A coherent feed-forward loop drives vascular regeneration in damaged aerial organs of plants growing in a normal developmental context

Plants are prone to numerous injuries in their lifespan, owing to their sessile lifestyle. They are subjected to injuries caused by biotic factors such as pathogen attack and herbivory. Abiotic factors such as damaging weather conditions can also cause tissue damage. Unhealed wounds can compromise plant fitness and survival, and tissue-healing mechanisms have evolved to counteract the damage. Following wounding, regenerative responses may be restricted to local healing in the form of cell proliferation or may entail complete regeneration of damaged tissue or organ (Ikeuchi et al., 2016; Galliot et al., 2017). The capacity of plants to regenerate the complete body plan in vitro from excised tissue is a powerful demonstration of the versatility of plant regeneration processes and forms the basis for many horticultural applications (Kareem et al., 2015; Ikeuchi et al., 2016; Radhakrishnan et al., 2018).

In stem, cellular, molecular and hormonal interactions at wound sites coordinate wound healing and restore vasculature (Flaishman et al., 2003; Asahina et al., 2011; Pitaksaringkarn et al., 2014; Melnyk et al., 2015; Mazur et al., 2016). Auxin is important for vascular tissue regeneration in multiple plant species (Sachs, 1968, 1969, 1981, 1991). The canalization models that underlie this regeneration process rely on the potential of auxin to induce correctly polarised auxin transporters together with activation of vascular cell fate determinants (Wenzel et al., 2007; Donner et al., 2009; Ohashi-Ito et al., 2013). In the growing tips of shoots and roots, damaged meristematic cells are replaced using positional cues from neighbouring cells (van den Berg et al., 1995; Reinhardt et al., 2003).

In roots, regeneration involves reactivation of embryo-specific genes, proper reallocation of root cell-fate determinants and integration of auxin, cytokinin and jasmonate signals (Xu et al., 2006; Efroni et al., 2016; Marhava et al., 2019; Zhou et al., 2019).

Laser ablation and root tip resection studies have shown that stem cell activation is a vital step for regeneration of lost cells and entire organs (van den Berg et al., 1995; Xu et al., 2006; Marhava et al., 2019; Zhou et al., 2019). The stem cell regulators PLETHORA1 (PLT1) and PLT2 are essential for the re-establishment of quiescent centre (QC) cells upon laser ablation and for the regeneration of primary and lateral root tips following resection (Xu et al., 2006; Durgaprasad et al., 2019). PLT1 and PLT2 are induced by PLT3, PLT5 and PLT7 activity to regulate stem cell activation during lateral root development (Du and Scheres, 2017). In the shoot, members of the PLT family along with the transcription factor AINTEGUMENTA (ANT) regulate the development and phyllotaxis of aerial organs (Prasad et al., 2011; Krizek, 2015). PLT factors also regulate hormone-mediated de novo shoot regeneration (Kareem et al., 2015).

Although several studies have addressed specific regeneration processes in specific plant parts or in excised organs and have implicated certain factors regulating these processes, our knowledge of the underlying molecular mechanisms of wound repair in aerial organs is limited (Ikeuchi et al., 2018). It is largely unknown how wound repair in leaf tissue relates to the normal developmental programme. Here, we investigate vascular reprogramming after leaf damage from the viewpoint that tissue reprogramming may require stem cell factors identified in other regeneration contexts. We reveal an essential role of members of the PLETHORA (PLT)/AINTEGUMENTA (ANT) gene family in activating regeneration responses. PLT genes act through CUP-SHAPED COTYLEDON2 (CUC2) to repair wounds and regenerate vascular tissue in damaged aerial organs. Furthermore, we show that the PLT-CUC2 module acts through local auxin biosynthesis, and is required for proper repolarisation of PIN1 auxin efflux facilitators and reprogramming of vascular identity in aerial organs. The PLT-CUC2 module is strictly required for regeneration of leaf vasculature, but is not essential for the normal development of closed vein loops in the absence of perturbations.

PLT3, PLT5 and PLT7 collectively regulate tissue culture-mediated in vitro shoot regeneration and will from here on be referred to as PLT3,5,7. PLT3,5,7-regulated root stem cell regulators establish pluripotency in calluses and PLT3,5,7-regulated shoot-promoting factors act in response to external hormonal cues to induce regeneration of the complete plant body (Kareem et al., 2015).

Interestingly, PLT3,5,7 genes are expressed in the shoot during development and positioning of aerial organs (Prasad et al., 2011; Krizek, 2015). To assess whether PLT3,5,7 function is required for repairing damaged inflorescence and leaf tissue without external hormonal cues, we determined whether expression of these genes is induced as a natural response to injuries that growing plants are likely to encounter, such as local abrasions in the stem, partial stem incisions and midvein injuries in the leaf blade. These injuries were made without detaching any organ from growing Arabidopsis plants. After local abrasion that damaged the epidermal and sub-epidermal layers, including vascular tissue, in inflorescence stem (Fig. 1A,A′, Fig. S1A-D), PLT7::PLT7-vYFP was induced 12 h post-injury, prior to any apparent regeneration response (Fig. 1B,B′). The expression peaked at 36 h (Fig. 1C,C′, Fig. S1F,F′). In response to partial incision of the inflorescence stem (Fig. 1D,D′), PLT7::PLT7-vYFP expression was upregulated at both ends of the incised stem, with relatively higher expression in the upper end after 6 h (Fig. 1E,E′). The high level of expression continued for 12 h (Fig. 1F,F′). At 12 h, upregulated expression expanded beyond the partial slit, and at 24 h it became confined to a narrower domain in the vicinity of the incision (Fig. 1F-G′). Transcript levels of PLT7 were consistent with the fusion protein expression data and remained upregulated at 24 h (Fig. S1E). Similarly, when the midvein of a growing leaf blade was wounded, cells in the vicinity displayed pronounced upregulation of PLT7::PLT7-vYFP 12 h post-injury (Fig. 1H,I). In response to injury, PLT3::PLT3-vYFP and PLT5::PLT5-vYFP also showed upregulation of expression in the vicinity of the wound albeit with some differences in the timing of their activation and in spatial distribution (Fig. 1J,K, Figs S1G-P′, S2 and S3). In response to leaf incision, although both PLT3 and PLT7 were expressed in close proximity to the wound, PLT5 was expressed predominantly in the vascular tissue near the damage (Fig. 1H-K, Fig. S3A,B).

