A genomic and expression compendium of the expanded PEBP gene family from maize.

The phosphatidylethanolamine-binding proteins (PEBPs) represent an ancient protein family found across the biosphere. In animals they are known to act as kinase and serine protease inhibitors controlling cell growth and differentiation. In plants the most extensively studied PEBP genes, the Arabidopsis (Arabidopsis thaliana) FLOWERING LOCUS T (FT) and TERMINAL FLOWER1 (TFL1) genes, function, respectively, as a promoter and a repressor of the floral transition. Twenty-five maize (Zea mays) genes that encode PEBP-like proteins, likely the entire gene family, were identified and named Zea mays CENTRORADIALIS (ZCN), after the first described plant PEBP gene from Antirrhinum. The maize family is expanded relative to eudicots (typically six to eight genes) and rice (Oryza sativa; 19 genes). Genomic structures, map locations, and syntenous relationships with rice were determined for 24 of the maize ZCN genes. Phylogenetic analysis assigned the maize ZCN proteins to three major subfamilies: TFL1-like (six members), MOTHER OF FT AND TFL1-like (three), and FT-like (15). Expression analysis demonstrated transcription for at least 21 ZCN genes, many with developmentally specific patterns and some having alternatively spliced transcripts. Expression patterns and protein structural analysis identified maize candidates likely having conserved gene function of TFL1. Expression patterns and interaction of the ZCN8 protein with the floral activator DLF1 in the yeast (Saccharomyces cerevisiae) two-hybrid assay strongly supports that ZCN8 plays an orthologous FT function in maize. The expression of other ZCN genes in roots, kernels, and flowers implies their involvement in diverse developmental processes.

The phosphatidylethanolamine-binding protein (PEBP) genes are found in all three major phylogenetic divisions of prokaryotes, archaea, and eukaryotes (Banfield et al., 1998;Hengst et al., 2001;Chautard et al., 2004). Their conserved sequences indicate an ancient common origin of a basic protein functional unit. The seminal PEBP described was isolated from bovine brain as a soluble 23-kD basic cytosolic protein (Bernier and Jolles, 1984). Animal PEBP proteins were subsequently shown to act as Raf kinase inhibitors (Yeung et al., 1999;Lorenz et al., 2003;Odabaei et al., 2004) in signaling cascades that control cell growth (Keller et al., 2004). Mouse PEBP exerts activity as a Ser protease inhibitor (Hengst et al., 2001). Overall, however, mammalian PEBPs are multifunctional proteins whose physiological roles are just beginning to be understood.
Plant PEBP-related genes were originally cloned from mutants with altered inflorescence architecture. These include Antirrhinum CENTRORADIALIS (CEN; Bradley et al., 1996), Arabidopsis (Arabidopsis thaliana) TERMINAL FLOWER1 (TFL1; Bradley et al., 1997), and tomato (Solanum lycopersicum) SELF PRUNING (SP; Pnueli et al., 1998). The Antirrhinum cen and Arabidopsis tfl1 mutations cause normally indeterminate inflorescences to terminate in a flower (Shannon and Meeks-Wagner, 1991;Bradley et al., 1996Bradley et al., , 1997. The tomato sp mutation changes the indeterminate growth habit to determinate, resulting in limited shoot growth and a bushy compact habit (Pnueli et al., 1998). These mutant phenotypes suggested these genes maintain the indeterminate state of the inflorescence meristem. CEN is activated in the inflorescence apex a few days after the floral transition, whereas TFL1 and SP are expressed in vegetative and inflorescence apices from very early stages where they delay the commitment to reproductive development (Bradley et al., 1997;Pnueli et al., 1998). The TFL1 and SP genes control both vegetative and reproductive phase durations by maintaining both apical and inflorescence meristem indeterminacy (Pnueli et al., 1998;Ratcliffe et al., 1998), while CEN controls only the inflorescence meristem (Bradley et al., 1996). Two distinct TFL1 homologs control pea (Pisum sativum) vegetative and reproductive phase duration: The LATE FLOWERING gene maintains shoot apical meristem indeterminacy, and DE-TERMINATE (DET) maintains inflorescence meristem indeterminacy (Foucher et al., 2003). TFL1 homologs with conserved function have now been identified in many eudicotyledonous (Amaya et al., 1999;Pillitteri et al., 2004;Sreekantan et al., 2004;Kotoda and Wada, 2005;Boss et al., 2006;Kim et al., 2006) and monocotyledonous plants (Jensen et al., 2001;Nakagawa et al., 2002;Zhang et al., 2005). The Arabidopsis TFL1 protein has recently been shown to be a mobile signal that moves from its localized transcriptional domain in the central region of the inflorescence meristem to the entire meristem, where it exerts its activity. The TFL1 protein is excluded from the floral meristem, allowing floral primordia to develop into a flower (Conti and Bradley, 2007).
The Arabidopsis FLOWERING LOCUS T (FT) gene is a member of the PEBP gene family, but it has an opposing function to TFL1: It promotes the transition from the vegetative to the reproductive phase (Kardailsky et al., 1999;Kobayashi et al., 1999). Critical amino acid residues differentiate FT and TFL1 proteins and dictate whether the protein acts as a floral activator or repressor. These residues include Tyr (Y85) in FT and His (H88) in TFL1 (Hanzawa et al., 2005), together with a stretch of 14 amino acid residues in exon 4 (Ahn et al., 2006). FT is a key activator of flowering, mediating both photoperiod and vernalization regulation (Baurle and Dean, 2006;Jaeger et al., 2006). The FT gene is transcribed in the leaf phloem (Takada and Goto, 2003), but the FT protein moves via the phloem to the shoot apex where it acts as a floral stimulus. This protein provides a molecular explanation for the long hypothesized mobile flowering signal, known as the florigen (Corbesier et al., 2007;Jaeger and Wigge, 2007;Mathieu et al., 2007). At the shoot apex, the FT protein interacts with the bZIP transcription factor FLOWER-ING LOCUS D (FD), and together they activate the meristem identity gene APETALA1, a MADS-box protein, triggering flower development Wigge et al., 2005).
The function of FT is highly conserved in plants. The rice (Oryza sativa) FT homolog encoded by Heading date3a (Hd3a; also known as OsFTL2) migrates from leaves to the shoot apical meristem to induce the floral transition (Tamaki et al., 2007). The tomato FT homolog SINGLE FLOWER TRUSS also regulates flowering time and shoot architecture by generating a graft-transmissible signal . In rice and perennial ryegrass (Lolium perenne), heading date quantitative trait loci (QTL) are associated with FT homologs Monna et al., 2002;Armstead et al., 2004). In wheat (Triticum aestivum) and barley (Hordeum vulgare) the VERNALIZATION LOCUS3 (VRN3) gene corresponds to an FT function (Yan et al., 2006;Faure et al., 2007). FT homologous genes accelerate floral development in transgenic woody plants, such as citrus (Poncirus trifoliata L. Raf.; Endo et al., 2005) and poplar (Populus spp.; Bohlenius et al., 2006), a quality that may speed tree breeding programs. In addition to regulating flowering time, the poplar FT homolog controls short-day fall-induced growth cessation and bud set (Bohlenius et al., 2006). In Norway spruce (Picea abies L. Karst.), an FT homolog mediates bud set and bud burst, indicating conservation of FT function in gymnosperms (Gyllenstrand et al., 2007).
Four additional PEBP genes were found in the Arabidopsis genome by homology: TWIN SISTER OF FT (TSF), BROTHER OF FT AND TFL1 (BFT), ARABI-DOPSIS THALIANA CENTRORADIALIS (ACT), and MOTHER OF FT AND TFL1 (MFT). TSF and MFT seem to act as floral integrators redundantly with FT (Yoo et al., 2004;Yamaguchi et al., 2005). The function of ACT and BFT remains unclear (Mimida et al., 2001). Small families of about six to eight genes are typical within eudicots such as Arabidopsis, tomato, poplar, and grape (Vitis vinifera; Mimida et al., 2001;Carmel-Goren et al., 2003;Kotoda and Wada, 2005;Carmona et al., 2007). Larger gene families were found in monocots such as rice (19 members), wheat (19), and maize (Zea mays; approximately 30; Chardon and Damerval, 2005). In this article we identified the likely complete set of maize PEBP-related genes, which we named Zea mays CENTRORADIALIS (ZCN) after CEN from Antirrhinum, the first PEBP-related plant gene described (Bradley et al., 1996). A combination of genomic, syntenic, phylogenetic, protein structural, and expression analyses revealed an expanded, diverse gene family. Included in this family are candidates for the probable functional orthologs of TFL1 and FT governing the floral transition, and also other members that may control developmental processes in roots, kernels, and flowers.

