Single-residue insertion switches the quaternary structure and exciton states of cryptophyte light-harvesting proteins

Significance

There is intense interest in determining whether coherent quantum processes play a nontrivial role in biology. This interest was sparked by the discovery of long-lived oscillations in 2D electronic spectra of photosynthetic proteins, including the phycobiliproteins (PBPs) from cryptophyte algae. Using X-ray crystallography, we show that cryptophyte PBPs adopt one of two quaternary structures, open or closed. The key feature of the closed form is the juxtaposition of two central chromophores resulting in excitonic coupling. The switch between forms is ascribed to the insertion of a single amino acid in the open-form proteins. Thus, PBP quaternary structure controls excitonic coupling and the mechanism of light harvesting. Comparing organisms with these two distinct proteins will reveal the role of quantum coherence in photosynthesis.

Abstract

Observation of coherent oscillations in the 2D electronic spectra (2D ES) of photosynthetic proteins has led researchers to ask whether nontrivial quantum phenomena are biologically significant. Coherent oscillations have been reported for the soluble light-harvesting phycobiliprotein (PBP) antenna isolated from cryptophyte algae. To probe the link between spectral properties and protein structure, we determined crystal structures of three PBP light-harvesting complexes isolated from different species. Each PBP is a dimer of αβ subunits in which the structure of the αβ monomer is conserved. However, we discovered two dramatically distinct quaternary conformations, one of which is specific to the genus Hemiselmis. Because of steric effects emerging from the insertion of a single amino acid, the two αβ monomers are rotated by ∼73° to an “open” configuration in contrast to the “closed” configuration of other cryptophyte PBPs. This structural change is significant for the light-harvesting function because it disrupts the strong excitonic coupling between two central chromophores in the closed form. The 2D ES show marked cross-peak oscillations assigned to electronic and vibrational coherences in the closed-form PC645. However, such features appear to be reduced, or perhaps absent, in the open structures. Thus cryptophytes have evolved a structural switch controlled by an amino acid insertion to modulate excitonic interactions and therefore the mechanisms used for light harvesting.

Light-harvesting complexes capture and funnel the energy from light using organic chromophore molecules that are bound to scaffolding proteins. The protein structure thereby sets the relative positions and orientations of the chromophores to control excitation transport. In other words, the protein plays a deciding role in building the “electronic Hamiltonian”—the electronic coupling between chromophores and the chromophoric energy landscape that directs energy flow. This strong connection between structural biology and physics means that ultrafast light-harvesting functions are under genetic and evolutionary control. Cryptophytes, a group of marine and freshwater single-celled algae, are an intriguing example, because one of their light-harvesting antenna complexes was completely re-engineered by combining a unique bilin-binding polypeptide with a single subunit from the ancestral red algal phycobilisome (1, 2). Here we report a further example of biological manipulation of this phycobiliprotein (PBP) light-harvesting system. We have discovered an elegant but powerful genetic switch that converts the common form of this PBP into a distinct structural form in which the mechanism underpinning light harvesting is vastly different—in essence because strong excitonic interactions within the PBP are switched from on to off.

The crystal structure of the cryptophyte PBP phycoerythrin PE545 from Rhodomonas CS24 showed that the protein is a dimer of two αβ monomers (3, 4), the β subunit of which has a globin fold (5, 6) and binds three linear tetrapyrroles (bilins), whereas the α subunit is a short, extended polypeptide with a single bilin chromophore. A prominent feature of this structure is the arrangement of the two central chromophores in van der Waals contact with each other on the pseudo-twofold axis, with each chromophore covalently linked to two cysteines on one of the β subunits (referred to as “β50/61”). This structural feature introduces excitonic coupling between the chromophores (3, 4). We are fascinated by this observation because it implies that if coherence plays a nontrivial role in light harvesting (7–12), it might be switched on and off (either dynamically or genetically) by controlling the separation, and hence excitonic coupling, between these two central chromophores.

