Mechanisms of light harvesting complex proteins in photoprotection of the brown tide alga

The ability to efficiently harvest light energy and protect against photodamage is crucial for the ecological success of phytoplankton. Aureococcus anophagefferens is widely distributed across the global oceans, causing large-scale brown tide blooms frequently observed in the coastal waters of the United States, South Africa, and China^1,2,3,4. These outbreaks pose serious threats to shellfish farming, fishing, and tourism industries, adversely affecting local economies and ecosystems³. Brown tides typically erupt in relatively enclosed, shallow and turbid estuaries characterized by low light intensity, high organic matter concentrations, and abundant trace elements⁵.

The competitive advantage of A. anophagefferens in low-light environments is critical for its ability to form brown tide blooms and derives from its unique cellular responses and gene regulation. This species achieves maximum growth rates under low light conditions⁶, enabled by its large repertoire of 62 light-harvesting protein genes, which are several times more abundant than in competing species, thereby enhancing light capture efficiency^5,7. These adaptations demonstrate its exceptional light-harvesting abilities.

Low-light-adaptive algae are typically more vulnerable to high-light stress^8,9. A. anophagefferens, living in shallow waters, must cope with occasional oversaturated light intensities through diel cycles or vertical mixing. In algae, photoprotection is achieved by dissipating excess light energy, mostly through non-photochemical quenching (NPQ), a mechanism shared by both algae and vascular plants¹⁰. Energy-dependent fluorescence quenching (qE), triggered by the proton gradient (ΔpH) across thylakoid membranes, represents the fastest NPQ component¹¹ and typically associates with the xanthophyll cycle^12,13,14. In addition, qE is related to the protein function of photosystem II subunit S (PsbS) in vascular plants, which belongs to the light-harvesting complex (LHC) family¹⁵. Photosynthetic algae exhibit similar photoprotection mechanisms, such as the LhcSR protein in green algae and the Lhcx protein in diatoms^16,17. However, the specific process by which the ΔpH ultimately triggers qE remains unclear. Diatoms utilize the xanthophyll cycle to respond to ΔpH but cannot directly trigger qE^12,18,19. The decreased pH in the thylakoid lumen triggers qE through the protonation of proteins, a process sensed by acidic amino acids^15,19.

In vascular plants, the PsbS protein with four transmembrane helices senses the ΔpH through two glutamate residues symmetrically positioned on the two loops on the lumenal side²⁰. Although the PsbS gene is also present in green algae, its function remains unknown²¹. The LhcSR or Lhcx proteins in algae contain acidic amino acid residues on the lumenal side, but unlike the PsbS protein, they both have a three-transmembrane helix structure, leaving their mechanism for sensing the proton gradient unclear. In C. reinhardtii, the D117, E221 and E224 of LhcSR3 are essential for NPQ induction¹⁹. However, the equivalent residues (D95 and E205) in Lhcx1 of the diatom Phaeodactylum tricornutum are not required for inducing NPQ²². Therefore, while acidic amino acids potentially play an important role, there is no consensus on their significance.

Here, we isolate and culture A. anophagefferens from a brown tide bloom, investigating how its light-harvesting system provides competitive advantages that enable the alga to form blooms. Brown tide is one of the harmful algal blooms creating devastating impacts on the aquatic ecosystem, the coastal environment, and economy, as well as public health concerns. Focusing on the numerous LHC proteins of A. anophagefferens, we examine the roles of these proteins, particularly their association with brown tide outbreaks in estuaries, and the underlying molecular mechanisms. We discover the critical roles of two glutamic residues in the LHC proteins of A. anophagefferens. Data from multiple experiments verify their connection to thylakoid acidification. In this study, we provide insight into the proton gradient response sites and common patterns of algal LHC proteins, which suggest significant implications for understanding the effects of light on brown tide blooms.

