Arabidopsis Homologs of the Petunia HAIRY MERISTEM Gene Are Required for Maintenance of Shoot and Root Indeterminacy

Maintenance of indeterminacy is fundamental to the generation of plant architecture and a central component of the plant life strategy. Indeterminacy in plants is a characteristic of shoot and root meristems, which must balance maintenance of indeterminacy with organogenesis. The Petunia hybrida HAIRY MERISTEM ( HAM ) gene, a member of the GRAS family of transcriptional regulators, promotes shoot indeterminacy by an undeﬁned non-cell-autonomous signaling mechanism(s). Here, we report that Arabidopsis ( Arabidopsis thaliana ) mutants triply homozygous for knockout alleles in three Arabidopsis HAM orthologs ( Atham1,2,3 mutants) exhibit loss of indeterminacy in both the shoot and root. In the shoot, the degree of penetrance of the loss-of-indeterminacy phenotype of Atham1,2,3 mutants varies among shoot systems, with arrest of the primary vegetative shoot meristem occurring rarely or never, secondary shoot meristems typically arresting prior to initiating organogenesis, and inﬂorescence and ﬂower meristems exhibiting a phenotypic range extending from wild type (ﬂowers) to meristem arrest preempting organogenesis (ﬂowers and inﬂorescence). Atham1,2,3 mutants also exhibit aberrant shoot phyllotaxis, lateral organ abnormalities, and altered meristem morphology in functioning meristems of both rosette and inﬂorescence. Root meristems of Atham1,2,3 mutants are signiﬁcantly smaller than in the wild type in both longitudinal and radial axes, a consequence of reduced rates of meristem cell division that culminate in root meristem arrest. Atham1,2,3 phenotypes are unlikely to reﬂect complete loss of HAM function, as a fourth, more distantly related Arabidopsis HAM homolog, AtHAM4 , exhibits overlapping function with AtHAM1 and AtHAM2 in promoting shoot indeterminacy.

Indeterminate growth, the continuing generation and growth of organs and tissues throughout the life cycle of an organism, is a fundamental component of postembryonic plant development. Vascular plants grow discontinuously throughout their life spans, repeatedly initiating new shoot and root systems. This capacity for growth throughout the life span permits plants to adaptively regulate their growth patterns in response to dynamic environments, since, as sessile organisms, they cannot relocate in response to environmental stressors. Indeterminate growth is also a fundamental aspect of the "life strategy" of vascular plants, endowing woody perennials with the capacity for individuals to achieve life spans of thousands of years.
In plants, indeterminate growth is the function of plant meristems. Lateral organs (leaves and floral organs) and stems are derived from shoot meristems, located at shoot apices. Root meristems, internal meristems located immediately above the columella of root apices, generate the radially organized tissues of the root. The primary shoot apical and root apical meristems arise during embryogenesis, while secondary meristems arise de novo during postembryonic development (McConnell and Barton, 1998;Laskowski et al., 2008). Meristems must balance two competing functions: specification of determinate tissues, which reduces the pool of undifferentiated and pluripotent cells; and maintenance of indeterminacy, which requires the retention of a pool of undifferentiated and pluripotent cells from which cells lost to differentiating tissues may be replaced. The dual-functional nature of meristems is reflected in meristem structure. Small populations of internally located cells function as organizing centers, signaling to maintain an undifferentiated state in adjoining meristematic cells. In shoot meristems, undifferentiated cells are located immediately above the organizing center and constitute the "central zone," while in the root, undifferentiated initial cells surround the organizing center. Cells displaced from the shoot meristem central zone or root meristem initial zone ultimately undergo differentiation (Dinneny and Benfey, 2008). Stuurman et al. (2002) identified the HAIRY MER-ISTEM (HAM) protein, a member of the GRAS family of transcription factors, as a component of a novel non-cell-autonomous signaling pathway maintaining shoot indeterminacy in Petunia hybrida. Wild-type Petunia plants produce as many as 19 leaves before transitioning to flowering. ham mutants exhibit cessation of lateral organ and stem production (meristem arrest) and differentiation of the shoot apical meristem into stem tissue following the production of six to 14 leaves (Stuurman et al., 2002). Arrest in lateral organ production in ham mutants is similar to the wus phenotype in both Arabidopsis (Arabidopsis thaliana) and Petunia, but differentiation of the shoot meristem is unique to ham mutants (Laux et al., 1996;Stuurman et al., 2002). HAM is expressed in the provasculature and internal regions of the meristem subtending initiating lateral organs. An upward expansion of HAM expression occurs below sites of lateral organ initiation and continues into the emerging organ, remaining contiguous with HAM expression in the stem provasculature. HAM expression is not reported in central zone meristem cells, and HAM expression in meristem L3 cells is sufficient to restore meristem function. Collectively, these data were interpreted by Stuurman and colleagues (2002) as consistent with HAM promoting shoot indeterminacy via a non-cell-autonomous pathway.
Arabidopsis orthologs of Petunia HAM are probable endogenous targets of posttranscriptional degradation by microRNAs (MIRs) 170 and 171 (Llave et al., 2002). MIRs are small (21-24 nucleotide) RNAs produced from endogenously encoded RNA precursors (miRs) that act to direct posttranscriptional silencing of target mRNAs (Baulcombe, 2004;Brodersen and Voinnet, 2006). Within the past decade, posttranscriptional regulation of expression by MIRs has emerged as a major regulatory mode of meristem regulation and organ patterning. mRNA of two Arabidopsis HAM orthologs is shown to be cleaved at the MIR171/170-binding site in inflorescence tissue, coincident with the highest detected levels of miR171 (Llave et al., 2002;Parizotto et al., 2004). Additional support for endogenous regulation of Arabidopsis HAM orthologs by MIR170/171 arises from the observation that mRNA levels of one ortholog are elevated in mutant plants in which MIR production is impaired (Vazquez et al., 2004).
Continuing genetic analyses of HAM function in regulating indeterminacy would be greatly furthered by the identification and characterization of Arabidopsis ham mutants. In this study, we present the results of a detailed characterization of shoot and root phenotypes resulting from loss of function in Arabidopsis homologs of Petunia HAM. We show that while the four Arabidopsis HAM homologs span the range of flowering plant HAM diversity, all four Arabidopsis HAM homologs promote shoot indeterminacy. Arabidopsis orthologs of Petunia HAM are shown to regulate root indeterminacy as well, placing HAM orthologs into a comparatively small set of meristem regulators that function in the regulation of both shoot and root meristems. These results expand upon our understanding of HAM protein function in postembryonic development and provide a foundation for both future genetic analyses of HAM function and characterization of the molecular phenotype of ham mutants.

