Main

The periderm is a protective outer tissue established during secondary growth in many seed plants. It consists of several layers containing three different cell types: the phellem, phellogen and phelloderm1,2 (Fig. 1a). The outermost cell type, phellem (also known as cork), differentiates from the outer daughter cells of dividing phellogen cells. The differentiating phellem cells deposit lignin and suberin in their cell walls to form a physical barrier to protect secondary tissues from biotic and abiotic stresses1. As the periderm is the interface between the plant’s internal tissues and its environment, it is prone to injury. Periderm is re-established at wound sites in tree trunks, potato tubers and some fruits (such as apple)3,4,5. Periderm regeneration at wound sites is critical to prevent water loss and pathogen entry through wounds4. In addition to being essential for survival, the phellem of some tree species (such as cork oak) has been used by humans as manufacturing material for thousands of years. Understanding the mechanisms of periderm re-establishment would therefore have substantial biological and economic value. Even though the formation of periderm at wounds has been studied for more than a century, its underlying mechanism is still largely unknown.

Fig. 1: Mechanical injury elicits periderm regeneration from vascular tissues in Arabidopsis roots.
figure 1

a, A cross-section of wild-type roots. Lignin (red) and suberin (yellow) were stained. b, Schematic of the wounding experiment. Roots within 5 mm below the root–hypocotyl junction were cut longitudinally (light brown, phellem; green dotted line, phellogen; dark brown, phelloderm; grey, phloem region; magenta dotted line, vascular cambium; blue, xylem region). c, The promoter activity of PER15 (left), AT3G26450 (centre) and WOX4 (right) at 1, 2 and 3 dai, respectively. In the intact periderm, the promoter activity of PER15, AT3G26450 and WOX4 was preferentially detected, respectively, in both the phellogen (PG) and young phellem (PL), in the dividing cells (presumably the phellogen), and in the phellogen and strongly in the phelloderm (PD). d, Cross-sections of wild-type roots at 1, 2 and 4 dai of 17-day-old roots. Col-0, Columbia-0. In c,d, orange and blue arrowheads indicate the normal periderm or the wound sites, respectively; red arrowheads point to thin cell walls parallel to the wound surface, indicating recent cell divisions at 2 dai; the insets are the magnified images of the wound sites. White rectangles mark empty corners of stitched images. e, proPXY:GUS signals at the wound site at 2 dai and 4 dai of 17-day-old roots. White arrowheads indicate the wound sites. f, proPXY:GUS signal strength at the wound site in the seedlings at 2 and 4 dai. n indicates the number of examined wound sites. Two-sided Fisher’s exact test was used (**P < 0.01). Venus signal intensities in c and intensity of suberin staining with Fluorol yellow in d are shown according to the colour scales on the right. The top (brighter) area of the scale represents a higher intensity of signals. White, SR2200 (cell wall). Fractions on the panels indicate the proportion of cross-sections showing a similar expression as in the images. Scale bars, 50 µm (a,c,d) and 0.5 mm (e).

Source Data

The accumulation of the gaseous hormone ethylene controls plant growth and development. When ethylene diffusion is limited by the surrounding environment, the increased concentration serves as a developmental signal for plant tissues. For instance, ethylene diffusion is sensed by root tips to monitor soil compaction; in compact soil, ethylene accumulates in and near the root, resulting in fortified growth6. Aerenchyma formation and internode elongation in rice are induced by the accumulation of ethylene caused by limited diffusion under waterlogged conditions7,8. Here we show that the Arabidopsis barrier tissue, the periderm, limits the diffusion of two gaseous molecules, ethylene and oxygen, and we propose that monitoring the accumulation or depletion of these two gases functions as a system for maintaining periderm integrity.

Arabidopsis roots regenerate periderm

To examine whether the periderm regenerates in Arabidopsis roots, we longitudinally cut the mature part of roots (Fig. 1b) and observed the morphological changes and the expression of the periderm reporter genes PEROXIDASE15 (PER15), PER49, PYK10-binding protein 1 (PBP1), AT3G26450, WUSCHEL RELATED HOMEOBOX 4 (WOX4) and AT1G14120 (refs. 9,10,11,12). We found that the reporters showing their expression in phellem (PER15, PER49, PBP1 and AT1G14120) were induced at the wound site 1 day after injury (dai; Fig. 1c and Extended Data Fig. 1a,c). Phellogen characteristics appeared at 2 dai, as indicated by the appearance of the phellogen-preferred reporter AT3G26450 and periclinal (that is, parallel to the cut surface) cell divisions at the wound site (Fig. 1c,d and Extended Data Fig. 1c). WOX4 expression was detected less frequently at the wound site at 2 dai (4 out of 29 sections) and more consistently at 3 dai (Fig. 1c and Extended Data Fig. 1c). At 4 dai, the surface-exposed cells showed lignification and suberization, indicating that phellem-like layers were established at the wound site (Fig. 1d and Extended Data Fig. 1b). Phellem-like cells were adjacent to the inner cells that were actively dividing, thus resembling phellogen. These observations show that periderm regenerates at the wound site in Arabidopsis roots.

Next we examined the functionality of the re-established barrier. We used proPXY:GUS, in which the GUS (β-glucuronidase) gene is expressed under the control of the PHLOEM INTERCALATED WITH XYLEM (PXY) promoter, which is active in the vascular cambium and xylem parenchyma, the inner tissues of the mature root13 (Extended Data Fig. 1d). Collected roots were incubated in a buffer containing the GUS substrate X-Gluc. We reasoned that an intact periderm would inhibit the entry of X-Gluc into the mature root, resulting in reduced GUS signal. Consistent with this, the GUS signal was stronger in the wounded region than in the intact region of proPXY:GUS roots at 2 dai (Fig. 1e). GUS signal intensity became weaker at 4 dai, indicating that the integrity of the re-established periderm barrier was sufficiently restored to prevent X-Gluc penetration (Fig. 1e,f). The change in GUS signal levels was not caused by changes in promoter activity, as proPXY:erVenus fluorescence levels remained unchanged after the injury and during regeneration (Extended Data Fig. 1d,e). Taken together, our results show that phellem identity, as indicated by the expression of phellem markers, is established at 1 dai, followed by phellogen-like cell divisions at 2 dai and phellem differentiation (deposition of lignin and suberin) by 4 dai, coinciding with regained barrier function.

