The Small Molecule Hyperphyllin Enhances Leaf Formation Rate and Mimics Shoot Meristem Integrity Defects Associated with AMP1 Deficiency

A chemical genetic approach identified the drug hyperphyllin, which phenocopies mutation of AMP1, a member of the M28 carboxypeptidase family with novel plant-specific functions.

Abstract

ALTERED MERISTEM PROGRAM1 (AMP1) is a member of the M28 family of carboxypeptidases with a pivotal role in plant development and stress adaptation. Its most prominent mutant defect is a unique hypertrophic shoot phenotype combining a strongly increased organ formation rate with enhanced meristem size and the formation of ectopic meristem poles. However, so far the role of AMP1 in shoot development could not be assigned to a specific molecular pathway nor is its biochemical function resolved. In this work we evaluated the level of functional conservation between AMP1 and its human homolog HsGCPII, a tumor marker of medical interest. We show that HsGCPII cannot substitute AMP1 in planta and that an HsGCPII-specific inhibitor does not evoke amp1-specific phenotypes. We used a chemical genetic approach to identify the drug hyperphyllin (HP), which specifically mimics the shoot defects of amp1, including plastochron reduction and enlargement and multiplication of the shoot meristem. We assessed the structural requirements of HP activity and excluded that it is a cytokinin analog. HP-treated wild-type plants showed amp1-related tissue-specific changes of various marker genes and a significant transcriptomic overlap with the mutant. HP was ineffective in amp1 and elevated the protein levels of PHAVOLUTA, consistent with the postulated role of AMP1 in miRNA-controlled translation, further supporting an AMP1-related mode of action. Our work suggests that plant and animal members of the M28 family of proteases adopted unrelated functions. With HP we provide a tool to characterize the plant-specific functions of this important class of proteins.

Arabidopsis ALTERED MERISTEM PROGRAM1 (AMP1, At3G54720, MEROPS ID: M28.007) belongs to the Zn²⁺-dependent metalloproteases of the M28B peptidase family (Helliwell et al., 2001). Family members are found in various multicellular organisms and share the following protein motifs: an N-terminal transmembrane domain, a protease-associated domain, and a M28 peptidase motif followed by a transferrin receptor dimerization domain (Davis et al., 2005; Mesters et al., 2006). In strong contrast to the wealth of genetic data positioning AMP1 as a crucial component for proper plant development and hormonal responses, a coherent understanding of its biochemical function(s) is lacking.

The most prominent defect of AMP1 loss-of-function mutants is hypertrophic activity of the shoot apical meristem (SAM). Mutant embryos form a larger SAM with supernumerary cotyledons and development of true leaf primordia starts before germination (Conway and Poethig, 1997; Vidaurre et al., 2007). During the vegetative growth phase the enlarged mutant shoot apex generates leaves at a much higher pace and with altered phyllotaxis (Chaudhury et al., 1993; Nogué et al., 2000a). Moreover, vegetative SAM enlargement and increased organ formation rate correlate and might be at least partially driven by a strong tendency to generate ectopic stem cell niches in the SAM periphery (Huang et al., 2015). Similar SAM-related phenotypes were also observed in mutants of AMP1 orthologs of corn (Zea mays), rice (Oryza sativa), and Lotus japonicus, suggesting that its role in shoot development is conserved among flowering plants (Suzuki et al., 2008; Kawakatsu et al., 2009; Suzaki et al., 2013).

Additional phenotypes of amp1 in seemingly unrelated processes were described including constitutive photomorphogenesis, ecotype-dependent alterations in germination and flowering time (Chaudhury et al., 1993; Lee, 2009; Griffiths et al., 2011), synergid to egg cell conversion in the embryo sac (Kong et al., 2015), suspensor proliferation in the presence of an intact embryo (Vidaurre et al., 2007), increased capacity for somatic embryogenesis (Mordhorst et al., 1998), and elevated abiotic stress resistance (Shi H et al., 2013; Shi Y et al., 2013; Yao et al., 2014).

AMP1 mutant plants exhibit obvious alterations in the biosynthesis of, and response to, plant hormones. However, to explain the range of phenotypes by a defect in one of the classical hormone pathways turned out to be difficult. Cytokinin (CK) biosynthesis has been shown to be up-regulated in amp1 (Chin-Atkins et al., 1996; Nogué et al., 2000b; Saibo et al., 2007; Huang et al., 2015). Whereas the increased CK levels appear to be responsible for de-etiolation in the dark, increased shoot branching, and enhanced tolerance against nitric oxide (Liu et al., 2013), it has been recently shown that they are a consequence rather than a cause of the abnormal SAM phenotypes found in the mutant (Huang et al., 2015). Depending on the subset of phenotypes analyzed, several studies also reported alterations in other hormonal pathways including ethylene, gibberellin, abscisic acid, and auxin (Saibo et al., 2007; Vidaurre et al., 2007; Griffiths et al., 2011; Shi H et al., 2013; Shi Y et al., 2013; Yao et al., 2014).

The unique pleiotropic mutant phenotype might result from a multifunctional role of AMP1 exerting distinct functions in unrelated processes similar to human GCPII (see below), e.g. by the independent use of individual protein domains. However, this assumption is not supported by the phenotypic similarity of the relative high number of characterized amp1 alleles, where separation of individual phenotypes has not yet been described. Thus, it is more likely that AMP1 either acts in a signaling pathway that controls several processes analogous to known plant hormones, or else fulfills a basic cell biological or metabolic house-keeping function important for numerous regulatory pathways. A recent study supports the latter hypothesis, showing that endoplasmic reticulum (ER)-associated translation of miRNA-targeted transcripts is limited by the presence of AMP1 and its paralog LAMP1, pointing toward a general de-regulation of miRNA-regulated pathways in the mutant (Li et al., 2013). However, further specification of AMP1 molecular function by genetic means was so far not successful. Loss-of-function double mutants of the cytochrome P450 genes CYP78A5/KLUH and CYP78A7 as well as gain-of-function alleles of the multidrug and toxic compound extrusion transporter gene ZRIZI bear phenotypic similarities to amp1. However, their genetic interactions and levels of phenotypic overlap have not been described in detail and the biochemical functions of the corresponding gene products are unidentified (Wang et al., 2008; Burko et al., 2011).

