The Gransden wild-type strain from Physcomitrella patens Bruch and Schimp was used for this study. Gametophytes were either grown on solid BCD media in petri plates, or on soil fertilized by liquid BCD media, in a growth chamber with 16 h light and 8 h dark regime (Cove et al., 2009). Sporophyte growth was induced by growing 4–5 weeks old mature gametophytes in a chamber held at 15°C under 8 h of light and 16 h of darkness.

Individual mature Physcomitrella gametophytes were imaged using an Olympus SZX7 stereomicroscope (Olympus Corporation, Tokyo, Japan) equipped with an Infinity 2 camera system and Infinity Capture imaging software (Lumenera Corporation, Ottawa, Canada).

Resin-embedded naked stems were prepared for transmission electron microscopy (described below) and cut into semi-thin (250–500 nm) sections, stained with toluidine blue, and imaged using a Nikon Eclipse TE300 inverted microscope (Nikon Corporation, Tokyo, Japan) equipped with an Olympus DP26 camera and cellSens digital imaging software (Olympus Corporation, Tokyo, Japan).

Individual mature Physcomitrella gametophores were fixed in 4.0% glutaraldehyde and 1.0% formaldehyde in 0.05 M sodium phosphate buffer, pH 7.2 for 2 h at room temperature. The leaves were gently separated from several stems under a stereomicroscope while still in the fixative fluid, and both leafy and naked stems were dissected into ∼2 mm pieces. The leaves and stems were further fixed in fresh, buffered fixative for 2 h at room temperature, post-fixed with 1% osmium tetroxide (OsO₄) for 2 h at room temperature, and en bloc stained with 0.2% uranyl acetate overnight at 4°C. The tissues were then serially dehydrated in acetone, gradually infiltrated with Spurr’s resin over 5–7 days, and flat-embedded in molds or on Teflon-coated glass slides in a 60°C oven. The tissues were embedded in various orientations to facilitate longitudinal and transverse sectioning. The resin blocks were trimmed, cut into 70–100 nm sections with a diamond knife (Diatome Inc., Hatfield, PA, United States) in a Leica Ultracut R microtome (Leica Microsystems, Vienna, Austria), mounted onto formvar-coated copper slot or 100-mesh grids (Electron Microscopy Sciences, Washington, PA, United States), and post-stained with 2% methanolic uranyl acetate and Sato’s lead citrate. The sections were then imaged at 80 kV using JEOL 1200EX transmission electron microscope (JEOL Ltd., Tokyo, Japan) equipped with SIA model L3C bottom-mount air-cooled CCD camera (Scientific Instruments and Applications, Inc., Duluth, GA, United States) and Maxim DL5 imaging software (Diffraction Ltd., Ottawa, ON, Canada). All the fixative chemicals were purchased from Electron Microscopy Sciences (Washington, PA, United States).

Individual gametophores and sporophytes were fixed with 4% glutaraldehyde in 0.05 M sodium phosphate buffer, pH 7.2 for 4–6 h at room temperature, post-fixed in 1% OsO₄ for 2 h at room temperature, serially dehydrated with acetone, and critical point dried (Balzers Union, Principe de Liechtenstein). If the tissues were to be dissected, the critically dried samples were cut in a desirable fashion using a fresh double-edged razor blade under a stereomicroscope. The samples were then mounted onto aluminum stubs, gold-palladium sputter-coated (Technics, Alexandria, Virginia), and imaged at 15 kV using JEOL JSM-6300 scanning electron microscope (JEOL Ltd., Tokyo, Japan) equipped with IXRF digital acquisition system (IXRF Systems, Inc., Austin, TX, United States). All the fixative chemicals were purchased from Electron Microscopy Sciences (Washington, PA, United States).

The scanning and transmission electron micrographs (SEMs and TEMs, respectively) were then processed using Adobe Photoshop CC software (Adobe Systems Incorporated, San Jose, CA, United States). Linear adjustments (brightness, contrast, and levels) were made to the entire images to enhance clarity.