Ethylene diffusion promotes regeneration

Next we examined the mechanisms underlying periderm regeneration. As plants sense injury as stress14, we investigated whether stress-related hormones affect the induction of PER15, one of the periderm genes induced early in regeneration. proPER15:erVenus seedlings were treated with methyl jasmonate, abscisic acid or 1-aminocyclopropane-1-carboxylate (ACC), a precursor of ethylene, for 1 dai. Whereas neither methyl jasmonate nor abscisic acid treatment affected PER15 induction, ACC treatment significantly reduced induction (Fig. 2a,b and Extended Data Fig. 1g,h). Using other reporter lines, we found that the induction of most periderm markers was also reduced by ACC treatment at 1 dai (Extended Data Fig. 2a,b,e). We also found that PER15 and PBP1 expression and phellem formation were not affected in intact roots following ACC treatment (Extended Data Fig. 2c,d,f,g). Therefore, it seems that there are other signals for periderm development in intact tissues. Although ACC could act independently from ethylene signalling15,16, ACC treatment did not affect PER15 induction at 1 dai in ethylene insensitive 2 (ein2-1) and ethylene response 1 (etr1-3) mutants. We also showed that ethylene treatment reduced PER15 and PBP1 induction at the wound site (Extended Data Fig. 2l,m,o). Taken together, these results indicate that canonical ethylene signalling is required to suppress periderm gene induction (Extended Data Fig. 2h–m,o). We next assessed whether ACC treatment affects suberized cell formation at the wound site. Whereas suberized cells in control roots formed a continuous layer at the wound site, treatment with ACC after injury occasionally resulted in discontinuous suberized cell layers or callus-like structures at the wound site (Fig. 2c,d). The discontinuous suberized cell layers at the wound site were also found following ethylene treatment (Extended Data Fig. 2n,o). To examine the functionality of the re-established barriers in ACC-treated roots, we tested X-Gluc penetration. Even though PXY promoter activity was repressed following ACC treatment at 4 dai, proPXY:GUS signals near the wound site were stronger in ACC-treated roots than in the untreated control at 4 dai (Extended Data Fig. 1d–f). The combination of stronger GUS signals and reduced promoter activity indicates that the barrier is less functional in ACC-treated roots. Altogether, these results show that ethylene impedes periderm re-establishment at the wound site.

Fig. 2: Ethylene diffusion through the wound reduces ethylene signalling and triggers periderm regeneration.
figure 2

a, Cross-sections of 18-day-old proPER15:erVenus roots at 1 dai grown on MS plates supplemented without (mock) or with 10 µM ACC (ACC) for 1 dai. b, The proportion of cells at the wound site showing Venus yellow fluorescent protein (YFP) signal intensities above the threshold was quantified at 1 dai in mock- or ACC-treated 18-day-old proPER15:erVenus roots. c, Cross-sections of 21-day-old wild-type roots at 4 dai grown on MS (mock) or 10 µM ACC-supplemented MS (ACC) plates for 4 dai. White box marks empty corner of stitched image. White arrowheads indicate gaps in the suberized cell layer. d, The density of suberized cells at the wound site was quantified in mock- and ACC-treated 21-day-old wild-type seedlings at 4 dai. e, Cross-sections of 17-day-old RPS5A:erVenus-EBF1UTR roots without injury or 8 h after the injury. White arrowheads indicate an increase in Venus signals. f, Concentration of ethylene emitted from roots 3 h after injury. The mature part of the wild-type roots was intact (control) or injured (wounded). FW, fresh weight. g, Cross-sections of 18-day-old proPER15:erVenus roots at 1 dai grown without (cut) or with lanolin (cut + lanolin) at the wound for 1 dai. h, The proportion of cells at the wound site showing Venus signal intensities above the threshold was quantified in 18-day-old proPER15:erVenus seedlings at 1 dai. Two-tailed Wilcoxon rank-sum test was used in b,d,h, and two-tailed Welch’s t-test was used in f (*P < 0.05, **P < 0.01). n indicates the number of examined cross-sections in b,d,h and the number of repeats in f. For descriptions of the different elements for all box plots, see Methods, ‘Statistics and reproducibility’. Venus signal intensities in a,e,g and intensity of suberin staining with Fluorol yellow in c are shown according to the colour scales on the right. The top (brighter) area of the scale represents a higher intensity of signals. White, SR2200 (cell wall). Scale bars, 50 µm.

Source Data

As it has been reported that ethylene production is increased after mechanical injury17, the suppression of periderm regeneration by ethylene seems counter-intuitive. To examine the ethylene signalling level, we generated a dynamic ethylene reporter line, RPS5A:erVenus-EBF1UTR, in which erVenus fused with the 3′ untranslated region (UTR) of EIN3-BINDING F BOX PROTEIN 1 (EBF1) is expressed under the control of the constitutive RPS5A promoter (Extended Data Fig. 3a,e). As erVenus-EBF1UTR translation is inhibited by the EIN2 protein in the presence of ethylene18, the Venus signal will be detected only in cells in which the ethylene signalling level is low. Thus, low Venus fluorescence levels indicate high ethylene signalling levels. We validated this reporter line in both the root apical meristem and intact or wounded root secondary tissues using ACC and inhibitors of ethylene biosynthesis or signalling, aminoethoxyvinylglycine (AVG) and AgNO3, respectively. Validation confirmed that RPS5A:erVenus-EBF1UTR accurately reports ethylene signalling levels (Extended Data Figs. 3b,c,f–i and 4a). Following periderm injury, RPS5A:erVenus-EBF1UTR signal intensity was greater (that is, ethylene signalling lower) in the phloem parenchyma near the wound at 2 dai (Extended Data Fig. 3d,k). In the control line RPS5A:erVenus, expression was not affected by injury, indicating that the increase in the RPS5A:erVenus-EBF1UTR signal was because of enhanced erVenus-EBF1UTR translation rather than enhanced RPS5A promoter activity (Extended Data Fig. 3d,j). These observations demonstrate that ethylene signalling is reduced after wounding, which is consistent with the negative role of ethylene in periderm regeneration.