The strong adaptability of A. anophagefferens to light changes

Similar to populations in the US and Africa,A. anophagefferens frequently blooms in shallow estuaries of China, particularly in the Bohai Sea. We conducted year-round sampling in Bohai Sea from 2013 to 2014 (Fig. 1a). Surface light intensity varied considerably across sampling sites, with approximately one-third receiving less than 200 μmol photons m⁻² s⁻¹ (Fig. 1b). Light attenuation increased with depth: about two-thirds of sites recorded below 200 μmol photons m⁻² s⁻¹ at one meter depth, while 91.43% of sites fell below 100 μmol photons m⁻² s⁻¹ at two meters. Field surveys revealed that A. anophagefferens maintained a uniform vertical distribution, whereas competing phytoplankton species exhibited varying distribution patterns. Dinoflagellates tended to prefer the surface with stronger light intensity. In contrast, the light intensity preferences of diatoms varied among species, resulting in non-uniform vertical distributions (Fig. 1c). These findings demonstrate A. anophagefferens’ exceptional adaptation to light variability. Its ability to adapt to low light helps them survive in the light-deficient coastal waters⁵, while its uniform abundance from the sea surface to depths suggests strong photoprotection. This strong photo-plasticity is exceptional. The only comparable case, best to our knowledge, is the cyanobacterium Prochlorococcus, which differentiates into different ecotypes: low-light-adapted strains that thrive in the deep euphotic zone and high-light-adapted strains that thrive at the surface²³. They adjust the light capture efficiency through special pigments (Chl b and divinyl Chl a) and proteins (Pcb proteins), enabling them to survive widely in the ocean²⁴. However, studies have found that this niche differentiation involves genetic divergence. In coral symbiotic dinoflagellates, those adapted to high-light environments can avoid photodamage through photoprotective mechanisms, while those adapted to low-light environments are more susceptible to photoinhibition²⁵. Among vascular plants, Hedera helix is sensitive to changes in light intensity and can rapidly induce non-photochemical quenching (NPQ) shortly after a slight increase in light intensity²⁶. From this perspective, A. anophagefferens and H. helix share similar low-light response characteristics. Both species are well adapted to thrive in low-light environments, are sensitive to minor fluctuations in light intensity, and possess strong photoprotective mechanisms.

High LHC gene abundance in A. anophagefferens with unique glutamic acid residues and evidence of their role in photoprotection

The strong photoadaptation of A. anophagefferens observed from the uniform vertical abundance distribution is consistent with the high number of LHC protein genes (AaLHC) in its genome⁵. To examine the role of AaLHC proteins in the photoprotection, we conducted a response analysis for the 62 AaLHC found in A. anophagefferens to different irradiance environments. Two expression patterns emerged (Fig. 2a). The 11 AaLHC in Group I showed increased expression levels when transferred from low light to high light, and from darkness to light of white and blue spectra. In contrast, the 51 AaLHC in Group II exhibited the opposite trend, with expression levels decreasing under high light stress and only increasing after a longer period of transfer from darkness to a lighted environment. Among them, the four genes AaLhc27, AaLhc37, AaLhc42, and AaLhc52 in Group I were more sensitive to short-term exposure. Additionally, weighted gene co-expression network analysis (WGCNA) identified the 11 AaLHC almost all in the same module (blue module, Supplementary Tables 1, 2), which were associated with short-term high light stress and short-term exposure to white and blue light, suggesting that these genes respond rapidly to changes in the light environment (Fig. 2b). We further analyzed these four significantly responsive genes using qPCR. Results showed that their expression was upregulated immediately after cells were exposed to high light, and dropped quickly when the cultures were returned to a low light environment (Fig. 2c). These genes were also upregulated when the algal cultures were transferred from darkness to low light or from white light to blue light (light stress in comparison, which has a shorter wavelength and higher energy). This result suggests that AaLhc27, AaLhc37, AaLhc42, and AaLhc52, and potentially others in Group I AaLHC as well, likely play the same role in light response.

We further analyzed their amino acid sequences to identify the reasons behind the differences in photoprotective capabilities between the two groups of AaLHC. We noted that the AaLHC proteins in Group I all had two conserved glutamic acid (E) residues (henceforth termed AaLhc2E), while AaLHC proteins in Group II had none (in most cases) or one (Supplementary Fig. 1). Further structural simulations revealed that these two E residues were located on the lumenal loop and the C-terminal domain within the thylakoid membrane (Fig. 2d). Moreover, the two glutamic residues were located very closely to each other (<3.0 Å apart) in all the 11 AaLhc2Es. The proximity potentially facilitates structural adjustments of the protein when they are protonated, as documented for LhcSR in the chlorophyte C. reinhardtii¹⁹. Based on the key role of the proton gradient (ΔpH) across the thylakoid membrane in NPQ, we manipulated ΔpH in A. anophagefferens by adding NH₄Cl to the algal culture to disrupt ΔpH, following a previously reported protocol¹². The results showed that without NH₄Cl treatment, NPQ could be induced by light and relaxed in the dark. NH₄Cl addition rapidly led to a NPQ decrease (Fig. 2e). This result indicates that potentially, the cross-membrane proton gradient may be modulated or sensed involving the two E residues in the luminal loop, thereby regulating NPQ.

