Enzymes contributing to the hydrogen peroxide signal dynamics that regulate wall labyrinth formation in transfer cells

Abstract

Transfer cells are characterized by an amplified plasma membrane area supported on a wall labyrinth composed of a uniform wall layer (UWL) from which wall ingrowth (WI) papillae arise. Adaxial epidermal cells of developing Vicia faba cotyledons, when placed in culture, undergo a rapid (hours) trans-differentiation to a functional epidermal transfer cell (ETC) phenotype. The trans-differentiation event is controlled by a signalling cascade comprising auxin, ethylene, apoplasmic reactive oxygen species (_apoROS), and cytosolic Ca²⁺. Apoplasmic hydrogen peroxide (_apoH₂O₂) was confirmed as the _apoROS regulating UWL and WI papillae formation. Informed by an ETC-specific transcriptome, a pharmacological approach identified a temporally changing cohort of H₂O₂ biosynthetic enzymes. The cohort contained a respiratory burst oxidase homologue, polyamine oxidase, copper amine oxidase, and a suite of class III peroxidases. Collectively these generated two consecutive bursts in _apoH₂O₂ production. Spatial organization of biosynthetic/catabolic enzymes was deduced from responses to pharmacologically blocking their activities on the cellular and subcellular distribution of _apoH₂O₂. The findings were consistent with catalase activity constraining the _apoH₂O₂ signal to the outer periclinal wall of the ETCs. Strategic positioning of class III peroxidases in this outer domain shaped subcellular _apoH₂O₂ signatures that differed during assembly of the UWL and WI papillae.

A polarized extracellular hydrogen peroxide signal, generated by cooperative activities of spatially localized class III peroxidases and a catalase, regulates wall labyrinth assembly in trans-differentiating epidermal transfer cells.

Introduction

Transfer cells are characterized by a unique wall labyrinth (WL). In general, transfer cell WLs are one of two architectural types, flange or reticulate (Talbot et al., 2002), with reticulate being the predominant type found throughout the plant kingdom (Andriunas et al., 2013). Flange WLs are comprised of parallel arrays of complex bar-like structures resembling annular secondary wall thickenings of tracheary elements (Fig. 4; Talbot et al., 2002). In contrast, the reticulate type consists of a uniform wall layer (UWL) from which wall ingrowth (WI) papillae arise (Fig. 1D; Andriunas et al., 2013). As they protrude into the cytoplasm, WI papillae can branch and fuse to form a complex WL of repeating fenestrated layers of wall material (Figs 2, 5, 6; Talbot et al., 2001). The contours of these layers are lined with an amplified plasma membrane surface area that confers an enhanced nutrient transport capacity at key apo-/symplasmic interfaces. Examples where reticulate transfer cells can be found include interfaces at the soil/root (Schikora and Schmidt, 2002) and biotroph/host (e.g. Rodiuc et al., 2014; Bartlem et al., 2014), loading/unloading sites of vascular pipelines, and maternal/filial junctures of developing seeds (e.g. Andriunas et al., 2013; Yuan et al., 2016).

Fig. 4.

Effect of pharmacological blockade of ascorbate peroxidase or catalase activities on the _apoH₂O₂ catabolic flux of trans-differentiating epidermal transfer cells of V. faba cotyledons. Following specified culture times on MS medium, cotyledons were transferred to MS medium containing ±5 mM 5,5′-dithiobis-2-nitrobenzoic acid (DNTP; ascorbate peroxidase inhibitor) or ±10 mM 3-aminotriazole (3-AT; catalase inhibitor) plus pharmacological agents to block _apoH₂O₂ biosynthesis, and held for 1 h at 4 °C. Thereafter, _apoH₂O₂ catabolic fluxes were computed from _apoH₂O₂ catabolic rates (pmol min^–1), measured using Amplex Red ±inhibitors in the reaction solution at 26 °C, expressed on a cotyledon adaxial surface area (mm²) basis. Data are means ±SEs for 10–15 replicate cotyledons. Statistical differences for treatment means against control (**P<0.01).

Fig. 1.

Effect of _apoROS on the formation of the wall labyrinth (A) or wall ingrowth papillae alone (B), temporal profiles of _apoH₂O₂ biosynthetic/catabolic fluxes (C), and spatial distribution of the _apoH₂O₂ signature (D–F) in adaxial epidermal transfer cells (ETCs) of cultured V. faba cotyledons. (A) Cotyledons were cultured for 15 h on MS medium alone (control; Ctrl) or in the presence of scavengers for O₂^– and ·OH, namely 0.5 mM n-propyl gallate (PG) and 10 mM sodium benzoate (SB), respectively. An additional treatment included the n-propyl gallate-exposed cotyledons receiving an H₂O₂ supplement. Epidermal peels were scored for percentage epidermal cells with WI papillae. (B) Cotyledons were cultured for 9 h on MS medium to allow formation of the UWL before exposure to media containing ±O₂^– and H₂O₂ scavengers (see A) for a further 6 h, and scored for percentage epidermal cells with WI papillae. (C) Temporal profiles of biosynthetic/catabolic fluxes of _apoH₂O₂ in the presence/absence of 5 mM NaN₃ computed from _apoH₂O₂ biosynthetic/catabolic rates (pmol min^–1), measured using Amplex Red, and expressed on a cotyledon adaxial surface area (mm²) basis. (D–F) Light micrographs of transverse sections of trans-differentiating ETCs showing polarization of the _apoH₂O₂ signature to their outer periclinal wall detected as a brown-yellow precipitate of 3',3-diaminobenzidine during (D) 0–2 h, (E) 5–7 h, and (F) 11–13 h of cotyledon culture. Dark blue arrowheads indicate the brown-yellow 3',3-diaminobenzidine precipitate; the light blue arrowhead in (D) indicates an intercellular space. Data for (A) and (B) are means ±SEs for a minimum of five replicate cotyledons with 150 cells examined per replicate; for (C), a minimum of six replicate cotyledons per time point. Mean values with the same letter are not significantly different (P>0.05); statistical differences for treatment means against control (**P<0.01). Scale bar=10 μm.

Fig. 2.

Levels of biological control at which _apoH₂O₂ biosynthesis/catabolism was regulated during trans-differentiation of adaxial epidermal transfer cells of cultured V. faba cotyledons. (A) _apoH₂O₂ biosynthesis: cotyledons cultured in MS medium alone (control; Ctrl) for 0, 0.5, or 6 h were transferred into MS medium containing ±100 μM 6-methylpurine (6MP), 100 μM cycloheximide (CHX), or 100 μM CHX+100 μM 6MP for 4 h at 4 °C and then transferred into MS medium ±treatments at 26 °C for a further 0.5, 5.5, or 6 h. (B) _apoH₂O₂ catabolism: cotyledons cultured in MS medium alone for 0, 0.5, or 6 h were transferred into MS medium ±the H₂O₂ biosynthesis inhibitors 5 mM salicylhydroxamic acid (SHAM) or 5 mM SHAM+100 μM 2-bromoethylamine hydrobromide (BEA) as well as ±100 μM CHX or 100 μM 6MP, and held for 4 h at 4 °C. Cotyledons were then transferred into MS medium containing H₂O₂ biosynthesis inhibitor(s) ±CHX or 6MP at 26 °C for a further 0.5, 5.5 (SHAM), or 6 h (SHAM+BEA). Thereafter, biosynthetic/catabolic fluxes were computed from _apoH₂O₂ biosynthetic/catabolic rates (pmol min^–1), measured using Amplex Red, expressed on a cotyledon adaxial surface area (mm²) basis. Data are means ±SEs for six replicate cotyledons. Statistical differences for treatment means against control (**P<0.01).

Fig. 5.