Altogether, our data indicate the existence of a mechanism to reduce ethylene signalling following wounding. This led us to reason that ethylene gas is released from the wound into the surrounding environment. Lignin and suberin in differentiated phellem normally prevent gas exchange through the surface, allowing ethylene to accumulate in unwounded secondary tissues. Following periderm injury, ethylene leaks through the wound, decreasing ethylene signalling and thus triggering periderm regeneration. To test this hypothesis, we examined expression changes in the ethylene signalling reporter line at earlier time points, as ethylene diffusion should occur right after injury. RPS5A:erVenus-EBF1UTR signal intensities increased in the exposed distal phloem region already at 8 h and more clearly at 11 h after injury, demonstrating a rapid reduction in ethylene signalling levels after injury (Fig. 2e and Extended Data Fig. 4b). We also measured ethylene emission by roots using gas chromatography. We detected higher ethylene emission from wounded roots compared with control roots at all measured time points (Extended Data Fig. 4c). The ethylene concentration was significantly higher 3 h after injury (Fig. 2f), indicating an increase in emitted ethylene immediately after injury. We also investigated reporter gene expression and periderm re-establishment when gas leakage was prevented by covering the wound with lanolin or Vaseline or by submerging the seedlings in liquid Murashige and Skoog (MS) medium. The RPS5A:erVenus-EBF1UTR signal was weaker at the wound site in covered roots than in roots with uncovered wounds at 1 dai (Extended Data Fig. 5b). The reduction was not due to reduced promoter activity, as RPS5A:erVenus remained unaltered after covering the wound site (Extended Data Fig. 5a). These observations demonstrate that a high level of ethylene signalling is maintained when a wound is physically sealed. We consistently found that most periderm marker induction at 1 dai and suberized cell layer formation at 4 dai failed to occur when the wound was sealed with lanolin immediately after wounding (Fig. 2g,h and Extended Data Fig. 5c–e). PER15 activation was also decreased at 1 dai when seedlings were submerged in liquid MS immediately after injury to limit gas exchange (Extended Data Fig. 5f). Altogether, these results indicate that ethylene diffusion from the wound leads to a decrease in its signalling level, and this reduction promotes periderm regeneration.

Oxygen flows into tissues after injury

As ACC treatment did not fully inhibit suberized cell formation at the wound site, we inferred that there might be other, probably gaseous, regulators involved in periderm regeneration. We considered oxygen as a candidate. Oxygen is consumed in cellular respiration during tissue growth, but external oxygen cannot easily enter secondary tissues because of the poor permeability of phellem layers. We therefore reasoned that secondary tissues are normally under physiological hypoxic conditions and that oxygen flows into the tissue through the wound following periderm injury and promotes periderm regeneration.

To investigate this hypothesis, we generated transcriptional reporter lines of PLANT CYSTEINE OXIDASE (PCO) genes (proPCO1:erVenus and proPCO2:erVenus) whose expression is increased under hypoxic conditions19. The PCO1 and PCO2 promoters were active in mature roots, and reporter expression was detected more strongly in the vascular region and the periderm, and weakly in the distal phloem parenchyma (Fig. 3a and Extended Data Fig. 6a). This result supports the hypothesis that inner cells in the mature root experience hypoxic conditions. Next we examined hypoxia signalling in the root after injury. The activity of both PCO1 and PCO2 promoters was weaker in whole tissues at 2 dai compared with immediately after injury (Fig. 3b and Extended Data Fig. 6b). We detected a reduction in PCO1 and PCO2 promoter activities already 16 h after the injury (Extended Data Fig. 6c), suggesting that oxygen entry quickly alleviates the hypoxic conditions. To examine whether the periderm affects the entry of oxygen, we measured the steady-state oxygen level by inserting an oxygen microsensor20 into wild-type secondary tissue (Fig. 3c). After the measurement, we partially peeled off the periderm and inserted the microsensor into the peeled region. Comparing the measurements before and after the removal of the periderm, we found that the oxygen level was higher in the root after the removal of the periderm, indicating that the periderm prevents the entry of oxygen (Fig. 3c and Extended Data Fig. 6d,e). Altogether, these data support the idea that oxygen enters through the wound and inhibits the hypoxic response in secondary tissues.

Fig. 3: Oxygen entry through the wound suppresses hypoxia signalling and promotes periderm regeneration.
figure 3

a,b, Cross-sections of 14-day-old intact proPCO1:erVenus roots (a) and 21-day-old proPCO1:erVenus roots right after injury (b, top) or at 2 dai (b, middle); and quantification of proPCO1:erVenus signals in the vascular region at 0 and 2 dai (b, bottom). Two-tailed Welch’s t-test was used (**P < 0.01). c, The quasi-steady-state oxygen concentration was measured and quantified before and after the periderm peeling. Paired Wilcoxon rank-sum test was used (*P < 0.05). d, Cross-sections of 25-day-old proPER15:erVenus roots at 1 dai grown on MS (mock) or 10 µM ACC-supplemented MS (ACC) plates in ambient air, or on MS (hypoxia) or 10 µM ACC-supplemented MS (ACC + hypoxia) plates under 5% oxygen concentration for 1 dai. e, The proportion of cells at the wound site showing Venus signal intensities above the threshold was quantified at 1 dai in mock ACC-treated, hypoxia-treated and ACC + hypoxia-treated proPER15:erVenus roots. Kruskal–Wallis test followed by Dwass–Steel–Critchlow–Fligner pairwise comparisons was used (different letters indicate statistically significant differences between two groups; P < 0.01). f, Cross-sections of 21-day-old ate1-2;ate2-1 roots at 4 dai grown on MS (mock) or 10 µM ACC-supplemented MS (ACC) plates for 4 dai. g, The proportion of suberized cell formation at the wound site in mock and ACC-treated 21-day-old wild-type and ate1-2;ate2-1 roots at 4 dai. Two-sided Fisher’s exact test was used to test for significant differences between mock- and ACC-treated ate1-2;ate2-1 (**P < 0.01). h, Schematic of periderm integrity surveillance mechanisms mediated by ethylene and oxygen diffusion. n indicates the number of examined cross-sections in b,e,g and the number of repeats in c. Venus signal intensities in a,b,d and intensity of suberin staining with Fluorol yellow in f are shown according to the colour scales on the right. The top (brighter) area of the scale represents a higher intensity of signals. White, SR2200 (cell wall). Scale bars, 50 µm.