Identification and Characterization of the ZCN Gene Family
Extensive searches of public and proprietary transcript and genomic databases with six Arabidopsis PEBP proteins as BLAST queries (TFL1, FT, ATC, BFT, TSF, and MFT) identified 25 ZCN genes that were completed sequences from the appropriate bacterial artificial chromosome (BAC) clones (Table I). One of these, ZCN23, appears to be a pseudogene because only exon 4 was found in the 200-kb sequence of the corresponding BAC. All complete 25 maize genes included DNA fragments previously identified by Chardon and Damerval (2005). However, the final number of ZCN genes will be established upon completion of the maize genome sequencing project.
The nearest genetic markers for each ZCN gene were determined from BAC contigs and used to position the ZCN genes on the maize genetic map (Table I; Fig. 1). In addition to in silico mapping, the chromosome location of each ZCN gene was confirmed by PCR using the oat (Avena sativa)-maize addition lines (Supplemental Fig. S1; Supplemental Table S1), which are a set of 10 oat lines each carrying a single maize chromosome addition (Kynast et al., 2001). ZCN2, ZNC13, ZCN20, ZCN21, and ZCN26 mapped near centromeres (Fig. 1).
All the maize ZCN gene family members appeared to have generally low expression judging by the small number of ESTs in all the databases queried. cDNAs were obtained by reverse transcription (RT)-PCR using pairs of gene-specific primers designed close to the start and stop codons (Supplemental Table S2). No cDNAs were obtained for ZCN13, ZCN21, and ZCN24, evidently because no expressing tissues were found ( Table I). The intron/exon structures of the ZCN genes were determined from the alignments of the cDNA and genomic sequences (Fig. 2). If a cDNA was not obtained, the genomic sequence was used to determine the probable gene structure by alignment to the conserved coding regions and intron positions exhibited by other gene family members. Of the ZCN genes found, 22 of 24 have a conserved genomic structure consisting of four exons. The exceptions, ZCN2 and ZCN20, have three exons that apparently resulted from fusion of exons 1 and 2. Most ZCN genes have a compact gene structure in the range of 1 to 3 kb. The sizes of exon 2 and exon 3 are generally 63 and 42 bp, respectively, which are also the conserved exon sizes among other plant PEBP-related genes (Carmel-Goren et al., 2003;Zhang et al., 2005;Carmona et al., 2007). ZCN14, ZCN18, ZCN21, and ZCN24 are the largest genes, with sizes between 5 to 9 kb, as a result of expanded introns. The complete amino acid sequence for each gene was deduced from its cDNA sequence and in a few cases its genomic sequence (Table I).