The root stem cell regulators PLT1, PLT2 and WOX5, which are activated by PLT3,5,7 during tissue culture-mediated in vitro shoot regeneration (Kareem et al., 2015), were not expressed in growing leaves and stems in response to injuries (Fig. S4).

Aerial organs of growing plants are subject to substantial wear and tear and PLT3,5,7 expression is rapidly activated in response to injuries (Fig. 1, Figs S1-S3). We therefore asked whether PLT3,5,7 genes are required for wound repair and tissue regeneration in stems and leaves growing in the normal developmental context of Arabidopsis.

We mimicked physical abrasion by damaging the epidermis, sub-epidermal layers and vascular tissue locally (see Materials and Methods for details; Fig. 2A,A′) in a growing inflorescence stem of wild-type as well as plt3;plt5-2;plt7 mutant plants. In the wild type, we noticed a healing response in the form of a visible mass of proliferating cells (callus-like growth) throughout the wound at 2 days after abrasion (daa), which became more prominent at 4 daa (Fig. 2B, Fig. S5A,B). Later, callus-like growth completely covered and sealed the wound. The inflorescence stem regained its growth following the repair process. In contrast to wild type, the healing response was severely reduced in injured plt3;plt5-2;plt7 inflorescence stems and the wound-sealing process was not completed in the triple mutant inflorescence stem (Fig. 2A′,B′, Fig. S5A′,B′). Importantly, the inflorescence stem development of uninjured mutant was comparable to that of wild type (Fig. S5I,J).

Next, we made a partial slit in the inflorescence stem of wild-type and plt3;plt5-2;plt7 mutant plants disrupting both vascular connections and ground tissue (Fig. 2C,C′,E,E′). Twenty-four hours after the incision, the wounded parts adhered in the wild-type inflorescence stem (Fig. S5C,D). Subsequently, cell proliferation was observed as indicated by visibly swollen tissues at cut ends followed by regeneration of vascular tissues at 4 days after cut (dac) (Fig. 2D,F). Subsequent restoration of growth and physiological functions were demonstrated by the development of new flowers and siliques (Fig. S5E). In contrast to wild type, in which the wound was healed on the fourth day, the plt3;plt5-2;plt7 triple mutant displayed severely reduced callus-like growth at the wound site and ∼49% inflorescence stems failed to regenerate vascular tissue (Fig. 2C′-F′, G, Fig. S5C′). Our data demonstrate the role of PLT3,5,7 in activating a healing response in the form of callus-like growth and vascular regeneration to restore damaged tissue in a growing inflorescence stem.

Restoration of vasculature is a long-known feature of stem regeneration, and we investigated whether this response also occurred in leaves. We made a local injury in the midvein of a growing young wild-type leaf of a 5 days post-germination (dpg) plant (see Materials and Methods). To keep the developmental stage uniform, we injured the first pair of young leaves, which displayed midvein formation but not fully developed lateral veins at the time of injury (Fig. 2H,H′, Fig. S6A,A′). The injuries either (1) damaged the midvein without making an opening or (2) completely disconnected the midvein leaving a gap between the vascular strands. In both the cases, cells in the vicinity of the midvein experienced mechanical perturbations due to the pressure applied by the needle. Four days post-injury (dpi), wild-type leaves repaired both types of injuries. In case (1), where the break was incomplete, the injured midvein was repaired and new vascular cells regenerated to restore the physiological connection (Fig. S6E). In case (2), where there was a complete disconnection, we observed regeneration of the vascular strand either connecting together the cut ends of the midvein or connecting the cut end of the midvein to a lateral vein (Fig. 2I,J, Fig. S6F,G). Strikingly, after local injury in the midvein of young wild-type leaf blades, ∼80% of the samples regenerated vascular tissue in response to incision (Fig. 2L). The regenerating vascular cells often bypassed the damaged area and reunited with the lower half of the midvein making a D-shaped loop around the wound site similar to Sachs' observation of vascular regeneration around the wound site in the epicotyl stem of pea plants (Sachs, 1981) (Fig. 2I). Alternatively, they formed a new connection to the nearest lateral vein (Fig. 2J). The non-regenerating lower vascular strand degenerated after residual proliferation at the cut end (Figs S5L and S6B). We followed vascular regeneration from the time of injury to distinguish between the vascular strand reuniting the midvein regenerating from the cut end and the recruitment of a pre-existing lateral vein developed during leaf growth (Fig. S6A-D′). When the injury left a hole in the leaf blade exceeding 400 µm between the cut ends of the midvein, we rarely observed any vascular regeneration (Figs S6H,I and S7A). Such injuries left behind only a disorganised mass of cells (Fig. S6H). We therefore restricted our subsequent analysis to leaf blade injuries that completely disconnected the midvein leaving a gap well under 400 µm between the cut ends.

