CRK5 Protein Kinase Contributes to the Progression of Embryogenesis of Arabidopsis thaliana

The fine tuning of hormone (e.g., auxin and gibberellin) levels and hormone signaling is required for maintaining normal embryogenesis. Embryo polarity, for example, is ensured by the directional movement of auxin that is controlled by various types of auxin transporters. Here, we present pieces of evidence for the auxin-gibberellic acid (GA) hormonal crosstalk during embryo development and the regulatory role of the Arabidopsis thaliana Calcium-Dependent Protein Kinase-Related Kinase 5 (AtCRK5) in this regard. It is pointed out that the embryogenesis of the Atcrk5-1 mutant is delayed in comparison to the wild type. This delay is accompanied with a decrease in the levels of GA and auxin, as well as the abundance of the polar auxin transport (PAT) proteins PIN1, PIN4, and PIN7 in the mutant embryos. We have previously showed that AtCRK5 can regulate the PIN2 and PIN3 proteins either directly by phosphorylation or indirectly affecting the GA level during the root gravitropic and hypocotyl hook bending responses. In this manuscript, we provide evidence that the AtCRK5 protein kinase can in vitro phosphorylate the hydrophilic loops of additional PIN proteins that are important for embryogenesis. We propose that AtCRK5 can govern embryo development in Arabidopsis through the fine tuning of auxin-GA level and the accumulation of certain polar auxin transport proteins.


Introduction
Embryogenesis is the first stage of the plant life cycle, which initiates with fertilization and zygote development and it terminates with the maturation of the embryo. The three phases of embryogenesis are the proembryo stage, the early, and the late embryogenesis. Embryo polarity is determined during the proembryo stage with asymmetric division of the zygote [1], while morphogenesis characterizes early embryogenesis from the globular to the heart stage [2]. The late embryogenesis is marked by maturation (expansion of the cotyledons and the axis of the embryo without cell divisions) and ends in

The Atcrk5-1 Mutant Exhibits a Considerable Delay in the Progression of the Phases of Embryogenesis
We could recognize a notable size difference when comparing the wild type Arabidopsis thaliana (Col-0) and the Atcrk5-1 mutant dry seeds (containing bent-cotyledon embryo stages) ( Figure 1A, pictures). We could determine significantly smaller seed sizes for the mutant when the dried seed contours were measured ( Figure 1A, diagram). After two days' imbibition of dried seeds at 4 • C, we carefully pressed out the inverted stage embryos from the seeds by forceps, and measured the average length of the embryo axis from the shoot apical meristem (SAM) to the root apical meristem (RAM) in both of the genotypes. We found that the length of the axis was significantly shorter in Atcrk5-1 embryos as compared to the wild type Col-0 ones ( Figure 1B). Additionally, the bending angles of hypocotyl hooks of these bent-cotyledon-stage embryos were also measured. It is notable that the hypocotyl hook of the Atcrk5-1 mutant embryos was already more opened at this stage when compared to the wild type ones ( Figure 1C; see Reference 61. for more details).

The Atcrk5-1 Mutant Exhibits a Considerable Delay in the Progression of the Phases of Embryogenesis
We could recognize a notable size difference when comparing the wild type Arabidopsis thaliana (Col-0) and the Atcrk5-1 mutant dry seeds (containing bent-cotyledon embryo stages) ( Figure 1A, pictures). We could determine significantly smaller seed sizes for the mutant when the dried seed contours were measured ( Figure 1A, diagram). After two days' imbibition of dried seeds at 4 °C, we carefully pressed out the inverted stage embryos from the seeds by forceps, and measured the average length of the embryo axis from the shoot apical meristem (SAM) to the root apical meristem (RAM) in both of the genotypes. We found that the length of the axis was significantly shorter in Atcrk5-1 embryos as compared to the wild type Col-0 ones ( Figure 1B). Additionally, the bending angles of hypocotyl hooks of these bent-cotyledon-stage embryos were also measured. It is notable that the hypocotyl hook of the Atcrk5-1 mutant embryos was already more opened at this stage when compared to the wild type ones ( Figure 1C; see Reference 61. for more details). . The diagram shows the quantification of the seed size in wild type and mutant lines after two days' imbibition of dried seeds at 4 °C. The seed contours were measured by ImageJ. The averages are from 100 independent seeds from the wild type and mutant, respectively. Standard errors (SE) are also shown. The mutant value indicated by asterisk is significantly lower compared to the wild type (Student's t-test: p < 0.01, n = 100). (B) Embryo axis length measured from shoot apical meristem (SAM) until root apical meristem (RAM) was calculated in bent-cotyledon-stage embryos arised from the (A) experiment. Pictogram shows how the SAM-RAM axis was measured. All values are averages of at least 100 bent-cotyledon embryos. Standard errors (SE) are also shown. Asterisk depicts significant difference between the wild type and mutant embryos (Student's t-test: p < 0.05, n = 100). (C) Differences in the hook angles of wild type and mutant embryos at the bent-cotyledon-stage (after two days' imbibition of seeds at 4 °C). The averages and standard errors (SE) are shown. The mutant value is significantly different in comparison with the wild type (Student's t-test: * p < 0.01, n = 75). Pictogram shows the mode of measuring embryo hook angles (C). All experiment was repeated three times.
Thereafter, the individual developmental stages of the embryos were tested to find possible morphological differences between the wild type and mutant embryos. Microscopic images of embryos photographed by a CELL-R Olympus Microscope, are presented in Figure 2. The embryos were isolated from siliques1-11 (where S1-S11 represent silique position numbers from the youngest Figure 1. Differences in seed and embryo sizes and embryo bent-cotyledon hook angles of wild type (Col-0) and mutant (Atcrk5-1) lines. (A) Pictures show dry seeds of the wild type (Col-0) and mutant (Atcrk5-1). The diagram shows the quantification of the seed size in wild type and mutant lines after two days' imbibition of dried seeds at 4 • C. The seed contours were measured by ImageJ. The averages are from 100 independent seeds from the wild type and mutant, respectively. Standard errors (SE) are also shown. The mutant value indicated by asterisk is significantly lower compared to the wild type (Student's t-test: p < 0.01, n = 100). (B) Embryo axis length measured from shoot apical meristem (SAM) until root apical meristem (RAM) was calculated in bent-cotyledon-stage embryos arised from the (A) experiment. Pictogram shows how the SAM-RAM axis was measured. All values are averages of at least 100 bent-cotyledon embryos. Standard errors (SE) are also shown. Asterisk depicts significant difference between the wild type and mutant embryos (Student's t-test: p < 0.05, n = 100). (C) Differences in the hook angles of wild type and mutant embryos at the bent-cotyledon-stage (after two days' imbibition of seeds at 4 • C). The averages and standard errors (SE) are shown. The mutant value is significantly different in comparison with the wild type (Student's t-test: * p < 0.01, n = 75). Pictogram shows the mode of measuring embryo hook angles (C). All experiment was repeated three times.
Thereafter, the individual developmental stages of the embryos were tested to find possible morphological differences between the wild type and mutant embryos. Microscopic images of embryos photographed by a CELL-R Olympus Microscope, are presented in Figure 2. The embryos were isolated from siliques1-11 (where S1-S11 represent silique position numbers from the youngest to oldest) of greenhouse plants of the same age. A sequential shift could be observed in the stages of Atcrk5-1 embryo development from the globular/early heart embryo stages onward (from silique4; S4) when compared to the wild type embryos ( Figure 2). Moreover, in the S10-S11 siliques, the Atcrk5-1 embryos considerably differed in morphology from the wild type ones (Figure 2). to oldest) of greenhouse plants of the same age. A sequential shift could be observed in the stages of Atcrk5-1 embryo development from the globular/early heart embryo stages onward (from silique4; S4) when compared to the wild type embryos ( Figure 2). Moreover, in the S10-S11 siliques, the Atcrk5-1 embryos considerably differed in morphology from the wild type ones ( Figure 2).

