Light-controlled carotenoid transfer between water-soluble proteins related to cyanobacterial photoprotection

Abstract

Carotenoids are lipophilic pigments with multiple biological functions from coloration to vision and photoprotection. Still, the number of water-soluble carotenoid-binding proteins described to date is limited, and carotenoid transport and carotenoprotein maturation processes are largely underexplored. Recent studies revealed that CTDHs, which are natural homologs of the C-terminal domain (CTD) of the orange carotenoid protein (OCP), a photoswitch involved in cyanobacterial photoprotection, are able to bind carotenoids, with absorption shifted far into the red region of the spectrum. Despite the recent discovery of their participation in carotenoid transfer processes, the functional roles of the diverse family of CTDHs are not well understood. Here, we characterized CTDH carotenoproteins from Anabaena variabilis (AnaCTDH) and Thermosynechococcus elongatus and examined their ability to participate in carotenoid transfer processes with a set of OCP-derived proteins. This revealed that carotenoid transfer occurs in several directions guided by different affinities for carotenoid and specific protein–protein interactions. We show that CTDHs have higher carotenoid affinity compared to the CTD of OCP from Synechocystis, which results in carotenoid translocation from the CTD into CTDH via a metastable heterodimer intermediate. Activation of OCP by light, or mutagenesis compromising the OCP structure, provides AnaCTDH with an opportunity to extract carotenoid from the full-length OCP, either from Synechocystis or Anabaena. These previously unknown reactions between water-soluble carotenoproteins demonstrate multidirectionality of carotenoid transfer, allowing for efficient and reversible control over the carotenoid-mediated protein oligomerization by light, which gives insights into the physiological regulation of OCP activity by CTDH and suggests multiple applications.

Abbreviations

∆NTE

OCP with the NTE deleted, holoprotein

Anabaena

Anabaena variabilis PCC 7120

AnaCTDH

CTDH from Anabaena variabilis

AnaOCP

OCP from Anabaena variabilis

apo

apoform

BV

blue-violet fraction

CAN

canthaxanthin

CTD

C-terminal domain

CTDH

CTD homolog of OCP

ECN

echinenone

FRP

fluorescence recovery protein

HCP

helical carotenoid protein, NTD homolog of OCP

holo

holoform

Native PAGE

PAGE under nondenaturing conditions

NTD

N-terminal domain

NTE

N-terminal extension (residues 1–20 comprising the αA helix)

OCP^AA

OCP with amino acid substitutions Y201A and W288A mimicking the photoactivated state, holoprotein

OCP-CTD

CTD of OCP, holoprotein

OCP

orange carotenoid protein, holoprotein

RV

red-violet fraction

SEC

size-exclusion chromatography

Synechocystis

Synechocystis sp. PCC 6803

SynOCP-NTD

NTD of SynOCP (amino acids 1–164), holoprotein

SynOCP

OCP from Synechocystis sp.

TeCTDH

CTDH from Thermosynechococcus elongatus

Introduction

Carotenoids are hydrophobic pigment molecules derived from the isoprenoid metabolism and are involved in various processes allowing living organisms to interact with light and perceive visual information [1] but also are valuable dietary components [2]. The over 700 carotenoids in nature not only provide for the polychromatic coloration of fruits, plants, algae, and animals [3-5] but more importantly, they are responsible for light harvesting, photosensory, and photoprotective reactions as well as defense mechanisms against oxidative stress [6, 7]. Their tremendous role notwithstanding, there is a limited number of water-soluble carotenoid-binding proteins described to date [5, 8], and knowledge about carotenoid transport and carotenoprotein maturation processes is still limited.

The unique carotenoid transfer mechanism between water-soluble proteins related to cyanobacterial photoprotection was discovered in 2017 [9, 10]. Many cyanobacteria, the photosynthetic microorganisms that transformed our planet because of the pioneering oxygen production [11], employ the so-called orange carotenoid protein (OCP) for protecting their sensitive photosystems against the harmful effects of excessively absorbed light [12-14]. OCP is a blue light-triggered photoswitch employing a ketocarotenoid as cofactor, which upon absorption of a blue photon (450–500 nm) undergoes photoconversion from the basal, dark-adapted orange state (OCP^O) to the active red state (OCP^R), the latter being able to interact with the phycobilisome antenna complexes to consequently quench their fluorescence and prevent photodamage [15].

Orange carotenoid protein is a ~ 35 kDa water-soluble protein subdivided into two structural domains which encapsulate a single xanthophyll molecule in a common central cavity. The xanthophyll is 3′-hydroxy-echinenone (ECN) when purified from native cyanobacteria (e.g., Synechocystis sp. PCC 6803, further Synechocystis, or Arthrospira maxima [16, 17]), but photoactivity is also preserved with only 4(4′)-ketolated xanthophylls (either mono-ketolated ECN, or the diketolated canthaxanthin, CAN) [18-20], whereas insertion of β-carotene or (3,3′-hydroxylated) zeaxanthin results in nonphotoconvertible protein [20]. The OCP^O→OCP^R photoactivation is triggered by the absorption of blue light, making OCP an excellent light intensity sensor over the whole range of ambient light intensities [21, 22]. The OCP^R→OCP^O back-conversion occurs spontaneously in the dark, but it can be accelerated by the fluorescence recovery protein (FRP) [23-25]. FRP controls the activity of OCP via direct protein–protein interactions with the photoactivated OCP and protein variants mimicking this state [26-30].

