Clonal seedling cuttings of H. polysperma, a heterophyllous amphibious plant, were used. Three plants were planted into each vinyl pot (220 ml) containing soil (Leaf Pro Soil normal, Leaf Corporation, Gunma, Japan). Ten pots were placed into each of two glass tanks (W30 cm × D30 cm × H40 cm). The pots were partially submerged in tap water such that the water level was kept below the soil surface. Plants were grown at 25°C and illuminated with a 150-W metal halide lamp (Funnel 2 8000K, Kamihata, Hyogo, Japan) with a photosynthetic photon flux density (PPFD) of 200 μmol m^–2 s^–1, 8-h light and 16-h dark. After 2 weeks of initial growth, plants were separated into two environmental treatments: terrestrial and submerged. The terrestrial treatment maintained the initial growing condition. The submerged treatments were achieved by adding 25 L of tap water to the tank and maintaining the same light and temperature conditions as described for the terrestrial treatment. The temperature, pH, and alkalinity in the submerged treatment at day time were 25°C, 7.38, and 0.3 mEq L^–1, respectively. Leaves that developed after the terrestrial and submerged treatments were regarded as terrestrial and submerged leaves, respectively. The uppermost fully expanded leaves from both terrestrial and submerged treatments were used for subsequent experiments.

Underwater oxygen evolution was monitored using a liquid phase oxygen electrode (OXYG-1, Hansatech, Norfolk, United Kingdom). Four leaf disks (6-mm diameter, total projected area of 113 mm²) were placed in an oxygen electrode chamber (DW1/AD, Hansatech, Norfolk, United Kingdom) containing measurement buffer whose composition was modified for every experiment. The temperature inside the oxygen electrode chamber was maintained at 25°C by submerging the chamber in a low-temperature water bath (NCB-1,200, EYELA, Tokyo, Japan). The DIC response curve for the net oxygen evolution rate was obtained for NaHCO₃ values between 10 and 1,650 μM. In addition to NaHCO₃, the measurement buffer contained 0.1 mM phosphate buffer (pH 6.3), 1.5 mM KCl, 1.0 mM NaCl. Photosynthesis was started by illumination at an irradiance of 285 μmol photons m^–2 s^–1 after dark acclimation for over 15 min. The DIC concentration increased immediately after injecting NaHCO₃. The light response curve of the net oxygen evolution rate was measured for PPFD values from 0 to 820 μmol photons m^–2 s^–1 in the presence of 10 mM NaHCO₃. The initial slopes were calculated in low DIC (10–150 μM NaHCO₃) and light (0–125 μmol photons m^–2 s^–1) regions. After measuring the underwater photosynthetic rate, leaves were collected to analyze the chlorophyll content.

The leaf chlorophyll concentration was determined as described previously (Porra et al., 1989). The leaf disks for measuring underwater photosynthesis were incubated in N,N-dimethylformamide at 4°C in the dark overnight. The absorbance of the extract was measured at three wavelengths: 750.0, 663.8, and 646.8 nm. The chlorophyll concentration and content per unit leaf area were calculated from the absorbance readings.

To investigate the DIC form used for photosynthesis, the photosynthetic oxygen evolution rate was measured at pH 6.3 and 8.3. DIC constituents (CO₂ and HCO₃^–) present in the media at pH 6.3 or 8.3 existed as a level of 1:1 or HCO₃^– only, respectively. The pH of the measurement buffer (10 mM NaHCO₃, 1.5 mM KCl, and 1.0 mM NaCl) was adjusted by adding 1 M HCl or 1 M NaOH. The inhibitor experiment was performed by a previously described procedure (Horiguchi et al., 2019). For inhibitor experiments, stock solutions (10 mM) of acetazolamide (AZ) and ethoxyzolamide (EZ) were prepared by dissolution in 50 mM NaOH. AZ and EZ inhibit the external CA and internal CA, respectively. Tris(hydroxymethyl)aminomethane (TRIS), an inhibitor of HCO₃^–/H⁺ symport, and 4,4′-diisothiocyanatostilbene-2,2′-disulfonate (DIDS), an anion exchange protein inhibitor, were prepared daily. The photosynthetic oxygen evolution rate was measured in the HCO₃^– condition (pH 8.3) in the presence of 0.1 mM AZ, 0.1 mM EZ, 50 mM TRIS, or 0.3 mM DIDS.