The role of catalase in regulating the polarized distribution of the _apoH₂O₂ signal in trans-differentiating epidermal transfer cells (ETCs) of cultured V. faba cotyledons. (A–F) Light micrographs of transverse sections of trans-differentiating ETCs showing the distribution of the _apoH₂O₂ signature in the absence (A, C, E) or presence (B, D, F) of a catalase inhibitor, 3-aminotriazole (3-AT), during 0–2 h (A, B), 5–7 h (C, D), and 11–13 h (E, F) of cotyledon culture. Cotyledons were cultured for 0 (A, B), 5 (C, D), or 11 h (E, F) on MS medium alone before being transferred into MS medium containing ±10 mM 3-AT and held for 1 h at 4 °C. Thereafter, cotyledons were transferred into MS medium containing 3',3-diaminobenzidine ±3-AT at 26 °C for a further 2 h. _apoH₂O₂ distribution was detected as a brown-yellow 3',3-diaminobenzidine precipitate (dark blue arrowheads); light blue arrowhaeads indicate precipitates or air accumulated in intercellular spaces. (G) Percentage of total pixel intensities of 3',3-diaminobenzidine precipitates per ETC detected in their outer periclinal (OP), anticlinal (Anti), and inner periclinal (IP) walls. Data are means ±SEs of a minimum of four replicate cotyledons with 50 cells measured per replicate. Scale bar=10 μm.

Fig. 6.

Effect of pharmacological blockade of _apoH₂O₂ biosynthetic enzymes on the subcellular localization of _apoH₂O₂, detected as electron-dense precipitates of cerium perhydroxides, in the outer periclinal wall and plasma membrane (PM) of trans-differentiating adaxial epidermal transfer cells (ETCs) of cultured V. faba cotyledons. (A, C, E) Representative electron micrographs of ETC transverse sections exhibiting cerium perhydroxide precipitates and (B, D, F) their accompanying total relative _apoH₂O₂ levels overlaying the outer periclinal cell wall and associated PM. Cotyledons were cultured for (A) 0, (C) 5, or (E) 11 h on MS medium alone before being transferred into MS medium containing ±specified pharmacological agents, and held at 4 °C for 1 h followed by 1 h at 26 °C. Thereafter, cotyledons were transferred to MS medium containing 5 mM CeCl₃ together with ±specified pharmacological agents and ±10 mM ascorbic acid (negative control) for 1 h at 26 °C. Total pixel intensities of cerium perhydroxide precipitates overlaying their outer periclinal wall (green arrowheads) and associated PM (pink and red arrowheads, between and ensheathing WI papillae, respectively) were collected from binary images containing only the black cerium perhydroxide precipitates on a white background (e.g. see A', C', and E') corrected for background and expressed on a per cell width basis (B, D, F). Orange arrows mark CeCl₃ precipitates formed outside the original cell wall commonly observed in plant cells as an artefact (Lherminier et al. 2009). Cyt, cytoplasm; OW, original cell wall; PM, plasma membrane; UWL, uniform wall layer; WI, wall ingrowth papillae. Data are means ±SEs of a minimum of four replicate cotyledons, and 10 cells were examined per replicate. Scale bar=500 nm. Statistical differences for treatment means against control (**P<0.01).

Nematode-induced giant cells and developing monocot and eudicot seeds have provided valuable insights into signals and mechanisms regulating transfer cell WL formation (e.g. Andriunas et al., 2013; Cabrera et al., 2014; Thiel, 2014; Yuan et al., 2016; Diaz-Manzano et al., 2018). Induction of WL formation in nematode-induced giant cells and developing seeds is preceded by an accumulation of auxin in transfer cell precursor cells (Andriunas et al., 2013; Cabrera et al., 2014; Yuan et al., 2016). The auxin spike acts to regulate WL formation exclusively through driving a transcriptionally dependent burst in transfer cell-specific ethylene production and a downstream ethylene signalling cascade (Andriunas et al., 2013; Cabrera et al., 2014; Thiel, 2014). In contrast, for adaxial epidermal cells of developing Vicia faba cotyledons trans-differentiating to functional epidermal transfer cells (Andriunas et al., 2013), ethylene induces production of a polarized apoplasmic reactive oxygen species (_apoROS) signal and, in concert with the _apoROS, a cytosolic calcium ([Ca²⁺]_cyt) signal (Andriunas et al., 2012; Xia et al., 2012; Zhang et al., 2015a, b). The _apoROS signal activates cell wall biosynthesis and defines outer lateral polarity for WL deposition (Andriunas et al., 2012). Plumes of elevated [Ca²⁺]_cyt determine foci at which WI papillae are deposited through remodelling an actin-dependent delivery of vesicles carrying cargoes of wall building materials (Zhang et al., 2015a, 2017a). Strikingly, a recent ETC-specific transcriptome study in which each signal was silenced by selective pharmacological blockade found that the _apoROS signal exerted the greatest regulatory influence over gene expression (Zhang et al., 2017b).

Given the strong transcriptional influence of _apoROS in trans-differentiating ETCs of V. faba cotyledons (Zhang et al., 2017b) and unanswered issues embodied in our previous study of ROS-regulated WL formation (Andriunas et al., 2012; Xia et al., 2012; see below), it was considered timely to verify the ROS responsible for regulating WL formation and identify the enzymes forming the polarized _apoROS signal and their cellular/subcellular organization. The studies by Andriunas et al. (2012) and Xia et al. (2012) did not exclude the possibilities that: (i) the superoxide ion radicle (O₂^–) or the hydroxyl radicle (∙OH) regulate deposition of the WL (Kärkönen and Kuchitsu, 2015); (ii) at a concentration of 100 μM, diphenyleneiodonium chloride (DPI) could well have inhibited ROS biosynthetic enzymes other than respiratory burst oxidase homologues (rbohs), such as class III peroxidases (Prxs), copper amine oxidases (CAOs), and polyamine oxidases (PAOs; Kärkönen and Kuchitsu, 2015); and (iii) whether ROS action on WI papillae formation was direct or indirect resulting from regulating UWL deposition which forms an essential platform for WI papillae construction (Xia et al., 2017). In addition to addressing these questions, we made use of the ETC transcriptome to identify gene candidates expressing H₂O₂ biosynthetic and catabolic enzymes (Zhang et al., 2017b). The activities of candidate enzymes were pharmacologically blockaded to evaluate their role in generating the _apoROS signal. Pharmacological blockade combined with histochemical localization of _apoH₂O₂-formed precipitates of 3',3-diaminobenzidine (Andriunas et al., 2012) and CeCl₃ (Lherminier et al., 2009) were used to deduce the cellular and subcellular distribution, respectively, of the biosynthetic/catabolic enzymes forming the _apoH₂O₂ signal in the trans-differentiating ETCs.

Overall these investigations led to the conclusion that _apoH₂O₂ was the ROS essential for both UWL and WI papillae deposition. The _apoH₂O₂ signal was produced by a composite of ROS biosynthetic enzymes that varied temporally and included rboh, PAOs, Prxs, and CAOs. Prxs, acting through Cu/Zn superoxide dismutases (Cu/ZnSODs), represented the key enzyme groups contributing to _apoH₂O₂ levels (Minibayeva et al., 2015). Catalase (CAT) functioned as the sole _apoH₂O₂ catabolic/scavenging enzyme (Podgórska et al., 2017) and was responsible for localizing the _apoH₂O₂ signal to the outer periclinal wall of trans-differentiating ETCs. Within this domain, the _apoH₂O₂ signal was localized to subcompartments of the PM and WL.

Materials and methods

Plant growth conditions and cotyledon culture

For details of Faba bean (V. faba L. cv. Fiord) plant growth and culture of their developing cotyledons, see Zhang et al. (2017a).