In the basal OCP^O state, the protein adopts a compact structure [31] with the all-α N-terminal domain (NTD) and the mixed α-helical/β-sheet C-terminal domain (CTD) that belongs to the widespread family of nuclear transport factor 2 (NTF-2)-like proteins (Pfam 02136). The OCP^O structure is stabilized by multiple protein–protein interactions across the NTD-CTD interface, by the interdomain linker, and by the N-terminal extension (NTE), which directly contacts the CTD [31], as well as the protein–carotenoid interactions including two H-bonds between the keto oxygen atoms of the carotenoid and the hydroxyl group of Tyr201 as well as to the imino nitrogen of Trp288 in the CTD (amino acid numbering according to Synechocystis OCP) [31, 32]. These amino acids are highly conserved in all known OCP sequences, as their mutation severely interferes with the ability to photoswitch and changes spectroscopic and other physicochemical properties of the protein [28, 33, 34]. A recent dynamic crystallography study proposed that the aforementioned H-bonds are the first to break down during OCP photoactivation [35]. Likewise, the Trp288Ala mutation of Synechocystis OCP was shown to cause protein transformation into an OCP^R-like form with separated protein domains being able to constantly interact with FRP [26, 28]; the double Tyr201Ala/Trp288Ala mutation caused a similar effect [27, 34]. Very recent time-resolved spectroscopy studies revealed a complex hierarchy of times associated with changes in carotenoid and protein components in OCP upon its photoactivation [34, 36].

Separation of the NTD and CTD accompanying OCP photoactivation is in perfect agreement with the modularity of OCP structure and function [37]. Reassembly of the recombinantly produced individual domains, which stably exist as individual carotenoproteins in vitro, allowed researchers to reconstruct metastable photoactive OCP-like species [10, 38], suggesting that a similar process could have arisen in evolution giving birth to the first ancestral OCPs [10, 38], and to reveal the possibility of carotenoid shuttling between the domains [10].

For both domains of OCP, multiple extant homologs exist in nature: up to nine clades of NTD homologs (called helical carotenoid proteins, HCPs) [39, 40] and at least two clades of C-terminal domain homologs (CTDHs) [41, 42], which exist in different cyanobacteria in various combinations regardless of whether full-length OCPs are present or not [12, 39, 41]. For example, T. elongatus BP-1 contains only one HCP (clade HCP4/5), one CTDH (clade 1), and no OCP, whereas Anabaena sp. PCC 7120 (further Anabaena) contains four HCP paralogs (HCP1–HCP4), one CTDH (clade 2), and one OCP. OCPs themselves can belong to different clades (e.g., OCP1, OCP2, and OCPx) [43].

Although some HCPs [39, 40] and CTDHs [41, 44] have been structurally and biochemically characterized, their function in vivo remains largely speculative. It was demonstrated that the isolated HCPs of Anabaena may have different roles in photoprotection: HCP4 interacts with phycobilisomes and induces excitation energy quenching, HCP2 and HCP3 are good quenchers of singlet oxygen, and HCP1 might serve as a carotenoid carrier [40]. CTDHs are very good singlet oxygen quenchers, efficiently extract carotenoids from membranes, and transfer them to the apoforms of OCP and HCPs [41], as initially demonstrated for the recombinant Synechocystis OCP-CTD (also called C-terminal OCP-related Carotenoid Protein, COCP) [9, 10]. The C-terminal tail of CTDH was proposed to be crucial for carotenoid binding, uptake from membranes, and transfer to other proteins [44]; however, the molecular details and prerequisites for these processes remain insufficiently understood.

By comprehensive characterization of the apo- and holoforms of CTDHs from Anabaena and T. elongatus and their interaction with various individual OCP domains and mutant variants, we found that carotenoid transfer is multidirectional, depending on the relative carotenoid affinities of the transfer partners and the efficiency of protein–protein interactions. Of note, this study demonstrates a photoinduced, reversible carotenoid shuttling between OCP and CTDH, suggesting the hypothesis that CTDH·OCP interactions may contribute to an OCP-related photoprotective mechanism in CTDH-containing cyanobacteria independently of the FRP-based process. In addition, these results forward the future design of light-triggered protein–protein interaction cascades and synthetic biology applications, as well as the construction of antioxidant nanocarriers.

Results

Characterization of the apo and holoforms of CTDH proteins

The CTDH from Anabaena (AnaCTDH) produced in Escherichia coli in the absence of carotenoids (AnaCTDH(apo)) was purified under reducing conditions as a predominantly monomeric species (apparent Mw ~ 17 kDa; calculated monomer Mw 16.1 kDa) with a propensity to form disulfide-linked dimers (apparent Mw ~ 33 kDa) upon oxidation (Fig. 1) due to the presence of Cys103, which is typical of clade 2 CTDHs, in agreement with [41]. AnaCTDH produced in an ECN/CAN-producing E. coli strain [18] yielded a violet-purple carotenoprotein [41], which according to preparative size-exclusion chromatography (SEC) dissected into two poorly resolved fractions of different colors (red-violet, ‘RV’, or blue-violet, ‘BV’) in the elution profile (Fig. 1A). Analytical SEC of the fractions obtained during the preparative SEC confirmed the overlap of at least two carotenoid-containing peaks (Fig. 1B), characterized by differences in the visible absorbance spectrum (Fig. 1C).

Figure 1

Characterization of AnaCTDH preparations obtained from Escherichia coli producing ECN/CAN. (A, B) 2D SEC of the AnaCTDH holoprotein preparation [A—preparative SEC on a Superdex 200 26/600 column at a 1.6 mL·min⁻¹ flow rate monitored using protein (297 nm) and carotenoid-specific (500 nm) absorbance, B—analytical SEC on a Superdex 200 Increase 5/150 column at a 0.45 mL·min⁻¹ flow rate monitored at either 280 or 540 nm]. Arrows indicate the direction of the run, apparent Mw values for the main peaks are indicated in kDa based on column calibration. The inset shows a segment of the preparative SEC column during the run. (C) Absorbance spectra of the two main fractions of AnaCTDH(holo) obtained, with the natural color of the samples indicated in the insert. Note the presence of the ~ 409 nm peak characteristic of only the ECN-bound form of AnaCTDH. Chemical formulae of ECN and CAN are depicted below the spectra. (D) Thin layer chromatogram of the two fractions, RV and BV, revealing the predominant carotenoid types [53]. (E) Native PAGE analysis of the RV and BV fractions showing different electrophoretic mobility (marked by arrowheads). Note the slight cross-contamination due to incomplete separation of these fractions on preparative SEC.