Protein Phylogenetic Analysis and Rice Synteny
Phylogenetic analysis was used to infer the evolutionary relationship between the 24 maize and 19 rice PEBP proteins. The six Arabidopsis PEBP proteins and the wheat and barley FT functionally homologous proteins corresponding to VRN3 (Yan et al., 2006), were included as phylogenetic references. Protein alignments (ClustalW algorithm) revealed dispersed protein conservation, but with higher conservation over the ligandbinding site (Banfield and Brady, 2000). The most conserved protein segments beginning with EDL/DPL and ending with RRR/GGR residues (PEPB domain pfam01161; Supplemental Table S3) were used for the phylogenetic analysis (Kumar et al., 2004). The PEBP proteins form three well-marked subfamilies, which we designate after the three Arabidopsis reference genes, TFL1-like, MFT-like, and FT-like (Fig. 3). The dendogram topology of the plant PEBP proteins is consistent with a previous report (Chardon and Damerval, 2005).
The rice proteins, with the exception of OsTFL8, formed monophyletic subgroups with one or two maize proteins. The previously reported rice-specific OsFTL5 and OsFTL11 (Chardon and Damerval, 2005) are now associated with ZCN16 and ZCN17, respectively. Among the 24 maize ZCN proteins, 14 ZCN proteins appeared as seven paired monophyletic subgroups: ZCN1-ZCN3, ZCN4-ZCN5, ZCN7-ZCN8, ZCN9-ZCN10, ZCN13-ZCN21, ZCN18-ZCN24, and ZCN19-ZCN25. These gene pairs likely represent paralogs, a product of the tetraploid ancestry of maize (Gaut, 2001;Bruggmann et al., 2006). The TFL1-like subfamily is composed of six maize and four rice proteins. Maize ZCN1-ZCN3-ZCN6 proteins group with rice proteins RCN1 and RCN3. ZCN1 and ZCN3 genes seem to be a recent duplication, because they share 89.3% nucleotide identity within exons and 79.4% nucleotide identity within introns (Supplemental Fig. S2). The other monophyletic group is composed of maize proteins ZCN2-ZCN4-ZCN5 and rice RCN2-RCN4, the latter thought to represent a duplicated rice chromosomal segment (Chardon and Damerval, 2005). The ZCN4-ZCN5 gene pair does not share much intron homology, suggesting a less recent duplication. The MFT-like subfamily is the smallest, composed of three maize and two rice genes: ZCN9-ZCN10 grouped with OsMFT2, and ZCN11 grouped with OsMFT1.
The FT-like subfamily is the largest, composed of 15 maize and 13 rice genes. This subfamily forms two large monophyletic groups that we named FT-like I and FT-like II. The FT-like I group is of particular interest because it includes the key floral activators from Arabidopsis, FT, and TSF (Huang et al., 2005;Yamaguchi et al., 2005). The FT-like I group is divided into two main subgroups. One subgroup is composed of the Arabidopsis FT and TSF plus seven maize and five rice proteins, including maize gene pairs ZCN18-ZCN24 and ZCN19-ZCN25 and rice OsFTL4 and OsFTL6. The other subgroup contains maize ZCN14 and ZCN15 within the well-supported monophyletic grouping of known monocot floral activators, including rice OsFTL2 (Hd3a) and OsFTL3 (Hd3b; Kojima et al., 2002), wheat and barley VRN3, which are named TaFT and HvFT (Yan et al., 2006). The FT-like II group is composed of six maize and five rice proteins, but no Arabidopsis proteins, which suggests diversification and expansion of this subgroup after the monocoteudicot radiation. The ZCN7-ZCN8 pair forms a grouping with ZCN12 and OsFTL9. ZCN13-ZCN21 and ZCN26 are similar to OsFTL13 and OsFTL12, respectively. Only OsFTL8 does not directly group with any of the maize proteins.
Synteny analysis was performed to assess the correspondence between the phylogenetic groupings of the PEBP proteins and the prevailing rice-maize syntenic context for their gene locations, recognizing that synteny or departures therefrom can exist between these complex cereal genomes. The maize gene map positions and their syntenic rice counterpart(s) are summarized in Figure 1. In cases of a single maize-to-rice gene phylogenetic correspondence, the synteny was present and unambiguous. These examples are ZCN11-OsMFT1, ZCN2-RCN2, ZCN16-OsFTL5, ZCN17-OsFTL11, ZCN20-OsFTL7, ZCN14-OsFTL1, ZCN12-OsFTL9, and ZCN26-OsFTL12. In cases where duplicated maize genes corresponded to a single rice gene, both maize duplicated segments might be considered syntenic to a single rice locus. These examples are ZCN9/ZCN10-OsMFT2, ZCN13/ZCN21-OsFTL13, ZCN18/ZCN24-OsFTL4, and ZCN19/ZCN25-OsFTL6. In one case, a single maize gene (ZCN15) corresponded to tandemly duplicated rice genes (Hd3a/Hd3b), and the syntenic context was apparent. More challenging for syntenic interpretation  (Table I). Centromeric regions are shown as black rectangles. Pseudogene, ZCN23, is marked by an asterisk (*). Seven duplicated pairs of genes are marked by various colors: ZCN1/ ZCN3, blue; ZCN4/ZCN5-ZCN19/ZCN25, red; ZCN9/ZNC10, burgundy; ZCN18/ZCN24, orange; ZCN7/ZCN8, green; ZCN13/ZCN21, framed red over the centromeres. Rice genes (their chromosomes in parentheses) with the highest probability of synteny are shown on the left side of the maize chromosome.
are genes involved in apparently ancient segmental duplications prior to the rice-maize divergence. The large duplicated segments of rice chromosomes 2 and 4, carrying RCN2-OsFTL5 and RCN4-OsFTL6 genes, respectively, appeared to be syntenic to maize genes ZCN4-ZCN21 and ZCN5-ZCN19 on chromosomes 2 and 10, but this synteny remains somewhat ambiguous. A similar situation exists for the rice RCN1-RCN3 genes that reside in duplicated segments of rice chromosomes 11 and 12. The closely related maize genes ZCN1, ZCN3, and ZCN6 mapped to duplicated segments of chromosomes 3, 10, and 4, and synteny was difficult to assign. We were also not able to determine clear rice synteny with the maize pair ZCN7/ZCN8. The complete maize genome sequence may help clarify these and other remaining inconsistencies.

