New homologues of Brassicaceae water-soluble chlorophyll proteins shed light on chlorophyll binding, spectral tuning, and molecular evolution

Abstract

Type-II water-soluble chlorophyll (Chl) proteins (WSCPs) of Brassicaceae are promising models for understanding how protein sequence and structure affect Chl binding and spectral tuning in photosynthetic Chl–protein complexes. However, to date, their use has been limited by the small number of known WSCPs, which also limited understanding their physiological roles. To overcome these limitations, we performed a phylogenetic analysis to compile a more comprehensive and complete set of natural type-II WSCP homologues. The identified homologues were heterologously expressed in Escherichia coli, purified, tested for assembly with chlorophylls, and spectroscopically characterized. The analyses led to the discovery of previously unrecognized type-IIa and IIb subclass WSCPs, as well as of a new subclass that did not bind chlorophylls. Further analysis by ancestral sequence reconstruction yielded sequences of putative ancestors of the three subclasses, which were subsequently recombinantly expressed in E. coli, purified and characterized. Combining the phylogenetic and spectroscopic data with molecular structural information revealed distinct Chl-binding motifs, and identified residues critically impacting spectral tuning. The distinct Chl-binding properties of the WSCP archetypes suggest that the non-Chl-binding subclass evolved from a Chl-binding ancestor that most likely lost its Chl-binding capacity upon localization in the plant tissues with low Chl content. This dual evolutionary trajectory is consistent with WSCPs association with the Kunitz-type protease inhibitors superfamily, and indications of their inhibitory activity in response to various forms of stress in plants. These findings suggest new directions for exploring the physiological roles of WSCPs and the correlation, if any, between Chl-binding and protease inhibition functionality.

Introduction

Most chlorophyll (Chl)-binding proteins are transmembrane complexes associated with the photosynthetic membranes of plant and algal cells, and photosynthetic bacteria [1]. Only a few are water-soluble, including the bacteriochlorophyll-binding Fenna–Matthews–Olson proteins of green sulfur bacteria, the Chl-binding peridinin–Chl a-proteins of dinoflagellates, and the water-soluble Chl-binding proteins (WSCPs) of Chenopdium (type-I), and Brassicaceae (type-II) types [2]. WSCPs are unique in that they are not bound to the thylakoids and do not participate in photosynthetic reactions. Furthermore, the sequences and structures of type-II WSCPs classify them as part of the Kunitz soybean trypsin inhibitor (STI) superfamily [3], and some were shown to inhibit serine [4] and/or cysteine [5,6] proteases. Their high levels of expression in response to various plant stress conditions such as drought, salinity, and heat [7–12], led to suggestions of their possible role in plant defense against herbivores [5,13], or in regulation of endogenous protease activities [4,14,15]. However, it is unclear how Chl binding is related to these functions. Others have suggested that they play a role in Chl metabolism [2,7,16], and/or Chl photoactivity [17,18], but such functions have not been substantiated [17].

Irrespective of their undetermined physiological role, type-II WSCPs are gaining more attention in recent years as model systems and spectroscopic benchmarks for understanding spectral tuning, and excited-state dynamics in multi-Chls protein complexes [19–25]. The high-resolution molecular structures of two representative type-II WSCPs [26,27] revealed symmetric homotetramers in which each monomeric subunit contains a single Chl bound to a ~20 kDa single-chain protein. Recently developed methods for assembling recombinant WSCP apo-proteins with natural and synthetic Chl derivatives [28–30], now render the complex an ideal system for rigorous spectroscopic and biochemical studies. Indeed, by combining these methods with X-ray crystallography and site-directed mutagenesis, Bednarczyk et al. [26], and Palm et al. [31] elucidated some molecular determinants of Chl absorption band shifts, and Chl a/b-binding selectivity, respectively.

A major limitation of using type-II WSCPs as a general model for Chl-binding proteins, is the lack of structural and spectral diversity among the few known WSCP homologues. Those characterized to date are classified into subtypes IIa and IIb, based on the characteristic absorption band at 673 and 664 nm of the respective Chl a complexes, and the higher Chl b-binding affinity of type-IIb WSCP [2,25]. But, this subclassification is based on a very limited set of Chl–protein complexes. Thus far, only a single representative of the type-IIb subclass was identified and characterized, namely, WSCP from Virginia pepperweed (Lepidium virginicum; LvWSCP) [32,33], and there are no more than three known and characterized type-IIa WSCPs, that is, the WSCPs from Arabidopsis thaliana (AtWSCP), cauliflower (Brassica oleracea var. botrys; CaWSCP), and Japanese wild radish (Raphanus sativus var. raphanistroides; RshWSCP) [7,14,34,35]. Two other WSCPs from rapeseed (Brassica napus) and Brussels sprout (B. oleracea var. gemmifera) are highly homologous (each sharing more than 94% sequence identity) to CaWSCP [36,37]. Several other putative type-IIa WSCPs were identified but were not fully characterized [6,8].

