Alonsoa meridionalis O. Kuntze (Scrophulariaceae) and Asarina barclaiana Pennell (Plantaginaceae) were grown on pot soil in a controlled growth chamber (Sanyo Gallenkamp, Loughborough, Leicester UK) at 20°C under a 16 h light/8 h dark cycle and a photon flux of 500 μmol m⁻² s⁻¹ and 0.035% CO₂. Barley (Hordeum vulgare L.) was grown hydroponically at similar conditions. Potato plants (Solanum tuberosum L. cv. Désirée) were grown in a greenhouse under 12 h of supplemented artificial light of 400 μmol m⁻² s⁻¹ at 26°C and 12 h of dark at 18°C. All studies on whole leaves and on single cell samples were performed using mature fully expanded leaves (usually the third leaf from the top of the shoot). Samples of leaf tissue for immunolocalization studies were taken from the middle parts of the leaves. For TEM analyses of shoot apices, all species were sown from seeds and grown on soil in greenhouse under artificial light of 150 μmol m⁻² s⁻¹, at 16 h light/8 h dark cycle, at 21°C in the light and at 19°C in the dark. Apices of 14–30 day-old seedlings were fixed.

Leaves of A. meridionalis and A. barclaiana were detached from the plants 3 h after the beginning of the light period and placed in continuous light conditions with a photon flux of 150 μmol m⁻² s⁻¹ for 24 h, while the petioles were kept in a 2 mM CaCl₂ solution to favor the sealing of the phloem with callose (King and Zeevaart, 1974). Sugar concentrations were determined in single cell samples from leaves and in extracts of whole leaves after 24 h of light exposure.

Single cell sap was extracted from individual cells from the upper epidermis and palisade mesophyll by glass microcapillary technique (Tomos et al., 1994). Prior to use, a microcapillary was back-filled with low-viscosity water-saturated paraffin oil (Sigma). Ejection of the single cell sample under oil allowed the determination of sugars (glucose, fructose, and sucrose) by an enzymatic assay described by Koroleva et al. (1997, 1998). The measurements were highly reliable in a concentration range between 2 and 200 mM.

Soluble carbohydrates were extracted from leaves with 80% ethanol at 80°C for 1 h. The extraction was repeated two times, and the extracts were combined, vacuum-dried, dissolved in ultra-pure water (Millipore), syringe-filtrated (0.45 μm cellulose-acetate; Schleicher and Schuell, Dassel, Germany), and stored at −80°C. For carbohydrate analysis by high-performance liquid chromatography (HPLC), an anion exchange column (CarboPAC10; Dionex Corp, Sunnyvale, CA, USA) was used. The column was eluted isocratically with 80 mM NaOH (J.T. Baker, England) with a flow rate of 1 ml min⁻¹ using the LC-9A pump from Shimadzu (Kyoto, Japan). Sugars were detected by a thin layer of amperometric cell (ESA, Model 5200, Bedford, United States) with a gold electrode and a pulse amperometric detector (Coulochem II, Bedford, USA). The evaluation of chromatograms was performed using the integration program Peaknet 5.1 (Dionex, Idstein, Germany).

Shoot apices were fixed with 2.5% glutaraldehyde in 0.1 M potassium phosphate buffer pH 7.2 and post-fixed in 1% osmium tetroxide in the same buffer. During dehydration in a graded ethanol series followed by a graded acetone series, the material was stained en bloc with 1% uranyl acetate in 70% ethanol for 1 h, and then embedded in epon resin (Sigma Aldrich, MO, United States). Ultrathin sections (60–70 nm) were cut with a diamond knife (Diatome, Switzerland) using a Leica EM UC7 ultramicrotome (Leica Microsystems, Wetzlar, Germany), double stained on grids with 1% uranyl acetate and 3% lead citrate, and observed at 120 kV with a Libra 120 Plus electron microscope (Zeiss, Germany). The PD connecting cells of L1, L2, or L3 layers were counted manually in three apices for each species. For each apex, 8–25 cells per cell boundary (L1/L1, L1/L2, L2/L2, L2/L3, and L3/L3, respectively) were analyzed.

