Initial characterisation of MtPHYA
MtPHYA (Medtr1g085160) is predicted to encode a 1124 aa protein containing the important domains that comprise the N-terminal photosensory core module and the C-terminal regulatory region typical of PHYA-like proteins (Additional file 1: Figure S1) [27]. It is 79% identical to Arabidopsis PHYA, but shares highest homology with PHYA-like protein sequences from other temperate legumes including pea (95%), red clover (95%), chickpea (93%) and lotus (89%) (Additional file 1: Figure S1).
We analysed MtPHYA gene expression in different tissues and environments by qRT-PCR using gene-specific primers 2F and 2R (Fig. 2a, Additional file 2: Table S1). MtPHYA was detected in a wide range of tissues 2 h after dawn, including cotyledons, leaves, apical buds, open flowers and roots (Fig. 1a) and was slightly more abundant in SD than in LD photoperiods (Fig. 1b). Analysis of MtPHYA expression in a developmental time-course in LD in leaves and shoot apices indicated that it did not change significantly in leaves, except for a 2-fold increase prior to flowering, but its expression declined after flowering (Fig. 1c). In contrast, in shoot apices, there was a steady increase in expression through development and it continued to rise after flowering (Fig. 1d).
Medicago flowering is promoted by prolonged winter cold (vernalization, V) followed by long-day (LD) photoperiods [10]. To assess if vernalization has a direct effect on MtPHYA transcript, we analysed its expression in a vernalized seedling long-day (VsLD) time course. Seeds were germinated, grown in LD, then 14-d-old seedlings were vernalized at 4 °C and afterwards returned to warm LD. MtPHYA was expressed in leaves and apices of seedlings before, during and after vernalization at similar levels, although at a significantly lower level than in germinated seeds (Fig. 1e-f). This indicates that MtPHYA expression is not directly regulated by cold. The high abundance of MtPHYA in etiolated germinated seeds (Fig. 1e), is consistent with other plants including pea where PHYA transcript accumulates to a much higher level in the dark than in the light [29, 30, 33].
Two independent Mtphya mutant lines have reduced sensitivity to far-red light compared with wild type R108 seedlings
To investigate the function of MtPHYA in Medicago plant development, we analysed two independent Tnt1 retrotransposon-tagged R108 mutant lines, Mtphya-1 (NF1583) and Mtphya-2 (NF3601), which have insertions facing in opposite directions in the 5′ UTR of MtPHYA (Fig. 2a). To analyse if the Tnt1 insertions affected the full-length transcript of MtPHYA, we used primers 3F and 3R (Fig. 2a, Additional file 2: Table S1) to amplify cDNA fragments from wild type R108 and the two mutant lines. The primers amplified cDNA fragments from both R108 and Mtphya-1, but none from Mtphya-2 indicating that the latter was a knockout mutant (Fig. 2b). Direct sequencing of the PCR product from R108 indicated that the intron upstream of the ATG in R108 was spliced out using the 5′ splice donor site at position − 522 to generate a 3533 bp cDNA. However, sequencing of the PCR product from Mtphya-1 indicated alternative splicing of this upstream intron, which utilized a 5′ splice donor site at position − 611. This led to splicing out of the Tnt1 insertion and adjacent 89 bp in the 5’UTR of Mtphya-1, which resulted in amplification of a slightly shorter cDNA (3444 bp). As measured by qRT-PCR using primers 2F and 2R, Mtphya-1 and Mtphya-2 had a statistically significant ~ 3-fold and ~ 11-fold reduction, respectively, in MtPHYA gene expression, compared with wild-type R108 plants (Figs. 2c, 4a).