Figure 2.
Embryogenesis of the wild type Arabidopsis thaliana Col-0 and the Atcrk5-1 mutant. Bright field microscopic images from a CELL-R Olympus Microscope. Embryo shapes were visualized after chloral hydrate treatment. S1-S11 = silique developmental stages where S1 represents the youngest and S11 the oldest siliques formed after pollination. As compared to wild type (A) embryos, there is a shift in Atcrk5-1 mutant (B) embryo development initiated from the globular/early heart embryo stages found in silique5 (S5). Red arrow represents the start and sequential direction of delay. Note that the wild type (Col-0) bent-cotyledon-stage embryos have 180° hook bending, while the mutant embryos at the same developmental stage (Atcrk5-1) have much less (around 130°) bending angle.
(Scale bars = 10 µm for S1-S7 and 100 µm for S9-S11). Figure 3 indicates quantitative analysis of the progression of embryo development in wild type and Atcrk5-1 siliques. The distribution of sequential embryo developmental phases in the S1-S11 siliques is indicated in percentage by color coding the different stages. These quantitative data indicate that, in the wild type, the S4/S5 siliques contained basically globular, early heart and heart embryos, while in the same type of siliques of the mutant mainly had the eight-cell, globular, and, at a lesser extent, early heart embryos. The wild type S6 siliques mainly consisted of torpedo and early torpedo embryos, while the mutant S6 siliques were dominated by heart-stage embryos but still also contained eight-cell and globular type embryos. The S7 siliques of the wild type had mostly late torpedo and, at a lesser extent, upright embryos, while the mutant S7 siliques still mainly had embryos in the globular and heart (and a few percentage of early torpedo) embryo stages. In the S8/S9 siliques of the wild type, there were mainly torpedo, upright, and walking stick (app. 90° cotyledon bending) stage embryos. In contrast, the mutant S8/S9 siliques had still embryos in the early heart, heart, early torpedo, and torpedo stages with a few percentage of upright ones. The wild type S10/S11 siliques already contained principally walking stick and bent-cotyledon embryos, but the S10/S11 siliques from the mutant only contained upright and walking stick embryos ( Figure 3).

Figure 2.
Embryogenesis of the wild type Arabidopsis thaliana Col-0 and the Atcrk5-1 mutant. Bright field microscopic images from a CELL-R Olympus Microscope. Embryo shapes were visualized after chloral hydrate treatment. S1-S11 = silique developmental stages where S1 represents the youngest and S11 the oldest siliques formed after pollination. As compared to wild type (A) embryos, there is a shift in Atcrk5-1 mutant (B) embryo development initiated from the globular/early heart embryo stages found in silique5 (S5). Red arrow represents the start and sequential direction of delay. Note that the wild type (Col-0) bent-cotyledon-stage embryos have 180 • hook bending, while the mutant embryos at the same developmental stage (Atcrk5-1) have much less (around 130 • ) bending angle. (Scale bars = 10 µm for S1-S7 and 100 µm for S9-S11). Figure 3 indicates quantitative analysis of the progression of embryo development in wild type and Atcrk5-1 siliques. The distribution of sequential embryo developmental phases in the S1-S11 siliques is indicated in percentage by color coding the different stages. These quantitative data indicate that, in the wild type, the S4/S5 siliques contained basically globular, early heart and heart embryos, while in the same type of siliques of the mutant mainly had the eight-cell, globular, and, at a lesser extent, early heart embryos. The wild type S6 siliques mainly consisted of torpedo and early torpedo embryos, while the mutant S6 siliques were dominated by heart-stage embryos but still also contained eight-cell and globular type embryos. The S7 siliques of the wild type had mostly late torpedo and, at a lesser extent, upright embryos, while the mutant S7 siliques still mainly had embryos in the globular and heart (and a few percentage of early torpedo) embryo stages. In the S8/S9 siliques of the wild type, there were mainly torpedo, upright, and walking stick (app. 90 • cotyledon bending) stage embryos. In contrast, the mutant S8/S9 siliques had still embryos in the early heart, heart, early torpedo, and torpedo stages with a few percentage of upright ones. The wild type S10/S11 siliques already contained principally walking stick and bent-cotyledon embryos, but the S10/S11 siliques from the mutant only contained upright and walking stick embryos ( Figure 3). Therefore, embryos in the S11 siliques of the Atcrk5-1 mutant resembled those in the S9 siliques of the wild type, which supported that, at the end of the investigations, the embryo development was delayed approx. by two phases in the mutant as compared to the wild type Col-0 (Figures 2 and 3). Furthermore, the hypocotyl hook of the embryos in the bent-cotyledon-stage of S10/S11 siliques was considerably less closed in the mutant than in the wild type ( Figure 4A). We found significant decrease in the axis length when we compared the SAM-RAM distance of the mutant embryos to that of the wild type Col-0 in the S10-S11 siliques, ( Figure 4B). Therefore, embryos in the S11 siliques of the Atcrk5-1 mutant resembled those in the S9 siliques of the wild type, which supported that, at the end of the investigations, the embryo development was delayed approx. by two phases in the mutant as compared to the wild type Col-0 (Figures 2,3). Furthermore, the hypocotyl hook of the embryos in the bent-cotyledon-stage of S10/S11 siliques was considerably less closed in the mutant than in the wild type ( Figure 4A). We found significant decrease in the axis length when we compared the SAM-RAM distance of the mutant embryos to that of the wild type Col-0 in the S10-S11 siliques, ( Figure 4B). Quantification of embryo axis length differences. Asterisks depict significant differences between the wild type and mutant embryos (Student's t-test: * p < 0.05, n ≥ 50). Scale bars = 100 µm.
The relative expression of genes coding for the transcription factors LEC1, LEC2, FUS3, and ABI3 involved in embryo maturation and dormancy regulation processes were investigated to further Figure 3. Quantitative analysis of embryogenesis in wild type Arabidopsis thaliana (Col-0) and Atcrk5-1 mutant in correlation to silique development. Distribution of the different embryo developmental stages from S1 until S11 depicted in percentage (%). A notable shift in Atcrk5-1 embryo development can be observed from the globular embryo stages in silique 4 (S4). Minimum 50 seeds were investigated for both genotypes for each embryo developmental stage of green-house-grown plants. The experiments were repeated three times with the same results.  . Quantitative analysis of embryogenesis in wild type Arabidopsis thaliana (Col-0) and Atcrk5-1 mutant in correlation to silique development. Distribution of the different embryo developmental stages from S1 until S11 depicted in percentage (%). A notable shift in Atcrk5-1 embryo development can be observed from the globular embryo stages in silique 4 (S4). Minimum 50 seeds were investigated for both genotypes for each embryo developmental stage of green-house-grown plants.
The experiments were repeated three times with the same results.
Therefore, embryos in the S11 siliques of the Atcrk5-1 mutant resembled those in the S9 siliques of the wild type, which supported that, at the end of the investigations, the embryo development was delayed approx. by two phases in the mutant as compared to the wild type Col-0 (Figures 2,3). Furthermore, the hypocotyl hook of the embryos in the bent-cotyledon-stage of S10/S11 siliques was considerably less closed in the mutant than in the wild type ( Figure 4A). We found significant decrease in the axis length when we compared the SAM-RAM distance of the mutant embryos to that of the wild type Col-0 in the S10-S11 siliques, ( Figure 4B). Quantification of embryo axis length differences. Asterisks depict significant differences between the wild type and mutant embryos (Student's t-test: * p < 0.05, n ≥ 50). Scale bars = 100 µm.
The relative expression of genes coding for the transcription factors LEC1, LEC2, FUS3, and ABI3 involved in embryo maturation and dormancy regulation processes were investigated to further Figure 4. Comparison of hypocotyl bending and the SAM-RAM distance of bent-cotyledon-stage embryos. (A) Embryo length was measured from SAM to RAM in silique10 and silique11-derived embryos after preclearing in Hoyer's solution. The SAM-RAM axis is indicated with red line. Note that the axes of the Atcrk5-1 embryos are shorter in comparison to those of the wild type. (B) Quantification of embryo axis length differences. Asterisks depict significant differences between the wild type and mutant embryos (Student's t-test: * p < 0.05, n ≥ 50). Scale bars = 100 µm.
The relative expression of genes coding for the transcription factors LEC1, LEC2, FUS3, and ABI3 involved in embryo maturation and dormancy regulation processes were investigated to further support the delay in embryo development. Small (S1-S4) and medium-sized (S5-S7) siliques of the wild type (Col-0) and mutant (Atcrk5-1) plants were collected and analysed by qRT-PCR. The transcription factors LEC1, LEC2, and FUS3 act at late embryogenesis; therefore, we did not notice gene expression in small siliques. In the medium size siliques (from S5-S7), these genes were significantly less expressed in the Atcrk5-1 mutant than in the control ( Figure 5A-C). The expression level of the ABI3 transcription factor gene was also dropped at each silique stage in the Atcrk5-1 mutant ( Figure 5D). The reduced transcription factor expressions in medium-sized siliques are in accordance with the delayed embryogenesis in the seeds of the Atcrk5-1 mutant.