HAM Genes Underwent Expansion of Homolog Diversity and Elevated Rates of Evolution in Flowering Plants
Earlier phylogenetic analyses demonstrate that HAM genes have a long evolutionary history in land plants. HAM homologs are present in the genomes of representative species from the moss, lycophyte, and fern lineages (Floyd and Bowman, 2007;E.M. Engstrom, unpublished data). Among completely sequenced plant genomes, the moss Physcomitrella patens and the lycophyte Selaginella moellendorffii each possesses a single HAM homolog, while the flowering plants rice (Oryza sativa) and Arabidopsis each possesses four HAM homologs (Bolle, 2004;Tian et al., 2004;E.M. Engstrom, unpublished data). Expansion of homology diversity suggests the acquisition of new functions and/or functional subspecialization (Lynch, 2007). To determine whether the expansion of HAM homologs observed in rice and Arabidopsis is broadly characteristic of flowering plants as a group, or alternatively is a trait that arose independently in discrete lineages within flowering plant diversity, we undertook a phylogenetic analysis of HAM proteins from 23 flowering plants, two gymnosperms, and single representative species from the moss, fern, and lycophyte lineages (Supplemental Table S1). Despite the likelihood that this analysis does not include the complete set of HAM homologs from species for which complete genomic sequence is not yet available, expansion and diversification of HAM homologs are evident across the monocot and core eudicot lineages, with as many as eight HAM genes in the poplar (Populus spp.) and soybean (Glycine max) genomes (Fig. 1A). Diversification of flowering plant HAM genes initiated with a split into two major clades (HAM I and HAM II) prior to the divergence of the monocot and core eudicot lineages. Core eudicots have retained HAM homologs from both clades, while monocots retain HAM homologs of the HAM II clade only. The rate of amino acid sequence change in flowering plant HAM homologs is Three Arabidopsis HAM homologs, AtHAM1 (At2g45160), AtHAM2 (At3g60630), and AtHAM3 (At4g00150), were previously identified as targets of posttranscriptional regulation by MIRs 170/171 (Llave et al., 2002;Rhoades et al., 2002). The MIR170/171binding sequence, 5#-GATATTGGCGCGGCTCAAT-CA-3#, is perfectly conserved within HAM II eudicots, and within moss, lycophyte, and gymnosperm HAM homologs, strongly indicating that MIR regulation is an ancestral trait of HAM genes that arose before the divergence of the moss and vascular plant lineages (Fig. 1, A and B;Axtell et al., 2007). AtHAM1, AtHAM2, and AtHAM3 are located in the HAM II clade, along with Petunia HAM, and are more closely related to one another than to other members of the HAM II clade. AtHAM1, AtHAM2, and AtHAM3, therefore, appear to be paralogs derived from a relatively recent set of gene duplication events and are orthologs of Petunia HAM. AtHAM1, AtHAM2, and AtHAM3 exhibit comparable levels of protein sequence identity over their alignable GRAS domains with Petunia HAM, ranging from 52% for AtHAM2 to 58% for AtHAM3 (Fig. 1C).
The fourth HAM homolog, AtHAM4 (At4g36710), resides in the HAM I clade (Fig. 1A). With the exceptions of HAM homologs from Aquilegia and Vitis, all members of the HAM I clade for which complete sequence is available exhibit lack of conservation of the ancestral MIR-binding sequence. The AtHAM4 sequence homologous to the MIR170/171-binding sequence diverges from the ancestral HAM MIR-binding sequence at six of 21 nucleotides, making it highly unlikely that AtHAM4 is regulated by MIR170/171 (Fig. 1B). Protein sequence identity of AtHAM4 to Petunia HAM is comparable in degree to the much more distantly related Physcomitrella HAM homolog (Fig. 1C). Determining the domains of AtHAM ortholog expression must account for MIR170/171-mediated posttranscriptional regulation. Promoter::reporter fusion constructs generally do not reflect posttranscriptional regulation, while RNA detection methods optimally should distinguish between full-length mRNA and products of MIR-directed cleavage. Llave et al. (2002) report the expression of full-length AtHAM2 and AtHAM3 in leaf, stem, and inflorescence tissue, detected by RNA gel blot (Llave et al., 2002). Expanding upon this work, we undertook to amplify segments of HAM orthologs AtHAM1, AtHAM2, and AtHAM3 by reverse transcription (RT)-PCR from cDNA derived from entire vegetative shoots, mature rosette leaves, inflorescence stem and flowers, fully expanded but unripe siliques, and roots using primer sets that discriminate full-length transcripts from MIR170/171 cleavage products. AtHAM1, AtHAM2, and AtHAM3 are consistently amplified from all tissues surveyed (Supplemental Fig. S1), consistent with all three Arabidopsis HAM orthologs functioning in root, vegetative shoot, and reproductive shoot tissues.
Petunia HAM gene expression in both vegetative and inflorescence shoot meristems is associated most strongly with differentiating organ anlagen and provasculature, with expression highly reduced or excluded from central and apical meristem zones, and extending into the interior tissues of developing lateral organ primordia (Stuurman et al., 2002). To determine if and to what extent the patterns of expression of Arabidopsis HAM orthologs are similar to that of Petunia HAM, we performed in situ hybridization experiments to determine the localization of AtHAM1 expression in shoot apices of 12-d-old wild-type seedlings. In situ hybridization reveals the expression of AtHAM1 in young leaves and differentiating leaf primordia, with the level of expression elevated in interior regions of leaf primordia ( Fig. 2A). Within the shoot meristem, AtHAM1 expression exhibits a gradient of expression, with the lowest level of expression in the meristem L1 layer and increasing expression pro- Figure 1. (Continued.) tree is rooted with a set of 12 DELLA proteins, although for visual simplicity only the Arabidopsis GA-INSENSITIVE protein is shown. Proteins encoded by genes that retain a perfectly conserved MIR-binding sequence are colored blue; proteins encoded by genes in which the MIR-binding sequence is imperfectly conserved are colored red. The phylogenetic locations of inferred losses in MIR-binding sequence conservation are indicated by slanted orange bars. The two largest monophyletic clades of flowering plant HAM proteins are designated HAM I and HAM II. B, Evolution of the MIR-binding sequence in flowering plant HAM genes. The MIR170/171-binding site sequence of AtHAM3 (Llave et al., 2002) is shown, along with the homologous sequences of HAM I and HAM II genes that deviate from the ancestral MIR-binding sequence and the conserved MIR-binding sequences of Pinus, Selaginella, and Physcomitrella. Nucleotides conserved with the ancestral MIR-binding sequence are colored blue; nucleotides that deviate from the ancestral MIR-binding sequence are colored red. C, Relative amino acid sequence identity of Arabidopsis homologs to Petunia HAM. A more distantly related HAM homolog from Physcomitrella is included for comparison. Percentage of pairwise amino acid identity between the aligned C-terminal GRAS domain with Petunia HAM, excluding alignment gaps, is indicated to the left of each Arabidopsis and Physcomitrella homolog. Indicated GRAS domain subunits follow the criteria proposed by Tian et al. (2004).
gressing downward into L3 meristem cells. Immediately basal of the shoot meristem boundary, there is a sharp decrease or cessation in AtHAM1 expression, though more laterally, AtHAM1 expression is maintained in differentiating stem provasculature.
HAM genes are expressed in roots of both Petunia and Arabidopsis (Stuurman et al., 2002;Supplemental Fig. S1). To determine the expression patterns of AtHAM orthologs in the root at high spatial resolution, we examined transcriptional profiles of individual root cell types for all three Arabidopsis HAM orthologs, utilizing AREX, the Arabidopsis Gene Expression Database (Birnbaum et al., 2003;Brady et al., 2007;Carlsbecker et al., 2010;Sozzani et al., 2010). Root expression patterns of AtHAM1, AtHAM2, and AtHAM3 share considerable overlap with regard to the Figure 2. Arabidopsis HAM orthologs are expressed in meristematic and differentiated tissues of both the shoot and the root. A, In situ localization of AtHAM1 in a Ler shoot apex in a median longitudinal section. The L1 meristem cell layer is indicated with the arrowhead. A strong signal is consistently detected in lateral organ primordia. Signal is also detected in the meristem itself, with the highest level of signal present in the basal meristem regions and reduced or no signal in evidence in the uppermost cell layers of the central meristem region. AtHAM1 expression in the provasculature is indicated with the arrow. B, Expression maps of Arabidopsis HAM orthologs in root tissue, from The Arabidopsis Gene Expression Database (Birnbaum et al., 2003;Brady et al., 2007). Darker hues reflect higher relative expression levels within the root and between AtHAM orthologs. C, Relative expression levels of AtHAM orthologs in specific cell and tissue types. Values graphed are means of three replicates of normalized expression levels derived from mixed-model ANOVA analysis profiled by microarray profiling. Specific cell and tissue types are indicated, along with the marker employed to delineate spatial expression patterns in parentheses. Data shown are derived from the analysis reported by Brady et al. (2007), with the exceptions of cortex/endodermal initial (CEI; Sozzani et al., 2010) and mature endodermis (Mat ENDO; Carlsbecker et al., 2010). QC indicates the quiescent center.
cell types and tissues in which they are expressed, but they exhibit ortholog-specific relative expression levels ( Fig. 2, B and C). Within the meristem region, all three orthologs are expressed in quiescent center, cortex/ endodermal initials, and endodermis, cortex, and stele cell files. In differentiating and mature tissues, all three orthologs are expressed in columella, root cap, epidermis, cortex, endodermis, and stele. Moreover, all three orthologs exhibit a striking pattern of expression in the epidermis, with significantly elevated expression in trichoblast epidermal cell files relative to atrichoblast epidermal cell files in differentiating and mature root. This pattern is inverted within the meristem, with atrichoblast cell files exhibiting AtHAM ortholog expression, while trichoblast cell files show low or no AtHAM ortholog expression. Ortholog-specific differences in relative expression levels are greatest in differentiating and mature trichoblast, where AtHAM2 predominates, in phloem and protophloem, where AtHAM3 predominates, and in developing xylem, where AtHAM2 again predominates. Within the meristem region, AtHAM3 expression is elevated relative to AtHAM1 and AtHAM2 in a radial root section three cells in height, at the transition zone between the root meristem and elongation zone. AtHAM2 expression is elevated relative to AtHAM3 in a longer bipartite cross section of the root meristem, immediately adjacent to and below the band of elevated AtHAM3 expression, while AtHAM1 expression is elevated relative to AtHAM3 in a cross section of root meristem located several cells above the quiescent center and overlapping with the band of elevated AtHAM2 expression. Our phylogenetic analysis of HAM proteins suggests that the Petunia ham phenotype of shoot apical meristem arrest and differentiation is most likely to be recapitulated in Arabidopsis by loss-of-function mutants of AtHAM1, AtHAM2, and AtHAM3. We identified insertional mutant alleles, predicted to confer complete loss of function, for AtHAM1, AtHAM2, and AtHAM3 (Supplemental Fig. S2A). Consistent with the insertions generating null loss-of-function (knockout) alleles, wild-type transcripts are not detectable by RT-PCR in homozygous insertion allele backgrounds (Fig.  3, A and B). As Atham2-1 and Atham3-1 both reside in the Landsberg erecta (Ler) background, we elected to use the Ler genotype as a wild-type reference in our characterizations of Atham1,2,3 mutants.