To gain a better understanding of the interrelation between protein structure, chromophore arrangement, quantum coherence, and biological evolution, we have determined crystal structures of three cryptophyte PBPs: phycocyanin PC645 from Chroomonas sp. (strain CCMP 270), PC612 from Hemiselmis virescens (strain CCAC 1635 B), and PE555 from Hemiselmis andersenii (strain CCMP 644) (13). The PC645 dimer has the same architecture as PE545, which we call the “closed” form, in which the two central β50/61 chromophores are in physical contact. In contrast, the structures from the two Hemiselmis species, PC612 and PE555, show a dramatically different dimer structure in which the αβ monomers have been rotated by ∼73° compared with the closed form. In this open form, the central β50/61 chromophores are separated by a water-filled channel. The results of 2D electronic spectroscopy (2D ES) from all three cryptophyte PBPs are reported here. We conclude that the mechanism of light harvesting and whether effects arising from electronic coherence are significant depend strongly on the structure. It is notable that each of these complexes harvests light differently but apparently successfully.

From the protein sequences and structures, it appears that the Hemiselmis proteins with the open form differ from the cryptophyte closed-form proteins by the insertion of a single amino acid in a conserved region just before the chromophore attachment site in the α subunit. This insertion results in a rotation of part of the chromophore, and this rotation, in turn, precludes the formation of the closed-form dimer, ultimately resulting in the new, open-form dimer structure. We compare the open and closed PBP structures using quantum chemical calculations and spectroscopic measurements. These findings open the door for determining the role of quantum coherence in a real-life biological system and gaining a better understanding of how these distinct light-harvesting proteins might have evolved from the elaborate ancestral phycobilisome structure.

Results

Crystal Structure of Phycocyanin PC645 from Chroomonas sp.

The crystal structure of phycocyanin PC645 from Chroomonas sp. CCMP270 was determined at 1.35-Å resolution (Fig. 1A, Fig. S1A, and Table S1). The molecule consists of an α₁β.α₂β dimer in which each α subunit is covalently linked to a mesobiliverdin chromophore (MBV α18), and each β subunit has a doubly linked dibiliverdin chromophore, DBV β50/61, and two singly linked phycocyanobilins, PCB β82 and PCB β158 (Table 1). The two DBV β50/61 chromophores are in van der Waals contact at the pseudo-twofold axis of the dimer where the pyrrole A rings are offset stacked (Figs. S1A and S2A).

Fig. 1.

Structures of the open and closed forms of cryptophyte PBPs. (A and B) Stereo cartoon diagrams of (A) the closed-form PC645 α₁β.α₂β dimer and (B) the open-form PC612 (αβ)₂ dimer. The α chains are colored blue and red; the corresponding β chains are magenta and cyan. In PC645, α₁ is blue, and α₂ red. Chromophores are shown as CPK models. The view is along the quasi-twofold axis with the two doubly linked chromophores proximal to the viewer. (C) Superposition of the stereo cartoon diagrams of the αβ monomers from all available cryptophyte PBPs. The β subunits are shown in light orange; α subunits are coded: PE545 α₁, green; PE545 α₂, lime green; PC645 α₁, magenta; PC645 α₂, purple; PE555 α₁, blue; PE555 α₂, cyan; PC612 α subunits (two chains in crystal structure), red and salmon. Chromophores are shown in atom colors. (D and E) Two orthogonal views of the closed-form PC645 dimer (monomers are shown in red and green) (D) and the open-form PC612 dimer (monomers shown in magenta and cyan) (E), where the 90° rotation between views is about the vertical axis in the plane of the page. (F) Electrostatic surface of PC612 dimer rotated 180° about the vertical axis in the plane of the page as compared with B.

Table 1.

Chromophores

Organism	Strain	PBP	α	β50,61	β82	β158
Rhodomonas sp. CS24	CS24	PE545	DBV	PEB	PEB	PEB
Hemiselmis andersenii	CCMP 644	PE555	PEB	DBV	PEB	PEB
Hemiselmis virescens	M1635	PC612	PCB	DBV	PCB	PCB
Chroomonas sp.	CCMP 270	PC645	MBV	DBV	PCB	PCB

The structure of the PC645 dimer is very similar to the previously published closed structure of phycoerythrin PE545 from Rhodomonas CS24 (73% sequence identity; rmsd 0.85Å on 453 C_α atoms) (3, 4).

Crystal Structure of Phyocyanin PC612 from H. virescens.

The crystal structure of phycocyanin PC612 from H. virescens CCAC 1635 was determined at 1.7-Å resolution (Fig. 1B, Fig. S1B, and Table S1). This light-harvesting complex also exists as an αβ.αβ dimer, but, in contrast to PE545 and PC645, it has nearly perfect twofold symmetry, and the two α subunit sequences are identical. The αβ monomers in PC612 are very similar to those observed in the closed-form structures (71% overall sequence identity with PC645; rmsd 1.11 Å on 213 C_α atoms; Fig. 1C). The positions of the chromophores in the αβ monomer are equivalent to those in the closed-form structures. The only chromophore difference between PC612 and PC645 is the α chromophore, which in PC612 is PCB instead of MBV (Table 1).

