ABSTRACT
Mediator is a conserved transcriptional co-activator that links transcription factors bound at enhancer elements to RNA Polymerase II. Mediator-RNA Polymerase II interactions can be sterically hindered by the Cyclin Dependent Kinase 8 (CDK8) module, a submodule of Mediator that acts to repress transcription in response to discrete cellular and environmental cues. The CDK8 module is conserved in all eukaryotes and consists of 4 proteins: CDK8, CYCLIN C (CYCC), MED12, and MED13. In this study, we have characterized the CDK8 module of Mediator in maize. The maize genome contains single copy genes for Cdk8, CycC, and Med13, and two genes for Med12. Analysis of expression data for the CDK8 module demonstrated that all five genes are broadly expressed in maize tissues, with ZmMed12a, ZmMed12b, and ZmMed13 exhibiting similar expression patterns. We performed a Dissociation (Ds) insertional mutagenesis, recovering two independent insertions in the ZmMed12a gene. One of these Ds insertions results in a truncation of the ZmMed12a transcript. Our molecular characterization of the maize CDK8 module, as well as transposon tagging of ZmMed12a, establish the basis for molecular and functional studies of these important transcriptional regulators in Zea mays.
INTRODUCTION
Transcriptional regulation plays an essential role in almost all aspects of development and physiology, including responses to the biotic and abiotic environment. One key regulator of transcription is Mediator, a multiprotein complex conserved from yeast to plants to animals, which was initially identified based on its requirement for transcription of virtually all protein-coding genes (Kelleher et al., 1990; Flanagan et al., 1991; Bourbon, 2008). The Core Mediator consists of Head, Middle and Tail domains, and typically functions as a transcriptional co-activator, linking transcription factors bound at upstream enhancer elements to RNA polymerase II (RNA pol II) (reviewed in Yin and Wang, 2014; Allen and Taatjes, 2015). The Head and Middle domains interact with RNA pol II, while the Tail domain is thought to interact with specific transcription factors (Tsai et al., 2014; Robinson et al., 2015; Plaschka et al., 2015; reviewed in Larivière et al., 2012). A fourth Mediator module shows transient association with Core Mediator and often acts to repress transcription. This Cyclin Dependent Kinase 8 (CDK8) module is composed of the proteins MED12, MED13, CYCLIN C (CYCC), and CDK8 (reviewed in Björklund and Gustafsson, 2005). In agreement with the variable association of the CDK8 module with Core Mediator, purification of Mediator from Arabidopsis thaliana yielded both conserved Core Mediator subunits, as well as subunits unique to Arabidopsis, but did not include components of the CDK8 module (Bäckström et al., 2007).
In yeast and animals, components of the CDK8 module can regulate transcription in several ways, with different subunits playing different roles. One mechanism for transcriptional repression involves steric inhibition, where the CDK8 module occupies the Core Mediator pocket that binds RNA pol II, thereby preventing interaction of Core Mediator and RNA pol II (Elmlund et al., 2006; Tsai et al., 2013). Transcriptional repression by this steric mechanism has the potential to be dynamic, as the occupancy of the RNA pol II binding pocket can be modulated during subsequent rounds of assembly of the Mediator-RNA pol II holoenzyme (reviewed in Allen and Taatjes, 2015). This steric mechanism involves all four units of the CDK8 module, with the MED13 subunit playing the most important role, interacting directly with the Middle domain of Core Mediator (Knuesel et al., 2009; Tsai et al., 2013). The MED13 subunit also serves an important function in regulation of CDK8 module stability: phosphorylation of a conserved phosphodegron site in MED13 can lead to recognition by a ubiquitin ligase complex, and subsequent degradation (Davis et al., 2013).
In Arabidopsis, components of the CDK8 module were initially identified by their requirement for development, and also affect the response to fungal pathogens and cellular stress. Mutations in CDK8 were identified as enhancers of the phenotype of the floral homeotic mutant hua1hua2, and thus were named hua enhancer 3 (hen3). hen3 mutants affect floral organ identity, as well as leaf size and cell shape, and the HEN3 protein was demonstrated to have CDK8 kinase activity (Wang and Chen, 2004). CDK8 regulates retrograde signaling from the mitochondria to the nucleus in response to H2O2 and cold stress (Ng et al., 2013). CDK8, as well as MED12 and MED13, are also required for the response to both fungal and bacterial pathogens (Zhu et al., 2014).