Single AtHAM knockout mutants and all three combinations of AtHAM double mutants (Atham1,2, Atham1,3, and Atham2,3 mutants) do not notably differ in their shoot phenotypes from the wild type (data not shown), consistent with considerable functional redundancy among AtHAM orthologs. Plants homozygous for knockout alleles in all three AtHAM orthologs (Atham1,2,3 mutants) exhibit a spectrum of abnormal shoot phenotypes. Atham1,2,3 mutants do not exhibit notable embryogenesis defects, as gauged by normal development of hypocotyl, cotyledons, and root and shoot meristems (Fig. 3C). The earliest evident phenotypic abnormalities in Atham1,2,3 mutants are deviations from the normal phyllotactic patterning of rosette leaves, which frequently are apparent as early as the emergence of the third and fourth leaves (Fig. 3D). At this stage in development, examination of sectioned shoot apices reveals that Atham1,2,3 shoot apical meristems are consistently broader and flatter than in the wild type (mild fasciation), although overall meristem size is not appreciably different (Fig. 3E). Fully expanded rosette leaves of Atham1,2,3 mutants typically exhibit less pronounced laminar growth and more pronounced leaf serration relative to wild-type expanded rosette leaves (Fig. 3F). Examination of the adaxial and abaxial epidermis of wild-type and Atham1,2,3 expanded rosette leaves by scanning electron microscopy reveals that epidermal cell surface area is significantly greater in both adaxial and abaxial epidermal pavement cells of Atham1,2,3 mutants relative to the wild type, while guard cells and trichomes are of comparable size in Atham1,2,3 rosette leaves relative to the wild type (Fig. 3G). Comparison of adaxial and abaxial epidermal surface characteristics in Atham1,2,3 mutants does not suggest leaf polarity defects.
Shoot phenotypes of Atham1,2,3 mutants are most evident following the transition to reproductive growth, whereupon wild-type Arabidopsis plants typically exhibit secondary shoots emerging from axils of both rosette and inflorescence leaves (Fig. 3H). In mature Atham1,2,3 mutant plants, secondary inflorescence stems rarely emerge from leaf axils of either rosette or primary inflorescence stem (Fig. 3H). Close examination of Atham1,2,3 rosette axils by scanning electron microscopy reveals discrete populations of comparatively small cells, consistent with axillary meristem formation and subsequent axillary meristem arrest prior to the initiation of organogenesis of subtending stem or lateral organs (Fig. 3I). Higher magnification of rosette axils reveals stomatal pores in the axillary "meristem" epidermis, indicating differentiation of meristem cells following meristem arrest (Fig.  3J). Occasional rosette and inflorescence axils produce single, radially symmetrical organs (Fig. 3K).
While arrest of the primary vegetative meristem has not been observed to date in Atham1,2,3 mutants, loss of indeterminacy occurs, with incomplete penetrance, in flower and primary inflorescence meristems. Atham1,2,3 mutants are significantly delayed in the development of mature flowers relative to the wild type, consistent with delayed inflorescence development and/or delayed transition from vegetative to reproductive development (Fig. 4A). A subset of Atham1,2,3 mutants exhibit loss of shoot indeterminacy following the transition to reproductive development, characterized by cessation of flower production and significant enlargement of the stem apex relative to the wild type ( Fig. 4B; Supple-mental Video S1). Loss of indeterminacy in the inflorescence meristem is preceded by a gradual loss of meristem indeterminacy in flowers (Fig. 4, B and C), initially evident by flowers with reduced or missing inner whorls (F4 in Fig. 4C), progressing to flowers that fail to initiate lateral organs (F3 and F2 in Fig. 4C), and culminating in flowers that lack both pedicle and lateral organ development (F1 in Fig. 4C). As with the vegetative shoot meristem, functioning Atham1,2,3 inflorescence meristems do not differ from the wild type with respect to meristem size, indicating that enlargement of the inflorescence apex in arrested Atham1,2,3 inflorescences is likely a consequence of meristem differentiation rather than meristem enlargement preceding arrest (Fig. 4D). Further paralleling vegetative meristem abnormalities, Atham1,2,3 inflorescences consistently exhibit aberrant phyllotaxis (Fig. 4E). However, functioning Atham1,2,3 inflorescence meristems exhibit a suite of abnormalities not observed in vegetative apices, including vacuolization of meristem cells, supernumerary meristem cell layers, and an indistinct boundary between meristematic and differentiating tissue zones (Fig. 4D).
Atham1,2,3 cauline leaves typically exhibit epinastic curling, indicative of greater adaxial leaf surface area relative to abaxial leaf surface area (Fig. 4E). Analysis by scanning electron microscopy did not reveal appreciable differences in epidermal cell size between Atham1,2,3 and the wild type on either the adaxial or abaxial cauline leaf surface (data not shown), indicating that curling results from an increase in cell number in Figure 3. Atham1,2,3 mutants exhibit arrest and differentiation of secondary meristems and altered structure and function of the primary shoot apical meristem. A, RT-PCR analysis of wild-type Wassilewskija (Ws) and a homozygous Atham1-1 mutant with primers flanking the T-DNA insertion site (Supplemental Table S2). Primers designed to amplify ACTIN2 cDNA were employed as a control for cDNA quality. B, RT-PCR analysis of wild-type Ler and homozygous Atham2-1 and Atham3-1 mutants with primers flanking the Ds insertion sites (Supplemental Table S2). Amplification from genomic DNA (gDNA) is employed as a reference for product size and primer set efficacy. AtHAM2 and AtHAM3 primer sets serve as reciprocal controls for cDNA quality in Ler, Atham2-1, and Atham3-1 genotypes. C, Ler and an Atham1,2,3 mutant at 3 d postgermination on sterile medium. Atham1,2,3 mutants exhibit elongation of the primary root, demonstrating the presence of a root meristem. The hypocotyl and cotyledons are evident in Atham1,2,3 mutants and do not differ significantly from the wild type in appearance, although epinastic curvature of the cotyledons is common in Atham1,2,3 mutants. D, Ler and an Atham1,2,3 mutant at 12 d postgermination. Postembryonic leaves are labeled (numbering of leaves 1 and 2 is arbitrary, as these two leaves arise roughly simultaneously). By emergence of the fourth leaf, deviations from wild-type phyllotaxis are apparent in many Atham1,2,3 mutants. E, Longitudinal section through shoot apices of Ler and an Atham1,2,3 mutant at 12 d postgermination. Atham1,2,3 mutants consistently exhibit broader and flatter primary shoot apical meristems relative to the wild type at 12 d postgermination, but significant differences in meristem size in Atham1,2,3 mutants relative to the wild type are not evident. F, Set of fully expanded rosette leaves of Ler and an Atham1,2,3 mutant. Leaf 8 of Ler is largely missing from this set. Atham1,2,3 rosette leaves typically exhibit reduced laminar expansion relative to the wild type. G, Epidermal surfaces of fully expanded rosette leaves of Ler and an Atham1,2,3 mutant imaged by scanning electron microscopy. Increases in average epidermal cell surface area relative to the wild type are evident on both the adaxial (Ad) and abaxial (Ab) leaf surfaces of Atham1,2,3 mutants. H, Ler and Atham1,2,3 mutant shoot phenotypes postflowering. Secondary shoots are typically in evidence emerging from axils of both the rosette and inflorescence at this stage in the wild type but are rarely observed in Atham1,2,3 mutants. I and J, Surface of an arrested rosette axillary meristem of an Atham1,2,3 mutant visualized by scanning electron microscopy. The main inflorescence stem (IS) is indicated for positional reference in I. The arrowhead in J indicates a fully differentiated stomata. K, Radially symmetrical multicellular structure, indicated with the arrow, emerging from an arrested axillary inflorescence meristem of an Atham1,2,3 mutant, visualized by digital optical microscopy. the adaxial domain relative to the abaxial domain. Sectioning of Ler and Atham1,2,3 mutant cauline leaves reveals a significant increase in cell number along the adaxial/abaxial axis in Atham1,2,3 mutants relative to the wild type (Fig. 4F). Wild-type patterning of vascular bundles demonstrates that adaxial/abaxial polarity is normal in Atham1,2,3 mutant cauline leaves (Fig. 4G).
Both Petunia HAM and AtHAM orthologs are expressed in root tissue, suggesting a role for HAM function in root development, although root abnor-  Fig. 3H). A flower has developed in the axillary meristem of the right-most cauline leaf at the position normally occupied by a secondary inflorescence shoot. Epinastic curling of cauline leaves is typical of Atham1,2,3 mutants. F, Cross sections through the cauline leaf lamina of Ler and an Atham1,2,3 mutant stained with toluidine blue and visualized by bright-field microscopy. Doublearrowed lines represent the approximate median thickness of adjacent cauline leaves. G, Vascular bundles of Ler and Atham1,2,3 cauline leaves. Xylem (Xy) is positioned adaxial to phloem (Ph) in Atham1,2,3 cauline leaves.
malities are not reported for Petunia ham mutants (Stuurman et al., 2002;Fig. 2). To determine if Atham1,2,3 mutants are altered in root growth, we germinated seeds on sterile medium and monitored primary root elongation and overall root morphology. Alongside Atham1,2,3 and wild-type Ler seedlings, shortroot2 (shr2) mutants were grown as a reference for root meristem arrest. Atham1,2,3 mutants exhibit a significant reduction in elongation of the primary root relative to Ler, comparable in degree to shr2 mutants (Fig. 5, A and B). At 9 d following germination, primary root apices of Atham1,2,3 mutants appear normal with respect to radial organization of the root meristem, but the size of Atham1,2,3 root meristems is significantly reduced relative to the wild type in both longitudinal and radial axes (Fig. 5, C-E). A subset of Atham1,2,3 root meristems exhibit starch staining in cells immediately adjacent to the quiescent center, at the position normally occupied by columella initials, indicating an incompletely penetrant phenotype of accelerated differentiation of the columella (data not shown). The primary root meristem is delineated in the longitudinal axis as the region extending from differentiating columella to the elongation zone, in which cell size remains relatively constant via continuing cell divisions. Atham1,2,3 root meristems are significantly smaller in the longitudinal axis relative to the wild type, indicating either a reduced rate of cell division in the meristem or accelerated cellular differentiation and elongation (Fig. 5F). Reduction in root meristem diameter correlates with a reduced number of cells in radial tissue layers of Atham1,2,3 roots relative to the wild type (Fig. 5E). By 10 d following germination, radial meristem organization and the quiescent center are no longer discernible in a subset of Atham1,2,3 mutant root apices, indicating loss of root indeterminacy (Fig. 5G). Two of 33 Atham1,2,3 mutant roots examined exhibited root bifurcation, with two root meristems located adjacent to one another at a single root apex (root fasciation; Fig. 5H).
Reduced primary root elongation in Atham1,2,3 mutants prior to loss of root indeterminacy could be reasoned a priori to result from (1) a reduction in the magnitude of root cell elongation or (2) a reduced rate of cell division in the root meristem. To determine if Atham1,2,3 mutants exhibit reduced root cell elongation, we measured the length of atrichoblast cells within the zone of root hair elongation from Ler and Atham1,2,3 mutants. Atrichoblasts of Atham1,2,3 are not significantly different in length from atrichoblasts of wild-type seedlings, demonstrating that the reduction in primary root length in Atham1,2,3 mutants is not attributable to reduced root cell elongation (Supplemental Fig. S3). We conclude that reduced root elongation in Atham1,2,3 mutants is the result of decreased rates of cell division in the root meristem.
Regulation of root meristem identity is the function of the quiescent center, and maintenance of the root quiescent center requires the generation of an auxin maximum at the root apex Grieneisen et al., 2007). Loss of root indeterminacy in Atham1,2,3 mutants could result from the inability to generate a wild-type auxin maximum at their root apices. If this were the case, we would predict that functioning Atham1,2,3 root apices may exhibit auxin maximum defects prior to full meristem arrest. To test this hypothesis, we visualized relative auxin levels in the primary root apices of Columbia and Atham1,2,3 mutants using the DR5::GUS auxin reporter system at 6 d following germination, when reduction in primary root elongation is readily discernible in Atham1,2,3 mutants (Ulmasov et al., 1997). Atham1,2,3 mutants exhibit root apex auxin maxima that are comparable to the wild type in spatial expression and intensity, consistent with Atham1,2,3 mutant root phenotypes not being principally a consequence of altered auxin transport at the root apex (Fig. 5I).

