Direct evidence that suspensor cells have embryogenic potential that is suppressed by the embryo proper during normal embryogenesis

Significance The suspensor is a temporary structure that undergoes programmed cell death during seed maturation. It has been suggested that suspensor cells have embryogenic potential that is suppressed by the embryo. Using an established in vivo living cell laser ablation system, we confirmed the embryogenic potential of the Arabidopsis suspensor and the role of the embryo proper in imposing suspensor cell identity. We also showed that auxin redistribution in suspensor cells after laser ablation of embryos may play an essential role in the initiation of suspensor embryos. The suspensor is a temporary supporting structure of proembryos. It has been proposed that suspensor cells also possess embryogenic potential, which is suppressed by the embryo as an effect of the embryo–suspensor interaction. However, data to support this hypothesis are not yet available. In this report, using an in vivo living cell laser ablation technique, we show that Arabidopsis suspensor cells can develop into embryos after removing the embryo proper. The embryo proper plays a critical role in maintaining suspensor cell identity. However, this depends on the developmental stage; after the globular embryo stage, the suspensors no longer possess the potential to develop into embryos. We also reveal that hypophysis formation may be essential for embryo differentiation. Furthermore, we show that, after removing the embryo, auxin gradually accumulates in the top suspensor cell where cell division occurs to produce an embryo. Auxin redistribution likely reprograms the fate of the suspensor cell and triggers embryogenesis in suspensor cells. Thus, we provide direct evidence that the embryo suppresses the embryogenic potential of suspensor cells.

The suspensor is a temporary supporting structure of proembryos. It has been proposed that suspensor cells also possess embryogenic potential, which is suppressed by the embryo as an effect of the embryo-suspensor interaction. However, data to support this hypothesis are not yet available. In this report, using an in vivo living cell laser ablation technique, we show that Arabidopsis suspensor cells can develop into embryos after removing the embryo proper. The embryo proper plays a critical role in maintaining suspensor cell identity. However, this depends on the developmental stage; after the globular embryo stage, the suspensors no longer possess the potential to develop into embryos. We also reveal that hypophysis formation may be essential for embryo differentiation. Furthermore, we show that, after removing the embryo, auxin gradually accumulates in the top suspensor cell where cell division occurs to produce an embryo. Auxin redistribution likely reprograms the fate of the suspensor cell and triggers embryogenesis in suspensor cells. Thus, we provide direct evidence that the embryo suppresses the embryogenic potential of suspensor cells.
suspensor | cell-to-cell interaction | embryogenesis | auxin | in vivo laser ablation T he suspensor is traditionally believed to be a supporting structure during plant embryo development that pushes the embryo proper into the endosperm cavity and connects it to the surrounding maternal and endosperm tissues to facilitate the transfer of nutrients and plant hormones. Therefore, it is supposed to be critical for the early development of the embryo (1)(2)(3)(4). The suspensor cells have characteristics of transfer cells (e.g., Phaseolus and Stellaria) (5,6) and are known to absorb and transport metabolites from the surrounding material tissue to the embryo proper (Phaseolus and Arabidopsis) (7,8). The cells are characterized by specialized organelles (Phaseolus Pisum, Ipomoea, Stellaria, and Tropaeolum), the polytene chromosome (Phaseolus and Alisma), and higher levels of RNA and protein synthesis compared with the embryo proper during the early development of the embryo, as well as active transcription (Phaseolus and Tropaeolum) (6,9). Suspensors in some plant species (Phaseolus, Tropaeolum, Cyfisus, and Arabidopsis) can synthesize or transport gibberellin or auxin, which promotes the development of the embryo proper during early embryogenesis (2,7,9).
Suspensors are temporary structures that degenerate after the globular embryo stage via programmed cell death; they do not participate in the development of later embryos, excluding hypophysis formation in Arabidopsis (10,11). Once a suspensor is formed, cells no longer divide and the cell morphology is highly specialized with features that are distinct from those of embryo cells. However, based on experimental data, an hypothesis was developed in the 1970s suggesting that suspensor cells still possesses embryogenic potential and may develop into an embryo if relieved from suppression by the embryo proper (12)(13)(14)(15). Based on radiation or acid treatment of the siliques or ovules, some pioneering studies showed that the active dividing embryo is more seriously injured than the highly differentiated suspensor, and a second embryo may be observed after several days of ovule culture. However, the exact origin of the second embryo has remained unclear and whether the radiation or the acid treatment leads to gene mutation in the suspensor cells has remained unknown (16)(17)(18).
