Role of C4 carbon fixation in Ulva prolifera, the macroalga responsible for the world’s largest green tides

Most marine algae preferentially assimilate CO2 via the Calvin-Benson Cycle (C3) and catalyze HCO3− dehydration via carbonic anhydrase (CA) as a CO2-compensatory mechanism, but certain species utilize the Hatch-Slack Cycle (C4) to enhance photosynthesis. The occurrence and importance of the C4 pathway remains uncertain, however. Here, we demonstrate that carbon fixation in Ulva prolifera, a species responsible for massive green tides, involves a combination of C3 and C4 pathways, and a CA-supported HCO3− mechanism. Analysis of CA and key C3 and C4 enzymes, and subsequent analysis of δ13C photosynthetic products showed that the species assimilates CO2 predominately via the C3 pathway, uses HCO3− via the CA mechanism at low CO2 levels, and takes advantage of high irradiance using the C4 pathway. This active and multi-faceted carbon acquisition strategy is advantageous for the formation of massive blooms, as thick floating mats are subject to intense surface irradiance and CO2 limitation.

T he Calvin-Benson cycle (C 3 ) is the dominant pathway of carbon fixation in marine algae 1 . Rubisco (ribulose-1, 5bisphosphate carboxylase), a key C 3 enzyme responsible for carbon fixation, requires inorganic carbon in the form of CO 2 2 . However, the demand for CO 2 in algal photosynthesis is generally higher than CO 2 concentrations in natural seawaters, e.g., the half-saturation constants for CO 2 of Rubisco in diatoms are 30-60 μM 3 but CO 2 concentrations in natural seawater are only 5-25 μM 4 . Thus, CO 2 can be an important factor limiting algal proliferation in the ocean.
Since the 1970s there is growing evidence that most species of prokaryotic and eukaryotic algae have developed CO 2 -concentrating mechanisms (CCMs) that enable accumulation of CO 2 via bicarbonate (HCO 3 − ) enzymolysis 3,5 . A dominant route in the CCMs is that intra-or extracellular dehydration of HCO 3 − is catalyzed by carbonic anhydrases (CA) to release CO 2 to increase reactions at the Rubisco site (Fig. 1a). Although traditionally associated with more advanced terrestrial plants, in recent decades a C 4 or C 4 -like pathway has been discovered in the green alga Udotea flabellum 6 and the marine diatom Thalassiosira weissflogii 7,8 , in which transmembrane HCO 3 − is not only catalyzed via CA to generate CO 2 but also involves C 4 enzymesphosphoenolpyruvate carboxylase (PEPCase) and PEPCase kinase (PEPCKase). CO 2 incorporated in this manner eventually enters the C 3 cycle to increase reactions at the Rubisco site (Fig. 1b). More recently, a freshwater macrophyte Ottelia alismoides, a constitutive C 4 plant and bicarbonate user, was shown to possess three different CCMs that can operate with the C 4 pathway in the same tissue, even though Kranz anatomy is absent 9,10 . Theoretically, the joint operation of CCMs and the C 4 pathway can greatly improve carboxylation efficiency of algal species in the use of HCO 3 − , and consequently enhance CO 2 accumulation at the Rubisco site (Fig. 1b).
Ulva prolifera, the main species involved in the massive "green tides" in the Yellow Sea 11 , exhibits remarkable capability for biomass accumulation. U. prolifera biomass increases rapidlymore than 60-fold within~50 days 12,13 . Growth rates in the field were generally higher than 28% per day at temperatures greater than 20°C 14 , and aerial cover of floating canopies exceed 30,000 km 2 across the Yellow Sea. Consistent with this, the involvement of the C 4 pathway in photosynthesis of U. prolifera has been suggested as an important mechanism to achieve such rapid biomass accumulation, based on two lines of evidence: (1) gene and enzyme analysis in U. prolifera revealed the existence and activity of the C 4 -related enzyme pyruvate orthophosphate dikinase (PPDKase) 15 ; and (2) tissue δ 13 C range of U. prolifera (−21.9 to −14.9‰) at 16 stations in the Yellow Sea indicated a mix of C 3 and C 4 pathways in carbon fixation 16 .
