By treating vacuolated microspores with stress factors, e.g., starvation or cold, as applied in this study, it is possible to change their developmental pathway from a gametophytic into a sporophytic one and to induce microspore embryogenesis. It is well known that the reprogramming of microspore pathway is possible only at a narrow window during microsporogenesis and in barley the efficient embryo induction is routinely initiated from ML uninucleate microspores. ML microspores of cvs. ‘Jersey’ and ‘Mercada’ used in our experiments went successfully through these reprogramming processes as they started mitotic divisions and formed embryos with a high frequency. Both cultivars exhibited a similar induction and regeneration potential of ca. 100 plants per 100,000 isolated microspores, however they differed significantly in their ability to produce green plants.
Regeneration of green plants depends on formation of chloroplasts, which in androgenic plants originate from proplastids enclosed in microspores that initiate in vitro culture. Proplastids present in microspores are programmed to differentiate into amyloplasts, however as we have previously showed, the stage of microspore development at which activation and differentiation into amyloplast occurred, varied between barley cultivars. We demonstrated that the molecular differentiation of plastids in barley microspores prior to in vitro culture affects the genotype ability to regenerate green plants in culture [18]. Cultivars producing mostly albino regenerants in androgenesis, such as ‘Mercada’, showed early activation of starch synthesis genes, differentiation of proplastids into amyloplasts and degradation of plastomes during microspore development in vivo. At the mid-to-late uninucleate (ML) stage, routinely used for culture initiation, microspores of cv. ‘Mercada’ contained a large proportion of amyloplasts filled with starch, while in microspores of cv. ‘Jersey’, producing mostly green regenerants, only proplastids with no distinctive morphology were present. Furthermore, in microspores of cv. ‘Mercada’ the activation of reserve starch synthesis genes occurred already at the early-mid (EM) stage of development which proceeds the stage used for culture initiation [18].
In the presented study we analysed plastid differentiation in microspores cultured in vitro. Using the same two barley cultivars ‘Jersey’ and ‘Mercada’ we examined the differentiation of plastids at the successive stages of in vitro culture, from the pre-treated microspores, through embryo induction and embryo formation phase to the plant regeneration. As mentioned above, in the ML microspores of cv. ‘Mercada’ almost a half of proplastids have already been differentiated into amyloplast, which were not present in ML microspores of ‘Jersey’. Since the fourth day of culture (4dC) we identified the appearance of starch-accumulating plastids also in dividing microspores and microspore-derived structures of cv. ‘Jersey’. However, the analysis of gene expression revealed a different function of these amyloplasts compared to ‘Mercada’. The expression of genes encoding reserve starch was markedly reduced, while the genes responsible for assimilatory starch synthesis were activated in ‘Jersey’ at the beginning of in vitro culture. The physiological function of reserve and assimilatory starch differs. The reserve starch is stored for the long period and used as the source of energy in the next generation whereas the assimilatory starch is temporarily accumulated during the day and is rapidly consumed to support biological activities in plant [63]. In both cultivars, the number of amyloplasts and expression of assimilatory starch synthesis genes reached the highest values on 21dC, when pro-embryo have been formed.
It should be noted that the increased number of amyloplasts in cv. ‘Mercada’ was observed before activation of genes related to the assimilatory starch synthesis. Therefore, we assumed that the majority of amyloplasts present in ML microspores of ‘Mercada’ were not degraded after pretreatment and culture initiation, and could divide as it was described in starch accumulating tissues, for example in Oryza sativa and Colocasia esculenta [64, 65]. Thus, the applied pre-treatment did not reverse the process of plastid differentiation in ML microspores of cv. ‘Mercada’, including starch synthesis and accumulation. However, the pre-treatment resulted in a decrease or even disappearance of differences between both cultivars in expression profile of most genes related to plastid transcription and translation. It also hindered the process of plastome degradation that occurs during pollen grain formation in vivo. The degradation of plastomes in ‘Mercada’ microspores was confirmed in this study by activation of DPD1 exonuclease responsible for degradation of plastid DNA during pollen development [66]. In ‘Mercada’ DPD1 was expressed throughout microsporogenesis, since the early stage of microspore development, which was not observed in cv. ‘Jersey’. During successive stages of in vitro culture both cultivars exhibited a low expression of DPD1 and a changing number of plastomes.
The number of plastid copies increased during embryo induction and formation phase in both genotypes and reached the highest value on 21dC, in accordance with the highest number of plastids observed at this time point.