Phenotypes of some mutants suggest that the embryo proper suppresses the developmental potential of the suspensor. When the embryo proper is abnormal, the suspensor cells can start dividing. Some mutant suspensors can develop into proembryos (e.g., twin, twin2, iyo) or globular embryo-like structures (e.g., SUS, raspberry). These data provide circumstantial evidence for this hypothesis; however, the role of the gene mutation itself or the abnormal embryo in triggering suspensor cell division requires further study (7,12,13,15,19,20). In addition, some available data are contradictory. Expressing LTP1::DTA (LTP1, lipid transfer protein 1; DTA, diphtheria toxin A chain) in the protoderm domain of embryos (21) to injure the embryo protoderm can induce excessive divisions in the suspensor (22). However, expression of the LTP1driven BARNASE (Bacillus amyloliquefaciens ribonuclease) gene in the embryo, which results in defects in the basal tier of the embryo and hypophysis, leads to no signs of deregulation of suspensor development (23). Thus, it remains unclear whether the suspensor-embryo interaction plays a role in determining cell fate transition of the suspensor.

Significance
The suspensor is a temporary structure that undergoes programmed cell death during seed maturation. It has been suggested that suspensor cells have embryogenic potential that is suppressed by the embryo. Using an established in vivo living cell laser ablation system, we confirmed the embryogenic potential of the Arabidopsis suspensor and the role of the embryo proper in imposing suspensor cell identity. We also showed that auxin redistribution in suspensor cells after laser ablation of embryos may play an essential role in the initiation of suspensor embryos.
The living cell microdissection system is a powerful approach for studying cell-cell interactions and testing the factors that may determine cell fate (24)(25)(26)(27)(28). Separation of the suspensor and embryo using this technique may be used to observe developmental fate in vivo and thus provide direct evidence of the suspensor-embryo interaction. It can also be used to determine whether suspensor cells possess embryogenic potential and whether the embryo suppresses that potential during embryogenesis. However, the embryo is embedded deeply in the ovule tissues, and there is no suitable technique for laser ablation of embryos in vivo at this time. Therefore, we attempted to establish such a technique. We applied this technique to confirm the embryogenic potential of suspensor cells and determine whether the embryo suppresses the potential during embryogenesis.

Results
Establishment of the in Vivo Living Cell Laser Microdissection Technique to Precisely Remove the Connection Between Embryo and Suspensor in Situ in Ovules. We first constructed the suspensor cell marker lines by transformation using ZC1::NLS:GFP. Several lines were selected with clear GFP signals observed in suspensor cells. In these lines, every cell of the suspensor could be observed in the ovule during the developmental process of the embryo (Fig. S1 A-E1). Based on the marker lines, we established an in vivo laser microdissection technique by optimizing the different conditions and carefully examining cut intensity, radius, and total time for the optimal conditions. This technique allows precise and efficient ablation of any cells of a suspensor to remove the connection between suspensor and the embryo proper at different positions and at different stages of living embryos.
The laser beam usually targets the GFP signal of the uppermost suspensor cell ( Fig. S2 A and A1) or the second cell from the top of a suspensor. Ablation of the chosen cell can be traced under a microscope during the manipulation process. To improve the efficiency and propriety of the ablation, microdissection was usually stopped immediately once the fluorescence of the ablated suspensor cell faded (Fig. S2B).
We also established a semi-in vivo embryogenesis system based on previous work in ovule culture (29). The morphology and structure of the embryos in cultured ovules were well developed and morphologically similar to their counterparts in vivo ( Fig. S3 A-E). The time course of the embryo development in cultured ovules is also shown in Fig. S3. After a 5-d culture, the embryos developed into torpedo stage (Fig. S3 F and G), and they could further develop into mature embryos in few days. The auxin distribution and the WOX5 (WUSCHEL related homeobox 5) expression in these embryos were identical to that of embryos developed in vivo ( Fig. S3 F-G1), indicating their physiological similarity. These experiments proved the reliability of the system for further experiments (Fig. S3H).