This work identified additional type-II WSCP sequences by implementing phylogenetic analysis, thereby significantly increasing the number of unique type-II WSCP homologues. Subsequent heterologous expression of selected sequences in Escherichia coli, assembly with Chl a and Chl b, and spectroscopic characterization led to the discovery of previously unrecognized type-IIa, and type-IIb WSCPs, and a new subclass, hereby denoted type-IIx, that is incapable of binding Chl. This extension of the type-II WSCPs class provides new insights into the critical sequence and structural determinants of Chl binding and spectral tuning. Moreover, biochemical and spectroscopic characterization of five archetypical WSCPs inferred from the phylogenetic tree by ancestral sequence reconstruction (ASR) demonstrated that the Chl-binding functionality was lost in nongreen plant tissues, at the expense of, or in parallel to an additional function, likely related to the protease inhibition activity. This apparently dual evolutionary trajectory for the Chl-binding and protease inhibition functionalities provides an evolutionary perspective for the physiological role of WSCPs. Furthermore, it highlights the role of cofactors binding/loss in the divergence of new functions.

Results

A comprehensive phylogenetic tree of type-II WSCPs

A homology search using the CaWSCP sequence as a reference, recovered 24 nonredundant candidate WSCP sequences. Their phylogenetic analysis using the STI sequence as an outgroup, resulted in a tree that split into two distinct branches (Fig. 1), one bearing a single clade that contained the outgroup, and the other bearing three additional clades, splitting from two nodes (IIxa and IIb with 98 and 99 bootstrap values, respectively, see Fig. 1). Since one clade included the type-IIb LvWSCP sequence, while all the sequences of known type-IIa WSCPs clustered into another, the respective clades were identified as type-IIb, and type-IIa. This phylogenetic tree architecture suggested a clear distinction between the close homologues of STI and type-II WSCPs. Furthermore, it revealed that type-IIa WSCPs share a common ancestor with WSCP of a third subclass, hereon referred to as type-IIx, while type-IIb WSCPs were phylogenetically distinct from these two subclasses.

Fig. 1.

Phylogenetic tree of type-II WSCPs. A set of 24 nonredundant sequences of type-II WSCP homologues were phylogenetically analyzed together with the sequence of STI as an outgroup. The tree with the highest log likelihood (−6022.0316) is shown. The bootstrap analysis values derived from 100 individual runs are indicated at each node. Selected nodes are indicated by name. Short names of the type-II WSCP clade members are given in parenthesis next to their Uniprot access id. Those expressed and characterized are in bold font. Those expressed and characterized in this work, including five reconstructed sequences predicted by ASR, are underlined. Analyses and graphical representation were done using the software package MEGA7 [40].

Chlorophyll-binding and spectroscopic properties of new type-II WSCP homologues

Heterologous expression in E. coli of the amino acids sequences of 11 representative type-IIa, IIb, and IIx WSCPs yielded water-soluble proteins for all except the type-IIa sequence, D2ZQS9, that could not be produced with sufficient purity and yield. The other 10 (denoted in boldface in Fig. 1) were purified, and tested for either Chl a or Chl b binding using the water-in-oil emulsion method [28,29]. Six of these (underlined in Fig. 1) were previously unrecognized and uncharacterized type-II WSCPs. The four sequences associated with type-IIa and the three associated with type-IIb WSCP assembled, as expected, with either Chl a or Chl b. The absorption spectra of the resulting complexes are presented in Fig. 2. Size-exclusion chromatography indicated that all the complexes were tetramers (Table 1). By contrast, the three candidate type-IIx subclass sequences did not bind Chl a or Chl b.

Fig. 2.

UV-visible absorption spectra of type-II WSCP-Chl complexes. Recombinant WSCP apo-proteins reconstituted with Chl a (A) and Chl b (B). All spectra were normalized to the absorption at 280 nm, thus the heights of the Chl absorption bands reflect Chl/protein absorption ratios.

Table 1. Size-exclusion chromatography-estimated MWs and aggregation states of WSCP- Chl complexes.