Atham1,2,3 Mutant Root Hairs Exhibit Elevated Levels of Branching and Transient Loss of Anisotropic Tip Growth
In examining Atham1,2,3 roots, we observed a significant number of branched root hairs (Fig. 6). Fifty-two percent of Atham1,2,3 root hairs are branched (n = 10, 23 SEM = 11%), while no branched root hairs were detected in the Ler roots examined. Root hair branches of Atham1,2,3 mutants emerge from swollen sections of the main root hair shaft, consistent with a transient reversion from anisotropic tip growth to isotropic growth coincident with or immediately preceding the initiation of a second site of tip growth. Swellings in the root hair shaft without associated branches are also common on Atham1,2,3 roots but are rarely noted on Ler roots. The frequency of root hair branching and root hair swelling in Atham1,2,3 mutants is not continuous along the length of the primary root but occurs in patches of highly elevated branching, interrupted by regions of root hairs exhibiting wild-type anisotropic growth.
AtHAM4 Genetically Interacts with AtHAM1 and AtHAM2 in Promoting Shoot Indeterminacy The probable lack of MIR regulation in AtHAM4, coupled with the degree of evolutionary divergence between the HAM I and HAM II lineages, suggest that AtHAM4 may have evolved novel functions and patterns of expression and may exhibit limited functional overlap with AtHAM1, AtHAM2, and AtHAM3 (Fig. 1, A  and B). To determine if and to what extent AtHAM4 exhibits functional redundancy with Arabidopsis HAM orthologs, we obtained a knockout allele of AtHAM4 ( Fig. 7A; Supplemental Fig. S2) and undertook to generate multiple mutant combinations between Atham1, Atham2, Atham3, and Atham4. Atham4 mutants exhibit no striking abnormalities in shoot development, although they are consistently of smaller overall stature relative to the wild type (Fig. 7B). Atham1,4 double mutants similarly exhibit no striking shoot abnormalities (Fig. 7C). However, Atham1,4; Atham2/+ mutants exhibit a range of variably penetrant phenotypes, from a loss of primary  . Root phenotypes of Atham1,2,3 mutants. A, Ler, Atham1,2,3, and shr2 plants grown for 12 d following germination on sterile medium. Atham1,2,3 mutants exhibit comparable reductions in primary root length relative to the wild type as shr2 mutants. B, Primary root growth rate of Ler and Atham1,2,3 mutants grown for 12 d following germination on sterile medium. Error bars indicate 23 SEM. WT, Wild type. C, Lugol-stained roots of Ler and an Atham1,2,3 mutant at 9 d following germination viewed by bright-field microscopy. Purple staining indicates starch granules in differentiated columella cells. The position of the quiescent center is indicated with the arrowheads. The reduction in primary root diameter in Atham1,2,3 mutants relative to the wild type is evident. D to H, Optical cross sections through primary root apices of Ler and Atham1,2,3 mutants visualized by confocal microscopy. D, F, G, and H show longitudinal cross sections through the root meristems of Ler and Atham1,2,3 mutants at 9 d following germination. The boundary between the root meristem and elongation zone is indicated in F, with the white arrowhead in the Atham1,2,3 panel, and is outside the frame of the Ler panel. E, Radial cross sections of Ler and Atham1,2,3 mutant meristematic zones at 9 d following germination. Epidermis (Ep) and cortex (C) are indicated. G, An Atham1,2,3 mutant at 10 d following germination shows loss of meristem indeterminacy. H, Longitudinal section through the root apex of an Atham1,2,3 mutant exhibiting root meristem bifurcation. I, Root apices of Columbia (Col) and Atham1,2,3 mutant roots expressing the pDR5::GUS auxin reporter construct stained with 5-bromo-4-chloro-3-indolyl-b-D-GlcUA and visualized by differential interference contrast microscopy. The intensity of blue staining is proportional to free auxin concentration. Shown are primary roots at 6 d following germination.
inflorescence shoot dominance to arrest and differentiation of the inflorescence meristem prior to flower initiation (Fig. 7C). Atham1,4; Atham2/+ mutants frequently exhibit a secondary meristem phenotype characterized by the production of a novel organ consisting of a leaf subtended by a stem similar in size to a flower pedicle (Fig. 7D), a structure observed less frequently in Atham1,2,3 mutants (data not shown). Atham1,2,4 mutants consistently exhibit an absence of secondary rosette stems and arrest and differentiation of flower and primary inflorescence meristems, similar to the most extreme phenotypes of Atham1,2,3 mutants (Fig. 7, E and F). Collectively, these results demonstrate significant functional redundancy between AtHAM4 and AtHAM1 and AtHAM2 in promoting shoot indeterminacy.

