Accelerated remodeling of the mesophyll-bundle sheath interface in the maize C4 cycle mutant leaves

C4 photosynthesis in the maize leaf involves the exchange of organic acids between mesophyll (M) and the bundle sheath (BS) cells. The transport is mediated by plasmodesmata embedded in the suberized cell wall. We examined the maize Kranz anatomy with a focus on the plasmodesmata and cell wall suberization with microscopy methods. In the young leaf zone where M and BS cells had indistinguishable proplastids, plasmodesmata were simple and no suberin was detected. In leaf zones where dimorphic chloroplasts were evident, the plasmodesma acquired sphincter and cytoplasmic sleeves, and suberin was discerned. These modifications were accompanied by a drop in symplastic dye mobility at the M-BS boundary. We compared the kinetics of chloroplast differentiation and the modifications in M-BS connectivity in ppdk and dct2 mutants where C4 cycle is affected. The rate of chloroplast diversification did not alter, but plasmodesma remodeling, symplastic transport inhibition, and cell wall suberization were observed from younger leaf zone in the mutants than in wild type. Our results indicate that inactivation of the C4 genes accelerated the changes in the M-BS interface, and the reduced permeability suggests that symplastic transport between M and BS could be regulated for normal operation of C4 cycle.


Results
Development of dimorphic chloroplasts in the maize leaf. We examined chloroplast structures in the maize leaf with TEM and electron tomography (ET) to characterize the ultrastructural dynamics involved in the C4 differentiation. The four locations along the developmental gradient of third leaves of 9-day-old maize seedlings were selected as defined in Li et al. (2010) (Fig. 1A). The location of leaf 2 ligule in leaf 3 was set as the reference point (0 cm), and maize leaves were cut perpendicular to the leaf axis at -4 cm, 0 cm, 4 cm, and 10 cm (Fig. 1B). Thin cross sections (0.2-0.3 mm) were isolated and preserved by high-pressure freezing and freezesubstitution for ET analyses. For consistency in comparing the four stages among genotypes, our examination was restricted to minor veins because they outnumber midveins and intermediate veins by 7 to tenfold 28 . We examined intermediate veins in the −4-cm samples, because minor veins do not extend to the young leaf tissue, and the cell organization of intermediate veins is more similar to that of minor veins than that of midveins. Intermediate and minor veins correspond to rank 1 and rank 2 intermediate veins, respectively, according to the classification by Sedelnikova et al. 5 .
We were able to discern the M and BS cells in the −4-cm cross section but the concentric arrangement indicative of Kranz anatomy was not apparent (Fig. 1C). In 0-cm cross sections, however, the vascular bundle, BS cells, and M cells outside the BS layer were visible (Fig. 1D). Proplastids in M and BS cells at −4 cm were indistinguishable: They were round with average diameters of 1.5-1.6 µm, and starch particles occupied most of their stroma (Fig. 1G,O) as reported in Majeran et al. 14 Chloroplasts of the two cell types in the 0-cm zone were larger and had more thylakoids than those of the −4-cm zone. M chloroplasts had grana stacks but few starch particles, whereas BS chloroplasts had starch particles, but grana stacks were rare, displaying signs of differentiation (Fig. 1H,K,L). The differing thylakoid morphologies of the M and BS cells in the 0-cm sections indicate that the two cell types express different sets of chloroplast proteins.
The chloroplast dimorphism of Kranz anatomy was clearly observed in 4-cm zone chloroplasts, (Fig. 1G-J). Grana stacks were abundant in M chloroplasts, but thylakoids were mostly unstacked in BS chloroplasts (Fig. 1I,M,N). Thylakoid architectures in the two cell types in the 4-cm sections persisted in 10-cm sections (Fig. 1J). When starch particles were stained with iodine, darkly stained particles accumulated in BS cells of leaf sections from 4-cm and 10-cm positions (Fig. 1O-R). The starch accumulation correlated with the extent of chloroplast development (Fig. 1K-N). Chloroplasts were ovoid in both cell types, but BS chloroplasts were more elongated than M chloroplasts (Fig. 1S).
