Metabolomics Approach to Reveal the Effects of Ocean Acidification on the Toxicity of Harmful Microalgae: A Review of the Literature

Abstract

Climate change has been associated with intensified harmful algal blooms (HABs). Some harmful microalgae produce toxins that accumulate in food webs, adversely affecting the environment, public health and economy. Ocean acidification (OA) is a major consequence of high anthropogenic CO₂ emissions. The carbon chemistry and pH of aquatic ecosystems have been significantly altered as a result. The impacts of climate change on the metabolisms of microalgae, especially toxin biosynthesis, remain largely unknown. This hinders the optimization of HAB mitigation for changed climate conditions. To bridge this knowledge gap, previous studies on the effects of ocean acidification on toxin biosynthesis in microalgae were reviewed. There was no solid conclusion for the toxicity change of saxitoxin-producing dinoflagellates from the genus Alexandrium after high CO₂ treatment. Increased domoic acid content was observed in the diatom Pseudo-nitzschia. The brevetoxin content of Karenia brevis remained largely unchanged. The underlying regulatory mechanisms that account for the different toxicity levels observed have not been elucidated. Metabolic flux analysis is useful for investigating the carbon allocations of toxic microalgae under OA and revealing related metabolic pathways for toxin biosynthesis. Gaining knowledge of the responses of microalgae in high CO₂ conditions will allow the better risk assessment of HABs in the future.

1. Introduction

Harmful algal blooms (HABs) refer to the rapid growth of phytoplankton, including cyanobacteria, dinoflagellates, raphidophytes, haptophytes, and macroalgae, which exerts harmful effects on the environment, human health, and the economy [1]. High microalgal biomass accumulation during HABs can cause hypoxia, suffocating surrounding aquatic organisms [2]. Some HAB species can produce algal toxins that accumulate in aquatic food webs [3]. Apart from causing the massive death of marine mammals and fish [4], algal toxins can lead to intoxication in humans. Approximately 50,000 to 500,000 cases of intoxication with a 1.5% mortality rate have been reported annually due to the consumption of contaminated shellfish or fish [5]. The health costs associated with HABs are substantial, ranging from around USD 90 to USD 12,000 for digestive and respiratory illnesses with different severities [6]. The aquacultural industry is the most vulnerable when considering the direct economic losses caused by HABs. From the 1980s to the 2010s, the aquacultural industry in Korea lost a total of USD 121 million due to HAB-induced fish and shellfish deaths [7]. In addition to the aquacultural industry, tourism is adversely affected by HABs. It was reported that the monthly revenue of coastal lodgings and restaurants decreased by USD 2.8 and 3.7 million, respectively, in Fort Walton Beach and Destin in Florida when HABs occurred [8]. To mitigate the negative effects of HABs, costly monitoring programs and mitigation measures are conducted in different regions, such as the US, Europe, and Australia, as well as by intergovernmental organizations [9].

The direct link between HABs and climate change was first affirmed in the Special Report on the Ocean and Cryosphere in a Changing Climate, published by the Intergovernmental Panel on Climate Change in 2019 [1]. Climate change has profoundly changed the aquatic environment over the years by increasing the surface temperature, acidifying water bodies, shifting nutrient availability, altering salinity, etc. Several studies have reviewed the association between climate change and intensified HABs [10,11,12,13]. In parallel with the spatial expansions and frequencies of HABs, the frequency and distribution of human intoxication by algal toxins have also increased globally [4]. According to the Harmful Algae Event Database (HAEDAT), 1598 HAB-related seafood poisoning cases were reported during the 20th century. A nearly five-fold increase in the number of cases (7843) was observed from 2001 to 2022. Intensified HABs can also further alter the food webs and disturb the ecosystem. The physiology, mortality, and toxicity of marine organisms at higher trophic levels were found to have changed when co-exposed to algal toxins and climate change stressors, as summarized by Griffith and Gobler [14]. Given the above-mentioned scenario, it is believed that the adverse effects of HABs will be aggravated in the future.

2. Algal Toxins

Intensified HABs of toxic microalgae may increase the algal toxins in aquacultures, subsequently raising the risks of seafood poisoning [15]. Consumption of contaminated seafood can lead to different poisoning syndromes, depending on the types of algal toxins accumulated in shellfish during HABs. As of December 2019, the incidence of paralytic shellfish poisoning (PSP) and diarrhetic shellfish poisoning (DSP) accounted for approximately one third of the reported cases of human intoxication, with PSP contributing 35% and DSP contributing 30% [16]. According to the data, the incidence of amnesic shellfish poisoning (ASP) was relatively low, at 9%, whereas both neurotoxic shellfish poisoning (NSP) and azaspiracid shellfish poisoning (AZP) each accounted for only 1% of cases.

2.1. Paralytic Shellfish Poisoning

The causative agents of PSP, saxitoxin (STX) and its 57 analogues, are collectively known as paralytic shellfish toxins (PSTs) [17]. STX is composed of a tricyclic 3,4-propinoperhydropurine backbone with two guanidine groups [18]. The analogues have been classified into different groups according to the side group moieties. PSTs are mainly produced by eukaryotic marine dinoflagellates from the genera Alexandrium, Gymnodinium, and Pyrodinium [19,20]. Prokaryotic freshwater cyanobacteria from the genera Anabaena, Cylindrospermopsis, Aphanizomenon, Planktothrix, and Lyngbya are producers of PSTs as well [21,22,23,24]. The toxicity levels of these analogues vary depending on the structures. Generally, the toxicity levels of STXs are inversely proportional to the degree of sulfation [17]. PSTs selectively and reversibly bind to receptor site 1 of the voltage-gated sodium channels in the nerves and muscles [25]. Due to the blockage of the channels, the propagation of action potential is terminated, leading to symptoms such as paralysis, burning sensations, and numbness [26]. In serious cases, death can result due to respiratory failure [27]. Most reported PSP cases were contributed by the toxic Alexandrium species, which are mainly present in Northern Europe, the Mediterranean, Northern Asia, East Asia, and North America [28,29,30,31]. Recently, the geographical distribution of Alexandrium has been expanded towards the poles [32].

2.2. Diarrhetic Shellfish Poisoning

Okadaic acid and its derivatives, the dinophysistoxins (DTXs), are lipophilic polyketides predominately produced by toxic dinoflagellates from the genera Prorocentrum and Dynophysis [33]. For human adults, DSP symptoms occur when a minimum dosage of 40 µg okadaic acid equivalents is consumed [34]. According to the Food and Agriculture Organization (FAO), the recommended okadaic equivalents for okadaic acid, DTX1, and DTX2 are 1.0, 1.0, and 0.5, respectively [35]. Diarrhea, nausea, vomiting, and abdominal pain are the common symptoms of DSP [36]. DSP is generally not life-threatening, and hospitalization is not required [37]. As potent inhibitors of serine/threonine phosphatases [38], okadaic acids show specifically high binding affinity to protein phosphatase 1 (PP1) and protein phosphatase 2 (PP2) [39]. The gastrointestinal symptoms of DSP may be a result of inhibited intestinal PP activities by okadaic acids [40]. In addition, the inhibiting effect of okadaic acids may induce the hyperphosphorylation of proteins that regulate the sodium secretion of intestinal cells [41]. The release of sodium from cells disturbs the osmotic gradient balance and subsequently causes the passive loss of fluids, leading to diarrhea. Both Prorocentrum and Dynophysis have expanded their niches in recent years. P. minimum is the most studied Prorocentrum species and is widespread in the Black Sea, the Baltic Sea, Lake Nakanoumi, the Mexican coast, and the Philippines [42]. The expansion of Dynophysis was observed in the coastal areas of the United States, Canada, and South Africa [13]. Moreover, increased abundances of Prorocentrum and Dynophysis in the Northeast Atlantic were reported [43].