Visualizing wall labyrinth formation

For TEM, tissue segments, with a 2×3 mm base of epidermal transfer cells, were fixed on ice for 4 h with 3% glutaraldehyde and 3% formaldehyde freshly prepared from paraformaldehyde containing 100 mM sucrose and 2 mM CaCl₂ in 25 mM cacodylate buffer (pH 7.0) followed by post-fixation, dehydration, and resin embedment (see Farley et al., 2000). Ultrathin sections were visualized with a JEOL 1200 EX II transmission electron microscope (JEOL, Japan). Percentages of ETCs with a UWL along with estimates of original and UWL volumes per ETC were determined (see Xia et al., 2017). For SEM (Phillips, The Netherlands) to visualize WI papillae, see Zhang et al. (2015c). For recording the mean percentage of cells containing WI papillae, see Zhou et al. (2010).

Hydrogen peroxide biosynthetic and catabolic flux assays

Amplex Red reagent (10-acetyl-3,7-dihydrophenoxazine; Invitrogen, Australia) was used to determine fluxes of _apoH₂O₂ biosynthesis and catabolism. The _apoH₂O₂ biosynthetic fluxes (pmol min^–1 mm^–2 of adaxial cotyledon surface) were determined following Andriunas et al. (2012).

To estimate _apoH₂O₂ catabolic fluxes (pmol min^–1 mm^–2 of adaxial cotyledon surface), cultured cotyledons were transferred into fresh medium containing pharmacological agents to block _apoH₂O₂ biosynthesis. After 1 h at 4 °C, cotyledon cultures were transferred to 26 °C. Then, at specified times, a set of cultured cotyledons were removed, blotted on paper towels to remove residual medium from their adaxial surface, and washed in 200 μl of Amplex Red reaction mixture for 1 min to displace all their _apoH₂O₂ (e.g. McDonald et al., 1995). Thereafter, captured _apoH₂O₂ levels were determined at a series of time points and fitted by linear regression to yield estimates of catabolic fluxes (see Supplementary Fig. S1 at JXB online).

Histochemical detection of hydrogen peroxide

Cellular distribution of _apoH₂O₂ in cultured cotyledons was visualized using 3',3-diaminobenzidine (Sigma, Australia) that generates an insoluble brown-coloured precipitate upon binding with H₂O₂. For the procedure, see Andriunas et al. (2012), modified as follows. Fresh transverse cotyledon sections (55–65 µm thick) were cut with a Vibratome (Leica). Micrographs of the sections were processed through the Photoshop CS6 level command to intensify the brown colour for optimal visualization (Supplementary Fig. S2). Fiji software (https://fiji.sc/) measured total pixel intensities of 3',3-diaminobenzidine precipitates (RawIntDen option) in the four walls of each ETC transverse section. Cells not exposed to 3',3-diaminobenzidine were recorded as background and used to correct total pixel intensities of the 3,3'-diaminobenzidine precipitates. Distribution of total pixel intensities in the outer (OPW) and inner (IPW) periclinal and anticlinal (1/2AW1 + 1/2AW2) wall were expressed as percentages of their summed total per ETC (i.e. OPW+IPW+1/2AW1 + 1/2AW2).

Subcellular localization of H₂O₂ was detected by CeCl₃ that, in the presence of H₂O₂, forms electron-dense precipitates of cerium perhydroxide using a method adapted from Lherminier et al. (2009). Cultured cotyledons were washed (3× 3 min) with 50 mM cacodylate buffer and incubated for 1 h at 26 °C in a freshly prepared solution of ±5 mM CeCl₃ in 50 mM cacodylate buffer (pH 7.2). Absence of CeCl₃ provided estimates of cell wall/PM background level, while CeCl₃ in 10 mM ascorbic acid served as a negative control. Following incubation, cotyledons were washed (4× 3 min) with 50 mM cacodylate buffer. Tissue segments were fixed for 2 h in 1.25% (v/v) glutaraldehyde and 1.25% (w/v) formaldehyde in 100 mM cacodylate buffer (pH 7.2), washed (3 ×10 min) with 100 mM cacodylate buffer, and post-fixed with 1% (w/v) osmium tetroxide in 50 mM cacodylate buffer for 1 h all at room temperature, followed by dehydration, resin embedment, and sectioning for TEM (see Xia et al., 2017). TEM micrographs of ETCs were optimized for total pixel intensity measurement by setting to a common cytoplasmic brightness (using the Expose option in Photoshop). These images were then adjusted to 25% of the thresholding histogram in Fiji software to obtain binary images containing black cerium perhydroxide precipitates on a white background. The free-hand selection tool of ImageJ was used to select specified cell wall and PM sectors in which total pixel intensity of cerium perhydroxide precipitates was determined (RawIntDen option).

RNA-seq analysis of genes encoding ROS enzymes

A transcriptome database of trans-differentiating ETCs and their underlying storage parenchyma cells of cultured V. faba cotyledons, annotated in Mapman Mercator and KEGG (Zhang et al., 2017b), was used to investigate expression profiles of transcripts with RPKMs (reads per kilobase of transcript per million mapped reads) >1 and encoding full-length sequences of proteins participating in _apoH₂O₂ biosynthesis and catabolism. Differentially expressed genes (DEGs) were defined as those exhibiting log₂fold change (FC) values >1 between two consecutive time points with a false discovery rate- (FDR) corrected P-value of <0.05 calculated using LimmaR (Ritchie et al., 2015). Subcellular localization of encoded proteins was predicted using ProtComp 9.0 plant (Softberry, USA) and WoLF PSORT. Only those proteins with a predicted apoplasmic to intracellular score ratio of ≥2-fold are reported. The presence of transmembrane domains (including both α-helices and β-barrels) and GPI (glycosylphosphatidylinositol) anchor modification sites was employed to test whether the apoplasmic proteins were PM localized by interrogating their full-length sequences using the following algorithms: TMHMM2.0 (http://www.cbs.dtu.dk/services/TMHMM/), PRED-DMBB2 (Tsirigos et al., 2016), and big-Pi plant (Eisenhaber et al., 2003). Unless specified otherwise, functions of the encoded proteins were inferred by best-fit percentage amino acid alignment with their closest Arabidopsis homologues using TAIR10 and Araport 11 databases.

Statistical analyses

Statistical differences between means were tested using a one-way ANOVA (P<0.05) or a Student’s t-test (*P<0.05; **P<0.01). Statistical analyses were performed in JMP 13 Statistical Software (SAS Institute).

Results

_apoH₂O₂ is the ROS responsible for regulating wall labyrinth formation

To address which _apoROS were active in regulating WL formation, cotyledons were cultured for 15 h in the presence/absence of the scavengers for O₂^– (n-propyl gallate; Zeng et al., 2017) or ·OH (sodium benzoate; Francoz et al., 2015), and scored for WL formation as assessed by the percentage of cells with WI papillae (Fig. 1A). Scavenging O₂^– inhibited WL formation by 76% while that of ·OH had no effect. A H₂O₂ supplement to the culture medium recovered the n-propyl gallate inhibition (Fig. 1A). Together with previously reported findings that the H₂O₂ scavengers CAT and reduced glutathione elicited comparable levels of WL inhibition (Andriunas et al., 2012), these responses are consistent with _apoH₂O₂ regulating WL formation. Whether H₂O₂ action on WI papillae formation was direct or indirect was addressed as follows. Cultured cotyledons were transferred to media containing ±n-propyl gallate once UWL formation was completed at 9 h, while recruiting cells to deposit WI papillae continued for the subsequent 6 h (Xia et al., 2017). In the presence of n-propyl gallate, further WI papillae formation was arrested (Fig. 1B). This response indicated that _apoH₂O₂ directly regulates both phases of WL formation.

_apoH₂O₂ biosynthesis is temporally variable but spatially invariant during wall labyrinth construction

To provide a framework to identify enzymes responsible for _apoH₂O₂ biosynthesis/catabolism, temporal profiles of _apoH₂O₂ biosynthetic and catabolic fluxes were measured during cotyledon culture (Fig. 1C). Blocking H₂O₂ catabolism by culturing cotyledons on NaN₃ (Arabaci and Usluoglu, 2013) exerted no influence on estimates of _apoH₂O₂ biosynthesis (Fig. 1C), indicating that unidirectional H₂O₂ biosynthetic fluxes were measured.