In contrast to wild-type leaves, in which ∼80% of the injured leaves regenerated vascular strands, only ∼40% of injured plt3;plt5-2;plt7 leaves could regenerate and the rest completely failed to regenerate vascular strands (Fig. 2H′,I′,L). In non-regenerating mutant leaves, lateral veins failed to connect to the midvein near the wound site (Fig. 2I′), but a disorganised mass of proliferating cells at the wound site was observed, mostly at the cut ends of upper vascular strands and on the epidermis (Fig. S7B-D). Many such leaves displayed poor growth and failed to develop properly (Fig. S7E). It is important to note that uninjured plt3;plt5-2;plt7 mutant plants did not display any defects in the formation of closed vein loops (consisting of midvein and secondary veins) compared with wild type but were severely impaired in vascular regeneration (Fig. 2L, Fig. S7F-I). With respect to leaf morphology, we did not observe any defects in the first pair of leaves (Fig. S5F,G). Among double mutant combinations, 70% of plt3;plt5-2 and plt5-2;plt7 double mutants regenerated vascular strands in response to injury, and only ∼64% of plt3;plt7 double mutant leaves regenerated vascular tissue (Fig. S8A).

The closely related AINTEGUMENTA (ANT) gene marks stem cells of root vascular cambium and acts redundantly with PLT3 and PLT7 during plant development (Krizek, 2015; Smetana et al., 2019). ANT is strongly expressed in the vascular tissue of young leaves (Fig. S8B). We therefore examined vascular regeneration in plt3;plt7;ant-4 triple mutant plants in response to midvein injury. Strikingly, none of the tested plt3;plt7;ant-4 seedlings regenerated vascular tissues, demonstrating an essential role of ANT with PLT3 and PLT7 in vascular regeneration (Fig. 2K,L, Fig. S8C). Taken together, our data reveal a previously unrecognised role of PLT3,5,7 and ANT in repairing damaged tissues during plant growth.

Because of the severity of shoot phenotypes in plt3;plt7;ant-4 (which produces only leaves but no stem) we chose the plt3;plt5-2;plt7 mutant, which develops normal leaves as well as an inflorescence stem comparable to wild type, to probe the mechanism of vascular regeneration using further assays (Prasad et al., 2011; Krizek, 2015).

Tissue/organ regeneration is closely linked to cellular reprogramming. We next asked whether PLT genes are sufficient to activate cellular reprogramming leading to enhancement of wound repair. Strikingly, inducible overexpression of PLT5 (35S::PLT5-GR) or PLT7 (35S::PLT7-GR) promoted multiple strand formation from the regenerating midvein in response to injury (Fig. 3A,A′,C,C′, Fig. S8D). Similarly, inducible overexpression of PLT5 or PLT7 enhanced wound repair at the cut ends of the detached organ and in response to inflorescence abrasion (Fig. 3B,B′,D,D′, Fig. S8E-F′). Consistent with the ability of PLT proteins to promote cell division upon wounding, transcripts of CYCLIN genes increased in growing seedlings upon inducible overexpression of PLT5 (35S::PLT5-GR) (Fig. S8G). These results suggest that PLT proteins are sufficient to promote wound repair and multiple vascular strand regeneration in response to injury.

We addressed whether PLT-like proteins from other plant species can trigger regeneration in Arabidopsis. Rice is a morphologically diversified monocot plant, whereas Arabidopsis is a dicot. Expression of the rice PLT-like gene OsPLT2 under the Arabidopsis PLT5 promoter in a plt3;plt5-2;plt7 mutant (plt3;plt5-2;plt7;AtPLT5::OsPLT2-vYFP) healed a damaged Arabidopsis plt mutant inflorescence stem by inducing cell proliferation as evident from upregulated expression of cell cycle progression markers (Fig. 3E-G, Fig. S8H,I). Furthermore, OsPLT2-vYFP rescued leaf vascular regeneration defects in plt3;plt5-2;plt7 suggesting that it is a functional homologue of Arabidopsis PLT genes (Fig. 3H-J).

Having established that PLT3,5,7 regulate wound repair and vascular regeneration in damaged aerial parts of the plant, we sought to define the molecular mechanisms underlying this regulation. Previously, we had shown that PLT3,5,7 direct tissue-culture-mediated in vitro shoot regeneration by activating root stem cell regulators and CUC2 (Kareem et al., 2015). Although we found no evidence for the participation of PLT1, PLT2 and WOX5 root stem cell regulators in the response to injuries in growing aerial organs (Figs S4 and S9A), CUC2 remains an attractive candidate for participation in wound repair. Therefore, we investigated whether CUC2 responds to mechanical injury and whether PLT3,5,7 act through CUC2 to repair wounds and regenerate vasculature.