2.3. Amnesic Shellfish Poisoning

ASP is caused by the consumption of domoic acid (DA)-accumulated shellfish. Red algae such as Chondria armata and diatoms from the genera Pseudo-nitzschia and Nitzchia (N. navis-varingica and N. bizertensis) are the primary producers of DA [44,45]. DA, a tricarboxylic amino acid, is analogous to glutamic acid and kainic acid [46]. Several DA isomers (isodomic acid A-H, 5′ epi-DA) have been identified [47]. With such structural similarity, Das can bind and activate three ionotropic glutamate receptor subtypes (N-methyl-D-aspartate (NMDA), kainate, and α-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA) receptors) located on the dendrites of postsynaptic cells [48,49]. An influx of calcium ions (Ca²⁺) into neurons is induced upon the activation of kainate and AMPA receptors [50], while an influx of both Ca²⁺ and sodium ions (Na⁺) is induced upon the activation of the NMDA receptor [51]. Desensitization is prevented by the low conformational mobility of DAs when docked to the receptors, leading to a continuous flow of cations to the postsynaptic cells [52]. DAs exert excitotoxic effects primarily through excess intracellular Ca²⁺, which triggers the mobilization of glutamate-containing vesicles towards the membrane surface, followed by the release of glutamate (Glu) into the synaptic cleft [53]. Excess Glu results in neurodegeneration and apoptosis [54,55]. Since neurons in the hippocampus, where the consolidation of memories takes place, are affected by DAs [56], symptoms such as short-term memory loss and anterograde amnesia may result [57]. Generally, the toxicity of DA isomers is lower than that of DA because of the lower binding affinity of DA isomers to glutamate receptors [58,59,60]. One of the major DA producers, Pseudo-nitzschia, has been detected in the Pacific Ocean and the Southern Ocean [61]. Expanded distributions of Pseudo-nitzschia and N. navis-varingica have been observed recently [44]. Several locations, including the Arctic, Angola, Singapore, Ukraine, and Pakistan, have recorded the presence of Pseudo-nitzschia for the first time, while N. navis-varingica has expanded to Malaysia, Australia, Indonesia, and the Philippines.

2.4. Neurotoxic Shellfish Poisoning

NSP is caused by the consumption of shellfish contaminated with brevetoxins (PbTx). PbTx are a set of cyclic polyether neurotoxins primarily produced by dinoflagellate Karenia brevis [62]. Besides K. brevis, recent research indicated PbTx production by raphidophyte Chattonella cf. verruculosa [63]. PbTx-1 and PbTx-2 are regarded as the parent molecules for other derivatives based on their different structural backbones. The toxicity of all identified natural derivatives is lower than that of the parent molecules [64]. PbTx bind specifically to receptor site five of the voltage-gated sodium channels in the nerves and muscles, with a preference for those located in the nerves [65,66]. The influx of Na⁺ ions into the cells is induced upon the activation of the sodium channels. Prolonged membrane depolarization persistently initiates the generation of action potentials in nerves and muscles, causing symptoms such as paresthesia, vertigo, and malaise [67]. Blooms of K. brevis occur nearly annually along the southwestern coast of Florida, which is notoriously known as the “Florida red tide” [68]. The low incident rate of NSP may be credited to the successful monitoring program of constantly occurring red tides. Apart from Florida, where regular red tides take place, a few large outbreaks of NSP have been reported in New Zealand and North Carolina [69].

2.5. Azaspiracid Shellfish Poisoning

Azaspiracids (AZAs) are polyether phytotoxins produced by some dinoflagellate species from the genera Azadinium and Amphidoma and the causative toxin of AZP [70]. AZA was named based on its unique spiro ring assemblies [71]. To date, over 60 AZA analogues have been identified [72]. They are different in the degrees of methylation and/or the number of hydroxyl groups and carboxyl groups that they possess. However, among the numerous analogues, only AZA1, AZA2, and AZA3 levels are monitored for regulatory purposes [33]. The symptoms of AZP are similar to those of DSP, including diarrhea, vomiting, nausea, and stomach cramps [73]. A minimum dosage ranging from 23 to 86 µg/person of AZA can have observable adverse health effects [74]. Unlike other phytotoxins, the molecular target(s) of AZAs has not been fully understood. Although the blocking of the hERG (human ether-à-go-go related gene) potassium channel by AZA1-3 was demonstrated in a recent study, relatively high concentrations of AZAs were required [75]. Therefore, there may be other molecular targets that have not been identified. The presence of toxigenic Azadinium species has been reported worldwide, including in the North Atlantic [76,77], Eastern South Atlantic [78], Mediterranean [79], Western Pacific [80], Eastern North Pacific [81], and Eastern South Pacific [82]. On the other hand, the occurrence of toxigenic Amphidoma species was only recorded in the North Atlantic [77,83,84].

3. Ocean Acidification

Ocean acidification (OA) is one of the major aspects of climate change and has indeed been significantly influencing the aquatic environment, where toxigenic microalgae live. Carbon dioxide (CO₂) emissions have increased dramatically since the Industrial Revolution, primarily due to the combustion of fossil fuels. At present, the recorded atmospheric CO₂ level is 417 ppm [85], which has increased by nearly 50% compared to the pre-industrial level. The atmospheric CO₂ concentrations will continue to rise until the end of this century unless a very stringent CO₂ emissions trajectory is satisfied (Representative Concentration Pathway 2.6) [86]. A large amount of atmospheric CO₂ traps and prevents heat from escaping the planet, driving global warming and subsequent climate change. As a natural carbon sink, the ocean has absorbed around 48% of anthropogenic CO₂ since the Industrial Revolution [87]. Given the elevated atmospheric CO₂ levels, more CO₂ has dissolved in the ocean. The dissolution of atmospheric CO₂ produces carbonic acid (H₂CO₃), which dissociates to form bicarbonate ions (HCO₃⁻) and hydrogen ions (H⁺). HCO₃⁻ further dissociates into carbonate ions (CO₃²⁻) and H⁺. The rise in the concentration of H⁺ in turn has acidified the ocean, resulting in OA. OA has altered the carbonate chemistry of the aquatic system. Excess H⁺ released from the dissociation of dissolved CO₂ creates an imbalance in the carbonate equilibrium. The equilibrium is attained by the natural buffering capacity of the ocean. The free CO₃²⁻ in the ocean binds with the excess H⁺ to produce more HCO³⁻. As a result, the dissolved CO₂ and HCO³⁻ concentrations increase while the CO₃²⁻ levels decrease under OA. Since the decrease in CO₃²⁻ levels leads to a lower saturation state of CaCO₃, OA is particularly detrimental to marine-calcifying organisms such as molluscs, coccolithophores, and corals [88].