Based on the culture system, the effect of ablation can be examined after a brief culture of ovules. Under our conditions, the connection between the suspensor and embryo proper was obviously broken in 63.64% (n = 77) of ovules (Fig. S2C). The first suspensor cell that was not obviously ablated or its viability recovered later during culture was also observed in 12.98% of ovules under the same conditions of ablation, probably due to the physiological and structural variation in different ovules. A total of 23.38% ovules shrank during our brief culture due to unknown reasons and were not used for evaluation.
After the microdissection and culture, 11.00% (n = 618) of ovules showed clear cell-division patterns of the suspensor. Compared with results for established Arabidopsis ovule culture systems without laser ablation, this percentage is quite high and sufficient for further analysis.

Suspensor Cells Could Develop into Secondary Embryos After Breaking
the Connection Between the Suspensor and Embryo. To determine the exact stages of embryonic development, we first investigated the time course of embryogenesis with reference to pollination time. Under our conditions, 28 h after pollination, 85.93% (n = 64) of embryos were at the two-celled embryo stage with an apical cell and a basal cell (Fig. S4). About 48 h after pollination, 88.46% (n = 78) of embryos were at the eight-celled stage with a four-or six-celled suspensor (Fig. S4). About 56 h after pollination, 77.91% (n = 86) of embryos were at the 16-celled embryo stage with a 7-or 8-celled suspensor (Fig. S4). Approximately 72 h after pollination, 85.57% (n = 97) of embryos were at the 32-celled embryo stage, and the suspensors still contained 7-8 cells (Fig. S4). About 96 h after pollination, 93.15% (n = 73) of embryos were at the heart stage, and the suspensor cell number had not changed (Fig. S4). This indicates that the suspensors already formed and the cells did not divide beginning at the 16-celled embryo stage (56 h after pollination).
Based on our results, three stages of embryos were chosen for microdissection with the in vivo laser ablation technique to break the connection between suspensor and embryo: 8-celled embryos (48 h after pollination), 32-celled embryos (72 h after the pollination), and heart-stage embryos (96 h after pollination). After the laser ablation ( These results confirm that, after the embryo-suspensor connection is removed, suspensors at stages before the globular embryo have embryogenic potential, despite the fact that the suspensor cell is still dividing (8-celled embryo) or has stopped division and differentiation (32-celled embryo). Thus, we confirmed the embryogenic potential of suspensor cells (at least in some developmental stages). Suspensors at stages after the heart-embryo stage no longer have embryogenic potential, although the embryo proper is similarly removed, indicating that suspensor cells at only some early stages (not all stages) possess embryogenic potential. Our data also confirm that the embryogenic potential is suppressed by the embryo proper because the second embryo generated from the suspensor occurs only after removing the embryo proper.

Developmental Prospects of Embryos at Different Stages After
Disconnection from the Suspensor. The in vivo embryo laser ablation system allows us to investigate the influence of the suspensor on the development of the embryo at different stages in the actual microenvironment in a seed. We chose to observe embryonic behavior at three stages of development after laser ablation of the suspensor. After breaking the connection between the embryo proper and the suspensor of eight-celled embryos (Fig. 1B) and after a 3-d culture, 11.69% (n = 77) of the embryos arrested ( Fig. 1C), about 50.65% (n = 77) divided into globular embryo-like structures (Fig. 1D), 20.78% (n = 77) developed into cell clusters with an irregular shape, and 16.88% (n = 77) became brownish (Fig. 1E). After a 7-d culture, only 3.51% (n = 57) of the embryos developed into heart embryo-like structures. Most of the embryo still arrested at the globular embryo stage (Fig. 1C). After breaking the connection in 32-celled embryos (Fig. 1G), 73.81% (n = 84) of the embryos divided into globular embryos after a 3-d culture (Fig.  1H), 5.95% (n = 84) arrested at this stage, 8.33% (n = 84) developed into irregular structures, and 11.90% (n = 84) became brownish. After a 5-d culture, 57.81% (n = 64) of the embryos developed into heart-stage embryos (Fig. 1I). After cutting off the suspensor (Fig. 1J), 90.91% (n = 43) of the heart-stage embryos developed into torpedo embryos after culture for 3 d (Fig. 1K). Development of the 8-and 32-celled embryos without suspensors was slower than that of embryos with suspensors, especially for the 8-celled embryos. However, the developmental speeds of the heartstage embryos with or without suspensors were similar.