Apoprotein		Holoprotein
Name	Monomer MW (kD)	Pigment	Elution Volume (mL)	MW (kD)	ΔMW (kd)	N	ΔN
LvWSCP	19 675	Chl a	13.5	115 368	14 940	5.9	0.8
		Chl b	13.6	110 898	14 570	5.6	0.7
PksWSCP	20 504	Chl a	13.7	107 615	14 292	5.2	0.7
		Chl b	13.4	121 011	15 395	5.9	0.8
AlpWSCP	18 612	Chl a	14.0	93 060	13 004	5.0	0.7
		Chl b	14.0	92 411	12 944	5.0	0.7
AtWSCP	19 773	Chl a	14.0	94 061	13 096	4.8	0.7
		Chl b	14.0	94 527	13 138	4.8	0.7
RshWSCP	21 676	Chl a	13.8	101 672	13 778	4.7	0.6
		Chl b	13.9	99 396	13 577	4.6	0.6
CaWSCP	19 155	Chl a	14.0	93 252	13 022	4.9	0.7
		Chl b	14.0	94 411	13 128	4.9	0.7
Ol1WSCP	19 672	Chl a	13.8	102 091	13 815	5.2	0.7

The Chl a-containing type-IIa WSCP complexes, including that formed with the previously uncharacterized Ol1WSCP, featured the typical type-IIa WSCP absorption spectrum with a Qy absorption peak at 673 nm. Of the two previously uncharacterized natural sequences belonging to the type-IIb clade, M4FJ52 from Chinese cabbage (Brassica rapa subsp. Pekinensis; PksWSCP) had almost the same spectrum as that of the typical type-IIb LvWSCP, with the characteristic blueshifted Qy band at 664 nm, while A0A087HRG5 from Arabis alpina (AlpWSCP) had an atypical spectrum, with the Qy redshifted to 670 nm, which is more similar to a type-IIa WSCP complex. Its binding affinity, as reflected by the yield of complex assembly, was significantly lower than all other type-IIa and type-IIb WSCPs.

The assembly of the natural WSCP homologues with Chl b and the spectral characteristics of the resulting complexes followed the same trend as Chl a, with the exception of CaWSCP and Ol1WSCP, both of which featured additional blue- and redshifted bands around the main Qy peak at 656 nm. These were likely due to a Chl b by-product formed in water-in-oil emulsions, since the additional bands were not observed in other reconstitution protocols [18,30,34].

Chlorophyll binding and spectral characteristics of archetypical WSCPs

To further explore the evolutionary trajectory of Chl binding and spectral properties of type-II WSCPs, we used ASR to generate a set of archetypical WSCP homologues. These represent putative ancestors of the type-IIa, type-IIb, and type-IIx subclasses (AnIIa, AnIIb, AnIIx, respectively), as well as the common ancestor of type-IIa and type-IIx WSCPs (AnIIxa), and the latest common ancestor of all three subclasses (AnIIxab). The genes of the ancestral sequences were synthesized, cloned into an expression vector, heterologously expressed in E. coli, and subjected to the same spectroscopic and Chl-binding analyses as the contemporary WSCPs (Fig. 3).

Fig. 3.

Amino acids sequences and UV-visible absorption spectra of type-II WSCP ancestors. (A) ASR-predicted sequences. Sites of mutations of AnIIa and AnIIb are indicated by bold underlined font. (B–d) Spectra of apo-proteins reconstituted with Chl a (A) and Chl b (B), and mutants of AnIIa and AnIIx reconstituted with Chl a (C). All spectra were normalized to the absorption at 280 nm, thus the heights of the Chl absorption bands reflect Chl/protein absorption ratios.

AnIIa and AnIIxa efficiently bound Chl a, as indicated by the ratio of absorption peaks at 673, and 280 nm (A673/A280), which resembled that of the natural type-IIa WSCPs. Upon assembly with Chl b, AnIIxa formed a complex with high binding affinity, and displayed the typical type-IIa WSCP absorption spectrum, whereas, assembly with AnIIa resulted in the same spectral artifact observed for natural CaWSCP and Ol1WSCP. AnIIx assembled with Chl a had the typical spectral characteristics of type-IIa WSCPs, but its A673/A280 ratio was ~ 30% smaller than those of AnIIa and AnIIxa complexes, indicative of partial binding. Its Chl b-binding affinity was low and similar to its binding affinity to Chl a, with the Qy band slightly blueshifted with respect to that of a typical type-IIa-Chl b spectrum. The Chl a complexes with AnIIb and AnIIxab had the typical type-IIb WSCP absorption spectra, but exhibited significantly lower binding affinity. AnIIb bound Chl b with a low affinity, while AnIIxab was incapable of binding Chl b altogether.