Moreover, no attempts were described to reveal AMP1 enzymatic activity by functional comparison with characterized members of the M28B peptidase family. The closest and best-studied AMP1 homolog in animals, human Glu carboxypeptidase II (HsGCPII; 28% amino-acid sequence identity to AMP1), also known as prostate-specific membrane antigen and folate hydrolase 1, exerts different enzymatic activities in a tissue-specific context. In neuronal tissues, HsGCPII cleaves the peptide n-acetyl-l-aspartyl-l-Glu (NAAG) into n-acetyl-l-Asp and l-Glu (Robinson et al., 1987; Klusák et al., 2009). NAAG is the most abundant neuropeptide in the brain and is thought to act as a reservoir for GCPII-dependent release of the neurotransmitter l-Glu. In the brush border of the small intestine the same enzyme also exerts folyl-poly-γ-Glu carboxypeptidase activity, which is thought to mediate dietary folate uptake (Pinto et al., 1996; Navrátil et al., 2014). The in vivo requirement of these biochemical activities could not be clearly demonstrated since knock-out mice of GCPII show normal development and behavior, most likely because of functional redundancy with the close paralog GCPIII (Bacich et al., 2002; Gao et al., 2015). GCPII is also expressed at lower levels in various other tissues and is strongly up-regulated in prostate cancer cells and the neo-vasculature of most solid tumors. However, its physiological function in these tissues and during tumor development is not understood and additional non-proteolytic activities of GCPII are discussed (Bařinka et al., 2012). Despite this lack of knowledge, GCPII has developed into a very important pharmacological target. Small molecule inhibitors of GCPII have been reported to attenuate neuronal damage associated with Glu excitotoxicity, a process linked to various acute and chronic disorders including stroke, Alzheimer’s dementia, and Parkinson’s disease (Bařinka et al., 2012). Additionally, based on its membrane localization and strong up-regulation in prostate cancer and metastatic tissues, GCPII constitutes a key target for prostate cancer diagnosis and therapy (Bařinka et al., 2012). Taken together, a better molecular characterization of members of this protease family is of high relevance for the plant sciences and beyond.

Here we present evidence that AMP1 and its human homolog HsGCPII are functionally divergent. To further elucidate AMP1 molecular function, we undertook a chemical genetic screen and identified the small molecule hyperphyllin (HP), which specifically mimics amp1-related phenotypes, including shoot meristem enlargement, ectopic SCN formation, and plastochron reduction. Consistent with the phenotypic resemblance, HP-treated wild-type plants displayed amp1-related tissue-specific changes of various marker genes and transcriptomic analysis revealed a significant overlap in changes of global gene expression. Amp1 was only marginally responsive to HP treatment in phenotypic and molecular analyses. Moreover, HP elevated the protein levels of the miRNA-regulated HDZIPIII transcription factor PHAVOLUTA (PHV) in wild type, further emphasizing an AMP1-related mechanism of action. HP represents, to our knowledge, the first chemical promoter of leaf formation and constitutes a useful tool to further characterize AMP1’s pivotal role in plant development.

RESULTS

No Evidence for Functional Conservation between AMP1 and HsGCPII

To analyze the level of functional conservation between AMP1 and its best-characterized homolog human Glu carboxypeptidase II (HsGCPII), we determined which of the amino-acid signatures shown to be required for HsGCPII substrate recognition and catalysis are conserved in the Arabidopsis protein. All of the zinc-coordinating residues as well as the catalytic Glu residue for peptide cleavage are present in AMP1, suggesting that it exerts peptidase activity as well as its human counterpart (Fig. 1A; Supplemental Fig. S1A). However, of the residues defining the substrate binding pockets S1 and S1′, only 4 out of 15 (27%) amino acids are conserved in AMP1, with a specifically high variation in the S1 site indicating that substrate-specificities might differ between the two proteins. To further investigate these differences, we expressed different versions of recombinant AMP1 protein to test its catalytic activity against the HsGCPII substrates NAAG and folyl-poly-γ-l-glutamic acids. However, we were not able to produce sufficient amounts of AMP1 protein by using either Escherichia coli or the insect cell expression system, which are routinely used for recombinant GCPII production (Tykvart et al., 2012). As an alternative approach to test for functional conservation between the two proteins, we expressed a MYC-tagged version of HsGCPII under control of the CaMV 35S promoter in amp1-1. We obtained several independent 35S::GCPII-MYC lines, which strongly expressed HsGCPII in amp1-1, but none of them showed phenotypic rescue of any of the known amp1-1 growth defects (Fig. 1, B and C). We also tested the effect of the well-characterized chemical HsGCPII inhibitor PBDA (4,4′-phosphinicobis(butane-1,3-dicarboxylic acid) (Zhou et al., 2005) on seedling development. Germination and growth for 10 d in liquid medium containing 50 μm PBDA did not induce any visible amp1-like shoot defects in wild type (Fig. 1D). Treatment on solid medium containing 100 μm PBDA caused reduced leaf expansion and inhibition of primary root growth (Supplemental Fig. S1B). Again, application of neither 10 μm nor 100 μm PBDA to wild type provoked any hypertrophic shoot phenotypes reminiscent of amp1-1 (Supplemental Fig. S1B). Taken together, we could not find any evidence for the conservation of the biochemical functions of HsGCPII and AMP1.

Figure 1.

Analysis of the degree of functional conservation between Arabidopsis AMP1 and human GCPII. A, List of amino-acid residues defining the indicated functional features in HsGCPII (Mesters et al., 2006) and their corresponding aa residues in AMP1 based on a protein sequence alignment shown in Supplemental Figure S1. Divergent amino-acid residues are indicated in green. B, Shoot phenotypes of Arabidopsis wild type, 35S::GCPII-MYC amp1-1 line 2 and amp1-1 at 14 d after germination (DAG). C, Immunoblotting of protein extracts of 2-week-old seedlings of the indicated 35S::GCPII-MYC amp1-1 lines using α-MYC antibody for GCPII-MYC detection. (Upper panel) Autoradiogram; (lower panel) Ponceau S staining as a loading control. D, Shoot phenotypes of Arabidopsis seedlings of the indicated genotypes grown for 10 d in liquid medium containing either only the solvent DMSO or 50μm PBDA, WT, wild type; pt, the amp1 allele primordia timing. Bars: 5 mm (B); 1 mm (D). AA, Amino acids.

Chemical Genetic Identification of HP, a Small Molecule that Causes SAM Phenotypes Reminiscent of amp1

To identify chemical tools to elucidate the biochemical function of AMP1 we performed a two-tier combinatorial small molecule screen (Supplemental Fig. S2). A quantity of 7800 compounds (Ion Channel Set, ChemBridge; www.chembridge.com) was applied in parallel to wild-type and amp1-1 seedlings and we screened for chemicals, which either (1) rescue the amp1-mutant seedling phenotype or (2) mimic the amp1-phenotype in wild type. Hits of the first category might include compounds with structural similarity to the product(s) of AMP1 catalytic activity, whereas hits of the second category might represent potential inhibitors of the enzyme (Supplemental Fig. S2). Both genotypes were germinated in liquid medium containing the chemicals and phenotypes were scored by visual inspection 10 d after germination. Whereas there was no hit of the first category, we identified one compound that triggered amp1-specific phenotypes in wild-type plants (Fig. 2). Since one main characteristic of amp1 is the enhanced vegetative leaf formation rate, which leads to more than twice the number of leaves compared to wild type at 10 d after germination (Fig. 2, A, I and M), we named the compound “hyperphyllin” (HP). Wild-type seedlings treated with 30 μm HP showed a 41% increase in leaf number compared to the solvent (DMSO) control (Fig. 2, E and M). The enhanced leaf formation rate of HP-treated seedlings was associated with an increased mitotic CYCLIN B1;1 (CYCB1;1)::GUS reporter expression in the shoot meristematic area and in young leaf primordia, similar to amp1-1 (Fig. 2, B, F, and J). HP not only affected leaf formation rate, it also caused a reduced leaf blade width, another feature of amp1-1 (Fig. 2, E and I). A third hallmark of amp1 seedlings is increased vegetative SAM size. Under our growth conditions, amp1-1 SAMs were 3.4 times the size of those of wild type (Fig. 2, C, D, K, L, and N). Application of HP enlarged the meristematic area in the wild-type shoot apex by a factor of 2.3 (Fig. 2, G, H, and N). Treated SAM domes showed an increased height as well as lateral expansion most likely caused by higher cell numbers in the L3 layer, which can also be found in a more extreme manner in amp1-1. Taken together, HP triggers a spectrum of shoot phenotypes typical for amp1.