Mutations in MED12 and MED13 were initially reported from a genetic screen for regulators of pattern formation in Arabidopsis embryogenesis, and were named center city (cct) and grand central (gct), to reflect the increased size of the shoot apical meristem (SAM) in these mutants. cct and gct mutants delay the timing of pattern formation during embryogenesis, rather than affecting pattern formation per se- the increased size of the SAM in cct and gct mutants can be attributed to its formation later in embryogenesis compared to the wild type (wt) (Gillmor et al., 2010). The delayed formation of the SAM may be related to auxin signaling, as both the med13 allele macchi-bou2 (mab2), and the med12 allele cryptic precocious (crp) act as enhancers of a mutation in the auxin dependent kinase PINOID (Furutani et al., 2004; Ito et al., 2011; Imura et al., 2012). Importantly for mechanistic studies of CDK8 module function in Arabidopsis, Ito et al. (2011) demonstrated that the MED13 and CDK8 proteins are both able to interact with Cyclin C, as has previously been demonstrated in Drosophila (Loncle et al., 2007). Consistent with studies showing auxin-related phenotypes for mutants in MED12 and MED13, a recent study showed that both of these genes, as well as CDK8, are involved in auxin transcriptional responses, and that the MED13 protein relays signals from the IAA14 protein to repress the auxin responsive transcription factors ARF7 and ARF19 (Ito et al., 2016).
In addition to affecting the timing of pattern formation in embryogenesis, MED12 and MED13 also regulate the timing of post-embryonic phase transitions in Arabidopsis. A dominant allele of med12 (named cryptic precocious (crp-1D)) was isolated in a genetic screen for enhancers of the early flowering phenotype conditioned by overexpression of the florigen FT (Imura et al., 2012). Loss of function mutants in crp/cct and gct show late flowering due to overexpression of the floral repressor FLOWERING LOCUS C (FLC), as well as decreased expression of the floral promoters FLOWERING LOCUS (FT), TWIN SISTER OF FT (TSF), SUPPRESSOR OF OVEREXPRESSION OF CONSTANTS 1 (SOC1), APETALA 1 (AP1) and FRUITFULL (FUL) (Imura et al., 2012; Gillmor et al., 2014). cct and gct mutants also misexpress seed specific genes during seedling development, and have an elongated vegetative phase due to overexpression of the microRNA miR156 (Gillmor et al., 2014), a master regulator of the vegetative phase in plants (Wu et al., 2009). Taken together, these results demonstrate that MED12 and MED13 act as master regulators of developmental timing in plants, regulating the timing of pattern formation in embryogenesis, the seed-to-seedling transition, vegetative phase change, and the transition to flowering (Gillmor et al., 2010; Ito et al., 2011; Imura et al., 2012; Gillmor et al., 2014).
Due to its importance in plant development and physiology, we have extended studies of the CDK8 module to the crop plant maize (Zea mays). Establishment of molecular and genetic resources for the study of the maize CDK8 module will allow evaluation of its role in the regulation of agricultural traits such as timing of flowering and seed development, as well as responses to biotic and abiotic stresses. One of the primary goals of this work was isolation of loss of function mutant alleles of maize CDK8 module-encoding genes. In maize, resources based on endogenous DNA transposons constitute the most accessible and widely-used technology for reverse genetics (McCarty and Meeley, 2009). The two major transposon systems used for gene tagging in maize are Activator/Dissociation (Ac/Ds) and Mutator (Mu) (Candela and Hake, 2008). These systems consist of an autonomous or master element that encodes a transposase (TPase) and a second non-autonomous or receptor element. The receptor elements are frequently derived from a master element by mutations within the TPase gene. Lacking TPase, non-autonomous elements are stable, unless mobilized by TPase supplied in trans by an autonomous element (Kunze et al., 1997). Ac is a member of the hAT transposon superfamily (named after the founding members hobo, Ac and Tam3; Calvi et al., 1991) and moves via a cut-and-paste mechanism (Bai et al. 2007), with a preference for transposition to linked sites, making the system ideal for local mutagenesis (Greenblatt, 1984; Dooner and Belachew, 1989; Brutnell and Conrad, 2003). To exploit the Ac/Ds system for reverse genetics, Ds elements have been distributed throughout the genome to provide potential “launch pads” for mutagenesis of nearby genes (Vollbrecht et al. 2010).