Source Data

Ethylene and hypoxia act additively

To test whether a reduction in hypoxia signalling is required for periderm regeneration, we grew proPER15:erVenus seedlings under 5% oxygen concentration for 1 dai. We found that hypoxia treatment significantly reduced PER15 induction at the wound site, suggesting that PER15 induction requires a decrease in hypoxia signalling (Fig. 3d,e). To further validate the role of hypoxia signalling, we used arginine transferase (ate1;ate2) and proteolysis6 (prt6) mutants that show a constitutively active hypoxia response because the target proteins for the N-degron pathway, which includes hypoxia signalling regulators, are stable in these mutants regardless of the oxygen level21,22,23,24. At 4 dai, ate1-2;ate2-1 roots often showed only half or no suberized layer formation at the wound site (Fig. 3f,g). These findings imply that maintaining a high level of hypoxia signalling inhibits periderm regeneration.

To examine whether ethylene and hypoxia signalling additively regulate periderm regeneration, we investigated PER15 induction following ACC treatment under hypoxia. As ACC oxidase requires oxygen to convert ACC into ethylene, we first studied whether ACC can increase ethylene signalling in the hypoxia conditions we used. ACC treatment under ambient conditions reduces primary root growth25, promotes root hair formation26 and stabilizes 35S:EIN3-GFP in the root tips27. We found that a 1-day treatment with ACC separately or combined with hypoxia showed similar growth defects and EIN3–GFP induction (Extended Data Fig. 6h), indicating that ACC treatment activates ethylene signalling even in our low-oxygen conditions. When plants were treated with ACC for 1 dai under hypoxic conditions, PER15 induction at the wound site was more severely suppressed than with ACC treatment or hypoxia alone (Fig. 3d,e). To further examine the combined effect, we treated prt6-5 and ate1-2;ate2-1 mutants with ACC after injury. Whereas untreated prt6-5 roots had suberized cell formation similar to the wild type at the wound site, ACC-treated prt6-5 roots showed more frequent formation of callus-like structures (Extended Data Fig. 6f,g). In addition, most ACC-treated ate1-2;ate2-1 mutant roots did not form suberized cells at the wound site (Fig. 3f,g). These results demonstrate that the reductions in ethylene and hypoxia signalling act additively in periderm regeneration (Fig. 3h).

We also examined whether the normal (that is, not regenerated) periderm formation is affected in ate1-2;ate2-1 mutants treated with ACC during secondary development. We did not find defects in suberization in the phellem, indicating that normal periderm formation is more robust than regeneration and is regulated by factors other than ethylene and oxygen (Extended Data Fig. 2g).

The additive effect between ethylene and hypoxia is consistent with the strong suppression of PER15 induction at the wound site when seedlings were submerged, which results in high levels of ethylene and hypoxia signalling after injury (Extended Data Fig. 5f). To examine whether lowering either ethylene or hypoxia signalling is sufficient to induce PER15 expression at the wound site of submerged seedlings, we used ein2-1 or etr1-3 and fully oxygenated liquid MS medium. Whereas proPER15:erVenus showed less frequent PER15 induction at the wound site in wild-type roots submerged in control liquid MS at 1 dai, we detected clear but variable PER15 induction following injury when the plants were submerged in oxygenated liquid MS, suggesting that supplementing oxygen partially rescues PER15 induction (Extended Data Fig. 7a,d). In the ein2-1 background, oxygenated liquid MS medium more stably rescued PER15 induction at the wound site compared with the case in the wild type (Extended Data Fig. 7b,e). We also observed a slight but not significant rescue in the etr1-3 background. (Extended Data Fig. 7c,f). Furthermore, we examined suberization of cells under submerged conditions. Liquid MS medium without gas supplementation reduced the level of suberized cell formation at the wound site in wild-type, ein2-1 and etr1-3. However, we found that when submerged in aerated or oxygenated liquid MS medium, the ethylene signalling mutant roots showed a significant increase in the density of suberized cells at the wound site compared to wild-type roots grown under liquid MS medium without gas supplementation (Extended Data Fig. 7g–j). Overall, these results indicate that lowering ethylene or hypoxia signalling rescues periderm regeneration under submerged conditions that prevent the rapid diffusion of gases.

Next we investigated the mechanism that terminates periderm regeneration. When the periderm is re-established at the wound site, gas diffusion once again becomes limited, which may be the cue to terminate regeneration. To determine whether diffusion of ethylene and oxygen is suppressed after periderm re-establishment, we observed the expression of ethylene and hypoxia signalling reporters. RPS5A:erVenus-EBF1UTR signal intensity decreased (that is, ethylene signalling increased) in the phloem parenchyma near the wound at 5 dai compared with 2 dai, although the RPS5A promoter activity remained unchanged (Extended Data Fig. 8a,b,e). The proPCO1:erVenus and proPCO2:erVenus signals were also stronger in whole tissues at 4 dai than at 2 dai (Extended Data Fig. 8c,d,f). Thus, both hypoxia and ethylene signalling returned to pre-injury levels, supporting the idea that diffusion of ethylene and oxygen is reduced after the periderm is re-established (Fig. 3h).

We next examined the phenotypes of ein2-1 and etr1-3 mutants, in which ethylene signalling is inactive even after periderm re-establishment, to determine whether this results in a failure to terminate periderm regeneration. At 6 dai, ein2-1 roots showed a significant increase in the density of suberized cells at the wound site (Extended Data Fig. 9a,b), suggesting a failure to precisely terminate the periderm regeneration. To study the possible role of wound-induced ethylene biosynthesis in regeneration, we used ethylene biosynthesis (AVG) and signalling inhibitors (AgNO3). These inhibitors had no effect on PER15 and PBP1 induction at 1 dai, with the exception of a slight reduction in PER15 expression following AVG treatment. These results indicate that ethylene biosynthesis following wounding may not be required for periderm gene activation and that the further reduction in ethylene signalling level does not enhance the induction of periderm genes (Extended Data Fig. 9c,d,f,g). However, consistent with the phenotype in ein2-1 roots, our findings showed an increased density of suberized cells at the wound site at 6 dai following the inhibitor treatment (Extended Data Fig. 9e,h). These findings support the hypothesis that ethylene signalling is required to precisely terminate the phellem differentiation process.

