Genome

There are 112 C2H2 type ZFPs listed in the Medicago transcription factor database (http://planttfdb.cbi.pku.edu.cn/index.php?sp=Mtr). Given the large number of C2H2 ZFPs reported in other species [15, 16, 27], for instance, 176 and 321 C2H2 ZFPs in the Arabidopsis and soybean genome, respectively [16, 27], we thought that there should be more C2H2 ZFPs in the M. truncatula genome. To date, very few C2H2 ZFPs have been reported in M. truncatula. Thus, a literature survey did not return much useful information regarding the C2H2 ZFP family in M. truncatula. To extensively identify C2H2 ZFPs, we employed a combined strategy and searched the genome databases MtrunA17r5.0 (https://medicago.toulouse.inra.fr/MtrunA17r5.0-ANR) and Mt4.0v2 (http://www.medicagogenome.org) [31,32,33]. Based on the reported C2H2 motif sequences pattern, we developed a custom script using the Regular Expression module in the programming language Python. As a result, 272 candidate C2H2 ZFPs containing 381 C2H2 motifs were found. To eliminate potential false positives, the candidates were further scanned in the Prosite protein database and PFAM platforms for confirmation. The candidates that passed the C2H2 motif scan in Prosite and PFAM platforms or showed significant similarity to Arabidopsis or rice C2H2 ZFPs were considered to be M. truncatula C2H2 type ZFPs. This step eliminated 55 candidates and resulted in 217 C2H2 ZFPs containing 328 C2H2 motifs in M. truncatula (Fig. 1, Additional file 1, Additional file 2 and Additional file 3).

MapChart diagram illustrating the distribution of C2H2 ZFPs in the M. truncatula genome. The black lines within the chromosomes indicate C2H2 ZFPs. The line linking two C2H2 ZFPs represents a pair of C2H2 ZFPs resulting from WGD. Different colors of the line and gene ID indicate different pairs of C2H2 ZFPs from WGD

TFIIIA is the ZFP that contain the most C2H2 motifs in plant species [14, 16, 34]. An initial examination of the identified 217 ZFPs found that TFIIIA was not present in the M. truncatula latest genome (Jemalong A17). We doubt that it may be caused by sequencing or annotation error. Using the genome sequence of another cultivar R108 (http://www.medicagohapmap.org/downloads/r108) and Affymetrix microarray probeset data, we identified and manually sequenced TFIIIA homolog in M. truncatula (described below).

In summary, we identified a total of 218 C2H2 ZFPs containing 337 C2H2 motifs (Fig. 1, Additional file 1, Additional file 2 and Additional file 3), far exceeding the genes listed in the plant transcription factor database. The C2H2 ZFPs identified in M. truncatula displayed variations in length and molecular weight, ranging from less than 100 to more than 1700 amino acids, and 7.3 to 197 kilodaltons (kDa), respectively. However, a large portion of the family (182 of the total of 218 ZFPs) was quite short (less than 500 amino acids). The identified C2H2 ZFPs also showed a wide range of the isoelectric point (PI) from 4.14 to 9.92, indicating the diversified physicochemical properties of the proteins. Many C2H2 ZFPs have splice variants, further adding the protein diversity of the gene family (Additional file 1).

According to the MtrunA17r5.0 and Mt4.0v2 genome databases, all C2H2 ZFPs are found on the chromosomes. Chromosome 1 harbored the most (59 C2H2 ZFPs), representing approximately 27% of the whole family, followed by chromosome 7 (36 C2H2 ZFPs) and chromosome 4 (35 C2H2 ZFPs). Chromosome 6 and chromosome 8 contained only 3 and 14 C2H2 ZFPs, respectively (Fig. 1, Additional file 4).

In M. truncatula C2H2 ZFP gene family, 46 local duplications containing two to six C2H2 ZFPs were found (Additional file 5). For instance, the neighboring genes MtrunA17_Chr3g0086321/Medtr3g023750 and MtrunA17_Chr3g0086331/Medtr3g023760 both encode C2H2 type ZFPs; a 6.5 kb fragment on chromosome 1 from 14.30 M to 14.31 M contains three C2H2 type ZFPs: MtrunA17_Chr1g0164501, MtrunA17_Chr1g0164511 and MtrunA17_Chr1g0164521 (Additional file 5). The local duplication of C2H2 ZFPs significantly increased the total number of C2H2 ZFPs in M. truncatula, and provided the potential for the subfunctionalization and diversification of C2H2 ZFPs.