Analysis of the carotenoid content by thin layer chromatography (Fig. 1D) revealed that the BV fraction predominantly contained CAN, whereas the RV fraction contained ECN. CAN bound to the AnaCTDH dimer showed the greatest red shift among all OCP-related species with an absorption maximum above 560 nm [41], which is by ~ 10 nm more red shifted compared to OCP-CTD(CAN) [9, 10]. These spectral properties hint at distinct differences in the specific environment formed by the corresponding protein matrix. Given the almost identical energy gap for S₀–S₂ excitation of ECN and CAN in organic solvents (~ 460 nm) [45], the about 30-nm bathochromic shift of AnaCTDH(CAN) relative to AnaCTDH(ECN) indicates different carotenoid configurations (conjugation length) and carotenoid–protein interactions. Another spectral signature characteristic of ECN-containing AnaCTDH species is a small peak at ~ 409 nm (Fig. 1C), probably corresponding to cis isomerization [46]. In contrast to the symmetric CAN molecule with two ketolated β-ionone rings, ECN cannot simultaneously form hydrogen bonds with Tyr and Trp residues (201 and 288 in Synechocystis OCP numbering, respectively) on both ends of the AnaCTDH homodimer. Of note, a similar hypsochromic shift was reported due to the Trp288Ala mutation in OCP-CTD [9], supporting the potential role of the H-bonds in spectroscopically relevant carotenoid–protein interactions within the CTD (or CTDH) protein. The different dimer stabilities and apparent sizes resulted in different electrophoretic mobility of AnaCTDHs from the RV and BV fractions on nondenaturing (native) PAGE (Fig. 1E).

On analytical SEC, AnaCTDH(CAN) eluted as dimers (Mw of 31–33 kDa) almost independent of protein concentration, whereas AnaCTDH(ECN) demonstrated a pronounced shift upon dilution corresponding to an Mw drop from ~ 29 to ~ 18 kDa (Fig. 2A) suggesting dimer dissociation. Most likely, the absence of the second keto-group in ECN hinders coordination of a second AnaCTDH monomer, as in the case of the OCP-CTD^W288A mutant, which is lacking one of the ketocarotenoid-coordinating residues [9]. This supports the idea that H-bonds between the ketogroups of CAN and the conserved Tyr/Trp residues in AnaCTDH (as well as in individual OCP-CTD [9]) contribute significantly to dimer stability. Notably, in agreement with [9, 41], we also observed a fraction of AnaCTDH(holo) monomers (~ 18 kDa) retaining carotenoid molecules, which exhibited absorbance spectrum with a maximum at ~ 505 nm, unusual for CTDH (Fig. 2B).

Figure 2

Analysis of the carotenoid-coordinating AnaCTDH species by SEC. (A) The concentration-dependent behavior of the BV and RV fractions. While the BV fraction containing mostly CAN is stable to dilution (inset on the left), the RV fraction enriched in ECN shows a remarkable dissociation upon dilution (the comparison is given in the inset on the right). (B) The analytical SEC profile of the apparently monomeric fraction obtained during preparative SEC. The color of the preparation and its visible absorbance spectrum are given in the inset.

The behavior of the recombinant CTDH from T. elongatus (TeCTDH), which represents the other phylogenetic clade of CTDHs (clade 1) containing a Phe residue in a position homologous to Cys103 of clade 2 [41], was similar to that of AnaCTDH. Even in the absence of Cys103, the TeCTDH apoform gave dimeric and monomeric peaks on the SEC profile showing two distinct bands on native PAGE representing two species in equilibrium with each other (Fig. 3A).

Figure 3

Characterization of TeCTDH samples. (A) Preparative SEC on a Superdex 200 26/600 column of either apoform or holoform of TeCTDH produced in Escherichia coli in the absence or presence of the carotenoids, respectively. The elution profile was run at a 1.6 mL·min⁻¹ flow rate while recording either protein- or carotenoid-specific absorbance (wavelengths are indicated). The insert shows the electrophoretic mobility of the apoform (1) and the holoform (2) TeCTDH under nondenaturing conditions. (B) Analytical SEC profiles of different fractions collected during the preparative SEC run. Forty microliters of each fraction were loaded on a Superdex 200 Increase 5/150 column and run at 0.45 mL·min⁻¹ flow rate while following either protein- or carotenoid-specific absorbance (wavelengths are indicated). The apparent Mw values corresponding to the main peaks were obtained from column calibration (indicated in kDa). (C) Absorbance spectra of the two representative fractions of TeCTDH containing predominantly CAN (BV) or ECN (RV). Note the presence of the ~ 407 nm peak characteristic of only the ECN-bound form of TeCTDH. (D) The analytical SEC profile of the apparently monomeric carotenoid-containing fraction of TeCTDH obtained during preparative SEC. The color of the preparation and its visible absorbance spectrum are given in the inset.

Like AnaCTDH, TeCTDH expressed in E. coli in the presence of ECN/CAN yielded on the SEC profile the main overlapping peaks corresponding to CAN- and ECN-containing dimers with different apparent size (Fig. 3B) and ~ 30-nm red-shifted visible absorbance spectra (Fig. 3C). The ECN-containing peak represented a dimer–monomer mixture sensitive to dilution, and a stable fraction of carotenoid-carrying TeCTDH monomers with an absorbance maximum at ~ 506 nm could be isolated (Fig. 3D).

Thus, the two CTDH proteins from different phylogenetic clades (1 and 2) [41] share the ability to coordinate not only diketolated CAN but also mono-ketolated ECN, by forming dimeric holoforms with substantially different stability and oligomerization behavior. Both paralogs are able to form monomeric species bearing carotenoids, which are spectrally characterized by the absence of vibronic structure and a significantly blue-shifted maximum compared to other OCP-related proteins described.