Because PHYA has a well-documented role in regulation of seedling photomorphogenesis [34, 35], we then analysed seedling de-etiolation responses of the two Mtphya mutant lines and wild-type R108 (Figs. 2d-e, 4b). Seeds were germinated in the dark overnight and then grown for three days either under far-red (FR) light, in the dark (D) or in white light (WL). Under continuous FR light, both the Mtphya mutants had longer hypocotyls with unexpanded cotyledons, compared to wild type R108 seedlings with short hypocotyls and expanded, green cotyledons (photographs shown for Mtphya-1, Fig. 2d). In contrast, in WL, the Mtphya mutants had short hypocotyls that were only slightly longer than wild-type R108 hypocotyls (photographs shown for Mtphya-1, Fig. 2d). In dark, the hypocotyls of both mutants were very long like dark-grown, wild-type R108 seedlings (photographs shown for Mtphya-1, Fig. 2d). When the hypocotyl lengths were plotted as a ratio of light to dark grown, the WL/Dark hypocotyl length ratios were low in the wild-type and both the Mtphya mutants (Figs. 2e, 4b). However, the FR/Dark hypocotyl length ratio in the Mtphya mutants was much higher than in R108 (Figs. 2e, 4b). These results indicate that the Mtphya mutants had reduced sensitivity to far-red light.
Mtphya mutants flower later than wild type R108 particularly in LD and VLD photoperiods
To investigate if the Mtphya mutations affect flowering time, we characterized the mutant plants for their flowering time phenotypes compared with wild type R108 plants. First, under vernalized long day photoperiod (VLD) conditions, both Mtphya-1 and Mtphya-2 mutant plants were late flowering compared with R108 (Figs. 2f-g, 4c-d). We then crossed homozygous Mtphya-1 mutants with wild-type R108 control plants and analysed the genotype and flowering time of the F1 plants and the segregating F2 population in VLD (Fig. 2f-h). The F1 progeny had a weak intermediate phenotype, flowering slightly later than wild-type R108, but much earlier than Mtphya-1 mutants. In the segregating F2 population of 217 plants, ~ one quarter (n = 50) were Mtphya-1 Tnt1 homozygotes and late flowering, ~ half (n = 114) were heterozygotes and flowered slightly later than wild-type R108 and wild-type segregants, and ~ a quarter (n = 53) were wild-type segregants and flowered like wild-type R108. Thus, the Tnt1 insertion in Mtphya-1 was tightly linked to the late flowering locus (within ≤1 cM).
The Tnt1 insertion in Mtphya-2 also showed 100% co-segregation with the late flowering phenotype. The pattern of inheritance in Mtphya-2 was analysed by characterising the flowering time of a segregating population from heterozygous, self-crossed parents in VLD. Out of 45 plants, about one quarter (n = 11) were homozygotes and all were late flowering, and ~ one quarter (n = 8) were wild-type segregants and early flowering like wild-type R108. The remaining plants (n = 26) were heterozygotes, which showed semi-dominance as observed for Mtphya-1, because they displayed an intermediate late flowering time phenotype (Fig. 4c-d).
To further investigate the role of MtPHYA in regulation of flowering, Mtphya-1 plants were grown under different photoperiodic conditions, with or without, vernalization treatment of germinated seeds (Fig. 2i-j). As expected, the R108 wild-type plants exhibited a strong response to photoperiod and vernalization, flowering most rapidly in vernalized long day (VLD) but flowered the latest under non-vernalized short day (SD) conditions. However, the Mtphya-1 mutants were strikingly impaired in their ability to respond to LD compared with wild type plants. Mtphya-1 mutants were delayed in flowering in both LD and VLD compared to R108, but flowered at a similar time to wild type R108 in VSD and SD. The mutants exhibited a late flowering day-neutral phenotype, particularly in vernalized conditions, as they flowered at a similar time in VLD and VSD. The Mtphya-1 mutants retained the ability to respond to vernalization because they flowered earlier in VLD than LD, and similarly, the VSD-grown mutants were earlier than the SD-grown plants. However, the response of Mtphya-1 to vernalization was slightly weaker compared with R108.