Fixation, embedding, and sectioning of mature leaves were performed as described by Stumpe et al. (2006). The anti-myosin VIII-antibody (Reichelt et al., 1999) was diluted in the ratio, 1:100 in tris-buffered saline (TBS) containing 1% (w/v) bovine serum albumin (BSA). The anti-calreticulin antibody (Baluska et al., 1999) was diluted in the ratio, 1:200 in the same buffer. In negative controls, the primary antibody was omitted. The secondary antibody, goat anti-rabbit Alexa Fluor 488 (Molecular Probes, Eugene, OR, USA), was diluted in the ratio, 1:500 in TBS with 1% BSA. Semi-thin sections (8–10 μm) were mounted using the ProLong Gold Antifade kit (Molecular Probes, OR, United States). Immunofluorescence was detected using a BX51 microscope (Olympus Deutschland GmbH, Hamburg, Germany). Images were captured with an objective UPlanApo 40×/0.85 ∞/0.11–0.23 using a ColorView II digital camera and Cell F^* image analytical software (Olympus Soft Imaging Solutions, Münster, Germany). Images shown in Figures 1A,B were obtained using AxioImager.Z1 Microscope (Carl Zeiss, Goettingen, Germany) and UPlanFI 100×/ 1.30 Oil ∞/0.17 C1 objective, an Axiocam 506 color digital camera and ZEN microscope software v. 3.3. Supplementary Videos 1–8 showing views of different cell junctions in leaves of the species under study were recorded using the same microscope.

Leaf sections (10 μm) were prepared from barley leaf pieces embedded in Steedman's wax as described by Stumpe et al. (2006) using an Automatic rotary microtome Microm HM 360 (Zeiss, Jena, Germany) with the blade lock assembly precooled to +4°C at an ambient temperature below +21°C. Single sections were placed on square coverslips of 1.5 × 1.5 cm size, coated with filtrated egg white (one section per coverslip). Sections were stretched in a drop of distilled water preheated to +40°C pipetted on the coverslips, and dried at +23°C for 2 h. Coverslips with sections were placed in Petri dishes, and dewaxing, rehydration, and blocking were performed as described by Stumpe et al. (2006). A primary anti-calreticulin antibody raised in rabbit (Baluska et al., 1999) was diluted in the ratio, 1:200 in TBS containing 1% (w/v) BSA. An anti-rabbit Alexa Fluor 488-conjugated secondary antibody produced in goat (Molecular Probes, Eugene, OR, USA) was diluted in the ratio, 1:500 in TBS with 10% BSA. In the negative controls, the primary antibody was omitted. The sections were incubated with the primary antibody for 1 h, washed three times with TBS for 15 min, incubated with the secondary antibody for 2 h, and washed three times with TBS for 10 min. After labeling, the coverslips were glued to the bottom of the inner side of Petri dishes of 35 mm diameter using melt wax as an adhesive. Prior to adhesion of sections, square hollows with a side length of ca. 1.5 cm matching the size of the coverslips had been carved into these Petri dishes with an incandescent spatula. The coverslips were put at the bottom of the Petri dishes using forceps in such a way that the sections in the Petri dish were located above the middle of the hollow. About 1 ml of glycerin was dropped into the Petri dish to cover the section. Fluorescence of Alexa Fluor 488 was examined with a Laser scanning microscope Leica TCS SP5 using 63× immersion with pZ about 276 nm and XY resolution of 150 nm. The visualization of the whole sections was performed using stereo microscope Leica TCS SP5 MP. Sections were stored at 4°C until they were processed for TEM.