To delineate thylakoid assembly more accurately, we carried out ET analysis. At −4 cm, plastids had simple thylakoid networks, and grana stacks consisted of two or three layers in both cell types ( Fig. 2A,B). In M chloroplasts in 0-cm sections, each granum had approximately five layers on the average, and grana stacks densely populated the stroma (Fig. 2C). Thylakoids in BS chloroplasts of the same stage had extended stroma lamellae and grana stacks consisting of 2-4 disks (Fig. 2D). M chloroplasts had thylakoids clearly distinct from those of BS chloroplasts at 4 cm (Fig. 2E,F). Grana stacks in M chloroplasts often had more than 10 disks interconnected by tubules surrounding the stacks. By contrast, thylakoids in BS chloroplasts were composed mostly of unstacked lamellae. Grana stacks consisted of only two or three layers in BS chloroplasts at 4 cm and they were smaller than those of younger BS cells (Fig. 2G,H). These results indicate that stroma lamellae expanded, and grana stacks shrank in BS chloroplasts during leaf maturation.  www.nature.com/scientificreports/

Localization of chloroplast proteins in M and BS cells by immunofluorescence. Our ET results
indicated that the chloroplast structures in M and BS deviated from each other in the 0-cm location and were fully differentiated by 4 cm (Figs. 1 and 2). We performed immunofluorescence localization of subunits in the thylakoid membrane protein complexes and of Rubisco to compare the structural dimorphism with macromolecular compositions of the chloroplasts. We utilized antibodies against PsbO (a subunit of PSII), Lhca (a subunit of the light harvesting complex of PSI), and the large subunit of Rubisco. None of these proteins were detected in the −4-cm samples (Fig. 3A,D,G). In 0-cm samples, we were able to image fluorescence from both M and BS chloroplasts, but significant differences between the two cell types were not detected (Fig. 3B,E,H). Differential enrichment of PsbO and the Rubisco large subunit in either M and BS chloroplasts was clearly seen in 4-cm samples (Fig. 3B,H). PSII and its light harvesting complexes contribute to the formation of grana stacks. Consistent with the proliferation and removal of grana stacks in M and BS chloroplasts, respectively, PsbO accumulated in M chloroplasts in the 4-cm regions (Fig. 3B,C). Lhca-specific fluorescence, indicative of PSI, was detected in both M and BS chloroplasts throughout the leaf developmental gradient (Fig. 3E,F). The Rubisco large subunit was detected primarily in BS chloroplasts in the 4-cm section (Fig. 3H,I). The cell-type specific accumulation of Rubsico suggests that the Calvin cycle operates primarily in BS chloroplasts in the 4-cm leaf cells.
Chlorophyll fluorescence from PSI is enriched in the spectral window above 700 nm, whereas that from PSII dominates the window below 700 nm; this feature has been utilized to visualize dimorphic chloroplasts in C4 plants 40,41 . We therefore tested whether PSII fluorescence is weaker from BS in the 4-cm leaf samples. PSII intensities were similar in M and BS cells in the −4-and 0-cm sections (Supplemental Fig. S1A,B). By contrast, PSII-specific fluorescence was lower in BS cells than M cells in 4-and 10-cm sections (Supplemental Fig. S1C,D). There were no differences in the fluorescence detection range of 720-800 nm (i.e., PSI) in chloroplasts from BS cells compared to those from M cells at any stage. Thus, the PSI-and PSII-specific chlorophyll fluorescence imaging agrees with the immunofluorescence localization data.