Because the quaternary arrangement of the two αβ monomers in PC612 is so distinct from the closed form observed in PC645 and PE545 (in Fig. 1, compare A with B and D with E), we refer to it as the “open” form. The two αβ monomers in the PC612 structure form a dome or cup-like structure which contains a central cavity (Fig. 1 B and F).

Crystal Structure of Phycoerythrin PE555 from H. andersenii.

The crystal structure of phycoerythrin PE555 from H. andersenii CCMP 644 was determined at 1.8-Å resolution (Table S1). The structure shows a near symmetric α₁β.α₂β dimer that is nearly identical to the open-state structure of PC612 (84% overall sequence identity; rmsd 0.92 Å on 464 C_α atoms). However, in terms of chromophores, PE555 contains three phycoerythrobilins (PEBs) that replace the PCBs of PC612 (Table 1).

αβ Monomer Structure and Chromophore Arrangement Are Conserved.

Each αβ monomer is composed of a β subunit with a globin fold and an extended α subunit which lies along the β subunit (Fig. 1C). Apart from loops (particularly around the α chromophore), there is little deviation in the αβ monomer when comparing PE545, PC645, PC612, and PE555 (rmsd of 0.78–1.6 Å over ∼210 C_α atoms). The sequences of the β subunits are highly conserved (79–92% identity). The sequences of the examined α subunit are much more divergent, with 23–82% identity, but their conformations still are very similar (rmsd of 0.8–2.6 Å).

Comparison of Open- and Closed-Form Quaternary Structures.

Although the protein structure and chromophore arrangement within the αβ monomer are conserved, the open and closed quaternary structures are radically different. The transformation that relates the open- and closed-form quaternary structures is a rotation of one αβ monomer by ∼73° around an axis perpendicular to the dimer pseudo-twofold axis (Fig. 1 D and E and Table S2). The centers of mass of the two αβ monomers are slightly more separated in the open form: The center of mass-to-center of mass separation for the closed form was 23.4 Å for PE545 and 22.2 Å for PC645, compared with 24.4 Å for the open form of PC612 and 24.8 Å for the open form of PE555.

Transitions between the two quaternary forms of the same protein are unlikely to occur, and there is no evidence that either closed- or open-form proteins are in equilibrium with a measurable monomer pool. In the closed-form dimer, the monomer–monomer interaction buries a substantial surface (PE545: 1,060 Ų per monomer; PC645: 1,230 Ų per monomer), indicating that the dimer is very stable. In the open-form dimer, the monomer–monomer interaction buries a smaller but still significant surface area (PE555: 618 Ų per monomer; PC612: 511 Ų per monomer). Although this result suggests that the open-form dimer may be less stable than the closed form, we see no evidence of monomer–dimer equilibrium on size-exclusion chromatography.

The main effect of the change from the closed to the open form is the separation of the central, doubly linked β50/61 chromophores. In the closed-form structures, these two chromophores are in van der Waals contact with the pyrrole A rings, which are offset stacked (closest approach of 3.8 Å between atoms in pyrrole A rings; Fig. 1A and Figs. S1A and S2A). However, in the open form, these two chromophores are well separated (closest approach between atoms in pyrrole A ring of 10.0 Å in PC612 and 11.0Å in PE555; Fig. 1B, Fig. S2 B and C, and Table 2).

Table 2.

Electronic couplings (unscreened) and center-to-center distances for selected phycobiliprotein pairs in PE545, PC645, PC612, and PE555

	PE545	PC645	PC612	PE555
Bilin pair (interdimer)	PEBβ50/61PEBβ50/61	DBVβ50/61DBVβ50/61	DBVβ50/61DBVβ50/61	DBVβ50/61DBVβ50/61
Center-to-center separation, Å	15.11	13.17	19.48	19.91
Coupling, cm−1	166	647	29	4
Bilin pair (interdimer)	DBVα20PEBβ50/61	MBVα18DBVβ50/61	PCBα20DBVβ50/61	PEBα20DBVβ50/61
Center-to-center separation, Å	22.74	23.49	30.43	29.30
Coupling, cm−1	−64	−74	−5	−8
Bilin pair (intradimer)	DBVα20PEBβ158	MBVα18 PCBβ158	PCBα20PCBβ158	PEBα20PEBβ158
Center-to-center separation, Å	20.55	18.34	18.23	19.35
Coupling, cm−1	51	151	146	68

Dimer Interface in the Closed Form.