The temporal profile of the _apoH₂O₂ biosynthetic flux exhibited two bursts. The first and highest peak occurred at 0.5 h of cotyledon culture, with a 3.5-fold increase compared with time zero (Fig. 1C). From 0.5 h to 6 h, the _apoH₂O₂ biosynthetic flux declined, reaching its lowest level (50% of the peak) at 6 h before increasing to reach a second peak at 7 h that plateaued up to 12 h. The second peak was 22% lower than the first (Fig. 1C). Broadly, the temporal pattern of _apoH₂O₂ biosynthesis was similar to that reported by Andriunas et al. (2012), but with a more extended decline after the first peak and an _apoH₂O₂ biosynthetic flux about three times higher. The _apoH₂O₂ catabolic flux was substantially less, and exhibited smaller temporal shifts compared with the biosynthetic flux. Thus, similar catabolic fluxes occurred from 0 to 0.5 h before a 2-fold rise at 3 h corresponding to the biosynthetic flux slowing. Thereafter, the catabolic flux dropped 25% by 4.5 h, after which it remained steady up to 12 h (Fig. 1C).

The variable temporal profile of _apoH₂O₂ biosynthesis was coupled with a progressive recruitment of cells to assemble the UWL (Xia et al., 2017) and WI papillae (Wardini et al., 2007). In contrast, the polarized _apoH₂O₂ signal, irrespective of its magnitude, was present in most adaxial epidermal cells early (0–2 h) in cotyledon culture and remained so throughout WL formation (Fig. 1D–F).

What level(s) of biological control regulate the temporally variable _apoH₂O₂ fluxes?

Cotyledons were cultured in the presence/absence of the RNA synthesis inhibitor 6-methyl purine and the protein synthesis inhibitor cycloheximide, to evaluate regulation of the first and second bursts in _apoH₂O₂ biosynthesis/catabolism (Fig. 1C). The _apoH₂O₂ biosynthetic flux was inhibited by 25% in the presence of 6-methyl purine and cycloheximide alone, and in combination, between 0.5 h and 6 h (Fig. 2A). Since all treatments elicited the same inhibition, this suggests that only 25% of the H₂O₂ biosynthetic enzymes were subjected to transcriptional control, while the remaining decline in the biosynthetic flux must have been governed post-translationally; a conclusion that equally applies to the up-regulated flux at 0–0.5 h and 6–12 h. In contrast, catabolism was unaffected by 6-methyl purine and cycloheximide across all ETC developmental phases, indicating that control of catabolism was at an entirely post-translational level (Fig. 2B).

Potential enzymes regulating _apoH₂O₂ biosynthesis/catabolism

Transcriptomes of the developing ETCs and underlying storage parenchyma cells were used to identify genes encoding enzymes that may contribute to generating the ETC-specific _apoH₂O₂ signal (Supplementary Table S1). Since transcriptional control of _apoH₂O₂ signal formation was partial and temporally restricted to between 0.5 h and 6 h of cotyledon culture (Fig. 2A), the transcriptome analysis is limited to identifying potential players. The ETC transcriptome contained transcripts encoding an array of extracellular enzyme groups participating in _apoH₂O₂ biosynthesis. These included an ETC-specific rboh DEG, 11 Prxs, of which five were ETC-specific DEGs, a lipoxygenase, a quinone reductase, three Cu/ZnSODs, two PAOs, and two CAOs, of which one was an ETC-specific DEG. In the case of _apoH₂O₂-scavenging enzymes, transcripts encoding an ascorbate oxidase, two ascorbate peroxidases, CAT1, and two glutathione peroxidases, were detected (Supplementary Table S1). Prediction scores for their subcellular localization strongly supported an extracellular localization. The rbohB, lipoxygenase1, quinone reductase1-like1, CAOα2, ascorbate peroxidase3, and five Prxs contained predicted trans-membrane domains consistent with a PM insertion (Table 1). None of the genes encoded predicted GPI anchor sites.

Table 1.

Genes encoding extracellular H₂O₂-related enzymes expressed in adaxial epidermal transfer cells of cultured V. faba cotyledons

Gene ID	Annotation	AT homologue ID	% identity	Prediction score for:		Transmembrane domain
				Apoplasmic	Intracellular
Biosynthetic genes
Respiratory burst oxidase homologue (rboh)
CL4633.C3	VfrbohB	AT1G09090	81%	47.15	11.59	Y (α)
Cell wall (class III) peroxidase (Prx)
U23221	VfPrx12	AT1G17695	74%	51.41	11.22	N
CL10736.C1	VfPrx15A	AT2G38390	71%	24.72	10.85	Y (α)
CL2921.C1	VfPrx15B	AT2G38390	69%	51.04	11.37	Y (α)
CL8458.C2	VfPrx17	AT2G22420	79%	50.98	11.73	Y (α)
U9439	VfPrx18	AT2G24800	80%	44.76	11.86	N
U23436	VfPrx31	AT5G40150	77%	44.79	11.65	Y (α)
U23079	VfPrx42	AT4G21960	91%	51.37	11.58	N
U20848	VfPrx51	AT4G37530	81%	50.91	11.52	N
CL8648.C2	VfPrx52A	AT5G05340	89%	50.45	11.7	Y (α)
CL5501.C1	VfPrx52B	AT5G05340	72%	50.78	11.58	N
CL6044.C1	VfPrx71	AT5G64120	78%	44.08	11.87	N
Lipoxygenase (LOX)
U19514	VfPLOX1	AT4G39730	74%	22.41	10.37	Y (α)
Quinone reductase (FQR)
U8743; 9235	VfFQR1-like 1	AT4G27270	87%	37.42	11.74	Y (β)
Cu/Zn Superoxide dismutase (SOD)
U22912	VfSOD1	AT1G08830	89%	23.82	8.48	N
U18651	VfSOD2	AT2G28190	80%	22.34	10.49	N
U9482	VfSOD3	AT5G18100	88%	23.24	9.71	N
Polyamine oxidase (PAO)
U18440	VfPAO2	AT1G65840	82%	19.84	9.37	N
CL7801.C1	VfPAO4	AT3G43020	84%	18.52	9.2	N
Copper-containing amine oxidase (CAO)
U9780	VfCAOα2	AT1G31690	76%	22.84	10.72	Y (α)
U27547	VfCAOδ	AT4G12290	79%	24.51	11.77	N
Catabolic genes
Ascorbate oxidase (AO)
U15565	VfAO1	AT5G21105	77%	44.44	11.95	N
Ascorbate peroxidase (APX)
U20322	VfAPX2	AT3G09640	83%	22.76	11.03	N
U30095	VfAPX3	AT4G35000	84%	21.57	9.42	Y (α)
Catalase (CAT)
U29765	VfCAT1	AT1G20630	92%	24.13	4.67	N
Glutathione peroxidase (GPX)
CL8849.C1	VfGPX6	AT4G11600	82%	26.72	9.61	N
CL8849.C2	VfGPX6B	AT4G11600	82%	26.82	9.61	N

Reported for each gene is their percentage amino acid identity of the encoded protein to the closest Arabidopsis (AT) homologue, predicted scores for an apoplasmic and intracellular localization, together with the former evaluated for the presence of an amino acid motif consistent with a transmembrane domain (including an α-helix or β-barrel; for more details, see the Materials and methods). The spatial–temporal expression profiles of these transcripts are presented in Supplementary Table S1.

