation-exhibiting mutants by counting the number of rosette leaves with or without abaxial trichomes and cauline leaves. Some mutants (an3 and fugu1) clearly showed delayed juvenile-to-adult phase change, while others (ant, fugu2, fugu3-D,fugu4-D and fugu5) did not(Fig. 7A). However, note that the an3 mutant showed rapid leaf production (shorter plastochron) and similar flowering time as the wild type(Horiguchi et al., 2005);therefore, the seemingly delayed phase change in an3 might be an indirect consequence of the shorter plastochron. By contrast, fugu1showed markedly delayed flowering (Fig. 7A, increased rosette leaf number), suggesting that both the juvenile-to-adult vegetative phase change and vegetative-to-reproductive phase change were delayed. The modes of delay in phase change in these mutants differed from each other.

The other mutant, axr1, was originally reported as one whose leaves have fewer but normal-sized cells(Lincoln et al., 1990). Subsequently, we reported that the first leaves of axr1 mutants have smaller cells compared with the wild type(Horiguchi et al., 2006b). To resolve this discrepancy, we determined the number and size of cells in the first, third and fifth leaves. In axr1-3 mutants, cell number is significantly decreased in the third and fifth leaves(Fig. 7B). Moreover, cell size in the first, third and fifth leaves was almost the same in this mutant; as a consequence, cell size in the first or fifth leaves of axr1-3 was smaller or larger, respectively, than the corresponding leaves of the wild type, whereas cell size in the third leaves of axr1-3 and the wild type was similar (Fig. 7C). These results reasonably explain the paradoxical results reported previously and suggest that AXR1 is required for the control of cell size during phase change. To investigate the relationship between SPL-regulated heteroblastic change in cell size and AXR1, we examined axr1-3 msc1-D double mutants. The double mutants showed increased cell number and slightly decreased cell size compared with the parental axr1-3 in all leaves tested (Fig. 7B,C),indicating that msc1-D and axr1-3 mutations have additive effects on cell number and size.

In this report, we described three msc mutants that have an increased cell number and decreased cell size. These mutants show acceleration in vegetative phase change. As adult leaves of the wild type have an increased cell number and a decreased cell size (Fig. 3A,B), precocious acquisition of the characteristics of adult leaves may be the cause of the phenotypes of these mutants. The results indicate that an unknown factor(s) underlying heteroblasty has an important role in the regulation of the cell number and cell size of leaves. Although an increase in cell number and a decrease in cell size in leaves at higher nodes have been observed in several species(Ashby, 1948; Granier and Tardieu, 1998; Cnops et al., 2004; Cookson et al., 2007), the cause of this phenomenon was unclear until the present study. In this report,we propose that miR156 and SPL genes are involved in heteroblastic change of cell number and size (see below). Our findings highlight the novel roles of SPL transcription factors in the regulation of heteroblasty.

A simple explanation for the increased cell number and decreased cell size in leaves at higher nodes is that a prolonged cell proliferation period in leaves at higher nodes may cause a shortening of the cell expansion period,resulting in the decreased final cell size. However, we cannot rule out another possibility: namely that a change in cell size is genetically separable from that in cell number. A notable example of this possibility is found in the rotunda2 (ron2) mutant, which is defective in the transcriptional co-repressor LEUNIG. In ron2, the cell size in the third leaf is as large as that in the first leaf of the wild type or ron2, whereas the cell number is almost the same as that of the third leaf of the wild type (Cnops et al.,2004). Another intriguing example is the axr1-3 mutant,in which cell size is almost the same in any leaf tested(Fig. 7C), suggesting that auxin signal transduction is required for the heteroblastic regulation of cell size. Analysis on these types of mutants in relation to the SPL genes will help us to distinguish these possibilities.

With regard to heteroblasty-dependent cell size control, note that cell size within each leaf at different nodes is fairly uniform in Arabidopsis (see Fig. S1 in the supplementary material), as also seen in sunflower leaves (Granier and Tardieu,1998). By contrast, many other heteroblastic traits are expressed heterogeneously within leaves produced during the transition from juvenile to adult phases. In these intermediate leaves, the distal part, which first arises from the shoot apical meristem, expresses more juvenile traits, whereas the proximal part, which arises later, expresses more adult traits(Kerstetter and Poethig,1998). The highly homogeneous cell size within an intermediate leaf, such as the fifth leaf (see Fig. S1 in the supplementary material),suggests that the cell size is controlled at the organ level, whereas other traits are controlled in a cell-autonomous manner.