pCUC2::3XVENUS as well as CUC2::CUC2-vYFP expression was detected in vascular tissue of young leaves in both wild-type and plt3;plt5-2;plt7 mutant plants although expression was reduced in the latter (Fig. 4A,A′, Fig. S9B-F,I,I′). The same CUC2 promoter was used to drive transcriptional and translational fusions. The detection of an expanded domain of expression of pCUC2::3XVENUS compared with CUC2::CUC2-vYFP can be largely attributed to 3XVENUS. Both reporter fusions used in this study can recapitulate the previously reported CUC2 expression at the leaf margin (Nikovics et al., 2006; Bilsborough et al., 2011) (Fig. S9G,H). In response to midvein damage in wild type, expression of both pCUC2::3XVENUS and CUC2::CUC2-vYFP was upregulated proximal to the wound 12 h post-injury followed by a broader domain of enhanced expression after 24 h (Fig. 4B,C, Fig. S9I-K). In contrast, there was no upregulation of the reporter near the wound site in plt3;plt5-2;plt7 (Fig. 4B′,C′, Fig. S9I′-K′). Similar patterns of changes were also observed at the transcript level in response to midvein injury (12 h post-injury) (Fig. S9L). Similarly, in damaged inflorescence stems, CUC2 transcripts were reduced in the plt3;plt5-2;plt7 compared with wild type (Fig. 4D). Furthermore, CUC2 transcripts rapidly increased in injured leaves upon inducible overexpression of PLT5 (35S::PLT5-GR) as well as of PLT7 (35S::PLT7-GR) even in the presence of the translation inhibitor cycloheximide, suggesting direct activation of CUC2 transcription by PLT5 and PLT7 (Fig. 4E, Fig. S9M). Consistent with these observations, PLT5 bound to the CUC2 promoter in a chromatin immunoprecipitation (ChIP) assay (Fig. S9N). In addition, DNA affinity purification sequencing (DAP-Seq) analysis identified the binding of PLT7 to the CUC2 promoter (O'Malley et al., 2016, http://neomorph.salk.edu/) (Fig. S10A). Furthermore, transient transfection of trans genes capable of producing PLT5 or PLT7 proteins and the CUC2 promoter-driven luciferase reporter in Nicotiana leaf induced reporter expression, further demonstrating that PLT5 as well as PLT7 can directly activate CUC2 transcription (Fig. 4F).

Because molecular data indicate that CUC2 acts downstream of PLT genes, we investigated whether PLT proteins require CUC2 activity for wound repair. Strikingly, inducible ectopic overexpression of PLT5 failed to promote wound repair at the damaged end of cuc2-3 (cuc2-3;35S::PLT5-GR) mutant tissues. The severely compromised wound repair that was observed at the cut ends remained unaltered upon PLT5 overexpression in cuc2-3 detached tissue, but not upon PLT5 overexpression in wild type (Wild type;35S::PLT5-GR), which enhanced wound repair at the cut ends (Fig. S10B-F). These results demonstrate that PLT proteins act through CUC2 to repair the wound.

We examined the role of CUC2 in leaf vascular regeneration by analysing loss-of-function mutants. Strikingly, vascular regeneration was severely impaired in both the recessive loss-of-function cuc2-3 mutant as well as in the cuc2-1D dominant mutant; 71% of cuc2-3 mutant and 81% of cuc2-1D mutant leaves failed to show any vascular regeneration in response to midvein injury (Fig. 4G). Notably, loss of CUC2 function did not cause any defect in the formation of closed vein loops formed by primary (midvein) and secondary (lateral) veins (Fig. S7F,G,J,K,L). Similarly, upon inflorescence stem incision, ∼78% cuc2-3 mutant and 92% of cuc2-1D mutant inflorescence stems failed to show any vascular regeneration (Fig. 4H). Finally, we asked whether CUC2 overexpression can rescue the vascular regeneration defect in plt3;plt5-2;plt7 mutant leaves. Strikingly, the regeneration efficiency (timings of regeneration and reunion of vascular strands) as well as frequency (number of plants) was restored upon CUC2 overexpression in plt3;plt5-2;plt7 to the level of wild type. New vascular strands regenerated and reunited by 4 dpi in the mutant, similar to wild type (Fig. 4I-J″). Moreover, CUC2 overexpression rescued the repair process in locally wounded plt3;plt5-2;plt7 inflorescence stems (Fig. 4K-K″). Taken together, these data demonstrate that PLT3,5,7 directly activate CUC2 transcription in response to injury and that the PLT-CUC2 module is required for wound repair and vascular regeneration in leaf and stem. Interestingly, the growth of inflorescence stems and leaves were similar in wild type and the cuc2-3 mutant (Fig. S5H,K).

CUC2 is implicated in the regulation of leaf margin development by directing PIN1 polarity and the resultant auxin distribution (Bilsborough et al., 2011) and PIN1 polarisation is crucial for the normal development of leaf vasculature (Scarpella et al., 2006). Hence, we next investigated whether the process of cell polarisation is regulated by the PLT transcription module during leaf vascular regeneration. We focused on in vivo vascular regeneration in developing leaves, which has not been explored. To this end, we examined the localisation of PIN1 in response to midvein injury in the leaf blade (Fig. S11C′-F′,G′-J′). Prior to wounding, we observed PIN1::PIN1-GFP expression in the procambium predominantly towards the basal end of young leaves in both wild type and mutant (Fig. S11A-F,G-J). In response to injury, we observed increased PIN1-GFP near wound sites in both wild type and mutant (Fig. S11E′,I′). To examine PIN1-GFP localisation in regenerating vascular cells, we generated transgenic lines harbouring both PIN1::PIN1-GFP and ATHB8::ATHB8-vYFP. ATHB8 specifically marks developing procambium cells in leaf (Scarpella et al., 2004). We observed both PIN1-GFP and ATHB8-YFP in developing procambium of 4-day-old leaves (Fig. 5A,B).