Terrestrial and submerged leaves were harvested 2 h after the onset of the light period and were stored at −80°C until enzymatic activities were measured. The frozen samples were homogenized in an ice-cold extraction buffer [50 mM Tris (hydroxymethyl) aminomethane (Tris), 15 mM MgCl₂, 5 mM dithiothreitol (DTT), 0.1 mM ethylenediaminetetraacetic acid (EDTA), 10% (w/w) polyvinylpolypyrrolidone, 5% (w/v) polyvinylpyrrolidone, and 10%(v/v) glycerol (pH 8.0)] with quartz sand using a cold mortar and pestle. Part of the homogenate was used to measure the chlorophyll content; the remainder was centrifuged at 12,000 × g, 15 min at 4°C. The supernatant liquid (the crude extract) was kept on ice during the measurements.

Phosphoenol pyruvate carboxylase activity was determined as described previously (Zhang et al., 2014). Briefly, PEPC activity in the crude extract was measured in 1 mL of reaction mixture [50 mM Tris, 15 mM MgCl2, 0.1 mM EDTA, 20 mM bicarbonate, 0.2 mM NADH, 1 mM phosphoenol pyruvate (PEP), and 5 U malate dehydrogenase (pH 8.0)]. The reaction was started by adding PEP after preincubating the reaction mixture and sample for 5 min at 25°C. The absorbance at 340 nm was recorded for 5 min, and the activity was calculated from the change in absorbance as a function of time.

Rubisco activity was determined by modifying a previously described procedure (Suzuki et al., 2007). The Rubisco activity in crude extracts was activated by incubation with 20 mM NaHCO₃ at 25°C for 5 min. The activated sample was injected into a reaction mixture [50 mM Tris, 15 mM MgCl₂, 0.1 mM EDTA, 10 mM bicarbonate, 0.2 mM NADH, 0.6 mM ribulose 1,5-bisphosphate, 5.0 mM phosphocreatine, 20 U of glyceraldehyde-3-phosphate dehydrogenase, 10 U of phosphoglyceric acid kinase, and 1 U of creatine phosphokinase (pH 8.0)] that was bubbled with N₂ gas. The activity was calculated from the decreasing absorbance at 340 nm as a function of time.

The CO₂ hydration and HCO₃^– dehydration activities of CA were determined by monitoring pH changes after the addition of substrate (Wilbur and Anderson, 1948; Kikutani et al., 2016). The CO₂ hydration reaction was initiated by the addition of ice-cold CO₂ saturated water to 20 mM Tris–HCl buffer (pH 8.4) containing the crude extract. The time required for the pH to drop from 8.3 to 8.0 was measured. The HCO₃^– dehydration reaction was initiated by the addition of ice-cold 50 mM NaHCO₃ to 50 mM MES buffer (pH 5.5) containing the crude extract. The time required for the pH to increase from 5.7 to 6.0 was measured. The CA activity was calculated as the Wilbur and Andersson unit (WAU): WAU = T0/T − 1, where T0 and T are the times required for the pH change in the absence and the presence of the sample, respectively.

Chlorophyll in leaf homogenates was extracted in the dark for 15 min with 80% (v/v) acetone on ice. The extract was centrifuged at 20,400 × g for 5 min at 25°C. The absorbance of the supernatant liquid was measured at 750, 663.6, and 646.6 nm to calculate the chlorophyll concentration (Porra et al., 1989).