The dimer interface interactions in the closed form are mediated by the α subunit, the α chromophore, or the β50/61 chromophore with no direct protein–protein interactions between the two β subunits. There are three key interaction sites. First, the pyrrole A rings of the two β50/61 chromophores pack against each other (Fig. S2A). In addition, the loop connecting helices hG and hH (the GH loop) of the β subunit and the C-terminal loop of α₁ pack against the face of the β50/61 chromophore from the α₂β monomer (Fig. S2A). The second interaction site is centered on pyrrole rings A and B of the α chromophore. These sit in a hydrophobic pocket formed by the C-terminal tail of the opposite α subunit, the C terminus of the opposite α subunit helix, and the loop connecting helix hB to hE of the opposite β subunit (Figs. 2A and 4A). Third, the α subunit helix makes polar interactions with the break between helices hA and hB and with helix hE in the opposite β subunit plus the opposite α subunit (Fig. 2 B and C).

Fig. 2.

Dimer contacts in the closed- and open-form structures. (A–C) Dimer contacts in the closed-form PC645 structure. (D–F) Dimer contacts in the open-form structures of PC612 (D and E) and PE555 (F). (A) Pyrrole rings A and B of the PC645 α chromophore (green carbon atoms) lie in a hydrophobic pocket formed by the C terminus of the neighboring α subunit (red cartoon) and the loop between helices hB and hE in the neighboring β subunit (cyan cartoon). (B) Helix from PC645 α1 (blue) makes polar side-chain contacts with residues from the opposite αβ monomer [β subunit helices hA, hB, and hE (cyan cartoon) and the α2 linker between β strand s2 and helix (red cartoon)]. Residues from the α1 subunit are labeled in black; residues from the neighboring αβ monomer are labeled in red and blue, respectively. (C) Helix from PC645 α2 (red) makes similar polar contacts with the opposite αβ monomer (hA and hB, magenta, and α1, blue). Residues from α2 are labeled in black; residues from the opposite αβ monomer are labeled in blue and magenta, respectively. (D) The open-form PC612 showing the hydrophobic contact between pyrrole ring A of the α chromophore (green) and the pocket in the opposite αβ monomer formed by β subunit helix hE (cyan) and the N terminus of the α subunit helix (red). (E) PC612 showing dimer interface between the C terminus of the α subunit helix (blue) and the GH loop from the opposite β subunit (cyan). Residues from the α subunit are labeled in black; residues from the β subunit are labeled in cyan. (F) The same interaction in PE555 between the helix from α2 (red) and the opposite GH loop (magenta). Residues from the α2 subunit are labeled in black; residues from the β subunit are labeled in magenta. Hydrogen bonds/salt links are shown as dotted lines.

Fig. 4.

Rotation of pyrrole ring A in the open-form α chromophore precludes formation of the closed form via a steric clash. (A) Stereo view of the packing interaction between pyrrole ring A and the opposite αβ monomer in the closed-form PC645. (B) Model of PE555 in which the two αβ monomers have been rotated so that they overlay the closed-form structure of PC645. Pyrrole ring A of the α subunit chromophore (green stick model) is rotated by ∼70° compared with that of PC645, and this rotation results in a steric clash with conserved Pro64 in the opposite β subunit.

Dimer Interface in the Open Form.

As in the closed form, all dimer interface contacts in the two open-form structures are mediated by either the α subunits or the α subunit chromophore with no direct interactions between the two β subunits. A major contact is made by the α chromophore where pyrrole ring A sits in a hydrophobic pocket in the opposite αβ monomer (Fig. 2D). The pocket is formed by the N-terminal portions of helix hE in the β subunit and the helix in the α subunit, plus the hydrophobic residue that precedes the α subunit helix by two residues (Met47 in PC612 α, Phe45 in PE555 α₁, and Met45 in PE555 α₂).