Maize ZCN Protein Structural Analysis
All PEBP proteins share a common structure, with a dominant feature being a central antiparallel b-sheet flanked by a small b-sheet on one side and two a-helices on the other. All family members also possess a highly conserved putative anion-binding site located at one end of the central b-sheet, supporting the idea that the proteins are functioning in part through forming complexes with a phosphorylated ligand (Banfield et al., 1998;Serre et al., 1998;Banfield and Brady, 2000;Ahn et al., 2006). The three-dimensional (3-D) structure of the CEN proteins suggested a role as a kinase regulator (Banfield and Brady, 2000), hinting that plant PEBP proteins may function as modulators of signaling cascades controlling plant growth and develop-ment. The critical structural variations among distinct members occur at the loops connecting these secondary structure elements and at the C-terminal regions.
To build 3-D models for the maize ZCN proteins, we used homology modeling with the Arabidopsis TFL1 (Protein Data Bank PDB:1wko A chain) and FT (PDB:1wkp A chain) proteins as structural templates. Out of 25 maize ZCN proteins, we focused on two representative candidates (ZCN2 and ZCN14) based on the highest sequence similarity to their Arabidopsis Figure 3. Phylogenetic analysis of the maize, rice, and Arabidopsis PEBP proteins. The protein sequences of 51 plant PEBP gene products including 45 Poaceae monocot (black) and six Arabidopsis (red) were limited to the conserved EDL/DPL...RRR/GGR region encompassing most of the complete proteins. The phylogenetic test used was the Bootstrap test by minimum evolution method, as implemented by MEGA version 3.1 (Kumar et al., 2004). Dendogram branches are labeled with percentage of 1,000 iterations supporting each branch. The scale bar at bottom reflects the frequency of amino acid substitutions between sequences as determined by the Poisson corrective distance model. counterparts. ZCN2 has 58% amino acid sequence identity to TFL1, and ZCN14 has 72% sequence identity to FT. The models were constructed with MODELER, and refined further by guidance from stereochemistry violation penalty rules. The overall refined models abided to their Arabidopsis templates with Ca RMSDs (root mean square deviations of Ca atomic coordinates) less than 1 Å (Fig. 4, A and B). Comparison between ZCN2 and TFL1 at the putative phosphatebinding site revealed that the key residues were 100% conserved, including Glu-109 (ZCN2)/Glu-112 (TFL1), His-87/90, Asp-71/74, Val-74/77, Leu-82/85, His-85/88, Asp-140/144, His-118/121, and Phe-120/123 (Fig. 4A). Similarly, ZCN14 and FT were also highly conserved in the binding site, again including Glu-107 (ZCN14)/Glu-109 (FT), His-85/87, Asp-69/71, Ala-72/ Val-74, . In addition, the previously identified structural features for distinguishing TFL1 and FT specificity, including the interacting triads, Glu-112-His-88-Asp-144 of TFL1 corresponding to Glu-109-His-85-Asp-140 of ZCN2 and Glu-109-Tyr-85-Gln-140 of FT corresponding to Glu-107-Tyr-83-Gln-138 of ZCN14 at the binding site entrance, were all well preserved between maize and Arabidopsis (Fig. 4, A and B). To investigate whether the binding site could accommodate a phosphorylated ligand, the bovine PEBP structure (PDB:1b7a) was superimposed, and then its cocrystallized ligand phosphoric acid mono-(2-amino-ethyl) ester (OPE in PDB structure) was transferred into the ZCN2 model. Further docking analysis regarding geometric complementarity and binding energy suggested that the OPE could fit well into the cavity without serious stereo conflict (Supplemental Fig. S3).
Conservation of ZCN protein primary structures was evaluated from the alignment of the ligand-binding motif and the external loop (Fig. 4C). All maize proteins possess invariable residues in the critical positions forming the ligand-binding pocket corresponding to Arabidopsis TFL1/FT residues Asp-74/71, Asp-75/73, Pro-78/75, His-90/87, and His-121/118 (Fig. 4C). The critical His-88 in TFL1 that defines the opposing repressor/activator activities of TFL1 and FT, respectively, is conserved in all six maize TFL1-like proteins. All other maize ZCN proteins have in this position Tyr, corresponding to Tyr-85 in Arabidopsis FT, with the exception of ZCN11 (MFT-like) and ZCN17 (FT-like I), that have Leu and Asn, respectively. The distinct feature of MFT-like proteins is substitution of conserved Glu-112/109 for Met in ZCN9 and ZCN10 or for Val in ZCN11. All the other maize proteins have Glu in this position.
The external loop encoded by exon 4 is important for the opposing activities of Arabidopsis TFL1/FT proteins (Ahn et al., 2006). The external loop is clearly defined in the alignment due to its variation in size between subfamilies (Fig. 4C). The loops of TFL1-like proteins are variable in size (14-17 residues) and have only two conserved residues despite the high amino acid conservation over the remainder of the proteins (100 invariant out of 180 total residues). The external loop of the MFT clade is also variable, ranging from 14 to 20 residues and having three invariant amino acids. The ZCN11 protein is the most diverged from the consensus with the longest (20 residues) external loop. The external loop of the FT-like subfamily has a uniform size of 14 amino acids with significant conservation of amino acids between members of the group. For instance, nine out of 14 residues are invariant in the FT-like I group, and eight out of 14/15 in the FT-like II group. This is consistent with the previous observation for the FT-like subfamily in other species (Ahn et al., 2006). In conclusion, the conservation of primary, secondary, and tertiary protein structures of maize ZCN proteins suggest their biochemical functions might also be preserved.