Pharmacological determination of enzyme groups contributing to the _apoH₂O₂ biosynthetic flux

Based on the ETC transcriptome (Table 1), genes encoding seven groups of enzymes were identified as potential contributors to the _apoH₂O₂ biosynthetic flux. For five groups, their contribution was evaluated by culturing cotyledons, for specified times, in the presence/absence of selected pharmacological inhibitors. These were: salicylhydroxamic acid (SHAM) which inhibits Prxs, lipoxygenases, and the alternative oxidase (Francoz et al., 2015; Chen and Fluhr, 2018); ibuprofen which inhibits lipoxygenase (Chen and Fluhr, 2018); DPI, which at low concentrations (<5 μM), specifically inhibits rbohs (Kärkönen and Kuchitsu, 2015); and 2-bromoethylamine hydrobromide (BEA) and N¹,N⁴-bis(2,3-butadienyl)-1,4-butanediamine (MDL 72527) which suppress CAO (An et al., 2008) and PAO (Takács et al., 2017) activities, respectively.

Responses to the pharmacological agents indicated that Prx, rboh, PAO, and CAO contributed to the temporal profile of the _apoH₂O₂ biosynthetic flux (Fig. 3A). That SHAM exclusively inhibited Prxs was concluded on the grounds that: (i) the _apoH₂O₂ biosynthetic flux was insensitive to the lipoxygenase inhibitor, ibuprofen, or a lipoxygenase substrate supplement of linoleic acid (Supplementary Fig. S3); and (ii) a SHAM (and a n-propyl gallate) inhibition of the alternative oxidase would be likely to cause an increase rather than a decrease in the _apoH₂O₂ biosynthetic flux (Fig. 3A) and associated WL formation (Fig. 1A). Throughout cotyledon culture, Prxs were the principal contributors to the _apoH₂O₂ biosynthetic flux, while the remaining enzyme groups underwent temporal change. For ETC precursor cells (t=0 h), CAOs (BEA) and rboh (DPI) acted in concert with Prxs. By the first _apoH₂O₂ burst (0.5 h), a combination of Prxs, rboh, and PAOs (MDL 72527) were operating. Between 3 h and 9 h of culture, the pharmacological treatments only identified Prx activity. During this phase, Prx activity was partially under transcriptional control (Supplementary Fig. S4). Thereafter, the CAO activity supplemented that of Prxs by 12 h of cotyledon culture (Fig. 3A).

Fig. 3.

Impact of pharmacological inhibitors of, or substrate supplementation for, specified H₂O₂ biosynthetic enzymes on _apoH₂O₂ biosynthetic fluxes by trans-differentiating epidermal transfer cells of cultured V. faba cotyledons. (A) Temporal responses of _apoH₂O₂ biosynthetic flux to inhibitors of specified H₂O₂ biosynthetic enzymes. Inhibitors used were: 5 mM salicylhydroxamic acid (SHAM) for cell wall peroxidase; 1 μM diphenyleneiodonium chloride (DPI) for respiratory burst oxidase homologue; 100 μM 2-bromoethylamine (BEA) for copper-containing diamine oxidases (CAOs); and 25 μM N¹,N⁴-bis(2,3-butadienyl)-1,4-butanediamine (MDL 72527) for polyamine oxidases (PAOs). ‘Mix’ indicates the effect of applying combinations of inhibitors against biosynthetically active enzymes at specific times points. These ‘mixes’ were for 0 h, SHAM+BEA+DPI; 0.5 h, SHAM+DPI+MDL 72527; 3, 6, and 9 h, SHAM alone; and 12 h, SHAM+BEA. (B) Effect of a supplementary supply of substrates for PAO (1 µM spermidine) and CAO (1 mM putrescine) on responses of the _apoH₂O₂ biosynthetic flux and the effect of inhibiting the putrescine response with 100 μM BEA. In all cases, cotyledons cultured for specified times were transferred to MS medium containing ±specified inhibitor or substrate for 1 h at 4 °C. Thereafter, _apoH₂O₂ biosynthetic fluxes were computed from _apoH₂O₂ biosynthetic rates (pmol min^–1), measured using Amplex Red ±inhibitors or ±substrate in the reaction solution at 26 °C and expressed on a cotyledon adaxial surface area (mm²) basis. Data are means ±SEs for 10–15 replicate cotyledons. Statistical differences for treatment means against control (*P<0.05; **P<0.01).

A closer inspection of the pharmacological data showed that the summed inhibition of each pharmacological agent at 0 and 0.5 h exceeded 100%, while those for 12 h approximated 100% of the control flux (Fig. 3A). Surprisingly, SHAM alone was found to exert a comparable inhibition to SHAM combined with the other inhibitors at 0 and 0.5 h (Fig. 3A). These responses suggested that CAO, PAO, and rboh must act coordinately to funnel their ROS product to the Prxs (Kärhönen and Kuchitsu, 2015). This conclusion is consistent with blocking Cu/ZnSOD activity in reducing O₂^– to H₂O₂ with diethyldithiocarbamate (Francoz et al., 2015). This treatment caused a comparable reduction in the _apoH₂O₂ biosynthetic flux to SHAM alone or in combination with diethyldithiocarbamate (Supplementary Fig. S5). Furthermore, the Cu/ZnSOD activity was found not to rate-limit the _apoH₂O₂ biosynthetic flux as indicated by the absence of any response when a Cu/ZnSOD supplement was added to the culture medium (Supplementary Fig. S5).

The combination of DPI and MDL 72527 at 0.5 h elicited an inhibition of the _apoH₂O₂ biosynthetic flux (i.e. to 20.3±1.3 pmol mm^–2 min^–1) identical to that when the inhibitors were applied alone (Fig. 3A). At a concentration of 1 μM, DPI was not expected to inhibit Prxs (Kärhönen and Kuchitsu, 2015) or PAOs (Gémes et al., 2016). This points to the possibility that VfrbohB and VfPAOs interact to fuel Prx-dependent _apoH₂O₂ biosynthesis. In contrast, CAO and Prxs functioned independently at 12 h (Fig. 3A).

A SHAM–DPI–MDL 72527–BEA-insensitive _apoH₂O₂ biosynthetic flux was present throughout cotyledon culture, peaking to a plateau between 6 h and 9 h before declining to pre-culture levels at 12 h (Fig. 3A). The absence of any additional flux inhibition in cotyledons exposed to a combination of diethyldithiocarbamate and SHAM indicated that the SHAM-insensitive flux was not occurring through a Cu/ZnSOD pathway (Supplementary Fig. S5). This excludes participation of quinone reductase (Table 1; Biniek et al., 2017).

We tested whether PAO and CAO activities were substrate limited (Kärhönen and Kuchitsu, 2015) by determining the response of the _apoH₂O₂ biosynthetic flux to culturing cotyledons on media containing spermidine or putrescine, substrates for PAO and CAO, respectively (Kärhönen and Kuchitsu, 2015). At concentrations in excess of 1 μM, spermidine proved to be phytotoxic, while the flux was unresponsive to lower concentrations (Supplementary Fig. S6; Fig. 3B). Similar responses to exogenously applied polyamines have been reported (e.g. Pandolfi et al., 2010). In contrast, putrescine substantially enhanced a BEA-sensitive flux between 6 h and 12 h (Fig. 3B). These responses were consistent with CAO activity across these culture times being regulated by putrescine secretion to the ETC apoplasm that commenced post-9 h. Thus, initiation of the second _apoH₂O₂ burst was mediated by Prx activity alone, and its contribution remained steady thereafter. However, by 12 h, a CAO-dependent H₂O₂ synthesis replaced the SHAM-insensitive flux (Fig. 3A).

Pharmacological determination of enzyme groups contributing to the _apoH₂O₂ catabolic flux

Contributions to the _apoH₂O₂ catabolic flux (Fig. 1C) by potential players identified in the ETC transcriptome (Table 1; Supplementary Table S1) were evaluated by culturing cotyledons on media containing their specific inhibitors. These were p-flurophenol for ascorbate oxidase (Santagostini et al., 2004), 5,5′-dithiobis-(2-nitrobenzoic acid) for ascorbate peroxidase (Anjum et al., 2016), mercaptosuccinate for glutathione peroxidase (Sajedi et al., 2012), and 3-aminotriazole for CAT (Bi et al., 2017). Of these inhibitors, only 3-aminotriazole impacted the _apoH₂O₂ catabolic flux by 75–82% across all phases of WL formation (Fig. 4; Supplementary Table S2), pointing to VfCAT1 being the predominant enzyme catalysing catabolism of _apoH₂O₂. The absence of any response of the _apoH₂O₂ catabolic flux to a H₂O₂ supplement points to CAT activity being substrate saturated (Supplementary Fig. S7).