Sections on the coverslips were washed three times for 5 min each in 0.1M phosphate buffer (pH 7.2-7.4) and fixed with 2.5% glutaraldehyde in the same buffer for 20 min on ice, washed again (3 × 5 min) in cold buffer and postfixed in 0.5% OsO₄ for 10 min on ice. Sections were then washed in cold water, dehydrated in a graded ethanol series (50, 70, and 96% for 5 min each) followed by equilibration in 100% acetone, and embedded in Epon EmBed medium grade (Sigma Aldrich, MO, United States). After polymerization, coverslips were detached from the resin blocks by dipping in liquid nitrogen. Outlines of the sections were clearly visible on the surfaces of the blocks. Blocks were trimmed and series of about 140 ultrathin sections were produced. Sections were mounted on formvar coated slot grids and stained with 2% uranyl acetate and 3% lead citrate. Sections were analyzed using JEM Jeol 1400 TEM equipped with Veleta side camera (Olympus Corporation, Tokyo, Japan) at 80 kV.

Fluorescent puncta corresponding to single PD or pit fields containing PD-localized calreticulin were counted manually. Altogether, nine types of cell–cell boundaries were analyzed. Numbers of calreticulin-labeled PD/pitfields were counted on the following borders: (1) between anticlinal cell walls of cells of the upper epidermis; (2) between periclinal cell walls of the upper epidermal cells and of palisade parenchyma cells; (3) between anticlinal cell walls of palisade parenchyma cells; (4) between palisade parenchyma and cells of the bundle sheath of minor veins; (5) between periclinal cell walls of palisade parenchyma cells and cell walls of the adjoining spongy parenchyma cells; (6) between spongy parenchyma and cells of the bundle sheath of minor veins; (7) between spongy parenchyma cells; (8) between spongy parenchyma cells and periclinal cell walls of lower epidermis cells; (9) between anticlinal cell walls of cells of the lower epidermis. For barley leaves, mesophyll cells of the adaxial and abaxial halves of the leaf blade were analyzed separately. In all species, only minor veins of higher (4th−5th) orders were analyzed; veins of lower orders (“major” veins) were excluded from the analysis.

For statistical verifications, numbers of immunolabeled PD/pitfields were counted on 10 sections per species, all sections originating from different leaves. In every section, analyses were performed in the following manner: within cells of the same tissue, where the stoichiometry of adjoining cells was typically 1:1 (epidermis/epidermis, palisade parenchyma/palisade parenchyma, spongy parenchyma/spongy parenchyma), immunolabeled PD/pitfields were counted for 10 intercellular boundaries per section (which resulted in 100 intercellular boundaries for 10 sections analyzed). Where two different tissues adjoined each other, the stoichiometry was different; e.g., in A. meridionalis, usually two or three palisade parenchyma cells bordered one upper epidermal cell. In these cases, the total number of immunolabeled PD/pitfields was counted on the periclinal cell walls of 10 epidermal cells bordering 20–30 parenchyma cells and normalized to one epidermal cell. This resulted in 100 epidermal cells analyzed on 10 sections per species. Similarly, the numbers of immunolabeled PD/pitfields between spongy parenchyma/palisade parenchyma, palisade parenchyma/bundle sheath, and spongy parenchyma/bundle sheath were calculated and normalized to a single cell of spongy parenchyma or palisade parenchyma, respectively. Lengths of cell walls in microns were determined on the same sections.

The significant differences in the numbers of immunolabeled PD/pitfields counted per cell surface unit was analyzed using one-way ANOVA (library “car”; Fox and Weisberg, 2019) with Tukey post hoc test, libraries “multcomp” (Hothorn et al., 2008) and “agricolae” (de Mendiburu, 2020). Statistical analysis and data visualization were performed using R 3.6.3 (R Studio Team, 2020) and RStudio (R Studio Team, 2020). R packages “dplyr” (Wickham et al., 2020), “ggplot2” (Wickham, 2016), “readxl” (Wickham and Bryan, 2019) and “knitr” (Xie, 2020) were used as well. For some data sets, log-transformation was applied to satisfy the assumptions of one-way ANOVA; square root transformation was used for countable data sets. The significanct differences in single cell concentrations of sugars were analyzed using Student's t-test.