The Delayed Development of the Atcrk5-1 Embryos Is Linked to Their Decreased Gibberellin Synthesis and Level
The delay in Atcrk5-1 embryo development started at the globular/early heart embryo stages found in silique4 (S4). The rapid expansion of wild type embryos takes place in this period (from S5 till S11) [3]. This was correlated with the elevated levels of bioactive gibberellins GA 1 and GA 3 reaching their maximum at the walking stick embryo stage [3]. Bioactive gibberellins regulate late embryogenesis via modulating DELLA protein level in the embryos. GA deficiency and DELLA protein accumulation both led to abnormal embryos that are characterized by a shortened embryo axis and unbended cotyledons [3]. Based on these observations, the delayed morphogenesis of the Atcrk5-1 embryos might be explained by the reduced level of bioactive GAs. Therefore, a GA-antibody-based Elisa kit measured the total GA content of medium-sized siliques (S5-to-S7) collected from mutant and wild type plants. It was found that the medium-sized siliques of the Atcrk5-1 mutant contained less GA than those of the wild type plants at the same age ( Figure 6A). We found that GA 3 treatment restored the SAM-RAM distance of the mutant embryos to the wild type level at S11-S12 stages when we treated greenhouse-grown wild type Arabidopsis and Atcrk5-1 mutant plants by 20 µM GA 3 from the early development of the inflorescence onward ( Figure 6B). Additionally, the hypocotyl bending phenotype of the Atcrk5-1 mutant embryos found in S10-S12 ( Figure 4) was also restored to the wild type level by the exogenous GA 3 treatment ( Figure 6C). support the delay in embryo development. Small (S1-S4) and medium-sized (S5-S7) siliques of the wild type (Col-0) and mutant (Atcrk5-1) plants were collected and analysed by qRT-PCR. The transcription factors LEC1, LEC2, and FUS3 act at late embryogenesis; therefore, we did not notice gene expression in small siliques. In the medium size siliques (from S5-S7), these genes were significantly less expressed in the Atcrk5-1 mutant than in the control ( Figure 5A-C). The expression level of the ABI3 transcription factor gene was also dropped at each silique stage in the Atcrk5-1 mutant ( Figure 5D). The reduced transcription factor expressions in medium-sized siliques are in accordance with the delayed embryogenesis in the seeds of the Atcrk5-1 mutant.  Table  S1.

The Delayed Development of the Atcrk5-1 Embryos Is Linked to Their Decreased Gibberellin Synthesis and Level
The delay in Atcrk5-1 embryo development started at the globular/early heart embryo stages found in silique4 (S4). The rapid expansion of wild type embryos takes place in this period (from S5 till S11) [3]. This was correlated with the elevated levels of bioactive gibberellins GA1 and GA3 reaching their maximum at the walking stick embryo stage [3]. Bioactive gibberellins regulate late embryogenesis via modulating DELLA protein level in the embryos. GA deficiency and DELLA protein accumulation both led to abnormal embryos that are characterized by a shortened embryo axis and unbended cotyledons [3]. Based on these observations, the delayed morphogenesis of the Atcrk5-1 embryos might be explained by the reduced level of bioactive GAs. Therefore, a GAantibody-based Elisa kit measured the total GA content of medium-sized siliques (S5-to-S7) collected from mutant and wild type plants. It was found that the medium-sized siliques of the Atcrk5-1 mutant contained less GA than those of the wild type plants at the same age ( Figure 6A). We found that GA3 treatment restored the SAM-RAM distance of the mutant embryos to the wild type level at S11-S12 stages when we treated greenhouse-grown wild type Arabidopsis and Atcrk5-1 mutant plants by 20 µM GA3 from the early development of the inflorescence onward ( Figure 6B). Additionally, the hypocotyl bending phenotype of the Atcrk5-1 mutant embryos found in S10-S12 ( Figure 4) was also restored to the wild type level by the exogenous GA3 treatment ( Figure 6C).  Table S1.
We also examined the gene expression levels of the gibberellin 20-oxidases (AtGA20ox2 and AtGA20ox3), which are highly expressed GA biosynthesis enzymes in dry and imbibed seeds [51] and they are responsible for converting the inactive intermediates of GA biosynthesis into bioactive forms [3]. The AtGA20ox2 and AtGA20ox3 genes have high expression during the expansion stage of embryogenesis (4-7 days after pollination, 4-7 DAP) [3]. The expression level of the gibberellin deactivation gene GA2ox4, which also has a high expression level in embryos at this stage (4-7 DAP; [3]), was also tested. Furthermore, the genes of DELLA proteins canonically known to participate in GA signaling during embryogenesis [3,32] were included into the gene expression study. Siliques of wild type (Col-0) and mutant (Atcrk5-1) plants were collected according to their developmental stages, like small (three DAP) and medium (five DAP) ones corresponding to embryo developmental stages S1-S4 and S5-S6-S7, respectively. The gene expression analysis was focused on the medium silique stage, where decreased GA content was observed in the Atcrk5-1 mutant. When considering the genes coding for DELLA proteins, we found a slight gene expression level elevation in RGL1 ( Figure 7A) and GAI ( Figure 7D) genes, while the expression levels of the RGL2, RGL3, and RGA ( Figure 7B,C,E) genes did not change in the Atcrk5-1 as compared to that of the wild type. However, the GA20ox2/GA20ox3 gene expression levels were decreased in the siliques of the Atcrk5-1 mutant in comparison to the wild type ( Figure 7F,G). This indicates that the insufficient conversion of the inactive GA intermediates into their bioactive forms might cause the fall-off in the GA level at the medium silique stage ( Figure 6B). The expression level of the GA2ox4 was also decreased ( Figure 7H) in the Atcrk5-1 medium siliques, which might relate to the reduced negative feedback of GAs on their biosynthesis. We also examined the gene expression levels of the gibberellin 20-oxidases (AtGA20ox2 and AtGA20ox3), which are highly expressed GA biosynthesis enzymes in dry and imbibed seeds [51] and they are responsible for converting the inactive intermediates of GA biosynthesis into bioactive forms [3]. The AtGA20ox2 and AtGA20ox3 genes have high expression during the expansion stage of embryogenesis (4-7 days after pollination, 4-7 DAP) [3]. The expression level of the gibberellin deactivation gene GA2ox4, which also has a high expression level in embryos at this stage (4-7 DAP; [3]), was also tested. Furthermore, the genes of DELLA proteins canonically known to participate in GA signaling during embryogenesis [3,32] were included into the gene expression study. Siliques of wild type (Col-0) and mutant (Atcrk5-1) plants were collected according to their developmental stages, like small (three DAP) and medium (five DAP) ones corresponding to embryo developmental stages S1-S4 and S5-S6-S7, respectively. The gene expression analysis was focused on the medium silique stage, where decreased GA content was observed in the Atcrk5-1 mutant. When considering the genes coding for DELLA proteins, we found a slight gene expression level elevation in RGL1 ( Figure 7A) and GAI ( Figure 7D) genes, while the expression levels of the RGL2, RGL3, and RGA ( Figure 7B,C,E) genes did not change in the Atcrk5-1 as compared to that of the wild type. However, the GA20ox2/GA20ox3 gene expression levels were decreased in the siliques of the Atcrk5-1 mutant in comparison to the wild type ( Figure 7F/G). This indicates that the insufficient conversion of the inactive GA intermediates into their bioactive forms might cause the fall-off in the GA level at the medium silique stage ( Figure 6B). The expression level of the GA2ox4 was also decreased ( Figure 7H Table S1.