Malate was extracted as previously described (Yang and Liu, 2015) and quantified by an enzymatic method. Leaves were sampled at dusk (the end of the light period) or dawn (the end of the dark period). Samples were ground in ice-cold perchloric acid using an ice-cold mortar and pestle. The homogenate was kept on ice for 20 min and centrifuged at 12,000 × g for 30 min at 4°C. The supernatant liquid was collected, and the pellet was extracted again in ice-cold 5% (v/v) perchloric acid. The supernatant liquids from both extractions were combined, and the pH was adjusted to pH 6.5 ± 0.2 using K₂CO₃. After incubation on ice for 10 min, the samples were centrifuged at 12,000 × g for 10 min at 4°C. The supernatant liquid was filtered (0.45 μm pore size), the volume was measured, and the samples were stored at −25°C before analysis. Malate in the leaf extracts was detected by an enzymatic method using an F kit (#139068, Roche Diagnostics K.K., Tokyo, Japan).

Leaf segments (3 mm × 3 mm) added to 2% (v/v) glutaraldehyde and 2% (w/v) paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) were placed in a desiccator connected to an aspirator and vacuum infiltrated to −0.09 Pa for 5 min. Infiltration was repeated until the segments were fully submerged, followed by fixation at 4°C for 3 h. The leaf segments were rinsed in 0.1 M phosphate buffer and dehydrated by an ethanol series (50, 70, 80, 90, 95, and 99.5%) followed last with acetone. The samples were infiltrated with the graded infiltration resin series, Spurr resin (Spurr Low Viscosity Embedding kit, Polysciences, Inc., PA, United States): acetone = 1:2, 1:1, 2:1 for 1.5 h, and Spurr’s resin overnight. The leaf segments were embedded in Spurr’s resin at 70°C overnight. Samples (1 μm) for light microscopy were sectioned using an ultramicrotome (Leica EM UC7, Leica, Wetzlar, Germany) at room temperature. Leaf cross sections were stained with 0.1% (w/v) toluidine blue in 0.1 M phosphate buffer (pH 7.0) and were observed using a light microscope (BX41, Olympus, Tokyo, Japan). Quantitative characteristics of leaf structure were measured using Image J ver. 1.45 s (National Institutes of Health, Bethesda, MD, United States).

Leaf morphological characteristics, including leaf area, leaf length, leaf width, and stomatal density were measured. The area, length, and width of the leaves were measured from leaf images scanned using Image J. The leaf thickness was measured from leaf cross section using Image J. Stomatal density was determined using the same microscope.

Proteomic analysis was performed as described previously (Perera et al., 2019). Briefly, three terrestrial (88 mg) and three submerged leaves (68 mg) were harvested. The leaves were immediately frozen in liquid N₂ and stored at −80°C. Soluble and membrane proteins were extracted from the frozen leaves (Perera et al., 2019). Samples were analyzed using Nano liquid chromatography-mass spectrometry (LC–MS/MS) with an UltiMate 3,000 (Dionex, Tokyo, Japan) and Q-Exactive Plus (Thermo Fisher Scientific, Tokyo, Japan). LC-MS/MS data were analyzed using the search engine Mascot Server v2.5.1. Spectral data were identified against NCBInr. Statistical analysis of protein spectral counts was conducted using the proteome software Scaffold v4.8.7 (Proteosome Software, United States).

The data were expressed as the mean ± standard deviation (SD). Student’s t test was performed to detect differences between the two groups using Microsoft Excel 2013 (Microsoft, WA, United States). One-way analysis of variance (ANOVA) and Tukey’s honestly significant difference (HSD) test were calculated using JMP statistical software v9.0.2 (SAS Institute, NC, United States) and used for multiple comparisons. The initial slopes of the DIC or light response curves were generated by the Levenberg–Marquardt method and compared using the Prism software v6.07 (GraphPad Software, CA, United States). We evaluated the significance of the effect of leaf, measurement DIC concentration, or light interactions on photosynthetic rate by using two-way ANOVA; multiple comparisons were performed by the Holm–Sidak test in the Prism software v6.07. Data were determined to be statistically different when p < 0.05.