To analyze the symplastic connections between cells of the leaf lamina in barley, potato, A. barclaiana, and A. meridionalis, antibodies raised against two PD-associated proteins were used, against myosin VIII (Radford and White, 1998; Reichelt et al., 1999) and calreticulin (Baluska et al., 1999; Schubert et al., 2013; Demchenko et al., 2014). Both antibodies resulted in similar labeling of PD in leaf tissues (Figure 1). In the vascular tissues, previous TEM studies had shown the presence of large PD fields at the intermediary cell/bundle sheath cell boundary in the minor veins of A. meridionalis leaves, and also smaller PD fields between modified intermediary cells and bundle sheath cells in the minor veins of A. barclaiana (Voitsekhovskaja et al., 2006). These PD fields were detected by immunofluorescence staining of calreticulin as shown in Figures 1A,B for A. meridionalis and A. barclaiana, respectively, as well as by myosin labeling (data not shown), but single PD within these fields could not be distinguished due to the small size of the cells and extremely high density of PD (Gamalei, 1991). In the parenchymatous tissues, however, PD were clearly recognized as multiple fluorescent puncta in cell walls that were sometimes seen as threads penetrating the cell walls, depending on the angle of the section (Figures 1C,D). In all the species studied, immunofluorescence revealed multiple PD between mesophyll cells (shown in Figures 1C,D for A. meridionalis and potato, respectively). In potato, but not in the other species, some non-specific binding of the antibodies to structures within chloroplasts occurred as can be seen in Figure 1D. This was the only non-specific labeling observed; except for this, the immunofluorescence patterns were similar to the PD-specific patterns reported in other studies with the same antibodies (Radford and White, 1998; Baluska et al., 1999; Reichelt et al., 1999; Schubert et al., 2013; Demchenko et al., 2014). Fields of fluorescent puncta like those found in the cell walls of mesophyll cells were also found between cells of neighboring epidermal cells and at the mesophyll/epidermis interface (shown in Figures 1E,F for A. barclaiana, Figures 1G,H for barley, Figures 1I,J for A. meridionalis; see also Supplementary Videos 1–8).

As the resolution limit of light microscopy is ca. 200 nm, and the diameter of PD is in the range of several tenths of nanometers (Robards, 1976), several closely adjoining PD in a pitfield would be detected as a single immunolabeled fluorescent punctum. Moreover, it has been reported that in certain plant tissues, PD do not contain calreticulin (Demchenko et al., 2014). In order to address the question whether punctate pattern of immunofluorescence staining of calreticulin was associated exclusively with PD, we performed TEM analysis of immunolabeled leaf sections as shown in Figure 2 for a barley leaf. A series of ultrathin sections (70 nm) were cut through the whole depth of a semi-thin (10 μm) section which had been imaged with the confocal laser scanning microscopy (CLSM) prior to TEM (a single scan; Figures 2B,C) to visualize fluorescent puncta corresponding to PD-localized calreticulin. A cell selected for TEM analysis is marked with an asterisk in Figures 2A–D. TEM analyses revealed single, twinned, and Y-shaped PD (Figure 2, TEM images 1–9) which were grouped in pitfields containing 3–7 PD (Figure 2, arrows 1, 2, 4, 5, 7–9), or present as single PD (Figure 2, arrow 6). In the upper part of the analyzed cell, four fluorescent puncta in the cell wall (arrows 1, 2, 4, 5) corresponded to pitfields with 3–7 PD, while no PD were detected by TEM in the cell wall site without immunolabeling (arrow 3). In the lower part of the same cell, TEM study revealed a single PD (arrow 6) and three pitfields (arrows 7–9) while only sites 6 and 8 showed some immunolabeling (Figure 2C). This could be explained by a slight bending of the section that probably resulted in the lower part of the cell which is out of the CLSM focal plane. Altogether, we concluded that immunolabeled calreticulin was confined to PD in the cell walls, and that immunohistochemistry was a reliable method to detect sites of symplastic connections between leaf cells.