In this study, we identify five genes encoding components of the CDK8 module in maize, present experimentally determined gene structures, and report expression of corresponding transcripts. We performed Ds mutagenesis of the gene ZmMed12a, identifying two novel insertional alleles, one of which results in a truncation of the ZmMed12a transcript. These insertional mutant alleles will enable determination of the biological roles of the CDK8 module in maize development and stress responses.
MATERIALS AND METHODS
Identification of maize CDK8 module genes
Maize CDK8 module genes were identified by BLAST searches using the predicted Arabidopsis thaliana protein sequences for HEN3/CDK8 (AT5G63610), CYCC1;1 (At5g48640), CCT/MED12 (At4g00450), and GCT/MED13 (At1g55325) available at TAIR (www.arabidopsis.org). Reciprocal BLAST searches were conducted between all maize and Arabidopsis sequences, to establish that the five maize genes ZmCDK8, ZmCycC, ZmMed12a, ZmMed12b, and ZmMed13 were the only full length CDK8 homologs present in maize.
Determination of coding sequences for ZmCDK8, ZmCycC, ZmMed12a, ZmMed12b, and ZmMed13
Multiple mRNA sequences with full-length coding sequences (as well as upstream and downstream untranslated regions) were identified from the NCBI database for both ZmCDK8 and ZmCycC. For CDK8, cDNAs for two alternative splice products were identified: EU968864, NM_001157457 and BT018448 correspond to one splice variant, and BT039744 and XR_552425 correspond to the other splice variant. For CycC, three independent cDNAs (BT040922, BT033427, and XM008652706) were identified for the one splice variant (shown in Figure 1). Two independent cDNAs (AY105730 and EU972675) represented another CycC splice variant with an identical coding sequence but with slight differences in the 3’UTR. A third splice variant was represented by a single cDNA (BT036293); this mRNA has two upstream ORFs, and encodes a truncated CycC protein. For ZmMed12a, ZmMed12b and ZmMed13, partial sequences were obtained from the maize database (maizegdb.org), which were then confirmed and extended by RT-PCR using RNA extracted from seedlings of the B73 inbred line. To confirm the ZmMed12a, ZmMed12b, and ZmMed13 gene models, we amplified cDNA products covering the entire predicted coding regions. Given their large expected size, ZmMed12a, ZmMed12b, and ZmMed13 cDNAs were amplified in multiple over-lapping fragments. Sequencing of cDNA products was generally consistent with gene models based on genomic sequence analysis, except in the case of ZmMed13, where a large intron not present in the maize genome sequence was discovered. Coding sequences were deposited in the NCBI database with the following accession numbers: ZmMed12a (KP455660), ZmMed12b (KP455661), and ZmMed13 (KP455662).
In addition, numerous short genes that are predicted to encode highly truncated ZmMed12 proteins of 199 to 431 residues were identified (Núñez-Ríos, 2012). These short ZmMed12 genes are predicted to encode the Med12 domain (pfam09497) and many have corresponding expressed sequence tags (EST) (B73 RefGen_v3), which do not cover the entire body of these short genes. Analysis of genomic sequences around these predicted coding sequences did not identify additional Med12 exons (data not shown), suggesting that these are indeed truncated versions of ZmMed12, and not mis-annotated genes with nearby exons that would constitute the middle and C-terminal portions of Med12 proteins.
Expression profiles of maize CDK8 module genes
Expression data from 22 maize tissues were obtained from http://qteller.com/qteller3/ on August 2014, in the form of Fragments Per Kilobase of transcript per Million (FPKM). In order to look for correlations between pairs of genes across the tissues, the data was log2 transformed (first adding 1, to avoid the logarithm of 0) and normalized using the normalizeQuantiles function from the limma package (Bolstad et al., 2003).
