Meiotically competent cells in plants differentiate late in floral ontogeny. Presently, the steps required for specification of meiotic competence in a tiny number of floral somatic cells are unknown. However, nuclear events in meiotic cells are readily distinguishable from those in mitotic cells, and the developmental program of archesporial cells and PMCs is independent of successful somatic cell development at least through progression into prophase I . The ameiotic1 mutants of maize demonstrate that conducting meiosis is distinguishable from the preceding PMC differentiation events such as cell and nuclear enlargement and a rounded cell shape. In all am1 mutant anthers, archesporial cells are normal and proliferate to make the normal numbers and morphology of highly enlarged PMCs present in 1.0 mm maize anthers; somatic cell differentiation is also cytologically normal. For all am1 alleles except am1-praI, the PMCs remain mitotic rather than switching to meiosis, suggesting that both repression of mitosis and activation or maintenance of meiotic programs could be required to enter meiosis and that AM1 contributes to these processes. Cells in the somatic layers are still dividing at 1.0 mm but most layers have stopped dividing by 1.5 mm ; the somatic cessation of mitosis does not require AM1.
Global transcriptome analysis reinforces distinctions in the anatomical phenotypes of am1-489 and am1-praI. Both mutants have few cytological and moderate transcriptome differences compared to fertile at 1.0 mm. At 1.5 mm fertile PMCs have entered meiosis while mutant PMCs have not; unsynchronized mitosis is observed in am1-489 PMCs, and these cells in am1-praI are still at the pre-meiotic interphase. The mutant transcriptomes are more distinctive compared to fertile anthers at 1.5 mm. A comparison of the am1-praI profiles at 1.0 and 1.5 mm confirms that these anthers are developmentally delayed at 1.0 mm and that a substantial fraction of the transcriptome expected at 1.0 mm appears later at 1.5 mm. These results indicate that AM1 has an early impact on the pace of PMC maturation that is critical for timely meiotic entry.
From profiling we discovered maize genes associated with meiotic entry and the L/Z transition in PMCs using three methods: analysis of meiosis-associated genes, K-median clustering, and comparison to transcripts detected in isolated normal PMCs. Many meiosis-associated transcript types identified in our transcriptome profiling study have been reported to play major roles in sister chromatid cohesion, synapsis, homologous recombination, and double-strand repair . Based on homology to defined proteins, a region shared between SWI1/DYAD1 and AM1 is predicted to have DNA binding ability like other PHD finger homeodomain proteins  and likely to be involved in regulating transcription. Our array data clearly demonstrate that AM1 attenuates the expression of other meiotic genes. Interestingly, none of the meiosis-associated genes are regulated in an absolute On/Off pattern, somewhat surprising given that am1-489 PMCs perform mitosis instead of meiosis. These results redefine a role of AM1 in the modulation of transcript accumulation for many critical meiotic genes rather than simply switching them on or off. This could also imply that many meiotic genes including those associated with DNA repair are expressed whether or not meiosis will proceed; this distinguishes plants from other well-studied eukaryotes.
One critical step to distinguish meiosis from mitosis is homologous chromosome pairing. Only during meiosis are homologous chromosomes held together by the formation of the synaptonemal complex and remain attached until meiosis prophase I has completed by the formation of chiasmata, the cytological manifestation of crossovers. Extensive studies have identified numerous genes involved in these processes after meiotic entry , and several genes highlighted in comparing am1 to fertile are in this category (Additional File 2). Transcript types affected only in am1-489 include those with homology to Afd1, RPA70, Ms5, and Msh5 gene families. Transcript differences found in both mutant anthers include those with homology to Skp1B, Rad54-like, Msh4, Sds, and Zyp1. AFD1, a maize REC8 homolog acting downstream of ASY1/HOP1, is associated with the maintenance of the axial/lateral elements of the synaptonemal complex and regulation of sister chromatid cohesion . Pawlowski et al.  found that AFD1 is loaded onto the chromatin in am1-praI but not in am1-1 PMCs. Both the array and qRT data showed a moderate yet significant reduction of Afd1 transcripts at 1.0 mm in am1-489, which is phenotypically similar to am1-1, but not in am1-praI. Dmc1, a RecA homolog, is involved in meiotic double strand DNA break repair . In plants Dmc1 transcripts are found in PMCs and ovule meiotic cells  and sporadically in mitotic cells . The maize Dmc1 homolog analyzed is down-regulated in am1-489 but not in am1-praI compared to fertile. Thus the single amino acid mutation in the am1-praI protein does not impact the transcript levels of Dmc1 or Afd1 genes. We can conclude that an important transcriptional regulation of Afd1 and Dmc1 occurs during meiotic entry but is not essential later during the L/Z transition. The status of the praI "meiocytes" is probably affected by another level of control for achieving a zygotene-like stage. Other mechanisms such as post-transcriptional regulation and/or nuclear entry of specific meiotic proteins might then be more important for meiotic prophase progression .
