Naturally occurring wounding reinitiates patterning in flower heads
Wounding is known to initiate cellular dedifferentiation and subsequent regeneration of organs in flower heads (Hernandez and Palmer 1988; Palmer and Marc 1982). Similarly, boron deficiency resulted in surface splits and development of involucral bracts and ray florets in the centre of the sunflower receptacle (Blamey 1976; Hernández 2002). In gerbera, such abnormalities in flower heads can result from natural damage caused by insect bites or by tissue compression during cultivation under standard greenhouse conditions (Fig. 1c, d). Depending on the size, shape and place of the damage, they may either result in ectopic formation of involucral bracts or ray florets in the middle of the head (Fig. 1c), or in a more severe case, the head surface may completely split (Fig. 1d). In both cases the regular phyllotactic pattern is disrupted, as ectopic bracts and ray florets forming along the wound edges take the places normally occupied by disc florets.
To better understand the changes in morphology and patterning caused by such damage, we collected naturally wounded head samples of gerbera and observed them using SEM (Fig. 1e, f). Regardless of the shape and size of the cracked area, new primordia formed right at the edges of the wound. These primordia showed uniform organ polarity, with the abaxial side of bracts or ray florets pointing towards the wounded site. This is consistent with the previous studies on sunflower heads suggesting that the wound rim acts as an organizing center for formation of new primordia (Hernandez and Palmer 1988; Palmer and Marc 1982). We also noticed subtle differences in gerbera compared to earlier results reported in sunflower. Firstly, the identity of the primordia surrounding the wound margin in gerbera was not uniform. Based on their morphology observed by SEM, the emerging primordia were either involucral bracts or floret primordia (Fig. 1e, f). Secondly, some primordia were fused at the base resulting in non-uniform size of the primordia (Fig. 1f). These differences may result from the irregular contour of the rim caused by natural damage as compared to the uniform circular (Hernandez and Palmer 1988) or straight cuts (Hernandez 2002) conducted to sunflower heads. Nevertheless, after patterning of the first primordia at the rim, the following primordia were equalized in size and the spiral organization of primordia resumed as the pattern progressed away from the wound margin (Fig. 1e, f). At the same time, the natural pattern in the parent head progressed normally, until meeting the pattern front originating from the wound site. Both pattern fronts continued as long as the undifferentiated meristem space was available (marked in yellow in Fig. 1e, f).
Growth and auxin responses at the site of mechanical wounding
To investigate the possible role of auxin at the wound site, we performed mechanical wounding to transgenic gerbera lines expressing the auxin reporter DR5rev::3xVENUS-N7 (Heisler et al. 2005) (Fig. 2). By utilizing modified syringe needles with different inner diameters, we were able to achieve either curved (Fig. 2a–d) or cylindrical wounds, creating a separated plug (Fig. 2e–g), to the yet undifferentiated growing meristems. By imaging the head meristems 6 days after the wounding, the wounded areas were found to be approximately 20–50 µm wide and 100–200 µm deep (Figs. 2a, e, S1), and thus comparable to those reported in sunflower treatments (Hernandez and Palmer 1988). We obtained confocal images from a total of seven wounded heads showing that wounding efficiently separated the treated area from the rest of the growing flower head (Fig. 2).
We analyzed the confocal images by focusing on cellular organization and DR5 expression at the wound site. The cells around the wound rim were found to be rearranged into three distinct regions. The first region, closest to the wound rim, is characterized by cells that show significant radial elongation and periclinal division planes in the direction parallel to the cut surface of the wound (Fig. 2b, region 1). This region comprises three to five layers of cells, which is consistent with the previous results in sunflower (Hernandez and Palmer 1988). Interestingly, the DR5 signals were completely absent in this region, indicating low auxin transcriptional output (Fig. 2c). Next to the elongated cells, a second region contains a ring of cells showing high expression of the DR5 reporter (Fig. 2b, c, region 2). DR5 signals in these cells started to localize into maxima, which formed approximately equidistant from each other (Fig. 2c). The adjacent auxin maxima, however, were not equal in size. This was seen in both the curved and the cylindrically wounded samples (Fig. 2c). Closest to the undifferentiated meristem center, cells in the third region remain undifferentiated and small in size (Fig. 2d, region 3). They also lack DR5 expression suggesting that these cells maintain similar identities (Fig. 2c). In heads with circular wounds, similar regional organization of cells were observed (Fig. 2e–g), except that after 6 days of recovery growth, the DR5 pattern had already progressed further towards the center of the isolated circle, and new maxima formed by stacking on top of the previously formed ones (Fig. 2g).
GhCLV3 expression is excluded from the wound margin
The elongation of marginal cells closest to the wound rim, and the emergence of a ring of DR5 signals next to them indicate that their cellular fates might have changed during the recovery growth. Thereby, we examined the expression of GhCLV3, as a marker for undifferentiated central zone of the meristem (Teng Zhang, Mikolaj Cieslak, Paula Elomaa and Przemyslaw Prusinkiewicz, unpublished results), in the mechanically wounded heads (Fig. 3). In the non-wounded control sample, GhCLV3 is expressed in the undifferentiated meristematic dome and maintains its expression after 4 days of culturing on medium (Fig. 3a). In the wounded sample, mechanical wounding was applied to the undifferentiated meristem. The wound on the meristem surface thus disrupted the area with high GhCLV3 expression (Fig. 3b). After 4 days of recovery growth, GhCLV3 expression was excluded from cells closest to the wound rim (Fig. 3c). The absence of GhCLV3 in these cells suggests that their identity has been changed, and their fate no longer corresponds with those in the undifferentiated central domain.
Laser ablation on epidermal layers recapitulates the effects of mechanical wounding
In order to precisely control the size and dimensions of the wounded areas, we optimized a laser ablation protocol for gerbera head meristems by using a confocal microscope equipped with the multiphoton laser. PI staining was applied to each sample to locate the individual cells. During the ablation process the residual PI stains the nuclei of the dead cells, thus allowing us to visualize the ablated area immediately (Fig. 4). After optimizing the parameters, we were able to achieve a dot shaped ablation area of five cells wide and two cells deep, i.e. considerably smaller than those achieved by mechanical wounding (Fig. S1). To create differentially shaped wounds, consecutive point ablations were made. We were able to produce either point-shaped (Fig. 4b), curve-shaped (Fig. 4f) or cylindrical (Fig. 4j) wounds by laser ablation, and follow their effects after 4 days (96 h) of growth on the culture medium.
The head meristems with laser ablations showed similar effects on cellular growth and auxin response as the mechanically wounded heads. Regardless of the shape of the ablated area, the zonal organization of cells and DR5 expression were similar around the wound rim (Fig. 4c, g, k). This indicates that ablation of cells in the epidermal layer is sufficient to reproduce the responses observed in the mechanically wounded heads where the wound extends deeper into the tissue. At the immediate vicinity of the rim, the cells were elongated and devoid of auxin signal. Interestingly, we found that phyllotactic patterning in the non-ablated area was not affected by the wounding close by (Fig. 4d, h). New DR5 maxima emerged in designated places determined by the previously formed neighbors, emphasizing that local interaction of primordia, rather than signals across the entire head meristems determines phyllotactic patterning in heads. Towards the center of the head, new DR5 signals started to emerge and at the 96 h timepoint the very first auxin maxima just started to initiate (Fig. 4c, d, g, h, k, l).