Distinct and common sequence and structural motifs in type-II WSCP subclasses

Multiple sequence alignment of the extended set of 18 nonredundant natural WSCP sequences and their five ASR-generated archetypes were mapped onto the representative type-IIa, and type-IIb molecular structures of CaWSCP [26], and LvWSCP [27,31]. Both sequence and structural alignments pointed to specific sites of Chl-protein and protein–protein interactions that may impact Chl-binding specificity, bound Chl spectra, and assembly of the homotetrameric complex. Specifically, we identified four distinct Chl-binding domains (CBDs; Figs 4 and 5A). The largest, CBD1, is a 17-residues β-hairpin that extends across the Chl’s macrocycle. A backbone nitrogen of a strictly conserved proline from one of the β-strands is the axial ligand to the central Mg atom, while the hairpin loop interacts with the Chl ring A. CBD2 is a loop that extends from CBD1 and forms a network of hydrogen bonds to carbonyl side groups of Chl rings IV and V. CBD3 and CBD4 consist of an 8- or 10-residue loop, and a 4- or 3-residue beta turn that interact with Chl rings II and I, respectively.

Fig. 4.

Sequence alignment of type-IIa, IIb, and IIx WSCPs. Only the regions including CBDs are shown (A full sequence alignment is available as Data S1). Residue numbers of CaWSCP and LvWSCP are indicated under the global alignment index. Positions interacting with Chls are colored olive green, those involved in monomer-monomer interactions and dimer–dimer interactions are colored red, and blue, respectively, while those that also interact with Chls are colored orange and green. Cyan and red triangles at the bottom of the alignment indicate the positions of point mutations in the AnIIa and AnIIx homologues, respectively.

Fig. 5.

Chl-protein, and protein–protein interactions in WSCPs. (A) Overlay of representative type-IIa and type-IIb WSCP structures (CaWSCP PDB id 5HPZ, and LvWSCP PDB id 2DRE, respectively) showing the bound Chl (green stick presentation) and CBDs in ribbon presentation, with interacting residues in stick presentation. The position of the first residue in each CBD is labeled, and the point mutations sites are indicated by triangles as in Fig. 4. For better clarity, the Chl phytil chain is not shown. CBD1, CBD2, CBD3, and CBD4 are colored orange, brown, blue, and pink, respectively. CaWSCP is in dark and LvWSCP in light shades. (B, C) Protein–protein interactions in CaWSCP (B) and LvWSCP (C) homotetramers. Interacting residues in the reference A subunits are shown in sphere representation in the same lighter shade coloring scheme of (A). Residues of the other tetramer subunits are in stick representation using the following coloring scheme: CBD2(B)-orange, CBD1(C)-orange red, CBD1(D)-dark red, CBD2(B)-cyan, CBD3(D)-sky blue, CBD4(B)-pink, CBD4(C)-magenta, and CBD4(D)-purple.

The same CBDs are also involved in inter-subunit interactions within the WSCP Chl–protein homotetrameric complexes, that can be viewed as comprising two homodimeric subunits. Thus, each homotetramer contains two identical intra-dimer interfaces within each dimeric subunit, and an inter-dimer interface between the A–B and C–D dimers (Fig. 5B,C). As indicated by the crystal structures, the intra-dimer interfaces of both type-IIa and IIb are highly homologous and involve interactions between the CBD2 domains. By contrast, the inter-dimer interface, and thereby, the relative orientations of the dimeric subunits, are different in type-IIa and type-IIb. Consequently, inter-subunit interactions between CBD1, CBD3, and CBD4 vary between type-IIa and type-IIb, as depicted in Fig. 5B,C.

As sequence variations are generally consistent with the molecular structural features, a comparison of the sequence variations of each CBD among the different subclasses, with the structural differences and similarities between the CaWSCP and LvWSCP structures, was performed. All the CBD sequences were more divergent in type-IIx than in the Chl-binding type-IIa and IIb WSCPs, with the exception of the more conserved β-strands of CBD1 that are the structural elements of the β-trefoil fold. The CBD2 sequences were highly conserved between the two Chl-binding type-IIa and type-IIb WSCPs, while the other CBDs were distinct to either type-IIa or type-IIb. More specifically, the CBD2 residues involved in hydrogen bonding to the Chl carbonyl and acetyl groups (T48 and Q53 in CaWSCP and T52 and Q57 in LvWSCP) were strictly conserved, while they were replaced by nonhydrogen-bonding residues in type-IIx WSCPs.