Gas-mediated barrier monitoring in shoot

The above results indicate that root periderm integrity is monitored through diffusion of ethylene and oxygen. We therefore examined whether the integrity of other barriers is monitored in the same way. Arabidopsis inflorescence stems do not develop periderm or suberized cell layers; instead, the epidermis with cuticle layer acts as a barrier28. When inflorescence stems were longitudinally cut (Fig. 4a), we found that expression of PER15 and PER49 (but not other examined periderm genes) appeared at the wound site 1 dai, and a suberized cell layer was established at the wound site 4 dai (Fig. 4b and Extended Data Fig. 10a). These observations indicate that a phellem-like layer is formed at the wound site in the inflorescence stems. Next we tested whether the re-establishment of a barrier in inflorescence stems is also mediated by the diffusion of gases. In accordance with this idea, sealing the wound with lanolin or Vaseline inhibited PER15 induction at 1 dai and suberized cell formation at 4 dai (Fig. 4c–e).

Fig. 4: Barrier integrity is monitored with gas-mediated surveillance system in shoot.
figure 4

a, Schematic of the wounding experiment in inflorescence stems. Inflorescence stems were longitudinally cut (orange bracket region). b, A cross-section of proPER15:erVenus (left) and wild-type (right) inflorescence stem at 1 and 4 dai, respectively. c, Cross-sections of proPER15:erVenus inflorescence stems at 1 dai grown without (cut) or with (cut + lanolin) lanolin or Vaseline at the wound for 1 dai. d, Quantification of Venus signal intensities at the wound site in proPER15:erVenus inflorescence stems at 1 dai. The wound was not (cut) or was (cut + lanolin) covered with lanolin or Vaseline right after the injury. Two-tailed Wilcoxon rank-sum test was used (**P < 0.01). e, Cross-sections of wild-type inflorescence stems at 4 dai. The wound was not (cut) or was (cut + lanolin) covered with lanolin for 4 dai. f, Concentration of ethylene emitted from inflorescence stems 3 h after injury. The wild-type inflorescence stems were not (control) or were (wounded) injured along the longitudinal axis. Two-tailed Welch’s t-test was used (*P < 0.05). g, RPS5A:erVenus-EBF1UTR signals in inflorescence stems at 2 dai. The inflorescence stems grown without (cut) or with (cut + Vaseline) Vaseline at the wound after the injury. h, Venus signal intensities in the cortical cells near the wound were measured and normalized with RPS5A:erVenus-EBF1UTR inflorescence stems grown without Vaseline. Two-tailed Wilcoxon rank-sum test was used (**P < 0.01). n indicates the number of examined cross-sections in d,h and the number of repeats in f. Venus signal intensities in b,c,g and intensity of suberin staining with fluorol yellow in b,e are shown according to the colour scales on the right. The top (brighter) area of the scale represents a higher intensity of signals. White, SR2200 (cell wall) in b,c,e. Fractions on the panels indicate the proportion of cross-sections showing similar expression to that in the images. Scale bars, 50 µm.

Source Data

To examine whether ethylene and/or oxygen diffuse through the wound to promote barrier re-establishment in inflorescence stems, we first investigated hypoxia signalling. Unlike in roots, PCO1 and PCO2 expression was maintained, and ate1-2;ate2-1 stems did not show defects in suberized cell layer formation following injury (Extended Data Fig. 10b,c). There was thus no clear indication of either reoxygenation following injury or a role for hypoxia signalling in barrier re-establishment in inflorescence stems. We next examined ethylene diffusion. We found that, as in roots, wounded stems emitted a higher concentration of ethylene than unwounded controls (Fig. 4f and Extended Data Fig. 10d). Furthermore, when wounds were sealed with Vaseline immediately after injury, we detected variable but weaker RPS5A:erVenus-EBF1UTR signals in the stems at 2 dai compared with controls (Fig. 4g,h and Extended Data Fig. 10e,h,i). These results indicate that gaseous molecules, such as ethylene, may diffuse through the wound. We therefore investigated the role of ethylene in barrier re-establishment. Whereas PER15 induction was less frequently detected at the wound site of inflorescence stems at 1 dai in ACC-treated seedlings, ein2-1 and etr1-3 inflorescence stems showed suberized cell layer formation similar to the wild type at 2 and 4 dai (Extended Data Fig. 10f,g,j). Considering that sealing the wound inhibited PER15 induction more strongly than ACC treatment and that there was no observable phenotype in ein2-1 and etr1-3, the diffusion of gaseous or volatile molecules other than ethylene could be necessary for barrier re-establishment in inflorescence stems. Altogether, these results support the idea that monitoring of barrier integrity through gas diffusion is also used to re-establish the barrier in the inflorescence stems.

Discussion

Previous studies have shown that wounding-induced ethylene biosynthesis is associated with periderm regeneration29. Here we show that mature Arabidopsis roots contain ethylene, and wounding leads to its release into the environment. This results in a reduction of ethylene signalling and, consequently, initiation of periderm regeneration. In addition, we demonstrate that hypoxia signalling is reduced owing to oxygen diffusion through the wound, leading to enhanced periderm regeneration. Whereas individual manipulations of ethylene and hypoxia have a modest effect on periderm regeneration, the combination of high ethylene and constitutive hypoxia signalling almost completely abolishes periderm regeneration. Overall, we show that the periderm integrity in the mature Arabidopsis root is monitored by the reciprocal diffusion of oxygen and ethylene, in and out of the root, respectively. During the formation of the Casparian strip and the embryonic cuticle, barrier integrity is monitored by specific localization of peptide hormone receptors and by spatial separation of peptide processing and perception, respectively30,31,32,33. The gas diffusion monitoring mechanism described here, which probably does not require a spatially restricted signalling mechanism, functions as a control system to initiate periderm regeneration and later terminate the process when hypoxia and ethylene signalling levels are restored owing to regained barrier integrity. However, because gases diffuse, precise positional cues for periderm formation are unlikely to be conveyed by these two gases. It is probable that other factors, such as peptides, other phytohormones and mechanical stresses34, operate together with the two gases to position the regenerating periderm. From this perspective, the two gases would establish a permissive environment for the other signals to enable barrier regeneration. Further studies are required to understand these details of periderm regeneration.