The expression values were selected for the 5 CDK8 module genes: CDK8 (GRMZM2G166771), CycC (GRMZM2G408242), Med12a (GRMZM2G114459), Med12b1 (GRMZM5G828278), Med12b.2 (GRMZM5G844080), Med13.1 (GRMZM2G053588), and Med13.2 (GRMZM2G153792). Since Med12b.1 and Med12b.2 as well as Med13.1 and Med13.2 are spliced versions of the same gene, the geometric mean was calculated to obtain an averaged estimate of their expression. These data were employed to produce Figure 2A, using the heatmap.2 function from the gplots package (Warns et al., 2015). All pair-wise combinations of the 5 genes across all tissues were plotted using the generic plot function in R (R Core Team, 2015) (Figure S5). The Pearson correlations for all possible pairs of genes were calculated with the cor function, and these data were used as the empirical null to calculate p-values. Correlations for CDK8 module genes were calculated separately. The blob plot in Figure 2B was generated with the corrplot for R.
Description of maize stocks
All stocks were maintained in the common genetic background of a color-converted W22 inbred line (Dooner & Kermicle, 1971). A stable source of Ac transposase was provided by Ac-immobilized (Ac-im), an Ac derivative which has lost 10bp at the 5’ end of the element, preventing excision (Conrad and Brutnell, 2005). Activity of Ac transposase was monitored using the mutable Ds reporter r1-sc:m3 that carries a Ds6-like insertion in the r1 locus that controls anthocyanin production in the aleurone and scutellum tissues (Alleman and Kermicle, 1993): when Ac transposase is present, excision of Ds from r1 restores gene function producing colored sectors (Brutnell & Dellaporta, 1994). The donor Ds (dDs) stock dDs-B.S07.0835 was generated by isolation of novel transpositions from r1-sc:m3 as previously described (Vollbrecht et al., 2010). Presence of dDs-B.S07.0835 was assayed by PCR as previously described (Vollbrecht et al., 2010) using a combination of the Ds end primer JSR05 and a primer specific to the genomic site of B.S07.0835 (5’- GACGCACACACGTCAGTATAG-3’). To generate the test-cross population, plants verified as carrying the donor dDs-B.S07.0835 with Ac-im in the genetic background were used as males to pollinate r1-sc:m3/r1-sc:m3 female plants.
Seedling screen for transposon insertions in ZmMed12a
Testcross progeny were germinated and screened for novel insertions of Ds in ZmMed12a using a PCR-based strategy. Tissue was collected between 7 and 10 days after planting from pools of 10-18 seedlings using a ≈3mm hole punch, and DNA was isolated following a CTAB-based extraction protocol (Weigel and Glazebrook, 2009). A total of 10 ZmMed12a gene-specific primers were designed, covering a region extending from 1.8kb upstream of the translational start to the stop codon. These were used in conjunction with the 5’ and 3’ Ds-end primers JSR01 and JGp3, respectively, to amplify DNA adjacent to novel Ds insertions in ZmMed12a (Table 1). Pools amplifying a product were de-convoluted by screening individuals separately; this second round of PCR used DNA extracted from a different seedling leaf than that sampled for the pool to reduce the chances of recovering somatic transposition events. The PCR products of the second PCR were cleaned (Sambrook and Russell, 2006) and the DNA concentration was adjusted for sequencing by the GENEWIZ Company (South Plainfield, New Jersey, USA). Seedlings carrying putative med12a insertional alleles were grown to maturity and propagated by both self-pollination and out-crossing to W22 and B73 inbred lines.
RT-PCR analysis of zmmed12a-1::Ds and zmmed12a-2::Ds alleles
DNA was extracted from 10 day old greenhouse grown seedlings of F2 populations segregating the 1::Ds and 2::Ds insertions. Seedlings were genotyped using primers to identify homozygous wild type and homozygous insertion alleles for 1::Ds (primer pair A5.12F and A5.12R for wild type and A5.12F and JGp3 for Ds insertion) and 2::Ds (primer pair C2.7F and C2.7R for wild type allele and C2.7R and JGp3 for Ds insertion). RNA was then extracted Trizol (Invitrogen) for wild type and homozygous insertion alleles. Reverse transcription was performed with SuperScript II (Invitrogen). PCR was performed with the following programs, using Kapa Taq Polymerase (Kapa Biosystems). TNC4-TNC5 primer pair: initial denaturation 95 °C 5′; 10 cycles of 95 °C 30″, 60 °C 30″ (-0.5 °C per cycle), 72 °C 45″; 27 cycles of 95 °C 30″, 55 °C 30″, 72 °C 45″; final extension 72 °C 5′. RS170- RS167 and ZmCDK primer pairs: initial denaturation 95 °C 5′; 30 cycles of 95 °C 30″, 60 °C 30″, 72 °C 1′; 72 °C 10′.