Direct carotenoid transfer from CTDH to OCP

Previous work showed that, when expressed in the E. coli strain producing a mixture of ECN and CAN [18], OCP-CTD binds almost exclusively CAN [10] and can transfer it to the OCP apoform [9]. Likewise, AnaCTDH was shown to transfer CAN to the apoforms of OCP, NTD, or HCPs, which may be negatively regulated by the formation of the disulfide bridge involving Cys103 [41]. Here, by using fully reduced protein preparations, we found that not only AnaCTDH(CAN) but also AnaCTDH(ECN) was able to transfer the embedded carotenoid to the OCP apoforms of both Synechocystis and Anabaena (SynOCP and AnaOCP). In both cases, this led to substantial spectral changes and appearance of pronounced vibronic structure, as typical for OCP^O (Fig. 4). In contrast to OCP-CTD, which under our conditions transferred ~ 84% of carotenoid to SynOCP and ~ 87% to AnaOCP, AnaCTDH was found to be a much less efficient carotenoid donor to both SynOCP [57% of CAN is transferred by AnaCTDH(CAN) and 68% of ECN is transferred by AnaCTDH(ECN)] and AnaOCP [only 31% of CAN is transferred by AnaCTDH(CAN) and 58% of ECN is transferred by AnaCTDH(ECN)] (Fig. 4).

Figure 4

Direct carotenoid transfer from OCP-CTD(CAN) (A, D), AnaCTDH(CAN) (B, E), or AnaCTDH(ECN) (C, F) to the apoforms of either SynOCP (A–C) or AnaOCP (D–F) followed by changes of absorption spectra at 20 °C during 3.5 h recorded with 10-min intervals. The protein concentration of both carotenoid donors and acceptors was 20 μm. Insets show natural colors of the samples before (b) or after (a) the transfer completion. Note that the process reached equilibrium in each case as no changes occurred after overnight incubation. Decomposition of the resulting spectra using the initial donor spectra and a spectrum of OCP^O as a reference allowed us to estimate the fraction of the initial carotenoid-bound species (shown with the filled area under the curve and the percentage of remaining carotenoid donors).

In agreement with the SEC data, the lower stability of AnaCTDH(ECN) dimers may explain the more efficient carotenoid transfer to both OCPs compared to AnaCTDH(CAN) (Fig. 4). The higher transfer efficiency from OCP-CTD(CAN) (also called COCP [10]) suggests that the specificity of NTD·CTD (or NTD·CTDH) interactions may be important for the carotenoid transfer and that the AnaCTDH(CAN) dimer might have a larger affinity for the carotenoid than OCP-CTD(CAN). This prompted us to test whether AnaCTDH can take up the carotenoid directly from OCP-CTD(CAN).

‘Horizontal’ carotenoid transfer from OCP-CTD to CTDH

Incubation of Synechocystis OCP-CTD(CAN) with AnaCTDH(apo) resulted in an efficient color change from RV, typical of OCP-CTD(CAN) [10], to BV, typical of AnaCTDH(CAN) (Fig. 1C), indicating at least partial CAN translocation from OCP-CTD to AnaCTDH. This carotenoid transfer between homologous proteins, which we tentatively call ‘horizontal’, was accompanied by a significant shift of the absorbance maximum from ~ 552 to ~ 569 nm (Fig. 5A). Recombinant OCP-CTD from Anabaena was also able to transfer the carotenoid to AnaCTDH, producing nearly identical spectral changes, with the final spectrum being equivalent to that of AnaCTDH(CAN) (Fig. 5B).

Figure 5

Horizontal carotenoid transfer from OCP-CTD(CAN) to AnaCTDH(apo). (A) Normalized absorption spectra of the SynOCP-CTD(CAN) sample and the product of carotenoid transfer to AnaCTDH(apo). The inset shows the natural color of the samples. (B) Equivalent spectra for AnaOCP-CTD(CAN) used as the carotenoid donor to AnaCTDH(apo). The spectrum of AnaCTDH(CAN) is given as a control. (C) Analysis of the horizontal carotenoid transfer at various AnaCTDH/OCP-CTD(CAN) molar ratios (indicated below) by native PAGE without staining. The gel was run after the completion of the transfer (ON, overnight) and thus represent near equilibrium conditions. (D) The same experiment was done for the carotenoid transfer to TeCTDH(apo) and both gels were analyzed by quantitative densitometry. Error bars represent standard deviation (n = 3). (E) Analysis of the horizontal carotenoid transfer from SynOCP-CTD(CAN) to AnaCTDH (111 μm per monomer each) at 37 °C at different time points by means of the unstained native PAGE. The main bands are marked by arrowheads. Note the intermediate band appearing very early upon mixing OCP-CTD(CAN) and AnaCTDH and putatively corresponding to their heterodimers sharing the carotenoid (ON, overnight).

The horizontal carotenoid transfer could be followed by native PAGE without requiring any staining, owing to the difference in electrophoretic mobility of the carotenoid donor and acceptor (Fig. 5C). Indeed, titration of a fixed OCP-CTD(CAN) concentration (111 μm per monomer) by increasing quantities of AnaCTDH(apo) resulted in a disappearance of the colored OCP-CTD(CAN) band with a concomitant appearance of the colored AnaCTDH(CAN) band, which differed by a tint of violet (Fig. 5C). Already at a 1 : 1 molar ratio and equilibrium conditions (> 24 h incubation), we could observe almost quantitative carotenoid transfer from OCP-CTD(CAN) to AnaCTDH and to TeCTDH, demonstrating in both cases nearly identical titration curves (Fig. 5D).

In the course of the transfer, we could observe the appearance of two colored bands of higher mobility on native PAGE gels, suggesting the presence of intermediates, presumably, OCP-CTD(CAN)·AnaCTDH heterodimers, which appeared within minutes after mixing of the samples and disappeared during further incubation time (Fig. 5E). The principal existence of such heterodimers of intermediate mobility sharing a carotenoid molecule implied that the CTD·CTDH interactions may mediate other types of carotenoid transfer between related proteins, as an alternative to the NTD·CTD(H) interactions [9, 41].

Inverse transfer of carotenoid to CTDH

Previously, it was shown that Synechocystis OCP-CTD can transfer its CAN to the NTD apoform, resulting in the formation of the metastable photoactive OCP^O-like species [10], whereas recombination of the holoforms of NTD (or HCP) with the apoforms of OCP-CTD (or CTDH) did not work [9, 10, 41], suggesting that carotenoid transfer is a unidirectional process from CTD(H) to NTD (or HCP) and full-length OCP.