The Auxin Level Is Decreased in the Atcrk5-1 Mutant Embryos in Comparison with the Wild Type
Ones.
It is known that there is crosstalk between auxin and GA during plant developmental regulation, including embryogenesis [52,60]. We pointed out a GA shortage in the Atcrk5-1 embryos at the globular/ heart/torpedo embryo stages (in medium siliques S5-S7, Figure 6B). Among others, GA is required for proper auxin transport in Arabidopsis [33]. Therefore, we tested the auxin distribution in the Atcrk5-1 mutant embryos during their development. The DR5::GFP marker (a synthetic promoter that responds to the presence of auxin linked to GFP gene [61]) was introduced into wild type ( Figure 8A) and Atcrk5-1 plants ( Figure 8B). Fluorescence microscopy was used to detect auxinregulated GFP fluorescence in various embryo stages. We found a strong decrease in the DR5::GFP

The Auxin Level Is Decreased in the Atcrk5-1 Mutant Embryos in Comparison with the Wild Type Ones.
It is known that there is crosstalk between auxin and GA during plant developmental regulation, including embryogenesis [52,60]. We pointed out a GA shortage in the Atcrk5-1 embryos at the globular/ heart/torpedo embryo stages (in medium siliques S5-S7, Figure 6B). Among others, GA is required for proper auxin transport in Arabidopsis [33]. Therefore, we tested the auxin distribution in the Atcrk5-1 mutant embryos during their development. The DR5::GFP marker (a synthetic promoter that responds to the presence of auxin linked to GFP gene [61]) was introduced into wild type ( Figure 8A) and Atcrk5-1 plants ( Figure 8B). Fluorescence microscopy was used to detect auxin-regulated GFP fluorescence in various embryo stages. We found a strong decrease in the DR5::GFP signal in the Atcrk5-1 embryos all along from the eight-cell stage until the walking stick stage of the embryos (Figure 8

Expression and Abundance of Auxin Efflux Proteins During Embryogenesis of Wild Type and Atcrk5-1 Mutant Plants
Auxin is known to determine cell division patterns and cell specification during all stages of embryogenesis, starting from the first division of the zygote and this is strongly dependent on PINs localization and activity [4,10,18,60]. The expression and abundance of the auxin efflux transporters PIN1, PIN3, PIN4, and PIN7 were investigated at various embryo developmental stages in order to investigate whether the delayed embryogenesis of the Atcrk5-1 mutant is associated with altered auxin transport pattern or efficiency. The PIN4 and PIN7 gene expressions were reduced in the small silique stage (S1-S4) of the Atcrk5-1 mutant ( Figure 9A,B), while the PIN1 gene expression was down regulated in the mutant siliques during the small and medium stages inclusive the torpedo stage (S1-S4 and S5-S7, Figure 9C).  Table S1. GFP-tagged PIN proteins were produced in wild type (Col-0) and mutant (Atcrk5-1) backgrounds to investigate their protein level and distribution. At early embryogenesis, PIN1 and PIN7 are required for normal embryo development. In the proembryo, PIN1 first localizes non-