Expression Patterns of the ZCN Genes
As a step toward elucidating maize ZCN gene function, we surveyed ZCN transcript accumulation across a wide range of organs and tissues using the massively parallel signature sequence (MPSS) RNA profiling database (Fig. 5). MPSS technology is an open-ended platform that quantifies the level of transcript accumulation of virtually all genes in a sample by counting the number of individual mRNA molecules produced from each gene in the form of a 17-mer tag signature sequence (Brenner et al., 2000). The 17mer tag distributions confirmed the low level of ZCN gene expression across all tissues, as predicted from their low EST abundance. ZCN14 and ZCN11 appeared to be the only highly expressed genes in the broad set of tissues queried. The MFT-like subfamily showed preferred expression in kernel tissues. The FTlike II group was expressed mainly in leaves. To refine our analysis of gene expression patterns depicted by MPSS, we performed RT-PCR in nine different tissues: root, stem, immature leaf, mature leaf blade, shoot apex, immature tassel, immature ear, embryo, and endosperm. For selected ZCN genes, a more detailed developmental analysis was performed. Gene expression was detected using RT-PCR with pairs of genespecific primers corresponding to the first and last exon of each gene (Supplemental Table S2). Thus, this primer design allowed us to detect alternative transcript splicing patterns.
The maize TFL1-like subfamiliy showed similar tissue expression patterns with expression in roots, young stems, immature ear, and tassel (Fig. 6A). Because the Arabidopsis TFL1 gene maintains shoot apical and inflorescence meristem indeterminacy, we focused on the expression of TFL1-like genes in developing shoot apices, tassels, and ears (Fig. 6B). The closely related ZCN1 and ZCN3 genes are expressed in shoot apices during vegetative development and in both tassel and ear primordia during reproductive development (Fig. 6B). ZCN1 and ZCN3 accumulated significant amounts of unspliced pre-mRNA at early stages. ZCN3-spliced mRNA appeared in significant proportions after the floral transition in the V9 stage tassel and ear primordia, whereas ZCN1 mRNA remained mostly unspliced. None of the other TFL1-like genes accumulated unspliced transcript.
ZCN6 mRNA was detected at lower levels in shoot apices and tassel primordia, but its expression is upregulated in developing ears. ZCN2 and ZCN6 were expressed before and after the floral transition like ZCN1 and ZCN3, but they show a distinct pattern of accumulation (Fig. 6B). ZCN2 mRNA accumulation gradually increased in shoot apices around the floral transition, peaking in V7 and V8 tassels and decreasing in later stages. ZCN6 had a complementary pattern, showing low transcript accumulation in V3 to V8 stage apices and a higher accumulation at the V9 stage. In developing ears, ZCN2 was active at all stages tested whereas ZCN6 was expressed at the V7 to V9 stages and decreased later. ZCN4 and ZCN5 transcripts accumulated after the floral transition in both tassel and ear primordia (Fig. 6B), suggesting they act specifically in the developing inflorescence after the floral transition, similar to snapdragon (Antirrhinum majus) CEN (Bradley et al., 1996) and pea DET (Foucher et al., 2003). Five out of the six TFL1-like maize genes were expressed in roots (Fig. 6A).
MFT-like ZCN transcripts accumulated predominantly in kernels with the exception of ZCN11, which accumulated in both kernel and seedling tissues (Fig.  6C). Developmental profiling of the MFT-like genes was performed on whole kernels starting from 2 to 8 d after pollination (DAP) and on dissected embryo and endosperm tissues from kernels at 10 to 26 DAP. ZCN9 mRNA was detected in the embryo after 10 DAP and faint expression was found in the endosperm after 14 DAP. ZCN10 mRNA was detected only in the embryo after 14 DAP with a significant proportion of unspliced transcript. ZCN11 transcript accumulated in ovules before pollination and continued to accumulate in both embryo and endosperm at all postpollination stages (Fig. 6D).
The FT-like subfamily is a complex group composed of 15 maize ZCN genes in two groups named FT-like I and FT-like II. The FT-like I group is composed of seven maize genes that are further divided into two subgroups, one with Arabidopsis FT-TSF, and the other formed by the FT functionally conserved monocot homologs (Fig. 3). The duplicated ZCN19-ZCN25 pair of genes was expressed in roots. ZCN16 and ZCN20 produced unspliced mRNA in many tissues. ZCN17 is actively transcribed in roots and stems. ZCN18 is expressed in the stem and leaves producing a mixture of spliced and unspliced transcript. ZCN24 mRNA was not detected in this set of tissues (Fig. 6E). ZCN14 and ZCN15 are grouped tightly with FT homologous monocot floral activators (Fig. 3), thus being the most favorable candidates for possessing FT function. However, their expression patterns do not support their function as floral activators. In the broad set of tissues surveyed, ZCN14 mRNA was detected in the tassel and ear primordia and the endosperm, whereas ZCN15 expression was not detected in these samples (Fig. 6E). The original ZCN15 EST was found in a pedicel cDNA library. In developing kernels ZCN15 mRNA was detected in whole kernels between 4 and 8 DAP and in the dissected pedicels from later stage kernels (Fig. 6F). Dissected pedicels often contain the basal endosperm. Therefore, ZCN15 transcript accumulates in kernels after fertilization, most likely in the basal layer of the endosperm. This expression pattern is inconsistent with ZNC15 as a floral activator. ZCN14 transcript was detected in ovules and early developing kernels, but its level gradually decreased after 12 DAP (Fig. 6F). Because ZCN14 is the only FT-like gene whose transcript accumulates in the ear and tassel primordia, we investigated its expression in these tissues as well (Fig. 6F). In shoot apices, weak transcript accumulation is detected at the V5 stage, coinciding with the floral transition. ZCN14 mRNA levels increased in developing tassels at the V7 stage and remained constant up to the V9 stage. ZCN14 transcript accumulated in V7 stage ear buds and later developing ear stages. The ZCN14 expression pattern is quite different from Arabidopsis FT and rice Hd3a, which are transcribed exclusively in leaf blades (Corbesier et al., 2007;Jaeger and Wigge, 2007;Mathieu et al., 2007;Tamaki et al., 2007). Thus ZCN14 is unlikely a floral activator in maize.
The FT-like II group is composed of six maize genes. Three maize genes of this group, ZCN8, ZCN12, and ZCN26, are expressed predominantly in leaf blades (Fig. 6G). Three other genes seem to be nonfunctional or with a limited function because they produce either unspliced mRNA (ZCN7) or their expression was not detected in any tissues tested (ZCN13 and ZCN21). We investigated further the developmental expression pattern of ZCN8, ZCN12, and ZCN26 from FT-like II and ZCN14 and ZCN18 from FT-like I in stems, immature leaves, and leaf blades at the V2 to V9 stages   (Fig. 6H). We also included ZCN7 to test mRNA splicing during development; however, only unspliced ZCN7 RNA was detected in stems and immature leaves and no RNA was found in leaf blades at various developmental stages. Its paralog ZCN8 also produced unspliced transcript in stems and immature leaves, but in expanded leaf blades, ZCN8 mRNA was completely spliced (Fig. 6H). This finding indicates that ZCN8 activity may be regulated by mRNA splicing. Importantly, the spliced ZCN8 transcript was detected in leaf blades before the floral transition at the V3 stage. Its level increased at the V4 stages just before the floral transition and stayed high afterward. This expression pattern makes ZCN8 an attractive candidate as a floral activator. Another leaf-specific gene, ZCN12, was found to be activated in leaf blades at the V7 stage after the floral transition and continued to be expressed at later stages (Fig. 6H). ZCN26 is expressed predominantly in leaf blades and produced a significant amount of unspliced pre-mRNA. ZCN18 was expressed mainly in the stem with weaker expression in leaf blades. ZCN14 mRNA was barely detected by RT-PCR in immature leaves and in leaf blades at vegetative V4 and V5 stages, but after the floral transition, the ZCN14 mRNA level increased and was reliably detected in leaf blades (Fig. 6H).

Yeast Two-Hybrid Interactions between DLF1 and Maize FT-Like Proteins
In Arabidopsis the floral activator FT interacts with the bZIP transcription factor FD to activate the floral identity genes inducing flower development . FT and FD protein interaction was demonstrated genetically and in the yeast (Saccharomyces cerevisiae) two-hybrid system Wigge et al., 2005). Of the maize FT-like ZCN genes the mostly likely candidate for FT function is ZCN8. Its expression pattern is consistent with a role as the floral activator. In maize the bZIP transcription factor DLF1 is an activator of the floral transition and has function similar to the Arabidopsis FD (Muszynski et al., 2006). We tested interactions between DLF1 and ZCN8, ZCN12 and ZNC14 proteins in the yeast two-hybrid system (Table  II). The DLF1 protein shows self interaction that is typical for bZIP transcription factors that form dimers through Leu zippers. None of the ZCN proteins showed self interaction. Of three ZCN proteins tested, only ZCN8 interacts with DLF1 in both combinations as bait and prey (Table II; Supplemental Fig. S4). ZCN14 interacts with DLF1 as bait but not as prey, indicating a weaker or ambiguous interaction. This result strongly supports the hypothesis that the ZCN8 gene functions as a floral activator in maize similar to the Arabidopsis FT and rice Hd3a.