The morphology of H. polysperma leaves that developed under terrestrial or submerged environments was strikingly different (Figure 1A). Submerged leaves were significantly narrower and thinner compared with terrestrial leaves (Table 1). H. polysperma has stomata on both the abaxial and adaxial leaf surfaces. The abaxial stomatal density of submerged leaves was significantly lower than that of terrestrial leaves; however, adaxial stomatal density was not significantly different between terrestrial leaves and submerged leaves (Table 1). Submerged leaves had a significantly lower chlorophyll a and b content compared with terrestrial leaves, but the ratios of chlorophyll a to chlorophyll b (a/b) were similar (Table 1).

We measured the net photosynthetic O₂ evolution rate (Pn) relative to the chlorophyll content and leaf area under submerged conditions when DIC or light intensity was varied. Submerged leaves generally had a higher underwater Pn at pH 6.3 in response to an increased DIC content than terrestrial leaves at DIC concentrations above150 μM NaHCO₃ (Figure 1B). Furthermore, the initial slopes of the DIC response curves from 10 to 150 μM NaHCO₃ were 0.09 ± 0.02 and 0.26 ± 0.04 μmol O₂ mg^–1 h^–1 μM^–1 for terrestrial and submerged leaves, respectively (mean ± SD, p < 0.001). Submerged leaves had a higher underwater Pn in response to increased light intensity compared with terrestrial leaves (Figure 1C). The initial slopes of the light response curves from 0 to 125 μmol photons m^–2 s^–1 were 0.29 ± 0.06 and 0.60 ± 0.07 μmol O₂ mg^–1 h^–1 PPFD^–1 for terrestrial and submerged leaves, respectively (mean ± SD, p = 0.003). In contrast, underwater Pn per unit leaf area responded to DIC and light differently than underwater Pn expressed relative to the chlorophyll content. Terrestrial leaves had a larger underwater Pn on a per leaf area basis compared with submerged leaves at high DIC levels and high light intensities (Supplementary Figure 1). There were no significant differences in the initial slopes of underwater Pn in response to DIC content or light intensity between terrestrial and submerged leaves.

To investigate the HCO₃^– affinity of H. polysperma, we measured the underwater Pn at pH 6.3 at which the concentrations of CO₂ and HCO₃^– are equivalent and at pH 8.3 at which the concentration of HCO₃^– is dominant. The underwater Pn relative to the Chl content was highest in the submerged leaves at pH 6.3 and lowest in the terrestrial leaves at pH 8.3 (Figure 2A). The submerged leaves had a significantly higher underwater Pn compared with the terrestrial leaves in both DIC conditions. The Pn at pH 6.3 was significantly higher than at pH 8.3 in both leaf types. The HCO₃^– affinity values calculated as the ratio of the Pn at pH 8.3 to the Pn at pH 6.3 were 0.3 ± 0.06 and 0.72 ± 0.11 for terrestrial and submerged leaves, respectively (mean ± SD, p = 0.001).

We also determined the CO₂ hydration and HCO₃^– dehydration reactions from CA activity in terrestrial and submerged leaves. The CO₂ hydration reaction in terrestrial leaves was significantly higher than the CO₂ hydration reaction in submerged leaves or the HCO₃^– dehydration reaction in submerged and terrestrial leaves (Supplementary Figure 2). There was no difference between the CA activities other than the CO₂ hydration reaction in terrestrial leaves. The HCO₃^– dehydration/CO₂ hydration ratio in the submerged leaves was significantly higher compared with that in the terrestrial leaves (Figure 2B).

Because submerged leaves had an increased HCO₃^– affinity, we examined the mechanism of HCO₃^– use by an inhibitor experiment. EZ, an inhibitor of internal CA, and DIDS, an inhibitor of the SLC family of anion exchangers, significantly inhibited the photosynthesis of submerged leaves (Figure 2C). The underwater Pns of submerged leaves was not sensitive to AZ and or TRIS, both of which inhibit external CA and HCO₃^–/H⁺ symport, respectively.

A total of 3,239 proteins were identified by proteomic analysis of which 186 proteins were differentially expressed between terrestrial and submerged leaves. One hundred proteins were significantly downregulated in submerged leaves compared with the terrestrial leaves, and 86 proteins were significantly upregulated (Supplementary Tables 1, 2). In this study, a protein identified as a HCO₃^– transporter in algae was not detected in H. polysperma. Among seven CA proteins detected in H. polysperma (Table 2), the amount of chloroplast β-CA was significantly lower in the submerged leaves compared with the terrestrial leaves.