The other major dimer interface interaction is centered on the C terminus of the α subunit α helix, which makes polar contacts with the GH loop in the opposite β subunit (Fig. 2 E and F). Additionally, in the PE555 structure, the C-terminal residue of α₂, Leu62, makes van der Waals contacts with the C-terminal tail of α₁ and the GH loop of the opposite β subunit (Fig. 2F and Fig. S2C). This interaction is adjacent to the polar interface mentioned above.

Sequence and Structural Changes Around the α Chromophore Dictate Quaternary Structure.

Given the structural conservation of the αβ dimer, it is not immediately obvious why the two Hemiselmis PBPs assemble in the open-form dimer rather than the closed form. An examination of the sequences of the α subunits shows that in the Hemiselmis PBPs an aspartic acid (Asp18) has been inserted in the highly conserved FDxRGC motif that links the first β strand to the α chromophore attachment site (Fig. 3C). In the closed-form structures, this motif forms a network of hydrogen bonds that determine the orientation of pyrrole ring A in the α chromophore with respect to the β sheet (Fig. 3A). The insertion of an aspartic acid in the Hemiselmis α subunits alters the hydrogen bonding network (Fig. 3B). The net effect of this change is that the plane of pyrrole ring A with respect to pyrrole ring B in the α chromophore is rotated in a counterclockwise fashion by ∼29° in the closed form, whereas it is rotated in a clockwise fashion by ∼40° in the open form (Table S3). Thus, the sequence insertion results in an ∼69° rotation in pyrrole ring A of the open-form αβ monomer as compared with the closed form.

Fig. 3.

Insertion of an aspartic acid in the open-form α subunit results in the rotation of pyrrole ring A in the α chromophore as compared with the closed form. (A) Stereo view of the α chromophore pocket of the closed-form PC645 showing the hydrogen bonding network. (B) Identical view showing the open-form PC612 pocket. (C) Structure-based alignment of all mature α subunit sequences reported in this paper. The red arrow indicates the chromophore attachment site; the blue arrow indicates Glu16 that coordinates the central pyrrole nitrogens in the open-form structures. Red type indicates identity; blue type indicates similarity. Breaks in the alignment mark the ends of secondary structure elements. Note that the helix length is variable at its C terminus.

How does the rotation of pyrrole ring A in the α subunit chromophore determine the quaternary structure? In both open and closed forms, pyrrole ring A of the α chromophore makes a major contribution to the dimer interface (see Figs. 2A and 4A for the closed form and Fig. 2D for the open form). Superposition of the open-form αβ monomer on a closed-form quaternary structure shows that the rotation of pyrrole ring A results in a steric clash with conserved Pro64 in the opposite β subunit (Fig. 4B). Thus, the open-form αβ monomer is prevented from adopting the closed-form dimer structure.

Quantum Chemical Calculations Reveal Excitonic Switching.

Using the high-resolution crystal structures of the PBPs, we calculated the gas-phase couplings (Table S4) and transition dipole moments (Table S5). Note that these electronic couplings do not include dielectric screening effects that tend to reduce their magnitude by a factor of about two (14). For the central, doubly linked β50/61 chromophores, the center-to-center separation increases when comparing closed and open-form PBPs by 6.3Å in the phycocyanin proteins (compare PC645 with PC612; Table 2) and by 4.8Å in the phycoerythrin proteins (compare PE545 with PE555; Table 2). This increased separation dramatically weakens electronic couplings in the open-form PBPs compared with those determined for the closed structures (Table 2). In particular, the very strong Coulombic interaction calculated within the central β50/61 bilin pair of the closed-form structures of PC645 (647 cm⁻¹) and PE545 (166 cm⁻¹) is absent in the open-form PBPs, PC612 (29 cm⁻¹) and PE555 (4 cm⁻¹) (Table 2). Previous work has shown that the strong couplings calculated for the closed-form PC645 and PE545 are consistent with spectroscopic data (15, 16). Steady-state spectra for the PBPs are documented in Fig. S3. These data include circular dichroism spectra, which are good experimental indicators of exciton interactions. However, the primary derivative-like feature in the PBP spectra comes from subtle interactions among the peripheral chromophores, not the central dimer, and therefore is present in spectra of both the closed and open forms of the protein. A decrease is observed for all interdimer αβ–αβ couplings, whereas the intradimer αβ couplings are relatively unaffected by the different quaternary structure (Table 2 and Table S4).

Spectroscopy of the Closed-State PC645 Versus the Open-State PC612.