OA (increased pCO₂/decreased pH) alters the carbon chemistry of the aquatic environment and acidifies the water. Microalgae have shown great adaptability towards different abiotic stress factors [89,90]. Therefore, it is believed that microalgae will respond and acclimate to OA by regulating their metabolic activities, which may in turn affect their toxicity. However, the effects of OA on microalgal toxicity and the underlying mechanisms have not been well characterized. Apart from leading to shellfish poisoning in humans, several studies have indicated the bioaccumulation of algal toxins in predators of microalgae and the biomagnification of algal toxins through food webs, which cause disease or even death among marine organisms [91,92,93,94,95]. A recent study suggested that the bioavailability of STX would increase under global warming and ocean acidification [96]. Although the bioavailability of STX and other phycotoxins in high CO₂ conditions has not been fully examined, the possibility that more toxins can be accumulated along the food chain and harm the health of marine organisms and humans should not be neglected [96]. Predicting the toxicity of microalgae under OA is therefore important to evaluate the public health concerns and ecological impacts of intensified HABs in future high atmospheric CO₂ scenarios, which will provide insights for policymakers to improve the monitoring programs for HABs. Considering the above, this paper reviewed the effects of OA on the toxicity of microalgae, and the underlying mechanisms proposed.

3.1. Effects of OA on the Toxicity of Microalgae

No studies that investigated the effects of OA on the toxicity changes in microalgae that produce okadaic acid and azaspiracids could be found in the literature. Therefore, the scope of the review was limited to microalgae that synthesize STXs, DA, and PbTx. The experimental setups and significant findings of the reviewed studies are summarized in

Table 1. Summary of reviewed studies on the effects of OA on toxigenic microalgae.