The downregulated proteins included proteins associated with biochemical CCMs: β-CA, PEPC, aspartate aminotransferase (AspAT), NAD-malate dehydrogenase (NAD⁺-MDH), NADP-malic enzyme (NADP-ME), and alanine aminotransferase (AlaAT) (Figure 3A). Additionally, proteins related to the photosynthetic electron transport system were present in lower amounts in submerged leaves compared with terrestrial leaves (Supplementary Table 1). In contrast, the upregulated proteins included proteins associated with the Calvin cycle: Rubisco, phosphoglycerate kinase (PGK), fructose-bisphosphate aldolase (ALDO), transketolase (TK), and phosphoribulokinase (PRK) (Figure 3B). The Rubisco large subunit and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were among the proteins that were both up- and down-regulated.

Proteomics analysis implied that H. polysperma switches its photosynthetic metabolism in response to moving between terrestrial and submerged environments. We compared biochemical CCM activities between terrestrial and submerged leaves. The malate content was significantly higher at dusk than at dawn in both leaf types (Table 3). PEPC and Rubisco activities in the submerged leaves were significantly lower compared with those in the terrestrial leaves (Figures 4A,B). Moreover, submerged leaves had significantly lower PEPC/Rubisco ratios than terrestrial leaves (Figure 4C).

Figure 5 compares the anatomy of leaf cross sections prepared from H. polysperma terrestrial and submerged leaves. Terrestrial leaves had cells that varied in both shape and size. In cross section, the upper layer of MCs in terrestrial leaves were vertically long, whereas those in the lower layer were horizontally long or spherical (Figure 5A). In contrast, submerged leaves had horizontally long or spherical cells but not vertically long cells (Figure 5B). We observed that both upper and lower MCs in terrestrial leaves have a chloroplast; however, some lower MCs in submerged leaves had no chloroplast (Figures 5A,B). Chloroplasts were absent in the epidermal cells of both terrestrial and submerged leaves. Kranz anatomy was evident in cross sections of terrestrial leaves but not in submerged leaves (Figures 5A,B); however, chloroplasts in BSCs were not located near the vascular bundles as is typical for the Kranz anatomy in some C4 plants.

The analysis of cross sections from terrestrial and submerged leaves identified significant differences in leaf structure (Figure 6). The MC size was significantly higher in submerged leaves than in terrestrial leaves (Figure 6A). The BSC and epidermal cell sizes were significantly smaller in the submerged leaves compared with the terrestrial leaves (Figures 6B,C). Therefore, the submerged leaves had a significantly higher MC/BSC size ratio compared with terrestrial leaves (Figure 6D). On the other hand, there were no differences in the percent of tissue coverage of a cross section between terrestrial and submerged leaves (Supplementary Table 3).

HCO₃^– use mechanisms are distinguished by inhibitor sensitivity in underwater photosynthesis. In a previous study, four inhibitors of HCO₃^– utilization (AZ, TRIS, EZ, and DIDS) were used to distinguish HCO₃^– use mechanisms in aquatic phototrophs (Beer et al., 2002; Hellblom and Axelsson, 2003; Fernández et al., 2014; Rubio et al., 2017; Tsuji et al., 2017). In marine diatoms, DIDS, an anion exchanger inhibitor, suppressed high-DIC-affinity photosynthesis in the pennate diatom P. tricornutum and the centric diatom Chaetoceros muelleri, but AZ, an external CA inhibitor, did not affect high-DIC-affinity photosynthesis (Tsuji et al., 2017). Tsuji et al. (2017) also reported that AZ reduced HCO₃^– affinity in the pennate Cylindrotheca fusiformis and the centric Thalassiosira pseudonana, but DIDS did not affect HCO₃^– affinity. The DIDS-sensitive group converted HCO₃^– to CO₂ after transport of HCO₃^– into cells; the DIDS-insensitive group facilitated CO₂ diffusion by external CA. Submerged leaves of H. polysperma had an enhanced affinity for HCO₃^– compared with terrestrial leaves (Figure 2A). The ratio of HCO₃^– dehydration to CO₂ hydration by CA was higher in the submerged leaves than in the terrestrial leaves (Figure 2B). Furthermore, EZ, internal CA inhibitor, and DIDS inhibited the underwater Pn of submerged leaves, AZ and TRIS; HCO₃^–/H⁺ symporter inhibitor did not (Figure 2C). These results suggest that H. polysperma transports HCO₃^– into leaves by a DIDS-sensitive transporter and converts HCO₃^– to CO₂ by an intercellular CA. H. polysperma leaves grown under submerged conditions achieved a higher photosynthetic rate by a biophysical CCM in low DIC levels (Figure 1B); therefore, submerged plants accumulated more proteins associated with the Calvin cycle compared with plants grown under terrestrial conditions (Figure 3B). HCO₃^– use mechanism in H. polysperma is an effective way to acclimate to submerged environment.