Flowering Plant HAM Genes Are Likely to Possess Both Core Ancestral and Angiosperm-Specific Functions
The presence of HAM homologs in the genomes of the basal plants Physcomitrella and Selaginella, and conservation of the domain structure and MIR-binding sequence among distantly related HAM proteins, suggest that aspects of flowering plant HAM function may be derived from the common ancestor of bryophytes and vascular plants and that ancestral functions may be shared among extant flowering and basal plants. Determining what the core ancestral HAM function is must await genetic ablation of HAM homologs in Physcomitrella and Selaginella and/or the complementation of flowering plant ham mutants with basal plant HAM homologs.
The dramatic expansion in HAM homolog diversity in flowering plants strongly suggests the evolution of novel HAM functions or functional HAM subspecialization in angiosperms, while elevated rates of evolution in flowering plant HAM homologs indicate a refinement of HAM function in response to novel selective pressures. Supporting the model of acquisition of novel functions by at least a subset of flowering plant HAM genes is the probable loss of MIR-mediated regulation on multiple, independent occasions, suggesting a strong selective pressure in flowering plants that favored the expansion of HAM expression. Modification of expression domains may be a major evolutionary force preceding the acquisition of novel functions (Matsuno et al., 2009), and modification to MIR-binding sequences is proposed to be an evolutionarily significant source of variation generation (Chen and Rajewsky, 2007;Ehrenreich and Purugganan, 2008). Observed sequence divergences in HAM MIRbinding sequences are likely to reflect an absence of MIR-mediated posttranscriptional regulation, although binding of MIRs to their target sequences may be tolerant of low levels of internal and 5# nucleotide mismatch, and the possibility of novel miRs in the genomes of many of the species surveyed cannot be discounted (Mallory et al., 2004). The ancestral MIRbinding sequence is perfectly conserved in all eudicot members of the HAM II clade for which complete sequences are available, indicating a function common to all eudicot HAM II proteins that exerts strong selective pressure to conserve the ancestral MIRbinding sequence and limit HAM expression. Loss of MIR regulation may be preceded by the disruption of some component of HAM function that removes this constraint, or, alternatively, the loss of MIR regulation may occur in HAM paralogs that evolve nonoverlapping patterns of expression with their ancestral miR regulators.
All Four Arabidopsis HAM Homologs Promote Shoot Indeterminacy Atham1,2,4 mutants exhibit stronger shoot meristem arrest phenotypes than Atham1,2,3 mutants, demonstrating that AtHAM4 contributes significantly to the maintenance of shoot indeterminacy and functions redundantly with AtHAM1 and AtHAM2. These results suggest that some degree of HAM-mediated indeterminacy may be retained in triple mutant com- binations and that the generation and analysis of Atham1,2,3,4 quadruple mutants will be required to define the full contribution of AtHAMs to the maintenance of indeterminacy. The variable penetrance of meristem-arrest phenotypes in Atham1,2,3 and Atham1,4; Atham2/+ mutants is consistent with the retention of HAM function, the degree of HAM function retained being stochastic in the loss-of-function backgrounds examined. Similarly, the variably penetrant phenotype of Petunia ham, coupled with the observation that eudicots with completed genomes all possess at least two HAM homologs, suggest that at minimum a second HAM homolog resides in the Petunia genome, which functions redundantly with the originally identified HAM gene.