As the xanthophyll cycle is coupled with and crucial for NPQ¹¹, we investigated the potential concerted action of them. We detected all xanthophyll pigments in A. anophagefferens, including trace amounts of violaxanthin (Vx), antheraxanthin (Ax), and zeaxanthin (Zx), and observed diadinoxanthin (Dd) de-epoxidation to diatoxanthin (Dt) under high light irradiation (Supplementary Fig. 2a). Furthermore, treatment with NH₄Cl inhibited the de-epoxidation state (DPS), indicating that the Dd-Dt cycle, rather than the Vx-Ax-Zx cycle, responds to pH changes. In addition, the conversion of fucoxanthin (Fx) to 19′-butanoyloxyfucoxanthin (19′-BFx)⁷ also appeared to occur under high light but was unaffected by ΔpH (Supplementary Fig. 2b), indicating that the role of Fx-to-19′-BFx conversion in photoprotection is independent of the ΔpH-modulated NPQ pathway.

Similarity of AaLHC with glutamic acid residues to the photoprotective Lhcx/LhcSR in other algae

As Lhcx/LhcSR is the best known photoprotective protein complex and shared by Archaeplastids, Stramenopiles, Alveolates and Rhizaria²⁷, we compared AaLHCs with it and other light-harvesting proteins. Our phylogenetic analysis results showed that the 62 AaLHC proteins were divided into three categories, which clustered with Lhcf, Lhcr, and Lhcx/LhcSR, respectively (Fig. 3a). Among them, all the 11 AaLhc2E proteins clustered with Lhcx/LhcSR. This indicates that the two E residues are ancient and play a crucial functional role.

Alignment analysis of the 11 AaLhc2E proteins and the Lhcx/LhcSR proteins (Fig. 3a) revealed that transmembrane helices 1 and 3 were relatively conserved (Fig. 3b). The two conserved E residues of the three-transmembrane Lhcx/LhcSR were located at the ends of the helices extending into the lumen. Differently, the two E residues (E122 and E226) in the four-transmembrane PsbS of vascular plants are both located on the loop in the thylakoid lumen²⁸. So far, it has been unclear whether algal LHC proteins all possess conserved acidic residues that respond to pH and are crucial to NPQ. In C. reinhardtii, residues D117, E221, and E224 in LhcSR3 are considered essential for inducing NPQ¹⁹. In the model diatom P. tricornutum, residues with potential protonation functions (D95 and E205 in Lhcx1) appear to be unnecessary for NPQ, while tryptophan residues located in the transmembrane helices may be essential^22,29. Therefore, it is necessary to experimentally demonstrate that the two conserved glutamate residues in AaLhc2E are directly involved in inducing NPQ in A. anophagefferens. Such experimental evidence is presented in the next section.