As a papilionoids species, M. truncatula underwent a whole-genome duplication (WGD) 58 million years ago (Myr) [35,36,37]. WGD gave rise to gene pairs, which are located in syntenic genome regions and have been detected by genome-level analysis [35]. By gene position analysis, MCScanX analysis, and comparison with a previous dataset [38], we found that 4 pairs of C2H2 ZFPs derived from WGD (Fig. 1, Additional file 6). In agreement with the duplication, the C2H2 motif type and arrangement were identical between the two paralogs from the gene pairs (Additional file 6).

To gain basic phylogenetic insights into the relationship of C2H2 ZFPs in M. truncatula, we attempted to perform sequence alignment and reconstruct the phylogenetic tree for the C2H2 ZFPs. However, the sequence similarity analysis indicated that, except for close homologs or subsets such as MtrunA17_Chr8R0156030/Medtr8g466760 and MtrunA17_Chr3g0081131/Medtr3g011990 that share a significant similarity, members of the C2H2 ZFP family are rather different in sequence and structure, as less than 15% similarity was detected for more than 97% of genes, suggesting the functional diversification of C2H2 ZFPs (Additional file 7).

The Arabidopsis homolog of TFIIIA has been previously reported to regulate the transcription of 5S RNA genes [16, 34]. The M. truncatula TFIIIA homolog was not found by BLAST search using Arabidopsis TFIIIA as the query protein against both Jemalong A17 Mt4.0v2 and MtrunA17r5.0 genome sequence, probably because that the M. truncatula TFIIIA gene model has not yet been predicted in the genome sequence. A BLAST search against the M. truncatula Affymetrix Microarray probesets found that probeset Mtr.14440.1.S1_at, which corresponds to a genome region located on scaffold MWMB01000024 of M. truncatula R108 genome (http://www.medicagohapmap.org/downloads/r108), is highly similar to AtTFIIIA, suggesting that the TFIIIA homolog was also present in the M. truncatula R108 genome. BLAST searches in other plant species also revealed TFIIIA homologs in other plant genomes. To confirm the existence of TFIIIA in Medicago R108 genome, we amplified and sequenced the coding region of M. truncatula TFIIIA homolog. The identified M. truncatula TFIIIA homolog is consistent with the Hidden Markov Model (HMM) profile built with TFIIIA homologs from other plants by HMMER (http://hmmer.org/), confirming the identification of TFIIIA homolog in M. truncatula. Gene sequences analysis indicated that, similar to Arabidopsis TFIIIA, M. truncatula and other plant TFIIIA homologs consisted of nine C2H2 zinc fingers (Fig. 2a). Phylogenetic and sequence similarity analysis showed that all plant TFIIIAs shared a high sequence identity (Fig. 2a), suggesting a conserved function in regulating the transcription of 5S RNA genes. In Arabidopsis TFIIIA, the 2–4 and 5–9 fingers were arranged in two tandem finger arrays, whereas the first C2H2 zinc finger was relatively separated at the N-terminus. This pattern was also observed in other plant TFIIIA homologs, though the link sequence between the first and second finger array was a little shorter, especially in monocot plants (Fig. 2a). Nevertheless, this similar motif arrangement pattern suggests a conserved evolutional ancestor of plant TFIIIAs.

Phylogenetic analysis and motif arrangement of TFIIIA and TRM1 homologs. The top part of (a) and (b) were the unrooted maximum-likelihood tree of TFIIIA and TRM1 homologs, respectively. The blue and green branches indicated monocot and dicot plants respectively. The bottom part of (a) and (b) are motif arrangements of TFIIIA and TRM1, respectively. The black line represents the protein sequence. The black rectangle indicates the C2H2 motif, and the red rectangle indicates the EAR motif