Algal Species (Strains)	OA Modeling Method	Experimental Parameters a		Significant Outcomes b	Reference
		Control	High pCO2 Treatment
STXs-Producing Microalgae
Alexandrium fundyense (NBP8) isolated from Northport Bay	Laboratory culture CO2 gas bubbling	Experiment No. 1 pH: 8.085 ± 0.009 AT: 1824 ± 87 μmol L−1 Calculated pCO2: 41 ± 2 Pa DIC: 1600 ± 80 μmol L−1 Length of experiment: 15 days	Experiment No. 1 pH: 7.629 ± 0.031 AT: 1678 ± 123 μmol L−1 Calculated pCO2: 122 ± 6 Pa DIC: 1606 ± 116 μmol L−1 Length of experiment: 15 days	Experiment No. 1 Growth rate increased *	[97]
		Experiment No. 2 pH: 8.068 ± 0.013 AT: 1858 ± 104 μmol L−1 Calculated pCO2: 44 ± 4 Pa DIC: 1639 ± 100 μmol L−1 Length of experiment: 27 days	Experiment No. 2 pH: 7.741 ± 0.007 AT: 1975 ± 128 μmol L−1 Calculated pCO2: 110 ± 7 Pa DIC: 1868 ± 125 μmol L−1 Length of experiment: 27 days	Experiment No. 2 Total cellular toxicity increased * GTX1/4 increased *	[97]
		Experiment No. 3 pH: 8.075 ± 0.057 AT: 1994 ± 355 μmol L−1 Calculated pCO2: 46 ± 2 Pa DIC: 1790 ± 329 μmol L−1 Length of experiment: 15 days	Experiment No. 3 pH: 7.545 ± 0.034 AT: 1966 ± 231 μmol L−1 Calculated pCO2: 177 ± 9 Pa DIC: 1941 ± 243 μmol L−1 Length of experiment: 15 days	Experiment No. 3 Growth rate increased *	[97]
		Experiment No. 4 pH: 8.096 ± 0.032 AT: 2122 ± 123 μmol L−1 Calculated pCO2: 47 ± 2 Pa DIC: 1885 ± 102 μmol L−1 Length of experiment: 12 days	Experiment No. 4 pH: 7.592 ± 0.022 AT: 1977 ± 240 μmol L−1 Calculated pCO2: 162 ± 12 Pa DIC: 1977 ± 240 μmol L−1 Length of experiment: 12 days	Experiment No. 4 Growth rate increased * Total cellular toxicity increased * GTX1/4 increased *	[97]
Alexandrium fundyense (CCMP2304) isolated from the Bay of Fundy		Experiment No. 5 pH: 8.118 ± 0.008 AT: 2167 ± 53 μmol L−1 Calculated pCO2: 45 ± 1 Pa DIC: 1912 ± 44 μmol L−1 Length of experiment: 24 days	Experiment No. 5 pH: 7.873 ± 0.033 AT: 2186 ± 91 μmol L−1 Calculated pCO2: 87 ± 4 Pa DIC: 2033 ± 71 μmol L−1 Length of experiment: 24 days	Experiment No. 5 Growth rate: increased *	[97]
		Experiment No. 6 pH: 8.041 ± 0.012 AT: 1729 ± 56 μmol L−1 Calculated pCO2: 44 ± 1 Pa DIC: 1539 ± 49 μmol L−1 Length of experiment: 12 days	Experiment No. 6 pH: 7.547 ± 0.028 AT: 1889 ± 90 μmol L−1 Calculated pCO2: 169 ± 4 Pa DIC: 1845 ± 84 μmol L−1 Length of experiment: 12 days	Experiment No. 6 Growth rate increased *	[97]
		Experiment No. 7 pH: 8.086 ± 0.048 AT: 2121 ± 213 μmol L−1 Calculated pCO2: 48 ± 3 Pa DIC: 1888 ± 183 μmol L−1 Length of experiment: 12 days	Experiment No. 7 pH: 7.556 ± 0.038 AT: 2184 ± 167 μmol L−1 Calculated pCO2: 191 ± 4 Pa DIC: 2138 ± 157 μmol L−1 Length of experiment: 12 days	Experiment No. 7 Growth rate increased *	[97]
Alexandrium minutum (AM-1) isolated from the South China Sea	Laboratory culture CO2 gas bubbling	400 ppm CO2 treatment Length of experiment: 37 days	800 ppm CO2 treatment Length of experiment: 37 days	400 vs. 800 ppm CO2 Total cellular toxicity increased **	[98]
		400 ppm CO2 treatment Length of experiment: 37 days	1200 ppm CO2 treatment Length of experiment: 37 days	400 vs. 1200 ppm CO2 Total cellular toxicity increased **	[98]
Alexandrium catenella (A-11c) isolated from Jalama Beach	Laboratory culture CO2 gas bubbling	380 µatm CO2 treatment pH: 8.169 Calculated pCO2: 285 µatm DIC: 1957 μmol L−1 Length of experiment: 7 days	750 µatm CO2 treatment pH: 7.925 Calculated pCO2: 571 µatm DIC: 2103 μmol L−1 Length of experiment: 7 days	380 vs. 750 µatm CO2 Growth rate increased *** Total cellular toxicity increased ***	[99]
Alexandrium ostenfeldii (AON13) isolated from Ouwerkerkse Kreek	Laboratory culture CO2 gas bubbling prior to cell inoculation c	400 µatm CO2 treatment pH: 8.23 ± 0.02 Calculated pCO2: 213 ± 11 µatm DIC: 1188 ± 41 μmol L−1 Length of experiment: 14 days	1000 µatm CO2 treatment pH: 7.82 ± 0.01 Calculated pCO2: 527 ± 30 µatm DIC: 1168 ± 56 μmol L−1 Length of experiment: 14 days	400 vs. 1000 µatm CO2 STX increased *	[100]
Alexandrium ostenfeldii (AON15) isolated from Ouwerkerkse Kreek		400 µatm CO2 treatment pH: 8.07 ± 0.07 Calculated pCO2: 357 ± 36 µatm DIC: 1120 ± 21 μmol L−1 Length of experiment: 14 days	1000 µatm CO2 treatment pH: 7.75 ± 0.08 Calculated pCO2: 676 ± 55 µatm DIC: 1117 ± 10 μmol L−1 Length of experiment: 14 days	400 vs. 1000 µatm CO2 Growth rate increased ** Total cellular toxicity decreased *** GTX increased *	[100]
Alexandrium ostenfeldii (AON5.26) isolated from Ouwerkerkse Kreek		400 µatm CO2 treatment pH: 8.17 ± 0.05 Calculated pCO2: 286 ± 13 µatm DIC: 1085 ± 15 μmol L−1 Length of experiment: 14 days	1000 µatm CO2 treatment pH: 7.86 ± 0.03 Calculated pCO2: 479 ± 18 µatm DIC: 1126 ± 2 μmol L−1 Length of experiment: 14 days	Growth rate increased *** GTX increased *** STX and C1/2 decreased ***	[100]
Alexandrium tamarense (Alex5)	Laboratory culture CO2 gas bubbling	380 µatm CO2 treatment pH: 8.27 ± 0.07 AT: 2439 ± 1 μmol L−1 Calculated pCO2: 315 ± 57 µatm DIC: 2117 ± 41 μmol L−1 Length of experiment: 8 days	800 µatm CO2 treatment pH: 7.97 ± 0.10 AT: 2434 ± 2 μmol L−1 Calculated pCO2: 706 ± 154 µatm DIC: 2245 ± 37 μmol L−1 Length of experiment: 8 days	380 vs. 800 µatm CO2 Total cellular toxicity decreased * Non-sulfated STX analogues decreased * Di-sulfated STX analogues increased *	[101]
		380 µatm CO2 treatment pH: 8.27 ± 0.07 AT: 2439 ± 1 μmol L−1 Calculated pCO2: 315 ± 57 µatm DIC: 2117 ± 41 μmol L−1 Length of experiment: 8 days	1200 µatm CO2 treatment pH: 7.83 ± 0.12 AT: 2418 ± 1 μmol L−1 Calculated pCO2: 995 ± 248 µatm DIC: 2283 ± 34 μmol L−1 Length of experiment: 8 days	380 vs. 1200 µatm CO2 Total cellular toxicity decreased * Non-sulfated STX analogues decreased * Mono-sulfated STX analogues increased * Di-sulfated STX analogues increased *	[101]
Alexandrium tamarense (Alex2)		380 µatm CO2 treatment pH: 8.27 ± 0.02 AT: 2384 ± 4 μmol L−1 Calculated pCO2: 305 ± 20 µatm DIC: 2111 ± 14 μmol L−1 Length of experiment: 8 days	800 µatm CO2 treatment pH: 7.90 ± 0.03 AT: 2390 ± 6 μmol L−1 Calculated pCO2: 810 ± 60 µatm DIC: 2229 ± 33 μmol L−1 Length of experiment: 8 days	380 vs. 800 µatm CO2 No significant changes in growth rate, total cellular toxicity, and toxin profile	[101]
		380 µatm CO2 treatment pH: 8.27 ± 0.02 AT: 2384 ± 4 μmol L−1 Calculated pCO2: 305 ± 20 µatm DIC: 2111 ± 14 μmol L−1 Length of experiment: 8 days	1200 µatm CO2 treatment pH: 7.75 ± 0.04 AT: 2386 ± 9 μmol L−1 Calculated pCO2: 1167 ± 112 µatm DIC: 2279 ± 14 μmol L−1 Length of experiment: 8 days	380 vs. 1200 µatm CO2 No significant changes in growth rate, total cellular toxicity, and toxin profile	[101]
DA-producing microalgae
Pseudo-nitzschia multiseries (CCMP2708) isolated from Eastern Canada	Laboratory culture CO2 gas bubbling	22 Pa CO2 treatment pH: 8.40 ± 0.03 Calculated pCO2: 22 ± 2 Pa DIC: 1970 ± 4 μmol L−1 Length of experiment: around 4 to 6 weeks	41 Pa CO2 treatment pH: 8.19 ± 0.02 Calculated pCO2: 40 ± 3 Pa DIC: 2066 ± 11 μmol L−1 Length of experiment: around 4 to 6 weeks	22 vs. 41 Pa CO2 Growth rate increased *** Carbon fixation rate increased **	[102]
		22 Pa CO2 treatment pH: 8.40 ± 0.03 Calculated pCO2: 22 ± 2 Pa DIC: 1970 ± 4 μmol L−1 Length of experiment: around 4 to 6 weeks	74 Pa CO2 treatment pH: 7.96 ± 0.01 Calculated pCO2: 73 ± 1 Pa DIC: 2177 ± 6 μmol L−1 Length of experiment: around 4 to 6 weeks	22 vs. 74 Pa CO2 Growth rate increased *** Carbon fixation rate increased *** Cellular DA content increased ***	[102]
		41 Pa CO2 treatment pH: 8.19 ± 0.02 Calculated pCO2: 40 ± 3 Pa DIC: 2066 ± 11 μmol L−1 Length of experiment: around 4 to 6 weeks	74 Pa CO2 treatment pH: 7.96 ± 0.01 Calculated pCO2: 73 ± 1 Pa DIC: 2177 ± 6 μmol L−1 Length of experiment: around 4 to 6 weeks	41 vs. 74 Pa CO2 Growth rate increased *** Cellular DA content increased **	[102]
Pseudo-nitzschia fraudulenta isolated from Ventura County	Laboratory culture CO2 gas bubbling	200 ppm CO2 treatment pH: 8.43 Calculated pCO2: 198 ppm DIC: 1965 μmol L−1	360 ppm CO2 treatment pH: 8.23 Calculated pCO2: 357 ppm DIC: 2107 μmol L−1	200 vs. 360 ppm CO2 Growth rate increased *** Carbon fixation rate increased	[103]
		200 ppm CO2 treatment pH: 8.43 Calculated pCO2: 198 ppm DIC: 1965 μmol L−1	765 ppm CO2 treatment pH: 7.95 Calculated pCO2: 764 ppm DIC: 2249 μmol L−1	200 vs. 765 ppm CO2 Growth rate increased *** Carbon fixation rate increased	[103]