Expression and Abundance of Auxin Efflux Proteins During Embryogenesis of Wild Type and Atcrk5-1 Mutant Plants
Auxin is known to determine cell division patterns and cell specification during all stages of embryogenesis, starting from the first division of the zygote and this is strongly dependent on PINs localization and activity [4,10,18,60]. The expression and abundance of the auxin efflux transporters PIN1, PIN3, PIN4, and PIN7 were investigated at various embryo developmental stages in order to investigate whether the delayed embryogenesis of the Atcrk5-1 mutant is associated with altered auxin transport pattern or efficiency. The PIN4 and PIN7 gene expressions were reduced in the small silique stage (S1-S4) of the Atcrk5-1 mutant ( Figure 9A,B), while the PIN1 gene expression was down regulated in the mutant siliques during the small and medium stages inclusive the torpedo stage (S1-S4 and S5-S7, Figure 9C).
GFP-tagged PIN proteins were produced in wild type (Col-0) and mutant (Atcrk5-1) backgrounds to investigate their protein level and distribution. At early embryogenesis, PIN1 and PIN7 are required for normal embryo development. In the proembryo, PIN1 first localizes non-polarly but, at the globular stage, it becomes localized to the basal membrane of provascular cells facing the hypophysis [10]. During early embryogenesis, PIN7 first localizes at the apical side of suspensor cells and at the 32-cell (globular embryo) stage PIN7 polarity changes and the protein becomes associated with the basal membrane of suspensor cells. PIN1 and PIN7 polarization changes correlate with the reversion of the auxin flow [4,10,12].
Auxin is known to determine cell division patterns and cell specification during all stages of embryogenesis, starting from the first division of the zygote and this is strongly dependent on PINs localization and activity [4,10,18,60]. The expression and abundance of the auxin efflux transporters PIN1, PIN3, PIN4, and PIN7 were investigated at various embryo developmental stages in order to investigate whether the delayed embryogenesis of the Atcrk5-1 mutant is associated with altered auxin transport pattern or efficiency. The PIN4 and PIN7 gene expressions were reduced in the small silique stage (S1-S4) of the Atcrk5-1 mutant ( Figure 9A,B), while the PIN1 gene expression was down regulated in the mutant siliques during the small and medium stages inclusive the torpedo stage (S1-S4 and S5-S7, Figure 9C). Figure 9. Relative expression level of auxin efflux carrier PIN genes at different silique stages quantified by qRT-PCR. The relative transcript levels were determined in small siliques (S1-S4, 3DAP) and medium siliques (S5-S7, 5DAP) for (A) PIN4, (B) PIN7, (C) PIN1. Ubiquitin1 (UBQ-1) was used as an internal control. The values are means (+/−) SD of two independent biological repeats. Student's t-test: * p < 0.05, ** p < 0.01, *** p < 0.001. Primers used are listed in the Table S1. GFP-tagged PIN proteins were produced in wild type (Col-0) and mutant (Atcrk5-1) backgrounds to investigate their protein level and distribution. At early embryogenesis, PIN1 and PIN7 are required for normal embryo development. In the proembryo, PIN1 first localizes non- Figure 9. Relative expression level of auxin efflux carrier PIN genes at different silique stages quantified by qRT-PCR. The relative transcript levels were determined in small siliques (S1-S4, 3DAP) and medium siliques (S5-S7, 5DAP) for (A) PIN4, (B) PIN7, (C) PIN1. Ubiquitin1 (UBQ-1) was used as an internal control. The values are means (+/−) SD of two independent biological repeats. Student's t-test: * p < 0.05, ** p < 0.01, *** p < 0.001. Primers used are listed in the Table S1.
The pattern of PIN1/PIN7 distribution seems to be unaffected by the Atcrk5-1 mutation (Figures 10  and 11). Figure 10 shows PIN1-GFP abundance in wild type (A) and Atcrk5-1 (B) embryos from the eight-cell stage to the walking stick embryo stage. The PIN1-GFP signal intensity is lower in the Atcrk5-1 mutant compared to the wild type in all embryogenesis stages, but especially in the early ones (up to the heart stage). Likewise, the PIN7-GFP signal intensity was drastically decreased in the Atcrk5-1 embryos at each embryo stage ( Figure 11). The pattern of PIN1/PIN7 distribution seems to be unaffected by the Atcrk5-1 mutation ( Figures  10 and 11). Figure 10 shows PIN1-GFP abundance in wild type (A) and Atcrk5-1 (B) embryos from the eight-cell stage to the walking stick embryo stage. The PIN1-GFP signal intensity is lower in the Atcrk5-1 mutant compared to the wild type in all embryogenesis stages, but especially in the early ones (up to the heart stage). Likewise, the PIN7-GFP signal intensity was drastically decreased in the Atcrk5-1 embryos at each embryo stage ( Figure 11).   The abundance and distribution of the PIN3 and PIN4 proteins was also investigated. PIN3 is expressed at the basal pole of heart stage embryos during embryo development [10]. For PIN3-GFP, we did not find any abundance or localization alterations in the Atcrk5-1 mutant when compared to the wild type during embryo development ( Figure S1). The PIN4 protein first appears during early embryogenesis at the globular stage in the descendants of the hypophysial cell and in the provascular initials of root meristem [10]. We have found that the PIN4-GFP signal intensity was also decreased in the Atcrk5-1 embryos in comparison to the wild type, like in the cases of PIN1-GFP and PIN7-GFP ( Figure S2).

Expression and Abundance of Auxin Influx (AUX1) Proteins During Embryogenesis of Wild Type and Atcrk5-1 Mutant Plants
Auxin influx is also required for proper embryogenesis. AUX1, LAX1, and LAX2 are required for both shoot and root pole formation of embryos, in cooperation with PIN efflux carriers [17,24]. AUX/LAX protein expressions start from the heart stage in Arabidopsis embryos. We compared the localization pattern of the AUX1::YFP signal in the wild type and mutant (Atcrk5-1) embryos during different phases of embryo development ( Figure 12). We found that the AUX1::YFP expression level was decreased in the Atcrk5-1 mutant in all of the embryo developmental stages in comparison to the wild type. Moreover, the distribution of the AUX1::YFP protein was more restricted at the root pole in the mutant background.

AtCRK5 Can Phosphorylate the Auxin Efflux Proteins PIN1, PIN4 and PIN7 in vitro
The auxin efflux PIN1, PIN3, PIN4, and PIN7 proteins are all supposed to participate in auxin transportation during embryogenesis [17,20,60,62]. The localization and activation of the various PIN proteins are known to be affected by their phosphorylation status [15,[63][64][65]. We have already published that AtCRK5 phosphorylates the PIN2 and PIN3 auxin efflux transporters and this may have a role in the gravitropic response of the Arabidopsis root upon gravistimulation and the hypocotyl hook bending during skotomorphogenesis [57,59]. The decreased abundance of PIN1-GFP, PIN4-GFP, and PIN7-GFP in Atcrk5-1 mutant embryos might reflect a defect in PIN protein stability and/or localization. We speculated that all PIN proteins might be targets of phosphorylation by the AtCRK5 kinase based on the high sequence homology between different PIN hydrophilic loop regions. We cloned the hydrophilic loop domain regions of PIN1, PIN4, and PIN7 into a bacterial protein expression vector in order to support this hypothesis. After transformation and expression into Escherichia coli cells, we purified the GST (Glutathione S-Transferase) tagged GST-PIN1HL and MBP (Maltose Binding Protein) tagged MBP-PIN4HL and MBP-PIN7HL. In the case of GST-PIN1HL, we performed an in vitro radioactive kinase assay while using the purified AtCRK5 protein kinase. Figure 13 shows the result of this phosphorylation assay.
Furthermore, a non-radioactive kinase assay was performed to determine whether the AtCRK5 kinase could phosphorylate the PIN4HL and/or the PIN7HL loop regions. The proteins of the kinase reaction were separated in a denaturing polyacrylamide gel and the GST-PIN1HL, MBP-PIN4HL, and MBP-PIN7HL protein bands were excised and sent for analysis by mass spectrometry after staining with Coomassie dye. AUX/LAX protein expressions start from the heart stage in Arabidopsis embryos. We compared the localization pattern of the AUX1::YFP signal in the wild type and mutant (Atcrk5-1) embryos during different phases of embryo development (Figure 12). We found that the AUX1::YFP expression level was decreased in the Atcrk5-1 mutant in all of the embryo developmental stages in comparison to the wild type. Moreover, the distribution of the AUX1::YFP protein was more restricted at the root pole in the mutant background.