Catalase regulates polarity of the _apoH₂O₂ signal

Blockading H₂O₂ catabolism had no influence on estimates of the _apoH₂O₂ biosynthetic flux (Fig. 1C). This hinted that the two processes could be spatially separated between the various wall orientations of trans-differentiating ETCs. For instance, localization of biosynthesis to the outer periclinal wall with catabolism restricted to the anticlinal and inner periclinal walls could account for the observed temporally stable distribution of _apoH₂O₂ to the outer periclinal wall (Fig. 1D–F). This hypothesis was tested by determining the effect of inhibiting VfCAT1 activity on the _apoH₂O₂ distribution between ETC wall orientations detected using the H₂O₂ histochemical stain, 3',3-diaminobenzidine. In contrast to the controls, the _apoH₂O₂ distribution between wall orientations in ETCs of cotyledons cultured on 3-aminotriazole was depolarized (Fig. 5B, D, F, G). This response is consistent with VfCAT1 activity being restricted to anticlinal and inner periclinal ETC walls across WL formation. A spatial separation of VfCAT1 from the H₂O₂ biosynthetic enzymes is also consistent with the sharp demarcation between the presence/absence of _apoH₂O₂ at the outer periclinal and anticlinal wall junctions (Fig. 5A, C, E).

_apoH₂O₂ biosynthetic enzymes are localized to sub-compartments of the outer periclinal wall

The temporal profile of biosynthetic enzymes generating the _apoH₂O₂ signal varied (Fig. 3A). Whether these enzymes were localized to sub-compartments in the outer periclinal cell wall or the PM was examined by determining the impacts of pharmacologically blocking selected H₂O₂ biosynthetic enzymes on the distribution of _apoH₂O₂ biosynthesis sites detected as electron-dense precipitates of cerium perhydroxide. This analysis was carried out at the two peaks of, and trough between, the _apoH₂O₂ biosynthesis flux (Fig. 1C).

For the first peak in _apoH₂O₂ biosynthesis, cerium perhydroxide precipitates in control ETCs indicated two sites of _apoH₂O₂ production, namely the PM and the original wall where _apoH₂O₂ exhibited a distinct increased accumulation towards its outer surface (Fig. 6A). Estimates of total cerium perhydroxide levels indicated that a 2.7-fold higher _apoH₂O₂ production occurred at the PM compared with the original wall (Fig. 6B). Pharmacological blockade by SHAM indicated that Prxs were responsible for _apoH₂O₂ biosynthesis at both sites, while rboh (DPI) and PAO (MDL 72527) selectively regulated the original wall Prx activity (Fig. 6B; see also text linked with Fig. 3A). Thereafter, _apoH₂O₂ biosynthesis was confined to the ETC PM during formation of the UWL (Fig. 6C, D) and WI papillae (Fig. 6E, F). For the former, _apoH₂O₂ was evenly distributed along the PM, with Prxs accounting for 69% of _apoH₂O₂ production (Fig. 6C, D; see also Fig. 3A). During WI papillae deposition, two domains of PM-localized _apoH₂O₂ biosynthesis developed; ensheathing and between WI papillae, with the former exhibiting a 2-fold higher activity (Fig. 6E, F). Responses to pharmacological blockades indicated that Prxs and CAO (BEA) were present in both domains, with their respective activities contributing 66% and 31% to the _apoH₂O₂ pool sizes in either domain (Fig. 6F). In the case of the PM ensheathing WI papillae, _apoH₂O₂ biosynthesis, and hence Prx/CAO activity, was constrained to tips of WI papillae best visualized in WIs reaching their full size (Supplementary Fig. S8).

Peroxidases generate the _apoH₂O₂ signal regulating assembly of the uniform wall layer and wall ingrowth papillae

To identify the _apoH₂O₂ biosynthetic enzymes regulating WL formation, cotyledons were cultured for 15 h on media containing a pharmacological inhibitor of each enzyme. The impact on the percentages of cells forming WI papillae indicated that WL assembly was sensitive to Prx-derived _apoH₂O₂ alone (Fig. 7A). Under these conditions, a small population of epidermal cells (20%) escaped to deposit a partially developed UWL (volume reduced by 77%; see Fig. 7C and D compared with E). This developmentally compromised UWL supported a highly depleted coverage of equally developmentally compromised WI papillae (Fig. 7B, D, F versus E, G respectively; H). In contrast, development of the original outer periclinal wall was unaffected (Fig. 7C) despite enlarging during the initial _apoH₂O₂ burst (Fig. 1C; Xia et al., 2017). Exposing cotyledons to SHAM once UWL assembly was completed by 9 h of cotyledon culture (Xia et al., 2017) blocked WI papillae construction, suggesting a direct action of _apoH₂O₂ on WI papillae assembly (Fig. 7B). Therefore, Prx-generated _apoH₂O₂ was essential for forming both the UWL and WI papillae while having no effect on original wall construction.

Fig. 7.

Effect of pharmacological blockade of specified H₂O₂ biosynthetic enzymes on wall labyrinth formation in adaxial epidermal transfer cells (ETCs) of cultured V. faba cotyledons. (A) Cotyledons were cultured for 15 h on MS medium ±5 mM salicylhydroxamic acid (SHAM) for cell wall peroxidases; 1 μM diphenyleneiodonium chloride (DPI) for respiratory burst oxidase homologues; 100 μM 2-bromoethylamine (BEA) for copper-containing diamine oxidases; and 25 μM N¹,N⁴-bis(2,3-butadienyl)-1,4-butanediamine (MDL 72527) for polyamine oxidases. Peels of ETCs were processed for SEM analysis and scored for the percentage of cells with WI papillae. (B–H) The inhibitory effect of SHAM on components of the WL (UWL and WI papillae) was assessed by culturing cotyledons for either 15 h on MS medium ±SHAM or 9 h on MS medium alone prior to transfer to MS medium ±SHAM for another 6 h. Thereafter, cotyledons were processed for SEM or TEM. (B) Percentage of cells forming a UWL and WI papillae in the presence of SHAM. (C) Inhibitory effect of SHAM on UWL volume. (D–G) TEM (D, E) and SEM (F, G) images of the outer periclinal wall of ETCs after a 15 h culture in the absence (D, F) or presence (E, G) of SHAM, illustrating the developmentally compromised UWL and WI papillae size (E versus D) and coverage (G versus F). (H) WI papillae coverage in the presence of SHAM. Data are means ±SEs of a minimum of six replicate cotyledons with 150 cells examined per replicate. Mean values with the same letter are not significantly different (P>0.05); statistical differences for treatment means against control (**P<0.01). Scale bar=500 nm for (D) and (E), 10 μm for (F) and (G).