Pseudo-nitzschia australis (HAB 200) isolated from northern Monterey Bay	Laboratory culture CO2 gas bubbling	pH 8.1 treatment Length of experiment: 7 days Exponential growth phase pH: 8.14 ± 0.01 AT: 2232 ± 9 μmol L−1 Calculated pCO2: 406 ± 8 µatm DIC: 2032 ± 11 μmol L−1 Stationary growth phase pH: 8.14 ± 0.01 AT: 2239 ± 25 μmol L−1 Calculated pCO2: 410 ± 2 µatm DIC: 2052 ± 21 μmol L−1	pH 8.0 treatment Length of experiment: 8 days Exponential growth phase pH: 8.03 ± 0.01 AT: 2248 ± 25 μmol L−1 Calculated pCO2: 550 ± 8 µatm DIC: 2093 ± 21 μmol L−1 Stationary growth phase pH: 8.04 ± 0.01 AT: 2254 ± 14 μmol L−1 Calculated pCO2: 537 ± 19 µatm DIC: 2107 ± 17 μmol L−1	pH 8.1 vs. pH 8.0 No significant response	[104]
		pH 8.1 treatment Length of experiment: 7 days Exponential growth phase pH: 8.14 ± 0.01 AT: 2232 ± 9 μmol L−1 Calculated pCO2: 406 ± 8 µatm DIC: 2032 ± 11 μmol L−1 Stationary growth phase pH: 8.14 ± 0.01 AT: 2239 ± 25 μmol L−1 Calculated pCO2: 410 ± 2 µatm DIC: 2052 ± 21 μmol L−1	pH 7.9 treatment Length of experiment: 8 days Exponential growth phase pH: 7.93 ± 0.02 AT: 2255 ± 20 μmol L−1 Calculated pCO2: 719 ± 34 µatm DIC: 2136 ± 25 μmol L−1 Stationary growth phase pH: 7.93 ± 0.02 AT: 2279 ± 14 μmol L−1 Calculated pCO2: 732 ± 28 µatm DIC: 2175 ± 13 μmol L−1	pH 8.1 vs. pH 7.9 Maximum carbon fixation rate increased **	[104]
		pH 8.0 treatment Length of experiment: 8 days Exponential growth phase pH: 8.03 ± 0.01 AT: 2248 ± 25 μmol L−1 Calculated pCO2: 550 ± 8 µatm DIC: 2093 ± 21 μmol L−1 Stationary growth phase pH: 8.04 ± 0.01 AT: 2254 ± 14 μmol L−1 Calculated pCO2: 537 ± 19 µatm DIC: 2107 ± 17 μmol L−1	pH 7.9 treatment Length of experiment: 8 days Exponential growth phase pH: 7.93 ± 0.02 AT: 2255 ± 20 μmol L−1 Calculated pCO2: 719 ± 34 µatm DIC: 2136 ± 25 μmol L−1 Stationary growth phase pH: 7.93 ± 0.02 AT: 2279 ± 14 μmol L−1 Calculated pCO2: 732 ± 28 µatm DIC: 2175 ± 13 μmol L−1	pH 8.0 vs. pH 7.9 Maximum carbon fixation rate increased *	[104]
		pH 8.1 treatment Length of experiment: 7 days Exponential growth phase pH: 8.14 ± 0.01 AT: 2232 ± 9 μmol L−1 Calculated pCO2: 406 ± 8 µatm DIC: 2032 ± 11 μmol L−1 Stationary growth phase pH: 8.14 ± 0.01 AT: 2239 ± 25 μmol L−1 Calculated pCO2: 410 ± 2 µatm DIC: 2052 ± 21 μmol L−1	pH 7.8 treatment Length of experiment: 9 days Exponential growth phase pH: 7.81 ± 0.03 AT: 2292 ± 3 μmol L−1 Calculated pCO2: 980 ± 64 µatm DIC: 2213 ± 11 μmol L−1 Stationary growth phase pH: 7.84 ± 0.01 AT: 2348 ± 17 μmol L−1 Calculated pCO2: 929 ± 12 µatm DIC: 2271 ± 16 μmol L−1	pH 8.1 vs. pH 7.8 Growth rate decreased **** Maximum carbon fixation rate increased *	[104]
		pH 8.0 treatment Length of experiment: 8 days Exponential growth phase pH: 8.03 ± 0.01 AT: 2248 ± 25 μmol L−1 Calculated pCO2: 550 ± 8 µatm DIC: 2093 ± 21 μmol L−1 Stationary growth phase pH: 8.04 ± 0.01 AT: 2254 ± 14 μmol L−1 Calculated pCO2: 537 ± 19 µatm DIC: 2107 ± 17 μmol L−1	pH 7.8 treatment Length of experiment: 9 days Exponential growth phase pH: 7.81 ± 0.03 AT: 2292 ± 3 μmol L−1 Calculated pCO2: 980 ± 64 µatm DIC: 2213 ± 11 μmol L−1 Stationary growth phase pH: 7.84 ± 0.01 AT: 2348 ± 17 μmol L−1 Calculated pCO2: 929 ± 12 µatm DIC: 2271 ± 16 μmol L−1	pH 8.0 vs. pH 7.8 Growth rate decreased **** Maximum carbon fixation rate increased *	[104]
		pH 7.9 treatment Length of experiment: 8 days Exponential growth phase pH: 7.93 ± 0.02 AT: 2255 ± 20 μmol L−1 Calculated pCO2: 719 ± 34 µatm DIC: 2136 ± 25 μmol L−1 Stationary growth phase pH: 7.93 ± 0.02 AT: 2279 ± 14 μmol L−1 Calculated pCO2: 732 ± 28 µatm DIC: 2175 ± 13 μmol L−1	pH 7.8 treatment Length of experiment: 9 days Exponential growth phase pH: 7.81 ± 0.03 AT: 2292 ± 3 μmol L−1 Calculated pCO2: 980 ± 64 µatm DIC: 2213 ± 11 μmol L−1 Stationary growth phase pH: 7.84 ± 0.01 AT: 2348 ± 17 μmol L−1 Calculated pCO2: 929 ± 12 µatm DIC: 2271 ± 16 μmol L−1	pH 7.9 vs. pH 7.8 Growth rate Decreased ****	[104]
Pseudo-nitzschia spp. in Gullmar Fjord	Mesocosm, addition of CO2-saturated seawater	380 µatm CO2 treatment Length of experiment: 111 days	1000 µatm CO2 treatment Calculated pCO2: 760 ± 175 µatm Length of experiment: 111 days	380 vs. 1000 µatm CO2 Cellular DA content increased *	[105]
PbTx-producing microalgae
Karenia brevis (CCFWC-126) isolated from the Gulf of Mexico	Laboratory culture CO2 gas bubbling	150 µatm CO2 treatment Before inoculation pH: 8.51 ± 0.04 AT: 2320 ± 3 μmol L−1 Calculated pCO2: 118.9 ± 0.1 µatm DIC: 1679 ± 2 μmol L−1 Mid exponential phase pH: 8.63 ± 0.05 AT: 2428 ± 64 μmol L−1 Calculated pCO2: 102.9 ± 16.8 µatm DIC: 1714 ± 4 μmol L−1	400 µatm CO2 treatment Before inoculation pH: 8.12 ± 0.02 AT: 2348 ± 18 μmol L−1 Calculated pCO2: 446.6 ± 31.0 µatm DIC: 2008 μmol L−1 Mid exponential phase pH: 8.22 ± 0.03 AT: 2376 ± 14 μmol L−1 Calculated pCO2: 355.6 ± 30.4 µatm DIC: 1982 ± 26 μmol L−1	150 vs. 400 µatm CO2 No significant response	[106]
		150 µatm CO2 treatment Before inoculation pH: 8.51 ± 0.04 AT: 2320 ± 3 μmol L−1 Calculated pCO2: 118.9 ± 0.1 µatm DIC: 1679 ± 2 μmol L−1 Mid exponential phase pH: 8.63 ± 0.05 AT: 2428 ± 64 μmol L−1 Calculated pCO2: 102.9 ± 16.8 µatm DIC: 1714 ± 4 μmol L−1	780 µatm CO2 treatment Before inoculation pH: 7.89 ± 0.01 AT: 2309 ± 1 μmol L−1 Calculated pCO2: 812.7 ± 0.5 µatm DIC: 2089 ± 1 μmol L−1 Mid exponential phase pH: 7.95 ± 0.04 AT: 2342 ± 19 μmol L−1 Calculated pCO2: 753.4 ± 78.8 µatm DIC: 2103 ± 4 μmol L−1	150 vs. 780 µatm CO2 No significant response	[106]
		400 µatm CO2 treatment Before inoculation pH: 8.12 ± 0.02 AT: 2348 ± 18 μmol L−1 Calculated pCO2: 446.6 ± 31.0 µatm DIC: 2008 μmol L−1 Mid exponential phase pH: 8.22 ± 0.03 AT: 2376 ± 14 μmol L−1 Calculated pCO2: 355.6 ± 30.4 µatm DIC: 1982 ± 26 μmol L−1	780 µatm CO2 treatment Before inoculation pH: 7.89 ± 0.01 AT: 2309 ± 1 μmol L−1 Calculated pCO2: 812.7 ± 0.5 µatm DIC: 2089 ± 1 μmol L−1 Mid exponential phase pH: 7.95 ± 0.04 AT: 2342 ± 19 μmol L−1 Calculated pCO2: 753.4 ± 78.8 µatm DIC: 2103 ± 4 μmol L−1	400 vs. 780 µatm CO2 No significant response	[106]
Karenia brevis (Wilson) isolated from John’s Pass	Laboratory culture CO2 gas bubbling prior to cell inoculation2	250 ppm CO2 treatment Length of experiment: 9 days Before inoculation Calculated pCO2: 241.2 µatm DIC: 1727.3 ± 33 μmol L−1 At the beginning of the experiment AT: 2082.8 ± 34.3 μmol L−1 Calculated pCO2: 134.2 µatm At the end of the experiment AT: 2198.6 ± 62.0 μmol L−1 Calculated pCO2: 80.5 µatm DIC: 1562.6 ± 14.5 μmol L−1	350 ppm CO2 treatment Length of experiment: 9 days Before inoculation Calculated pCO2: 318.4 µatm DIC: 1733.9 ± 163.7 μmol L−1 At the beginning of the experiment AT: 2021.3 ± 33.6 μmol L−1 Calculated pCO2: 115.6 µatm At the end of the experiment AT: 2250.5 ± 0.6 μmol L−1 Calculated pCO2: 55.1 µatm DIC: 1509.9 ± 259.7 μmol L−1	No significant response	[107]
		250 ppm CO2 treatment Length of experiment: 9 days Before inoculation Calculated pCO2: 241.2 µatm DIC: 1727.3 ± 33 μmol L−1 At the beginning of the experiment AT: 2082.8 ± 34.3 μmol L−1 Calculated pCO2: 134.2 µatm At the end of the experiment AT: 2198.6 ± 62.0 μmol L−1 Calculated pCO2: 80.5 µatm DIC: 1562.6 ± 14.5 μmol L−1	1000 ppm CO2 treatment Length of experiment: 9 days Before inoculation Calculated pCO2: 1131.9 µatm DIC: 1986 ± 17.2 μmol L−1 At the beginning of the experiment AT: 2076.6 ± 6 μmol L−1 Calculated pCO2: 438.9 µatm At the end of the experiment AT: 2224.7 ± 10.7 μmol L−1 Calculated pCO2: 166.1 µatm DIC: 1750.8 ± 21.1 μmol L−1	250 vs. 1000 ppm CO2 Growth rate increased ***	[107]
		350 ppm CO2 treatment Length of experiment: 9 days Before inoculation Calculated pCO2: 318.4 µatm DIC: 1733.9 ± 163.7 μmol L−1 At the beginning of the experiment AT: 2021.3 ± 33.6 μmol L−1 Calculated pCO2: 115.6 µatm At the end of the experiment AT: 2250.5 ± 0.6 μmol L−1 Calculated pCO2: 55.1 µatm DIC: 1509.9 ± 259.7 μmol L−1	1000 ppm CO2 treatment Length of experiment: 9 days Before inoculation Calculated pCO2: 1131.9 µatm DIC: 1986 ± 17.2 μmol L−1 At the beginning of the experiment AT: 2076.6 ± 6 μmol L−1 Calculated pCO2: 438.9 µatm At the end of the experiment AT: 2224.7 ± 10.7 μmol L−1 Calculated pCO2: 166.1 µatm DIC: 1750.8 ± 21.1 μmol L−1	350 vs. 1000 ppm CO2 Growth rate increased ***	[107]
Karenia brevis (SP1) isolated from South Padre Island		300 ppm CO2 treatment Length of experiment: 9 days Before inoculation Calculated pCO2: 299 ± 157.8 µatm At the beginning of the experiment Calculated pCO2: 211.8 ± 32 µatm At the end of the experiment Calculated pCO2: 99 ± 32.7 µatm	1000 ppm CO2 treatment Length of experiment: 9 days Before inoculation Calculated pCO2: 1084 ± 14.1 µatm At the beginning of the experiment Calculated pCO2: 398.6 ± 3.4 µatm At the end of the experiment Calculated pCO2: 219 ± 11.1 µatm	300 vs. 1000 ppm CO2 Growth rate increased	[107]