Critical residues to chlorophyll binding and spectral tuning in type-II WSCPs

The CBD sequences of the Chl-binding, and nonbinding WSCP archetypes AnIIa, and AnIIx, differed by only four amino acids. Of these, only residue A33 in AnIIa CBD1 which was replaced by valine in AnIIx, was in direct contact with the bound Chl. Two other residues in AnIIa, that is, N36 and N91 in CBD1, and CBD3, respectively, replaced by lysine in AnIIx, are unlikely to affect Chl binding since they are not involved in either protein-Chl or protein–protein interactions. Residue Y52 in AnIIa CBD2, replaced by aspartate in AnIIx, and conserved in all type-IIa WSCPs, did not interact directly with Chl, but interacted with many residues at the inter-subunit interface formed within the WSCP complex (Fig. 5B,C). Unexpectedly, the adjacent residue, Q53, that is hydrogen bound to the Chl, and strictly conserved in all Chlbinding WSCPs, was retained in AnIIx but replaced by nonhydrogen bonding residues in the descendant Chl nonbinding type-IIx WSCPs. To further explore these trends, we constructed four point mutants: AnIIaA33V, AnIIaN36K, AnIIxD52Y, and AnIIxQ53L and analyzed their assembly with Chl a (Fig. 3C). As expected, the Chl-binding capacity of AnIIaN36K remained unchanged with respect to AnIIx, but was reduced in AnIIaA33V, most likely due to unfavorable steric interactions of the bulkier valine with the nearby Chl macrocycle. The Chl-binding capacity of AnIIxQ53L was further reduced, but not completely lost. By contrast, AnIIxD52Y Chlbinding capacity recovered to close to that of AnIIa. This implies that the inter-subunit protein–protein interactions at CBD2 have the most critical effect on Chl binding and complex assembly, whereas Chl-protein interactions have a significant, albeit, minor effect.

The discovery of three new type-IIb WSCPs in addition to LvWSCP, revealed that the latter is the only one with a tryptophan (W154) in CBD4. Our previous work [26] implicated the displacement of this tryptophan sidechain with respect to its analogue, W151 of CaWSCP, as the main cause for the spectral variations between type-IIa and IIb WSCPs. We demonstrated that this displacement is caused by a single mutation of alanine A34 in CaWSCP to asparagine N38 in LvWSCP, which shifts the tryptophan hydrogen bond from the backbone oxygen of A34 to the δ1-oxygen of N38. Here, we found that while the A34-W151 pair is indeed strictly conserved in all the natural sequences identified as type-IIa, as well as in their putative ancestors, AnIIa, and AnIIxa, the N38-W154 pair is unique to LvWSCP. In all the other sequences identified as type-IIb, W154 was replaced by phenylalanine or tyrosine, while N38 was replaced by serine or isoleucine. Consequently, the Qy absorption peak was at 673 nm in all type-IIa WSCP–Chl a complexes, but varied in type-IIb WSCPs from 664 nm in LvWSCP and PksWSCP, to 670 nm in AlpWSCP (Fig. 2). Notably, the type-IIb ancestors, AnIIxab and AnIIb retained W154 but included a serine in place of N38, yet still showed a Qy absorption peak at 664 nm. This implies that serine may replace asparagine in hydrogen bonding to the tryptophan. To verify this, we replaced A34 in CaWSCP, and N38 in LvWSCP by a serine. As expected, the Qy peak position of the A34S CaWSCP mutant was blueshifted 5–668 nm, but redshifted only 2–666 nm in the N38S LvWSCP mutant (Fig. 6).

Fig. 6.

Absorption spectral shifts of Chl a in CaWSCP A34S, and LvWSCP N38S point mutants. The positions of each Qy band’s absorption peak are indicated. Spectra were normalized to have the same peak absorption. Dashed, and solid lines indicate the spectra of native and mutants WSCPs, respectively.

Discussion

Our phylogenetic analysis extended the very limited set of type-II WSCPs to include 18 unique WSCP homologues. While it remains a limited collection, it is most likely that it includes all of the currently known, phylogenetically distinct type-II WSCP sequences because the phylolgentic algorithm picked these 18 sequences as a distinct clade from a larger set of 24 homologues with ≥ 23% sequence identity. The phylogenetic subclassification was consistent with type-IIa and IIb subclass spectral and Chl-binding characteristics, expanding them to now include seven and four homologues, respectively. Surprisingly, it identified a new subclass, type-IIx, which comprised seven homologues that are incapable of binding Chl. Despite the small number of sequences in each subclass, they were all well aligned, and contained distinct domains that correlated well with structural motifs specific to each subclass (Figs 4 and 5), ultimately enabling prediction of archetypical sequences by ASR. Correlation of the sequence variations among the different subclasses with the physical properties of selected contemporary WSCP homologues and their ASR-predicted ancestors, provided many new insights into the determinants of Chl binding, complex assembly, and spectral tuning in WSCPs, as well as their evolutionary history.