Discussion

Our previous studies presented findings consistent with a polarized _apoROS signal initiating biosynthesis of cell wall materials and regulating their polarized deposition to form the UWL and possibly WI papillae in ETCs of cultured V. faba cotyledons (Andriunas et al., 2012; Xia et al., 2012). This conclusion was confirmed and further advanced by demonstrating that of _apoROS, O₂^–, H₂O₂, and ·OH, _apoH₂O₂ was the sole ROS regulating polarized deposition of the UWL and WI papillae (Fig. 1A, B). In pollen tubes and root hairs, _apoH₂O₂ performs a similar function in being part of the regulatory complex directing their polarized growth (Mangano et al., 2016). In these two polarized growth systems, _apoROS signal generation is under transcriptional control (e.g. Mangano et al., 2017). In contrast, for trans-differentiating ETCs, the two successive bursts (minutes; hours) in the _apoH₂O₂ biosynthetic flux, along with the accompanying catabolic flux, were under post-translational control while the intervening biosynthetic phase was co-regulated at the transcriptional and post-translational levels (Figs 1C, 2A). Broadly this pattern of control is comparable with oxidative bursts associated with wounding, abiotic stress, and pathogen attack (Baxter et al., 2014). Nevertheless, a point of difference is that the ethylene-induced polarized _apoH₂O₂ signal (Andriunas et al., 2012) formed in most epidermal cells within 30 min (Fig. 1C, D). Thus, _apoH₂O₂ signal formation was temporally decoupled from successive (hours) recruitment of ETCs exhibiting a Ca²⁺ signal (Zhang et al., 2015b). This suggests that ROS and Ca²⁺ signalling may not be integrated into cell–cell waves across the epidermal cell population to induce their trans-differentiation to an ETC phenotype as described for systemically acquired acclimation triggered by abiotic stresses (Choi et al., 2017).

Within the framework outlined above, the questions described herein focused on identifying the _apoH₂O₂ biosynthetic and catabolic enzymes generating the _apoH₂O₂ signal, their potential localization in the ETC WL or PM, and how their coordinated activities shaped the polarized signal to drive WL formation.

Biosynthetic and catabolic _apoROS enzymes form the temporally dependent _apoH₂O₂ signal

The ETCs contained transcripts encoding a large array of _apoH₂O₂ biosynthetic (21) and catabolic (6) enzymes with predicted apoplasmic or PM locations (Table 1). The enriched presence of _apoROS enzymes in ETCs of developing cotyledons is a feature of reproductive organs in which a repository of developmental and defence ROS signalling options can be rapidly and selectively deployed (Cosio and Dunand, 2010; Francoz et al., 2015). VfrbohB, VfPrx15A, VfPrx18, VfPrx52A, VfPrx52B, VfPrx71, and VfCAOδ, were ETC-specific DEGs (Supplementary Table S1). Of these, only the Arabidopsis homologue, AtPrx18, has been shown to have a direct role in reproductive development (Kumar et al., 2012), while AtCAOδ has been implicated in fruit ripening (Tavladoraki et al., 2016). Moreover, their known functions in seed after-ripening (AtrbohB; Müller et al., 2009) and lignin polymerization (AtPrx52 and AtPrx71; Shigeto and Tsutsumi, 2016) do not align closely with the polysaccharide composition and synthesis of WLs (Vaughn et al., 2007; Xia et al., 2017). This circumstance is not unsurprising. First, the percentage amino acid identities for most of the expressed VfPrx genes with their Arabidopsis homologues fall short of the >90% considered necessary for being truly orthologous (Table 1; Welinder and Larssen, 2004). Secondly, most control was exercised at the post-translational level, rendering it impossible to discern which transcripts were translated into catalytically active proteins (Fig. 2A).

In this context, selective pharmacological blockade of encoded _apoROS enzymes suggested that biosynthetic activities of VfrbohB, Prxs, Cu/ZnSODs, PAOs, and CAOs, in combination with catabolism by VfCAT1, coordinately generated the polarized ETC-specific _apoH₂O₂ signal (Fig. 1, 3–5; Supplementary Fig. S5; Andriunas et al., 2012). While CATs are generally localized to peroxisomes, AtCAT1 can be located in cell walls (Podgórska et al., 2017) and, relevant to this study, is primarily expressed in reproductive tissues (Mhamdi et al., 2018). Similar to their role in generating the oxidative burst in pathogen defence (e.g. O’Brien et al., 2012), Prx activity dominated _apoH₂O₂ biosynthesis throughout ETC development (Table 1; Fig. 3; Supplementary Fig. S3) and accounted for the partial transcriptional control operative between the two successive oxidative bursts (Fig. 2A; Supplementary Fig. S4). To achieve this outcome, Prxs functioned in their oxidative cycle producing O₂^– that was subsequently reduced to _apoH₂O₂ catalysed by one or more of the Cu/ZnSODs with a predicted apoplasmic localization (Table 1; Supplementary Fig. S5; Minibayeva et al., 2015; Podgórska et al., 2017). Since Cu/ZnSOD activity was in excess for _apoH₂O₂ biosynthesis (Supplementary Fig. S5), the Prxs probably acted as the primary regulator of the _apoH₂O₂ biosynthetic flux.

Two distinctive types of Prx activity were present at 0 and 0.5 h of cotyledon culture, namely Prxs that acted redundantly (66%) or cooperatively (30%) with CAOs (0 h) or VfrbohB plus PAOs (0.5 h) to generate the _apoH₂O₂ signal (Fig. 3A). Rbohs produce apoplasmic O₂^– using NADPH as an electron donor while CAOs and PAOs generate _apoH₂O₂ by oxidizing mainly putrescine and spermidine, respectively (Podgórska et al., 2017). Possible Prx/CAO or Prx/PAO cooperative networks could function as feed-forward mechanisms whereby the amine oxidases channel their H₂O₂ product to Prxs for oxidation to O₂^– for subsequent Cu/ZnSOD reduction to H₂O₂ (Roach et al., 2015). The role of VfrbohB in _apoH₂O₂ biosynthesis is more difficult to deduce from our findings. Inhibition of Prxs by 1 μM DPI is unlikely as the SHAM-sensitive _apoH₂O₂ biosynthesis was DPI insensitive at 0 h (Fig. 3A). The responsiveness of flavin-containing PAOs and CAOs at this DPI concentration is unknown but unexpected (Gémes et al., 2016). There is a growing body of evidence showing that _apoROS biosynthetic enzymes can function cooperatively to regulate _apoH₂O₂ production as shown for coupling of rboh/Prx (Lee et al., 2013), CAO/Prx (Roach et al., 2015), and rboh/PAO (Gémes et al., 2016). Whether cooperativity could extend to a rboh/PAO/Prx/Cu/ZnSOD assembly is uncertain and particularly so to account for Vfrboh and PAOs eliciting the same response alone or in combination in regulating SHAM-sensitive Prx activity (Fig. 3A). These responses are not consistent with the feed-forward rboh/PAO loop sustaining _apoROS homeostasis of roots in response to salt stress (Gémes et al., 2016).

During transitioning to the second burst in _apoH₂O₂ production between 3 h and 9 h of cotyledon culture, Prx activity alone catalysed _apoH₂O₂ generation (Fig. 3A). By 12 h, however, both Prx (66%) and CAO (31%) contributed to the _apoH₂O₂ biosynthetic flux (Fig. 3A). Plant CAOs preferentially oxidize putrescine and cadaverine, producing 4-aminobutanal to form H₂O₂ and NH₃ (Tavladoraki et al., 2016). Potential contributors to the CAO-dependent _apoH₂O₂ flux were the extracellularly predicted, but functionally uncharacterized, Arabidopsis homologues, VfCAOα2 and VfCAOδ, with the latter being expressed as an ETC-specific DEG (Table 1; Supplementary Table S1; Tavladoraki et al., 2016). The post-9 h CAO-catalysed _apoH₂O₂ production arose from a post-translationally regulated increase in CAO activity (Fig. 2A). In part this was accounted for by easing of a substrate supply limitation imposed by release rates of putrescine to the ETC apoplasm (Fig. 3B; Cona et al., 2006).