TRM1 was first cloned and characterized in the C4 plant maize (Zea mays), in which TRM1 binds to the rbcS-m3 gene and plays a transcriptional repressor role [39]. The TRM1 homolog has also been identified in Arabidopsis [16]. Like TFIIIA, TRM1 was not initially found by BLAST search in the M. truncatula Mt4.0v2 genome database using either the Arabidopsis or maize TRM1 protein sequence as the query. However, a BLAST search against M. truncatula MtrunA17r5.0 genome database found that MtrunA17_Chr2g0277381 shows high similarity to Arabidopsis TRM1. We found that an Affymetrix probeset Mtr.25069.1.S1_s_at showed high similarity to TRM1 protein, suggesting that TRM1 was also present in the M. truncatula genome. To confirm the existence of TRM1 in M. truncatula, we amplified and sequenced the M. truncatula TRM1 homolog. Like maize TRM1, the M. truncatula TRM1 homolog contained five C2H2 motifs (Fig. 2b). The C2H2 motifs of TRM1 could be separated into two classes based on the linker length (Fig. 2b). The first four motifs formed a tandem repeat separated by 29 or 30 linker residues. The fifth motif was 65 residues away from the fourth motif (Fig. 2b). HMM analysis indicated that the M. truncatula TRM1 is consistent with the HMM profile built with known TRM1 homologs by HMMER. A sequence search in other plant species also identified putative TRM1 homologs that shared a high sequence identity and conserved phylogenetic relationship, suggesting a conserved role of TRM1 in plant species (Fig. 2b). Interestingly, the N-terminus of TRM1 from monocot species was a little shorter than its counterparts from dicot species. In rice, maize, and sorghum, the N-terminus, which started from the first residue to the first C2H2 motif, was 61 or 63 amino acids in length, whereas it ranged from 76 to 80 amino acids in dicot plants (Fig. 2b).

Surprisingly, a putative EAR motif-like sequence, ‘LKLHLK’, was found within the fifth C2H2 motif of M. truncatula TRM1 and TRM1 homologs in other plant species, including Arabidopsis, soybean, rice, maize, and sorghum (Fig. 2b).

The 218 C2H2 type ZFPs identified in the M. truncatula genome contained 337 individual C2H2 motifs. We aligned M. truncatula C2H2 motifs and classified them into five major groups based on the conserved core sequences within the C2H2 domains (Additional file 8).

Q-type C2H2s were found in 93 motifs from 71 genes. Q-type C2H2 had the core sequence ‘QALGGH’, where ‘H’ is the first histidine of the C2H2 motif. This type of C2H2 zinc finger has been previously reported to be specific to plants [1, 12, 15, 16, 25]. In addition to the core sequence, the three residues between the two histidines also showed the high conservation in certain Q-type motifs. Regarding the residues between the two histidines, we further subclassified Q-type C2H2s into six subgroups, including 28 motifs with the consensus ‘QNA’ between the two histidines (Q1 subgroup), 22 motifs with ‘(K/R)AS’ (Q2 subgroup), 12 motifs with ‘MR(R/K)’ (Q3 subgroup), 10 motifs with ‘MN(I/V)’ (Q4 subgroup), 10 motifs with ‘KR(C/S)’ (Q5 subgroup), and 11 motifs without the consensus sequence between the two histidines (Q6 subgroup) (Additional file 8).

In some C2H2 motifs, the core sequences of the Q-type motif were slightly or moderately modified, such as ‘RALGGH’ or ‘QGLGGH’. These C2H2s have also been reported in other plant species, and herein were designated as the Q-Modified (QM) type. In the M. truncatula genome, 57 QM-type C2H2 motifs were identified from 45 ZFPs. Based on the pattern of the core sequences and the methods used in a previous report [15], we classified the QM-type C2H2 motifs into nine subgroups (Additional file 8).

In nineteen ZFPs, a new type of conserved C2H2 zinc motif was found. Starting with phenylalanine (F), this type motif was even more conserved than the Q-type. Of a total of 23 amino acids, 16 were identical across all the genes (Fig. 3, Additional file 8). In Arabidopsis, this type of C2H2s was found in indeterminate-domain (IDD) genes, such as IDD10/JKD (AT5G03150), BALDIBIS (AT3G45260), IDD5 (AT2G02070) and IDD7 (AT1G55110). The first IDD gene ID1 was cloned in maize where it controlled the transition to reproductive growth [40]. The IDD family in plants consists of multiple members, with different roles in regulating plant growth or metabolism. Sequence comparison indicated that M. truncatula ZFPs containing this type of C2H2s demonstrated high similarity to Arabidopsis IDDs. For instance, MtrunA17_Chr8g0342421/Medtr8g017210, which contains this type of C2H2 zinc finger, shows a high similarity to Arabidopsis IDD11 (AT3G13810). Because of the prevailing presence in IDD genes, this type of zinc finger was named IDD-type C2H2. Although IDD genes have not been previously reported in M. truncatula, the identification of IDD-type C2H2 confirmed the presence and putative conserved function of this important gene family in M. truncatula.