Figure 2.

HP treatment phenocopies amp1-associated shoot defects. A, E, and I, Comparison of Arabidopsis wild-type (A and E) and amp1-1 (I) seedlings grown for 10 d in liquid medium containing either only the solvent DMSO (A and I) or 30μm HP (E). B, F and J, Comparison of CYCB1;1::GUS activity in Arabidopsis wild-type (B and F) and amp1-1 (J) seedlings grown for 10 d in liquid medium containing either only the solvent DMSO (B and J) or 30μm HP (F). C, G and K, Median longitudinal SAM sections of Arabidopsis wild-type (C and G) and amp1-1 (K) seedlings grown for 10 d in liquid medium containing either only the solvent DMSO (C and K) or 30μm HP (G). Scanning electron micrographs of SAM areas of Arabidopsis wild-type (D and H) and amp1-1 (L) seedlings grown for 10 d in liquid medium containing either only the solvent DMSO (D and L) or 30μm HP (H). M, Quantification of leaf number in mock-treated wild type (WT + DMSO), wild type treated with 30μm HP (WT + HP), and mock-treated amp1-1 (amp1-1 + DMSO) at 10 DAG grown in liquid medium. Values are means ± se (n ≥ 30). N, Quantification of SAM size from median longitudinal sections of mock-treated wild type (WT + DMSO), wild type treated with 30μm HP (WT + HP), and mock-treated amp1-1 (amp1-1 − DMSO) at 10 DAG grown in liquid medium. Values are means ± se (n ≥ 3). Bars: 2 mm (A, E, and I); 1 mm (B, F and J); 50 μm (C, D, G, H, K and L).

Structural Requirements of HP Activity

HP (n-[4-amino-2-chlorophenyl]-2,4-dichlorobenzamide) consists of two substituted benzene rings connected by an amide bond (Fig. 3A). The leaf formation-enhancing activity of HP started at a concentration of 10 μm and peaked at 50 μm with higher concentrations being less effective due to a general toxic effect (Fig. 3B). To assess the minimal structural requirements of HP activity, we compared a number of analogs for their effect on leaf organ formation at a concentration of 50 μm (Fig. 3, A and C). The unsubstituted benzanilide (I2) and 3-chloro-benzanilide (I1) were inactive, whereas 2-chloro-benzanilide (A3) exerted weak activity (Fig. 3C). An additional fluor in ortho-position of ring 2 (2-chloro-6-fluoro-n-phenylbenzamide, A2) resulted in a more active compound, which already showed a measurable effect at 1 μm concentration and a comparable peak activity to HP at 50 μm (Fig. 3, B and C). Compound A1 (4-chloro-n-[2,6-difluorophenyl]benzamide) showed the highest potency in this analysis, which reached already 50% of its maximal activity at 1 μm concentration (Fig. 3, B and C). To better visualize the structure/activity relationship of the tested analogs, we performed a structural clustering analysis using the ChemMine software (Backman et al., 2011). Based on this algorithm, HP is most closely related to A1 and the two other active compounds A2 and A3 build a second distinct branch (Fig. 3C). All four active compounds have halogen substituents in ring 2 in ortho- and/or paraposition. If ring 2 is either unsubstituted (I2), halogenated in meta (I1), or the halogen in ortho is exchanged to an amino group (I3), activity is lost. In contrast, there is considerable structural freedom at ring 1.

Figure 3.

Structure-activity relationship of HP and its analogs. A, Chemical structures of HP and its active (A1, A2, A3) and inactive (I1, I2, I3) analogs. B, Effects of HP and analogs A1 and A2 on leaf number in Arabidopsis wild-type seedlings grown for 10 d in liquid medium containing the indicated concentrations of the compounds. Values are means ± se (n ≥ 10). C, Correlation of structure-based hierarchical clustering of HP and its analogs with their respective effects on the number of leaves formed at a concentration of 50 μm. Values are means ± se (n ≥ 30). Hierarchical clustering was generated by the ChemMine clustering tool (see “Materials and Methods”).

Activities of SAM Markers Are Highly Similar in amp1 and HP-Treated Wild-Type Plants

To test whether the observed morphological similarities between HP-treated wild-type plants and amp1 also correlate with changes in underlying molecular processes, we analyzed the expression of marker genes known to be specifically altered in the mutant. The SAM boundary marker KLUH/CYP78A5, which controls leaf formation rate, is massively up-regulated in amp1-1 (Zondlo and Irish, 1999; Helliwell et al., 2001; Wang et al., 2008). To assess whether KLUH/CYP78A5 expression is also affected by HP, we used the pKLU::GUS reporter (Anastasiou et al., 2007). In amp1-1, pKLU::GUS activity was not only found in a much broader domain depicting the enlarged SAM area, the staining also appeared to be more intense compared to the mock-treated wild-type control (Fig. 4A). In HP-treated wild-type plants, the expression of the reporter was also clearly expanded and stronger compared to the mock control, but not to the same extent as in amp1-1 (Fig. 4A). Another marker with an altered expression pattern in amp1 is the organizing center (OC) defining transcription factor WUSCHEL (WUS; Huang et al., 2015). The zone of WUS::GUS activity was substantially broadened in both amp1-1 and in drug-treated wild-type SAMs compared to the mock-treated wild-type control (Fig. 4B). Moreover, the previously reported amp1-specific formation of ectopic OCs in the SAM periphery (Huang et al., 2015) was phenocopied in wild type by HP treatment (Fig. 4, C and D). Under the growth conditions used, around 20% of amp1 plants showed more than one WUS::GUS expression domain in the primary SAM. Notably, this characteristic behavior was also observed in 7% of HP-treated plants, but absent in the mock-control (Fig. 4C). The expanded and partially ectopic expression of the OC marker WUS in amp1 and HP-exposed wild type was accompanied with the formation of enlarged and partially fragmented CLAVATA3 (CLV3)-expressing stem cell pools in the shoot apex (Fig. 4E). Furthermore, HP triggered ectopic CLV3::GUS activity in vascular tissues of petioles, hypocotyl, and roots, another specific characteristic of amp1 (Fig. 4E).

Figure 4.

SAM marker activities in Arabidopsis HP-treated wild-type plants are highly reminiscent of those of amp1. Comparison of SAM marker activities in Arabidopsis mock-treated wild type (WT + DMSO), wild type treated with 30μm HP (WT + HP), and mock-treated amp1-1 (amp1 + DMSO) at 10 DAG grown in liquid medium. A, pKLU::GUS activity. B, WUS::GUS activity. C, Percentage of plants showing ectopic WUS expression foci in the SAM (n ≥ 100). D, 30μm HP-treated wild-type and mock-treated amp1-1 seedling with ectopic WUS expression foci in the SAM. E, CLV3::GUS activity in the SAM (upper panel), in the shoot (second panel), and root vascular tissues (third panel). Inset in upper panel shows HP-treated wild-type SAM with an ectopic CLV3::GUS-expressing stem cell pool. Bars: 500 μm (A); 200 μm (B and D); 500 μm (E, upper and second panel); and 200 μm (E, third panel).