Crucial role of E residues in the function of AaLHC for NPQ

As a functional genetic system is not yet available for A. anophagefferens, we chose a diatom model for the functional demonstration of AaLhc2Es. Based on the conservation of transmembrane helices 1 and 3 of the Lhcx/LhcSR among photosynthetic species, we compared the LHC protein sequences of A. anophagefferens and the model diatom Thalassiosira pseudonana. Results showed that the transmembrane regions of LHC proteins between these two species are also highly similar (Fig. 4a). We used T. pseudonana as the model to determine the function of a AaLhc2E for photoprotection. We overexpressed AaLhc27 in T. pseudonana (Tp_AaLhc27E). To verify the function of the E residues, we mutated E102 and E210 in AaLhc27E to Q102 and Q210, respectively, and included them in the heterologous expression experiment (Tp_AaLhc27Q) (Fig. 4b, Supplementary Fig. 3). We chose to replace the glutamic acid (E) residues with glutamine (Q) to minimize biochemical alterations except for the charge change. As a result, the AaLhc27E-expressing T. pseudonana strain (Tp_AaLhc27E) exhibited a greater NPQ under high light, while the AaLhc27Q-expressing strain (Tp_AaLhc27Q) displayed a smaller change in NPQ induction compared to the wild-type (Fig. 4c), indicating that the E residues play a crucial role in the function of AaLHC27 for NPQ. The capacity of NPQ is closely related to the expression of photoprotective LHCs and xanthophyll pigments bound to them^30,31. Compared to the wild-type and Tp_AaLhc27Q, the DPS of Tp_AaLhc27E increased more rapidly upon light induction (Fig. 4d), indicating that the xanthophyll cycle is involved in the AaLHC27E-based NPQ regulation. Notably, the centric diatom T. pseudonana exhibits lower DPS values than the pennate diatom P. tricornutum³². Apart from the influence of growth conditions and light history³³, this difference may result from the distinct topologies of NADPH-dehydrogenases (type-1 and type-2)³². The suppression of DPS elevation upon DCMU addition (Fig. 4d) suggests that T. pseudonana likely possesses a non-membrane-spanning type-2 NADPH dehydrogenase, which is probably not involved in generating the lumenal proton gradient that facilitates DDE-mediated de-epoxidation. Whether this enzyme releases protons on the stromal side and, together with abundant NADPH, promotes the epoxidation of Dt also warrants further investigation. Furthermore, our growth measurements indicated that the promoted high light adaptation by AaLHC27E enabled the AaLhc27E-expressing T. pseudonana to grow better under higher light conditions (200 and 800 µmol photons m⁻² s⁻¹) compared to the AaLhc27Q-expressing strain and the wild type strain (Fig. 4e). Given that T. pseudonana was transformed with only one AaLhc2E, whereas A. anophagefferens possesses eleven AaLhc2Es, this suggests that the multiple AaLhc2Es in A. anophagefferens may provide greater advantages in dissipating excess energy and growing well under intense and fluctuating light conditions.

To shed light on how proton gradient dynamics may alter AaLhc2E structures and modulate NPQ, we conducted fourier transform infrared (FTIR) spectroscopy analysis of AaLHC27 under varying pH conditions. D₂O (D = deuterium) detergent buffer was used to avoid interference from O–H bending absorption, which overlaps with the Amide I region corresponding to the C = O vibrations in the protein backbone. Therefore, the FTIR results will be discussed in terms of pD instead of pH (pD = pH + 0.4). At pD 7.5, the FTIR spectrum of AaLHC27E exhibited a prominent band at 1570 cm⁻¹, which originated from the stretching vibrations of the deprotonated carboxylic acid in the deprotonated E residues (Fig. 5a)³⁴. Notable alteration in the FTIR spectrum was detected at pD 5.0, where a broad positive band in the region of 1700 to 1750 cm⁻¹ was observed, indicative of the presence of deuterated carboxyls (COOD). The characteristic of protonated (deuterated) carboxyls suggests that nearly all E residues were protonated at pD 5.0. In addition, comparing the FTIR spectra at pD 5.0 and pD 7.5 revealed changes in the Amide I region from 1600 to 1700 cm⁻¹, where a negative band at 1640 cm⁻¹ and a broad positive band at 1690 cm⁻¹ indicated a change in the Amide I band, and hence a conformational change in the secondary structure of the AaLHC27E. In contrast, only a minimal change was detected in the FTIR spectra of the AaLHC27Q mutant under different pD conditions. Clearly, mutating the two E residues to Q abolished proton response sensitivity while the general secondary structure of the protein was not altered.

At acidic pH, the E102 and E210 of AaLHC27 were in close proximity to I217 and I116 (2.4 Å), respectively, which likely contributes to the formation of stable interactions (Fig. 5b). Similarly, the two E residues of AaLhc2E proteins could form stable hydrogen bonds with amino acid residues on the lumenal side, and these residues are located in the C-terminal domain and the lumenal loop, respectively (Fig. 5c). Therefore, in an acidic lumenal environment, the glutamic residues on helix1 and helix3 become protonated and can easily approach the amino acid residues of the other within a very short range to establish stable interactions, thereby leading to changes in the protein’s secondary structure (Fig. 5d). Through 1000 ns of constant pH molecular dynamics (CpHMD) simulations, the dynamic stability changed little from a neutral environment to an acidic environment (Supplementary Fig. 4, Supplementary Tables 3–6). To further observe the major motion states of the four systems (AaLHC27E_pH_4.5, AaLHC27E_pH_7, AaLHC27Q_pH_4.5, and AaLHC27Q_pH_7), the slowest motions were captured using time-lagged independent component analysis (tICA), and Markov state models (MSMs) were built to examine the differences in state transitions across the systems. The MSM models showed that each state transition varied in time and conformation across the four systems. Specifically, in the AaLHC27E system, both neutral and acidic conditions led to the formation of three metastable states, while the AaLHC27Q system formed four metastable states under both conditions. The most significant structural difference among the four systems was in the lumenal side (e.g., residues 118-130). Under both acidic and neutral conditions, the AaLHC27Q system exhibited a clear helix secondary structure. In contrast, the AaLHC27E system exhibited a partial loop structure along with the helix structure under both conditions (Supplementary Figs. 5, 6).