4,4′-Diisothiocyanatostilbene-2,2′-disulfonate inhibits HCO₃^– uptake and/or underwater photosynthesis in plants ranging from algae to higher plants (Nakajima et al., 2013; Fernández et al., 2014; Huang et al., 2020). Huang et al. (2020) first reported the presence of direct HCO₃^– uptake via a DIDS-sensitive SLC4 transporter in the higher plant O. alismoides. Here, we report that the underwater Pn of H. polysperma-submerged leaves was also inhibited by DIDS; this is the first report of DIDS inhibiting the photosynthetic gas exchange of a higher plant. DIDS inhibits not only the SLC4 family of transporters but also members of the SLC26 gene family in mammals (Soleimani et al., 2001). HCO₃^– transporters associated with biophysical CCMs have been identified in both the SLC4 and SLC26 gene families; SLC4 is present in a marine diatom (Nakajima et al., 2013) and an aquatic angiosperm (Huang et al., 2020), and BicA, a member of the SLC26/SulP transporter family, is present in cyanobacteria (Price and Howitt, 2011). Nevertheless, neither SLC4 nor SLC26 anion exchangers annotated as HCO₃^– transporter were identified in our proteomics analysis without regard to the expression levels. These results suggest that H. polysperma may move HCO₃^– into leaves via a novel and as yet unidentified DIDS-sensitive HCO₃^– transporter.

In algae, inorganic carbon uptake within the cells consumes light energy (Tchernov et al., 2003; Burnap et al., 2015; Wang et al., 2015). To photosynthesize in MCs, H. polysperma transports HCO₃^– into leaves via the epidermal cells. Because H. polysperma does not have chloroplasts in the epidermal cells (Figure 5), energy for HCO₃^– uptake may have to be supplied by MCs. Submerged leaves had higher initial slopes of their photosynthetic light response curves on a chlorophyll basis, indicating more efficient light use (Figure 1C). The chlorophyll content per leaf area in the submerged leaves was significantly lower than that in the terrestrial leaves, but the light use efficiency on the basis of leaf area was constant (Table 1 and Supplementary Figure 1B). These results imply that the light use efficiency of submerged leaves was high to provide sufficient light energy for HCO₃^– uptake.