HP Mediates CK-Independent Effects

Plants mutated in amp1 show elevated CK biosynthesis, which results in enhanced CK-responsive reporter activity particularly in vascular-associated tissues (Chaudhury et al., 1993; Nogué et al., 2000b; Saibo et al., 2007; Huang et al., 2015). However, this defect in CK biosynthesis appears to be a consequence rather than a cause of the shoot phenotypes of amp1 (Huang et al., 2015). To test whether HP affects CK homeostasis, we used the CK-responsive ARABIDOPSIS RESPONSE REGULATOR5 (ARR5)::GUS reporter as a read-out (Fig. 5A). Almost identical to amp1, we observed augmented activity of ARR5::GUS in the SAM, hypocotyl, and petioles of HP-treated seedlings. This raised the question whether HP exerts CK activity on its own. This scenario seemed rather unlikely, since HP has no obvious structural similarity to any of the known CKs. To further exclude this possibility, we compared the effect of exogenously applied transzeatin and HP on leaf formation in a dose-response experiment. In contrast to HP, we did not observe any leaf number promoting effect of transzeatin under our growth conditions in a concentration range from 1 μm to 75 μm (Fig. 5, B and C). Correspondingly, transcriptomic analysis of HP-treated plants (see below) revealed only a marginal overlap, with the published transcriptomic responses of CK-treated plants strongly supporting a model in which HP does not directly exert CK activity.

Figure 5.

HP is not a CK analog. A, ARR5::GUS activity in Arabidopsis mock-treated wild type (WT + DMSO), wild type treated with 30 μm HP (WT + HP), and mock-treated amp1-1 (amp1 + DMSO) at 10 DAG grown in liquid medium. B, Shoot phenotypes of mock-treated (WT + DMSO), 30μm HP-treated (WT + HP), and 25μm transzeatin-treated (WT + zeatin) wild-type plants at 10 DAG grown in liquid medium. C, Effects of HP and transzeatin on leaf number in wild-type seedlings grown for 10 d in liquid medium containing the indicated concentrations of the compounds. Values are means ± se (n ≥ 30). Bars = 1 mm.

HP-Treated Wild-Type and amp1 Plants Show Significantly Overlapping Transcriptomic Alterations

If the observed amp1-mimicking effect of HP is based on molecular interference with the AMP1 pathway, one would expect overlapping transcriptomic aberrations in the mutant and in wild type after compound treatment. To test this assumption, we performed a genome-wide transcript analysis using the Arabidopsis ATH1 Genome Array (Affymetrix; www.affymetrix.com). Wild-type seedlings were germinated and grown for 10 d in liquid medium in the presence of either DMSO or 30 μm HP, whereas amp1-13 seedlings were grown under the same conditions in medium containing only DMSO. HP-treatment of wild type altered the expression of 1526 genes (978 induced/548 repressed) whereas 953 genes (645 induced/308 repressed) were mis-expressed in amp1-13 (Fig. 6A). Remarkably, between these two groups there was an overlap of 608 genes (40% of the HP altered genes and 64% of amp1-13 altered genes). Moreover, only three out of these 608 genes showed opposite misexpression between the two samples.

Figure 6.

HP-treated Arabidopsis wild-type and amp1 plants have significantly overlapping transcriptional responses. A, Venn diagram showing the number of overlapping and non-overlapping genes that exhibited altered expression in amp1-13 (blue) and in wild type in response to HP treatment (green) and percentage of total misregulated genes from the two samples falling in the overlapping group. B, Diagram displaying the fraction of hormone-specific marker genes misregulated in HP-treated wild type and in amp1-13. ACC, ethylene (1-amino-cyclopropane-1-carboxylic acid); CK, cytokinin (transzeatin); BL, brassinosteroid (brassinolide); IAA, auxin (indole-3-acetic acid); ABA, abscisic acid; MJ, jasmonic acid (methyl jasmonate).

To assess whether the transcriptional changes in the two samples show specific hormonal signatures, we determined the frequency of marker genes for six different hormones (Nemhauser et al., 2006). Surprisingly, in both samples less than 2% of CK-responsive marker genes were present, showing a similar low overlap as with ethylene and brassinosteroid-dependent marker genes (Fig. 6B). In contrast and in accordance with recent reports (Shi H et al., 2013; Shi Y et al., 2013), there was a relatively high frequency of misregulation of ABA-specific marker genes in amp1-13, which we also found in HP-treated wild type. Notably, the strongest overlap was observed with genes specifically induced by methyl jasmonate (Fig. 6B; 14% present in the HP-treated wild-type sample/12% present in the amp1-13 sample).

To identify specific biological and molecular processes commonly affected, we performed an enrichment analysis of gene ontology (GO) terms with the group of 608 overlapping genes using the ReviGO software (Supek et al., 2011). We found a significant enrichment of genes associated with stress responses triggered by xenobiotic compounds (organic substance, chemical stimulus) and biotic interactions (defense, chitin, wounding, jasmonic acid, salicylic acid, immune response; Supplemental Fig. S3). More-specific processes such as lipid localization and transport, photosynthesis, cell wall organization, and trehalose metabolism, also emerged as significantly affected in this analysis.

HP Responses Are Substantially Alleviated in the amp1 Mutant Background

To further analyze the specificity of HP for the AMP1-regulatory pathway, we tested the effect of HP on the leaf formation rate of amp1-1 and amp1-13. Both alleles did not show an HP-induced rise of leaf number (Fig. 7A). In contrast, clavata1-1 (clv1-1) and clavata3-2 (clv3-2) displayed a significantly elevated leaf count in the presence of the drug, indicating that HP is generally effective in increased meristem-size mutants (Fig. 7A). We also tested the effect of HP on a mutant in the AMP1 paralog LAMP1. In this assay, lamp1-2 showed an already slightly higher leaf number under control conditions compared to wild type but reached similar organ numbers in the presence of HP (Fig. 7A). Thus, whereas amp1 is insensitive to the promoting effect of HP on leaf number, lamp1 is only barely resistant. We also did not observe a gain in amp1-1 SAM size after HP treatment, but instead a slight but insignificant negative impact similar to the leaf number analysis in the mutant (Fig. 7, B and C).

Figure 7.