These results show that the suspensor is critical for the early development of embryos. Embryos older than the 16-celled stage could independently develop in ovules without a suspensor, indicating that nutrient transportation by the suspensor (33) is no longer absolutely required for embryonic development at this stage.
Remaining Suspensor Cells Attached to Embryos After Laser Ablation Cannot Generate New Suspensors but May Ensure Hypophysis Formation.
During normal embryogenesis, the uppermost suspensor cell develops into a hypophysis and finally becomes incorporated into the embryo proper. Signaling between two domains (the suspensor and the embryo proper) connected by the hypophysis is required for precise patterning of the embryo's apical-basal axis (34). Our laser microdissection technique created embryos with only the uppermost suspensor cell or with the top two suspensor cells. This allowed us to observe the fate of these remaining suspensor cells and to observe their influence on embryonic development.
This manipulation is difficult because we need to ensure that the remaining suspensor cells are intact and living to study their developmental behavior. Based on our conditions, embryos with some intact suspensor cells were obtained in 27.66% (n = 94) of ovules at the eight-celled embryo stage (Fig. 1B), and the uppermost cell of the suspensors were precisely removed in 28.72% (n = 94) of ovules at this stage (Fig. 1C). This allows for a comparison between the two types of material. Globular embryos with one or two suspensor cells were obtained in 34.25% (n = 73) of ovules (Fig. 1I), and precise removal of all suspensor cells was successfully performed in 36.99% (n = 73) of ovules at this stage (Fig. 1G).
When there are some remnant cells of suspensor, the embryo is more likely to develop normally ( Fig. 1I; Fig. 2 D-F) and the apical-basal axis can be established, as in nonablated embryos, no matter how the embryos are oriented in the seeds (Fig. 1 I and  K). As a reference, the original suspensor represents the original apical-basal orientation (Fig. 1K). Development of the hypophysis in these embryos was similar to embryos without laser ablation (n = 26; Fig. 1 I and K; Fig. 2 A-F and L), indicating that the intact suspensor is not required for hypothesis differentiation, but is influenced by their attachment to embryonic cells. However, when the top cell of the suspensor was ablated, embryos developed further but did not generate a hypophysis from adjacent embryo cells (n = 56; Fig. 2 G-I and M). We typically observed an absence of basal structure in these embryos (Fig. 2 G-I). Thus, unlike that in embryos with suspensor cells (Fig. 2F), in these embryos, the WOX5::GFP signal representing hypophysis (Fig. 2 J and K) was never observed. Consistent with this result, cell organization in the basal end of the embryos differed from that of normally developed embryos, and only cells derived from the embryo proper were observed around the absent hypophysis region (Fig. 2M). These embryos lacking the primary embryonic root structure were usually arrested in the globular stage (n = 34; Fig. 1D; Fig. 2 I and J), suggesting that the top suspensor cell may play a critical role in embryo differentiation.
When one or two suspensor cells remained attached to the embryos after laser ablation, the suspensor never regenerated into a new and intact suspensor (n = 56) during culture under the same conditions. Auxin Accumulates at the Free End of the Suspensor After Laser Ablation.
We used DR5rev::GFP as the auxin marker to investigate the distribution of auxin after microdissection and during culture of the suspensors and embryos. Proembryos in vivo showed different levels of auxin distribution in the suspensor and embryo domains. The auxin signal was observed mainly in embryos and rarely in suspensors ( Fig. 3 A and A1). After laser ablation of the embryo, the auxin in the suspensor was gradually redistributed and accumulated mainly in the first suspensor cell after a 12-h culture. Later, auxin accumulation showed an obvious polar gradient from the free end of the suspensor to the bottom end after a 24 h culture (Fig. 3 B and B1;  Fig. S7), and at the same time the top cell started to divide, which functioned as a novel apical cell. This suggests that auxin redistribution may play a critical role in the cell-fate transition from the suspensor cell to an initial embryo cell.