^a A_T: total alkalinity, DIC: dissolved inorganic carbon. Length of experiment does not include pre-acclimation period of culture. ^b Significant outcomes refer to the significant differences in the growth rates, carbon fixation rates, and/or the toxicity observed between the corresponding control and high CO₂ treatment. * = p ≤ 0.05, ** = p ≤ 0.01, *** = p ≤ 0.001,**** = p ≤ 0.0001. ^c Seawater was manipulated once by CO₂ gas bubbling before cell inoculation.

.

3.1.1. STXs-Producing Microalgae

Dinoflagellates from the genus Alexandrium are the most studied STX producers. However, there is not yet any solid conclusion drawn for the toxicity change in Alexandrium at elevated pCO₂. Increases in cellular toxin levels were detected in A. catenella, A. minutum, and A. fundyense after high CO₂ treatment [97,98,99]. It is believed that elevated pCO₂ will permit microalgae to perform photosynthesis when more substrates are available for their unsaturated carbon-fixing enzyme ribulose-1,5-bisphosphate carboxylase-oxygenase (Rubisco) [108]. Dinoflagellates may be particularly sensitive to elevated pCO₂ as they possess type II Rubisco, which has a higher affinity with O₂, a competitive inhibitor of CO₂ [109]. Since carbon is essential for the synthesis of biomolecules, including phycotoxins, increased production of fixed carbon at elevated pCO₂ may thus contribute to the higher toxicity of microalgae [99]. However, the carbon fixation rates of A. catenella, A. minutum, and A. fundyense after high CO₂ treatment were not measured to support this hypothesis.

Another plausible explanation for the enhanced production of STXs is the increased availability of STX precursors [98]. To initiate the synthesis of STX, arginine (Arg), methionine (Met), and acetate are required [110]. In the study performed by Lian, Li, He, Chen, and Yu [98], the concentrations of Arg and Met were quantified in A. minutum at elevated pCO₂. It was found that the content of Arg and Met increased at the beginning of the experiment and then decreased gradually. The increase in the Arg level was related to the enhanced activity of argininosuccinate synthase, an enzyme responsible for a rate-limiting step in Arg synthesis. However, the mechanisms behind the rise in the Met level remained undetermined. Since the STX content increased after the decline in Arg and Met levels with a slight delay, it was suggested that elevated pCO₂ had indirectly promoted STX production by increasing Arg and Met supplies.