HP responses are substantially alleviated in the amp1 mutant background. A, Quantification of leaf number in the indicated genotypes at 10 DAG grown in liquid medium containing either only the solvent DMSO or 30μm HP. Values represent means ± se (n ≥ 15) Statistical significance between treated and untreated samples of the same genotype were calculated with the Student’s 2-tailed t-test; the resulting p-values are shown. B, Scanning electron micrographs of SAM areas of amp1-1 seedlings grown for 10 d in liquid medium containing either only the solvent DMSO or 30μm HP. C, Quantification of SAM area from scanning electron micrographs of mock-treated (DMSO) 30μm HP-treated amp1-1 seedlings (HP). Values are means ± se (n ≥ 5). D, Comparison of GUS activities of indicated reporter lines in mock-treated (amp1-1 + DMSO) and 30μm HP-treated (amp1-1 + HP) amp1-1 seedlings at 10 DAG grown in liquid medium. (Right) WUS::GUS panel shows plants with multiple OCs including the frequency of appearance (n ≥ 20). E, Venn diagram showing the number of overlapping and nonoverlapping genes that were HP-regulated in wild type (blue), constitutively misregulated in amp1-13 (green), and HP-regulated in amp1-13 (red) based on the transcriptomic analysis using Arabidopsis Gene 1.1 ST Array Strips. Bars = 50 μm (B).

Next, we monitored how AMP1-specific marker genes, whose expression responded to HP in wild type, react in amp1-1 in the presence of the drug. The enhanced and ectopic activities of pKLU::GUS, ARR5::GUS, and CLV3::GUS in amp1-1 were subtly further intensified by HP application; however, the level of change was far weaker than that observed in wild type (Fig. 7D). The increased WUS::GUS activity in amp1-1 primary stem cell pools and the OC amplification rate was unaffected by HP (Fig. 7D).

To further dissect to which extent HP-induced molecular responses are dependent on AMP1 function, we repeated the transcriptomic analysis expanded for a set of HP-treated amp1 samples using the new Ara Gene 1.1 ST Array Strips (Supplemental Table S3). In this experiment, we again detected a substantial overlap between the group of HP-regulated genes in wild type (142/496: 29%) with the set of misregulated genes in amp1-13 (142/302: 47%; Fig. 7E). In comparison to ATH1 chip analysis (Fig. 6A), the overall number of genes with altered expression in the different samples was considerably smaller (approximately only 30%), consistent with previously observed differences in signal strength between the two microarray platforms (Kakei and Shimada, 2015). Notably, in contrast to wild type, amp1 to amp13 reacted only marginally to HP application with only 60 genes affected in total. Of these, 27% (16 genes) were also present in the group of genes regulated by HP in wild type but not changed in amp1-13 under control conditions. Thus, in the absence of AMP1, HP provokes only very limited transcriptional responses, concordant with the remarkably mild phenotypic drug effects on the mutant.

HP Elevates Protein Levels of a miRNA-Regulated Member of the HDZIPIII Family

Because it has recently been shown that protein levels of miRNA-regulated genes, like transcription factors of the HD-ZIPIII family, are up-regulated in amp1 due to enhanced translation by ER-localized ribosomes (Li et al., 2013), we also examined the effect of 30 μm HP or 30 μm A1 on PHV-MYC protein levels in a 35S::PHV-MYC line. Both compounds increased PHV-MYC levels as found in untreated amp1-1 plants supporting an AMP1-dependent mode of function of these chemicals (Fig. 8A). This effect was also observed for YFP-tagged PHV in a 35S::PHV-YFP line, although to a lesser extent (Fig. 8A). The transcript levels of each of the transgenes were only slightly different between control and HP-treated samples, suggesting that the observed changes at the protein level derived at least partially from a post-transcriptional effect of the compound. (Fig. 8B).

Figure 8.

HP elevates PHV protein levels in wild type. A, Immunoblotting of protein extracts of 10-d-old 35S::PHV-MYC and 35S::PHV-YFP seedlings in the indicated genetic backgrounds grown in liquid medium containing only the solvent DMSO (−), 30μm hyperphyllin (HP), or 30-μm A1 (A1). Upper panel autoradiogram: PHV-MYC detection using an anti-MYC antibody. Lower panel autoradiogram: PHV-YFP detection using an anti-GFP antibody. Coomassie Blue staining of the membrane is shown as a loading control. Normalized relative signal intensities are indicated. B, qPCR analysis of PHV expression in 10-d-old seedlings of the indicated lines. The SE was calculated from three biological replicates. UBC was used as an internal control. CBS: loading control.

HP Shows Efficient Uptake Characteristics and High Stability in Planta

Cellular uptake and intracellular stability are important parameters affecting compound activity. We followed the accumulation of HP in seedlings grown in liquid medium over time. HP was rapidly and efficiently incorporated in wild type, and intracellular concentrations reached a plateau after 24 h (Supplemental Fig. S5). The in planta levels exceeded those of the medium by severalfold. HP appeared to be relatively stable and not metabolized once taken up by the plant, since concentrations stayed close to peak levels even after 50 h of incubation. Finally, we could not find any differences between wild type and amp1-13 in this assay, suggesting that the unresponsiveness of amp1 to HP is not due to altered uptake or fate of the compound in the mutant.

DISCUSSION

Based on our protein sequence comparison, complementation approach, and pharmacological analysis, we conclude that AMP1 might exhibit peptidic hydrolase activity as its human homolog HsGCPII, but that the substrate specificity or other nonenzymatic key functions are most likely different between both proteins. Although we cannot fully exclude that the unsuccessful complementation of amp1 by HsGCPII was due to incorrect processing or subcellular targeting of the human protein in a plant cell environment, this does not seem to be a general issue (Hosein et al., 2010). Functional divergence between AMP1 and HsGCPII is not too surprising: (1) both proteins show a low overall sequence identity of only 28%; and (2) to our knowledge, no physiological role was ascribed to NAAG (spaglumic acid) in plants, and NAAG treatment of wild-type and amp1 plants did not have any obvious phenotypic effects (Helliwell et al., 2001). (3) Plants synthesize folates de novo, and the majority of the in planta folate pool is poly-glutamylated. A family of γ-glutamyl-hydrolases was characterized in tomato (Solanum lycopersicum) and Arabidopsis, and they have been shown to regulate cellular folate content and folate sequestration in the vacuole (Orsomando et al., 2005). Except for lower seed set, no obvious phenotypes were described in gain- and loss-of-function alleles of corresponding genes, indicating that general alterations in the homeostasis of folate poly-glutamates are not directly linked to altered shoot meristem activity (Akhtar et al., 2010). However, this does not exclude that AMP1 might have an ER-specific role in folate-deglutamylation. Nevertheless, our data support a model, in which plant and animal GCPIIs adopted different functions during evolution, and that the plant-specific biochemical roles of this enzyme class have yet to be identified.

HP treatment of wild-type plants imitates the amp1-specific spectrum of phenotypes at the morphological, cellular, and molecular levels. Based on this unique combination of phenotypes and the observed resistance of amp1 to the chemical, we assume that HP either acts directly in the AMP1 regulatory pathway or affects a parallel pathway that converges on a common molecular target. Notably, the strength of HP-induced phenotypes, particularly on SAM size, leaf formation rate, and frequency of ectopic stem cell pool formation, is significantly weaker compared to amp1. However, this might be caused by the postembryonic application of the chemical. AMP1 mutant plants show massive SAM enlargement and presence of true leaf primordia in the mature embryo, giving them a head-start in developing SAM defects in the seedling stage (Conway and Poethig, 1997; Vidaurre et al., 2007). Long-term treatment of mother plants during flower and silique development could be used to verify this assumption. We further found that HP and its analogs only induce obvious amp1-related, mainly shoot-specific phenotypes when applied in liquid growth conditions, where shoot tissues get in direct contact with the chemical. Most likely, HP has to be present in the shoot to induce the respective phenotypes and is potentially not sufficiently transported from root to shoot tissues. This is supported by the absence of weakly acidic groups in the hitherto characterized active HP analogs, which have been reported to facilitate phloem mobility to allow systemic transport in the plant (Xuan et al., 2013). Testing alternative application methods combined with future studies of transport characteristics of known and novel HP derivatives might help to overcome these constraints.