Early Division Pattern of the Top Suspensor Cell After Laser Ablation
May Mimic the Original Apical Cell. To compare the top suspensor cell and original apical cell during embryo generation, we observed their cell division patterns. During early embryogenesis, one of the most important events is the orientation of the first cell division, which usually determines the developmental fate of the cell, especially in apical or basal cells. In Arabidopsis, the division of apical cells is vertical and the division of basal cells is transversal.
We found that the first vertical cell division usually occurred in the uppermost cell of the suspensor, which marks the cell differentiated into an embryo (Fig. 3 D and D1). The division could be modified because the top suspensor cell was irregularly shaped (Fig. 3 E and  E1). When the cytoplasm of the top suspensor cell was thin after the laser ablation and remained undivided, the second suspensor cell acted as the initial embryo cell. In this case, the first vertical cell division was observed in this embryogenic cell (Fig. 3 F and F1), mimicking the behavior of the original apical cell. However, after the first cell division, subsequent cell divisions did not exactly follow the stationary pattern of the first embryo. This may be due to the irregular cell shape compared with the original apical cell of the proembryo.
In Vitro Culture of Suspensors Detached from the Embryo Does Not Generate Embryos. Next, we further examined whether polar auxin transport plays an important role in triggering embryogenic development of the suspensor by an in vitro culture system. We cultured the isolated suspensors of both Arabidopsis (Fig. 4 A and  B) and Brassica napus (Fig. 4 C and D) because we have efficient in vitro embryo culture systems for Brassica, which can be used as a control for our evaluation of the embryogenic potential of suspensor cells. The results showed that isolated suspensors without connections to the embryo proper did not generate new embryos. The suspensors typically shrank and ultimately degenerated after a 7-d culture (Fig. 4 E and F) (n = 87).
We also examined different concentrations and combinations of plant hormones in the culture medium. We found that the suspensor cells developed into cell clusters and callus in the presence of 1.0 mg/L 2, 4-D and 0.3 mg/L kinetin (Fig. 4 G and H), but did not develop into embryos in any media that we examined. Hence, the embryogenic potential of suspensor cells was observed only in vivo and not in vitro when the suspensor was detached from both the embryo and the maternal tissue, although auxin was present in the growth environment. Thus, we propose that auxin polar transportation from maternal tissue to the suspensor cell might be essential for the generation of a second embryo.
We then cultured the laser-ablated ovules in the medium containing the auxin polar transport inhibitor NPA (N-1-naphthylphthalamic acid). After a 5-d culture, 61.54% of the suspensors became brownish and shrank, 30.77% arrested their development, and only 7.69% showed irregular cell division (n = 39). None of the suspensor cells differentiated into an embryo. This result again suggests that the polar auxin transport plays a critical role in triggering embryogenic potential of the suspensor cells.

Discussion
In Vivo Living Cell Laser Ablation Technique Confirms the Embryogenic Potential of Suspensor Cells, Which Is Suppressed by the Embryo During Normal Embryogenesis. After fertilization, the zygote asymmetrically divides into two cells: the apical cell, which develops into an embryo proper, and the basal cell, which mainly forms a suspensor. Suspensor cells soon differentiate into specific cell types and no longer divide. The suspensor was traditionally considered a bridge between maternal tissue and the embryo for nutrient transport and a support structure for spatial positioning of the embryo in a seed. In some early reports, suspensor cells were shown to form an embryo after certain physical or chemical treatments of the embryo proper (16)(17)(18). This was also observed in some embryo-defective mutants (7,12,13,15,19,20). Thus, it was proposed that the suspensor cells maintain embryogenic potential that is suppressed by the embryo during normal embryogenesis. Based on available data, a two-way communication model has been proposed to explain the possible interactions between the embryo proper and the suspensor. The suspensor provides nutrients and growth regulators to the embryo proper. Conversely, the embryo releases negative regulators that suppress the embryogenic potential of suspensor cells, thereby maintaining suspensor cell identity (3). Because traditional methods are unable to eliminate the effects of the remaining embryo proper and gene mutations in suspensors on suspensor cell identity, more direct evidence supporting the hypothesis is required. Fortunately, the in vivo living cell laser ablation system can be used to explore this hypothesis. Precise laser ablation allows exact microdissection and ensures that cells after laser ablation can grow in the same microenvironment to minimize the effects of manipulation and thus offers a unique opportunity for observing the in vivo behavior of embryo or suspensors during development. Using this approach, we confirmed that Arabidopsis suspensor cells have embryogenic potential. After detaching from the embryo proper, the top suspensor cell may develop into a second embryo with the same morphology as the original embryo. This suggests that, during normal embryogenesis in a seed, the embryogenic potential of suspensor cells is suppressed by the embryo proper. Once this suppression is removed, the top suspensor cell can develop into an embryo. This seems to be a "back-up system" for seed formation established during plant evolution to ensure embryonic generation if the apical lineage fails to form an embryo.