Individual gametophores or sporophyte-bearing gametophores were vertically floated on ∼5 μL of 250 μg.mL^-1 5(6) carboxyfluorescein diacetate (CFDA) or 250 μg.mL^-1 HPTS solution inside PCR tube strips, and incubated for 1, 3, or 24 h in a dark humid chamber. The plants were then rinsed in excess volume of double-distilled water at least three times, and directly whole-mounted on slides with water as a mounting medium, and imaged with an AMG EVOS FL Cell Imaging epifluorescent system (Life Technologies Inc., United States) using in-built 4X/0.13 NA objective lens, GFP excitation (470 nm) and emission (525 nm) light cube, and brightfield optics. Negative controls were performed by floating the plants in water only. The sporophyte-bearing gametophores were identically processed, except that most of the leaves were excised prior to imaging. For CFDA tracer study in individually excised leaves, single leaves were carefully removed from the stem using fine forceps under a stereomicroscope, and processed identically as above except only 1 – 2 μL of CFDA solution was used.

Aniline blue staining of plasmodesmata-associated callose of Physcomitrella leaves were performed according to the protocol described by Zavaliev and Epel (2015), using 0.01% (w/v) aniline blue in 0.01 M K₃PO₄, pH 12. The leaves were then whole-mounted in the staining solution itself and imaged with a Leica TCS SP5 AOBS Spectral Confocal microscope (Leica Microsystems) equipped with a 20×/0.7 NA Plan-Apo dry objective lens, DIC optics, and Leica confocal software. Plasmodesmata-associated aniline blue staining was imaged (512 pixels × 512 pixels) at an excitation wavelength of 405 nm, and the emission signal was detected between 475 and 525 nm for aniline blue, and between 643 and 730 nm for chlorophyll autofluorescence. Negative controls were performed by excluding aniline blue.

Entire gametophores and sporangia-bearing gametophores were immediately immersed in at least 20-fold excess volume of FAA (10% v/v 37% formaldehyde : 5% v/v acetic acid: 50% v/v 200-proof ethanol : 35% v/v water) fixative, and placed under a gentle vacuum for 10 – 15 min at room temperature. The fixation was then allowed to continue overnight at 4°C. Following this, the tissues were dehydrated in a graded ethanol series, exchanged with xylene to clear the tissues, and gradually infiltrated with paraplast. The tissues were then embedded in paraplast in desired orientations. Using a Leica RM2155 rotary microtome (Leica Microsystems, Germany),¹ the tissue blocks were cut into 10 μm thick sections and mounted on poly-L-Lysine coated slides. The tissues were then deparaffinized, rehydrated, and exposed to an antigen retrieval buffer (Biogenex, United States)² in a 65°C water bath for 20 min. After washing in water, the endogenous peroxidase activity was quenched using 3% (v/v) H₂O₂. The tissues were washed with Phosphate Buffered Saline, pH 7.2 with 0.01% (v/v) Tween-20 (PBST), and blocked with 1% (w/v) casein in PBST for 60 min at room temperature. After washing with PBST three times, the tissues were exposed to anti-AVP1, anti-SUS, or anti-AHA3 rabbit polyclonal antibodies at 1:1000 dilutions overnight at 4°C. Negative controls were concurrently performed using pre-immune serum at 1:1000 dilutions in PBST, or by excluding the primary antibody. After two washes with PBST and one wash with PBS, the signal was developed using the SignalStain^® Boost IHC HRP anti-rabbit detection system (Cell Signaling Technology, United States),³ following the manufacturer’s instructions. To develop the signal when mouse anti-serum was used, the DAKO EnVISION anti-rabbit anti-mouse system was used according to the manufacturer’s instructions (DAKO Inc.).⁴ The tissues were then run through a dehydrating ethanol series, and were permanently mounted. Images were acquired with a Zeiss Axioskop light microscope (Carl Zeiss Inc.) equipped with 10× Achroplan 0.25 NA Zeiss, 40× Plan Neo-Fluar 0.75 NA Zeiss, and 100× 1.30 NA Oil Plan Neo-Fluar objective lenses, phase and differential interference contrast optics, Olympus DP72 camera system, and Olympus cellSens^® imaging software. Alternatively, a Nikon Eclipse E600 microscope equipped with Nikon Plan Fluor 10× 0.30 NA, 20× 0.50 NA, 100× 1.30 NA Oil objective lenses (Nikon Instruments Inc.), Omax A355OU camera (OMAX Microscopes), and ToupView software (ToupTek Photonics) was used.