In this context it will be also important to further resolve the functional relationship among HP, AMP1, and AMP1’s paralog, LAMP1. Genetic analysis revealed that AMP1 and LAMP1 have partially redundant functions in shoot development and the miRNA-dependent control of translation rate (Li et al., 2013; Huang et al., 2015), with AMP1 playing a dominant role most likely due to its spatially and temporally broader expression pattern compared to LAMP1. At the morphological level and also in respect to global transcriptional responses, amp1 was almost insensitive against HP, giving the implication that LAMP1 function is not overly compromised by the compound. Surprisingly, we did not observe any obvious increased susceptibility of lamp1 to the organ-promoting properties of the drug. Whether this ostensibly controversial response also results from the above-mentioned restricted postembryonic application, or tissue-specific functional limitations of the drug compared to constitutive elimination of AMP1 function by mutation, cannot be answered at the moment.

We found a remarkable overlap in transcriptomic changes between HP-treated wild-type samples and amp1. Although increased CK levels have been postulated as a hallmark of amp1, we only found a minor fraction of CK marker genes to be misregulated in amp1 and the HP-treated wild-type sample. This finding is consistent with amp1-related microarray data from previous studies using different developmental stages and mutant alleles. In amp1 developing seeds as well as 14-d-old plants, the fraction of CK-regulated factors in the pool of mis-expressed genes was only marginal (Griffiths et al., 2011; Shi Y et al., 2013). This is also in line with recent evidence that the majority of shoot meristem defects in amp1 are also present in a CK-insensitive background (Huang et al., 2015). One explanation for this discrepancy regarding the elevated CK levels detected in the mutant could be that increased CK abundance in amp1 is restricted to specific tissues. In support of this assumption, the ARR5::GUS reporter showed up-regulation mainly in vascular-associated tissues of amp1. This might disguise global transcriptional responses at the whole plant level. Alternatively, general CK responses might be suppressed in an AMP1-deficient background. Importantly, HP-induced transcriptional responses show a similarly low impact on the CK pathway concordant with our hypothesis that the compound affects plant development in an AMP1-related manner.

Our transcriptomic analysis revealed that HP application and AMP1 deficiency alter the expression of a substantial overlapping set of ABA-responsive genes. Although these changes are consistent with recent studies where amp1 displayed elevated ABA responses and was more tolerant to ABA-related abiotic stresses (Shi H et al., 2013; Shi Y et al., 2013; Yao et al., 2014), we do not yet understand the relevance of these alterations on the observed SAM phenotypes and their meaning in terms of the potential molecular functions of AMP1 and HP. Nevertheless, it will be interesting to test whether HP treatment confers tolerance to abiotic stresses.

Another outcome of our genome-wide expression comparison is the obvious coinciding effects of missing AMP1 function and HP treatment on lipid-related processes including alterations in jasmonate responses. Further studies have to be done on the physiological cause and functional relevance of these transcriptional changes to clarify whether this information can be used to better position AMP1 and/or HP activity at the molecular level.

Thinking about HP’s mode of action, one plausible scenario is that HP directly targets the enzymatic activity of AMP1, acting as a competitive or noncompetitive inhibitor. However, so far we were unable to produce evidence for physical interaction between HP and AMP1, mainly because of the difficulties involved in generating recombinant AMP1 in sufficient quantity and quality to perform binding studies. HP does not contain residues found in known metalloprotease inhibitors such as zinc-chelating groups resistant to hydrolysis, including hydroxamate, sulphonate phosphonate, phosphinate, phosporamidate, or urea (Zhou et al., 2005; Pavlicek et al., 2014). Based on its small size, HP might rather bind to a subdomain of the substrate-binding pocket of AMP1 similar to the n-phenylbenzamide GW9662, which inhibits the binding of fatty acid ligands to peroxisome proliferator-activated receptor gamma (Chandra et al., 2008).

Thus, a future key goal will be the development of experimental strategies to assess the level of physical HP-AMP1 interaction. Testing different versions of closely related orthologs of AMP1 in different expression systems might allow the production of sufficient recombinant protein for binding studies and protease activity assays, and for gathering structural information about AMP1. This step will also be critical to further compare the functional relationship among AMP1, LAMP1 and their animal homologs in the M28 family of metalloproteases, such as in testing whether the recently identified arene-binding site of GCPII is conserved in AMP1; and if yes, whether it plays a role in HP-AMP1 interaction (Zhang et al., 2010). In case it turns out that HP directly targets AMP1, it will be a valuable tool to further establish the biochemical function of the protein. It could also be used to further dissect the diverse roles of AMP1 in plant development by manipulating AMP1 activity in a spatially and temporally controlled manner. Moreover, it will be interesting to test to what extent HP is able to modulate other AMP1-dependent processes including egg cell fate determination, somatic embryogenesis, or ABA-mediated abiotic stress resistance (Mordhorst et al., 1998; Shi H et al., 2013; Shi Y et al., 2013; Kong et al., 2015).

In case HP does not act on AMP1 directly, it may serve as a useful agent in search for additional components of the AMP1-regulatory pathway. To our knowledge, HP is the first small molecule compound, which accelerates leaf formation rate, a process not well understood at the molecular level (Lee and Jackson, 2009). It should thus also help to interrogate and specify the communication events between established organ primordia and incipient leaf anlagen regulating plastochron length.

MATERIALS AND METHODS

Plant Material and Growth Conditions

Arabidopsis L. (Heynh) ecotypes Columbia (Col-0) and Landsberg erecta (Ler) were used in this study. Amp1-1 (Col-0; N8324), amp1-13 (Col-0; N522988), pt (Ler; N235), lamp1-2 (Col-0; SM_3.22750), clv1-1 (Ler; N45), and clv3-2 (Ler; N8066) were ordered from the European Arabidopsis Stock Centre (NASC, Nottingham Arabidopsis Stock Centre; http://www.arabidopsis.info/). Origin of transgenic lines: CYCB1;1::GUS (in Col-0) was provided by John Celenza (DiDonato et al., 2004); CLV3::GUS and WUS::GUS (in Ler) were received from Thomas Laux (Gross-Hardt et al., 2002); pKLU::GUS (in Ler) was provided by Michael Lenhard (Anastasiou et al., 2007); and ARR5::GUS (in Col-0) was provided by Joe Kieber (D’Agostino et al., 2000). Combinations of mutants and reporter lines were obtained by crossing individual lines and genotypes were verified phenotypically and by PCR genotyping. Homozygosity of reporters was determined by unanimous GUS signal presence (n ≥ 30).