IDD-type C2H2 motifs. a Alignment of IDD-type C2H2 motifs from M. truncatula. b Signature of IDD-type C2H2 motifs

Z-type motifs represent C2H2s that display highly conserved core sequences and do not belong to Q, QM, and IDD-type C2H2s. We identified 41 C2H2 motifs from 25 ZFPs, which were classified as Z-type (Additional file 8). Based on the pattern of the core sequences within the C2H2 motifs, Z-type C2H2s were further classified into eight subgroups, with 2 to 15 individual motifs in each subgroup (Additional file 8). Previous reports grouped the C2H2s that did not contain any conserved core sequences to C-type [1, 15, 26]. In this work, we identified 127 C-type C2H2s.

The C2H2 ZFPs identified in M. truncatula were naturally classified into groups based on the C2H2 zinc finger numbers in ZFPs. Among the 218 C2H2 ZFPs, 151 genes had a single C2H2 motif, representing 69% of the family. The dominant abundance of single C2H2 motif genes in the M. truncatula genome is consistent with the ZFP family in other plant species [16], and highlights the difference in the C2H2 ZFP family between plant and animal genomes, wherein single C2H2 motif genes are rarely found [7]. Among the 67 ZFPs with multiple C2H2s, 38 genes had two zinc fingers, 16 had three zinc fingers, 9 had four zinc fingers, one (TRM1) had five zinc fingers, two have six and one (TFIIIA) had nine zinc finger C2H2 motifs (Additional file 9).

The majority of C2H2 ZFPs found in the M. truncatula genome contained only one C2H2 motif. Among a total of 218 ZFPs found, 151 had only one C2H2 zinc finger motif. The EAR motif connects transcription suppressor TPL/TPL-like proteins, and thus enables transcription repression functions of the coding genes. The EAR motif is often found in ZFP genes. Based on the presence of the EAR motif, we categorized the 151 single C2H2 ZFPs into three subgroups, i.e., ZFPs without an EAR, ZFPs with one EAR and ZFPs with multiple EARs. Among 151 single C2H2 ZFPs, 74 contained no EAR, 60 contained one EAR, and 17 contained multiple EAR motifs (Additional file 10).

The 74 single C2H2 ZFPs without an EAR motif varied significantly in length, ranging from 69 to 1890 amino acids (Fig. 4a). The distribution of the C2H2 motif was also variable, from the very N-terminus to the end of the C-terminus (Fig. 4a). These features highlight the sequence divergence of this subgroup, and imply the functional diversification among members. Unfortunately, none of the members have been characterized to date. To better understand the putative functions of the genes in this subgroup, we performed a sequence analysis and homolog search in other species. The resultant gene annotation showed that the function of this subgroup was significantly diversified. Some members were putative zinc finger transcription factors or C2H2-like zinc finger proteins, such as MtrunA17_Chr1g0151261/Medtr1g015930 and MtrunA17_Chr3g0104431/Medtr3g463270; some members were annotated as a helicase domain-containing protein such as Medtr3g014030 and Medtr1g037750; and some were predicted as the VEFS-Box of polycomb proteins, such as MtrunA17_Chr1g0196821/Medtr1g090240 and MtrunA17_Chr5g0399611/Medtr5g013150 (Fig. 4a). Fifteen IDD-type C2H2 motifs were found in this subgroup (Fig. 4b), which was highly similar to Arabidopsis IDD genes, suggesting that the members containing the IDD-type C2H2 motifs in M. truncatula might have similar functions to their Arabidopsis homologs.

Representative single C2H2 ZFP in M. truncatula. a Single C2H2 ZFP without the EAR motif. b IDD homologs. c PALM1 and RSD type C2H2 ZFPs. The black line represents the protein sequence. The black and red rectangles indicate C2H2 and EAR motifs, respectively

MtrunA17_Chr8g0355141/Medtr8g043980, with a deduced length of 746 resides, is highly similar to Arabidopsis SERRATED LEAVES (SE, AT2G27100) [41], suggesting that it may be the M. truncatula homolog of SE. Arabidopsis PCF11P-SIMILAR PROTEIN 4 (PCFS4, AT4G04885) belongs to the C2H2 ZFPs gene family and plays a role in regulating flowering time [42]. The M. truncatula PCFS4 homolog has not yet been identified. We found that a member of this subgroup, MtrunA17_Chr6g0452881/Medtr6g011450, with a C2H2 motif located at the C-terminus, was highly similar to Arabidopsis PCFS4 and shared comparable C2H2 motif arrangement. This result suggests that MtrunA17_Chr6g0452881/Medtr6g011450 is probably the homolog of PCFS4.