AtCRK5 Can Phosphorylate the Auxin Efflux Proteins PIN1, PIN4 and PIN7 in vitro
The auxin efflux PIN1, PIN3, PIN4, and PIN7 proteins are all supposed to participate in auxin transportation during embryogenesis [17,20,60,62]. The localization and activation of the various PIN proteins are known to be affected by their phosphorylation status [15,[63][64][65]. We have already published that AtCRK5 phosphorylates the PIN2 and PIN3 auxin efflux transporters and this may have a role in the gravitropic response of the Arabidopsis root upon gravistimulation and the hypocotyl hook bending during skotomorphogenesis [57,59]. The decreased abundance of PIN1-GFP, PIN4-GFP, and PIN7-GFP in Atcrk5-1 mutant embryos might reflect a defect in PIN protein stability and/or localization. We speculated that all PIN proteins might be targets of phosphorylation by the AtCRK5 kinase based on the high sequence homology between different PIN hydrophilic loop regions. We cloned the hydrophilic loop domain regions of PIN1, PIN4, and PIN7 into a bacterial protein expression vector in order to support this hypothesis. After transformation and expression into Escherichia coli cells, we purified the GST (Glutathione S-Transferase) tagged GST-PIN1HL and MBP (Maltose Binding Protein) tagged MBP-PIN4HL and MBP-PIN7HL. In the case of GST-PIN1HL, we performed an in vitro radioactive kinase assay while using the purified AtCRK5 protein kinase. Figure 13 shows the result of this phosphorylation assay. Figure 13. AtCRK5 phosphorylates the PIN1 hydrophilic loop in vitro. We performed in vitro radioactive phosphorylation assay with His6-AtCRK5 and two substrates: Myelin Basic Protein (MBP) as a positive control, and GST-PIN1-HL loop. White asterisks indicate the phosphorylation event on MBP (* asterisk) and GST-PIN1HLloop (** asterisk) proteins, respectively. We carried out glutathione S-Transferase (GST) column purification after the kinase reactions shown in the last three columns to remove the HIS6-CRK5 kinase because the His6-CRK5 and the GST-PIN1 protein sizes are nearly identical and we could not distinguish the phosphorylation signals.
Furthermore, a non-radioactive kinase assay was performed to determine whether the AtCRK5 kinase could phosphorylate the PIN4HL and/or the PIN7HL loop regions. The proteins of the kinase reaction were separated in a denaturing polyacrylamide gel and the GST-PIN1HL, MBP-PIN4HL, and MBP-PIN7HL protein bands were excised and sent for analysis by mass spectrometry after staining with Coomassie dye.
We could identify some potential phosphorylation sites for the PIN1, PIN4, and PIN7 hydrophilic loop. AtCRK5 protein kinase can phosphorylate the PIN1 protein at Ser-252 or Ser-253 sites. The phosphorylation site is at Ser-271 in the case of the PIN4 protein. In the case of PIN7 protein, two phosphopeptides were detected in the digest, the Ser-431 and the Ser-277/Ser-278, respectively Figure 13. AtCRK5 phosphorylates the PIN1 hydrophilic loop in vitro. We performed in vitro radioactive phosphorylation assay with His6-AtCRK5 and two substrates: Myelin Basic Protein (MBP) as a positive control, and GST-PIN1-HL loop. White asterisks indicate the phosphorylation event on MBP (* asterisk) and GST-PIN1HLloop (** asterisk) proteins, respectively. We carried out glutathione S-Transferase (GST) column purification after the kinase reactions shown in the last three columns to remove the HIS6-CRK5 kinase because the His6-CRK5 and the GST-PIN1 protein sizes are nearly identical and we could not distinguish the phosphorylation signals.
We could identify some potential phosphorylation sites for the PIN1, PIN4, and PIN7 hydrophilic loop. AtCRK5 protein kinase can phosphorylate the PIN1 protein at Ser-252 or Ser-253 sites. The phosphorylation site is at Ser-271 in the case of the PIN4 protein. In the case of PIN7 protein, two phosphopeptides were detected in the digest, the Ser-431 and the Ser-277/Ser-278, respectively ( Figure S3).

The AtCRK5 Protein Kinase Controls the Gibberellin Level Influencing Seed Size and Embryogenesis
Ca 2+ has a pivotal role as a secondary messenger in many abiotic and biotic stress responses in plants. Environmental cues trigger Ca 2+ concentration alterations in plant cells that are perceived by calmodulin, several calcium-dependent protein kinases (CDPK superfamily members) and calcineurin B-like proteins [66][67][68][69][70][71][72][73]. The Calcium-Dependent Protein Kinase superfamily members (CDPKs) are among the main effectors of Ca 2+ signal transduction in plants [71,72]. Members of the CDPK superfamily-including the Ca-dependent protein kinases (CDPKs), the CDPK-related protein kinases (CRKs), the Ca 2+ and calmodulin-activated kinases (CCaMKs), and the sucrose non-fermenting1-related kinases3 (SnRK3s)-in addition to abiotic/biotic stress responses play important roles in many aspects of the regulation of plant growth and development [57][58][59]67,[72][73][74]. In Arabidopsis, the CRK5 protein kinase is active in most Arabidopsis organs and it is certainly involved in the establishment of the proper auxin gradient necessary for gravitropic response of the root and for proper bending of the hypocotyl hook in Arabidopsis [57,59]. CRK5 is produced in the embryos at all embryo developmental stages and it is localised to the plasma membrane ( Figure S4).
Here, we report a significant difference between the dry seed and embryo sizes of the wild type Arabidopsis thaliana (Col-0) and the Atcrk5-1 mutant (Figure 1). These processes were investigated in more detail in the Atcrk5-1 mutant to clarify the role of the AtCRK5 kinase in seed and embryo development. Microscopical investigation of the characteristic developmental stages of embryos developing within the siliques revealed that embryo development was strongly delayed in the Atcrk5-1 mutant as compared to the wild type from the globular stage onward in the stage4-to-stage11 siliques (S4-to-S11). This embryo developmental period is related to the accumulation of bioactive gibberellins, GA 1 and GA 3 . The accumulation of gibberellic acid in seeds is characterized with two-peaks that correspond to two specific stages of seed development. The first peak is associated with the stage of embryo differentiation when GA promotes cell growth and expansion (until heart embryo stage, S1-S6), and the second GA maximum takes place at the end of the maturation phase when gibberellins trigger proteolytic enzyme activity for the mobilization of endosperm resources for germination [3,4]. Bioactive gibberellins regulate late embryogenesis via modulating the DELLA protein level in the embryos.
GA deficiency and DELLA protein accumulation both lead to abnormal embryos characterized by a shortened embryo axis and unbending cotyledons [3]. This resembles to the phenotype of the late-stage Atcrk5-1 mutant embryos: we also pointed out that the SAM-RAM distances are shorter in the embryos of the Atcrk5-1 mutant as compared to the wild type ones and the cotyledons are unbended ( Figure 4). Therefore, it was supposed that the observed delay in the embryogenesis of Atcrk5-1 might be explained with the shortage or reduced level of bioactive GAs. In agreement, a decreased GA level could be measured in the medium sized (S4-S7) siliques of the Atcrk5-1 mutant in comparison to the wild type. Moreover, the expressions of the key genes of GA metabolism were downregulated in the Atcrk5-1 embryos during this developmental period ( Figure 7F-H). We found that the treatment restored the SAM-RAM distance of the mutant embryos to the wild type level observed at S11-S12 stages when we employed 20 µM GA 3 to the inflorescence of greenhouse-grown wild type Arabidopsis and Atcrk5-1 mutant plants to investigate the effect of exogenous GAs on their embryogenesis ( Figure 6A). Additionally, the delayed bending phenotype of the Atcrk5-1 mutant embryos found in S10-S12 (Figure 4) was also restored to close to normal by the GA treatment ( Figure 6C).
Our results indicate that the accumulation of GAs during the early phase of embryogenesis is missing or limited in the Atcrk5-1 mutant embryos, which probably cause problems in embryo differentiation and delay the later embryo developmental stages (as depicted in Figures 2 and 3).