Our results clearly indicate the crucial role of the two luminal E residues in AaLHC27 for NPQ in A. anophagefferens. Given the highly similarity in sequence and simulated structure, including the spatial proximity of the two glutamate residues, the other 10 AaLhc2E proteins likely serve the same function. However, experimental proof remains to be obtained in future research. Further research is also required to understand if the structurally conserved E residues in LHCs in other algae play the same role. It is worth noting that the total of 11 AaLhc2E proteins (and 62 total LHC proteins) in A. anophagefferens far exceeds that of many other algae but is comparable to some lineages living in highly light-variable environments. The intertidal brown alga Ectocarpus siliculosus and the polar ocean-adapted diatom Fragilariopsis cylindrus both contain 11 Lhcx-like/Lhcx proteins^35,36. The intertidal environment undergoes extreme light changes daily. The polar environment experiences complete light to complete dark annually, but F. cylindrus flourishes and forms blooms there³⁷. Emiliania huxleyi, which thrives and forms large-scale blooms in habitats ranging from the equator to the subarctic, possesses even more (17) LHCs with photoprotective functions³⁸, although the apparent plasticity within species can potentially be attributed to genetic differentiation into distinct geographic populations. Our findings provide insights into a molecular mechanism by which LHCs mediate photoprotection in A. anophagefferens, with implications in other variable light-adapted lineages.

Our results on AaLhc2Es also have significant ecological implications for the brown tide alga A. anophagefferens, which thrives and forms massive blooms in turbid coastal waters with dynamic light radiation, high organic matter and metals concentrations⁵. Harmful algal blooms occur when the causative alga outgrows co-existing species due to competitive advantages in photoenergy harvesting (E), nutrient acquisition (N), defense against environmental stress, grazing, and microbial attacks (D), and sometimes sexual reproduction (S)- known as ENDS drivers- with photoprotection being one significant D component³⁹. While the high number of LHC genes in A. anophagefferens have been linked to the species’ ability to grow in a dim light environment⁵, we demonstrate here that at least some are crucial for protection against photodamages. Combined with the substantial genetic capacity for utilizing organic nutrients, trace metal, and vitamins^5,40,41, the large LHC protein repertoire means A. anophagefferens is strongly equipped in “END” and poised to thrive in its coastal habitat and form brown tides.

Sampling design and algal cultivation

A total of 20 phytoplankton sampling sites were established in the brown tide outbreak area (the northwestern part of the Bohai Sea, China) (Fig. 1a, Supplementary Table 7). Samples were collected monthly from June 2013 to May 2014. At each site, phytoplankton samples were collected from the surface layer (1 m) and the bottom layer (1 m above seafloor), and preserved in 1.5% Lugol’s solution. Phytoplankton species identification and counting were conducted under a microscope (Olympus CX31, Japan). Dominant species was classified based on the dominance index (Y ≥ 0.02). Light intensity was measured for 35 sampling sites at the water surface, 1 meter below the water surface, and at the bottom. Light intensities were measured with a PAR sensor (Onset HOBO MX2202, USA).