We propose that miR156-mediated regulation of SPL genes,including SPL15 and SPL3, -4 and -5, is involved in heteroblastic regulation of cell number and size for the following reasons. The spl15-1D/msc1-D mutant has a mutation in the miR156 target site of the SPL15 gene, and expression levels of SPL15 are elevated in this mutant(Fig. 2A-D). In addition, miR156-overexpressing plants showed severe defects in heteroblastic change in cell number and size (Fig. 4E,F). Transgenic plants overexpressing the miR156-insensitive form of SPL3 or SPL4 have increased cell number and decreased cell size, particularly in the first leaf(Fig. 4E,F). Moreover, in the psd-16/msc2 and sqn-6/msc3 mutants, expression levels of a few SPL genes are significantly higher than in the wild type(Fig. 5B).

In the rdr6, sgs3 and zip mutants, in which heteroblasty is accelerated but miR156 levels are not changed(Peragine et al., 2004), some leaves have an increased cell number but do not show reduced cell size, except for the fifth leaf of sgs3 (Fig. 4A,B). The results indicate that these genes have only minor effects on the heteroblastic regulation of cell number and size. Because SPL3 expression levels were slightly (1.2- to 1.5-fold) elevated in these mutants (Peragine et al.,2004), such increases could affect cell number and size.

Because tasiR-ARF expression is not detected in rdr6,sgs3 and zip mutants(Peragine et al., 2004), tasiR-ARF-mediated regulation of ETT/ARF3 and ARF4is unlikely to be involved in heteroblastic change of cell number and size. This differs from the control of trichome production on the abaxial side of leaves, which is regulated by both SPL genes and by ETT/ARF3and ARF4 (Hunter et al.,2006; Wu and Poethig,2006). Under our experimental conditions, rdr6 and sgs3 clearly showed acceleration of abaxial trichome emergence, as previously reported (Peragine et al.,2004), although zip did not show this effect for unknown reasons (see Fig. S3 in the supplementary material). These facts suggest that heteroblastic regulation of abaxial trichome production and cell number and size are under the control of different pathways. This is further supported by the result that neither the arf3 nor the arf4 single mutant,which exhibit a delay in abaxial trichome emergence(Hunter et al., 2006), showed altered cell number or size compared to the wild type(Fig. 4C,D). The arf3-2 arf4-2 double mutant also showed no change in cell size(Fig. 4D). Although cell number in this latter mutant was reduced, heteroblastic change in cell number occurred in a similar manner to that in the wild type(Fig. 4C,D). The reduced cell number could have been caused by disturbed leaf polarity, because the arf3-2 arf4-2 double mutant shows defects in leaf adaxial/abaxial polarity and altered leaf morphology(Pekker et al., 2005).

spl15-1D/msc1-D is particularly intriguing, because a single synonymous nucleotide substitution occurred in the miR156 target site of the SPL15 gene (Fig. 2A). Transcriptional regulation of the gene should be the same in spl15-1D/msc1-D and in the wild type, suggesting that miR156-mediated cleavage of SPL15 transcripts indeed plays a role in control of heteroblasty. This finding is consistent with a recent report that transgenic plants expressing the miR156-resistant SPL9 gene under the control of its own promoter show significantly accelerated phase change (Wang et al.,2008). As only SPL15 mRNA level was upregulated in spl15-1D/msc1-D, the mutated SPL15 could have been the primary cause of the mutant phenotypes. In addition, because miRNAs, including miR156, also affect the translation of target genes(Gandikota et al., 2007; Brodersen et al., 2008), SPL15 protein level might be upregulated in msc1-D.