Seeds were surface-sterilized with 70% ethanol + 0.05% SDS for 3 min, rinsed with 96% ethanol for 1 min, and subsequently dried in a laminar flow hood. For the small-molecule screen, GUS-reporter assays, the microarray experiment, leaf formation rate, and shoot apical-meristem analyses, the sterile seeds were germinated and grown in 96-well plates (approximately 5 seeds per well; Nunclon Delta Surface, cat. no. 163320; Thermo Fisher Scientific/Nunc, Rochester, NY) in 100μl liquid half-strength Murashige and Skoog medium (MS; Duchefa Biochemie, Haarlem, Netherlands) containing 1% Suc. After stratification at 4°C in darkness for 2 d, seeds were placed in a growth chamber (BrightBoy GroBank, cat. no. BB-XXL; Clf Plant Climatics, Wertingen, Germany) and grown at 26°C grown under long d conditions (16 h, 80 μm m⁻² s⁻¹ cool-white light/8 h dark). For the phenotypic analysis of 35S::GCPII:MYC lines, plants were grown on soil in a growth chamber under long d conditions (16 h 120 μm m⁻² s⁻¹ cool-white light/8 h dark at 18°C to 22°C). Trans-zeatin and PBDA were ordered from Sigma-Aldrich (St. Louis, MO).

Gene Constructs and Transformation

PCR was performed with proofreading thermostable polymerase, and all clones were confirmed by sequencing. To create 35S::GCPII-MYC, the human GCPII ORF was amplified by PCR with primers PSMAF-EcoRV and PSMAR-BamHI and subcloned into pGEM-T Easy (Promega, Madison, WI). The fragment was excised via EcoRV and BamHI and ligated into pGWR8 (Rozhon et al., 2010), resulting in p35S::GCPII. A 6xMYC-tag (from pGWR8-MYC) was inserted into the NotI site of p35S::GCPII to create p35S::GCPII-MYC. To create 35S::PHV-MYC and 35S::PHV-YFP, the PHV ORF was amplified from cDNA (Col-0 seedlings) by PCR with primer transcription factors PHAVOLUTA (PHV) ORF F (EcoRV) and PHV ORF R (NotI). The fragment was subcloned into pGEM-T Easy (Promega). Subsequently, the PHV ORF was transferred via EcoRV and NotI into the pGWR8 backbone to generate p35S::PHV. The YFP ORF (from pGWR8-YFP) and 6xMYC-tag (from pGWR8-MYC) sequences were subcloned into the NotI site of p35S:PHV to create p35S::PHV-YFP and p35S::PHV-MYC, respectively.

Using the floral dip method, amp1-1^−/− plants were transformed with p35S::GCPII:MYC, and amp1-1^−/+ plants were transformed with p35S::PHV:YFP and p35S::PHV:MYC, respectively. At least 10 independent transformants were generated for each line and the resulting T2 lines were confirmed for single-transgene insertion sites based on the 3:1 segregation of the selection marker and propagated for further analysis. For 35S::PHV:YFP and 35S::PHV:MYC, segregating wild-type and amp1-1 plants were isolated, which were isogenic and homozygous for the transgene.

Small Molecule Screen

For the primary screen, 7800 chemicals of the Ion Channel Set (ChemBridge; http://www.chembridge.com/index.php) were tested under the growth conditions described above at a final concentration of 50 μm in DMSO [1% (v/v)]. Five seeds per well were germinated and grown in the presence of the compounds until d 10, when visual inspection of leaf number and shoot apical-meristem size was done with a stereomicroscope (model no. SZX10; Olympus, Melville, NY). Hyperphillin (HP) and structural analogs were re-ordered for HP (5839245), A1 (7679986), A2 (5578458), and I3 (6955571), from ChemBridge; and A3 (S571202), I1 (S883042), and I2 (442470), from Sigma-Aldrich. To correlate structural features of compounds with their activities, the ChemMine Clustering and Data Mining Web Tools were used (http://chemmine.ucr.edu/).

Leaf Number Analysis

The number of visible leaves was recorded after examination of the shoot apex area under the stereomicroscope (2× magnification) 10 d after germination. In each experiment, the mean number of emerged leaves from at least 10 plants was calculated.

GUS-Staining

Plants were submerged in GUS staining buffer (Jefferson et al., 1987) containing 1 mm 5-bromo-4-chloro-3-indolyl-β-d-GlcA (Duchefa Biochemie), 100 mm sodium P (pH 7.0), 0.5 mm potassium ferricyanide, 0.5 mm potassium ferrocyanide, 10 mm EDTA, and 0.1% (v/v) Triton X-100. Seedlings were incubated at 37°C for individual periods of time depending on the reporter strength. After staining, the samples were dehydrated in 70% ethanol. Seedlings were analyzed with a stereomicroscope (model no. SZX10; Olympus) equipped with a digital camera (model no. DP26; Olympus).

Histology

For histological analysis, 10 d-old seedlings were fixed overnight at 4°C in FAA [10% (v/v) formaldehyde, 5% (v/v) acetic acid, and 50% (v/v) ethanol]. After fixation, samples were dehydrated in an ethanol series and embedded in Technovit 7100 (Heraus Kulzer, South Bend, IN), according to the manufacturer’s instructions, as described in De Smet et al. (2004). A series of 5–7-µm-thick longitudinal sections was made with a rotary microtome (Reichert-Jung/Reichert Technologies, Depew, NY). Sections were transferred to microscopic slides (Marienfeld-Superior/Paul Marienfeld, Königshofen, Germany), stained with Ruthenium Red (Sigma-Aldrich), and photographed with a digital camera (model no. PM 20; Olympus) mounted on an Axiophot microscope (Carl Zeiss, Oberkochen, Germany).

Scanning Electron Microscopy

Seedlings were incubated in fixative (50% ethanol, 10% acetic acid, 5% formaldehyde) overnight at 4°C and then dehydrated through a graded ethanol series up to 98% ethanol and supercritical point-dried using an electron-multiplying CPD300 (Leica Microsystems, Wetzlar, Germany). Dried seedlings were dissected and mounted on conductive adhesive tabs (PLANO, Wetzlar, Germany) under a stereomicroscope (cat. no. SZX10; Olympus). Samples were subsequently examined using a T-3000 tabletop scanning electron microscope (Hitachi, Tokyo, Japan).

Transcriptomic Analysis Using Arabidopsis ATH1 Genome Arrays

The experiment was conducted with 10-d-old seedlings of the indicated genotypes incubated with either DMSO [1% (v/v)] or 30 μm HP under the growth conditions described above. Total RNA was extracted from three biological replicates for each treatment using the RNeasy Kit (Qiagen, Hilden, Germany). RNA quality control, cDNA synthesis, labeling, and hybridization on a GeneChip Arabidopsis ATH1 genome array (Affymetrix, www.affymetrix.com) were conducted at the Nottingham Arabidopsis Stock Centre according to standard Affymetrix protocols. The raw microarray data were preprocessed and normalized using the robust multiarray average method (Irizarry et al., 2003); the data statistical significance was assessed by using the moderated t-test (Smyth, 2004); and the p-value was adjusted with a Benjamini and Hochberg false discovery rate ≤ 0.05 (Benjamini and Hochberg, 1995). Genes with adjusted p-values less than 0.05 were filtered with regard to their differential expression (2-fold change, 95% confidence). The gene ontology terms analysis was done using AgriGO and ReviGO web-based tools (Supek et al., 2011). The microarray data were deposited in NCBI’s Gene Expression Omnibus (National Center for Biotechnology Information; http://www.ncbi.nlm.nih.gov/).