In order to provide a quantitative estimate of the abundance of sites of symplastic contacts between various types of cells and tissues of leaves of the four species under study, the numbers of fluorescent puncta (representing immunolabeled PD/pitfields) per length of the cell wall in microns were counted at the boundaries between all cell types except for vascular tissues. An example of a transverse section used for counting is shown in Figure 1D for the palisade cells on the left. For this procedure, we used the anti-calreticulin antibody for immunolabeling. Altogether, nine types of intercellular boundaries were analyzed. In the leaves of all species, we distinguished between upper and lower epidermis, palisade and spongy mesophyll, and cells of the bundle sheath around minor veins (major veins were excluded from the analyses). In barley leaves where only one type of mesophyll cells is present, “upper” (adaxial) and “lower” (abaxial) layers of the mesophyll were analyzed separately and are designated here as “palisade” and “spongy” to simplify the comparison with the other species. Based on leaf development, secondary PD are exclusively expected to be found at epidermis/mesophyll boundaries, while other boundaries analyzed in this study are expected to be connected by both primary and secondary PD (Satina and Blakeslee, 1941; Poethig, 1987). The numbers of immunolabeled PD/pitfields and the length of the cell walls for all cell types examined in this study are shown in Table 1.

In the three dicot species analyzed, the highest numbers of fluorescent puncta per cell–cell boundary (corresponding to immunolabeled PD/pitfields) were found in the palisade mesophyll. Here, the average numbers ranged from 14 puncta per cell–cell boundary in A. meridionalis to 38 in potato (Table 1). A somewhat lower abundance of symplastic connections was observed for spongy parenchyma cells and between palisade and spongy parenchyma (Table 1). In the monocot barley, the highest numbers (17 puncta per cell boundary) were observed between the mesophyll cells of the abaxial part of the blade, which significantly differed from the numbers for the adaxial blade part (Table 1). In all the four species, mesophyll cells were symplastically connected to each other to the highest extent as compared to other tissues, while the lowest numbers (0.2–1.5 puncta per cell boundary) were found in leaves of barley and potato for epidermis/epidermis (upper and lower) and upper epidermis/palisade parenchyma boundaries (Table 1).

When the lengths of the cell walls were taken into account, the highest values (1–2 fluorescent puncta corresponding to immunolabeled PD/pitfields per micron of cell wall length) were observed between cells of the spongy parenchyma in all species (Figure 3A). A cross-section of the leaf of A. barclaiana indicating the position of the analyzed cell types within the leaf lamina is shown in Figure 3B. The values of other cell types were much lower (0.25–0.60 immunolabeled PD/pitfields μm⁻¹ CW length). Strikingly low symplastic connectivity (0.02–0.10 immunolabeled PD/pitfields μm⁻¹ CW length) was found for upper epidermis/upper epidermis and upper epidermis/palisade mesophyll boundaries in barley and potato, as well as for lower epidermis/lower epidermis and lower epidermis/spongy mesophyll boundaries in barley. At the same time, values for boundaries of epidermal cells in A. meridionalis and A. barclaiana did not differ from those for other cell types, being in the range of 0.3–0.6 immunolabeled PD/pitfields μm⁻¹ CW length (Figure 3A).

In angiosperms, shoot apical meristems (SAMs) are organized in distinct cell layers originating from the initial cells. Cells of the L1 and sometimes L2 layers, forming the tunica, undergo anticlinal divisions while cells of the L3 layer do not show a distinct pattern of divisions (see also Supplementary Figure 1 for L1 in A. barclaiana and A. meridionalis SAMs). Based on this model, it is easy to conclude that secondary PD are present at L1/L2 and at L2/L3 boundaries, and primary PD connect cells of the same layer (although the formation of secondary PD between cells derived from the same layer is possible; Fitzgibbon et al., 2013). PD at the borders of cells belonging to L1 and L2 layers, as well as at L1/L2 and L2/L3 boundaries, were quantified in SAMs of A. barclaiana, A. meridionalis, barley, and potato by means of TEM, because the small size of the SAM cells would not allow reliable determinations using immunohistochemistry as was performed for leaves. The patterns of PD distribution and their frequencies were generally similar in the SAMs of all species studied (Figure 4). The frequencies of PD between cells of the L1 layer (designated as primary PD) were significantly lower than the frequencies of PD between cells of L1 and L2 (secondary PD) in potato and barley, but not in the other species, at the level of p < 0.05. A general tendency to form more secondary PD than primary PD per cell wall length unit was apparent for all species.