While the physiological relevance of the horizontal carotenoid transfer discovered here is unclear, yet, the possibility of carotenoid translocation from OCP-CTD to AnaCTDH strongly suggested high affinity of the latter to carotenoids and supported the idea that carotenoid transfer can occur in many directions, involving both NTD·CTD(H) and CTD·CTD(H) interactions. In line with this, it was found recently that AnaCTDH can retrieve carotenoid from AnaHCP1 (but not HCP2, HCP3, nor HCP4 [40]) suggesting that HCP1 may serve as an additional carotenoid carrier [44]. Therefore, we questioned whether the apparently high affinity of AnaCTDH may allow it to retrieve the carotenoid from other OCP-related proteins containing the NTD.

First, we examined the ability of CTDHs to retrieve the carotenoid directly from SynOCP-NTD and AnaHCP1 holoproteins. Under the conditions used, AnaCTDH extracted up to 86% of carotenoid from SynOCP-NTD and 55% of carotenoid from AnaHCP1 (Fig. 6A,B), as judged from spectral decomposition, confirming the high carotenoid affinity of AnaCTDH and implying favorable NTD·CTDH interactions. Under identical conditions, TeCTDH extracted 55% of carotenoid from AnaHCP1, whereas on the contrary, uptake from SynOCP-NTD was barely efficient (~ 3%; Fig. 6C,D). These findings emphasize the notion that the transfer process is not only guided by the affinity for carotenoid but also by protein–protein interactions.

Figure 6

Inverse carotenoid transfer from Synechocystis OCP-NTD (A, C) or Anabaena HCP1 (B, D) to either Anabaena (A, B) or Thermosynechococcus elongatus CTDH (C, D) followed by absorbance spectroscopy. The carotenoid donors (AnaHCP1, SynOCP-NTD) were present at concentrations equivalent to O.D. = 0.3 at 520 nm and AnaCTDH(apo) was added in a twofold molar excess to initiate the transfer. The samples were incubated at 23 °C for 40 min. Panel E demonstrates decomposition of the final spectrum from panel A used to derive the respective contributions from the reference spectra (indicated in %) and by this means to estimate the efficiency of the carotenoid transfer. Decomposition of other spectra was done similarly.

Given the modular organization of OCP [37] and the idea of domain separation accompanying OCP photoactivation [18, 33, 47-49], we next questioned whether AnaCTDH can retrieve carotenoids from full-length OCP variants.

Indeed, upon mixing of AnaCTDH(apo) with the analog of photoactivated OCP with separated domains, the OCP^AA mutant (Y201A/W288A) [27, 34], we observed pronounced changes in the absorption spectrum and a 30-nm bathochromic shift of its maximum from ~ 525 nm, characteristic of OCP^AA, to ~ 553 nm, characteristic of AnaCTDH(holo) (Fig. 7A). Spectral decomposition using the reference spectra of AnaCTDH(CAN) and OCP^AA indicated that, under these conditions, up to ~ 80% of the carotenoid was transferred in the opposite direction compared to the previously characterized process (i.e. CTD/CTDH → HCP/NTD/OCP [9, 10, 41]). This number is close to the efficiency of the transfer from SynOCP-NTD to AnaCTDH(apo) (Fig. 6A), which indirectly indicates that in the OCP^AA case, the carotenoid is extracted from the NTD, in line with the fact that upon OCP photoactivation, the carotenoid is translocated to the NTD [36, 48, 49]. The open conformation of OCP^AA may favor protein–protein interactions and carotenoid transfer to AnaCTDH, guided by the affinity to a carotenoid.

Figure 7

Inverse transfer of carotenoid to AnaCTDH studied by absorption spectrometry (A, B), native PAGE (C), and analytical SEC (D, E). Carotenoid transfer from OCP^AA (A) or ∆NTE-OCP^O (B) to AnaCTDH(apo) was followed at 30 °C over 4 h by recording absorption spectra each 6 min. AnaCTDH(apo) concentration was 16 μm, concentrations of carotenoid donors were adjusted to a similar absorbance at 500–525 nm. Note that the process reached equilibrium in each case as no changes occurred after overnight incubation. Arrows indicate the direction of spectral changes and the black line in panel B represents the difference spectrum (final–start). The vibronic structure characteristic of OCP^O in the ∆NTE^O case is marked by dashed lines with the position of the main bands indicated. The color gradient roughly corresponds to the sample colors. (C) Carotenoid transfer from various NTD-containing OCP-related proteins to AnaCTDH monitored by native PAGE without staining. The samples before (b) and after (a) the completion of the transfer (1 h at 37 °C) were analyzed. The products/intermediates of the carotenoid transfer are marked by asterisks. The position of the AnaCTDH(CAN) dimer band is shown by an arrowhead on the right. (D, E) Inverse carotenoid transfer from OCP^AA to AnaCTDH followed by subjecting 30-μL aliquots of the reaction mixture to analytical SEC monitored by protein-specific (D) and carotenoid-specific (E) absorbance. The position of AnaCTDH dimers and monomers (dashed lines), apparent Mw of the peaks (in kDa), and the direction of changes in the course of carotenoid transfer (arrows) are indicated.

The inverse transfer of carotenoid to AnaCTDH(apo) could be observed not only from purple OCP^AA but also from the orange compact OCP form lacking the NTE, that is, from the ∆NTE^O variant, albeit with limited efficiency (15–20% according to spectral decomposition; Fig. 7B). The product of carotenoid transfer, AnaCTDH(holo), gave a pronounced contribution to the absorbance at ~ 550 nm (Fig. 7B) and could be easily discriminated on an unstained native PAGE (Fig. 7C). In Fig. 7C, carotenoid transfer from SynOCP-NTD to AnaCTDH is also shown for comparison. Notably, besides the band matching the AnaCTDH(holo) control, the band with lower electrophoretic mobility could also be detected, especially in the case of the carotenoid transfer from OCP-NTD, most likely representing products and intermediates of the reaction, respectively (marked by two asterisks in Fig. 7C).