Mtphya mutants have a very short primary axis in LD and VLD photoperiods compared with wild type
In addition to displaying delayed flowering, both the Mtphya mutant plants were more compact than wild type R108, with a strikingly short primary axis (see photographs of Mtphya-1 mutant in Fig. 3a-b). This phenotype was more pronounced in LD photoperiods (both LD and VLD) than in SD or VSD conditions in the Mtphya-1 mutant compared with R108 (Fig. 3a-d). Measurements of primary axis lengths, taken at different stages of development, indicated that both the Mtphya mutants exhibited the short axis phenotype compared to R108, prior to and after flowering (Fig. 3e-f, Fig. 4g). To analyse the inheritance of the short primary axis phenotype, we also measured primary axis length in VLD in the population segregating for the Mtphya-2 mutation (Fig. 4f), previously analysed for flowering time (Fig. 4c-d). The short axis phenotype was only observed in the Mtphya-2 homozygous segregants (Fig. 4f). Thus, there was co-segregation between the short primary axis phenotype and the late flowering phenotype (Fig. 4c-d). These results indicate that the late flowering time defect and the short primary axis phenotype are both caused by mutations in the MtPHYA gene.
The delayed flowering of Mtphya-1 in LD and VLD is associated with a decrease in expression of LD-induced MtFTs, MtFULs and MtE1L
Next, we analysed the molecular basis of the altered long day photoperiod flowering time response of Mtphya-1 mutants. To do this, we analysed the expression of candidate Medicago circadian clock and flowering-time genes in leaves of the Mtphya-1 mutant and wild type plants in LD and SD (Fig. 5) and in VLD (Fig. 6) two hours after dawn (ZT2) using qRT-PCR.
The expression level of GI was similar in LD and SD in wild type R108 and was reduced by ~ 3-fold in the Mtphya-1 mutants compared with R108 in LD, but not in SD (Fig. 5). GI was reduced by ~ 2-fold in the mutant in VLD (Fig. 6). TOC1a level increased by ~ 4-fold in the Mtphya-1 mutant in SD compared with R108, but was unchanged in LD (Fig. 5) at ZT2. However, it was reduced by ~ 2-fold in the mutant in VLD compared with R108 control (Fig. 6a). The other Medicago clock-related genes homologous to ELF3, ELF4, LUXa, LUXb and LHY were analysed only in VLD, but were not changed compared with wild type R108 (Fig. 6a). The expression of a candidate flowering-time gene, FKF1-like, was also not changed in the mutant compared with wild type in VLD (Fig. 6a).
The expression of the three LD-induced MtFT genes, MtFTa1, MtFTb1 and MtFTb2 [16] were strikingly reduced in the Mtphya-1 mutant. MtFTb1 and MtFTb2 were highly expressed in LD but undetected under SD in R108, consistent with previous findings [16]. However, both genes were strongly decreased by ~ 16- and ~ 19-fold, respectively, in the Mtphya-1 mutant compared with wild type in LD (Fig. 5). In VLD, MtFTa1, MtFTb1 and MtFTb2 were strongly decreased by ~ 39-, 24- and 11-fold, respectively in the Mtphya-1 mutants compared with wild-type R108 (Fig. 6b). In contrast, the fourth FT-like gene tested, MtFTa2, was weakly reduced by ~ 2-fold in the mutant in VLD (Fig. 6b).
MtFULb was previously shown to be under photoperiodic control because it was expressed at higher levels in LD than in SD [17]. A similar result was observed here (Fig. 5). MtFULa was also photoperiodically regulated with strongly elevated expression in the leaves of wild type R108 under LD compared with SD (Fig. 5). The Mtphya-1 mutation had a particularly strong effect on the expression of MtFULa because it was decreased by ~ 15-fold, while MtFULb was reduced by ~ 3.5-fold in the Mtphya-1 mutant compared with wild type in LD. Similarly, in VLD, MtFULa was reduced by ~ 8.3-fold and MtFULb by ~ 3-fold in the Mtphya-1 mutant compared with wild type R108 (Fig. 6c).