As for the 60 single C2H2 ZFPs containing a single EAR motif, 59 of them had the C2H2 motif located upstream and EAR downstream. The C2H2 zinc finger was close to the N-terminus, whereas the EAR is placed at the far end of C-terminus, as represented by the reported PALM1 and RSD. This subgroup of C2H2 ZFPs was typically short, with 50 (83%) members less than 350 amino acids in length (Fig. 4c).

Except for PALM1 (MtrunA17_Chr5g0400571/Medtr5g014400) and RSD (MtrunA17_Chr7g0239441/Medtr7g063220), all the genes in this subgroup have not been characterized. By similarity analysis, we identified several genes that are very similar to Arabidopsis characterized C2H2 ZFPs. MtrunA17_Chr1g0175851/Medtr1g054450, which carried one C2H2 at N-terminus and EAR at C-terminus, appeared to be a close homolog of Arabidopsis ZFP8 (AT2G41940), soybean ZFP gene Glyma.09G107400 and rice C2H2 ZFP LOC_Os08g36110. MtrunA17_Chr5g0435711/Medtr5g080660 and MtrunA17_Chr1g0184381/Medtr1g070265 are very similar to Arabidopsis RABBIT EARS (REB, AT5G06070) [43] and SUPERMAN (SUP, AT3G23130) [19], respectively. Consistently, all these homologs had a similar motifs structure, i.e., one C2H2 and one EAR located at comparable positions (Fig. 4c, Additional file 10).

Medtr1g110710 is the only single C2H2 ZFPs with an EAR motif located upstream of the C2H2 zinc finger (Fig. 4d, Additional file 10). Medtr1g110710 is highly similar to and appears to be the homolog of Arabidopsis JAGGED (JAG, AT1G68480), which plays an important role in shaping lateral organs and promoting leaf tissue development [44, 45]. Sequence analysis also revealed that Medtr1g110710 was highly similar to soybean Glyma.10G273800, rice STAMENLESS 1 (SL1, LOC_Os01g03840) and a few other closely related genes in other plant species that share a similar gene structure to Medtr1g110710. SL1 has been reported to be the homolog of JAG [46]. Instead of regulating leaf development, SL1 controls floral development, demonstrating the functional divergence of the homologs [46]. Taken together, JAG, Medtr1g110710, and other homologs may represent a special cluster of C2H2 type ZFPs with an EAR motif upstream of the C2H2 zinc finger.

There were 17 single C2H2 ZFPs containing multiple EARs, including 16 genes containing two EARs and one gene containing three EARs (Additional file 10). The 16 ZFPs containing two EARs included nine ZFPs with the C2H2 motif placed between the two EARs, and another seven ZFPs with the C2H2 motif located downstream or upstream of the EARs (Additional file 11).

In the M. truncatula genome, 67 ZFPs harbored multiple C2H2 motifs, including 38 ZFPs containing two C2H2s, 16 ZFPs containing three C2H2s, 9 ZFPs containing four C2H2s, MtTRM1 (five C2H2s), 2 ZFPs containing six C2H2s and MtTFIIIA (nine C2H2s). ZFPs with multiple C2H2s have been widely found in other plant species [1, 15, 16, 27]. Based on the spacer length between motifs, the organization of C2H2s can be classified into dispersedly distributed motifs or tandem arrays [16, 27]. Using this method, we classified the 67 ZFPs with multiple C2H2s into two subgroups, i.e., 15 ZFPs contained tandem C2H2s and 52 ZFPs lacked tandem C2H2s.