AtCRK5 is a General Regulator of Auxin Distribution Potentially via the Phosphorylation of Several PINs
Auxin production, signaling, and the establishment of auxin gradients have a central role in embryo development and morphogenesis [4,[9][10][11][12]. When considering the delay in the progression of embryogenesis in the Atcrk5-1 mutant seeds, the distribution of the GFP-expressing auxin sensor DR5, and that of the GFP-tagged auxin transporters PIN1, PIN3, PIN4, PIN7, and AUX1 was also investigated during the whole process of embryogenesis in both phenotypes. We found a much weaker DR5::GFP signal in the Atcrk5-1 embryos than in the control ones in all embryo developmental stages ( Figure 8B), which indicated auxin shortage in the embryos. When the abundance of the auxin efflux transporter PIN proteins PIN1, PIN4, and PIN7 was checked in the Atcrk5-1 embryos, we also observed strong decline in their level. We also found a low auxin influx protein (AUX1::YFP) level in the Atcrk5-1 mutant in all of the studied embryo developmental stages. Auxin itself regulates the expression and stability of its transporters [75,76]. Therefore, a lower level of AUX1 might be the indirect consequence of PIN phosphorylation. The auxin influx carriers AUX1/LAX1/LAX2 have a crucial role in the formation of the embryo shoot and root poles, and they act redundantly to specify embryonic root and shoot pole identity and development [17,24], so any malfunction of these auxin transporters results in morphological changes in embryo shape. Actually, we found a reduction in the SAM-RAM distances of the Atcrk5-1 embryos at the end of the embryogenesis (in S11-S12; Figures 4 and 5), which is probably the consequence of the impaired cooperation of the auxin efflux and influx carrier's PINs/AUXs in the absence of the AtCRK5 protein kinase. The auxin-dependent polarization of embryos necessitates balances of efflux and influx auxin transport mechanisms [17].
The phosphorylation status of these transporters determines the abundance and asymmetrical distribution of PIN proteins in the plasma membrane, which in turn affects the polar auxin transport [77]. Various kinases are known to control the distribution of auxin through the phosphorylation of the PIN efflux carriers or the ABC transporters [78]. Among others, the D6PROTEIN KINASE and PINOID/WAG AGCVIII kinases are types of kinases that are known to activate the PIN-mediated auxin efflux [65]. Previously, we have shown that the AtCRK5 kinase can phosphorylate in vitro the hydrophilic loops of various PIN proteins, like PIN2 and PIN3 [57,59]. Based on in vitro kinase assays and phospho-peptide analysis data, here we also report the possible phosphorylation sites of AtCRK5 in PIN1, PIN4, and PIN7. We assume that these transporters are also the targets of the AtCRK5 kinase. The AtCRK5-mediated phosphorylation of PIN proteins might increase their stability and maintain their optimal abundance at the various embryo stages. The role of AtCRK5 in the regulation of auxin transport controlling PIN protein abundance has been described during root gravitropic responses [57] and hypocotyl bending [59]. Our results suggest that AtCRK5 can have a general role maintaining the stability of several polar auxin transport (PAT) proteins, thus affecting auxin transport efficiency during various developmental processes, including embryogenesis, skotomorphogenesis, and gravitropism.

The AtCRK5 Protein Kinase is Involved in Hormonal Crosstalk Influencing Embryogenesis
Several data suggest that there is crosstalk between auxin transport and GA signaling [31,33,50,79]. It is already known that auxin promotes GA biosynthesis via the regulation of the expression of the two enzymes AtGA3ox and AtGA20ox being responsible for the rate limiting steps of GA biosynthesis [3,50]. This control mechanism is independent of the feedback regulation mediated by GAI and RGA DELLA proteins, and it involves Aux/IAA and ARF signaling elements [4,50]. It has also been pointed out that impaired auxin transport resulted in the stabilization of the DELLA repressor RGA (REPRESSOR OF ga1-3) of the GA pathway in Arabidopsis [80]. Therefore, the limited auxin transport in the Atcrk5-1 embryos might be responsible for the observed reduction in their GA synthesis/level. Gibberellin shortage was shown to result in the formation of abnormal embryos at the S10-S11 silique stages with shortened embryo axis and unbending cotyledons [3]. The same characterizes the embryos of the Atcrk5-1 mutant at this developmental stage. Abnormal embryo development of Atcrk5-1 could be rescued by exogenous GA treatment (Figure 6), which indicated the primary role of GA shortage in the embryo phenotype of the mutant. Interestingly, we previously found an unclosed hypocotyl hook phenotype of Atcrk5-1 seedlings during skotomorphogenesis, which might be explainable by the fact that the Atcrk5-1 mutant embryos had already displayed reduced hypocotyl bending during their late embryogenesis ( Figures 1, 2 and 4 and [59]).
It was reported that the GA signaling pathway interacts with PIN-protein-dependent auxin transport in embryo cotyledon development and root gravitropism [33]. Auxin transport impairment was observed in the GA biosynthesis and signaling mutant ga1, which was explained by a decrease of PIN auxin efflux carrier abundance [33]. GA deficiency resulted in targeting PIN2 for vacuolar degradation, leading to reduction of auxin transport. The PIN protein levels could be restored in ga1 to the wild type level by adding exogenous GA. Therefore, the auxin transport impairment in the Atcrk5-1 embryos might be related to the impaired GA biosynthesis, resulting in GA deficiency, similarly as it was found in [33].
Our results suggest that, AtCRK5 can also govern the embryo development in Arabidopsis through the fine tuning of auxin-GA levels via the stability/abundance of the polar auxin transport (PAT) proteins in addition to its regulatory role in root gravitropic responses and hypocotyl bending [57,59], as indicated in Figure 14. of PIN auxin efflux carrier abundance [33]. GA deficiency resulted in targeting PIN2 for vacuolar degradation, leading to reduction of auxin transport. The PIN protein levels could be restored in ga1 to the wild type level by adding exogenous GA. Therefore, the auxin transport impairment in the Atcrk5-1 embryos might be related to the impaired GA biosynthesis, resulting in GA deficiency, similarly as it was found in [33].
Our results suggest that, AtCRK5 can also govern the embryo development in Arabidopsis through the fine tuning of auxin-GA levels via the stability/abundance of the polar auxin transport (PAT) proteins in addition to its regulatory role in root gravitropic responses and hypocotyl bending [57,59], as indicated in Figure 14. Figure 14. Hypothetical model about the involvement of the AtCRK5 protein kinase in the coordination of embryo morphogenesis in Arabidopsis thaliana. The auxin transport proteins having central role in the establishment of auxin gradients controlling embryo morphogenesis are potential targets of AtCRK5. The phosphorylation of these transporters by AtCRK5 increases their stability. In the absence of AtCRK5 their level is decreased without affecting their distribution pattern. The auxin gradients therefore are properly established in the Atcrk5-1 mutant but at lower auxin maxima in the corresponding regions. The lower auxin concentration in these regions limits gibberellin synthesis, via the decreased expression of key genes of gibberellin metabolisms. The lower GA level results in lower expression of the genes coding for auxin transport proteins strengthening the phenotype. An auxin-transport-independent effect of the AtCRK5 kinase on gibberellin metabolism cannot be excluded. In summary, in the absence of AtCRK5, embryogenesis progresses at correctly organised but lower auxin and gibberellin levels delaying but not preventing the processes of morphogenesis.

Plant Material and Growth Conditions
All of the plants used in this study are in Arabidopsis thaliana (L.) Columbia-0 ecotype (Col-0) background. The Atcrk5-1 mutant has been described previously [57,58]. The auxin sensor DR5::GFP  The phosphorylation of these transporters by AtCRK5 increases their stability. In the absence of AtCRK5 their level is decreased without affecting their distribution pattern. The auxin gradients therefore are properly established in the Atcrk5-1 mutant but at lower auxin maxima in the corresponding regions. The lower auxin concentration in these regions limits gibberellin synthesis, via the decreased expression of key genes of gibberellin metabolisms. The lower GA level results in lower expression of the genes coding for auxin transport proteins strengthening the phenotype. An auxin-transport-independent effect of the AtCRK5 kinase on gibberellin metabolism cannot be excluded. In summary, in the absence of AtCRK5, embryogenesis progresses at correctly organised but lower auxin and gibberellin levels delaying but not preventing the processes of morphogenesis.
When it was necessary, the imbibed seeds were transferred onto plates containing half strength Murashige and Skoog medium (1/2 MS) with 0.5% sugar (Molar Chemicals Kft, Hungary), 0.8% phytoagar (Duchefa Biochemie, Haarlem, The Netherlands), pH: 5.7. After seed transfer (AST), the plates were kept in white light for 5 h to stimulate and synchronize seed germination The plates were put into short day condition (SD, 8h light/16h dark cycle, 22 • C, 100 µmol photons m −2 s −1 light intensity). For GA rescue experiments, the developing siliques on the inflorescences of one-month-old greenhouse plants were sprayed once in every day with 20 µM GA 3 (Phylaxia, Budapest, Hungary) until silique number 14 (S14) was obtained.