During the first 12 h following incision, we did not observe regenerating vascular cells expressing both PIN1-GFP and ATHB8-YFP near wound sites (Fig. 5C,D). Regenerating procambium cells marked with ATHB8-YFP and polarised PIN1-GFP were observed after 24 h near wound in wild type (Fig. 5E). In contrast, we did not observe regenerating procambium cells expressing polarised PIN1-GFP or ATHB8-YFP near the wound after 24 h in plt3;plt5-2;plt7 plants, demonstrating that cells surrounding the damaged site failed to re-specify the PIN1 polarity in the mutant (Fig. 5F). These data suggest that failure of re-establishment of polar auxin transport within 24 h may contribute to impaired vascular regeneration in the plt triple mutant. We next examined whether lack of directional auxin flow in the damaged plt3;plt5-2;plt7 mutant leaves altered the distribution patterns of the auxin response. We examined the auxin response using the auxin reporter pDR5rev::3XVENUS-N7 in both wild-type and plt3;plt5-2;plt7 mutant plants. Prior to injury, we did not observe any difference in distribution patterns or levels of the auxin response in leaves between these two genotypes (Fig. 5G,G′, Fig. S11K-M′). In both wild type and plt3;plt5-2;plt7 an increase in pDR5rev::3XVENUS-N7 signal in the tissue proximal to the wound was observed at 24 h post-injury (Fig. 5H,H′,I,I′). However, a further enhanced auxin response was confined to an area near the wound site by 48 h only in wild type (Fig. 5J). In contrast to wild type, the triple mutant failed to show such confined expression of pDR5rev::3XVENUS-N7 signal in response to injury (Fig. 5J′). The distribution patterns and levels of auxin response in uninjured developing mutant leaves compared with wild type did not change, further substantiating the specific role of PLT3,5,7 in response to injury (Fig. 5G,G′, Fig. S11K-M′). Taken together, our results show that PLT3,5,7 are needed for re-specification of polarised vascular cells to facilitate vascular tissue regeneration.

Local auxin biosynthesis has been implicated in root haustoria formation and associated vascular development during host-parasite interaction (Kokla and Melnyk, 2018). We therefore asked whether PLT-CUC2 module regulate wound repair and vascular regeneration by modulating local auxin biosynthesis genes. Interestingly, local auxin biosynthesis genes are downregulated in plt3;plt5-2;plt7 mutant callus (A.K. and K.P., unpublished data). PLT proteins are also known to control phyllotaxis by regulating one of the auxin biosynthesis genes, YUCCA4 (YUC4) (Pinon et al., 2013). Similarly, YUC4 expression was upregulated in response to midvein injury (12 h post-injury) in growing wild-type leaves, unlike in the plt3;plt5-2;plt7 leaves, in which the transcript level was reduced (Fig. 6A). In addition to damaged leaves, YUC4 transcripts were also reduced in damaged plt3;plt5-2;plt7 inflorescence segments (Fig. S12A). Conversely, YUC4 transcripts were rapidly increased in injured tissues upon PLT5-GR induction (4 h) even in the presence of the translation inhibitor cycloheximide, suggesting direct activation by PLT5 (Fig. 6B). Because molecular data suggests that YUC4 acts downstream of PLT genes, we investigated whether PLT proteins require YUC4 activity to trigger cellular reprogramming. Strikingly, inducible overexpression of PLT5 as well as PLT7 failed to trigger any ectopic cellular reprogramming in the yuc4;yuc1 mutant background (yuc4;yuc1;35S::PLT5-GR or yuc4;yuc1;35S::PLT7-GR), unlike in the wild-type background (Wild type;35S::PLT5-GR; or Wild type;35S::PLT7-GR) (Fig. S12B,C). Similarly, PLT5 as well as PLT7 overexpression failed to promote wound repair at damaged ends, demonstrating that PLT proteins act through YUC4 during reprogramming and wound repair (Fig. S12D-G).

We explored whether, in addition to PLT genes, CUC2 might also contribute towards regulating the local auxin biosynthesis in response to injury. YUC4 transcripts were not upregulated in response to midvein injury in the cuc2-1D single mutant (Fig. 6A) and its transcript levels were rapidly increased upon CUC2-GR induction even in the presence of the translation inhibitor cycloheximide, suggesting direct activation of YUC4 expression by CUC2 (Fig. 6C). Consistent with the likelihood of direct activation of YUC4 transcription by CUC2 inferred from our results, DAP-Seq analysis indicated the binding of CUC2 to the YUC4 promoter (Fig. S12H) (O'Malley et al., 2016, http://neomorph.salk.edu/). Next, we examined whether, like PLT proteins, CUC2 also requires downstream YUC4 activity to promote vascular regeneration. Ectopic overexpression of CUC2 promoted vascular regeneration in the leaf and resulted in regeneration of multiple vascular strands from the wound site in the wild type (Wild type;35S::CUC2-3AT) (Fig. 6D). In contrast to wild type, ectopic overexpression of CUC2 failed to promote regeneration of multiple vascular strands from the wound site in the yuc4;yuc1 mutant (yuc4;yuc1;35S::CUC2-3AT) (Fig. 6D-F). Injured leaves in yuc4;yuc1;35S::CUC2-3AT seedlings either did not regenerate any vascular strand or occasionally displayed a single file of regenerating vascular cells as was observed in yuc4;yuc1 (Fig. 6E,F, Fig. S12I). These data demonstrate that, like PLT genes, CUC2 also acts through YUC4 to promote wound repair and vascular regeneration.