RESULTS
The maize genome encodes all four components of the CDK8 module of Mediator
A previous effort to identify Mediator genes from many plant species identified a single maize homolog for all four CDK8 module genes (CDK8, CYCC, MED12 and MED13) (Mathur et al., 2011). In order to conclusively define the number and identity of CDK8 module homologs in maize, we performed BLAST searches to identify all maize gene models (B73 reference genome v3; www.maizesequence.org) whose putative protein products exhibit a high degree of similarity to the entire predicted Arabidopsis proteins of the CDK8 module of Mediator: CDK8 (encoded by HEN3) (Wang and Chen, 2004); CYCC1;1 or CYCc1;2 (Wang et al., 2004); MED12 (encoded by CCT/CRP) (Gillmor et al., 2010; Imura et al., 2012); and MED13 (encoded by GCT/MAB2) (Gillmor et al., 2010; Ito et al., 2011) (Table 2). Using the translated experimentally verified coding sequences for all maize CDK8 module genes (see below), all potential orthologous relationships were further validated by reciprocal searching of the Arabidopsis genome using maize sequences, and by inspection of the next-best-hit in both Arabidopsis-to-maize and maize-to-Arabidopsis searches (data not shown).
A single maize gene (GRMZM2G166771) was identified as a potential ortholog of HEN3/CDK8, and designated ZmCDK8. Two different full-length splice products were identified for this gene (EU968864 and BT039744), predicted to encode a full-length and a truncated maize CDK8 protein (Figure 1A; Figure S1). The full-length ZmCDK8 protein is 471 amino acids (AA), and shows 73% identity with the 470 AA Arabidopsis CDK8 protein, and 43% identity with the 464 AA human CDK8 protein (Figure S1). The smaller ZmCDK8 protein is 385 AA, primarily because of a truncation of the C terminal domain, and shows 75% identity with Arabidopsis CDK8, and 43% identity with human CDK8. This truncation occurs after the CDK8 kinase catalytic domain (cd07842), and is thus unlikely to interfere with the kinase function of the protein (Figure S1).
Although Arabidopsis CYCC is encoded by a tandem-duplicated gene pair (Wang et al., 2004), a single potential maize ortholog of CYCC (GRMZM2G408242) was identified, and designated ZmCycC. Figure 1B shows the splice product represented by the full-length cDNA clone BT040922 (Figure 1B). The 257AA BT040922 protein is 42% identical to human CycC and 67% identical to Arabidopsis CycC1;1 (Figure S2), and contains the Cyclin domain (cd00043) that is present in human and Arabidopsis CycC (Figure S2).
BLAST searches using the Arabidopsis CCT/MED12 protein identified two putative full- length maize genes (GRMZM2G114459 on chromosome 1, and the split gene GRMZM5G828278 / GRMZM5G844080 on chromosome 9), which were designated ZmMed12a and ZmMed12b. Partial cDNA sequences were publicly available for ZmMED12a and ZmMed12b; these sequences, as well as coding sequences predicted by the maize database, were used to experimentally determine mRNA sequences for both genes by RT- PCR. The exon-intron structure of both genes is very similar, with the only differences occurring in the length and position of exons 2, 3 and 4 (Figure 1C&D). These splicing differences lead to several small insertions or deletions in the N-terminal portions of the ZmMed12 proteins, with ZmMed12a encoding a protein of 2193AA, and ZmMed12b encoding a protein of 2202AA; the two ZmMed12 proteins are 91% identical (Figure S3). ZmMed12a is 19% identical to human Med12, and 46% identical to Arabidopsis MED12; ZmMed12b is 20% identical to human Med12, and 46% identical to Arabidopsis MED12 (Figure S3). The region of highest identity is that comprising the Med12 domain (pfam09497), located at the N-terminus of the Med12 proteins (Figure S3).