Approximately 300 mg of Physcomitrella gametophores were immediately ground into fine powder with liquid N₂ on a cold mortar and pestle. The liquid N₂ was allowed to evaporate, and the fresh frozen powder was transferred in ∼200 μL aliquots to screw-cap eppendorf tubes. To each tube, 1 mL of 10% (v/v) Trichloroacetic acid in -20°C acetone was added, and proteins were allowed to precipitate overnight at -20°C. The samples were centrifuged at 10,000 × g for 30 min at 4°C, followed by removal of the supernatant. The pellet was washed with -20°C acetone containing 0.07% β-mercaptoethanol, vortexed, and centrifuged at 10,000 × g for 10 min at 4°C. The washing, vortexing, and centrifugation steps were repeated four more times. After the final centrifugation step, the supernatant was removed and the pellet dried in a tabletop vacuum for ∼30 min. Total protein was solubilized by adding Laemmli’s buffer to the pellet. Prior to storage at -80°C the mixture was vortexed and centrifuged at 5,000 × g for 5 min at 4°C three times.

The solubilized protein extracts from were then run on a 10% SDS-polyacrylamide gel, transferred to a PVDF membrane, blocked with 5% (w/v) non-fat milk in Tris buffered saline with 0.01% Tween-20 (TBST), and probed with H⁺-PPase-specific polyclonal sera or anti-AHA3 polyclonal sera at 1:1000 dilution overnight at 4°C. After three washes with TBST, the membrane was developed using the Bio-Rad Alkaline Phosphatase Immun-Blot^® Colorimetric Assay kit (Bio-Rad Inc., United States)⁵ according to the manufacturer’s instructions.

Approximately 100 mg of gametophores were immediately ground to fine powder using liquid N₂ in pre-chilled mortar and pestle. The liquid N₂ was allowed to evaporate, and the fresh frozen powder aliquoted into five pre-chilled screw-cap eppendorf tubes. Then 250 μL of 80% ethanol was added to each tube, vortexed, incubated at 80°C for 20 min with shaking (100 rpm) in Eppendorf Thermomixer R (Eppendorf Co.), and centrifuged at 14,000 rpm for 5 min at 4°C. The supernatants from the tubes were transferred to fresh pre-cooled eppendorf tubes, and these tubes were kept on ice. To the pellet, 150 μL 80% ethanol was added, vortexed, incubated at 80°C for 20 min with shaking (100 rpm) in Eppendorf Thermomixer R, followed by centrifugation at 14,000 rpm for 5 min at 4°C. The supernatants from the tubes were then pooled with the respective supernatants from the first step. Finally, 250 μL of 50% ethanol was added to the pellet, vortexed, incubated at 80°C for 20 min in the Eppendorf Thermomixer R, and centrifuged at 14,000 rpm for 5 min at 4°C. Again, the supernatants were pooled into respective tubes. The soluble carbohydrate content, i.e., sucrose, glucose, and fructose, was spectrophotometrically determined using the Megazyme^® Sucrose/D-Fructose/D-Glucose Assay Kit (Megazyme Inc.) following the manufacturer’s instructions.