These two lines of evidence, however, are not sufficient to characterize a full C 4 pathway in U. prolifera. That pathway requires the participation of the key C 4 enzymes, PEPCase and PEPCKase. PEPCase (in the cytoplasm) supports catalysis of carboxylation of phosphoenolpyruvic acid (PEP), which uses HCO 3 − as the inorganic carbon substrate leading to C 4 compound synthesis. PEPCKase (in the chloroplast) is a decarboxylase, generating CO 2 and insuring efficient transfer of CO 2 to Rubisco 17 . PPDKase is responsible for catalyzing the regeneration of PEP in the cytoplasm, but it is not the only way for the C 4 pathway to acquire PEP 17 . Some researchers even believe that PPDKase might not assist with net CO 2 fixation but rather has a role in protection against photoinhibition 18 . Therefore, examining the activities of PEPCase and PEPCKase is a necessary step to confirm the existence of C 4 pathway in U. prolifera.
The 13 C/ 12 C distributions in plant tissue records the integrated pattern of photosynthetic carbon acquisition 19,20 ; for example, high CO 2 uptake leads to low δ 13 C values (e.g., <−30‰ 21 ), whereas high HCO 3 − uptake gives high δ 13 C values (e.g., >−10‰ 22 ). Therefore, tissue δ 13 C is an important index using to distinguish C 3 and C 4 activity. The 13 C/ 12 C values in seaweeds, however, often display a wide range that makes definition of the C 4 contribution uncertain 20,21 . The variation in the ratios depends on ambient temperature, light, salinity, nutrients and CO 2 concentrations in seawater 20,22 . Growth and respiration in seaweeds 20,23 can also lead to discrimination and fractionation of 13 C/ 12 C. As canopies of U. prolifera drift in the Yellow Sea, environmental conditions and physiological activity change, forcing variations that may mask clear evidence for C 4 activity.
To resolve this uncertainty, controlled experiments are needed to examine the correlation between the activities of key enzymes and photosynthetic products to ascertain the activity of C 4 pathway and/or CA mechanism in U. prolifera. Although several environmental factors (e.g., temperature, light, salinity, HCO 3 − and CO 2 supply, and nutrient availability) may affect the functional expression and activity of CA mechanism and C 4 pathway, light intensity and pCO 2 are regarded as the most important two under the condition with sufficient nutrient supply 3,5,24 . Massive floating algal mats in nutrient enriched Yellow Sea are not only exposed to strong surface light, but the thick biomass also can reduce the dissolution of CO 2 from the air and lead to CO 2 limitation. Consequently, the HCO 3 − systems operated by CA mechanism and/or C 4 pathway would likely become active. Here, we describe the results of three outdoor culture experiments designed to examine the daily variations of key C 3 (Rubisco), CA (extra-and intra-cellular CA) and C 4 -metabolic enzymes (PEPCase, PEPCKase, and PPDKase) in U. prolifera. The relationships between the activities of major enzymes and photosynthetic products were analyzed via the corresponding variations in tissue δ 13 C, aiming to identify the participation of C 3 and C 4 photosynthetic pathways and CA mechanism in U. prolifera. The results exhibited the coexistence of C 3 and C 4 pathways and a CA-supported HCO 3 − mechanism in U. prolifera. The correlated variations of light intensity and pCO 2 with enzyme activity indicated that the C 4 pathway was most active under high light but that the CA mechanism became active at low CO 2 levels. The joint operation of the C 4 pathway and a CA-supported HCO 3 − mechanism in U. prolifera greatly improved efficiency of inorganic carbon fixation and illustrated why massive biomass accumulation can be formed in a short period, when thick floating mats are subject to intense surface irradiance and CO 2 limitation in the Yellow Sea.
The third experiment was conducted on a sunny day, aiming to identify the contribution of CO 2 and HCO 3 − in carbon fixation.