The general features of the absorption spectra of PC645, PC612, and PE555 (Fig. 5 A–C) are consistent with the chromophore compositions (Table 1). Previous modeling of the spectroscopy of PC645 suggests a model for the absorption band positions (the center positions of absorptions indicated in Fig. 5D) (16). The three different chromophore types (DBV, MBV, and PCB) provide the primary spectral broadening and establish an energy funnel from the core to periphery of the complex (17). PC612 is similar to PC645 in that its absorption spectrum is broad, with two distinct peaks, and absorbs in the same region, but it does not contain MBV chromophores (Table 1). Based on the relative absorption energies of the individual bilins and the assignment of chromophore absorption energies in PC645, we expect that the higher energy peak is dominated by DBVs and that the lower energy peak is dominated by PCBs (Fig. 5E) (18).

Fig. 5.

Electronic spectroscopy of the closed- and open-form PBPs. (A–C) Electronic absorption spectra of closed-form PC645 (A) and open-form PC612 (B) and PE555 (C) PBPs. Spectra were recorded at 295 K (red trace) and 77 K (blue trace). (D–F) Representative 2D electronic spectra at 295 K for the closed-form PC645 (D, at waiting time T = 55 fs), the open-form PC612 (E, at waiting time T = 100 fs) and the open-form PE555 (F, at waiting time T = 100 fs). The spectra are the real part of the total signal, plotted with 33 evenly spaced contours. The estimated exciton energies of the chromophores are plotted on the 295-K absorption spectra which are superimposed onto the excitation and emission axes. (G–I) Magnitude of the 2D ES amplitude at selected cross-peaks as a function of waiting time taken as a trace from the absolute value 2D ES spectra (the first 15 fs are omitted to avoid possible nonresonant solvent response). Cross-peak coordinates (excitation, detection) in nanometers are approximately PC645: (570, 600), PC612: (550, 600), and PE555 (540, 560). Error bars indicate one SD as determined from three trials for PC645 and PC612 and six trials for PE555.

Two types of spectral shift are evident in the PC645 spectrum. First, the DBV chromophores are positioned closely in the closed-form crystal structure and are particularly strongly coupled (Table 2). This electronic coupling splits the DBV absorption bands into the two exciton states labeled DBV(+) and DBV(−). Second, the degeneracy of the PCB absorption bands is broken. Atomistic modeling of PE545 indicates that spectral shifts are caused mainly by perturbations of the chromophore conformation with smaller effects caused by electrostatic interactions with the local protein environment (15). Neither of these two spectral broadening features appears to be evident in the PC612 spectrum. For example, the absorption spectrum recorded with the sample at 77 K (Fig. 5B) clearly reveals only two bands and a vibronic tail on the blue side of the spectrum. Exciton splitting is absent in the open-form structure of PC612 (Table 2) because the doubly linked β50/61 DBV chromophores are widely separated.

PC645 has proven to be a remarkable system because of the clear coherent oscillations seen in cross-peaks in the 2D ES as a function of pump-probe waiting time (10, 19, 20). Complementary nonlinear experiments also have identified these coherences (21, 22). A representative total, real 2D ES spectrum is shown in Fig. 5D. The spectrum shows numerous features of interest, most distinctly an off-diagonal cross-peak (located at excitation wavelength 570 nm and signal wavelength 600 nm in Fig. 5D) that oscillates strongly as a function of the waiting time (Fig. 5G) (10). An extensive analysis of the oscillations has been reported (19). Using a procedure involving the separation of the total spectrum into its rephasing and nonrephasing components (20), we concluded that the oscillations involve both vibrational and electronic coherences (19).

Unlike PC645, little is known about the photophysics of the open-form PBPs PC612 and PE555. For PC612, we specifically photoexcited the DBV states to compare the 2D ES measurements directly with those of PC645. The 2D ES total, real spectrum (Fig. 5E) shows a rectangular feature, centered at the DBV bleach, suggesting substantial coupling between vibronic transitions throughout this spectral region; however, a striking oscillating cross-peak like that noted for PC645 is not evident in these data. In Fig. 5H, a trace at an off-diagonal position (excitation wavelength 550 nm and signal wavelength 600 nm in Fig. 5E) shows damped oscillations that have frequencies consistent with the vibrational beats (19).

Spectroscopy of the Open-State PE555.