A single maize gene was identified corresponding to GCT/MED13 (split gene GRMZM2G053588 / GRMZM2G153792), and designated ZmMed13. Partial cDNA sequences were publicly available for ZmMed13; these sequences were used as the basis for RT-PCR experiments to identify full-length mRNA and coding sequences, which demonstrated that ZmMed13 encodes a protein of 1892 AA, with 20% identity to human Med13, and 49% identity to Arabidopsis MED13 (Figure 1E & Figure S4).
Maize CDK8 module genes are expressed throughout development
In other organisms where the CDK8 module has been studied, the gene pairs CDK8 and CyclinC; and Med12 and Med13, have similar expression patterns and mutant phenotypes (Yoda et al., 2005; Loncle et al., 2007; Gillmor et al., 2010; Gillmor et al., 2014). In order to determine whether the CDK8 / CycC and Med12 / Med13 genes have similar expression patterns in maize, we used publicly available RNA sequence data to quantify CDK8 module gene expression in different tissues and at different developmental stages (see Materials and Methods). As seen in the heatmap in Figure 2A, CycC was expressed at much higher levels in all tissues than the other CDK8 module genes, with CDK8 and Med12a the next highest expressed genes, and Med13 and Med12b with the lowest expression levels
In order to more precisely compare tissue-specific expression between the different CDK8 module genes, we made pairwise comparisons for all five genes (Figure 2B & Figure S5). Expression was most highly correlated for Med13 and Med12b (Pearson’s r = 0.93), where the expression ratio between the two genes was close to 1 (compare dotted red line for r, with solid black line representing a 1:1 expression ratio) (Figure 2B & Figure S5). Med12a and Med12b (r = 0.77); Med12a and Med13 (r = 0.7); and CDK8 and Med12a (r = 0.76) also had high Pearson’s coefficients for pairwise comparisons (Figure 2B & Figure S5). By contrast, CycC showed almost no correlation with any of the other CDK8 module genes (Figure 2B & Figure S5). The fact that CycC shows little expression correlation with the other CDK8 module genes, and is expressed at higher levels than CDK8, and many times higher than Med13, Med12a and Med12b, suggests that CycC may play more varied roles in development and physiology than the other CDK8 module genes.
Maize Med12 is encoded by the duplicated gene pair ZmMed12a and ZmMed12b
The high degree of similarity between ZmMed12a and ZmMed12b suggests that they are the result of a recent duplication event (Figure S6). ZmMed12a and ZmMed12b are located in homologous regions of the genome (1S and 9L, respectively), which derive from a polyploidy event that occurred 5-12 million years ago, sometime after the divergence of maize and sorghum lineages. Although gene loss has reduced the number of genes in present-day maize close to pre-duplication levels, in certain cases both syntenic paralogs have been retained (Schnable et al., 2011). Further inspection revealed a sorghumMed12 gene (Sb01g050260; SbMed12) to be present in a region on Chromosome 1L syntenic to the two maize ZmMed12 containing regions. Moving up- and downstream from SbMed12, micro-synteny was conserved, although, typically, for any given sorghum gene only one candidate ortholog was identified in maize, in either the 1S or 9L region, presumably as the result of gene-loss within paralog pairs following whole genome duplication (Fig. 3).
Reverse genetics strategies to target maize CDK8 components
To initiate functional analysis of the maize CDK8 module, we identified publicly available seed stocks carrying Ac/Ds or Mu family transposons inserted into, or close to, maize CDK8 module encoding genes (Table 3). On the basis of this search, we selected ZmMed12a as our first target for reverse genetics: at ˜56kb, the closest potential Ds donor was nearer to ZmMed12a than to any of the other genes. In addition, the availability of a well-characterized med12 mutant in Arabidopsis provides possibility for comparative study (Gillmor et al., 2010; Imura et al., 2012; Gillmor et al., 2014). Finally, the retention of two Med12 syntenic paralogs in maize suggests that the roles of ZmMed12a and ZmMed12b are functionally different, a question which can be addressed by characterization of maize med12 mutant alleles.