A. anophagefferens used in this study was isolated from the coastal waters of Qinhuangdao in the Bohai Sea in July 2012 during a brown tide event. It was preserved in the algal collection of the Research Center of Harmful Algae and Marine Biology at Jinan University, China. Cultures were grown in artificial seawater supplemented with f/2 nutrients⁴² at 20 °C, under a light intensity of 50 μmol photons m⁻² s⁻¹ with a 12:12 light:dark cycle, and experiments were initiated during the logarithmic growth phase. T. pseudonana (CCMP 1335) was cultivated under the same conditions. Wild-type and two genotypes of T. pseudonana were maintained in semi-continuous cultures under 50, 200, and 800 μmol photons m⁻² s⁻¹. Cultures were diluted daily according to their respective growth rates. Steady state was reached at final dilution rates of 0.37, 0.60 and 0.50 d⁻¹ under 50, 200 and 800 μmol photons m⁻² s⁻¹, respectively. To accurately calculate growth rate, one-mL samples were collected daily before and after dilution from each culture and fixed using Lugol’s solution. Cells were counted using a hemocytometer under an inverted microscope (BX53, Olympus, Tokyo, Japan). Growth rates (μ, in d⁻¹) were calculated as: μ = (lnN₂ – lnN₁) / (t₂ – t₁), where N₂ and N₁ represent the cellular concentrations at t₂ (before dilution on day 2) and t₁ (after dilution on day 1), respectively.

Experiment with different light conditions and fluorescence analyses

Different light environments were used for the study of algal responses to light, including low light (LL, 50 μmol photons m⁻² s⁻¹), high light (HL, 800 μmol photons m⁻² s⁻¹), white light (WL, 100 μmol photons m⁻² s⁻¹), blue light (BL, 100 μmol photons m⁻² s⁻¹ at 450 nm), and darkness (D). The growth of T. pseudonana and its AaLhc-overexpressing strains were conducted under an irradiance of 50, 200, and 800 μmol photons m⁻² s⁻¹. The determination of NPQ capacity was conducted using a Phyto-PAM Phytoplankton Analyzer (Walz, Germany). After different algal strains were calibrated to the same chlorophyll a concentration, kinetic NPQ measurements were initiated. In the experiment on A. anophagefferens, NPQ was induced by 10 min of 764 μmol photons m⁻² s⁻¹, followed by a 10-min recovery phase in darkness to provide information on the relaxation kinetics. After 5 min high light exposure, 1 mM and 20 mM NH₄Cl were added, respectively. In the experiment on T. pseudonana and its AaLhc-overexpressing mutants, NPQ was induced by 30-min exposure to a photon flux of 764 μmol photons m⁻² s⁻¹, followed by a 30-min recovery phase in the dark. The experiment was conducted with 3–5 parallel samples. NPQ was calculated as Fm/Fm’ −1.

Pigment analysis

A. anophagefferens in the exponential growth phase was collected for pigment detection at 0 h, 0.5 h, and 1 h under HL and LL irradiation. NH₄Cl was added 25 min after HL exposure. Samples of T. pseudonana and its mutants in the exponential growth phase were taken at 0 h and 0.5 h under HL irradiation and after an additional 0.5 h in darkness. Under identical experimental conditions, 40 μM DCMU was added to inhibit linear electron transport. The samples obtained by filtration (25 mm GF/F membrane, 0.7 µm pore size) were immediately flash-frozen in liquid nitrogen and stored at −80 °C. Pure methanol was added for pigment extraction. Pigment analysis was performed using a 1200 Series HPLC system (Agilent Technologies, Santa Clara, CA, USA) equipped with an automatic sampler (Agilent 1100 series G1329A). The system was fitted with a reversed-phase Inertsil C8 column (150 × 4.6 mm, 5 μm particle size; GL Science) and eluted with mobile phase A (methanol: acetonitrile: 0.25 M pyridine solvent, 50: 25: 25, v: v: v) and mobile phase B (methanol: acetonitrile: acetone, 20: 60: 20, v: v: v)⁴³. Elution was performed following the method described in ref. ⁴⁴. with a flow rate of 0.9 mL min^-1.

RNA-seq

In the high-light induction experiment, algae cells cultured under LL were transferred to a HL environment for 1 h, 3 h, and 6 h. Additionally, cells that had been kept in darkness for 12 h were transferred to WL (1 h and long-term) and BL (1 h and long-term). Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, US). After the total RNA was qualified, RNA-seq libraries was sequenced in BGI-tech (Shenzhen, China) using an Illumina NovaSeq 6000 (San Diego, CA, USA) at Sangon Biotech (Shanghai, China). Sequencing quality was assessed using FastQC and MultiQC^43,45. For the significance of gene expression difference, a cut off of: qValue < 0.05 and |log2FC | ≥ 1 was used. RNA-seq data were uploaded to the NCBI Sequence Read Archive (SRA) database (PRJNA680884 and PRJNA787041).