To determine the uptake of photosynthate by the heterotrophic sporangia, ¹⁴CO₂ labeling was performed on sporophyte-bearing gametophytes. Physcomitrella patens were grown on 10 of the 18 wells in an 18-well plate under 16 h/8 h light/dark period at 22°C until the gametophores attained maturity (∼5–6 weeks) and then transferred to 8 h/16 h light/dark period at 15°C until the gametophores bore sporangia. Approximately 20 sporangia-bearing gametophores were used for each time point. The experiment was performed when the sporangia were still green. The 18-well plate containing the plants was kept under a halogen lamp (∼500 lumens) for 2 h to acclimatize the plants to the lighting conditions, and then kept in a sealed transparent box prior to radioisotope labeling. A micro-blower (Pelonis Technologies Inc.) powered by an external battery pack was affixed inside the box to homogeneously circulate the air, and later ¹⁴CO₂. ¹⁴CO₂ was injected into the box by mixing 2.5 μL of NaH¹⁴CO₃ (specific activity: 56 mCi/mmol, concentration: 2.0 mCi/mL; MP Biomedicals) with an excess of 80% lactic acid in a syringe. Photosynthesis was allowed to take place for 20 min, after which the ¹⁴CO₂ was vented, and the box kept in the dark. At specific time points (1, 3, 5, 9, and 25 h) two colonies were taken out of the wells, and only the sporangia were excised under a stereomicroscope and immediately transferred to two scintillation vials (∼15 sporangia per vial) containing 100 μL of 80% methanol to stop metabolism. Then 50 μL of commercial bleach was added to the vials, and finally 5 mL of scintillation cocktail (MP Biomedicals) was added to each of the vial. After the mixture had cleared, scintillation counting was performed in a Beckman Coulter LS6500 liquid scintillation counter.

Given that both CFDA and HPTS moved freely through the gametophore, with their movement only halted at the gametophore-sporophyte interface, we argued that exogenously supplied radiolabeled sugars would also move similarly. [U¹⁴C]-sucrose (specific activity: 432 mCi.mmol^-1; concentration: 100 μCi.mL^-1; MP Biomedicals) and [U¹⁴C]-maltose (specific activity: 600 mCi.mmol^-1; concentration: 100 μCi.mL^-1; American Radiolabeled Chemicals Inc.) were prepared at final concentrations of 0.1 μCi.mL^-1 in a solution containing 0.5 mM CaCl₂, 0.25 mM MgCl₂, and 1 mM “cold” sucrose. 2,4-DNP was prepared in the same solution, with or without [U¹⁴C]-sucrose at final concentrations of 5, 1, and 0.5 mM. DEPC was prepared identically at final concentrations of 1.7 and 0.5 mM. For all the assays, individual sporophyte-bearing gametophores were floated on a 5 μL drop of experimental solution in PCR tube strips inside a humid chamber, and allowed to take up the assay solution for 4 h in the dark. For [U¹⁴C]-sucrose and [U¹⁴C]-maltose assays, the mosses were individually pretreated on 5 μL drops of solution containing 0.5 mM CaCl₂ and 0.25 mM MgCl₂ for 1 h before transferring them to solutions containing radiolabeled sugars. Similarly, for the inhibitor treatments, the mosses were incubated on 5 μL drops of solution containing 0.5 mM CaCl₂ and 0.25 mM MgCl₂ with aforementioned inhibitor concentrations for 1 h before transferring them to solutions with both labeled and unlabeled sucrose and the specific inhibitor. After 4-h long incubation in the dark, individual sporangium was excised under a stereomicroscope and immediately transferred to a scintillation vial containing 100 μL of 80% methanol. Then 50 μL of commercial bleach was added to the vials, and finally, 5 mL of scintillation cocktail (MP Biomedicals) was added to each of the vials. After the mixture had cleared, scintillation counting was performed in a Beckman Coulter LS6500 liquid scintillation counter.