Do HAM Proteins Function Non Cell Autonomously?
Petunia HAM is inferred to promote shoot indeterminacy by a non-cell-autonomous mechanism on the bases of (1) the absence of detectable HAM expression in much of the shoot meristem and (2) the complementation of the ham phenotype by L3-restricted HAM expression (Stuurman et al., 2002). Non-cell-autonomous functioning is well established for the related GRAS protein SHR, and the VHIID and PYFRE do- Figure 7. Shoot phenotypes of Atham4, Atham1,4; Atham2/+, and Atham1,2,4 mutants. A, RT-PCR analysis of wild-type Columbia (Col) and a homozygous Atham4-1 mutant with primers flanking the T-DNA insertion site (Supplemental Table S2). Primers designed to amplify ACTIN2 cDNA were employed as a control for cDNA quality. B, Columbia and Atham4 mutant shoot phenotypes postflowering. Atham4 mutants consistently exhibit reduced stature relative to the wild type but do not otherwise exhibit obvious abnormal phenotypes. C, Shoot phenotypes of Atham1,4 mutant and Atham1,4; Atham2/+ genotypes. Atham1,4 mutants do not exhibit obvious shoot phenotype abnormalities. Shoot phenotypes of Atham1,4; Atham2/+ plants are highly variable. The middle plant is an Atham1,4; Atham2/+ plant exhibiting reduced dominance of the primary inflorescence stem, indicated by the arrow, relative to wild-type plants.
The right-most plant is an Atham1,4; Atham2/+ plant exhibiting an absence of secondary growth in the rosette and arrest of the primary inflorescence meristem prior to the initiation of flowers. D, Novel organ formation in the inflorescence axil of an Atham1,4; Atham2/+ plant. At the position normally occupied by a secondary inflorescence stem, a leaf subtended by a short stem resembling a flower pedicle is evident. The orientation of the adaxial leaf surface is away from the adjoining inflorescence stem. E, Shoot phenotypes of Atham1,2,4 mutants.
The left-hand panel shows an Atham1,2,4 mutant with arrested flower and inflorescence meristems. The inflorescence meristem is indicated by the arrow. No secondary stems are in evidence emerging from the rosette. The right-hand panel shows an Atham1,2,4 mutant inflorescence terminating in a single "trumpet-leaf" cauline leaf. F, Inflorescence apices of Atham1,4 and Atham1,2,4 mutants visualized by scanning electron microscopy. mains, conserved among SHR and HAM proteins, are required for intracellular movement of SHR (Nakajima et al., 2001;Gallagher and Benfey, 2009). AtHAM1 expression in Arabidopsis shoot apices overlaps significantly with the reported expression of Petunia HAM but clearly extends farther toward the Arabidopsis meristem apex than is reported for Petunia HAM, overlapping with expression domains of meristem autonomous regulators of indeterminacy such as WUSCHEL and SHOOTMERISTEMLESS (Long et al., 1996;Mayer et al., 1998). In the root, the potential for non-cell-autonomous promotion of root indeterminacy by AtHAMs appears low, as AtHAM1, AtHAM2, and AtHAM3 are expressed nearly constitutively in the root. In situ hybridization may fail to detect transcripts expressed at very low, yet potentially functional, levels; therefore, inferences that depend upon in situ hybridization to establish an absence of expression must be regarded cautiously. The data presented in this study do not directly inform our understanding of whether AtHAMs function cell autonomously or non cell autonomously. However, the detailed characterization of Atham loss-of-function phenotypes described here provides a foundation for developing future experiments to directly test the model of HAM functioning via a non-cell-autonomous pathway.
What Is the Cellular Function of HAM Proteins?
The highly pleiotropic phenotype of Atham1,2,3 mutants, including expanded anticlinal cell divisions in developing cauline leaves, reduced levels of cell division in root meristems, and root hair branching, coupled with the broad expression patterns of AtHAM1, AtHAM2, and AtHAM3 in shoot and root tissue demonstrate that HAM function is not limited to promoting organ indeterminacy and may inform the development of models for HAM function at the cellular level. The diverse suite of ham loss-of-function phenotypes may reflect HAM function in the transmission of a specific stimulus that regulates a broad spectrum of cellular and developmental processes, such as a hormone. In this context, it is notable that within the GRAS protein family, HAM proteins are most closely related to DELLA proteins, transcriptional regulators whose function and stability are mediated by gibberellins (Harberd et al., 2009;E.M. Engstrom, unpublished data). Alternatively, HAM proteins may regulate a specific cellular function that has broad developmental consequences, such as cell cycle progression. Canonical cell cycle regulatory proteins regulate stem cell identity, indeterminacy, and organ patterning in roots, suggesting that regulation of indeterminacy may occur at least in part through the regulation of cell cycle progression Sozzani et al., 2010). Moreover, the GRAS proteins SCARECROW and SHR are not only similarly required for the maintenance of root indeterminacy (Di Laurenzio et al., 1996;Helariutta et al., 2000;Sabatini et al., 2003) but also regulate the transcription of a set of cell cycle regulatory proteins, including RETINOBLASTOMA RELATED, D cyclins, and E2F factors in developing leaves and D cyclins and cyclindependent kinases in the root (Dhondt et al., 2010;Sozzani et al., 2010). Regulation of cell cycle progression is a plausible candidate for a cell-level HAM function that could explain much of ham loss-of-function phenotypes, and the expression of cell cycle regulators in Atham backgrounds warrants examination. ham1,2,3 Mutant Phenotypes Overlap Considerably with fasciata1 and fasciata2 Phenotypes FASCIATA1 (FAS1) and FAS2 encode components of the chromatin assembly factor 1, which regulates nucleosome assembly, chromatin silencing, and homologous recombination (Endo et al., 2006;Kirik et al., 2006;Ono et al., 2006;Schö nrock et al., 2006). "Fasciation" refers to phenotypes attributable to multifurcating shoot apices, resulting in supernumerary, fused lateral organs and flattened, ridged stems (Worsdell, 1905). In addition to exhibiting fasciation phenotypes, fas1 and fas2 exhibit highly pleiotropic shoot and root phenotypes, aspects of which significantly parallel Atham1,2,3 mutant phenotypes, including broadened and relatively flat shoot apical meristems, ectopic vacuolization in the meristem, abnormal phyllotaxis, enlarged epidermal cell size, impaired root growth, and occasional meristem arrest in fas2 mutants (Reinholz, 1966;Ottoline Leyser and Furner, 1992;Kaya et al., 2001;Exner et al., 2006). Further paralleling AtHAM lossof-function phenotypes, fas1 and fas2 mutant phenotypes are restricted to postembryonic development and exhibit enhanced branching of trichomes, similar to the enhanced branching of root hairs exhibited by Atham1,2,3 mutants (Kaya et al., 2001;Exner et al., 2006). Fasciation of the root, as is observed in Atham1,2,3 mutants, has not been reported to date in fas1 or fas2 mutants that we are aware of. As DNA modification enzymes, FAS1 and FAS2 are plausible candidates for physical interactions with HAM transcription factors.