We showed that injuring the epidermis of the Arabidopsis stem results in the formation of a phellem-like layer in the exposed tissue. Even though barrier integrity in the stem also seems to be monitored by gas diffusion, ethylene and hypoxia signalling do not have major roles in this process. Similar to in the Arabidopsis stem, formation of the periderm or suberized cells is elicited in some fruits when there is a crack in the waxy cuticle layer in the exocarp, the outermost barrier cell type35,36. It would thus be interesting to study whether similar mechanisms of monitoring gas leakage are used for barrier re-establishment in different plant organs and species. Previous reports show that plants adapt their growth and development to their environment on the basis of the accumulation of ethylene in compacted soil and waterlogged conditions6,7,8. Considering this together with our findings, we propose that monitoring the accumulation or depletion of gaseous molecules inside tissues or in the surrounding environment is a common strategy for controlling plant development.

Methods

Plant materials and growth conditions

Col-0 was used as the wild type unless stated otherwise. ein2-1, etr1-3, ate1-2;ate2-1, prt6-5, proWOX4-erYFP, proPXY:GUS, proPXY:erVenus and 35S:EIN3-GFP have been described previously10,13,27,37,38,39,40. ate1-2;ate2-1 mutants were reciprocally backcrossed into Col-0 for three generations and transfer DNA insertions were confirmed by genotyping to obtain double-homozygous mutants.

Seeds were sown on half-strength MS (Duchefa) plates supplemented with 0.05% MES (Duchefa), 1% agar (Duchefa) and 1% sucrose (Duchefa), pH 5.8. MS medium also contained vitamins for all experiments except for ethylene measurement from roots. After incubation at 4 °C for more than 2 days, the plates were moved to the growth chamber (22 °C; 16 h light, 8 h dark). The day that the plates were placed in the growth chamber is defined as day 0. The seedlings were moved to the new MS agar plate around day 7; for the experiment in shoots, the seedlings were transferred to soil 1 or 2 weeks after germination and were grown in a greenhouse (16 h light, 8 h dark) unless stated otherwise. For examining normal periderm, we used 14- or 16-day-old seedlings as periderm is established at that age. For assessing regeneration, we usually used 17- to 21-day-old seedlings.

The Arabidopsis Genome Initiative locus codes for the genes are as follows: ATE1, AT5G05700; ATE2, AT3G11240; EBF1, AT2G25490; EIN2, AT5G03280; EIN3, AT3G20770; ETR1, AT1G66340; PBP1, AT3G16420; PCO1, AT5G15120; PCO2, AT5G39890; PER15, AT2G18150; PER49, AT4G36430; PRT6, AT5G02310; PXY, AT5G61480; RPS5A, AT3G11940; WOX4, AT1G46480.

Cloning

From the Col-0 genome, the 643-base-pair (bp) EBF1 3′ UTR with its downstream 72-bp sequence (from +2780 to +3495) was amplified by PCR using H122_EBF1_URT_F and H123_EBF1_URT_R primers18 (Supplementary Table 1), which have a 25-bp overlap with the erVenus and attL2 sequence, respectively (EBF1 3′ UTR). erVenus/pDONR plasmid (30 ng) was amplified with the EBF1 3′ UTR PCR product (600 ng) by PCR. The amplified PCR product was digested with DpnI enzyme at 37 °C for 2 h and was used for transformation into Escherichia coli (DH5α) by electroporation (erVenus–EBF1 3′ UTR/pDONR). RPS5A:erVenus–3AT/VD8034GW-mTurq and RPS5A:erVenus–EBF1UTR–3AT/FRm43GW were generated from RPS5A/pDONR, 3AT/pDONR, FRm43GW, VD8034GW-mTurq and erVenus/pDONR or erVenus–EBF1 3′ UTR/pDONR, respectively, by MultiSite Gateway LR clonase reaction.

From the Col-0 genome, promoter sequences of 1.1 kilobases (kb) of PCO1 (from −958 to +198) and 2.5 kb of PCO2 (from −2213 to +298) were amplified by PCR using H68_proPCO1_F and H69_proPCO1_R, or H70_proPCO2_F and H71_proPCO2_R primers, respectively23 (Supplementary Table 1). The amplified PCR products were cloned into the p1R4z-pDONR vector by BP reaction (proPCO1/p1R4z-pDONR, proPCO2/p1R4z-pDONR). proPCO1:erVenus–3AT/FRm43GW and proPCO2:erVenus–3AT/FRm43GW were generated from erVenus/pDONR, 3AT/pDONR41, FRm43GW42, and proPCO1/p1R4z-pDONR or proPCO2/p1R4z-pDONR, respectively, by MultiSite Gateway LR clonase reaction.

The plasmids generated in this work were introduced into Col-0.

Surgical injury of the periderm and chemical treatment

Roots within 5 mm below the root–hypocotyl junction were used for the wounding experiment unless stated otherwise. Under a dissection microscope, the shoot was pulled upwards slightly to create tension in the roots; the roots were longitudinally cut with a razor blade. As roots where the cut reached the vascular cambium region tended to form a callus-like structure instead of the wound periderm at 4 dai, we focused on sections in which the depth of the cut reached between the phloem parenchyma contacting the periderm to the phloem-side cambium in the following analysis unless stated otherwise. For the analysis of promoter induction at the wound site, we also excluded the sections in which the cut was just at the primary phloem pole because reporter expression tended not to be induced. To peel off the periderm for the oxygen level measurement, we made a shallow cut on the surface of the roots tangentially; the cut edge at the wound site was grasped with forceps and pulled towards the root tip.

For chemical treatment, 100 mM ACC (Merck) in water, 100 mM abscisic acid (ABA; Duchefa) in ethanol, 100 mM jasmonic acid (JA; Sigma-Aldrich) in dimethylsulfoxide, 100 mM AVG (Sigma-Aldrich) in water and 100 mM AgNO3 (Sigma-Aldrich) in water were prepared as stock solutions. Right after the injury, seedlings were moved to plates supplemented with 10 µM ACC, 10 µM JA, 10 µM ABA, 10 µM AVG or 0.5 mM AgNO3. For ACC treatment and its mock control, the plates were sealed with surgical tape and Parafilm. In all other experiments, plates were sealed only with surgical tape. For ABA and JA treatment, the MS agar plates containing an equivalent volume of ethanol or DMSO were used as the controls. For ethylene treatment, plates were placed in a 2-l container and 600 µl of 10% ethylene gas was injected (final concentration: 30 ppm). To ensure a consistent ethylene concentration, the container was opened and ethylene was re-injected every 1 or 2 days.