High-pressure freezing (HPF) and freeze-substitution (FS): Haustoria from gametophytes with fully grown sporophytes were excised under a drop of 1-hexadecene, and quickly loaded into interlocking brass planchettes, filled with 1-hexadecene, and subjected to HPF in a BalTec HPM 010 high-pressure freezer. The cryofixed haustoria were then transferred to a pre-chilled -80°C FS cocktail containing anhydrous acetone and 0.2% (w/v) non-methanolic uranyl acetate. The FS was allowed to continue for 96 h at -80°C, after which the FS solution was allowed to warm to 4°C over a period of 12 h. The anhydrous acetone was exchanged with anhydrous ethanol several times, and the tissues were gradually infiltrated with medium grade London Resin (LR) White (Electron Microscopy Sciences, United States) at 4°C: 25% (v/v) LR White (12 h), 50% (v/v) LR White (12 h), 75% (v/v) resin (12 h), 100% LR White (12 h), 100% LR White (12 h), 100% LR White (12 h). The samples were transferred to gelatin capsules, filled with fresh LR White and polymerized for 24 h at 50°C in longitudinal orientation. The polymerized resin blocks were then trimmed, ultrathin sectioned to gold reflectance (90 – 100 nm) using a Leica Ultracut R ultramicrotome (Leica Microsystems, Germany), and mounted on formvar-coated nickel slot grids.

For immunogold labeling, the nickel slot grids containing the sections were floated on drops of blocking solution (1% w/v casein in PBST) for 1 h at room temperature. The grids were then sequentially transferred to drops of PBST before being floated on drops of anti-AVP1 or anti-SUS polyclonal antibodies diluted 10-fold in the blocking solution. The antibody incubation step was performed at 4°C overnight. Following this, the grids were washed five times with PBST, and then transferred to drops of anti-rabbit IgG conjugated to either Aurion Ultrasmall 0.8 nm (H⁺-PPase) or 10 nm (SUS) gold particles (Electron Microscopy Sciences, United States) diluted 25- to 50-fold in the blocking solution. The grids were further washed in PBST five times, and rinsed with double-distilled water three times, before being air-dried overnight. Silver enhancement of Aurion Ultrasmall immunogold labeled sections was performed for 3 min in a dark room by using HQ SILVER^TM Enhancement kit (Nanoprobes Inc.) following the manufacturer’s instructions. The sections were then imaged with a JEOL 1200EX transmission electron microscope (JEOL Corp., Japan) operating at 80 kV without post-staining. Linear image adjustments like contrast, brightness, and levels were adjusted on entire grayscale electron micrographs using Adobe Photoshop CC (Adobe Inc., United States).

The amino acid sequences for H⁺PPase and SUS orthologs in P. patens were identified by BlastP in www.cosmoss.org or www.gramene.org, using the reference sequences from A. thaliana,⁶ Vigna radiata,⁷ and Oryza sativa.⁸ The amino acid sequences were first aligned using MUSCLE built-into the MEGA5 software, and then the evolutionary history was inferred by using the Maximum Likelihood method based on the JTT matrix-based model (Jones et al., 1992). The bootstrap consensus tree was deduced from 500 replicates. Evolutionary analyses were conducted in MEGA5 (Tamura et al., 2011).

Although mosses undergo an alternation of generations like all land plants, they differ from vascular plants in that the haploid gametophytic stage, rather than the diploid sporophytic stage, is the dominant phase of the life cycle. In Physcomitrella, the gametophores (Figure 1A) emerge from protonemal filaments and are affixed to the substrate by basal filamentous rhizoids. There is a single apical meristematic cell that gives rise to young phyllids (Figures 1B,C), as the gametophore undergoes vertical growth. At the apex of the monoecious gametophore, archegonia containing the female gamete or egg (Figure 1D) and antheridia containing male gametes, or sperm form. Upon fertilization, an apical sporophyte (Figures 1E,F) is formed within each archegonium. The diploid tissue in this structure, the capsule undergoes meiosis to form numerous haploid spheroidal spores, with a characteristic exine layer (Figures 1G,H).