PD at the M-BS boundary acquire sphincters and cytoplasmic sleeves. It was previously shown that PD connecting an M and BS cell pair has a collar around its desmotubule, termed a sphincter; the sphincter is located on the M side 23    www.nature.com/scientificreports/ and Supplemental Fig. S2). PD in the 0-cm cell wall were longer than those in the −4-cm cell wall. But they remained simple PD lacking cytoplasmic sleeves or electron-dense collars around the desmotubule. Sphincters were observed in PD of 4-cm and 10-cm leaf samples (Fig. 4A). When visualized with ET, the doughnut-shaped sphincters appear to surround the desmotubule and press the plasma membrane outward, enlarging the neck region. Cytoplasmic sleeves were observed in the BS half of PD in 4-cm sections as were sphincters in the M side ( Fig. 4B,C). These structural changes in PD correlated with chloroplast differentiation between 0-cm and 4-cm.
We then examined PD that join the same cell types to determine whether they undergo changes similar to those observed for PD that join M and BS cells. The sphincters were discerned in PD that connect two M cells in 4-and 10-cm sections. However, sphincters were not observed in PD connecting two BS cells (Supplemental Fig. S2). The PD connecting two M cell walls had sphincter collars on both sides as previously observed 28 , indicating that the attachment is derived from the M cell. The thickness of the BS cell wall increased from −4 to 4 cm, but the M cell wall stopped thickening at 0 cm (Supplemental Fig. S2).
PD maturation at the M-BS boundary occurs in younger leaf zones of C4 mutant leaves. Because biogenesis of dimorphic chloroplasts is essential for C4 photosynthesis, the PD remodeling might be related to the C4 cycle. To test this notion, we compared PD in ppdk-1, a null allele of C4 PPDK in maize and dct2 (dct2-1 in Weissman et al. 2016 7 ) of which malate transport into BS chloroplasts is inhibited. PD developed more rapidly in the maize mutant lines than in wild-type plants (Fig. 4D-I). PD in the ppdk-1 and the dct2 mutant leaves elongated as the cell wall thickened from −4 to 0 cm. ET imaging revealed that sphincter assembly had begun before 0 cm in the mutant leaves (Fig. 4D,I). The PD in the 0-cm mutant leaves had acquired cytoplasmic sleeves. These morphological changes were not detected in wild-type PDs until 4 cm (Fig. 4A-C). Sphincter sizes were larger in ppdk-1 PD than wild-type PD at the same leaf positions (Fig. 4G) probably as sphincters developed earlier in the mutant PD. However, widths of cytoplasmic sleeves did not differ between mutant and wild-type plants (Fig. 4H). The accelerated PD changes were observed in the dct2 mutant ( Fig. 4I) and two other ppdk mutant alleles, ppdk-2 and ppdk-3 (Supplemental Fig. S3).
Maize plants homozygous for ppdk-1 and dct2 mutations are seedling lethal, but their development from germination through 9 days of growth (i.e., the time of sampling) was relatively normal (Supplemental Fig. S4). The Kranz anatomy was not affected by the deficiency in C4 pathways, although their leaves were pale green. Chloroplast proteins were differentially enriched in M and BS chloroplasts between the 0-cm and 4-cm stages in the mutant leaves as in the wild-type leaf (Supplemental Fig. S5). Transcriptomic datasets from ppdk-1 and dct2 mutant leaves are available and we performed gene expression correlation test using the RNA-seq datasets 7,8 . Correlation coefficients of genes were calculated from their expression profiles at the four leaf positions in wild type and the mutant lines. The mutations did not cause large scale transcriptomic alterations in mutant leaves (Supplemental Fig. S4).

Reduced symplastic dye movement at the M-BS boundary correlated with PD maturation.