In contrast to PC645 and PC612, the PE555 absorption spectrum is narrow and nearly featureless, showing minimal change between the room temperature and 77-K spectra (Fig. 5C). The presence of two distinct types of chromophores in PE555 (singly bonded PEBs and doubly bonded DBVs) suggests that there could be at least two distinct absorption energies. By fitting the room temperature and 77-K absorption spectra using two Gaussians, we obtained estimates of the absorption energies. Based on the relative site energies of the PEBs and DBVs in PE545, the higher-energy shoulder is attributed to the PEBs, and the lower-energy, main absorption band is attributed to DBVs. This ordering agrees with previous assignment (Fig. 5F) (23, 24). The narrow, congested absorption spectrum suggests that the absorption bands of individual chromophores overlap significantly and that exciton splitting and energy shifts caused by different bilin conformations are minimal. As in PC612, this observation is consistent with the idea that the open-form complex is more symmetric than the closed-form complex. The 2D ES total, real spectrum (Fig. 5F) corroborates this expectation, appearing qualitatively similar to that recorded for PC612. Oscillations in the amplitude were weak compared with the other complexes (Fig. 5I; see the trace for excitation wavelength 540 nm and signal wavelength 560 nm in Fig. 5F) and appear most consistent with vibrational coherences.

Discussion

Crystallographic analyses reveal two very different quaternary structures that are adopted by distinct cryptophyte PBPs: the open and the closed forms. To date, both open-form PBPs come from Hemiselmis spp., whereas the two closed-form PBPs come from two distinct cryptophyte subclades (25). The αβ monomer structure is conserved in all cryptophyte PBP structures that have been determined. This structural conservation includes the arrangement of chromophores in both α and β subunits. The interface between the two αβ monomers forming the dimer depends mainly on contacts between α subunits and particularly the α subunit chromophore, which makes key contacts across the dimer interface. Thus, the α subunit mediates dimer formation and hence determines the quaternary structure of the cryptophyte light-harvesting protein.

Sequence and structural analysis shows that the insertion of an aspartic acid residue just before the covalent chromophore attachment site (Cys20) in the Hemiselmis α subunits results in a rotation of the first pyrrole ring A of the α chromophore by ∼69° compared with the closed form. This rotation precludes the assembly of the Hemiselmis αβ monomers into the closed-form dimer, because such assembly would result in a severe steric clash. It appears that the observed open form is the next available dimeric state in terms of minimizing free energy. The buried surface area between monomers in the open-form dimer is approximately half that observed for closed-form dimers, making the open form a less stable structure. However, we note that we have observed only dimers with no evidence of free αβ monomers.

The closed-form cryptophyte PBP dimers are clearly asymmetric in structure and α subunit sequence, with the long α subunit providing a C-terminal extension that mimics the loop between helices G and H in the β subunit and interacts with the doubly linked β50/61 chromophore (4). In contrast, the open-form Hemiselmis PBP dimers are nearly symmetric in structure and sequence. Only minor structural differences are observed in PE555, and no significant differences are observed in PC612.

The key difference between the two quaternary forms in terms of chromophore arrangement is the van der Waals contact between the two doubly linked β50/61 chromophores on the pseudo-twofold axis of the closed-form PBPs. This arrangement is unique to closed-form cryptophyte PBPs, and it results in the strong excitonic coupling of these two chromophores (Table 2). This pairing of chromophores is completely disrupted in the open-form quaternary structure, where these two chromophores are separated by a water-filled central channel, reducing the excitonic coupling (Table 2).

The common chromophore across each of the PBPs investigated in this work is DBV, in each case the doubly linked β50/61 chromophore. Consequently, strong electronic coupling among the central bilins is the key distinguishing feature of the closed- versus open-form PBPs with respect to light harvesting. Spectral shifts resulting from a combination of local environment and conformational effects extend the PC645 absorption ∼25 nm further to the red than that of PC612. PC645 further incorporates MBV chromophores to absorb at ∼600 nm. In comparison, the absorption cross-section of PE555 is unusually narrow, about half the spectral width of PC645. The great diversity of solutions to light harvesting in the cryptophyte algae, in particular the combination of different chromophores and significantly different structural combinations, are quite extraordinary.