Our data suggest that, in addition to PLT proteins, CUC2 can also activate YUC4 expression during vascular regeneration. Activation of YUC4 by PLT5, PLT7 and CUC2 indicates a feed-forward loop controlling local auxin biosynthesis (Fig. 6B,C, Fig. S12J). PLT5-GR can only moderately activate YUC4 expression after 4 h induction when the function of CUC2 and of the redundantly acting CUC1 is lost (in damaged cuc1-5;cuc2-3 tissues) (Fig. 6H), indicating that increased transcription of YUC4 in wild-type damaged leaves may be an output of a coherent feed-forward loop during tissue regeneration. We further provide genetic evidence for the feed-forward loop: inducible overexpression of PLT7 or PLT5 can still increase vascular regeneration by 18% and 24%, respectively, in response to midvein injury in the cuc2-3 mutant (Fig. S12K).

We further investigated this regulatory interaction by analysing the genetic interaction between PLT genes and CUC2. Strikingly, we found synergistic interaction between PLT genes and CUC2 during wound repair and vascular regeneration. Cumulative loss of PLT and CUC2 function in the plt3;plt5-2;plt7;cuc2-3 mutant resulted in severely compromised wound repair at the cut end of the detached plant organ compared with the plt3;plt5-2;plt7 or cuc2-3 mutant (Fig. S13A). In addition to dramatically reduced frequency of wound repair in plt3;plt5-2;plt7;cuc2-3 mutant, we could barely observe any proliferating callus-like cells at the damaged ends in plt3;plt5-2;plt7;cuc2-3 mutant organs (Fig. S13B-E). The YUC4 transcript level was further reduced in the plt3;plt5-2;plt7;cuc2-3 mutant compared with the plt3;plt5-2;plt7 or cuc2-3 mutant (Fig. S13F). Similarly, seedlings heterozygous for plt and cuc2 alleles, plt3^+/−;plt5-2^+/−;plt7^+/−;cuc2-3^+/− displayed hypersensitivity to leaf midvein injury compared with plt3^+/−;plt5-2^+/−;plt7^+/− or cuc2-3^+/− (Table S1, Fig. S7L). These data substantiate the regulation of YUC4 expression by PLT proteins and CUC2 in a coherent feed-forward loop during wound repair and vascular regeneration.

Consistent with the importance of activation of YUC4 expression for regeneration, ∼40% of yuc4 single mutant and 87% of yuc4;yuc1 double mutant leaves failed to regenerate vascular tissue in response to midvein injury (Fig. 6G). Strikingly, the uninjured yuc4 single mutant develops a fully grown midvein without any discontinuity and there is no significant difference in the formation of closed vein loops compared with wild type (Fig. S7F,G,M). Although midvein formation in the yuc4;yuc1 mutant was normal, as in the wild type, the number of loops surrounding the midvein was reduced (Fig. S7F,G,N). Strikingly, reconstitution of YUC4 expression in the endogenous PLT5 domain (PLT5::YUC4-vYFP) in the plt3;plt5-2;plt7 mutant as well as in the cuc2-1D mutant rescued the vascular regeneration in injured leaves to a large extent (Fig. 6I-M, Fig. S13G-I). These data provide compelling evidence for the functional significance of PLT-CUC2 module-dependent activation of local auxin biosynthesis in controlling vascular regeneration. Remarkably, reconstitution of YUC4 expression in the cuc1-5;cuc2-3 (cuc1-5;cuc2-3;PLT5::YUC4-vYFP) mutant, which generates cup-shaped cotyledons but no leaf or stem, rescued post-embryonic development with fully developed rosette leaves (Fig. S14).

Multicellular organisms display the ability to regrow damaged tissues and organs. Unlike many animals, in which regeneration potential is restricted to specific cell lineages, plants repair and rebuild damaged tissues throughout the body. In this study, we have investigated the mechanism of wound repair across aerial parts of the plant body, identifying PLT/AIL transcription factors, well known for their role in stem cell maintenance, as regulatory triggers for this process. We demonstrate that activation of CUC2 transcription by PLT3,5,7 is a key regulatory mechanism of wound repair and vascular regeneration: (1) PLT factors bind to the CUC2 promoter and directly activate the transcription of CUC2; (2) PLT factors require downstream CUC2 activity during wound repair; and (3) reconstitution of CUC2 expression under a heterologous promoter in plt3;plt5-2;plt7 triple mutants rescues vascular regeneration. We provide evidence that PLT proteins and CUC2 activate the transcription of local auxin biosynthesis gene in a feed-forward loop to drive vascular regeneration: (1) both PLT and CUC2 require downstream YUC4 activity as ectopic overexpression of PLT proteins as well as of CUC2 fails to trigger regeneration response in the yuc4;yuc1 mutant; (2) reconstitution of YUC4 expression under a heterologous promoter in the plt triple mutant as well as in the cuc2-1D mutant rescues the vascular regeneration defects; and (3) PLT proteins and CUC2 act synergistically to activate YUC4 transcription and repair the damaged tissues, which involves induction of vascular identity and proper polarisation of the polar auxin transporter PIN1.