To investigate consequences of the PD remodeling on transport capacity, we carried out a carboxyfluorescein-diacetate (CFDA) movement assay. After leaves were fed with CFDA solution from their base, cross sections (1-2 mm thick) at 0, 4, and 10 cm were dissected. In 0-cm sections of wild-type leaves, both M and BS cells had CFDA fluorescence (Fig. 5A,B). By contrast, the fluorescence was highly confined to BS cells in 4-and 10-cm samples (Fig. 5A,B). These results indicate that PD permeability is reduced in the more mature leaf tissues. In ppdk-1 and dct2 mutant leaves, CFDA was limited to BS cells in all three locations (Fig. 5C,D). CFDA was detected in M cells of wild-type and mutant leaves in all three stages when the dye solution also contained 2-deoxy-d-glucose (DDG), a callose synthase inhibitor ( Fig. 5E-G), indicating that callose synthesis is required for the reduction in dye transport. When we calculated the ratios of fluorescence intensities in M-BS cell pairs, the largest differences were detected in 0-cm samples from wild-type and mutant leaves because the dye failed to diffuse into M cells from BS cells at this stage ( Fig. 5B-D,H). We concluded that C4 development in the maize leaf involves a decrease in symplastic permeability between M and BS cells and that the decrease is associated with structural changes in PD.

Suberin deposition in the BS cell wall in younger leaf zones of C4 mutant leaves. The BS cell wall
accumulates suberin, and the hydrophobic stratum blocks apoplastic diffusion of photosynthetic metabolites 31 . Since this barrier could restrict solute exchange between M and BS exclusively through PD, we assayed for suberin deposition in the cell wall of the four-leaf zones. The BS cell wall in 0-cm cross sections did not exhibit fluorescence. Only xylem cells in the vascular bundle were stained in the sections (Fig. 6A-C). BS cell walls in 4-cm sections were positive for suberin staining, and the fluorescent outline visualized the BS cell layer outside the vascular bundle ( Fig. 6A-C). By contrast, suberin was clearly observed in BS cell walls of 0-cm leaf sections from ppdk-1 and dct2 plants (Fig. 6D-I). The staining became stronger at later leaf developmental stages of wildtype as well as the mutants (Fig. 6J).
A recent microscopy study revealed that suberin deposition in the Arabidopsis root endodermal cell wall involves the accumulation of vesicular tubules in the gap between the plasma membrane and the secondary cell wall 42 . The complex membranous carriers appeared in non-endodermal cells where suberization was ectopically induced. We examined TEM samples from wild-type and the C4 mutant leaves (−4, −2, 0, 2, and 4 cm sections) to determine whether similar apoplastic membrane tubules are seen in the suberizing BS cell wall (Supplemental Fig. S6). In wild-type leaves, round plasma membrane ingrowths bulging with vesicular tubules appeared in BS cells from 0 and 2 cm sections. But in 4 cm sections where BS suberization is evident, we did not find such structures. By contrast, we detected extracellular pockets of vesicular tubules in the BS cell wall from −2 cm sections www.nature.com/scientificreports/ www.nature.com/scientificreports/ of ppdk-1 and dct2 mutant leaves. These extracellular carriers disappeared in 0 cm sections in the mutants, in agreement with their early suberization (Fig. 6A-I).
We examined expression levels of genes that play roles in suberin synthesis or export in the RNA-seq datasets from ppdk-1 and dct2 7,8 . Maize genes associated with cell wall suberization were identified with the MapMan tool for functional classification (http:// gabi. rzpd. de/ projects/MapMan/). Most of the identified genes exhibited strong transcriptional activities in the −4-or 0-cm zones (Fig. 6K,L). When transcript levels in wild-type and the mutant lines were compared, suberin-related genes were more upregulated in the ppdk-1 (6 out of 8) and dct2 (6 out of 10) leaves than wild-type leaves, in agreement with the suberin staining results. To evaluate temporal expression profiles more accurately, we carried out qRT-PCR analysis of CYP68B (also known as RALPH; GRMZM2G162758) and ABCG (GRMZM2G054332) genes in six leaf regions. Transcripts from the two genes were most abundant in -2-cm specimens of ppdk-1 and dct2 leaves and in 0-cm specimens of wild type leaves (Fig. 6M,N), indicating that the suberin-related genes were more strongly upregulated in the zone (−2 to 2 cm zone) encompassing the 0 cm leaf segment where we observed suberin staining in the mutant BS cell wall. When we evaluated maize genes that the MapMan tool identified as PD components, there were no significant differences in expression profiles in the mutant compared to wild-type leaves (Supplemental Fig. S7 and Dataset 1).