WGCNA

The gene expression matrix, which contained HL and LL (1 h, 3 h, and 6 h), was employed to perform the gene co-expression network analysis using WGCNA⁴⁶, followed by the identification of the modules that clustered with LHC-encoding genes. Similarly, the gene co-expression network analysis was also performed within BL (1 h and long-term), WL (1 h and long-term) and D (darkness).

Real-time quantitative PCR (RT-qPCR)

In the high-light induction experiment, the process involves transferring from LL to HL for 0.5 h, 1 h, 2 h, and 3 h, then exposing to LL again (0.5 h, 1 h, 2 h, 3 h), followed by HL exposure (0.5 h, 1 h, 2 h, 3 h). For low light experiments, the transfer is from D to LL for 1 h, 2 h, and 3 h, with the LL + D group receiving LL exposure for 1 h followed by D for 1 h and 2 h. In the blue light induction experiment, the transfer is from WL to BL for 1 h of exposure, followed by 1 h of WL exposure. For the validation of the exogenous genes AaLhc27E and AaLhc27Q in T. pseudonana, HL exposure was applied to the algae cells cultured under LL conditions for 0.5 h, 1 h, and 3 h. Algal cells were collected by centrifugation and performed RNA extraction. RNA was reverse transcribed using the HiScript II 1st Strand cDNA Synthesis Kit (Vazyme-Biotect, China). Reverse transcription PCR was performed using 50 ng cDNA with the AceQ qPCR SYBR Green Master Mix (Vazyme-Biotect, China). RT-qPCR was performed with the CFX Connect Real-Time PCR Detection System (Bio-Rad, CA, USA). The 18S rRNA and ACT4 were used as reference genes for experiments with A. anophagefferens and T. pseudonana, respectively. The gene-specific primers used for this study were listed in Supplementary Table 8. Experiments were repeated three times independently.

Prediction of the LHC protein structure

AlphaFold3 [https://alphafoldserver.com/] was used to predict the three-dimensional structures. The 50 AaLHC prediction models are selected based on predicted template modeling (pTM) score (range from 0.62 to 0.92) for the best ranking. A pTM score above 0.5 indicates that the predicted fold might be the true structure⁴⁷. The structures were visualized using PyMOL [http://www.pymol.org].

Phylogenetic analysis

A total of 151 amino acid sequences of LHC proteins were obtained by the NCBI database (https://www.ncbi.nlm.nih.gov/) and JGI genome portal (https://genome.jgi.doe.gov/) from A. anophagefferens, Bigelowiella natans, Chlamydomonas eugametos, C. reinhardtii, Cyclotella cryptica, I. galbana, Karlodinium micrum, Mesostigma viride, Micromonas sp., Ostreococcus lucimarinus, Ostreococcus tauri, P. tricornutum, Physcomitrella patens, Scenedesmus obliquus and T. pseudonana (Supplementary Table 9)⁴⁸. These sequences were aligned using MUSCLE (v5.1)⁴⁹ and Clustal Omega (v1.2.4)⁵⁰ with default parameters. The maximum likelihood (ML) phylogenetic tree was estimated using IQ-TREE (v1.6.5)⁵¹. The recommended best-fitting model LG + F + R5 was selected by using ModelFinder in IQ-TREE (v1.6.5). ML analysis was run with 1000 replicates via a bootstrap test. WebLogo was used to build sequence logos from LHCs⁵².