Maize Contains an Expanded PEBP Gene Family
Growing evidence indicates that PEBP-related proteins are potent and sometimes mobile regulators of plant architecture, development, and seasonal growth adaptation (Bohlenius et al., 2006;Yan et al., 2006;Conti and Bradley, 2007;Corbesier et al., 2007;Tamaki et al., 2007). A survey study on this important gene family in cereals was published earlier, but maize genes were represented largely by raw sequences available from public EST and GSS (genome sample survey) databases (Chardon and Damerval, 2005). In this article we redress this gap by identifying and curating the entire maize PEBP gene family, including complete gene and protein structures, evolutionary relationships with rice and Arabidopsis, and spatiotemporal specific gene expression patterns.
We identified 25 ZCN genes, although one is likely a pseudogene. Eudicot genomes such as Arabidopsis and poplar have about a half-dozen PEBP genes (Mimida et al., 2001;Carmel-Goren et al., 2003;Carmona et al., 2007). Monocots are trending higher, with at least 19 known in rice (Chardon and Damerval, 2005). The phylogenetic analysis revealed three major ZCN subfamilies, the TFL1-like, MFT-like, and FT-like, as in other plant genomes, reinforcing the idea that the monocoteudicot last common ancestor had at least three PEBP members (Chardon and Damerval, 2005). Gene distribution between these rice subfamilies is 4:2:13 (TFL1-like:MFT-like:FT-like). The ZCN subfamily distribution is 6:3:15, which is nearly double the number relative to the presumed monocot ancestor. The larger FT-like subfamily fell into two groups with the FT-like II group remaining monocot specific.
The maize PEBP gene family expansion (to 25 members) compared to rice (19 members) could be explained by the ancient tetraploid ancestry of maize, in which a genome duplication occurred after divergence from the rice and maize common ancestor, followed by subsequent diploidization en route to modern maize Figure 6. (Continued.) kernels, pedicel (ZCN15), and in developing tassels and ears (ZCN14; F). Expression patterns of FT-like II genes in a broad set of tissues (G) and in developing stems and leaves (H). Horizontal arrows indicate spliced mRNA. White stars indicate the floral transition. Asterisks (*) mark genes with no detectable expression. V3 to V10 indicates the vegetative stages of plant development defined according to the appearance of the leaf collar of the uppermost leaf. The superscript numbers 1 and 2, mark the early and later stages of V4 and V7, respectively, as judged by the shorter internode length at the early stages. Beginning of the floral transition was determined by the elongation of the shoot apical meristem followed by the appearance of the spikelet pair meristem that typically takes 1 to 2 d. Tissues are abbreviated as: EM, Embryo; EN, endosperm; IL, immature leaves; IT, immature tassel; IE, immature ear; LB, leaf blade; SAM, shoot apical meristem; R, root; S, stem.  (Paterson et al., 2004;Swigonova et al., 2004;Bruggmann et al., 2006). This diploidization process is thought to have shed at least 50% of the duplicated genes . The seven duplicated pairs of ZCN genes may have survived this diploidization contraction, contributing to the higher overall family gene count. Syntenic relationships between rice and maize PEBP genes were sometimes confusing due to gene duplications and deletions during evolution of each species; however, for 13 rice genes their syntenic maize ZCN counterparts were found (Fig. 1).

Most of the ZCN Genes Have Distinct Expression Patterns
Most of the ZCN genes appear to be functional, because all but one gene predicts a credible complete open reading frame, and all but three genes (ZCN13, ZCN21, and ZCN24) have transcripts detected in at least one tissue of 13 tested by RT-PCR. Although the three poorly expressed genes may be heading toward evolutionary degeneracy, their open reading frames are complete, and important cryptic expression cannot be ruled out. Further, the duplicated ZCN13 and ZCN21 genes are located near centromeres 5 and 2, where expression may be suppressed due to epigenetic silencing spreading from the heterochromatin. ZCN20 is also centromeric and shows very low expression.
The presence of duplicated genes raises the question about their functional redundancy. According to evolutionary models, duplicated genes may undergo different selection processes: nonfunctionalization where one copy loses function, hypofunctionalization where one copy decreases in expression or function, neofunctionalization where one copy gains a novel function, or subfunctionalization where the two copies partition or specialize into distinct functions (Lynch and Conery, 2000;Otto and Yong, 2002;Duarte et al., 2006). These evolutionary fates may be indicated with divergence in expression patterns or protein structure. We compared the expression patterns of the seven duplicate ZCN gene pairs that were not duplicated in the rice genome. Three maize gene pairs from the FT-like subfamily suggest non/hypofunctionalization. These dominant/submissive pairs are ZCN19/ZCN25, ZCN18/ZCN24, and ZCN7/ZCN8. Both ZCN19/ ZCN25 are expressed in roots, but ZCN19 has a relatively high level whereas ZCN25 transcript is barely detected. ZCN18 is expressed in stems and leaves, but a ZCN24 transcript was not found in any tissue tested. The ZCN8/ZCN7 pair could be nonfunctionalization involving differential splicing. Both ZCN8 and ZCN7 are transcribed, but ZCN7 transcript remains unspliced in all tissues tested, whereas ZCN8 mRNA is completely spliced in the leaf blade where function is anticipated.
Possible subfunctionalization trends are inferred by expression patterns for gene pairs ZCN1/ZCN3, ZCN4/ZCN5, and ZCN9/ZCN10, which showed clear expression pattern shifts. ZCN1 and ZCN3 are both expressed in the shoot apices before the floral transition and in the ear and tassel primordia after the floral transition, but they produce different levels of fully spliced mRNA. ZCN1 transcripts are mostly unspliced, whereas ZCN3 produces mostly spliced mRNA, and accumulates this fully spliced mRNA in ears and tassels at later developmental stages. ZCN4 and ZCN5 are expressed in developing ears and tassels after the floral transition, but their expression is activated at somewhat different stages of development. The duplicated genes ZCN9 and ZCN10 are expressed in kernels, but ZCN9 is expressed in the embryo and endosperm, whereas ZCN10 is embryo specific. The other duplicated pair ZCN13/ZCN21 may have undergone bilateral nonfunctionalization, as apparently neither gene is expressed.