Discussion
The timeline of chloroplast diversification in M and BS cells included assembly of thylakoids with distinct architectures, and cell type-specific accumulation of chloroplast proteins at 4 cm. The 0-to 4-cm sector corresponded to the region in which PD across M and BS acquire sphincters and cytoplasmic sleeves. Symplastic dye movement was impeded after the structural changes of PD in a callose-dependent manner, implying that the mature PD can be gated. The BS cell wall became suberized in parallel with the PD changes. In ppdk-1 and dct2 mutant leaves, PD remodeling and suberin deposition were finished in the 0-cm zone, implying that genes for the modifications in the M-BS interface are activated earlier if the C4 cycle is disrupted (Fig. 7).
Our TEM and ET results from the wild type agree with previous structural, transcriptomic, and proteomic studies of maize C4 leaf development. Using cell-type specific RNA-seq, Tausta et al. demonstrated that genes involved in the C4 cycle and PSII-related genes begin to be differentially expressed in M cells and BS cells from the sink to source transition zone 43 . The transition zone is close to our 0-cm position, where we discerned varying ultrastructural features in chloroplasts. The 0-cm position is the zone at which light availability drastically increases for photosynthesis because leaf 3 cells are wrapped inside the older leaves below this position. In a proteomic analysis of maize BS cells and vascular bundle samples, normalized levels of PSII subunits decreased in samples 4.5 cm from the leaf 3 ligule, suggesting that cyclic electron transport becomes dominant in the bundle sheath at this location 14 . This location overlaps with the 0-cm zone in our classification, and we found that grana stacks were scarcer in BS chloroplasts than in M chloroplasts in this section. We detected structural dimorphism of the M and BS chloroplasts in 0-cm sections. Fluorescence microscopy-based protein localization methods failed to distinguish M and BS chloroplasts in 0-cm sections, however; probably because differences in protein compositions were too small for detection by immunolabeling.
De Bellis et al. (2021) reported that extracellular membrane structures accumulate in suberizing Arabidopsis root endodermal cells 42 . The root endodermis and BS in the leaf are anatomically equivalent in that they are cell layers encircling the vascular tissue 44 . They also acquire a diffusion barrier made of suberin at maturity. SCARE-CROW is a transcription factor that controls the root endodermis differentiation. SCARECROW is expressed in BS cells and its mutation affected the BS organization, suggesting that genetic programs for the root endodermis and BS cells in the leaf are conserved 9,45,46 . We discerned round plasma membrane ingrowths in suberizing BS cells analogous to the apoplastic vesicular tubules in the Arabidopsis endodermal cells (Supplemental Fig. S6). Our data showed that the leaf BS cells and root endodermal cells export substances for the cell wall suberization via similar vehicles.
Abnormal suberin layers in the BS cell wall of mutant Setaria viridis leaves caused loss of CO2 from BS cells, decreasing their C4 photosynthetic efficiency 39 . The concentration of CO2 to BS chloroplasts is expected to be affected in ppdk-1 or dct2 leaf cells as C4 cycle is aberrant in them. The reduced CO2 partial pressure in the mutant BS cells may stimulate a regulatory mechanism that activates suberin synthesis and transport (Fig. 6). However, we did not observe a shift in the timing of cell type-specific chloroplast differentiation and protein expression in the mutant leaves (Fig. 7, Supplemental Figure S5). It is possible that the genetic program for constructing dimorphic chloroplasts may be less sensitive or slower in responding to the defects in C4 cycle.
PD structures and permeability evolve as the cells that they connect undergo differentiation 47,48 . PD transport is regulated to coordinate organ development and tissue patterning, because non-cell autonomous signaling molecules and transcription factors diffuse through PD [49][50][51] . A recent ET study showed that nascent PD in Arabidopsis root meristem cells lack cytoplasmic sleeves, but sleeves do form in PD of columella cells derived from meristem cells 52 . Contrary to the idea that molecular transport happens through the cytoplasmic sleeves, CFDA and GFP diffused among meristem cells despite the finding that PD in these cells lack gaps. Our connectivity assay data are consistent with the observation. CFDA crossed the M-BS boundary in the 0-cm leaf samples, where cytoplasmic sleeves were resolved.