The inverse transfer of carotenoid could also be readily monitored by SEC. For example, mixing of OCP^AA and AnaCTDH(apo) resulted in an increase of the ~ 37-kDa peak corresponding to the carotenoid-bound AnaCTDH dimers and a concomitant decrease in the ~ 51-kDa peak of OCP^AA (followed by 560 nm) and the ~ 18-kDa peak of the AnaCTDH(apo) (followed by 280 nm; Fig. 7D,E), indicating physical carotenoid translocation from one protein to the other, in agreement with the data from native PAGE.

Photoinduced carotenoid shuttle between OCP and CTDH

Since we found that mixing of AnaCTDH(apo) and the OCP^AA mutant, which mimics the quenching-competent OCP^R form [27, 34], resulted in carotenoid translocation into AnaCTDH, as seen from the appearance of violet forms, we questioned if this process could be induced by actinic light, which is known to trigger domain separation in wild-type OCP [18, 33, 47-49]. Mixing of wild-type OCP^O with ApoCTDH(apo) had absolutely no effect on the absorption spectrum of the samples. This indicates that, as long as the OCP domains are most of the time closed, AnaCTDH cannot extract carotenoids, and the transfer occurs only upon destabilization of the compact OCP structure (either by the NTE removal or by mutations of the key Tyr/Trp residues in the CTD coordinating the carotenoid [28, 34, 50]). We then tested whether carotenoid transfer to AnaCTDH(apo) could be induced by photoactivation of OCP^O and performed experiments using SynOCP and AnaOCP holoproteins (Fig. 8). In both cases, after the completion of O–R conversion (it took ~ 10 s to reach the equilibrium), further slow changes of absorption in the red region of the spectrum were observed, associated with the accumulation of the violet forms in the course of the inverse carotenoid transfer process from OCP^R to CTDH (Fig. 8, ‘O’, ‘R’, and ‘V’ states). After ~ 30 min of continuous illumination, the absorption of the samples coincided with that of AnaCTDH(ECN) (see Fig. 1C), without any signatures of the orange or red OCP forms.

Figure 8

Photoinduced carotenoid shuttling between holoforms of SynOCP (A), AnaOCP (B) and AnaCTDH. AnaCTDH(apo) was added to a solution of the dark-adapted OCP^O holoforms to reach concentration ratio equal to four CTDH for each carotenoid molecule. Experiments were conducted at 30 °C, constant stirring and equivalent photon flux densities of actinic blue light (450-nm light-emitting diode, LED). In order to show hierarchy of rate constants and better comparison of A and B, zero on the timescale corresponds to the beginning of both light activated inverse (black squares) and direct carotenoid transfer (red circles), which was observed after the end of illumination by actinic light. Dashed and dotted lines show the main individual components of kinetics, obtained by fitting of the data by a sum of decaying exponential functions. (C) Schematic representation of the photoinduced carotenoid shuttle mechanism between OCP and CTDH. OCP^O expressed in Escherichia coli predominantly coordinates ECN [10]. Note that the process occurs via the intermediary formation of heterodimers. NTE is shown by a red rectangle, small violet and magenta circles designate key Trp and Tyr residues, respectively, involved in carotenoid–protein interactions by H-bonding to carotenoid keto-groups. NTD and CTD are shown in gray and white, respectively, CTDH is shown in light blue.

Thus, Anabaena CTDH can gradually extract the carotenoid from the photoactivated forms of OCP as soon as the domains are separated; quantitatively similar results were obtained for Synechocystis OCP and Anabaena OCP (see Fig. 8A,B). This directly suggests that potent modulatory effects are possible, when OCP and CTDH co-occur in vivo. Importantly, the process was reversible: after the actinic light was turned off, we observed a gradual recovery of the orange OCP forms as the result of the direct carotenoid transfer from AnaCTDH(holo) to the OCP apoforms (Fig. 8A,B, red circles).

Discussion

Lipophilic carotenoids are synthesized at and localized in membrane compartments, whereas OCP and related proteins are water-soluble carotenoproteins, which raises the question how carotenoid delivery to various destinations proceeds through aqueous media. Previous works have shown that the C-terminal OCP domain and its homologs, CTDHs, can efficiently retrieve carotenoids from membranes, and dimerize subsequently while encapsulating the carotenoid molecule. This suggests that these proteins can serve as carotenoid carriers to supply their cargo downstream to the carotenoid acceptors [9, 10, 41]. HCPs were proposed to be an example of such acceptors [41].

Developing further the recent finding that HCP1 from Anabaena can transfer its carotenoid to AnaCTDH(apo) [44], we demonstrate here that carotenoid transfer between OCP-related proteins is a multidirectional process at least in vitro. Both symmetric CAN- and asymmetric ECN-bound CTDH dimers take part in the initially characterized ‘direct’ carotenoid transfer to OCP from both Synechocystis and Anabaena similar to the CAN-bound OCP-CTD, leading to the formation of the photoactive OCP species (Fig. 4). However, the properties of the ECN- and CAN-bound CTDH dimers are remarkably different in terms of spectral characteristics, stability, and apparent size (Figs 1-3), which can probably explain why ECN is transferred more efficiently (Fig. 4). The substantially lower stability and the concentration-dependent dissociation of the CTDH dimers coordinating the asymmetric ECN (Fig. 2) with only one ketolated β-ionone ring resembled the lower stability of the OCP-CTD^W288A mutant lacking one of the H-bond donor residues [9]. This corroborates the fact that the H-bonds to the conserved Tyr/Trp residues play a significant role in carotenoid–protein interactions. Thus, the nature of the carotenoid is an important factor affecting the direction and efficiency of carotenoid transfer as well as the CTDH dimer stability.

While the binding mode and the configuration of the bulky lipophilic carotenoid within the individually stable, but relatively small monomeric CTDH species is completely unclear, the very existence of the latter may imply roles as intermediates of the carotenoid uptake and delivery processes [44].