On the other hand, significant decreases in STX content were detected in both A. tamarense and A. ostenfeldii at elevated pCO₂, resulting in lower toxicity [100,101]. Van de Waal, Eberlein, John, Wohlrab, and Rost [101] isolated the RNA of A. tamarense cultured at different pCO₂ (180, 380, 800, 1200 ppm) to perform microarray-based gene expression analysis. When comparing the 800 ppm group to the 380 ppm group (control), 1238 differentially expressed genes (DEGs) were found. Genes associated with amino acid transport and metabolism were downregulated, implying that fewer Arg and Met molecules were synthesized and available for STX biosynthesis at elevated pCO₂.

The differences in the toxicity of Alexandrium species may be caused by genetic variations between species. The gene expression patterns of the toxic A. fundyense and A. tamarense strains revealed by Taroncher-Oldenburg and Anderson [111] indicated the high interspecific variations between them. In parallel with this previous finding, A. fundyense and A. tamarense responded differently under OA. While A. fundyense promoted STX production at elevated pCO₂ [97], A. tamarense inhibited STX production by reducing the synthesis of STX precursors [101]. The Alexandrium species are likely to regulate their metabolism differently under OA due to the interspecific variations in genetic expression, thus contributing to the toxicity differences.

Regarding the STX composition of Alexandrium, increased GTX1/4 is a common change observed in A. catenella and A. tamarense, although the toxin profiles of Alexandrium are highly varied [99,101,112]. According to the gene expression profile of A. tamarense, sulfur metabolism was differentially regulated under high CO₂ treatment [101]. The significant upregulation of a putative sxtN homologue encoding a sulfotransferase involved in synthesizing sulfated STXs was found [113,114,115,116]. In contrast, genes encoding sulfatases that are responsible for the hydrolysis of sulfate esters were downregulated. Together, they might promote the transformation of non-sulfated STXs to sulfated STXs while inhibiting the transformation of sulfated STXs to non-sulfated STXs. Moreover, the downregulation of genes encoding sulfite reductase was reported. Inhibition of assimilated sulfur into amino acids might result in more sulfur being allocated for synthesizing sulfated STXs [117,118]. In parallel with these, lower non-sulfated STX content but higher sulfated GTX1/4 and C1/2 content of A. tamarense at elevated pCO₂ were observed. Nevertheless, increases in both non-sulfated and sulfated STX levels were exhibited in A. catenella after high CO₂ treatment [99], suggesting that the toxin composition could be altered by different metabolic pathways besides sulfur metabolism.

3.1.2. DA-Producing Microalgae

The diatom Pseudo-nitzschia has been extensively studied compared to other DA producers. Generally, both laboratory and mesocosm studies revealed increased cellular DA content of Pseudo-nitzschia at elevated pCO₂ [102,103,104,105]. In line with the assumption that excess fixed carbon may be shunted for toxin biosynthesis in microalgae under OA [99], an increased growth rate, carbon fixation rate, and cellular DA content were reported in P. fraudulenta at elevated pCO₂ [103]. Nonetheless, there were no consistent findings of P. australis [104]. The cellular DA content of P. australis was largely unchanged despite the increased carbon fixation rate at elevated pCO₂.

For DA-producing Pseudo-nitzchia, environmental pH may also play a role in toxin biosynthesis by affecting cellular enzymatic activities, metal speciation, and the composition of symbiotic bacteria [119]. The intracellular pH of diatoms may be modified or maintained by regulating the metabolism under acidification [120]. Although research on the intracellular pH change of Pseudo-nitzchia at elevated pCO₂ is lacking, the effects of acidification on DA biosynthesis should not be neglected. Lundholm, Hansen, and Kotaki [119] proposed that there might be an optimal pH for DA biosynthesis; however, it remains to be determined as an increased cellular DA level was detected under both low and high pH conditions in Pseudo-nitzschia [102,103,119,121]. The DA biosynthetic pathway has been modeled recently based on the transcriptome sequencing of P. multiseries [122]. The intracellular pH of Pseudo-nitzschia may be modified under acidification, which possibly affects the activities of putative DA biosynthetic (Dab) enzymes including terpene cyclase (dabA), hypothetical protein (dabB), alpha-ketoglutarate-dependent dioxygenase (dabC), and CYP450 (dabD), causing changes in cellular DA content.

The toxicity and availability of metals are likely to be affected by a decreased pH. For example, a rise in the concentrations of free copper (Cu²⁺) and dissolved iron (Fe³⁺) was expected at lower pH levels due to increased solubility [123]. While the Cu²⁺ level is generally closely related to its toxicity in phytoplankton [124], Fe³⁺ is one of the essential micronutrients for microalgal growth. Although the ecological and physiological roles of DA have not been well established, it was suggested that DA might be a trace metal chelator involved in iron acquisition and copper detoxification, given its similar structure to phytosiderophores [125]. Therefore, increases in Cu²⁺ and Fe³⁺ levels under an acidic environment may stimulate the DA production of Pseudo-nitzschia. However, both the cellular and dissolved DA content of P. multiseries remained largely unchanged after exposure to different copper levels [126]. On the other hand, a positive relationship between the iron concentration and total DA content in P. multiseries was indicated [127]. Thus, enhanced DA production at elevated pCO₂ may be mainly contributed by the increased concentration of Fe³⁺ rather than Cu²⁺ under acidification.

Symbiotic bacteria are related to DA biosynthesis in Pseudo-nitzschia. Xenic cultures of P. multiseries exhibited increased DA content when compared with axenic cultures [128]. While free-living bacteria were likely to be incapable of producing DA autonomously [129,130], it is difficult to obtain a conclusion for epiphytic bacteria due to the difficulties in isolating them from the diatoms. Research on the bacteria–phytoplankton interaction mechanisms for DA biosynthesis in Pseudo-nitzschia is lacking. However, it was proposed that the attached bacteria might exchange metabolites that are used for DA biosynthesis with Pseudo-nitzschia [130]. Since bacterial abundance and diversity were shown to be affected by a decreased environmental pH [131], symbiotic bacteria that promote DA biosynthesis may become more abundant at lower pH levels given the observation of increased cellular DA content in Pseudo-nitzschia.

3.1.3. PbTx-Producing Microalgae

Dinoflagellate K. brevis is the primary producer of PbTx and its response to OA has been studied. Elevated pCO₂ had no significant effect on the cellular PbTx content of K. brevis [106,107]. Although K. brevis shifted its inorganic carbon preference from HCO₃⁻ to CO₂ and the half-saturation constant (K_1/2) for CO₂ increased under high CO₂ treatment, the growth rate, carbon fixation rate, and cellular carbon composition remained largely unchanged [106]. The relative insensitivity of K. brevis to elevated pCO₂ was surprising as high CO₂ availability had been thought to be especially beneficial to dinoflagellates that possess inefficient type II Rubisco [109].