The activities of key C 3 enzyme (Rubisco), C 4 enzymes (PEP-Case, PEPCKase, and PPDKase), and CA enzymes (extra-and intra-cellular CA) and their responses to varying light levels were measured over the course of the day (Fig. 4). Similarly, the variations of pCO 2 , HCO 3 − , and tissue δ 13 C were also measured (Fig. 5). The intent of this trial was to examine variations in key enzymes that represent the three photosynthetic routes, in relation to variations in photosynthetic products (tissue δ 13 C).
From 10:00 to 12:00 as irradiance increased, tissue δ 13 C increased 10‰ (Fig. 5b), suggesting a link between C 4 pathway and carbon fixation using HCO 3 − . The activity of PEPCase and PEPCKase and tissue δ 13 C declined markedly at 2 p.m. (Fig. 4b), likely a result of decreased light intensity (Fig. 5b), consistent with the results from experiments 1 and 2, indicating the importance of strong light induction for C 4 pathway occurrence, rather than low pCO 2 . (3) Intra-cellular CA became more active in the afternoon (Fig. 4c) when pCO 2 was below 150 μatm (Fig. 5a). Throughout the experiment, the value of tissue δ 13 C kept increasing after 2 p.m. (Fig. 5b), although the activities of PEPCase and PEPCKase declined markedly (Fig. 4b).
Intracellular CA activity was negatively correlated to both pCO 2 (R 2 = 0.773, p = 0.049) and HCO 3 − concentration (R 2 = 0.717, p = 0.070), a result that indicated a CAdominant photosynthetic pathway late in the day when pCO 2 decreased. In contrast, extracellular CA was not active and remained unchanged during the experiment (Fig. 5b), indicating the species mainly used the diffusion of HCO 3 − in cytosol via intracellular CA (Fig. 1b).
These results suggest that U. prolifera used a HCO 3 − source via a combination of C 4 pathways and the CA mechanism in response to changes in irradiation intensity and CO 2 level.

Discussion
The results of this study not only demonstrated the coexistence of C 3 and C 4 pathways and the CA mechanism in U. prolifera but also indicated that the C 4 pathway catalyzed by PEPCase and PEPCKase were most active in U. prolifera photosynthesis under high light. In this regard, the photosynthetic machinery of U. prolifera could be similar to that previously described for the green alga U. flabellum 6 and the marine diatom T. weissflogii 8 . The CO 2 delivered to the Rubisco site comes from three sources: the available CO 2 from seawater, HCO 3 − catalyzed by the CA mechanism, and HCO 3 − catalyzed by PEPCase and PEPCKase via the C 4 pathway (Fig. 1b). The joint operation of the C 4 pathway and the CA mechanism greatly improve efficiency of inorganic carbon fixation in U. prolifera, and consequently, accelerate the rate of biomass accumulation (Fig. 1b). This multifaceted photosynthetic mechanism has important implications for the ability of this species to grow rapidly when there is high irradiance, and facilitates the formation of the massive floating green tide mats observed in the Yellow Sea 13 .