Determinants of chlorophyll binding, complex assembly, and spectral characteristics

Sequence and structural alignment revealed four distinct CBDs (Figs 4 and 5). Of these, CBD2 was highly conserved in Chl-binding type-IIa and IIb WSCPs, suggesting it a primary determinant of Chl-binding and Chl-protein complex assembly. More specifically, we found that inter-subunit interactions between CBD2 domains, particularly around the conserved CBD2 tyrosine, were the most critical to the assembly of Chl–protein complexes. However, their disruption did not completely abolish Chl binding. As indicated by the AnIIa and AnIIx archetypes and their contemporary descendants, complete loss of Chl binding is a cumulative effect of unfavorable Chl-protein and protein–protein interactions. Similarly, comparison of the type-IIb ancestors, AnIIxab and AnIIb, with the type-IIa ancestors, AnIIxa and AnIIa, suggests that CBD3 affects Chl a/b-binding selectivity, while the hairpin loop of CBD1 (residues: 33–43 of CaWSCP, and 37–47 of LvWSCP) plays a role in Chl spectral tuning. The type-IIb ancestors sharing the same CBD1 hairpin loop sequence bound Chl a weakly, and formed complexes which featured a typical type-IIb absorption spectrum (Fig. 3). Yet, only AnIIb was capable of binding Chl b, whereas AnIIxab that has a CBD3 more typical of type-IIa and similar to that of AnIIxa, did not bind Chl b. This is consistent with the different lengths of the type-IIa and type-IIb CBD3 loops that contain a strictly conserved tryptophan (W87, W90 in CaWSCP, LvWSCP, respectively), which interacts with the Chl ring B. In type-IIa, the loop is extended by two residues as compared to type-IIb, with proline, and glutamate or aspartate (P88, and D/E92 in CaWSCP) inserted at the first, and fourth positions, respectively following the conserved tryptophan. Mutations in this region were shown to affect the binding preferences for Chl a vs Chl b in type-IIa and type-IIb WSCPs [31].

Extension of the Chl-binding WSCPs database, particularly by the discovery of new members of the type-IIb subclass, sheds light on the tuning mechanisms of Chl spectra by the protein environment. The role of W154 of CBD4 in redshifting the Chl Qy absorption band was previously described [26]. Yet, here, we found that its substitution by phenylalanine or tyrosine in type-IIb WSCPs was the more common alternative for blueshifting the Qy band than the hydrogen bonding network that moves the tryptophan away from the Chl in LvWSCP. Furthermore, in the newly discovered type-IIb AlpWSCP, the Chl a Qy peak position was significantly redshifted to 670 nm from the 664 nm peak observed for the other type-IIb, that is, LvWSCP and PksWSCP, suggesting that other Chl-protein interactions affect Chl spectral tuning. These may occur around the hairpin loop of CBD1 since its sequence in AlpWSCP was significantly different from both the respective type-IIa, and type-IIb loops (Fig. 4). Additional structural information from other type-II WSCP homologues will still be necessary to gain a deeper and more generally applicable understanding of the molecular mechanisms affecting both spectral tuning and Chl a/b binding.

Implications to WSCP evolution and physiological roles

The discovery of the new type-IIx Chl nonbinding subclass of type-II WSCP, and the Chl binding and spectral properties of their putative archetypes suggest two trends in type-II WSCP evolution. One trend yielded higher Chl a-, and Chl b-binding affinity, as reflected in the increased Chl a-binding affinity of AnIIxa and AnIIb as compared to their common ancestor, AnIIxab. Only the descendants, but not their ancestors, were capable of binding Chl b, suggesting that this ability was gained later in evolution than the ability to bind Chl a. The second evolutionary trend toward the loss of Chl-binding function was reflected in the split of the IIxa node, represented by the Chl-binding AnIIxa WSCP, into the respective type-IIa and type-IIx Chl-binding, and nonbinding subclasses. While the high binding affinity of AnIIxa was maintained in the descendant type-IIa ancestor, AnIIa, and its contemporary WSCP descendants, it was significantly reduced in the descendant type-IIx ancestor, AnIIx, and completely abolished in its contemporary WSCP descendants. This evolutionary trend suggests that type-II WSCPs may have at least one more function in addition to Chl binding. While the specific nature of this function, and its physiological role remain to be determined, the sequence and structural homology between type-II WSCPs and STI [3] and other plant protease inhibitors [38,39], as well as biochemical and physiological evidence [4–12], suggest that they may serve as serine- and/or cysteine protease inhibitors to protect against biotic or abiotic stress [5,7–13], or to regulate endogenous protease activities [4,14,15].

How the protease inhibition activities of type-II WSCPs are related, if at all, to Chl binding remains to be explored. Since many plant species maintain several genes of various type-II WSCP homologues, and given the plasticity and functional diversity of the WSCP β-trefoil fold [38,39], it is reasonable to assume that different homologues have acquired different functions in different plant tissues. This hypothesis is supported by the findings of Halls et al. [15], who identified a type-II WSCP homologue that co-purified with proaleurain maturation protease from cauliflower florets. This homologue, originally termed WSCP2, (Uniprot ID: Q3HYA1_BRAOB), differed from the Chl-binding CaWSCP that is expressed in cauliflower leaves. Its high sequence homology to Q42436 (88% sequence identity), assigned here to the type-IIx clade (Fig. 1), indicated that it is a type-IIx WSCP. This was confirmed in our laboratory by subjecting a recombinant WSCP2 protein to a Chl-binding assay, which indicated no Chl-binding capacity. The localization of a type-IIx WSCP in nongreen plant tissues strongly supports an evolutionary scenario in which type-IIx WSCPs evolved from a Chl-binding ancestor, but lost their Chl-binding capacity upon localization in plant tissues with low Chl content.