Transcriptomic Analysis Using Arabidopsis Gene Array Strips

The experiment was conducted with 10-d-old seedlings of the indicated genotypes incubated with either DMSO [1% (v/v)] or 30 μm HP under the growth conditions described above. Total RNA was extracted from three biological replicates for each treatment using the E.Z.N.A. Plant RNA Kit (OMEGA Bio-Tek, Norcross, GA). The sample concentration and purity of RNA were determined by spectrophotometry, and the integrity was confirmed using a 2100 Bioanalyzer with the RNA 6000 Nano Kit (Agilent Technologies, Santa Clara, CA). Whole-genome transcriptome analysis was conducted by hybridizing 12 biological samples of total RNA per treatment to Arabidopsis Gene 1.1 ST Array Strips (Affymetrix). All steps were conducted at the Nottingham Arabidopsis Stock Centre. Gene expression data were analyzed using the software Partek Genomics Suite (v. 6.6; Partek, Chesterfield, MO). The raw CEL files were normalized using the robust multiarray average background correction with quantile normalization, log base 2 transformation, and mean probe-set summarization with adjustment for GC content. Differentially expressed genes were identified by a two-way ANOVA. Differentially expressed genes were considered significant if the p-value was ≤ 0.05 at a fold-change of >2 or <−2. The microarray data were deposited in NCBI’s Gene Expression Omnibus.

Quantitative Real-Time RT-PCR

For qPCR, about 50 mg of seedlings was collected and shock-frozen in liquid nitrogen. After total RNA extraction with the E.N.Z.A. Plant RNA Mini Kit (OMEGA Bio-Tek), DNase I (Thermo Fisher Scientific, Waltham, MA) treatment, and cDNA synthesis with the RevertAid first-strand cDNA synthesis kit (Thermo Fisher Scientific), qPCR was performed with a Eppendorf Realplex2 Mastercycler (Eppendorf, Hamburg, Germany) using SensiFAST SYBR Lo-ROX Mix, 2× (Bioline USA, Taunton, MA) and specific primers for the mRNAs of interest. Data were normalized to UBC (AT5G25760) and measured in at least three technical replicates.

Protein Preparation and Immunoblotting

Plant material (50 mg) was flash-frozen in liquid nitrogen and homogenized with a Retsch mill (Verder Scientific, Newtown, PA). A quantity of 200 μl extraction buffer (62.5 mm TRIS pH 6.8, 125 mm DTT, 2.5% SDS, 12.5% glycerol, 0.01% bromophenol blue) was added and samples were incubated at 95°C for 2 min. The samples were centrifuged at 14,000g for 5 min and 10 μl of the supernatants separated by SDS-PAGE (10% gel) and semi-dry-blotted onto a polyvinylidene difluoride membrane (Millipore, Billerica, MA). The membrane was blocked with blocking buffer (5% skim milk powder dissolved in 0.05% Tween 20, 150 mm NaCl, and 10 mm TRIS/HCl, pH 8.0). For PHV-YFP detection, the membrane was probed with a mouse anti-GFP-horseradish peroxidase antibody (1:5000; Miltenyi Biotec, Bergisch Gladbach, Germany) and signals were detected using the ECL Select Detection Reagent (GE Healthcare, Marlborough, MA). For PHV-MYC detection, membranes were probed with a mouse antic-Myc antibody (1:5000; Santa Cruz Biotechnology, Santa Cruz, CA). Alkaline phosphatase-conjugated goat anti-mouse IgG (Sigma-Aldrich) diluted 1:5000 with blocking buffer was employed as a secondary antibody. For detection, the CDP-Star detection reagent (GE Healthcare) was used.

HP Uptake Assay

Whole seedlings were grown for 7 d on agar medium (half-strength Murashige and Skoog) and transferred to liquid medium (approximately 50–70 mg plant material/ml half-strength Murashige and Skoog medium). At certain time points, HP was added to a final concentration of 50 μm. After the treatment, plants were frozen in liquid nitrogen. For analysis, the plant material was ground to a fine powder and 1 ml of 40% ACN (acetonitrile) was added to each sample. The mixtures were spiked with 100 μl of a 100-μm 2-chloro-benzanilide (A3) stock solution as an internal standard. Extraction was performed in a thermomixer set to 60°C and 800 rpm for 1 h. The extracts were centrifuged for 5 min at 15,000g. The clear supernatants were diluted with 2-ml 20-mm TRIS/HCl pH 8.0 and loaded onto C8 100-mg solid phase extraction columns (Macherey-Nagel, Düren, Germany) conditioned with 1 ml 100% ACN and subsequently with 1 ml 20% ACN. The columns were washed with 1 ml 30% ACN. Subsequently, elution was performed with 1 ml 60% ACN and the eluates were directly used for HPLC. A LC-10 system (Shimadzu, Kyoto, Japan) equipped with a Symmetry 3.5 μm C18 100 × 4.6 mm column (Waters, Milford, MA) was used for HPLC analysis. The injection volume was set to 100 μl. Elution began with an isocratic flow of 1 ml/min of 67.7% solvent A (20 mm acetic acid set with NaOH to pH 4.8 in 15% ACN) and 32.3% solvent B (80% ACN) for 15 min. The concentration of solvent B was then raised linearly to 100% in 0.5 min and kept isocratic for another 2 min before reducing it to 32.3% within 0.5 min. Finally, the column was equilibrated with 67.7% solvent A and 32.3% solvent B for 7 min before injection of the next sample. The absorbance was recorded at 245 nm for quantification and 260 nm for purity check. The obtained chromatograms were analyzed with the software Clarity (DataApex, Prague, Czech Republic).

Supplemental Data

The following supplemental materials are available:

• Supplemental Figure S1. Protein sequence alignment of HsGCPII with AtAMP1 and analysis of PBDA treatment of solid medium-grown Arabidopsis plants.

• Supplemental Figure S2. Strategy and set-up of an AMP1-specific chemical genetic screen in Arabidopsis.

• Supplemental Figure S3. Enrichment analysis of GO terms of genes misregulated in both Arabidopsis HP-treated wild type and amp1-13.

• Supplemental Figure S4. Enrichment analysis of GO terms of HP-regulated genes in amp1-13.

• Supplemental Figure S5. HP shows efficient uptake characteristics and high stability in planta.

• Supplemental Table S1. Differentially expressed genes identified in the transcriptomic analysis using Arabidopsis ATH1 genome arrays.

• Supplemental Table S2. ReviGO GO enrichment analysis of genes misregulated in both HP-treated wild type and amp1-13.

• Supplemental Table S3. Differentially expressed genes identified in the transcriptomic analysis using Arabidopsis gene 1.1 ST array strips.

• Supplemental Table S4. List of primers used in this study.

Glossary

DAG

days (d) after germination

HP

hyperphyllin

OC, OCs

organizing center(s)

PHV

transcription factor PHAVOLUTA

SAM, SAMs

shoot apical meristem(s)

WUS

WUSCHEL