Seed/Embryo Size/Axis Determination, Embryo Isolation and GA Rescue Experiments
The wild type Col-0 and Atcrk5-1 seeds were sterilized in a standard manner and stored in vials for two days at 4 • C for seed/embryo size determination. 100 µL seeds from each vial were taken out onto a slide and covered with cover slips. Seed sizes (perimeter) in the intact seeds were quantified in wild type (Col-0) and mutant (Atcrk5-1) lines after two days of imbibition at 4 • C by photographing the seeds with a stereo microscope Nikon SMZ800 (Mitsubishi Corp., Minato, Tokyo, Japan) attached to a CCD camera (Nikon Cool pix 995, Japan). The photographs were then evaluated with the ImageJ software (NIH, Bethesda, MD, USA; (https://imagej.net/Fiji/Downloads)). The average values were taken from minimum 100 independent seeds from the wild type and mutant genotypes, respectively.
Embryo axis length, as measured from shoot apical meristem (SAM) until root apical meristem (RAM), was calculated in bent-cotyledon-stage embryos that arose from the previously described experiment. ImageJ measured at least 100 embryo axes. The differences in hook angles of wild type Col-0 and mutant Atcrk5-1 embryos after two days of imbibition at 4 • C were also measured. The bent-cotyledon-stage embryos were carefully pressed out from the seed coat by forceps under a stereo microscope Nikon SMZ800 (Mitsubishi Corp., Minato, Tokyo, Japan) and they were photographed by a CCD camera (Nikon Cool pix 995, Japan). Images were evaluated by the ImageJ software (NIH, Bethesda, MD, USA). The apical hook angles were defined according to the inset in Figure 1C. At least 100 wild-type and mutant seedlings were finally monitored in three biological repeats. Student's t-test was used for statistical analysis for all quantitative measurements. All of the experiments was repeated three times.

Total GA Measurement by Competitive GAs Elisa Assay
The septum bearing the seeds without valves that originated from medium sized siliques (S5-S7 from the top of inflorescence) were collected from six weeks old greenhouse grown wild type (Col-0) and mutant (Atcrk5-1) plants in the case of total GA content measurement from siliques arised seeds. The seeds were dissected from the siliques under stereo microscope Nikon SMZ800 (Mitsubishi Corp., Minato, Tokyo, Japan) and they were collected into Eppendorf tubes on dry ice. We used Retsch (Haan, Germany) tissuelyzer for grind the samples. The grounded samples were incubated overnight at 4 • C in 10mL of 80% methanol solution. After centrifugation with 12,000 rpm, 10 min. at 4 • C, the supernatants were transferred to new tubes, and they were completely lyophilized. We added 200 µl PBS to each sample and loaded 50 µl/well. Three technical repetitions were used. For ELISA measurement, we followed the protocol of the kit (ELISA Kit for Gibberellic Acid (GA), CEA759Ge, Cloud Clone Corp. Wuhan, China). Three biological repeats were performed for total GA content evaluation.

Embryo Morphology Monitored by Cell-R Microscopy
Twelve-twelve siliques (S1-S11/12 where S1 represents the youngest and S12 the oldest siliques after pollination; silique position number) with various developmental stages of embryos were collected from wild type Col-0 and Atcrk5-1 inflorescence of greenhouse plants. The seeds were dissected from siliques and, after preclearing-depending on the age of the siliques-they were incubated in Hoyer's solution for 4 h (for small siliques) or overnight (for medium and large siliques) [86]. Subsequently, seed developmental stages were determined by differential interference contrast microscope (Cell-R, Olympos Corporation, Tokyo, Japan).
For GA rescue experiments, the developing siliques on inflorescences of one-month greenhouse plants were sprayed once in every day with 20 µM GA 3 (Phylaxia, Budapest, Hungary) until silique number 14 (S14) was obtained. Afterwards, siliques were harvested and handled as described above.

Monitoring of Abundance of GFP/YFP Signals in Embryos by LSM Microscopy
The seeds of the GFP/YFP tagged lines of wild type Col-0 and Atcrk5-1 plants of various developmental stages were collected into 10% (v/v) glycerol from several silique stages from developing inflorescences of plants. Embryos were cautiously pressed out from the seeds by using a mild pressure of the glass cover. Afterwards, the abundance/distribution of GFP/YFP signals in the case of various stages of embryogenesis for the auxin sensor DR5::GFP, the auxin transporters (PIN1-GFP, PIN3-GFP, PIN4-GFP, PIN7-GFP, and AUX1::YFP), and the AtCRK5-GFP was analysed while using Olympus FV1000 confocal laser scanning microscopy (Tokyo, Japan). The fluorescence signals were monitored according to [57][58][59]. Generally, 5-5 embryos were investigated for both categories (wild type and mutant) from three independent experiments. The images were prepared while using the Adobe Photoshop and Illustrator software (Adobe Systems Incorporated, San Jose, CA, USA).

RNA Isolation and Real Time Quantitative PCR (qRT-PCR) for Embryo Gene Expression
The isolation of RNA was performed with a slightly modified CTAB-LiCl extraction method of [87]: 100 mg material was collected from 45-days-old green-house-grown plants from small (S1-S4), medium (S5-S7), and large (S8-S10) siliques of Col-0 and Atcrk5-1, respectively. cDNA synthesis of 1 µg of total RNA was carried out in 20 µL using RevertAid M-MuLV Reverse Transcriptase (Thermo Fischer Scientific, Waltham, MA, USA). Quantitative real time PCR (qRT-PCR) was carried out applying the SYBR Green master mix (Applied Biosystems, Thermo Fischer Scientific, Vilnius, Lithuania) with the ABI 7900 Fast Real Time System (Applied Biosystems, Thermo Fischer Scientific, Vilnius, Lithuania) while following the next protocol: 45 cycles at 95 • C for 15 s, followed by 60 • C for 1 min. The 2 −∆∆Ct method calculated the normalized relative transcript levels [88]. Reactions were made in triplicates and minimum two independent biological repetitions were performed. Ubiquitin-1 was used as an endogenous control. Table S1 lists the set for qRT PCR primers used in this study.

PIN4 and PIN7 Hydrophilic Loop Region Cloning
We amplified the corresponding cDNA sequences from Arabidopsis thaliana cDNAs applying the high fidelity Phusion polymerase (Thermo Fischer Scientific, Vilnius, Lithuania) while following the manufacturer's instructions ( Figure S5). Table S1 lists the specific primers used. After amplification, the BamHI-EcoRI enzyme-digested cDNA fragment was ligated into the pBluescript II SK plasmid. Sequencing justified the error free cloned fragments. The PIN1HL loop fragment was moved into the pGEX4T1 (Amersham, Little Chalfont, Great Britain) protein expression vector, while the PIN4HL and PIN7HL loop fragments were moved into the pMALp2 (New England Biolabs, Ipswich, MA, USA)

Statistical Analysis
The experiments were carried out with two or three independent biological repetitions, as indicated. Sample numbers are given for each experiment in the text/figure legends. The data are presented as the mean ± standard error (SE) calculated from the combined data of biological repetitions. Differences of the mutant from the wild type control were determined by Student's t-test and the significant differences were represented, as follows: * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.