Studies of the C 4 pathway in marine algal photosynthesis are very limited, but the results from a few species indicate that the occurrence of a C 4 pathway and its importance in carbon fixation have distinct species-specific expression. For example, inhibition of PEPCase or PEPCKase could reduce more than 90% of photosynthesis in T. weissflogii 8 and~50% of photosynthesis in U. flabellum 6 . In this study, the range of tissue δ 13 C (−19.1 to −17.4‰) implied an important contribution of HCO 3 − in carbon fixation in U. prolifera. We cannot specifically define the contribution of HCO 3 − fixed separately via the CA mechanism versus the C 4 pathway, but we found that the environmental factors inducing the activity of C 4 enzymes are different from the CA mechanism. The diurnal modalities of C 4 enzymes among the three experiments were consistent, although their concentrations were different depending on the ambient environmental conditions at the time of those experiments (Figs. 4 and 5). The repeated modes indicate that the highest PEPCKase activity occurred at maximal light (Figs. 2c and 4b); PEPCKase has a direct correlation with C 4 acid formation 6 . Previous studies on the diatom T. pseudonana also found that light intensity is more important for the C 4 acid accumulation than low pCO 2 concentrations 25,26 . In this study, tissue δ 13 C showed a marked increase (10‰) at noon, indicating photosynthesis supported by HCO 3 − (Fig. 5) and that the process is operated by C 4 pathway rather than C 3 pathway and CA mechanism (Fig. 4). In contrast, HCO 3 − and CO 2 supply play important roles in modulating activity of C 3 pathway and CA mechanism 27 . High CO 2 generally suppresses expression of a high-affinity CCM state 28 , a feature reflected in our result that the C 3 pathway was active at high pCO 2 concentrations but that the CA mechanism became dominant when pCO 2 was low (Figs. 4 and 5a). HCO 3 − transport rate and C 4 activity are energetically costly and require high photon flux densities and sufficient nutrient supply 29,30 to support synthesis of specific proteins. Excess light energy under strong irradiation however enables ATP to accumulate in the cells and leads to photoinhibition. An active C 4 pathway at noon indicates that U. prolifera has an efficient capability to dissipate excess light energy and ATP in cells. Photosystems I and II (PSI and PSII) are light-harvesting operators, and the energy flux they capture will be quenched and redistributed between PSII/PSI to achieve a balance and avoid irreversible damage to the photosynthetic systems 31 . An induction experiment with U. prolifera showed that the electron transport rate (ETR) between PSII/PSI was still high when light intensity reached 800 µmol photons m −2 s −1 32 . Xu and Gao 33 found that the ETR and net photosynthetic rate of U. prolifera remain high and stable even as light intensity increased to 2000 µmol photons variations.  These results explain the continuous carbon accumulation of U. prolifera exposed to light intensity > 1200 µmol photons m −2 s −1 for more than 2 h in sunny day experiments, which is about two-or threefold higher than that required to saturate the photosynthetic rate (600 µmol photons m −2 s −133 ). PPDKase might play an important role in dissipating excess energy and ATP in cells, since it is responsible for catalyzing the regeneration of PEP in the cytoplasm, where ATP is consumed during PEP formation from pyruvate (Fig. 1b). The basis for this speculation is from the experiments of Haimovich-Dayan et al. 18 who used RNA-interference to silence the single gene encoding PPDKase in P. tricornutum and found that the variations of PPDKase activity hardly affected net CO 2 fixation but were distinctly involved in dissipating excess energy and ATP in cells. In this study, we only detected PPDKase under sunny conditions, indicating its link with protection against photoinhibition. Based on our study and previous evidence, we propose that, of the three enzymes of C 4 metabolism, net CO 2 fixation in U. prolifera mainly is determined by PEPCase and PEPCKase, but energy dissipation depends on PPDKase.
Coastal eutrophication and high light supply in the Yellow Sea are important environmental conditions that favor the initiation of C 4 function of U. prolifera. During the summer (June-August) when the U. prolifera bloom is forming, dissolved inorganic nitrogen (DIN) concentrations are 10-80 µM near the coast and 1-15 µM offshore 16 , and the average light intensity in surface seawaters is~1072.2 µmol photons m −2 s −1 34 . In this study, the DIN concentrations in the culture medium were about 30 µM and the daily average light intensity on sunny days in experiments 1 and 3 was 1196 and 1011 µmol photons m −2 s −1 , respectively. These values fit within the range of environmental condition in the Yellow Sea. The pCO 2 supply is limited during the formation of green tide in the Yellow Sea, with a range of 300-450 μatm 35 , and moreover, it can be affected by phytoplankton consumption. Large floating mats formed by U. prolifera block air-sea exchange, thus, HCO 3 − uptake and assimilation could become important for rapid biomass accumulation. This was reflected in the tissue δ 13 C (culture samples: −19.5 to −17.4‰; field samples: −21.9 to −14.9‰) 16 .