The Physcomitrella stem is an anatomically complex structure, and possesses a well-developed central column of water-conducting cells or hydroids (Figures 1I–K), corroborating a previous report (Sakakibara et al., 2003). These cells are thin-walled, and are devoid of living protoplasm at maturity, presumably due to programmed cytoplasmic lysis (Figure 1K) (Hebant and Johnson, 1976). The midrib or costa of leaves, in cross-section, reveals that the central region of the midrib is mostly comprised of thick-walled parenchyma or stereid cells (Figure 1L). These stereid cells have narrow lumens, and do not retain living protoplasts at maturity (Figure 1M). Given the thickness of their cell walls, it seems plausible that the function of the stereid cells is limited to the mechanical support of the leaf. The thick walls of stereid cells show loose, fibrillar fine structure, and their function in internal conduction of water cannot be ruled out (Figure 1M). In the leaf costa, thin-walled hydroids lie abaxial to the stereids, and are mostly empty (Figure 1L). The leaves penetrate the stem in the cortical layer, just below the epidermis, forming a leaf trace (Figure 1N). The presence of small, thick-walled stereid cells, and thin-walled empty hydroids characterize the leaf trace (Figure 1N). These cells are presumably contiguous with their respective counterparts in the leaf midrib. The cross-section of Physcomitrella leaf costa shows a putative food-conducting cell, with plasmodesmatal connections to the laterally adjoining cells (Figure 1O). The stems of Physcomitrella have elongate putative food-conducting cells with oblique end walls that are profusely decorated with plasmodesmata (Figures 1P–Q1). The presence of plasmodesmata suggests that the mechanism of sucrose translocation via these food-conducting cells is symplasmic in nature (Raven, 2003).

Using a fluorescent tracer 5(6)-carboxyfluorescein diacetate (CFDA), we studied whether the translocation of photosynthate in the gametophytic generation of Physcomitrella was symplasmic in nature. After 1 h of application from the rhizoidal end, CFDA was found to have translocated to the tip of the leafy gametophores (Figures 2A,B), providing support for our ultrastructural observation (Figure 1Q). Upon observing the movement of CFDA in the gametophores 24 h after application, the fluorescence was spread throughout the stem (Figures 2C,E), and the leaves (Figure 2D). These results directly suggested that a symplasmic continuum exists in the Physcomitrella gametophyte.

We further analyzed the movement of CFDA in individually excised leaves, and 12 h after application of CFDA from the bottom of the leaves, the fluorescence was uniformly spread through all the leaf cells including the midrib (Figure 2F), while a negative control did not show any interfering autofluorescence (Figure 2G). This finding suggested plasmodesmatal interconnections among leaf cells, and upon callose-specific aniline blue staining, a distinct punctate fluorescent pattern was observed (Figures 2H–K). Considered a diagnostic test for the presence of plasmodesmata, aniline blue staining in the leaves of Physcomitrella gave further proof that the leaf cells of Physcomitrella are symplasmically connected. As expected, a negative control experiment showed no plasmodesmata-specific aniline blue staining (Figures 2L–O).

A previous study reported that a diffusion barrier exists between the gametophytic and sporophytic generation of Physcomitrella (Courtice et al., 1978). We fed the sporophyte-bearing gametophores with CFDA from the rhizoidal end, and 3 h after application, it was found that the movement of CFDA was halted at an interface separating the sporophyte from the gametophore (Figures 3A–C). Identical results were obtained when HPTS was used as a fluorescent tracer (Supplementary Figures S1A–C). To further verify that the movement of CFDA was symplasmic in the sporophyte-bearing Physcomitrella gametophore stem, we pretreated the plants with 2,4-DNP, a metabolic uncoupler, and found that CFDA was translocated unimpeded through the gametophore stem (Supplementary Figures S1F,G), and that a diffusion barrier existed between the haploid and diploid generations of Physcomitrella.

In bryophytes, the sporophyte-gametophyte junction called the haustorium, has long been known to harbor a band of transfer cells (Ligrone and Gambardella, 1988; Ligrone et al., 1993). The name of the cells itself implies their presumptive role in the ‘transfer’ of metabolites and nutrients. Light microscopic evidence for the existence of transfer cells at this interface in Physcomitrella exists (Xu et al., 2014), but electron microscopic evidence has been lacking. We analyzed the haustorium using a conventional chemical fixation regime, and the distinctive cytological features of transfer cells were evident (Figures 3D,E). Saliently, the transfer cells had intricate cell wall invaginations with numerous mitochondria in their immediate vicinity – morphological attributes seen as directly facilitating the energy-intensive nature of nutrient transfer via an increase in the absorptive surface area and proximally located sources of primary energy currency ATP (Figure 3E). A large central nucleus, several vacuoles, and plastids were also observed (Figures 3D,E).