3.2. OA May Affect the Central Carbon Metabolism of Toxic Microalgae

Based on the above, OA can affect the toxicity of some species of dinoflagellate Alexandrium and diatom Pseudo-nitzschia. On the other hand, dinoflagellate K. brevis was shown to be somewhat more resistant to OA and its toxicity was largely unchanged. The way in which they regulate central carbon metabolism under OA is likely to contribute to the changes in toxicity or to help to achieve homeostasis. OA was shown to be beneficial to P. fraudulenta as it promoted the carbon fixation rate by increasing inorganic carbon availability [103]. The increased growth rate and cellular toxin content gave rise to the hypothesis that excess carbon not used for growth might be shunted for toxin biosynthesis [99]. However, the positive relationship between the carbon fixation rate and toxin level does not always apply, as observed in P. multiseries [104]. All biomolecules are carbon-based, so the excess fixed carbon is not necessarily transported for toxin production. The detailed routes of carbon reallocation for toxin biosynthesis have not been elucidated either. Therefore, investigating the central carbon metabolism of Alexandrium and Pseudo-nitzschia is important to validate the hypothesis. For K. brevis, the acquisition of inorganic carbon was shown to be affected by OA despite the nearly unchanged growth rate, carbon fixation rate, and toxin content [106]. Since central carbon metabolism is highly conserved across phylogeny [132], comparing the central carbon metabolism of relatively resistant K. brevis to that of other sensitive microalgae at elevated pCO₂ may help to explain the unique response of K. brevis to OA. A decreased pH was proposed to affect the toxicity of Pseudo-nitzschia, though the exact mechanisms had not been demonstrated [119]. Increased inorganic carbon availability and decreased pH may have combined effects on the regulation of central carbon metabolism in microalgae, hence affecting toxin biosynthesis. Considering the above, studying the carbon fluxes in Alexandrium, Pseudo-nitzschia, and K. brevis is essential to deepen our understanding of the effects of OA on microalgae.

4. Future Research Directions

Manipulating seawater to a range of increased pCO₂ levels for OA experiments is not a straightforward task due to the interdependency of ocean carbon system components [133]. To predict the microalgal responses under OA accurately, the seawater used for microalgae cultivation should resemble the ocean carbon chemistry in future high CO₂ scenarios as closely as possible. Multiple seawater manipulation methods with distinct advantages and disadvantages have been applied to OA experiments. Most of the reviewed studies adopted CO₂ gas bubbling to model OA (Table 1). CO₂ gas bubbling is an efficient seawater manipulation method that exactly mimics the carbon chemistry of the ocean in future high atmospheric CO₂ scenarios (increased DIC without altering A_T) [134]. In addition, it is easy to implement and it can maintain the initial conditions in the long term [135]. Thus, CO₂ gas bubbling was adopted by most of the studies to investigate the effects of OA on microalgae. However, the turbulence induced by bubbling may affect the growth of phytoplankton to a different extent, especially Alexandrium spp. [136]. Hence, the effects of turbulence became an uncontrolled confounding variable, which increased the difficulties in generating reproducible results for the microalgal adaptive response to OA [137]. To minimize the effects of turbulence, a dialysis bag with a 3 kDa molecular weight cut-off should be used to enclose the microalgae, so as to reduce the mechanical disturbance introduced by the aeration [135]. Unfortunately, none of the reviewed studies mentioned the uses of dialysis bags in their methodology. Besides CO₂ gas bubbling, mixing CO₂-enriched water prior to cell inoculation, and the combined addition of HCl and Na₂CO₃/NaHCO₃, are other seawater manipulation methods that also closely resemble the carbon chemistry of the ocean in future high atmospheric CO₂ conditions [134,135]. These methods can be effective and reliable alternatives to CO₂ gas bubbling without the mechanical disturbance to microalgae. Thus, they are especially useful for studying the effects of OA on microalgae with high sensitivity to turbulence. Regardless of the method used, it is noteworthy that the biological processes of microalgae, such as respiration and photosynthesis, will affect the carbon chemistry of seawater, especially when the biomass is high [138,139]. Thus, it would be preferable to monitor at least two carbon chemistry parameters of pH, A_T, DIC, and pCO₂, in addition to temperature and salinity, throughout the experiments [140].

With appropriate methods to model OA and monitor the carbon chemistry of the seawater, metabolomics is a useful technique to investigate the metabolic changes of the microalgae in response to OA. This will help to reveal the biological regulation of toxin biosynthesis in microalgae under OA. Metabolomics is the investigation of a complete set of small molecules with molecular weights lower than 1500 Da, also known as the metabolome, of biological samples [141]. It has been increasingly applied in algal toxin research with technological advancements [142,143]. Metabolic flux analysis (MFA) is an effective metabolomic tool that examines the turnover rate of metabolites by using stable isotope tracers such as ¹³C, ²H, and ¹⁵N [144]. With the help of nuclear magnetic resonance (NMR) or mass spectrometry (MS), the isotope labeling patterns of intracellular metabolites are determined, which helps to construct a flux map, including reverse fluxes [145]. Compared to the measurement of metabolite levels, analyzing metabolic fluxes is more informative as it reveals the production and consumption rates of metabolites, which illustrate the biochemical events behind the changes in metabolite levels [146]. Based on the reviewed studies, it is hypothesized that the central carbon metabolism of toxic microalgae may be regulated differently under OA. The altered central carbon metabolism may then affect the toxicity of microalgae or relieve the stress brought by OA. Therefore, further studies on the central carbon metabolism of toxic microalgae are warranted. MFA has been applied for the investigation of photosynthesis and central carbon metabolism in microalgae, C₃, and C₄ plants recently in ¹³CO₂ pulse labeling, and the distinct carbon flux pattern of microalgae has been successfully identified [147]. In this light, carbon allocation in toxic microalgae can be investigated to reveal the metabolic pathways related to toxin biosynthesis and their regulation under OA by MFA in order to elucidate the underlying mechanisms of toxicity changes. When conducting MFA to study the effects of OA on the carbon allocations of microalgae, it is critical to note that CO₂ gas bubbling without the continuous control of the pH would alter the pCO₂, DIC, and pH of the seawater at the same time. As a result, it is difficult to determine which confounding variable causes the observed effects. Although CO₂ gas bubbling successfully mimics the carbon chemistry of seawater under OA observed in reality, this method fails to decouple different variables in the carbon chemistry to study the individual effects of the variables on the carbon allocations of microalgae. Therefore, other methods or systems that can decouple different variables of the carbon chemistry of seawater under OA should also be adopted to investigate in detail the biological mechanisms of the changes in carbon allocation of microalgae under OA when using the metabolomic approach (Figure 1) [148].

Knowledge of the relationship between environmental drivers and HABs is of fundamental importance to optimize the current mitigation of HABs [149]. Under climate change, it is undoubted that the HAB dynamic has been altered [1]. However, the physiological responses of toxigenic microalgae in changed climate conditions have not been well documented. A precise linkage between environmental factors and microalgal toxicity has not been confirmed due to the diverse responses displayed by microalgae belonging to the same genus or even the same species, as in the case of the STX-producing dinoflagellate Alexandrium [97,98,99,100,101,150]. MFA of the central carbon metabolism of microalgae can reveal the metabolic pathways that are responsible for the toxicity changes under OA. The identification of differentially regulated pathways will facilitate the discovery of universal biomarkers for potentially more toxic microalgae in future elevated atmospheric CO₂ scenarios. Given the diverse responses of microalgae to different environmental conditions, detecting universal biomarkers is a relatively efficient way to identify microalgae with increased toxicity under climate change. Monitoring and forecasting these microalgae can be prioritized to help establish early warning systems for the coastal tourism and aquaculture sectors, to minimize the economic losses caused by HABs [151,152,153]. Policymakers can guide the closure of beaches, fish farms, and shellfish-harvesting areas in an appropriate timeframe according to the early warning system, to protect public health while minimizing economic losses. Moreover, more research efforts can be implemented to understand the interactions between the potentially more toxic microalgae and other organisms in the food web under climate pressure. The findings can be used to assess the ecological risk of HABs in the future, which will aid policymakers to achieve a balance between environmental protection and economic development.