Concluding remarks

This comprehensive phylogenetic analysis of an extended set of type-II WSCP family identified three distinct subclasses of Chl-binding WSCPs. These findings, combined with spectroscopic data, and the currently available molecular structures of CaWSCP and LvWSCP, singled out candidate residues critical for Chl binding and complex assembly, and further established type-II WSCPs as a useful model system and benchmark for studying general principles of spectral tuning in photosynthetic Chl-binding proteins. While the functional rationale of spectral variations in WSCPs remains unclear, the same tuning principles may be used in photosynthetic Chl–protein complexes in which control of the transition energies of individual Chls is essential for proper functionality. The discovery of the non-Chl-binding type-IIx WSCP subclass and evidence that these are expressed in Chl-deficient, nongreen plant tissues, while the Chl-binding type-IIa, and type-IIb subclasses are expressed in green plant tissues, have far-reaching implications on the evolution and physiological roles of WSCPs. Phylogenetic evidence indicates that the loss of Chl-binding capacity occurred later in evolution, as this subclass branched from type-IIa WSCPs. This places Chl binding and other suggested WSCP functionalities in a new perspective, and raises new questions regarding the physiological role of Chl binders and nonbinders, and the correlation between Chl-binding and other biochemical activities of WSCPs. These questions will have to be addressed in future biochemical and physiological studies that should rely on the new classification of WSCPs presented here.

Materials and methods

Phylogenetic analysis of WSCP homologues

The sequence of CaWSCP (UniProt ID: Q7GDB3) was used as a template to search for homologues using the standard Basic Local Alignment Search Tool (Blast) on the EMBL-EBI web server (https://www.ebi.ac.uk/Tools/sss/ncbiblast/), set at default parameters. Sequences showing > 30% identity were retrieved. Sequences with ≥ 90% sequence identity were considered redundant. The 24 nonredundant sequences discovered were then subjected to multiple sequence alignment, phylogenetic analysis and ASR, using the software package MEGA7 [40]. Multiple sequence alignment was carried out using MEGA7’s built-in MUSCLE algorithm [41]. The complete alignment is provided as Data S1. The phylogenetic tree was inferred using the neighbor-joining algorithm [42]. The bootstrap resample test was performed 100 times to confirm the tree’s reliability. Based on the phylogenetic tree and ASR results, 11 natural protein sequences and five ancestor sequences were selected to be overexpressed in E. coli, and to be tested in a Chl-binding assay.

Protein expression and purification

Custom-synthesized DNA sequences encoding the aminoacid sequences of selected proteins, were purchased from Bio Basic Inc (Markham, ON, Canada). The genes were codon-optimized for overexpression in E. coli, and cloned in the pUC-57 vector. All sequences carried a stretch of 30 nucleotides of destination vector sequences at the 5′ end of the gene and at the 3′ end of the gene (after the stop codon), so that a single set of primers can be used for all genes. A single set of universal primers (forward and reverse) specific to the added vector sequences, was designed to amplify the genes of interest for subsequent cloning into the modified pET28 vector carrying a Thioredoxin-Profinity eXact™ tag (Bio-rad, Hercules CA, USA) [26,28,29,43]. Point mutations were introduced to this gene construct using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent, Santa Clara CA, USA). The modified pET28 vector was used for transforming BL21 (DE3) cells that were then grown at 37 °C in the presence of 30 μg·mL⁻¹ kanamycin, until an OD_{600 nm} of 0.6–0.8. Thereafter, protein overexpression was induced by addition of 200 μM isopropyl 1-thio-β-d-galactopyranoside. Following induction, cultures were grown at either 15 or 37 °C for 24 or 4 h, respectively. Cells were then pelleted and stored at −70 °C until processing. Proteins were harvested and purified by the Profinity eXact™ affinity chromatography system [26,28,29]. The eluted tag-less proteins were extensively dialyzed in 0.1 m sodium phosphate buffer (pH 7.8), and stored at 4 °C until further processing. The purity of the proteins was assessed by 12% SDS/PAGE, after which the gel was stained with InstantBlue™ solution (Sigma-Aldrich, St. Louis, MO, USA) and de-stained using tap water.