Given the proximity of numerous mitochondria to the cell wall invaginations (Figure 3E), we posited that P-type ATPases must be localized in the haustorium. Using polyclonal rabbit antibodies generated against the KLH-conjugated TISKDRVKPSPTPDS epitope of A. thaliana companion cell-specific AHA3 protein (DeWitt et al., 1991; DeWitt and Sussman, 1995; Pizzio et al., 2015), we tested the cross-reactivity of the antibody against total protein extracts from Physcomitrella by western blotting, and found a distinct band (>100 kDa) corresponding to P-type ATPase orthologs in the moss (Figure 3I). In silico analysis revealed 8 putative orthologs of ATPases in Physcomitrella, and that the antigenic epitope was indeed highly conserved (67–93%) in this moss (Supplementary Figure S3A). And as expected, P-type ATPases were prominently immunolocalized at the haustorium, suggesting a proton gradient energized uptake of nutrients via the transfer cells (Figures 3F,G). The specificity of the primary antibody was confirmed by excluding the primary antibody, which yielded no staining (Figure 3H).

To test for the physiological aspects of photosynthate movement between the two generations of Physcomitrella, we first analyzed the soluble carbohydrate content in the gametophore, and in agreement with previous reports (Allsopp, 1951; Thelander et al., 2009), found that sucrose was by far the predominant soluble carbohydrate (Figure 4A). Given the likely nutritional dependency of the sporophytic generation on the photoautotrophic gametophyte (Ligrone and Gambardella, 1988; Ligrone et al., 1993), we postulated that the growing sporangium of Physcomitrella was a strong heterotrophic sink tissue. While immature sporangia are green in color (Figure 1E), and likely photosynthetic, the surface area to volume ratio imparted by its spherical structure is not ideal for maximal photosynthetic capacity. We labeled immature sporophyte-bearing Physcomitrella gametophores with ¹⁴CO₂ for 20 min, and chased the radioactive pulse sequestered in the sporangia at several time points (1, 3, 5, 9, and 25 h) over a period of 25 h in the dark (Figure 4B). Within the 1st hour, a minimal amount of label was measured in the sporangia, suggesting that these structures were indeed minimally photosynthetic. Over the course of 25 h, it was found that more and more radioactive label had moved into the sporangia, and that the kinetics of the photosynthate uptake into the sporangia was likely saturable (Figure 4B).

The set of evidence obtained thus far painted a picture wherein the sporangia were heterotrophic sinks, and that the primary photosynthate sucrose was apoplasmically loaded via the transfer cells. This implied that, like in apoplasmically loading higher plant species like A. thaliana, the uptake of sucrose into the transfer cells was mediated by sucrose/proton symporters (SUTs) powered by the proton gradient generated by P-type ATPases. Therefore, we used a protonophore 2,4-DNP, and a SUT-specific inhibitor DEPC (Bush, 1993) to investigate their effects on the uptake of exogenously applied [U-¹⁴C]-sucrose into the growing sporangia. 2,4-DNP was effective in significantly diminishing sucrose uptake into the sporangium at all the concentrations (5, 1, and 0.5 mM) tested (Figure 4C). It has been known that DEPC specifically modifies the histidine residue located in the first extracellular loop near the second transmembrane domain of the SUT, and is accessible from the extracellular face of the plasma membrane (Bush, 1993; Lu and Bush, 1998), and upon bioinformatics analysis, it was found that of the six SUT paralogs harbored in the Physcomitrella genome, this histidine residue was fully conserved among all of them (Supplementary Figure S3C). Accordingly, application of 1.7 mM DEPC significantly decreased the exogenous sucrose uptake into the sporangium, while 0.5 mM DEPC, while reducing the mean sucrose uptake, was not significantly effective (Figure 4C).