Nine ZCN Genes Display Regulated RNA Processing
In plants regulated RNA processing and alternative splicing are common mechanisms of the posttranscriptional regulation of gene expression (Reddy, 2007). Nine ZCN genes show distinct patterns of transcript processing that appeared to be development and tissue dependent. The first pattern is displayed by ZCN16 and ZCN20 that produce low levels of unspliced pre-mRNA in many tissues accompanied by properly spliced transcript in one tissue, leaf for ZCN16, and tassel for ZCN20. The second pattern is displayed by ZCN10, ZCN18, and ZCN26 that are expressed at relatively high levels predominantly in one tissue type, embryo, stem, and leaf blade, respectively. They produced unspliced pre-mRNAs and two to three alternatively spliced transcripts per gene. The third pattern is displayed by the ZCN1/ZCN3 genes that showed developmentally regulated alternative splicing. ZCN1 pre-mRNA is mostly unspliced in the shoot apices before the floral transition and at early stages of ear and tassel development, whereas ZCN3 produces proportionally more fully spliced mRNA in these tissues, and fully spliced mRNA in ear and tassel at later stages that suggests a more prominent role for ZCN3 relative to ZCN1. The fourth and most intrigu- Table II. Yeast two-hybrid interactions between DLF1 and FT-like ZCN8, ZCN12, and ZCN14 proteins Interactions were assessed by transformed yeast colony growth on the SD/-Leu/-Trp/-His dropout media (Supplemental Fig. S4) using the reciprocal combinations of the bait pGBKT7 vector and the prey pGADT7 vector with insertions of the tested cDNAs. 1, Interaction; 2, no interaction; 6, a weak interaction only in one vector combination.
6 2 ing pattern is displayed by ZCN8, which apparently demonstrates tissue-specific RNA maturation. A low level of unspliced ZCN8 pre-mRNA is detected in stem and immature leaves at V3 to V9 stage seedlings, but fully spliced ZCN8 transcripts capable of producing a functional protein appeared in expanded leaf blades at the same seedling stages. Its paralog, ZCN7, produced unspliced pre-mRNA in stems and immature leaves, but no detectable transcripts in leaf blades. This indicates that ZCN8 may play a larger role relative to its duplicate ZCN7.
ZCN8 Is a Candidate Gene for the Floral Activator FT in Maize The Arabidopsis FT protein and its orthologs in other plant species serve as long-range developmental signals in activation of the floral transition Corbesier et al., 2007;Jaeger and Wigge, 2007;Lin et al., 2007;Mathieu et al., 2007;Tamaki et al., 2007). Which maize ZCN gene(s) plays a similar floral activator function? Of 15 ZCN genes in the FT-like subfamily, only ZCN8 and ZCN26 are expressed in leaf blades before and after the floral transition, and neither is expressed in the shoot apical meristem as would be expected for FT function (Fig. 6, G and H). ZCN8 is favored because it produces fully spliced mRNA in leaves, whereas ZCN26 produces several splicing variants. However, there are inconsistencies that warrant discussion. First, the ZCN8 locus has no clear syntenic counterpart in rice. Second, ZCN8 protein is not phylogenetically subgrouped with the FT homologous proteins that are floral activators in rice, barley, and wheat (Fig. 3). Third, ZCN8 has two substitutions in the putative ligand-binding domain compared to Arabidopsis FT: Val-120(FT)/Leu-118(ZCN8) and Gln-140(FT)/His-138(ZCN8). Based on structural analysis, it is unlikely that these two substitutions dramatically alter the ZCN8's binding capability. Val-120 of FT is at the bottom of the binding site and its side chain is approximately 6 Å away from the modeled OPE according to the bovine protein (PDB: 1b7a), thus leaving ample space to accommodate Leu-118 in ZCN8 without blocking the ligand binding. Gln-140 of FT is at the binding cavity rim and fully exposed to solvent (PDB: 1wkp). The change to His-138 in ZCN8 seemingly maintains the Gln's hydrophilic property, and might even preserve the potential capability to hydrogen bond to the incoming ligand through its nitrogen atom on the imidazolium ring, ND1 or NE2.
However, the yeast two-hybrid assay provides strong evidence that ZCN8 likely functions as the FT floral activator because only the ZCN8 protein interacts with the DLF1 bZIP transcription factor similar to interaction of the FT and the bZIP transcription factor FD in Arabidopsis . Moreover, ZCN8 is linked to marker umc2075 on chromosome 8 that maps in the vicinity of the flowering QTL vegetative-togenerative transition2 (Vladutu et al., 1999;Chardon et al., 2004), one of the strongest maize QTLs for flowering time. Further functional analysis of ZCN8 as the maize floral activator will require a genetic study of mutants and/or transgenic plants. It is also important to point out that none of the FT-like ZCN genes were expressed in immature leaves, the tissue where the major floral regulator indeterminate1 is expressed (Colasanti et al., 1998).
We identified six maize genes whose proteins formed a well-supported monophyletic subfamily with Arabidopsis TFL1. The important TFL1 protein structures are all preserved in the six maize homologous proteins including His-88, which is critical for repressing flowering (Hanzawa et al., 2005), and the divergent external loop (Ahn et al., 2006). Modeling 3-D structure of maize ZCN2 protein revealed that its overall fold structure, putative phosphate-binding site, and features distinguishing its floral repressive activities are all preserved in the maize TFL1-like proteins. Thus, all six maize homologous proteins may potentially function similar to TFL1. The expression patterns of all six genes are also consistent with their putative role in meristem maintenance, as all are expressed in the shoot apices or the developing inflorescence. According to their expression patterns, TFL1-like ZCN genes can be divided into two groups. One group of genes (ZCN1, ZCN3, ZCN4, and ZCN6) is expressed before and after the floral transition, suggesting functions in both vegetative and reproductive meristems. The second group, ZCN4 and ZCN5, is expressed only in the developing inflorescence after the floral transition, suggesting functions in maintenance of the inflorescence meristem. This suggestive partitioning of function between several genes for the independent maintenance of the vegetative and inflorescence meristem has been shown in Antirrhinum (Bradley et al., 1996), tobacco (Nicotiana tabacum; Amaya et al., 1999), and pea (Foucher et al., 2003).
The expression patterns of maize ZCN genes hint at roles in root development. The TFL1-like maize genes ZCN1, ZCN3, ZCN2, and ZCN5 are expressed in the root tips as fully spliced mRNA, suggesting a possible function in the root apical meristem. In addition, among the FT-like subfamily, ZCN19, ZCN25, and ZCN17 are expressed mainly in the roots. These could also be involved in root development perhaps as a mobile signal akin to FT.
Another likely role for some of the ZCN genes is in kernel development. The seed-specific expression of the maize MFT-like genes ZCN9 and ZCN10, and broader kernel plus other tissues expression of ZCN11, suggests these MFT-like genes function during seed development. Indeed, the Arabidopsis MFT gene is highly expressed in early developing seeds (Supplemental Table S4). This finding supports the hypothesis that Arabidopsis MFT may act during seed development in addition to its role as a redundant floral inducer (Yoo et al., 2004). Moreover, rice OsMFT1 and OsMFT2 display high level expression (by MPSS) in developing and germinating seeds (Supplemental Table S5). The MFT-like genes' EST distributions from barley, rice, and wheat are apparently derived from spike or kernel cDNA libraries (Chardon and Damerval, 2005), supporting the hypothesis of a kernel-preferred function for the MFT-like genes in the Poaceae. Two genes from the FT-like subfamily, ZCN14 and ZCN15, are also expressed in kernels. The highly restrictive expression of ZCN15 in the basal layer of the endosperm after fertilization indicates a possible novel function in developing endosperm. In short, genes from both MFT-like and FT-like subfamilies are expressed in developing kernels and may function in distinct aspects of kernel development.
ZCN14 expression in various tissues suggests its involvement in other developmental processes beyond flowering time. ZCN14 is activated after the floral transition in the tassel and ear primordia and expressed during flower development and later in ovules and kernels. To date, ZCN14 is the only FT-like gene with apparent expression in the inflorescence meristem. Rice OsFTL1 syntenic to ZCN14 has a similar pattern of expression in spikes and kernels (Izawa et al., 2002;Chardon and Damerval, 2005). Barley HvFT2 and a putative wheat ortholog appear to have similar expression in spikes and kernels (Chardon and Damerval, 2005;Faure et al., 2007). This raises the possibility of a role for ZCN14 and related orthologs in plant seed and flower development well after the floral transition.
In conclusion, this study of the maize PEBP gene family affirms the conserved protein structures and function in flowering time, but the gene expression results, including alternative splicing, point to likely adaptation and specialization of physiological roles, further investigation of which this study may help guide.
Maize (Zea mays) sequence subject sources were proprietary ESTs and their assemblies, publicly available ESTs, CDS, GSS, BACs, and The Institute for Genomic Research genomic GSS assemblies AZM_4, AZM_5 (http://maize.tigr. org/), and MAGI_4 (http://www.plantgenomics.iastate.edu/maize/). All potential hits to conserved regions of the PEBP gene family were assembled and curated, and additional rounds of searching were performed to extend the genomic and/or transcript sequences across the most complete possible transcript region. The 33 unique sequences obtained originally by this survey were used to design overgo probes for screening genomic BAC libraries. Selected BACs were sequenced and assembled. The exon/intron structures of the ZCN genes were deduced from alignments of cDNA and BAC genomic sequences using the program Sequencher 4.7 (Gene Codes Corporation).