Our study revealed a previously unrecognised role of ANT in vascular regeneration and demonstrated that a PLT-like gene from rice, a morphologically diverse grass species, could substitute the regeneration function of Arabidopsis PLT genes. These observations indicate that the activation of PLT gene promoters in response to mechanical injuries may be more important for the selection of regeneration-associated PLT genes than their protein sequence. In this context, it is relevant that distinct PLT transcription factors determine competence for regeneration in the root context (Durgaprasad et al., 2019).

In striking contrast to in vitro shoot regeneration cues (Kareem et al., 2015), PLT3,5,7 do not act through the root stem cell regulators PLT1, PLT2 and WOX5 to initiate repair of damaged aerial tissues of a growing plant. Rather, PLT proteins act through CUC2 by directly activating its expression (Fig. 6N). Interestingly, PLT genes and CUC2 act in a feed-forward loop to activate the expression of the auxin biosynthesis gene YUC4 (Fig. 6O). This circuit can act as a coherent feed-forward loop, which often serves as a signal persistence detector (Mangan et al., 2003), even though our analysis indicates that the regulatory logic at the promoter is not strictly an ‘AND gate’ (Alon, 2006). Regardless of the precise regulatory logic, the output of the circuit is the activation of YUC4. In that view, it is tantalising that the cellular defects associated with the malfunctioning of this circuit are the inability to redirect ground tissue cells to vascular identity and the inability to properly polarise PIN proteins. A regulatory feedback loop between auxin level, auxin flux and polarisation of auxin efflux carriers (PIN) has been proposed as a key regulatory mechanism of shoot branching, phyllotaxis and vascular tissue differentiation (Jönsson et al., 2006; Smith et al., 2006; Bayer et al., 2009; Schuetz et al., 2012; Mazur et al., 2016; Fujita and Kawaguchi, 2018). It is therefore conceivable that PLT-CUC2-dependent activation of YUC4 activates this feedback loop to drive vascular regeneration in damaged growing leaves (Fig. 6O). In summary, our study reveals PLT-CUC2 regulatory axis is specifically involved in controlling regeneration through induction of a local hormonal environment in response to injury.

Arabidopsis thaliana ecotype Columbia (Col-0) was used as wild type in this study. The origins of the mutants used in the study, such as the double mutants plt3;plt5-2, plt3;plt7 and plt5-2;plt7 and the triple mutant plt3;plt5-2;plt7 (Prasad et al., 2011), the double mutant yuc4;yuc1 (Pinon et al., 2013), the single mutant ant-4, double mutant ant-4;plt5-3 and the triple mutant plt3;plt7;ant-4 (Krizek, 2015), the single mutants cuc2-1D (Larue et al., 2009) and cuc2-3 (Hibara et al., 2006), and the double mutant cuc1-5;cuc2-3 (Hibara et al., 2006), have been described previously. Translational fusion constructs of PLT1::PLT1-vYFP, PLT2::PLT2-vYFP (Mähönen et al., 2014), PLT3::PLT3-vYFP, PLT5::PLT5-vYFP and PLT7::PLT7-vYFP (Prasad et al., 2011) have been described previously. 35S::PLT5-GR, 35S::PLT7-GR (Prasad et al., 2011), pCUC2::3XVENUS and 35S::CUC2-3AT (Kareem et al., 2015) have been described previously. Multisite gateway recombination cloning system (Invitrogen) using the pCAMBIA 1300 destination vector was used for cloning the translational fusion constructs, which were then introduced into the C58 Agrobacterium strain by electroporation and further transformed into wild-type or mutant Arabidopsis plants by the floral dip method (Clough and Bent, 1998) (see supplementary Materials and Methods for details on plasmid construction). DR5rev::3XVENUS-N7 expression was examined in wild-type and plt3;plt5-2;plt7 transgenic plants with the double marker pDR5rev::3XVENUS-N7,PIN1::PIN1-GFP line, which has been described previously (Pinon et al., 2013). In this study, only the YFP marker was analysed using a single YFP channel.

Arabidopsis thaliana seeds were surface sterilised with 70% ethanol and 20% bleach, followed by seven washes with sterile distilled water. Seeds were plated on half-strength Murashige-Skoog (MS) medium (pH 5.7) and grown vertically under 45 μmol/m²/s continuous white light at 22°C and 70% relative humidity.