Phylogenetic Analysis
Multiple sequence alignments were generated with the ClustalW program (Larkin et al., 2007) using Geneious Pro 4.6.4 software, employing a GONNET cost matrix, a gap opening penalty of 35, and a gap extension penalty of 0.75. Resulting alignments were then refined by eye using the Geneious Pro 4.6.4 software. GRAS proteins exhibit substantial sequence divergence in their N-terminal regions, precluding accurate alignment for roughly the N-terminalmost one-third of GRAS proteins. An aligned region corresponding to amino acids 358 to 721 of the Petunia hybrida HAM protein was extracted from the complete protein alignments and manually refined prior to phylogenetic analyses. Final protein alignments are available upon request from the corresponding author.
Phylogenetic analysis by the Bayesian inference method was performed with the MrBayes program (Huelsenbeck and Ronquist, 2001) via Geneious Pro 4.6.4 software, employing a Poisson amino acid rate matrix and a gamma rate variation. A total of 120,000 generations were generated, and every 1,000th tree was saved. Four independent chains were run with a temperature of 0.1, and the initial 30,000 trees were discarded as burn in. The analysis was

Plant Material
Arabidopsis (Arabidopsis thaliana) Atham1-1 seed was obtained from the INRA. The Atham1-1 insertion (FLAG_239F03; Samson et al., 2002) is localized 5# and in close proximity to position +742 relative to the translational start site in the Wassilewskija ecotype. Efforts to more precisely map the Atham1-1 insertion site have not been successful due to an inability to amplify fragments containing portions of the T-DNA. The wild-type allele of AtHAM1 was identified by PCR (Supplemental Table S3). Atham2-1 and Atham3-1 seed was provided by Dr. Venkatesan Sundaresan. The Atham2-1 insertion (SGT11982) is localized to position +1,303 relative to the translational start site in the Ler ecotype (Sundaresan et al., 1995). The Atham3-1 insertion (SGT13186) is localized to position +920 relative to the translational start site in the Ler ecotype (Sundaresan et al., 1995). Wild-type and Ds insertion alleles of AtHAM2 and AtHAM3 were identified by PCR (Supplemental Table S3). Insertion sites of Atham2-1 and Atham3-1 alleles were identified by sequencing PCR-amplified products derived from the Forward and Ds30 primer sets (Supplemental Table S3). Atham4-1 seed was obtained from the Arabidopsis Biological Resource Center. The Atham4-1 insertion (SALK_110871; Alonso et al., 2003) is localized to position +1,046 relative to the translational start site in the Columbia ecotype. shr2 seed (Helariutta et al., 2000) was obtained from the Arabidopsis Biological Resource Center. Atham2-1, Atham3-1, and Atham1,2,3 mutant lines have been deposited at the Arabidopsis Biological Resource Center.