Chlorophyll preparation

All the procedures involving Chls and Chl–protein complexes were carried out under dim green light illumination, provided by commercial green LED lamps. Chls were extracted from fresh lettuce leaves purchased at a local supermarket. Approximately 40–50 g lyophilized leaves were homogenized in 250 mL cold acetone, in a waring blender, at high speed, for 5–8 min, with intermittent pause and addition of cold acetone. The homogenate was filtered through a Whatman® (grade 2) filter and centrifuged at 1160 g for 10 min. A 1/7 volume of 1,4 dioxane was added to the clean collected supernatant, followed by dropwise addition of 150 mL water, while shaking, until onset of turbidity [44]. The mixture was kept in the dark, at −20 °C, for 2–3 h, centrifuged at 12860 g for 5 min, and stored for further processing. The precipitate was redissolved in 130 mL acetone, to which 20 mL 1,4 dioxane was added, followed by dropwise addition of 30–40 mL double-distilled water, a brief period of freezing at −20 °C and finally centrifuged as in the previous stage. The precipitate was then redissolved in diethyl-ether, filtered through a cotton pad and evaporated under vacuum at 25 °C. Chls were separated from other pigments by chromatography on DEAE sepharose (GE Healthcare, Chicago, IL, USA) [45]. The mixture of Chl a and b was separated by preparative TLC using a 100/6 (v/v) Pentane/isobutanol mixture as the developing solvent. The green zones corresponding to R_f values of 0.7–0.75 for Chl a, and 0.35–0.45 for Chl b, were scraped off of the TLC plate and Chls were extracted with acetone. The purity and concentration of Chl a and Chl b were further confirmed spectrophotometrically [46]. Absorption spectra were measured at room temperature using a JASCO 7200 spectrophotometer (JASCO, Tokyo, Japan).

Assembly of chlorophyll–protein complexes

Chlorophyll–WSCP complexes were assembled using the previously described water-in-oil emulsion method [28,29]. Briefly, tag-less, pure WSCP (1–1.3 mg·mL⁻¹) in 1 mL sodium phosphate buffer pH 7.8 (aqueous phase) was mixed with 5 mL of an organic phase (4.5% v/v Span 80 and 0.4% v/v Tween 80 in mineral oil) in a homogenizer to form an emulsion. A 10-fold molar excess of highly concentrated Chl a or Chl b dissolved in methanol, was added to the emulsion, thoroughly mixed, and incubated on ice for 1–1.5 h. The aqueous phase containing the assembled Chl-WSCP complex was recovered from the emulsion by repeated centrifugation and washing with mineral oil, followed by two final washes with diethyl-ether to remove unbound pigments. The aqueous phase containing the reconstituted complex was then loaded onto PD10 desalting columns (GE Healthcare) and eluted with 50 mM sodium phosphate buffer (pH 7.8). The samples were concentrated using 10 kDa molecular weight (MW)-cutoff Amicon ultracentrifugal filter units (Millipore, Burlington, MA, USA) to an OD of ~ 0.5–1.0 at its Qy band.

Size-exclusion chromatography of chlorophyll–protein complexes

The natural WSCP-Chl complex (~ 6 mL) in 50 mM sodium phosphate buffer (pH 7.8) with an O.D. of 0.4–0.7 at the Qy peak, was concentrated to 550 μL and injected into Superdex 200 10/300 GL column (GE Healthcare). The chromatogram was monitored at 280 nm for protein absorbance and at ~ 420 and 660 nm for Chl absorbance. The specific wavelengths for Chl absorbance were selected to match absorption maxima specific to the type of WSCP and bound Chl. The column was calibrated using a standard set of low MW markers (GE Healthcare). For both standard and sample runs, 50 mM sodium phosphate (pH 7.8) with 150 mM potassium chloride was used as running buffer.

Molecular weights of the protein–Chl complexes were calculated according to the following equation: MW = a · 10^{b(V_e–V_o)/(V_c–V_o)}, whereby V_e, V_o, and V_c are the protein’s elution volume, void volume, and column volume, respectively, and a and b are parameters obtained from a nonlinear fit of a standard curve of elution volumes vs. MWs. Aggregation numbers (N) were calculated by dividing the respective MW by the protein’s monomer MW, calculated from its amino acids sequence. The errors in MW and N (ΔMW and ΔN), were estimated from the standard errors of the a and b fit parameters. Nonlinear fitting was performed using the curve fitting module of the software igor 7 (Wavemetrics Inc., Lake Oswego, OR, USA).

Supplementary Material

Additional supporting information may be found online in the Supporting Information section at the end of the article.

Full sequence alignment

Data S1. Full multiple sequence alignment of WSCP sequences and STI outlier (aligned fasta file).

Abbreviations

ASR

ancestral sequence reconstruction

CBD

Chl-binding domain

Chl

chlorophyll

STI

soybean trypsin inhibitor

WSCP

water-soluble chlorophyll protein