As shown in Fig. S8, the embryogenic potential of the suspensor cell depends on the developmental stage. The suspensors of 8-and 32-celled embryos developed into second embryos, whereas those at or beyond the heart stage did not. This may be because suspensors at this stage are already entering the process of programmed cell death (35) and are unable to reprogram their fate, although the cells are still living.
Auxin Accumulation in the Top Suspensor Cell Likely Results in Cell-Fate Transition and the Generation of a Second Embryo. During normal embryogenesis, auxin polar transportation is from maternal tissue to the embryo via the suspensor, and, after the 32-cell embryo stage, the embryo proper synthesizes sufficient auxin and then transports the auxin downward to the hypophysis (33). Our previous work in roots revealed that auxin distribution in specific cells decided the cell fate and that changes in the auxin level triggered a cell-fate transition from a differentiated cell to an initial cell of a lateral primordial cell (36).
In this study, after removing the connection between the suspensor and early embryo, auxin transport was completely abrogated. In this case, auxin transportation from maternal tissue to the embryo is interrupted at the cutting end of the suspensor, and auxin accumulates in the uppermost cells of the suspensor instead of the embryo proper. The suspensor cell accumulating auxin then starts to divide and ultimately develops into a second embryo, indicating that redistribution of auxin likely resulted in the developmental fate  transition of the top suspensor cell. With the low level of auxin exiting the cells, the suspensor tends to become differentiated and functionally specialized. Once the embryo is removed and auxin transportation is interrupted, the suspensor cell-fate transition occurs when auxin concentrations increase in the cell.
Our in vitro culture of isolated suspensors and NPA-treated ovules supports the concept that upward, polar transport of auxin is critical for the cell-fate transition of top suspensor cells. Embryogenesis was never observed in isolated suspensors in the presence or absence of auxin. In the presence of plant hormones including auxin, the media triggered suspensor cell division and the formation of cell clusters but did not generate embryos. In addition, when laser-ablated ovules were cultured in the medium with NPA, suspensor cells never generated the second embryos. These evidences indicate that proper levels of plant hormones, and their polar transport, are critical to the cell-fate transition of suspensor cells for embryogenesis.
In fact, it has been reported that the inhibition of auxin responsive process in the suspensor caused excessive divisions in suspensor cells (37). It was also reported that some cell-autonomous and lineagespecific ARFs (auxin response factor) and their inhibitor IAA10 were involved in preventing suspensor cell-fate transition (37). In the normal embryogenesis process, auxin is obviously accumulated in the top suspensor cell or hypophysis initial cell (33). Interestingly, it is just this cell that changes its cell fate and finally develops into hypophysis. These works actually have already implied the involvement of auxin in suspensor cell-fate determination. Our work reveals that auxin accumulation in the newly formed top suspensor cell after laser ablation is coupled with its cell-fate transition to the embryogenic cell identity and thus provides experimental evidence for the above deductions. However, the auxin-related regulatory mechanism in this process still remains to be elucidated. The previous work on the cell-type dependent auxin response may offer a valuable clue to investigate the cell-fate transition of the suspensor cells. As proposed previously, in addition to auxin, the upper suspensor cell may receive other signals that could modify its auxindependent gene expression and regulate its cell-fate transition (37). Thus, the suspensor ARF and IAA10 activity and auxin-activated gene expression will be interesting to analyze after laser ablation.

Materials and Methods
Detailed information on material and methods is provided in SI Material and Methods, which includes detailed description on the plant material and growth conditions; in vivo living cell laser ablation (Fig. S9); culture of isolated suspensors, embryos, and ovules; ovule clearing and GUS staining; and the auxinlevel indication and evaluation.