Ulva prolifera assimilates CO 2 predominately via the C 3 pathway, takes up HCO 3 − via the CA mechanism at low CO 2 levels, and takes advantage of high irradiation by deploying the C 4 pathway. This adaptive and multifaceted carbon acquisition strategy in U. prolifera is obviously an advantageous biological approach to support fast biomass accumulation and improve oxygen production in the filaments to sustain respiration for algae while floating in thick mats 36 . The relative contribution of the C 4 pathway and CA mechanism in utilizing the HCO 3 − system and the environmental thresholds that determine which of these pathways dominate at different stages of bloom development, maturity, and decline need to be further elucidated.
Ulva prolifera blooms are initiated from the biomass of thalli dislodged in the intertidal zone where the alga is subject to very high diurnal ranges in temperature and light and can tolerate partial desiccation 37 . Evolutionarily, the expression of C 4 photosynthesis in U. prolifera maybe an adaptive response that enables the species to take advantage of rapidly changing conditions, and which also leads to formation of unusual massive blooms. The ability to harness the benefit of C 4 photosynthesis might also be a feature of other types of blooming macroalga and warrants investigation.

Methods
Culture experiments. Floating specimens of Ulva prolifera were collected from the Yellow Sea and, after cleaning epiphytes off using sterilized seawater, the sampled thalli were cultured in a laboratory incubator for a week prior to the outdoor experiments. Fifty grams of fresh thalli were placed into each of three replicate transparent plastic containers (53.5 cm × 39 cm × 32.5 cm), with 40 L of filtered seawater. To maintain ambient seawater temperatures in the containers, they were suspended in an outdoor seawater pond (60 × 100 m) at Muping Coastal Station of Chinese Academy of Sciences, China. Each experiment lasted 1 day from 08:00 to 18:00, without stirring. The first two outdoor culture experiments were carried out in July 2018, including a sunny day culture (experiment 1) and a cloudy day culture (experiment 2). The two experiments were to examine the responses of major enzyme activities of the C 3 and C 4 pathways (Rubisco, PEPCase, PEPCKase, and PPDKase) to diurnal variation of light intensity. The details are given below.
Algal sample was collected every 2 h from each of the three culture containers and stored in a liquid nitrogen tank prior to assays of enzyme activities. Light intensity, salinity, and air temperature were monitored during the culture using a light meter (TES-1339R, TES Electrical Electronic Corp., Taiwan A third outdoor experiment (again with three replicate culture containers as described above) was conducted in August 2019 to define the correlation between major enzyme activities and corresponding photosynthetic products. This experiment was carried out on a sunny day, from 08:00 to 18:00. Samples were collected from each of the three culture containers every 2 h and stored in liquid nitrogen for the assays of enzyme activities and tissue δ 13 C. Meanwhile, the changes of pCO 2 and HCO 3 − were measured in the container seawater, and light, salinity, and air temperature were monitored through the day. The initial concentrations of DIN and DIP in the culture medium were 30.9 and 0.16 μM, respectively. During the third experiment, salinity varied from 33.1 to 33.8, water temperature ranged from 26.2 to 29.6°C, and sunlight intensity from 50.4 to 1738 μmol photons m −2 s −1 .
Measurement of the pCO 2 and HCO 3 − concentrations. Twenty-five milliliters of water samples were taken from each of the three culture containers at each 2-h time interval and transferred to a fifty milliliters stoppered polyethylene bottle containing ten milliliters of a hydrochloric acid (HCl) standard solution (0.006 mol L −1 ). After the pH in the water sample was stabilized, HCl standard solution was added into the sample to adjust the pH between 3.40 and 3.90. The added HCl volumes and pH values were recorded and total alkalinity calculated as where A represents the alkalinity of the sample (mmol L −1 ); c HCl represents HCl standard solution (mol L −1 ); V W represents water sample volume (mL); V HCl represents added HCl volume (mL); a H þ represents the hydrogen ion activity corresponding to the solution pH; and f H þ represents the hydrogen ion activity corresponding to the pH and actual salinity of the solution. All data, including temperature and salinity, were input into CO2SYS 38 , which was used to calculate the pCO 2 and HCO 3 − concentrations.