For wound-induced natural regeneration experiments, all plants and explants were grown on hormone-free half-strength MS agar medium (Sigma-Aldrich). To study wound repair and vascular regeneration in growing inflorescence stems, 3-week-old seedlings were selected. Using a sterile razor blade, the stem region between the rosette leaves and the first or second cauline leaves was subjected to either peeling of the tissue layers including epidermis and sub epidermal layers (inflorescence abrasion) or partial incision (inflorescence incision) through the vascular tissues under a dissection microscope (Zeiss Stemi 2000). The observations were recorded 4 days after wounding. For the leaf vascular regeneration assay, to maintain uniformity we injured a single leaf belonging to the first pair of rosette leaves of 5 dpg seedlings. Plants of same developmental stage were chosen for incision. Fine-pointed sterile tweezers (Dumont tweezer, Style 5) were used to make a sharp incision in the midvein at the basal part of the leaf blade. To avoid ambiguity, incisions made elsewhere were not scored. The incisions were made from the abaxial surface of the leaf to ensure precise injury to the midvein. The injured leaf was left connected to the growing parent plant and it was protected from any further damage. Vascular regeneration was analysed in the injured leaf 4 days post-incision. These leaves were cut at the petiole using Vannas straight scissors (Ted Pella, 1340) without causing additional damage to the leaf blade. The leaf tissue was cleared using chloral hydrate (Sigma-Aldrich) (see supplementary Materials and Methods for further details of decolourisation and tissue clearing) and brightfield images were obtained to assess the regeneration outcomes. When newly formed vascular strands (identified by the distinct morphology of end-to-end connected xylem elements) connected the cut ends of the midvein to form a D-shaped loop or connected the damaged midvein to a lateral vein, the outcomes were scored as successful regeneration (Fig. 2I,J). To study healing in response to wounding in excised organs (leaf/root), excised explants were collected from 9 dpg seedlings and placed on hormone-free MS agar medium. Upon excision, continuous dexamethasone (Sigma-Aldrich) induction was provided until the tenth day post-excision for samples collected from transgenic lines harbouring steroid-inducible constructs. The corresponding mock-treated samples were incubated on MS plates containing an equal proportion of DMSO (same volume as dexamethasone). The plates were kept in the dark for the first 24-32 h and later shifted to continuous light. All the plates of regeneration experiments were incubated vertically in a plant growth chamber maintained at 22°C and 70% relative humidity under 45 μmol/m²/s continuous white light.

Brightfield and confocal laser-scanning microscopy imaging were performed as described previously (Kareem et al., 2015). Brightfield images of vascular regeneration in incised leaves were acquired using the brightfield mode in a Leica TCS SP5 II inverted confocal microscope and an Olympus BX63F fluorescence microscope after clearing the leaf sample (see supplementary Materials and Methods for details of decolourisation and tissue clearing). Confocal imaging of leaves and thick samples were performed using a Leica TCS SP5 II upright microscope and a Zeiss LSM 880 confocal laser-scanning microscope. Brightfield images acquired using Leica M205 FA fluorescence stereo microscope and confocal microscopes were adjusted for brightness and contrast. For confocal imaging, the cell boundaries of inflorescence stem samples were stained using 10 µg/ml propidium iodide (Sigma-Aldrich). Images were acquired using 10× air, 20× oil immersion, 20× air and 40× oil immersion objectives. The projection view of the images was reconstructed from the z-stacks with Leica LAS-AF software and Zeiss ZEN blue softwares. Images were compiled using Adobe Photoshop CS6 and Adobe Photoshop CC 2015. All image panels represent z-stacks unless mentioned. The area of callus formation at the cut end of detached organs was measured using ImageJ software.

Total RNA was extracted from samples (see supplementary Materials and Methods for further details of sample preparation) using the Nucleospin Plant RNA extraction kit (MN) and subjected to on-column DNase treatment according to the manufacturer's guidelines. cDNA was synthesised from 1 μg total RNA using a High-Capacity cDNA Reverse Transcription kit (Applied Biosystems). qRT-PCR was performed in a 25 μl reaction volume containing 12.5 μl SYBR Green PCR master mix (Takara Bio), 100 nM gene-specific primers (Table S1) and 100 ng cDNA in a CFX96 Touch Real-Time PCR Detection System. All reactions were performed with RNA derived from three independent biological replicates. Each biological sample was tested in technical triplicate. Data were normalised to ACTIN2 (ACT2). The transcript level in the control was normalised to 1. The expression of the gene of interest is represented with respect to the control (as performed by Kareem et al., 2015). The relative gene expression is represented as fold-change value by calculating −ΔΔC^T.

The luciferase assay was performed as described by Díaz-Triviño et al. (2017). Healthy Nicotiana benthamiana plants (3-4 weeks old) grown under long-day conditions (16 h light, 8 h dark) were used for agroinfiltration. The primers used for cloning are listed in Table S2. Competent cells of the C58 strain of Agrobacterium tumefaciens were used for the infiltration.

ChIP was performed by following the protocol as described by Yamaguchi et al. (2014) (see supplementary Materials and Methods for a brief description). ChIP-qPCR was performed using SYBR Premix (Clontech) to determine PLT5 protein occupancy on the CUC2 promoter region. The relative fold enrichment of CUC2 DNA was calculated by computing the enrichment in PLT5::PLT5-vYFP relative to plt3;plt5-2;plt7. ACTIN7 (ACT7) was used to normalise the results between the samples. The ChIP-qPCR reactions were performed in triplicate. The primers used for ChIP-qPCR are listed in Table S3.

Pearson's χ² test (regeneration assay analysis), Welch's two-sample t-test (qRT-PCR data analysis), Mann–Whitney U one-tailed test (luciferase assay) and Kruskal–Wallis χ² test (comparing the number of closed vein loops) were used for data analysis. The Holm–Bonferroni correction was performed for multiple analysis when using Pearson's χ² test. R programming was used for the statistical analyses.

We thank Dr Philip N. Benfey, Dr Elliot Meyerowitz and Dr Ottoline Leyser for their valuable suggestions on the early draft of the manuscript. We are grateful to Dr Charles Melnyk and Dr Utpal Nath for all the discussions and help. pG10-90:vYFP control vector for luciferase assay was received from Menno Pijnenburg. We are thankful to M. R. Krishna Prashanth for assistance with schematics and figures, B. Subhiksha for cloning AtPLT5::OsPLT2-vYFP and Sajesh Vijayan for help with the statistical analyses.

The peer review history is available online at https://dev.biologists.org/lookup/doi/10.1242/dev.185710.reviewer-comments.pdf