For the wounding experiment in shoot, we used an inflorescence stem whose length was between 10 and 15 cm. The region 1 to 2 cm below an inflorescence meristem was used for the surgery; we longitudinally cut the stem with a razor blade under a dissection microscope. For ACC treatment in shoot, seedlings were grown on MS agar plates for around 19 days. Inflorescence stems approximately 5 mm in length were used and a region of 2 to 3 mm below the inflorescence meristem was longitudinally injured. After the injury, seedlings were transferred to MS agar plates supplemented with or without 10 µM ACC. The plates were sealed with surgical tape and Parafilm.

To seal the wound site, lanolin was mixed with the equivalent volume of milliQ water. We applied lanolin or Vaseline to the wound after the injury. We also applied lanolin or Vaseline at the wound of control samples just before sampling to ensure that lanolin and Vaseline equally affected the histological processes.

For submergence treatment, the seedlings were first submerged into MS liquid medium not supplemented with sucrose, and their roots were injured. The seedlings were placed in the growth chamber for 24 h.

For oxygen treatment with reporter lines, MS liquid medium without sucrose was aerated with ambient air or oxygen gas for more than 10 min at 20 to 22 °C. Immediately before the surgical injury, the seedlings were submerged into aerated or oxygenated MS liquid medium and their roots were injured with a razor blade. The seedlings were transferred to glass medium bottles filled with aerated or oxygenated MS liquid medium and were placed in the growth chamber for 24 h. The method for oxygen treatment with wild-type and mutants was the same, with the exceptions that the seedlings were not submerged into MS liquid medium before the injury and they were placed in the growth chamber for 4 dai.

For hypoxia treatment, plants were grown under 5% oxygen conditions using a Whitley H85 hypoxystation (Don Whitley Scientific). Seedlings were treated for 24 h.

Oxygen measurement

Oxygen measurement was performed using O2 microsensors as previously described with modifications20. We used 24- or 25-day-old wild-type roots for the measurement. Shoot and roots below the region for the measurement were fixed on a metal mesh with plastic tape and rubber bands, and the metal mesh was set in a chamber. After we inserted an O2 microsensor (Unisense A/S) into the region 8 to 10 mm below the root–hypocotyl junction at the depth of 70 µm from the surface, the chamber was filled with aerated deionized water. Quasi-steady-state of oxygen concentration was measured under light (50–60 µeinstein) and dark conditions; we defined the quasi-steady state as a state in which a change in oxygen concentration was less than 5% of the total concentration for at least 5 min. The microsensor was retracted from the root, and the aerated water was drained. To measure oxygen levels when periderm integrity was compromised, the metal mesh was retrieved from the chamber, and the surface of roots used for the intact root measurement was peeled off as described above. The metal mesh was again set in the chamber and the microsensor was inserted at the depth of 50 µm in the surface-removed region. The measurement was performed in the same way as described above. Using the same roots for the measurement, the respiration and photosynthetic rate should be almost identical, and the oxygen intrusion rate would be the main cause of the oxygen level changes before and after the surgery. After the measurement, the roots were retrieved to perform histological analysis.

Ethylene measurement

We injured roots within 3 to 4 cm below the root–hypocotyl junction in 24-, 25- or 26-day-old seedlings and removed their shoot by cutting off the hypocotyl. The hypocotyl cut sites were sealed with Vaseline. Immediately, roots were moved into a vial (volume 37.5 ml) that had been partially filled with approximately 25 ml 1% agar in 1× PBS. Plants were kept in the growth chamber (22 °C; 16 h light, 8 h dark), and samples were taken at the indicated times. After detecting the level of ethylene, we measured the fresh weight of roots and the headspace of the vial. The used Vaseline weight was subtracted from the measured fresh weight, and the corrected weight was used for quantification. Around 20 seedlings were used for measurement. For inflorescence stems, regions within 10 to 15 cm below the inflorescence meristem were used for the measurement; 3- to 5-cm stems were collected, and the flower stalks were cut. The fresh weight was measured, and the top, bottom and the flower stalk cut sites were sealed with Vaseline. For wounding, the stems were cut in the longitudinal direction with a razor blade. Immediately, the inflorescence stems were placed into a vial partially filled with approximately 25 ml 1% agar in 1x PBS. Around 30 inflorescence stems were used for measurement. In both cases, to measure the amount of ethylene emitted, 1-ml samples from the headspace were collected using a syringe 3, 6 and 24 h after the vial was closed. Ethylene measurements were performed using a Clarus 480 gas chromatograph (PerkinElmer) carrying a HayeSep N (80–100 MESH) 584 column. The oven temperature was 100 °C, with the flame ionization detector temperature set to 150 °C. The flow rate of the nitrogen carrier gas was 20 ml min−1. Peak area was integrated and compared against a standard curve.

GUS staining and microtome sectioning

Roots from 0 to 3 cm below the root–hypocotyl junction with a wound were submerged in 90% acetone on ice for 30 min. The samples were washed with 50 mM sodium phosphate buffer twice and were submerged in the GUS staining solution under vacuum for 30 to 60 min. The samples were incubated with GUS staining solution (30 mM Na2HPO4, 20 mM NaH2PO4, 1.5 mM K4Fe, 1.5 mM K3Fe, 500 mg l−1 X-Gluc, 0.1% Triton) at 37 °C until sufficient GUS signals were detected. The roots were fixed in fixation solution (50 mM sodium phosphate (pH 7.4), 4% formaldehyde, 1% glutaraldehyde) at 4 °C overnight. To observe the samples from a lateral view under a dissection microscope, we kept the samples in 70% ethanol to remove chlorophyll.

Vibratome sectioning and staining methods

The samples were fixed in 4% paraformaldehyde in 1× PBS for 30–60 min and washed with 1× PBS twice. The samples were embedded in 4% agarose in 1× PBS and 200 µm-thick cross-sections were made using a vibratome. The cross-sections were stained in 1× PBS or ClearSee43 supplemented with 1 µl ml−1 Renaissance SCRI 2200 (SR2200; Renaissance Chemicals) to stain cell walls44. To stain lignin, the cross-sections were stained in ClearSee supplemented with 1 µl ml−1 SR2200 and 50 µg ml−1 Basic Fuchsin (Sigma-Aldrich). For suberin staining, the cross-sections were stained in ethanol supplemented with 0.01% Fluorol Yellow 088. Before observation, the cross-sections were washed with 1× PBS supplemented with 1 µl ml−1 SR2200. To visualize lignin and suberin at the same time, first the cross-sections were stained in ClearSee supplemented with 1 µl ml−1 SR2200 and 50 µg ml−1 Basic Fuchsin overnight. Next, suberin staining was performed as described above, and before observation, the cross-sections were washed with 1×PBS supplemented with 1 µl ml−1 SR2200. To observe the root tip, the samples were fixed in 4% paraformaldehyde in 1× PBS for 30 to 60 min and washed with 1× PBS twice. The samples were then stored in ClearSee supplemented with 1 µl ml−1 SR2200.