Sphincter modules have been reported in PD, and they are linked to inhibition of cell-to-cell transport. In dormant shoot apical cells of birch trees, symplastic transport is blocked, and PD of the inactive cells are clogged with electron-dense collars 53 . When the shoot apical cells are activated, the collars disappear, and amounts of callose at the PD are reduced. In the maize leaf, sphincters in PD spanning the M-BS cell wall are associated with slow photosynthesis in plants grown at low temperature 54 . We demonstrated that sphincters appear earlier in the ppdk-1 mutant leaves and that inhibition of callose synthase facilitated CFDA movement. Because sphincter and callose could obstruct cytoplasmic sleeves, these PD components may regulate permeability at the M-BS www.nature.com/scientificreports/ boundary. We speculate that the expedited PD modification and BS cell wall coating in the mutant leaves could be a feedback reaction to inhibition of CO 2 concentration in their BS cells due to aberrant C4 cycle.

Methods
Plant material and growth conditions. Zea mays B73 (wild type for ppdk-1, ppdk-2, and ppdk-3), W22 (wild type for dct2) and four mutant lines (ppdk-1, dct2, ppdk-2, and ppdk-3) were grown from seeds in a growth chamber with 12 h:12 h light/dark and 31 °C light/22 °C dark cycle with light intensity of 550 µmol/m 2 /s and relative humidity of 50% as previously described 9 . The inbred and mutant lines were obtained from the maize stock center (http:// maize coop. crops ci. uiuc. edu/) and they were grown under the institutional guidelines.  20 . We examined maize leaf samples preserved by high-pressure freezing for electron tomographic characterization of three-dimensional thylakoid structures. However, we employed chemically fixed samples for determining PD structures and starch particle analysis. We had to capture TEM micrographs from more than hundreds of PD and chloroplasts for statistical comparisons of PD and chloroplasts in multiple stages from several genetic backgrounds. It was impossible to find sufficient numbers of intact PD or chloroplasts in maize leaf samples processed by HPF because of freezing damages. Maize leaves (leaf 3) were dissected and cryofixed with an HPM100 machine (Leica Microsystems). The frozen samples were freeze substituted in 2% OsO4 in anhydrous acetone (Leica EM AFS2) at −80 °C for 24 h and slowly warmed from -80 °C to −45 °C over 36 h. After being brought to room temperature, the samples were washed with anhydrous acetone three times, and embedded in Embed-812 resin (Electron Microscopy Sciences; Cat. No. 14120) and cured at 60 °C for 24 h. For chemical fixation, maize leaf segments were dissected in 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) and incubated in the fixative solution overnight at 4 °C. After thorough washing (4-5 times) with 0.1 M cacodylate buffer (pH 7.4), the samples were post-fixed in 1% OsO 4 in deionized water for 2 h at room temperature. Excess OsO 4 was removed by washing with water (4-5 times), and the samples were dehydrated with a graduated acetone series in deionized water (from 10 to 100% acetone). Finally, they were embedded in Embed-812 resin (Electron Microscopy Sciences, Cat. No. 14120) and polymerized at 60 °C for 24 h. Ultrathin sections (90-nm thick) were collected on formvar-coated copper slot grids (Electron Microscopy Sciences; Cat. No. GS2010-Cu) and the sections were post-stained with uranyl acetate and lead citrate solutions. The sections were examined with a Hitachi 7650 TEM (Hitachi-High Technologies) operated at 80 kV.