Identification of novel Ds insertions into ZmMed12a
To use the Ac/Ds transposon system to generate mutant alleles of ZmMed12a, we first obtained donor Ds (dDs) stocks carrying the Ds element dDs-B.S07.0835, located 56.2 kb from ZmMed12a (acdstagging.org). The position of the linked Ds element was confirmed by PCR assay (see Materials and Methods) (Conrad and Brutnell, 2005). Presence of Ac-im in testcross progenitor seed stocks was monitored by somatic excision of a second Ds from the r1-sc:m3 marker locus, resulting in variegated spotting of the kernel aleurone and scutellar tissues (Figure 4A & B). Spotted kernels were planted and seedlings genotyped for the presence of dDs using a PCR assay (Materials and Methods). To generate novel germinal insertions into ZmMed12a, individuals carrying the dD and the Ac-im transposase source were used as males to pollinate T43 (r-sc:m3/r-sc:m3) females. A test cross population of 59 ears was obtained for the ZmMed12a screen (Figure 4A&B).
The test-cross population was screened for Ds insertions in ZmMed12a using combinations of gene specific and Ds specific PCR primers (see Materials and Methods). Pools of 10-18 seedlings were assayed for amplification of putative Ds-flanking junction products (see Figure 4C for example for the Zmmed12a-2::Ds insertion). Seedlings constituting the pools from which products were amplified were re-screened separately to identify positive individuals (Figure 4D). This second PCR was performed using DNA extracted from a leaf different from that used for the pool PCR to reduce the rate at which we recovered somatic transposition events. We screened a total of 3,049 seedlings and identified two novel insertions into ZmMed12a: zmmed12a-1::Ds, located 918bp upstream of the translational start, and zmmed12a-2::Ds located in exon 10 (Figure 1C). We performed additional PCR reactions to recover both flanks of the zmmed12a-1::Ds and zmmed12a-2::Ds insertions. Flanking DNA products were sequenced, confirming the location of the insertions and identifying characteristic 8bp target site duplications. The seedlings carrying the two novel zmmed12a insertional alleles were grown to maturity and propagated by both self-pollination and out-crossing. Progeny were germinated and genotyped, confirming the heritability of novel Ds insertions (Figure 4E).
The 2::Ds insertion results in a truncated ZmMed12a transcript
In order to determine the effect of these novel Ds insertions on the ZmMed12a gene, we performed RT-PCR analysis of plants homozygous for the wild type and Ds alleles, using primer pairs that amplify fragments in exon 9 (downstream of the 1::Ds insertion, and upstrearm of the 2::Ds insertion), and exon 12 (downstream of both Ds insertions) (Figure 4F). Both primer sets amplified fragments from wild type and the zmmed12a-1::Ds allele, suggesting that the 1::Ds insertion has no significant effect on the ZmMed12a transcript, a result which is not surprising, since this Ds insertion is upstream of exon 1. In the case of the zmmed12a-2::Ds allele, the primer pair in exon 9 produced an amplification product, while the primer pair in exon 12 showed no aplification from homozygous 2::Ds plants. This result demonstrates that the 2::Ds insertion causes production of a truncated version of the ZmMed12a transcript, likely causing a loss of function of the ZmMed12a gene.
DISCUSSION
In this study we have identified the five genes encoding the CDK8 module of Mediator in maize, determined their coding sequences, characterized their expression in maize tissues during development, and examined the synteny of maize and sorghum in the region of the Med12 genes. Additionally, we have mutagenized the ZmMed12a gene using the Ac/Ds transposon system created by Vollbrecht et al. (2010).
In our analysis of CDK8 module genes, we identified two alternative transcripts for CDK8 (Figure 1). One predicted CDK8 protein is significantly shorter than the other, lacking the C-terminal 86 AA. This truncation seems unlikely to affect enzyme activity per se, as the kinase domain is intact (Figure S1). However, the lack of this domain may alter regulation of the kinase activity. Alternatively, the truncation may modify the interaction of CDK8 with CycC, or affect the formation of the four protein CDK8 complex. This complex sterically inhibits the interaction of Core Mediator with RNA pol II, by making direct contact with Core Mediator (Tsai et al., 2013). In the case of CycC, one only one isoform was represented by multiple independent cDNAs. Only single splice products were identified for Med12a, Med12b and Med13 (Figure 1). One explanation for this is that there is indeed only one splice product for each gene in maize. It is also possible that the very large size of the mRNAs for these three genes (6-7 kb) makes cloning of multiple splice products difficult, due to technical difficulties in cloning such large cDNAs.