The three MtSOC1 genes were expressed at higher levels in LD than in SD in R108 as shown in previous work [20, 21]. However, the expression of MtSOC1a was not significantly changed in the leaves of the Mtphya-1 mutants compared with R108 in either LD or SD conditions. MtSOC1b and MtSOC1c decreased by ~ 3-fold in the mutant in LD but were unchanged in SD (Fig. 5). In VLD, the Mtphya-1 mutants showed a weak decrease in the expression of MtSOC1a and MtSOC1c (~ 2-fold) but not in MtSOC1b compared with R108 (Fig. 6d).
We observed that MtE1L is ~ 3.5 fold more abundant in LD than in SD in R108, indicating that it is also photoperiodically controlled (Fig. 5). The same pattern was observed for the expression of MtE1L in the Mtphya-1 mutant. However, the mutant plants showed a reduction in MtE1L transcript level by ~ 3-fold in LD and SD compared with R108 control and a similar pattern was observed under VLD (Fig. 6e).
There was no significant change in the transcript level of MtVRN2 [23] in the leaves of Mtphya-1 mutants compared with wild type (Fig. 6e).
Mtphya-1 Mte1l double mutants flower at the same time as the Mtphya-1 mutant in VLD
Since MtE1L expression is under LD photoperiodic control and its level decreased in the Mtphya-1 mutant, we tested genetically if MtE1L and MtPHYA promoted flowering via a common pathway. The Mte1l mutant line (NF16583) with a Tnt1 insertion in the MtE1L coding region was obtained [19]. The Mte1l single mutants were moderately delayed in flowering compared with wild-type R108 controls in VLD (Fig. 7). We then crossed Mtphya-1 with Mte1l and genotyped and scored the flowering time of the segregating F2 population from the cross. Three out of 59 (~ 1/16) F2 plants were Mtphya-1 Mte1l double mutants. These plants flowered late in VLD at a similar time to the single Mtphya-1 mutants (Fig. 7). Therefore, no additive effect on flowering time was seen indicating that Mte1l and Mtphya-1 were likely to be in the same pathway. The other Mtphya-1 homozygous plants, either wild type or heterozygous at MtE1L, were also similarly delayed in flowering comparable to Mtphya-1 mutant (Fig. 7). However, only a very weak effect on flowering time was seen in the one MtPHYA wild type/Mte1l homozygous F2 plant obtained, consistent with the mild effect of Mte1l on flowering in VLD (Fig. 7).
Mtphya-1 fta1 double mutants flower slightly later than either of the single mutants in VLD
Plants with a knockout mutation at MtFTa1 are late flowering with short primary axes and prostrate architecture [16], similar to the Mtphya mutants. In addition, MtFTa1 expression is strongly reduced in the Mtphya-1 mutant in VLD. Thus, to test if MtFTa1 and MtPHYA promote flowering in the same flowering time pathway, we crossed Mtphya-1 and fta1 mutants to generate a double mutant. In our initial experiment, we obtained six F2 plants, but genotyping indicated that none were double mutants. Therefore, a F2 Mtphya-1-homozygous/ FTa1-heterozygous plant was selected, self-crossed, and its F3 progeny were genotyped and scored.
We identified 9 Mtphya-1 fta1 double mutants out of 48 F3 plants. They flowered later (by ~ 2 weeks and with 3–4 more nodes) than the Mtphya-1-homozygous F3 plants (Fig. 8a-b). They also flowered later than Mtfta1 single mutants grown as controls (Fig. 8a-b). This indicates that the mutations at both loci caused a weak additive effect compared with the single mutants in VLD. In addition, the 24 Mtphya-1 homozygous/FTa1 heterozygous plants flowered slightly later than the Mtphya-1 homozygous/FTa1 wild-type plants (Fig. 8a-b). This suggests that fta1 and Mtphya-1 largely affect flowering in the same pathway, but also possibly under different pathways. Apart from the delay in flowering time, mutations at both loci also caused a weak additive effect on reduction of the primary axis length (Fig. 8c), with a slight reduction in the length of the longest secondary axis (Fig. 8d).