In algae, CA is a key component of the biophysical CCMs. In algae, CA mutants have lower photosynthetic DIC affinity and growth in low CO₂ conditions than wild type (Karlsson et al., 1998; Kikutani et al., 2016). These CAs are localized in the thylakoid lumen. Chloroplast CAs that are localized in the thylakoid lumen are assumed to supply CO₂ to the Calvin cycle. CA contributes not only biophysical CCMs but also biochemical CCMs. In C4 plants, CO₂ hydration activity of CA plays an essential role in primary CO₂ fixation by HCO₃^– production. Some studies revealed the CA contributions of MC and BSC to total CA activity. Amaranthus cruentus leaves and C4 eudicots, were isolated two CAs (Guliev et al., 2003). One associated with the chloroplast of BSCs is responsible for 8% of the total leaf CA activity; the other, found in the MC cytoplasmic fraction, represents 62% of the total leaf CA activity (Guliev et al., 2003). In corn (Zea mays, C4 grass) and green foxtail millet [Setaria viridis (L) P. Beauv., C4 grass], CO₂ assimilation rates in transgenic lines of loss of β-CA(s) were unchanged at ambient CO₂ but decreased at low CO₂ (Studer et al., 2014; Osborn et al., 2017). The β-CA mutant of S. viridis decreases growth and photosynthesis at ambient CO₂ (Chatterjee et al., 2021). On the other hand, pineapple (Ananas comosus), a CAM plant, increases the expression levels of three copies of CA at night time (Ming et al., 2015). Cytosolic β-CA activity may be partial to CO₂ hydration reaction contrary to chloroplast CA. The submerged leaves of H. polysperma increased the HCO₃^– dehydration/CO₂ hydration ratio of CA activity compared with that of terrestrial leaves (Figure 2B); however, the terrestrial leaves attained higher HCO₃^– dehydration and CO₂ hydration CA activities than those of the submerged leaves (Supplementary Figure 2). Seven differentially regulated proteins were identified as CAs by proteomic analysis (Table 2). The amount of chloroplast β-CA protein was significantly lower in the submerged leaves than in the terrestrial leaves, whereas the levels of other CAs did not change. In the case of other aquatic plants, P. lucens lack external CA (Staal et al., 1989); CA activity of E. canadensis is not influenced by the CO₂ concentration (Elzenga and Prins, 1988). In O. alismoides, external CA activity was greater in leaves acclimated to low CO₂ aquatic environment compared with that in leaves acclimated to high CO₂ aquatic environment; however, the expression of the four isoforms of putative α-CA1 was not significantly different between two CO₂ conditions (Huang et al., 2020). Our results suggest that the reduction in CO₂ hydration activity in the submerged leaves is caused by a decrease in the amount of β-CA, resulting in an increased HCO₃^– dehydration/CO₂ hydration ratio of CA activity in the submerged leaves.

Aquatic plants have different leaf anatomical characteristics compared with land plants: thinner leaves, cuticles, and cell walls, a few or no stomata, and epidermal cells with chloroplasts or chloroplasts facing the epidermis in MCs (Veen and Sasidharan, 2019). The thin leaves of aquatic plants improve gas exchange when leaves are submerged (Frost-Christensen et al., 2003). The submerged leaves of H. polysperma were thinner than the terrestrial leaves, and the stomatal density on the abaxial surface was lower than that of the terrestrial leaves (Table 1). The reduction in leaf thickness was achieved by decreasing the thickness of the MC layer (Figures 5A,B). Considering the results, H. polysperma appears to follow an acclimation mechanism for aquatic plants that facilitates passive CO₂ diffusion. Furthermore, both molecular size and diffusion resistance of HCO₃^– are larger than that of CO₂, thinner leaves, especially smaller epidermal cells (Figure 6C), are effective not only in CO₂ diffusion but also in HCO₃^– uptake. The submerged leaves had large MCs compared with terrestrial leaves (Figures 5A,B, 6A). These MCs in submerged leaves are expected to equip with big vacuoles. In submerged leaves of H. polysperma, some lower MCs were observed the absence of chloroplasts (Figure 5B). The submerged leaves may separate function between upper and lower MCs as in the cases of E. canadensis and P. lucens. For instance, upper MCs, which contain chloroplasts, perform photosynthesis using HCO₃^–; lower MCs, which have no or few chloroplasts, excrete or store OH^– to maintain pH homeostasis.