Spatiotemporal distribution of _apoROS enzymes in relation to _apoH₂O₂ and wall labyrinth formation

While the magnitude of the _apoH₂O₂ biosynthetic flux underwent substantial temporal changes, distribution of the _apoH₂O₂ signal to the outer periclinal wall of ETCs was temporally invariant (Figs 1D–F, 5A, C, E). In the absence of ROS catabolic enzyme activities, the _apoH₂O₂ is chemically stable (Supplementary Fig. S1) and, when supplied to cultured cotyledons in excess, readily permeated throughout all ETC wall orientations (Andriunas et al., 2012). These observations suggested that homeostasis of the polarized _apoH₂O₂ signal must depend upon catabolic removal of _apoH₂O₂ diffusing into ETC anticlinal walls by VfCAT1 (Figs 4, 5). This proposition was verified by pharmacological blockade of VfCAT1 activity resulting in a uniform redistribution of the _apoH₂O₂ signal throughout all ETC wall orientations (Fig. 5B, D, F, G). Furthermore, if _apoH₂O₂ catabolism was localized to the ETC anticlinal wall, then _apoH₂O₂ catabolic rates per ETC were estimated to exceed those of biosynthesis by a minimum of 17-fold throughout WL formation, ensuring complete consumption of _apoH₂O₂ entering the anticlinal wall. To our knowledge, this is the first demonstration of an _apoROS catabolic enzyme sculpturing a polarized _apoH₂O₂ signal in plant cells.

At key phases of _apoH₂O₂ generation, selective pharmacological blockade of _apoROS enzymes on their _apoH₂O₂ signatures, detected as cerium perhydroxide precipitates at the TEM level, was employed to deduce the sub-compartment(s) they occupy within the outer periclinal PM and cell wall domains (Fig. 6). These studies revealed that the putative VfrbohB/Cu/ZnSOD/PAO/Prx/Cu/ZnSOD assemblage, responsible for generating the smaller component of the first oxidative burst (Fig. 3A), generated an _apoH₂O₂ signature in the ETC outer periclinal wall that was most concentrated proximal to its exterior surface (Fig. 6A, B). This _apoH₂O₂ sub-compartmentation is consistent with ROS channelling (e.g. Lee et al., 2013) through this enzyme assemblage, culminating in Cu/ZnSOD-catalysed dismutation of O₂^– to _apoH₂O₂. In contrast, the accompanying ‘redundant’ Prx activity, generating a stronger _apoH₂O₂ signal at the PM (Fig. 6A, B), was presumably catalysed by one or more of the five predicted PM-localized Prxs functioning in series with another cell wall Cu/ZnSOD isoform (Table 1; Supplementary Fig. S5). Significantly, the ratio of PM- (46%) and cell wall- (54%) localized Prxs expressed in ETCs (Table 1) closely reflects the corresponding ratio reported for maize (Lüthje et al., 2011). Similar sub-compartmentalization of the _apoH₂O₂ signature between the PM and cell wall has been observed for defence responses (e.g. Lherminier et al., 2009). However, in this case, PM and cell wall _apoH₂O₂ signatures were separated temporally between the first and second oxidative bursts and generated by rboh and PAO, respectively.

Following the first oxidative burst reaching a peak at 0.5 h, the redundant Prxs contributed to the PM-localized _apoH₂O₂ signal throughout the remainder of ETC development (Fig. 3A, B). Their localization shifted from an even distribution across the entire outer periclinal PM region to one in which higher _apoH₂O₂ concentrations accumulated in PM regions ensheathing developing WI papillae coincident with the second oxidative burst (Figs 3A, 6C, D versus E, F). The _apoH₂O₂ signatures located between and ensheathing WI papillae were generated by independent Prx and CAO activities (Figs. 3, 6E, F). As WI papillae developed, the elevated _apoH₂O₂ signal became aggregated around their tips (Fig. 6E; Supplementary Fig. S8). While the mechanisms(s) responsible for this sub-compartmentation await resolution, comparable compartmentalization of Prx64 and rboh participate in Casparian Strip formation. This was shown to be guided by an extensive transmembrane polymeric platform formed by Casparian Strip domain proteins (CASPS) (Lee et al., 2013). In contrast, for trans-differentiating ETCs, _apoH₂O₂ would appear to direct spatial organization of lipid-enriched domains in the outer periclinal region of the PM (Zhang et al., 2019).

Embedded within this complex of _apoROS enzymes is a capacity to generate an _apoH₂O₂ signal that regulates formation of the UWL and WI papillae (Fig. 1; Andriunas et al., 2012). Selective pharmacological blockade of the VfrbohB/PAO/Prx assemblage, CAO, and PM-localized Prxs indicated that one or more of PM-localized Prxs were exclusively responsible for eliciting the _apoH₂O₂ signal directing construction of both UWL and WI papillae (Table1; Figs 6, 7). In the case of UWL formation (Fig. 7), our previous studies suggested that the PM-located _apoH₂O₂ signal acted by inducing a microtubule/actin-independent polarized vesicle docking system, embedded in sterol-enriched domains, located along the entire outer periclinal region of the ETC PM (Zhang et al., 2019). In contrast, the concurrent VfrbohB/PAO/Prx original wall-generated _apoH₂O₂ signal, residing in the cell wall matrix, played no role in UWL construction. Rather it may have caused cell wall glycoproteins to cross-link (e.g. Novakovic et al., 2018), accounting for the rapid cessation of ETC expansion by 3 h of cotyledon culture (Xia et al., 2017). The cell wall-localized Prx71 is a candidate to perform this function (Table 1; Raggi et al., 2015).

In contrast to CAO, the activity of PM-localized Prxs generated an _apoH₂O₂ signal, ensheathing WI papillae, of sufficient magnitude to drive their development (Supplementary Fig. S8; Figs 6E, F, 7B). The array of potential PM-localized _apoH₂O₂ targets accounting for assembly of WI papillae included the directed incorporation of sterols that form lipid-enriched microenvironments in which clusters of Ca²⁺-permeable channels are assembled (Zhang et al., 2019). In combination with ethylene, the _apoH₂O₂ signal activated Ca²⁺-permeable channels to form plumes of elevated [Ca²⁺]_cyt (Zhang et al., 2015a, b). These plumes in turn mediated remodelling of the actin network to direct a polarized delivery of vesicles carrying wall building materials to sterol-enriched docking loci at which WI papillae were constructed (Zhang et al., 2017a, b; 2019).

Amongst key questions remaining to be resolved are identifying mechanisms responsible for constraining _apoROS biosynthetic enzyme activity to the outer periclinal ETC PM domain and VfCAT1 to the anticlinal wall, and unravelling how Prxs and CAOs are redistributed within their PM domain during transition from UWL to WI papillae formation.

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. Procedures used to determine rates of non-catalytic and catalytic decay of _apoH₂O_2.

Fig. S2. Photoshop enhancement of 3',3-diaminobenzidine precipitates in fresh transverse sections.

Fig. S3. Evaluating potential contribution of lipoxygenases to _apoH₂O₂ biosynthesis.

Fig. S4. Levels of biological control regulating Prx activity.

Fig. S5. Relative roles of Cu/ZnSOD and Prx in _apoH₂O₂ biosynthesis.

Fig. S6. Evaluating substrate limitation of PAO.

Fig. S7. Evaluating substrate limitation of _apoH₂O₂ catabolism.

Fig. S8. Domains of PM-localized _apoH₂O₂ during WI papillae formation.

Table S1. Spatial–temporal expression profiles of transcripts encoding _apoH₂O₂-related enzymes.

Table S2. Evaluating contributions of ascorbate oxidase and glutathione oxidase to _apoH₂O₂ catabolism.

Glossary

Abbreviations:

apo

apoplasmic

BEA

2-bromoethylamine hydrobromide

[Ca²⁺]_cyt

cytoplasmic Ca²⁺ concentration

CAT

catalase

CAO

copper amine oxidase

DPI

diphenyleneiodonium chloride

ETC

epidermal transfer cell

MDL 72527

N ¹,N⁴-bis(2,3-butadienyl)-1,4-butanediamine

PAO

polyamine oxidase

Prx

class III peroxidase

rboh

respiratory burst oxidase homologue

ROS

reactive oxygen species

SHAM

salicylhydroxamic acid

Cu/ZnSOD

Cu/Zn superoxide dismutase

UWL

uniform wall layer

WI

wall ingrowth

WL

wall labyrinth