Microscopy and data analysis

A Leica 2500 microscope (Leica) was used for light microscopy images and Leica SP5 (Leica), Stellaris (Leica) and LSM880 (Zeiss) confocal laser scanning microscopes were used to detect GFP, YFP, Venus, Basic Fuchsin, Fluorol Yellow 088 and SR2200.

Confocal images were stitched by using an ImageJ plugin (Pairwise stitching)45. Occasionally, this resulted in the formation of empty corners in the figure panels. To visually separate the empty corners from the black background signal of the microscopy image, we filled these empty corners with white colour.

Fluorescence signal intensities at the wound site were quantified using Fiji (v1.53) and PlantSeg. We defined the wound site as a vascular tissue located within 30 µm from the wound surface for roots. For quantification in roots, the cell wall images of the wound site were extracted by using Fiji, and the extracted images were used for segmentation by PlantSeg46. After manual modification, the segmented images were used to count the cell number at the wound site and to measure the Venus–YFP signal intensity in each cell by Fiji. The Venus signal intensities were measured with the minimum signal threshold to exclude the background signals. The Venus-positive cells were defined by whether the signal intensity was above the threshold specific to each line. Five to six roots were used for each treatment in each experiment; a maximum of two cross-sections of the same root at different positions were used for quantification. Cross-sections with cuts reaching the vascular cambium were excluded from quantification because they tended to show unstable periderm gene induction. For Venus signal intensity quantification in intact proPER15:erVenus and proPBP1:erVenus roots in Extended Data Fig. 2f, signal intensities were measured in periderm from the quarter of the cross-section with the minimum signal threshold. For quantification of periderm reporter lines in Extended Data Fig. 1c, five distal phloem cells in the intact tissues or near the wound in each cross-section were used for Venus signal intensity measurement with the minimum signal threshold. For RPS5A:erVenus and RPS5A:erVenus-EBF1UTR quantification in Extended Data Figs. 3h,i, 5a,b and 8e, five to ten distal phloem parenchyma cells in each cross-section were used for Venus signal intensity measurement with the minimum signal threshold. At 8 and 11 h after injury or at 2 and 5 dai, distal phloem parenchyma cells near the wound were selected for signal measurement. The average of Venus signal intensities from five to ten cells was used for statistical analysis. For proPXY:erVenus quantification in Extended Data Fig. 1e, we circled the vascular cambium and xylem parenchyma region and measured Venus signal intensities within the circle with the minimum signal threshold to exclude the background signals. The signal intensities were normalized with those of uninjured roots in each repeat. For proPCO1:erVenus and proPCO2:erVenus signal quantification in Fig. 3b and Extended Data Figs. 6b,c and 8f, the vascular cambium, xylem parenchyma and phloem parenchyma region were circled, and we measured Venus signal intensities within the circle with the minimum signal threshold to exclude the background signals. For RPS5A:erVenus and RPS5A:erVenus-EBF1UTR quantification in Extended Data Fig. 3j,k, Venus signal intensities were measured with the minimum signal threshold and averaged from five epidermal cells in each roots. For RPS5A:erVenus and RPS5A:erVenus-EBF1UTR quantification in Fig. 4h and Extended Data Fig. 10h,i, 10 cortical cells or 15 cortical cells near the wound in each cross-section were, respectively, used for Venus signal intensity measurement with the minimum signal threshold. For Extended Data Fig. 10i, the signal intensities were normalized with those of control inflorescence stems in each repeat. T2 or T3 lines were used. For proPER15:erVenus quantification in shoots, the wound site within 40 µm from the wound surface was selected and the Venus signal intensities were measured with the minimum signal threshold in each cross-section. For quantification of suberized cell formation in inflorescence stems in Extended Data Fig. 10j, the length of the total wound site and the region covered with suberized cells were measured and the ratio of the suberized region was calculated.

proPXY:GUS signals at the wound site were classified into three categories under a stereo microscope: weak when there were no or faint signals, strong when the signals were close to saturation, or intermediate when the signals were clearly visible and weaker than the strong category.

The density of suberized cells at the wound site was calculated by dividing suberized cell number by the length of the wound site using Fiji. For the quantification, the area connected to the original periderm was excluded because it was occasionally hard to distinguish which suberized cells were originated from vascular tissue. The cross-sections showing callus-like structure formation were also excluded as suberized cell formation and morphology at the wound site is affected by callus-like structure. The mature suberized cells (more than 20 µm2) were counted. For Extended Data Fig. 5e, when the re-established periderm had three gaps in the suberized cell layer or three successive cell files did not show suberized cells, we defined it as a failure in suberized cell layer formation.

Data analysis was performed with MS Excel v2308, R (v2024.04.2) and Python 3.9.

Statistics and reproducibility

In the box plots, the 25th, 50th (central value) and 75th percentile are marked with horizontal lines within the box. The ends of the whiskers indicate the maximum and minimum values within 1.5× the interquartile range from the box ends. Outliers are shown above or below the whiskers. For quantification of suberized cell formation and gene induction at the wound site or gene expression level in the vascular region, each dot corresponds to a cross-section. For ethylene concentration, each dot indicates a repeat. All experiments were repeated at least three times (three times for Figs. 1a,c–e2c,e3a,b and 4b,c,e and Extended Data Figs. 1a,b,d,f,g, 2a–d,l,m3a,b(AVG and AgNO3),c,d4a,b, 5a–d,f6a–c,h7b,c,g–i8a–d9a,c–h and 10a–c,e,g; four times for Figs. 2a,g3d and Fig. 4f and Extended Data Figs. 2g,h–j,n3b(ACC) and Extended Data Fig. 10d; five times for Fig. 4g and Extended Data Figs. 4c and 5e; six times for Extended Data Figs. 6f7a and 10f; nine times for Fig. 3f; all are biological repeats). Exact P values are provided in Supplementary Table 2.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.