In our analysis of the relative expression of CDK8 module genes, we found CDK8 and CycC to be more highly expressed in all tissues than Med12a, Med12b or Med13. In particular, CycC showed the highest expression in all tissues, consistently 3-4 times higher even than CDK8 (Figure 2). This increased expression of CycC is consistent with roles of CycC beyond regulating transcription in tandem with CDK8 (the best known role for CycC) (Allen and Taatjes, 2015). In addition to regulation of transcription, CycC has been shown to promote the G0 to G1 cell cycle transition through phosphorylation of Retinoblastoma, allowing quiescent cells to enter the cell cycle. CycC achieves this through interaction with CDK3, a kinase that is not associated with transcriptional activation, but instead promotes cell cycle entry (Ren and Rollins, 2004). CycC has also been demonstrated to be a haploinsuficient tumor suppressor in mammals, whose loss of function in mice is lethal during embryogenesis (Li et al., 2014). The haploinsuficiency of CycC may require its mRNA or protein levels to be stably maintained, suggesting an explanation for its high levels in all the tissues that we examined (Figure 2 & Figure S6). Med12a, Med12b, and Med13 show much lower expression levels, which also vary considerably between different tissues (Figure 2 & Figure S6). The similar expression profiles for Med12 and Med13 in maize are consistent with Arabidopsis, where similar expression profiles for these two genes were reported (Gillmor et al., 2010; Ito et al., 2011; Imura et al., 2012; Gillmor et al., 2014). The widely varying expression levels for Med12 and Med13 in different tissues are consistent with various roles for these genes in development, both in primordia (where they show the highest expression), as well as in differentiating and more mature tissue.
In Arabidopsis, MED12 is a single copy gene, with mutant phenotypes in both development and pathogen responses (Gillmor et al., 2010; Imura et al., 2012; Gillmor et al., 2014; Zhu et al., 2014). In maize, however, two Med12 genes were identified. Sometime after divergence with sorghum, the maize lineage underwent whole genome duplication (Schnable et al., 2011). While in the majority of cases resulting additional gene copies have been lost, for ˜10% of the original gene set syntenic paralog pairs have been retained (Hughes et al., 2015). The genomic location of ZmMed12a and ZmMed12b is consistent with them representing such a paralog pair. In the region of synteny between maize and sorghum, other genes surrounding Med12 have been reduced to a single copy, suggesting that the retention of both paralogs of Med12 in maize may have functional significance. Our isolation of the 2::Ds insertional allele of ZmMed12a will allow us to test the functional importance of this gene. The truncation of the ZmMed12a transcript in the 2::Ds allele makes it very likely that this allele causes a loss of function: a T-DNA insertion in a similar location of the CCT (MED12) gene of Arabidopsis causes a strong loss of function phenotype, even when some aberrant transcript is produced (Gillmor et al., 2010; Gillmor et al., 2014)
One additional advantage of Ds as a mutagen is that novel transpositions occur into linked sites, meaning that the Ds insertions in ZmMed12a can be remobilized to create further allelic variation in ZmMed12a. In addition to mutant alleles that cause a complete loss of function, subsequent Ds mutagenesis of ZmMed12a may result in hypomorphic alleles that either reduce (but do not eliminate) the function of ZmMed12a, or that inactivate specific functional domains of Med12. Alleles that eliminate only certain parts of the Med12 protein could be especially useful in understanding the function of different domains of Med12, currently one of the most interesting, and least explored, aspects of Mediator biology.
AUTHOR CONTRIBUTIONS
Study designed by TN-R, KA, TPB, SG and RS. Data acquired and/or analyzed by TN-R, KA, ALA-N, DL-S, CM-C, MG-A, SG, RS. Manuscript written by TN-R, SG and RS, and approved by all authors.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.
FUNDING
This study was funded by CINVESTAV institutional funds to CSG and RJHS, CONACyT CB 2009 No. 133990 to MGA, CONACyT CB 2012 No. 151947 to RS, and NSF IOS-0922701 to TPB. CONACyT graduate fellowships supported TN-R, ALA-N (No. 339468), DL-S (No. 262808), and CM-C (No. 20729).
ACKNOWLEDGEMENTS
Thanks to Cei Abreu-Goodger for advice on analysis of gene expression data, and to Jessica Carcaño-Macías for managing seed stocks.