The 15 ZFPs with tandem arrays of C2H2 motifs included three ZFPs with two C2H2s, two ZFPs with three C2H2s, six ZFPs with four C2H2s, and two ZFPs with six C2H2s, together with MtTRM1 and MtTFIIIA (Fig. 5). The spacer between C2H2 motifs in an array varied from zero to 11 residues. In MtrunA17_Chr8g0392331/Medtr8g106220, there were no link residues between two C2H2s, resulting in two continuous C2H2 motifs at the N-terminus (Fig. 5). The links between two C2H2s in MtrunA17_Chr7g0225251/Medtr7g029095 and MtrunA17_Chr4g0075341/Medtr4g132670 consisted of 5 and 10 residues, respectively (Fig. 5). Medtr7g029090 and MtrunA17_Chr4g0075321/Medtr4g132610, which had three C2H2s, showed a different array pattern. In MtrunA17_Chr4g0075321/Medtr4g132610, the links consisted of both 10 residues and three C2H2s forming a continuous tandem array. By contrast, the third C2H2 in Medtr7g029090 was isolated from the array formed by the first and second C2H2 (Fig. 5).

ZFPs with an array of C2H2 motifs in M. truncatula. The black line represents the protein sequence and the black rectangle indicates C2H2 motifs

Among the nine ZFPs with four C2H2s, three of them (MtrunA17_Chr5g0403241/Medtr5g018850, Medtr8g078490, and MtrunA17_Chr8g0373841/Medtr8g078700) had the same array pattern. The first three C2H2s were tandemly linked by 10 or 9 residues, and the fourth C2H2 was isolated from the array. Notably, all three proteins exhibited similarities to Arabidopsis SU (VAR) 3–9-RELATED PROTEIN 5 (SUVR5, AT2G23740). Two of them, MtrunA17_Chr1g0199511/Medtr1g094740 and MtrunA17_Chr3g0116441/Medtr3g075210 are highly similar to Arabidopsis EARLY FLOWERING 6 (ELF6, AT5G04240) and RELATIVE OF EARLY FLOWERING 6 (REF6, AT3G48430), respectively. Despite the functional divergence, ELF6 and REF6 were highly similar to each other in sequence, and they both encoded Jumonji-class transcription factors containing four tandem C2H2 motifs at the C-terminus [47]. Similar to ELF6 and REF6, MtrunA17_Chr1g0199511/Medtr1g094740 and MtrunA17_Chr3g0116441/Medtr3g075210 also had an array of four C2H2 motifs located at the C-terminus, including a degraded C2H2, which has also been found in ELF6 and REF6 [16]. MtrunA17_Chr1g0199511/Medtr1g094740 and MtrunA17_Chr3g0116441/Medtr3g075210 are highly similar to soybean Glyma.20G181000 and Glyma.04G191900, the C2H2 arrays which have recently been reported [27].

For the 52 ZFPs without a tandem C2H2 array, the C2H2 zinc fingers were all separately located. To date, none of them has been characterized. A homolog search showed that three of them (MtrunA17_Chr1g0207451/Medtr1g106730, MtrunA17_Chr1g0152661/Medtr1g018420, and MtrunA17_Chr3g0134711/Medtr3g102980) had sequence similarity to Arabidopsis SALT TOLERANCE ZINC FINGER (STZ/AT10 AT1G27730) [48], suggesting that three ZFPs might mediate salt tolerance in M. truncatula. In addition, two members from a local tandem duplication, MtrunA17_Chr4g0049671/Medtr4g093260 and MtrunA17_Chr4g0049701/Medtr4g093270, exhibited similarity to members of the Arabidopsis HVA22 family, suggesting that they might also be involved in the stress response in M. truncatula.

To understand the spatial expression patterns of M. truncatula C2H2 ZFPs, we analyzed the tissue-specific expression datasets for the C2H2 ZFPs in M. truncatula. By sequence comparison between the transcripts of C2H2 ZFPs and M. truncatula Affymetrix microarray probesets, we found that 119 out of a total of 218 ZFPs had the corresponding microarray probesets, enabling us to perform further expression analysis on M. truncatula Gene Expression Atlas (https://mtgea.noble.org/v3/) (Fig. 6a, b, c) [49, 50].

Tissue-specific expression analysis of C2H2 ZFPs in M. truncatula. a - c The transcript abundances of the C2H2 ZFPs showed by heatmaps generated from the M. truncatula Gene Expression Atlas data. a shows the genes with low transcript abundances, (b) shows the genes with medium transcript abundances, and (c) shows the genes with high transcript abundances. d The qRT-PCR result of six selected C2H2 genes. Shown are means ± standard deviations for three biological replicates and three technical replicates of each biological replicate. The relative expression levels of tested genes were normalized by the geometric mean of three endogenous control genes. ‘N.A.’ indicated undetectable expression. The lowercase letters above the bar indicate significant differences (P