Early phase of chloroplast differentiation is regulated by checkpoints at proplastid development, including the stability of plastomes and effective transcription and translation processes occurring in proplastids [51, 67]. Decrease of the number of plastid genomes is frequently observed during chloroplast differentiation associated with plant development [68]. High demand for plastidial mRNAs and rRNAs within the cell is covered by the great number of plastids and many copies of plastomes within plastid [69]. The different plastome content between organs serves as a transcription regulation in plastids [70]. We observed a significant decline of the plastome copy number in both cultivars between 21dC and 35dC, i.e. during embryo differentiation phase. In cv. ‘Jersey’, in contrast to cv. ‘Mercada’, the plastome copy number increased during the next phase of embryo development, i.e. embryo body axis formation. Additionally, in cv. ‘Mercada’ since 35dC the copy numbers of individual genes differed from the values expected on the basis of gene localisation within plastome. During conversion of cv. ‘Mercada’ embryos and further development of regenerants, the average copy number of plastomes increased, however the number of copies of individual genes were even more divergent. This indicates that the copies of replicating plastome were inaccurate. The high number of proper plastid genomes is considered as a checkpoint for chloroplast differentiation [71]. It was proven in Arabidopsis that the instability of plastid genomes blocked further light-induced differentiation of proplastid into chloroplast [72]. Plastid genome of cv. ‘Mercada’ during androgenic culture showed high instability, which could influence the differentiation of proplastids into chloroplasts.
The transition from NEP- to PEP-dependent transcription in plastids is a crucial factor in chloroplast differentiation, as only plastidial transcription system is capable of providing a high level of transcripts of plastid-encoded genes including rRNAs, tRNAs, and genes for some ribosomal proteins involved in plastid translation [41, 73]. Mutation in each of PEP subunit genes caused an albino phenotype and a lack photosynthesis in tobacco (Nicotiana tabacum L.) [74, 75] while a knock-out of RpoTp (encoding NEP) in Arabidopsis resulted in delayed chloroplast development only [76]. The progress of chloroplast differentiation, which involves light, requires SIG2-dependent expression of plastid genes to activate PEP-dependent transcription in plastids. Among these genes is the tRNAGlu gene whose transcription product, after reaching a certain level, inhibits the activity of NEP and thus serves as a checkpoint for induction of chloroplast differentiation [45–47, 77]. In addition, the charged glutamyl-tRNA is the precursor for 5-aminolevulinic acid and promotes expression of photosynthesis-associated nuclear genes in retrograde signalling [45].
Expression profiling of Sig2 and tRNAGlu revealed significant differences between ‘Jersey’ and ‘Mercada’ cultivars, indicating different transcription activity of polymerases NEP and PEP. In cv. ‘Jersey’ the activation of PEP-depended transcription was observed already in converting embryos on 46dC. At this time point, the relative expression levels of Sig2 and tRNAGlu genes in differentiating ‘Jersey’ embryos increased 37-fold and 5-fold, respectively, compared to their levels on 43dC. Contrary to ’Jersey’, the PEP-dependent transcription during early stages of plant regeneration in cv. ‘Mercada’ was not observed. The relatively low expression level of tRNAGlu throughout plant regeneration indicates that NEP was still the dominant RNA polymerase in ‘Mercada’ plastids. The additional support for ongoing activity of NEP in in regenerating plants of cv. ‘Mercada’ was given by a high expression of rpoB gene that is dominantly transcribed by NEP. Furthermore, expression profile of Sig2 normalized to the reference genes ARF1 (ADP-ribosylation factor 1-like protein) and EF1 (Translation elongation factor 1-a), clearly showed that this nuclear gene was active only since 46dC and only in ‘Jersey’ embryos (Additional file 1: Figure S7). This raises the question about the factor (factors) that triggered the Sig2 activation in ‘Jersey’ embryos but did not act in ‘Mercada’. We assume that the mechanism leading to the lack of Sig2 expression in ‘Mercada’ may be related to the genome instability observed during embryo formation in this cultivar. As described above, since 35dC, the plastids of cv. ‘Mercada’ contained incorrect copies of plastic genomes and on 43dC they had the lowest number of genome copies during the whole in vitro culture. The integrity of plastid genome is considered as a checkpoint during early proplastid-to-chloroplast differentiation [78, 79]. Differentiation of chloroplasts depends on the effective nucleus-to-plastid (anterograde) and plastid-to-nucleus (retrograde) signaling and the lack of signal from plastids might result in the absence or deficiency of transcription activation in the nucleus. Nevertheless, the molecular mechanisms underlying the retrograde signal that activates Sig2 gene in the nucleus remain to be uncovered.