The occurrence of the so-called horizontal carotenoid transfer between the OCP-CTD and its homologs, demonstrated here (Fig. 5), indicated that not only NTD·CTD(H) but also CTD(H)·CTD(H) interactions may mediate the translocation of carotenoid between OCP-related proteins. This, in turn, suggests that differences in the affinity of homologous proteins with a similar three-dimensional structure [44] and, probably, in the interior of their carotenoid-binding cavities are important prerequisites of carotenoid transfer. The efficiency of carotenoid transfer seems to be also dependent on specific matching between the proteins involved. For example, TeCTDH was capable of carotenoid transfer from AnaHCP1, but not from SynOCP-NTD (Fig. 6), likely associated with the fact that OCP is absent from T. elongatus [12] and thus the corresponding NTD–CTDH interactions may not have been preserved during the CTDH evolution. Structural elements responsible for such specificity remain to be determined but most likely expand beyond the previously considered difference in the β5/β6 loop length or the presence of the CTDH clade-specific Cys103 residue [41, 44].

According to our observations, the ‘inverse’ carotenoid transfer process, which leads to the formation of the violet-colored dimeric CTDH proteins, can efficiently occur not only from NTD or HCP1 but also from the full-length analog of the photoactivated OCP, OCP^AA (Fig. 7A). The inability of CTDHs to retrieve the carotenoid from OCP^O indicated that carotenoid transfer from OCP occurs via domain separation. Indeed, we found that carotenoid transfer from OCP to CTDHs can be reversibly triggered by OCP photoactivation, which led to the carotenoid-stabilized CTD·CTDH dimerization (Fig. 8C). Thus, the affinity for carotenoids in different OCP forms is predefined in such a specific manner that allows carotenoid uptake from membranes by CTD(H), and delivery of the carotenoid into the full-length OCP, the latter retaining the carotenoid in the basal orange form, but readily and reversibly donating it to CTDHs upon photoactivation. Of note, the interprotein carotenoid transfer processes described in this study proceed via formation of intermediates (heterodimers) sharing the carotenoid molecule, before the final product is formed. While this makes accurate quantitation a difficult task, it can nevertheless be expected that further detailed thermodynamic analysis will yield rewarding results.

A reversible, light-controlled carotenoid translocation and carotenoid-mediated protein dimerization allows hypothesizing that a similar process may take place in vivo to modulate the efficiency of the OCP-regulated photoprotection by controlling the amount of the quenching OCP form in cyanobacteria expressing CTDH. This may be especially relevant for cyanobacterial strains harboring OCP and CTDH, but lacking FRP genes (See Table 1) [43], and warrants further detailed investigation of the expression levels of CTDH proteins relative to the levels of OCP and HCP. It is worthwhile to mention that, although both SynOCP and AnaOCP can extract carotenoids from membranes, OCP from Synechocystis, which lacks a CTDH gene, does it much more efficiently than OCP from Anabaena, where the CTDH gene is present [41]. Protein engineering can further be used to modify the described photoinduced carotenoid transfer process for the future design of light-driven protein–protein interaction cascades and other synthetic biology applications.

Table 1. Examples of cyanobacterial strains containing OCP (OCP2 or OCPX types [43]), HCP and CTDH genes but lacking FRP genes. Cyanobacterial strains containing OCP1 type were not considered because, utilizing the classical FRP-regulated system, those strains either contain FRP or do not contain FRP, nor CTDH. Morphological subsections: (a) Chroococcales; (b) Pleurocapsales; (c) Oscillatoriales; (d) Nostocales; (e) Stigonomatales. The data were taken from [43]

Species	OCP type	HCP type	CTDH	FRP
(a) Gloeobacter PCC 7421	X	9	+	−
(a) Gloeobacter JS1	X	9	+	−
(a) Chamaesiphon PCC 6605	X	1, 2	+	−
(a) Gloeocapsa PCC 7428	2	1, 2, 3, 4, 6	+	−
(a) Synechocystis PCC 7509	2, X	1, 2, 3	+	−
(b) Pleurocapsa PCC 7327	2, X	4, 5, 7	+	−
(b) Chroococcidiopsis PCC 7203	X	1, 2, 3, 4, 6	+	−
(c) Microcoleus FGP-2	X	3, 8	+	−
(c) Oscillatoria PCC 7112	X	3, 8	+	−
(c) Microcoleus PCC 9802	X	3, 8	+	−
(c) Kamptonema PCC 6506	X	3, 8	+	−
(c) Kamptonema PCC 6407	X	3, 8	+	−
(c) Crinalium PCC 9333	2	2, 3, 8	+	−
(c) Microcoleus PCC 7113	X	3, 8	+	−
(d) Scytonema PCC 7110	X	1, 2, 3	+	−
(d) Rivularia PCC 7116	X	1, 4	+	−
(d) Tolypothrix UTEX 2349	X	1, 3, 6	+	−
(d) Nodularia CCY9414	X	4	+	−
(d) Microchaete PCC 7126	X	1, 3	+	−
(d) Calothrix PCC 7507	X	1, 3	+	−
(d) Nostoc ATCC 29133	X	1, 2, 3, 6	+	−
(d) Cylindrospermum PCC 7417	X	3, 6	+	−
(d) Anabaena PCC 7122	X	3	+	−
(d) Anabaena PCC 7108	X	3, 4	+	−
(d) Trichormus 708	X	1, 3, 4	+	−
(d) Dolichospermum AWQC310F	X	4	+	−
(d) Dolichospermum AWQC131C	X	4	+	−
(e) Mastigociadopsis PCC 10914	X	1, 2, 3, 4, 6	+	−
(e) Fisherella PCC 7521	2	2, 4	+	−
(e) Fisherella JSK-11	2	4	+	−
(e) Fisherella PCC 7414	2	1, 2, 3, 4	+	−

We anticipate that the principles of the protein-to-protein carotenoid transfer mechanism reported here may be generally applicable to other related processes involving water-soluble carotenoproteins, especially given the ubiquity of the NTF-2-like family of proteins. In the future, our findings may inspire the development of stable carotenoid nanocarriers for increased carotenoid resistance against photodestruction and for targeted antioxidant delivery, for example, to retina tissue, where, in the macula lutea, an NTF-2-like STARD3 protein was shown to play a role in photoprotection thanks to its lutein-carrying ability [51].