The level of symplastic connectivity between epidermis and mesophyll in A. meridionalis and A. barclaiana was similar to that between mesophyll cells, in contrast to barley and potato, where symplastic connectivity was higher within the mesophyll than between the epidermis and mesophyll (Figure 3A). Thus, the question arose whether symplastic exchange of sugars between epidermis and mesophyll might occur in A. meridionalis and A. barclaiana. We therefore analyzed the concentrations of sucrose and hexoses in single epidermal and mesophyll cells in A. meridionalis and A. barclaiana during the day and during forced accumulation of sugars in leaves due to blockage of export via the phloem.

First, the levels of non-structural carbohydrates were measured in the whole leaves of A. barclaiana and A. meridionalis, and no pronounced changes were observed over the diurnal rhythm (Figures 5A,B). In order to cause the accumulation of soluble carbohydrates in the leaves, leaves were detached from the plants after 3 h of the light period and placed in continuous light for 24 h, while the petioles were kept in a 2 mM CaCl₂ solution to favor sealing of the phloem with callose (King and Zeevaart, 1974). This treatment disrupted sugar export from the leaves while photosynthesis continued. At the end of the light exposure period, concentrations of sucrose, glucose, and fructose had increased significantly in the detached leaves from A. barclaiana and A. meridionalis, as compared to their levels at the end of the day in normally functioning attached leaves (Figures 5A,B). Another major carbohydrate-conjugated compound, the iridoid glucoside antirrhinoside, found in A. barclaiana (Voitsekhovskaja et al., 2006), showed no significant accumulation upon export blockage from the leaves (Figure 5A); neither did mannitol, myo-inositol, galactinol, raffinose, and stachyose in leaves of both species (data not shown).

The concentrations of glucose, fructose, and sucrose in the epidermal cells of A. meridionalis and A. barclaiana were compared with those in mesophyll cells in the morning (after 3 h of illumination) and in the evening (after 11 h of illumination) at the single-cell level in plants growing under a 16 h light/8 h dark regime. The results are shown in Figures 5C,D. At both “morning” and “evening” time points, the levels of these sugars were in a similar range for epidermis and mesophyll cells in both species, except for the accumulation of glucose in mesophyll cells of A. barclaiana in the “evening”. When expressed on the basis of total hexose units, the “morning” concentrations of soluble sugars in mesophyll cells did not differ significantly from those in epidermal cells in these species and were in the range between 20 and 70 mM (Figures 5E,F). At the “evening” time points, sugar concentrations in mesophyll cells of both species were in the range between 20 and 100 mM and thus were significantly higher than those in epidermal cells (20–70 mM) at the level of p < 0.05 (Figures 5E,F).

When the assimilate export from the leaves was blocked by the detachment of the leaves, the epidermal cells in both plants accumulated soluble carbohydrates up to the levels measured in mesophyll cells as shown by the concentrations of sucrose and hexoses measured in the single cells of detached leaves after 24 h of illumination (Figures 5E,F). In A. barclaiana, the single-cell total sugar concentrations (calculated as a sum of hexoses and sucrose, expressed in hexose units) were 157 ± 40 mM in the epidermal cells and 175 ± 50 mM in the mesophyll cells. In A. meridionalis, these values were 126 ± 50 mM in epidermal cells and 113 ± 40 mM in mesophyll cells. These data indicate that in A. meridionalis and A. barclaiana leaves, the epidermis is able to accumulate soluble sugars.