The role of RNA polymerases seems to be crucial in regenerating green plants in microspore embryogenesis as the albinism is a phenomenon occurring solely in cereals that harbour only one nuclear-encoded polymerase (RpoTp) [48]. Dicots, that do not regenerate albino plants in androgenic culture, contain two nuclear-encoded polymerases: RpoTp and RpoTmp required for transcription occurring in plastids, including transcription of rRNA genes [39, 80]. The main consequence of the failed NEP-to-PEP transition in ‘Mercada’ plastids was the lack of activation of rRNAs transcription during embryo conversion and plant regeneration stages. In ‘Jersey’ embryos, where this transition took place, the relative expression of 16S and 23S genes encoding plastid rRNAs increased 20 to 30-fold between 43dC and 46dC and reached 300–500 times higher level in the regenerated plantlets on 55dC. When expression profiles of plastid rRNA genes normalized to the reference genes are observed throughout the whole androgenesis process, it is clearly seen that the increase in rRNA levels took place only in cv. ‘Jersey’ and was initiated on 43dC (Additional File 1: Figure S7). At this time point, when body axis in embryos become visible, embryo had been cultured for 8 days on regeneration medium, first at darkness and for 3 days in light. In contrast to ‘Jersey’, during regeneration of cv. ‘Mercada’ plants we did not observe any significant increase in the level of plastid-encoded rRNA transcripts between 43dC and 55dC. The 16S and 23S rRNAs are required for ribosome assembling and their lack results in ribosome depletion [50], while the proper translation occurring in plastids is necessary for induction of proplastid-to-chloroplast differentiation [81].
As a consequence of incorrect plastid biogenesis, genes encoding transcriptional factors GLKs, that are involved in light-induced chloroplast differentiation, as well as genes related to photosynthesis were not activated in regenerating ‘Mercada’ plants and albino regenerants of both cultivars. Similarly, barley albostrians mutant with depletion in plastidial ribosomes showed the reduced content of mRNAs for photosynthesis-related proteins [82].
Continuous activity of NEP in albino regenerants of cv. ‘Mercada’ resulted in accumulation of transcripts of plastid genes whose transcription is NEP-dependent. Also other genes involved in plastid biogenesis, such as import (Tic21, Toc159) and plastome replication stayed active. The increased expression of these genes indicate the response at the transcription level in order to regain plastid biogenesis and maintain chloroplast differentiation. Studies in Arabidopsis showed that the low activity of PEP in plastids, caused by depletion of PEP subunits or sigma factors, resulted in the increased transcription of genes dependent on NEP [45, 75, 83]. The increased level of PEP subunit transcripts encoded by rpoA, rpoB, rpoC1, rpoC2 genes was also revealed in barley albostrians and maize iojap mutants that lack plastid 70S ribosomes [84, 85].
Interestingly, plastids in mesophyll cells of albino plants of cv. ‘Mercada’ were more advanced in differentiation than in albino plants of cv. ‘Jersey’ that occasionally appeared among regenerants. TEM observations showed the presence of both, the prolamellar body and non-organized prothylakoid/thylakoid structures in the etioplast-like plastids of ‘Mercada’, while in ‘Jerey’ etioplasts only prolamellar bodies were present. Albino plants of cv. ‘Jersey’ exhibited a 2-fold lower average number of plastome copies in comparison to green regenerants and a significant deviation between numbers of individual gene copies. The low number of plastidial gene copies, including rRNA-encoding genes, provided a limited number of templates for transcription occurring in plastids, which in turn impeded the light-dependent chloroplast biogenesis and resulted in the arrest of plastids at early stage of differentiation. The various number of copies of specific genes observed in albino plants of cv. ‘Jersey’ might result from incomplete plastome replications and/or structural changes in plastid genome, as described in albino regenerants of many cereals [19–22].
Based on the comparison of green and albino regenerants of cvs. ‘Jersey’ and ‘Mercada’ it is worth noting that the time of activation rather than the level of expression of specific genes is crucial in regeneration of green plants. Expression of many of analysed genes including tRNAGlu, Sig2, Glk1, Glk2 increased in cv. ‘Mercada’ on 55th day of culture, yet the plastids observed in mesophyll cells of albino regenerants were already arrested at the early stage of development. The lack of the proper plastid biogenesis during embryo differentiation stage resulted in in the lack of proplastid-to-chloroplast transition and regeneration of albino plants.
It should be underlined that the presented here mechanism leading to formation of albino plants in androgenic culture is a consequence of plastid differentiation during pollen development in vivo. The proplastids, which initiated the programme of proplastid-to-amylopast differentiation, cannot be reversed by in vitro conditions. Therefore, microspores that contain such plastids at the stage of culture initiation will produce mostly albino regenerants in androgenesis. Additional support for this notion was provided by the induction of isolated microspore culture from the earlier stage of microspore development than routinely utilised. When microspores harbour only proplastids, it is possible to significantly increase the frequency of green regenerants in barley androgenesis [18].