Materials and methods

Protein expression and purification

AnaCTDH and TeCTDH with an uncleavable His₆-tag on the C terminus were cloned into the pCDFduet-1 vectors [41], TeCTDH with the N-terminal His₆-tag cleavable by 3C protease was cloned into the pQE81-M vector, Anabaena OCP with the N-terminal His₆-tag cleavable by 3C protease was cloned into the pQE81-M vector. The identity of the constructs and the presence of mutations were verified by DNA sequencing. The obtained plasmids were used to transform chemically competent cells. Protein expression was induced using 0.5 mm isopropyl-β-thiogalactoside (final concentration). Apoforms were either purified by isolation from apo-/holo-mixtures due to different hydrodynamic properties or expressed and purified from E. coli strains not producing carotenoids. Holoforms were expressed in ECN and CAN-producing E. coli cells described in detail earlier [18]. Synthesis of ECN/CAN was achieved by means of the pACCAR25∆crtXZcrtO plasmid carrying the crtO gene encoding a β-carotene ketolase [18], yielding a mixture of both carotenoids. All His₆-tagged proteins were purified by immobilized metal affinity chromatography (IMAC) and SEC to electrophoretic homogeneity and stored at 4 °C in the presence of 3 mm sodium azide [26-28, 50]. Where the construct allowed, the His-tag was removed by human rhinoviral 3C protease and subtractive IMAC [52].

Individual NTD and CTD of Synechocystis OCP devoid of the His₆-tags as well as His₆-tagged HCP1 or OCP-CTD from Anabaena were obtained as described in previous works [9, 26, 41]. The same applies for the wild-type SynOCP and OCP^AA and ∆NTE^O mutants thereof [26, 50]. Each protein was purified in at least three individual batches with qualitatively similar properties.

Protein concentrations were determined at 280 nm using calculated protein-specific molar extinction coefficients. The holoprotein preparations obtained exhibited typical visible-to-UV absorption ratios of 1.6–1.8, indicative of a low apoprotein content. In case of COCP, AnaCTDH(holo), and TeCTDH(holo), the ratio achieved ranged between 2.2 and 2.6 due to a different Trp content. Low apoform content could not be achieved in the case of OCP^AA and CTDH(holo) monomers and visible-to-UV absorption ratio were ~ 0.5 and ~ 0.1, respectively.

Thin layer chromatography analysis of the carotenoid content

Carotenoids were extracted from carotenoproteins by the addition of a twofold volume excess of pure acetone. Samples clarified by centrifugation were subjected to thin layer chromatography on silicagel plates (Silufol, Kavalier, Czechoslovakia) using a mixture of petroleum ether (85% v/v) and acetone (15% v/v) for 15 min and the results were recorded immediately to prevent oxidation of carotenoids. Previous work reporting R_f values for different carotenoids was used as a reference [53].

Native PAGE

Protein samples were analyzed by electrophoresis in the glycine-Tris gel system under nondenaturing conditions as described earlier [28, 54]. The gels were run at 350 V and then scanned prior to and after Coomassie brilliant blue staining. Quantitative densitometry was performed using imagej software v.1.48. Each experiment was repeated at least twice and the most representative results are demonstrated.

Analytical SEC

The oligomeric state of various AnaCTDH and TeCTDH fractions obtained during preparative SEC on a Superdex 200 26/600 column (GE Healthcare, Pittsburg, PA, USA), concentration-dependent dissociation of AnaCTDH(ECN) as well as the results of the inverse carotenoid transfer from OCP^AA to AnaCTDH(apo) were analyzed by SEC on a Superdex 200 Increase 5/150 columns (GE Healthcare) as described previously [10, 27, 28, 50]. The column was equilibrated with buffer SEC (20 mm Tris-HCl, pH 7.6, 150 mm NaCl, 0.1 mm ethylenediaminetetraacetate, 3 mm β-mercaptoethanol) and calibrated using BSA monomer (66 kDa), BSA dimer (132 kDa), BSA trimer (198 kDa), and α-lactalbumin monomer (15 kDa). The elution profiles were followed by simultaneous UV and visible absorption at indicated wavelengths. Typical results obtained in at least three independent experiments are presented.

Absorption spectroscopy

Steady-state absorption spectra and absorption time courses at 550 and 600 nm were recorded as described earlier [34, 41, 44]. A blue light-emitting diode (M455L3; Thorlabs, Newton, NJ, USA), with a maximum emission at 455 nm was used for photoconversion of the samples (actinic light for OCP^O → OCP^R photoconversion). The temperature of the samples was stabilized by a Peltier-controlled cuvette holder Qpod 2e (Quantum Northwest, Liberty Lake, WA, USA) with a magnetic stirrer. Amplitudes of photoconversion and the corresponding rates were estimated according to procedures described earlier [33]. To estimate the efficiency of the carotenoid transfer under the given conditions, spectral decomposition using reference spectra of carotenoid donors and carotenoid acceptors in the 100% holoform was performed in originpro 9.0 (OriginLab, Northhampton, MA, USA) by fitting the corresponding contributions of the donor and acceptor spectrum to the spectrum obtained at the end of the mixing experiment. All experiments were repeated at least two times and the most typical results are presented.

Acknowledgements

The authors are thankful to M. Moldenhauer and N.N. Tavraz (TU Berlin) for help with molecular cloning. The authors acknowledge the support of the Russian Science Foundation (grant No. 18-44-04002) and the German Research Foundation (DFG grant no. FR1276/5-1). The study was partially supported by the grant from Russian Foundation for Basic Research (No. 18-04-00691 to N.N.S.). Protein–protein interactions were studied in the framework of the Program of the Ministry of Science and Higher Education of the Russian Federation (N.N.S. group). This work was partially supported by grants from the Agence Nationale de la Recherche (ANR project RECYFUEL (ANR-16-CE05-0026)), by the European Union's Horizon 2020 research and innovation program under grant agreement no. 675006 (SE2B), by the Centre National de la Recherche Scientifique (CNRS) and the Commissariat à l'Energie Atomique (CEA). FM salary is financed by SE2B.

Conflict of interest

The authors declare no conflict of interest.

Author contributions

NNS – conceived the idea and coordinated the study; YBS, FM, EGM, and NNS – performed experiments; YBS, FM, EGM, AW, TF, DK, and NNS – analyzed and discussed the results; NNS wrote the paper with an input from all authors.