Growth Conditions and Measurements
Soil-grown plants were maintained at 20°C with a 16-h daylength under cool-white fluorescent bulbs (8.3 6 0.93 kilolux). For analysis of root phenotypes, seeds were germinated and maintained on vertically oriented plates containing 0.53 Murashige and Skoog salts, 1% Suc, and 0.7% phytagel (Sigma-Aldrich). For measurements of root length, photographs of entire seedlings were taken at intervals through a Zeiss Discovery V12 stereomicroscope equipped with a 0.633 objective and coupled to a Nikon D50 SLR digital camera. For measurements of root cell length, seedlings grown on vertical plates were removed, placed in a bath of 0.53 Murashige and Skoog salts on a coverslip, and sections of the zone of root hair elongation were photographed through a Zeiss Axiovert 40 CFL microscope equipped with a long-distance 403 Apoplan objective (numerical aperture [N.A.] 1.0) and coupled to a Nikon D50 SLR digital camera. Root and root cell lengths were measured using the Ruler Tool of Photoshop (Adobe).

Flowering Time Assay
Seeds of Ler and Atham1,2,3 mutant genotypes were surface sterilized, germinated on 0.53 Murashige and Skoog salts and 0.6% phytagel (Sigma-Aldrich), and maintained at 18°C with a 16-h daylength for 6 d. Seedlings for which the cotyledons had expanded and the first two postembryonic leaves were clearly visible were transplanted to soil (day 0) and maintained at 20°C with a 16-h daylength under cool-white fluorescent bulbs. Plants were examined daily for an additional 53 d, and the time, measured in days, at which petals emerged from the first flower was recorded for each plant.

Histology
For tissue sectioning in plastic resin, tissue was placed in 2% paraformaldehyde and 2.5% glutaraldehyde in 0.025 M phosphate buffer (sodium phosphate, pH 7.4), vacuum was applied for 30 min, and tissue was fixed overnight at 4°C. Tissue was then rinsed twice with 0.025 M phosphate buffer, postfixed with 1% osmium tetroxide in 0.025 M phosphate buffer for 30 min, and moved through an increasing acetone series (20% increments), each increment lasting a minimum of 1 h and ending with two exchanges of 100% acetone. Tissue was then infiltrated with 812 epoxy resin by sequential transfer through a series of increasing resin concentrations in acetone (1:2, 1:1, and 2:1, 100%) and embedded in 812 resin for 3 d at 70°C in molds. Sections of 900 nm were cut with a diamond knife on a MT6000-XL ultramicrotome (RMC), and individual sections were mounted on glass slides. Slides were placed on a 50°C hot plate for 1 min, and a drop of 1% toluidine blue/1% sodium borate was applied to each section and allowed to stain for 1 min. Slides were then rinsed with distilled water and examined and photographed using brightfield microscopy with a Zeiss Axiovert 40 CFL microscope equipped with a long-distance 403 Apoplan objective (N.A. 1.0) and coupled to a Nikon D50 SLR digital camera.
For scanning electron microscopy, tissue was placed in 1.2% glutaraldehyde in 0.025 M phosphate buffer (sodium phosphate, pH 6.8), vacuum was applied for 10 min, and tissue was fixed overnight at 4°C. Tissue was then rinsed twice with 0.025 M phosphate buffer for 1 h, postfixed with 0.5% osmium tetroxide in 0.025 M phosphate buffer for 24 h at room temperature, and moved through an increasing ethanol series (20% increments), each increment lasting a minimum of 1 h and ending with two exchanges of 100% ethanol. Ethanol was removed by critical point drying with a critical point drier (SAMDRI), and tissue was mounted to stubs with double-sided adhesive tape and sputter coated with gold-palladium alloy using a Hummer Sputtering System (Anatech). Samples were examined with either a Hitachi 4700 or a Hitachi S-570 scanning electron microscope.

Confocal Microscopy
Roots of plants at 5 to 15 d after germination were stained for 2 min with 10 mM propidium iodide and imaged by laser scanning confocal microscopy using a Zeiss LSM 510 microscope. Transverse sections were taken at the middle of the meristematic zone.

Starch Staining
Roots of plants at 5 to 8 d after germination were fixed in ethanol:acetic acid (3:1, v/v) for 3 min, stained with Lugol solution (Sigma) for 1 min, mounted in chloral hydrate, and imaged using a Leica DM 5000B compound microscope.

RT-PCR
Total RNA was extracted from tissues of 12-d-old seedlings with the RNeasy kit (Qiagen). RNA was DNase treated using the Turbo DNA-free kit (Ambion). Total cDNA was then prepared from approximately 1.6 mg of DNA-free total RNA using the SuperScript III kit (Invitrogen) with random hexamers according to the manufacturer's protocol. AtHAM1 cDNA was amplified with the AtHAM1 T-DNA flanking primer set, AtHAM2 cDNA was amplified with the AtHAM2 5# of Ds insertion primer set, and AtHAM3 cDNA was amplified with the AtHAM2 5# of Ds insertion primer set (Supplemental Table S2). AtHAM4 cDNA was amplified with the AtHAM4 T-DNA flanking primer set (Supplemental Table S3). All primer pairs amplify a segment of their respective target cDNAs containing their respective MIR-binding sites and consequently are selective to intact cDNA relative to MIR degradation products.

In Situ Hybridization
In situ hybridizations were performed following a modified protocol of Vielle-Calzada et al. (1999). A detailed protocol is available from the corresponding author upon request.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Expression domains of Arabidopsis HAM orthologs in both shoot meristem and root tissues.
Supplemental Figure S2. Loss-of-function alleles of Arabidopsis homologs of Petunia HAM.
Supplemental Figure S3. Reduction in primary root elongation in Atham1,2,3 mutants results from reduced rates of cell division in the root meristem.
Supplemental Table S1. DELLA and HAM subfamily proteins used for phylogenetic analysis in Figure 1.
Supplemental Table S2. Oligonucleotide primer sequences used for RT-PCR amplification of AtHAM gene fragments.
Supplemental Table S3. Oligonucleotide primer sequences used for PCR genotyping of Atham insertional alleles.
Supplemental Video S1. Arrested inflorescence and flower meristems of an Atham1,2,3 mutant visualized by digital optical microscopy.