In psd-16/msc2 and sqn-6/msc3 mutants, the expression levels of miR156 slightly decreased and those of a few SPLgenes increased. However, the significance of such a slight reduction in miR156 levels is not clear because only a subset of SPLgenes was affected in their mRNA accumulation in these mutants(Fig. 5B). The increased expression of SPL genes should cause an alteration in cell number and size in leaves. This is further supported by the fact that increased cell number and decreased cell size in these two mutants were effectively suppressed by the overexpression of miR156(Fig. 6). The PSD gene encodes exportin-t, which regulates tRNA processing and nuclear export, but does not affect accumulation or export of miRNA(Park et al., 2005). SQN encodes cyclophilin 40, a protein that associates with the Hsp90 chaperone complex (Berardini et al.,2001), but its precise molecular function in plants is unknown. Although how these genes affect the expression levels of miR156 or SPL genes is unknown, one possibility is that they regulate expression of a subset of SPL genes by unknown mechanisms. If this is correct, psd and sqn are useful mutants to clarify the mechanisms of heteroblasty operating upstream of the SPL genes.

We propose that the heteroblasty-promoting SPL genes, including SPL3, -4, -5 and -15, increase cell number and reduce cell size in leaves. However, how these genes regulate cell number and size is still unknown. An intriguing possibility is that SPLs function through auxin signaling. Nevertheless, the additive phenotypes of the axr1-3 msc1-D double mutant indicate that SPL15 and AXR1 function in at least partially non-overlapping pathways.

SPLs comprise a family of transcription factors that share the SBP domain,a DNA-binding domain first identified in a protein that binds to a promoter of the SQUAMOSA gene in Antirrhinum majus(Klein et al., 1996; Cardon et al., 1997; Cardon et al., 1999). The amino acid sequence of the SBP domain is highly conserved among SPL proteins, but the sequence outside the SBP domain is diverse(Cardon et al., 1999). Although they are thought to bind to similar DNA sequences(Cardon et al., 1999), the molecular functions or target genes of SPLs remain unclear. Among the 10 SPL genes that have the miR156 target site, SPL3,-4, -5, -9 and -15 are involved in the heteroblastic regulation of abaxial trichome production(Wu and Poethig, 2006; Schwarz et al., 2008). In this study, we found that heteroblastic change in cell number and size is also regulated by SPL3, -4, -5 and -15,indicating that these genes might have overlapping functions. Among six SPL genes without the miR156 target site, SPL14 is particularly interesting because the loss-of-function mutant of this gene seems to have a truncated juvenile vegetative phase(Stone et al., 2005). SPL14 could have an antagonistic function to other SPL proteins that promote vegetative phase change. Identifying the targets of SPL transcription factors in further investigations will be necessary.

Precise regulation of cell number and cell size is essential for leaf development. However, the genetic network controlling cell number and size in the context of leaf development is too complex to establish a unified view on its regulation. Instead, finding a particular pathway that confers unique effects on leaf development and comparing the relationship among individual pathways are needed. In the present study, we show that increased cell number and reduced cell size in msc mutants are caused by accelerated heteroblasty. In contrast to msc mutants, the compensation syndrome,in which cell number is reduced and cell size is increased, is not necessarily caused by a genetic pathway associated with heteroblasty, because most compensation-exhibiting mutants do not show delayed phase change(Fig. 7A). Two compensation-exhibiting mutants (an3 and fugu1) indeed show a delay in phase change. However, they had an increase in cell number and a decrease in cell size in leaves at higher nodes than the wild type (see Fig. S4 in the supplementary material), indicating that they are not deficient in the heteroblastic change in cell number and size. In addition, the miR156-overexpressing plant did not show a reduction in cell number and an increase in cell size in the first leaf(Fig. 4E,F). In other words,the first leaves might be in the ground state of the vegetative phase and cannot revert beyond it to a further immature phase. However, compensation syndrome can be seen in the first leaves (see Fig. S4 in the supplementary material). This indicates that cell number and cell size are controlled by at least two distinct genetic pathways, i.e. one associated with heteroblasty and one related to the compensation syndrome. Increasing numbers of genes identified from mutants with altered cell number/size will help us to understand the details of the genetic network and the mechanisms controlling cell number and size.

In summary, we conclude that heteroblasty has an important role in the regulation of cell number and cell size. Investigating the factor(s) that regulate cell number and size under the control of heteroblasty, particularly downstream of the SPL genes, is crucial for understanding the mechanisms that regulate cell number, cell size and organ size during leaf development.