Phylogenetic analysis has shown that Physcomitrella harbors both types II and III SUTs (Kühn and Grof, 2010; Reinders et al., 2012a), wherein the former subtype is localized to the plasma membrane and is generally implicated in phloem loading in monocots, while the latter are localized at the vacuoles and mediate sucrose uptake from the cytoplasm. Furthermore, given that the type II SUTs are also known to have a high affinity for maltose (Sivitz et al., 2005; Reinders et al., 2006, 2012b; Sun et al., 2009), we measured the uptake of [U-¹⁴C]-maltose into the sporangia, and found that the sporangia had taken up approximately threefold less maltose associated label than they did sucrose (Figure 4C). These results further strengthened our argument that the mechanism of sucrose movement into the sporangium via the transfer cells is empowered by ATPase generated proton gradient and is mediated by SUTs.

In apoplasmically phloem loading higher plant species, the sieve element-companion cell complexes are symplasmically isolated in the vascular bundles, and part of the actively symported sucrose is metabolized via the SUS pathway to supply ATP for the maintenance of proton gradient that, in turn, is used for sucrose symport via SUTs. There is a substantial body of evidence that implicates two enzymes – SUS and plasma membrane localized H⁺-PPase – as key components of this pathway (Lerchl et al., 1995; Paez-Valencia et al., 2011; Gaxiola et al., 2012, 2015; Pizzio et al., 2015; Regmi et al., 2015). As such, we immunolocalized these two enzymes in the haustorium/transfer cells at both the light and electron microscopy level (Figure 5).

H⁺-PPases are evolutionarily conserved proteins with a pyrophosphate-binding KAADVGADLVGKVE motif that is highly conserved among diverse organisms, including P. patens (Supplementary Figure S3B) (Rea and Poole, 1985; Rea et al., 1992). The antibody used in this study was raised against this motif in a rabbit (Park et al., 2005), and recognized the H⁺-PPase specific band (∼80 kDa) when immunoblotted against the total protein extract from Physomitrella (Figure 5D). In the sporophytic generation, marked by the development of the sporangium, a distinct immunohistochemical signal was observed at the sporophyte-gametophyte interface (Figures 5E–G). Specifically, localization of H⁺-PPases was evident at the transfer cells that characterize this filial junction (Figure 5G). Representative pre-immune and secondary antibody controls lacked any label (Figures 5H–J). Interestingly, the gametophore stem (Figure 5A) and leaves (Figures 5A,B) had relatively very weak signals. Pre-immune sera treated control, as expected, lacked any discernible signal (Figure 5C). A parallel immunohistochemical localization of H⁺-PPases using antisera raised against the KAADVGADLVGKVE motif in mouse (Paez-Valencia et al., 2011) was also performed, that further confirmed the transfer cell-specific localization of H⁺-PPases (Figure 5G).

Given the cytology of these transfer cells, and because these cells translocate sucrose (an endogenous cryoprotectant), and the need for immunogold labeling of these cells, we cryofixed the excised haustorium samples using HPF and freeze substitution (HPF/FS) regime. To determine the membrane-specificity of H⁺-PPases in transfer cells, silver-enhanced immunogold labeling was performed on high-pressure frozen samples (Figure 5K), and both the plasma membrane and the vacuoles were profusely decorated with H⁺-PPase-specific label (Figures 5L,M). Negative controls performed with pre-immune sera showed the specificity of the H⁺-PPase label (Supplementary Figures S2A–C). In parallel, immunohistochemical localization of SUS showed that it was prominently localized at the haustorium (Figure 5O), while the antibody-specific signal was very weak in the gametophyte stem and leaf (Figure 5N). The cross-reactivity of the antibody used against protein extracts from Physcomitrella was reported previously (Regmi et al., 2015). Negative controls that omitted the primary antibody showed no signal (Figures 5P,Q). Upon immunogold labeling of HPF/FS processed haustorium (Figure 5R) using the anti-SUS antibody showed clear labeling at the plasma membrane (Figures 5S,T), while the negative controls showed a clear lack of gold labeling (Supplementary Figures S2D,E).