Molecular dynamics (MDs) simulations

MDs simulations were performed using the Gromacs constant pH program⁵³, with the CHARMM36m force field selected, and the TIP3P water model was employed. The pHbuilder tool was used to generate topology files for CphHMD, with pH values set to 4.5 and 7, respectively, to determine all titratable protein residues in the protein. A cubic box with a distance of 1.5 nm from the protein was established, and water molecules were added. The pHbuilder tool was used to add an appropriate number of positive/negative ions (sodium ions/chloride ions) and buffer to ensure the system is a net-neutral system, and to generate the corresponding structural and topology files for the system. Four systems were constructed, namely AaLHC27E_pH_4.5, AaLHC27E_pH_7, AaLHC27Q_pH_4.5, and AaLHC27Q_pH_7, respectively. Each protein was solvated in a cubic box (1.5 nm buffer) with water molecules, followed by ion addition (Na⁺/Cl⁻) and buffer components to ensure charge neutrality. Energy minimization utilized the steepest descent algorithm (5000 steps) with Verlet cutoffs, PME electrostatics (rcoulomb = 1.2 nm, fourierspacing = 0.14), and force-switch vdW treatment (rvdw = 1.2 nm). Bond constraints were applied via LINCS. Equilibration comprised NVT (100 ps, leap-frog integrator, V-rescale thermostat at 300 K, tau_t = 0.5 ps) and NPT phases (100 ps, C-rescale barostat at 1 bar, isotropic coupling, tau_p = 5.0 ps, refcoord_scaling = com). Production runs employed CpHMD with identical parameters to NPT, excluding refcoord_scaling. During the 1000 ns production simulations, conformational snapshots were recorded at 1 ns intervals for all systems. System stability was assessed using built-in modules for root-mean-square deviation (RMSD), residue fluctuations (RMSF), radius of gyration (R_g), and solvent-accessible surface area (SASA). Markov state models (MSMs) and time-lagged independent component analysis (tICA) were computed via PyEMMA⁵⁴. To encode secondary structural integrity, protein backbone root-mean-square deviation (RMSD) relative to the energy-minimized reference structure was employed as the primary feature for tICA dimensionality reduction. This RMSD-based metric explicitly captures global conformational variations while preserving secondary structure signatures.

Construction of vectors for Lhc and overexpression

pTHa-K1 vector, containing the fcp8 promoter from T. pseudonana, was used for overexpression⁵⁵. The encoding gene sequence of AaLhc27 was cloned into the pTHa-K1 vector to generate a pTHa-AaLhc27E plasmid. The Mut Express II Fast Mutagenesis Kit V2 (Vazyme-Biotect, China) was used for site-directed mutagenesis of E102 and E210 to Q102 and Q210, respectively, to generate another pTHa-AaLhc27Q plasmid. Plasmids were transformed into wild-type T. pseudonana cells by microprojectile bombardment using the Bio-Rad Biolistic PDS-1000/He Particle delivery system (Bio-Rad, CA, USA)⁵⁶. With the 100 μg mL⁻¹ nourseothricin resistance screening, colonies appeared after 14 days. The colonies were verified for successful transformation through PCR sequencing.

Plasmid construction, expression, purification and refolding of AaLHC27E and AaLHC27Q

The pET-28a-AaLHC27E and pET-28a-AaLHC27Q vectors containing a C-terminal His6-tag were transformed into Escherichia coli BL21 (DE3) cells, respectively. All transformants were incubated in LB medium with 100 mg/L kanamycin at 37 °C. To induce protein expression, isopropyl-beta-D-thiogalactopyranoside (IPTG) was added to a final concentration of 1 mM when the bacterial solution was cultured to OD 600 = 0.6. Cells were incubated at 37 °C for 5 h, collected by centrifugation at 4 °C and resuspended in lysis buffer (50 mM Tris-base, 500 mM NaCl, 8 M Urea, pH 7.4, Sangon-Biotect, China). After sonication and centrifugation (10,000 × g, 4 °C, 30 min), the supernatant was collected. His-tag Protein Purification Kit (Beyotime, China) was used to purify the unfolded AaLHC27E and AaLHC27Q. Then exchange buffer to 50 mM Hepes, 0.1% LDS, pH 5.0 (Dialysis Membranes, 14 kDa, Beyotime, China) for removing the high concentration of urea and adjusting the pH. Recombinant proteins of AaLHC27E and AaLHC27Q were refolded following the method described in refs. ^57,58. SDS-page gel electrophoresis analysis (Sangon-Biotect, China) was carried out to check the yield of recombinant protein at each step (Supplementary Fig. 7).

FTIR spectroscopy

Samples of AaLHC27E and AaLHC27Q were loaded in a Specac Omni Cell demountable cell with calcium fluoride (CaF₂) windows with a 6 μm path length Mylar spacer. Infrared spectra were recorded using an FTIR spectrometer (IRTracer-100, Shimadzu, Japan) equipped with a Deuterated L-Alanine Triglycine Sulfate detector. Samples were solubilized in D₂O detergent buffer. Measurements for the pD 5.0 and 7.5 samples were conducted sequentially, and differential FTIR spectra were generated by normalizing to the integrated absorbance of the amide I band (1550–1750 cm⁻¹).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.