A series of semi-thick sections (250 nm) were collected on copper grids. After post-staining and gold particle coating, tilt series were collected with a 200-kV Tecnai F20 intermediate voltage electron microscope (+ 60° to −60° at an interval of 1.5° interval around two orthogonal axes). Tomogram calculation and 3D model rendering were performed with the IMOD software package as described previously 57,58 .
For morphometric analysis of thylakoids, stroma lamellae were defined as thylakoids consisting of single layer, and grana thylakoids were defined as thylakoids with membranes in contact with opposing membranes to constitute a stack. The largest number of disks in a granum was assigned as the number of lamellae. For PD quantitative analyses, outer diameters of sphincter rings were measured. Widths of cytoplasmic sleeves corresponded to the average width of the two gaps from the desmotubule to the plasma membrane (right and left sides) in each PD. In each panel, lower micrograph images correspond to higher magnifications of the boxed areas in upper micrographs. BS and xylem cell walls visualized by suberin staining are indicated with arrows (red arrows for 0-cm and yellow arrows for 4-and 10-cm images) and arrowheads, respectively. Note that BS cell walls in 0-cm ppdk-1 and dct2 samples stain for suberin staining (D,G), whereas BS cell walls in 0-cm WT sample do not (A). Scale bars: 10 µm. (J) Fluorescence intensities of BS cell walls in WT, ppdk-1, and dct2 leaf 0-cm (white bars), 4-cm (gray bars), and 10-cm (black bars) sections. Fluorescence values of BS cell walls were normalized to those of adjacent xylem walls. (K,L) Heat maps illustrating changes in transcript levels of genes involved in suberin synthesis or transport in ppdk-1 (K) and dct2 (L) leaves. All genes with FPKM values higher than 1 were included in the heat maps (Supplemental Fig. S7). Genes transcribed more actively in the mutant than WT in 0-cm are in blue letters. (M,N) qRT-PCRbased quantification of (M) CYP86B1 and (N) ABCG in six locations in the maize leaf. GAPDH was used as the reference gene. Error bars indicate SDs.  13 . For quantifying fluorescence intensities of chloroplasts, we loaded micrographs in ImageJ (ver. 1.53d), and outlines were drawn around chloroplasts to set regions of interest (ROI). Using the "Measure" command, we Chlorophyll autofluorescence imaging. Free-hand maize leaf cross-sections were mounted in water on glass slides (SuperFrost™, Thermo Scientific), and were examined using an SP8 confocal microscope to acquire the autofluorescence images of chloroplast and cell wall. After excitation with a 638 nm laser, emission wavelengths of 650-720 nm and 720-800 nm were imaged for PSII an PSI chlorophyll autofluorescence, respectively. The cell wall was visualized using its autofluorescence (excitation-405 nm, emission-420-480 nm).
Starch staining. Maize leaves (leaf 3) were dissected and cleared by incubating in 95% ethanol overnight.
The samples were stained with iodine potassium iodide (IKI) solution (6) and excess IKI solution was washed off by rinsing in deionized water. Free-hand cross-sections were examined under bright-field under a Meiji Techno MT4310L dissecting microscope (Meiji Techno Co.) equipped with a Leica MC120 HD microscope camera.
Carboxyfluorescein diacetate (CFDA) transport assay. Maize leaves (leaf 3) were cut at their bases and put them upright in a beaker containing CFDA solution (50 µg/mL; Sigma-Aldrich, Cat No. 21879-25MG-F) to feed the dye from their cut ends. After 1 h in the dark, three leaf segments (0 cm, 4 cm, and 10 cm) were collected by free-hand dissection. CFDA dye distribution in the leaf cross-sections was examined with a Leica TCS SP8 confocal laser-scanning microscope. We could not assess CFDA in BS-M pairs of −4 cm cross-sections (close to the cut end) consistently because CFDA infiltrated into the apoplastic space and overstained all cell types. For DDG treatment, leaf 3 samples were first put in a DDG solution (0.  Type-II PD refer to PD with cytoplasmic sleeves 55 . PD maturation and BS suberization are observed from the 0 cm zone of ppdk-1 and dct2 mutant leaves.