Tissue Preparations
Maize plants (genotype B73) were grown in the field. Vegetative growth stages (V1-V9) were defined according to the appearance of the leaf collar of the uppermost leaf (Muszynski et al., 2006). Shoot apices with one or two leaf primordia attached were dissected from seedlings at V3 to V9 stages under the dissecting microscope. Immature leaf blades were sampled from the plant whorls. Fully expanded green leaves were sampled as leaf blade. Lateral ear buds and ear primordia were sampled at the V6 to V10 stages. Kernels were collected after pollination at 2 d intervals. After 10 DAP the embryo and the endosperm tissues were dissected from the kernels.

RNA Isolation, RT-PCR, and cDNA Cloning
Small tissues such as shoot apices and ear buds were homogenized in 300 mL of TRIzol Reagent (Roche Diagnostics Corporation) using a 1.5 pestle (VWR KT479521-1590). Immature leaves and leaf blades were ground with a mortar and pestle in liquid nitrogen. Ground tissue (50 mg) was treated with 300 mL of TRIzol. Total RNA was isolated with TRIzol Reagent in combination with Phase Lock Gel (Brinkmann Instruments Inc.) according to the manufacturer's instructions. Complementary DNA synthesis was performed with Superscript First-Strand Synthesis system (Invitrogen). RT-PCR amplification was performed using Expand Long Template DNA polymerase (Roche). Primers were designed according to the predicted gene structure deduced from BAC sequences. Primers are shown in Supplemental Table S1. Two microliters of the cDNA reaction was used for PCR amplification in a 50 mL volume. The PCR conditions were 95°C for 2 min followed by 35 cycles at 94°C for 45 s, 58°C for 45 s, 72°C for 1 min, and a final extension of 72°C for 10 min. PCR products were cloned in pCR4-TOPO (Invitrogen) and sequenced.

Yeast Two-Hybrid Assay
The commercial kit MATCHMAKER GAL4 Two-Hybrid system 3 (Clontech) was used according the manufacturer's protocol. cDNAs of interacting genes were cloned into pGBKT7 (the Bait vector) and pGADT7 (the Prey vector) in reciprocal combinations. Both plasmids were transformed into the AH109 yeast (Saccharomyces cerevisiae) strain and plated on the dropout media SD/-Leu/-Trp. Individual colonies were then replated in parallel on two plates with dropout media SD/-Leu/-Trp and SD/-Leu/-Trp/-His. Colony growth on the His-minus media was indicative of protein interactions.

MPSS Analysis
The DuPont MPSS (Solexa) 17-mer expression libraries isolated from over 200 diverse maize tissues and developmental stages was queried with ZCN gene sequences. The MPSS samples were curated and grouped by tissuedevelopmental criteria, and from these groupings the mean parts per million for each MPSS 17-mer tag was calculated for 13 tissues, and then used to represent that ZCN gene's transcript abundance. Arabidopsis MPSS can be found at http://mpss.udel.edu/at/?/.

Protein Structural 3-D Modeling
The starting coordinates for molecular dynamics simulations and homologous modeling were taken from the x-ray structures PDB:1wko A chain at 1.80 Å resolution and PDB:1wpb A chain at 2.60 Å resolution. The initial structural coordinates of ZCN1 and ZCN14 were constructed using InsightII's MODELER module with its autoenergy minimization procedure (Accelrys). Before further analysis and molecular dynamics, all the structures were energy minimized under various constraints to relax the structure gradually, first in virtual vacuum with the crystal waters if applicable, and then subsequently in solvent water boxes.
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers EU241892 to EU241916 and EU241917 to EU241937.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Mapping of ZCN genes on oat-maize addition lines.
Supplemental Figure S2. Alignment of ZCN1 and ZCN3 genes across exon/intron sequences.
Supplemental Figure S3. Anion-binding site charge distribution in ZCN2 protein model structure.
Supplemental Figure S4. The yeast two-hybrid assay for interaction of the DLF1 and the FT-like proteins ZCN8, ZCN12, and ZCN14.
Supplemental Table S1. A list of primers for ZCN gene mapping on the oat-maize addition lines.
Supplemental Table S2. A list of primers for RT-PCR.
Supplemental Table S3. Multiple rice, maize, and Arabidopsis PEBP protein alignment used for phylogenetic analysis.
Supplemental Table S4. MPSS expression analysis of Arabidopsis PEBP genes.
Supplemental Table S5. MPSS expression analysis of rice MFT-like genes.