Enzyme assays. The frozen algal samples were ground with liquid nitrogen, with 0.1 g of the ground sample used for enzyme assays. The activities of Rubisco, PEPCase, and PPDKase were assayed using enzyme assay kits provided by Solarbio Life Sciences, China. The activity of PEPCKase was assayed using enzyme assay kits provided by Nanjing Jiancheng Bioengineering Institute, China. One unit of Rubisco, PEPCase, and PEPCKase activity is defined as the amount of enzyme consuming 1 nmol NADH per minute. One unit of PPDKase activity is defined as the amount of enzyme consuming 1 nmol NADPH per minute.
CA activity was assayed according to the method of Wilbur and Anderson 39 . Prior to storage in liquid nitrogen,~0.1-0.14 g of fresh algal thallus (algal weight) were weighed and then soaked in 2 mL of barbiturate buffer (20 mmol/L, pH = 8.4). Five milliliters of CO 2 -saturated water was injected to initiate the reaction. The reaction mixture was cooled in a 4°C-water bath. The extracellular CA activity was assayed by measuring the time of catalyzed reaction required for the pH to change from 8.3 to 7.3 (T c ). The same procedure was repeated using 2 mL of barbiturate buffer without algal thallus and the uncatalyzed time required for background pH change from 8.3 to 7.3 was recorded (T 0 ). The extracellular CA activity was calculated according to the following formula: Enzyme activity (U/g) = ((T 0 /T c − 1) × 10)/algal weight.
Then,~0.03-0.04 g of algal thallus was carefully ground in liquid nitrogen and soaked with 2 mL of barbiturate buffer (20 mmol/L, pH = 8.4) for the assay of total CA activity. The total CA activity was measured and calculated using the same method and formula as for extracellular CA activity. The intracellular CA activity was calculated by total CA activity minus extracellular CA activity.
Tissue δ 13 C measurement. After rinsing with 0.1 M HCl and washing with Milli-Q water, algal samples were freeze-dried and ground. Approximately 0.5-1 mg ground sample was placed into a 4 × 6 mm tin capsule for tissue δ 13 C analysis. The samples were analyzed using an isotope-ratio mass spectrometer (MAT 253, Thermo Scientific, USA) at the Yantai Institute of Coastal Zone Research, Chinese Academy of Sciences. Reference gases were calibrated against International Reference Materials (IAEA-CH6, IAEA-600 and EMA-P1). Results are expressed relative to Vienna PeeDee Belemnite. Replicate measurements of a laboratory standard (acetanilide, Thermo Scientific) analyzed with the samples indicated that analytical errors were <0.1‰ for δ 13 C.
Statistics and reproducibility. Linear regression analysis was carried out to test the dependence of enzyme activity on irradiance levels using data from both the sunny and cloudy day experiments (experiments 1 and 2). Pearsonʼs R 2 was used to determine the significance of the results. PEPCKase activity was transformed (χ 2 ) to ensure normality of residuals for the linear regression (Shapiro-Wilk test). Other variables did not require transformation. Two-way analyses of variance (ANOVA, factors = time of day and light condition) was carried out on data from experiments 1 and 2. These analyses compared the level of enzyme activity under sunny and cloudy conditions and between different time periods (three replicates at each time period). Comparisons for PEPCKase were made across all time periods from 08:00 to 18:00, and for Rubisco and PEPCase for time periods from 08:00 to 16:00. Bartletts test was used to check for homogeneity of variances. A square root transformation was required to correct the variance structure for Rubisco and PEPCase. The 18:00 time period was excluded for Rubisco and PEPCase as variance structure could not be corrected if it was included. Tukey's HSD was used for the pairwise comparisons. Correlation analyses were conducted on parameters measured in the third experiment for the time period from 08:00 to 14:00 (also three replicates) in order to examine the relationships between the activities of key enzymes and photosynthetic products to ascertain the activity of C 4 and/or CA pathways in U. prolifera. All analyses were undertaken using Addinsoft's statistical software XLSTAT.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data supporting the findings of this study and source data of main figures are provided as Supplementary Data. Received: 28 January 2020; Accepted: 13 August 2020;