RAD51, another RecA homolog, is required for homology searching, chromosome pairing, DNA strand transfer (D-loop formation), and double-strand break repair in meiosis and in DNA repair in somatic cells . RAD51 protein is absent in am1 mutants . Neither Rad51A nor Rad51A2 transcript levels are affected, therefore, post-transcriptional regulation of RAD51 protein levels is likely. RAD54 is a branch migration protein that is involved in dissociating D-loops formed by DMC1 during meiosis as well as D-loops formed by RAD51 though less efficiently . Interestingly, Rad54-like transcript is significantly reduced in both am1 mutants.
Skp1B of another known meiotic gene family is also down-regulated in both am1 mutants. The Arabidopsis Skp1-like1 (ASK1) gene was previously reported to perform a critical role in recombination during the leptotene to pachytene stages [24, 25]. ASK1 protein is important for many cell processes including modifying and reconstructing meiotic chromatin [24, 25] and telomere formation . Telomere bouquets are missing in am1-489 and are retarded or incomplete in am1-praI . Telomere clustering on the nuclear envelope is another crucial event of meiotic chromosome organization during prophase I that occurs at the L/Z transition in many organisms [64, 65]. Based on clustering analysis, Rad54-like and Skp1B expression patterns are very similar and dependent on Am1. Therefore, all three proteins might work in concert to advance meiocytes through the L/Z transition, the stage when am1-praI is arrested.
Other interesting genes differentially expressed in both mutants compared to fertile that have been implicated in the meiotic process include Always Early1 , Werner syndrome helicase , Zap-3 telomerase , and kinesin . Our analysis of these genes along with Skp1B, Rad54-like and Am1 shows a predominant expression in meiocytes (at least 5 times higher than whole anthers and at least 3 times higher than tapetum). This finding suggests that AM1 interacts directly or indirectly with these proteins in establishing or stabilizing meiotic chromosomes. Two good candidate genes for apomixis, Afd1 and Brassinosteroid insensitive1-associated receptor kinase genes [69, 70], were down-regulated in am1-489 anthers at 1.0 mm and in both mutant anthers at 1.5 mm, respectively.
According to Palmer , am1-1 anthers degenerate after one or 2 rounds of ectopic PMC mitosis and fail by the ~ 2.2 mm anther stage. am1-489 anthers also cease growth, and this is similar to most other pre-meiotic male-sterile mutants . Therefore, the continued growth of am1-praI anthers to 4-5 mm and their normal morphology indicate that the transcriptome differences are primarily causing defects in the PMCs and not significantly impairing somatic cell layers, despite inferred changes in the somatic and detected changes in the tapetal transcriptomes. We further conclude that meiotic entry by PMCs is sufficient to trigger somatic anther growth; the nature of the developmental signal is yet to be determined.
To more critically address the cell type specificity of transcriptome alterations, the 530 non-redundant genes differentially expressed in both am1 alleles were used to query gene lists derived from array profiling of laser microdissection-enriched PMCs or tapetal cells in normal anthers. Fifty-six percent were expressed at least 2-fold higher in PMCs compared to those found in the immediately adjacent tapetum. Our results indicate that either the complete absence of AM1 or the presence of excess aberrant AM1-PRAI protein has a profound impact on gene expression in the PMCs during the maturation phase from 1.0 to 1.5 mm anther length, including the period of meiotic entry up to the L/Z transition stage. Based on estimation of the contribution of PMC RNA to total anther RNA , we hypothesize that transcripts enriched 4-fold or higher could be PMC-specific. In fact, the highest enrichment observed was 9-fold, initially puzzling because PMCs are less than 1% of the cell population but should contain many unique transcripts associated with meiosis. After accounting for RNA contribution, however, enrichment of 4- to 5-fold in isolated PMC is the expectation.
Recently, multiple microarray or RNA-Seq analyses of purified meiocytes have been reported [5, 41, 42, 71]. Using manual methods or laser capture microdissection, meiocytes relatively free of contaminating somatic cells were obtained in sufficient numbers for analysis. The limitation of these rice and Arabidopsis analyses was that all stages of meiosis were represented; even if there is little or no de novo transcription during meiosis, transcripts present initially in PMCs may have decayed during meiosis, and the mixed meiocyte population is hence not ideal for analysis. In these studies from 800 to 1,586 meiocyte-specific transcript types were identified. For maize, we used whole anther analysis and comparison to an independent dataset of laser microdissected PMCs and tapetal cells in carefully staged 1.5 mm fertile anthers at prophase I, and we also used the comparison of am1-489 and am1-praI to discover meiosis-associated genes. Prophase I spans only one day of the 30 days in maize anther development. From our analysis, 297 PMC-enriched genes were identified as mis-regulated by both mutants within the narrow window of prophase I. Additionally, we identified genes mis-regulated at the start of PMC maturation, particularly in the am1-489 mutant, that are likely associated with the suppression of mitosis and early preparation for meiosis.
Our am1 transcriptome data can serve as a resource to decipher the complex networks of meiosis initiation in plants. Comparing results from the two am1 alleles provides insights into multiple control points.