In this study, we found that the submerged leaves of H. polysperma accumulated lower levels of proteins associated with biochemical CCM compared with the terrestrial leaves (Figure 3A). A result of the proteomics analysis implies that the terrestrial leaves use a biochemical CCM, and the submerged leaves do not. Biochemical CCMs include C4 photosynthesis and CAM, two pathways that require specific enzymes. Therefore, we measured the diel changes in malate content to confirm CAM activity in the leaves of H. polysperma. Terrestrial and submerged leaves did not accumulate malate at dawn (Table 3), suggesting that the terrestrial leaves did not perform CAM. Modified C4 photosynthetic pathways are identified as C3–C4 and C4-like photosynthesis in some plant species such as Flaveria (Mckown and Dengler, 2007), Eleocharis (Murphy et al., 2007), Heliotropium (Muhaidat et al., 2011), Moricandia (Schlüter et al., 2017), and Chenopodium (Yorimitsu et al., 2019). C3–C4 intermediate plants have extended BSCs and lower CO₂ compensation point. In Eleocharis, C3–C4 intermediate species exhibit weak C4 cycles that means that there is a higher rate of PEPC to Rubisco than that of the C3 species and localization of enzymes between MC and BSC (Murphy et al., 2007). In Heliotropium and Cheopodium C3–C4 intermediate species, abundant mitochondria and localization of glycine decarboxylase in extended BSCs suggested that they capture, concentrate, and re-assimilate CO₂ released by photorespiration (Muhaidat et al., 2011; Yorimitsu et al., 2019). This simple CCM is called C2 photosynthesis, glycine shuttle, and photorespiratory CO₂ pump (Lundgren, 2020). Interestingly, some Heliotropium and Chenopodium species exhibit a proto-Kranz type of photosynthesis that is functionally C3 but involves BSCs. Proto-Kranz plants have larger BSC size than MC size like the biochemical CCM plant; however, enzyme activities related to C4 photosynthesis are the same levels as C3 plants (Muhaidat et al., 2011; Yorimitsu et al., 2019). The biochemical and anatomical characteristics of H. polysperma leaves had unique features that do not differentiate between photosynthetic types. The ratio of PEPC to Rubisco increases with the transition from C3 to C4, leading to enhanced C4 activity. In the terrestrial leaves of H. polysperma, the ratio of PEPC to Rubisco was significantly higher than that in the submerged leaves (Figure 4C). However, the ratios of PEPC to Rubisco were 0.29 and 0.14 in the terrestrial and submerged leaves, a range in values that is typical for C3 plants (Murphy et al., 2007; Zhang et al., 2014). Anatomical analysis of H. polysperma found vascular bundles surrounding BSCs that contained few chloroplasts (Figures 5A,B). Nevertheless, the ratio of MC/BSC in a cross sectional area was 4.5 in the terrestrial and submerged leaves (Supplementary Table 3), a value that is identical to that in Flaveria, a C3–C4 plant (Mckown and Dengler, 2007), Heliotropium (Muhaidat et al., 2011), and the Kranz-like plant Chenopodium (Yorimitsu et al., 2019). In H. polysperma, the cell size ratios of MC/BSC were significantly different between terrestrial and submerged leaves (Figure 6D) at 0.5 and 1.3, respectively. These ratios are within the range for proto-Kranz- and non-Kranz-type plants (Yorimitsu et al., 2019). Our results imply that under terrestrial conditions, H. polysperma conducts incomplete C4 photosynthesis, likely proto-Kranz, and alters its photosynthetic metabolism to C3 photosynthesis when leaves are underwater.

In summary, we report a comprehensive photosynthetic mechanism to acclimate H. polysperma, a heterophyllous amphibious plant to a submerged environment that has limited CO₂. Leaves that developed in the submerged condition adopted two DIC uptake strategies: (i) passive CO₂ diffusion without external CA and (ii) active HCO₃^– transport by a DIDS-sensitive HCO₃^– transporter-like member of the SLC gene family. Furthermore, our results suggest that H. polysperma can alter its photosynthesis from a proto-Kranz type in terrestrial leaves to a C3 type in submerged leaves. To improve photosynthetic performance for increasing crop yields, many researchers have attempted to introduce a biophysical CCM or a C4 pathway into C3 crop plants (Price et al., 2013; Ermakova et al., 2020). Further characterization and understanding of the acclimation mechanisms to extreme stress conditions in H. polysperma will provide novel resources for improving stress resistance and photosynthesis in eudicot crop plants.