The ancestral cryptophyte alga acquired its chloroplast by engulfing and taming a red algal endosymbiont, which would have had at least a primitive phycobilisome (7, 24). Extant cyanobacterial and red algal phycobilisomes are complex structures made up of stacked rods of several types of trimeric PBPs (αβ)₃ rings attached to the stromal surface of the thylakoid membrane (26), however, the phycobilisome α subunits are globin proteins that are completely unrelated to the cryptophyte α subunits. At some point during the integration of the red algal plastid, the phycobilisome structure disappeared, and the original globin-fold α subunit was replaced by an unrelated polypeptide of unknown evolutionary origin and was retargeted to the thylakoid lumen (2). This replacement resulted in the first radical rearrangement in quaternary structure: the formation of a cryptophyte progenitor (αβ)₂ dimer (4). Here we report that among the cryptophytes there are two radically different forms of the α₁β.α₂β dimer, the open-form dimer, which appears to be confined to the Hemiselmis lineage, and the closed form. The emergence of two forms appears to have been caused by a single insertional/deletional mutation in the new cryptophyte α subunit.

A central biological question is whether the presence of long-lived electronic coherence in the light-harvesting proteins results in a selective advantage for the algae—for example, is coherence important for efficient light harvesting? If the emergence of long-lived electronic coherence gives cryptophytes containing the closed-form PBP a selective advantage over the ancestral cyanobacteria and red algae, it would seem that the Hemiselmis cryptophytes, with their open-form PBPs, have lost this advantage. Our results suggest that successful light harvesting can be achieved in diverse ways, with or without coherent molecular excitons delocalized over pairs of chromophores. Nevertheless, it is apparent that the excitonic interactions in the PBPs are switched profoundly (over an order of magnitude) by the structural change from open to closed and that this exciton switch is genetically controlled.

Materials and Methods

Growth of Cryptophytes.

Chroomonas sp.

Strain CCMP 270 (Provasoli-Guillard National Center for Marine Algae and Microbiota, Bigelow Laboratory for Ocean Sciences) was cultured in modified Fe medium (27) at 22–24 °C under continuous aeration with a 12-h light/12-h dark cycle at a light intensity of 80–100 mE⋅m⁻²⋅s⁻¹. When cultures reached a density of 5 × 10⁸ cells/L (4–6 wk), cells were harvested by centrifugation. The pellet was resuspended in 25 mM phosphate buffer (pH 7) and was stored at −80 °C.

H. andersenii.

CCMP 644 (Provasoli-Guillard National Center for Marine Algae and Microbiota, Bigelow Laboratory for Ocean Sciences) was grown in GSe medium (28) at 22–24 °C under a 12-h light/12-h dark cycle at a light intensity of 80–100 μE⋅m⁻²⋅s⁻¹. Cells were harvested after 4–6 wk by centrifugation and were stored at −80 °C.

H. virescens.

CCAC 1635B (Culture Collection of Algae at the University of Cologne) cultures were grown in aerated ASP-H medium (29, 30) at 16 °C under a 14-h light/10-h dark cycle with light intensities of ∼50 µmol photons⋅m⁻²⋅s⁻¹. Cultures were harvested by flow-through centrifugation and were stored at−80 °C.

Protein Purification.

Algal cell pellets were thawed, resuspended in two to three volumes of 25 mM phosphate buffer (pH 7), and homogenized with a Teflon glass homogenizer at 30 rpm. Cells were disrupted in a French press at 1,000 psi and centrifuged at 23,000 × g for 1 h at 4 °C. The supernatant was purified via ammonium sulfate cuts (0–50%, 50–60%, 60–70%, and 70–80%) by adding solid ammonium sulfate, stirring for 1 h at 4 °C, and centrifuging at 23,000 × g for 30 min at 4 °C. The 70–80% pellets were resuspended in 25 mM phosphate buffer (pH 7), filtered, dialyzed against the same buffer, and loaded onto a Q Sepharose HiLoad 26/10 anion exchange column (GE Healthcare). The fractions containing the majority of the light-harvesting protein were selected using the absorbance at 280 nm and concentrated on a 10-kDa Centriprep (Millipore). The protein was purified by size-exclusion chromatography using a Superdex 200 HiLoad 26/60 column (GE Healthcare). Proteins eluted as a single peak and were concentrated using a 10-kDa cutoff (Centriprep; Millipore) before snap freezing and storage at −80 °C.

Crystallization.

The proteins were crystallized using vapor diffusion under the following conditions: PC645, 20% PEG 4 k, 20% isopropanol, and 0.1 M sodium citrate (pH 5.6); PE555, 20% PEG 10 k in 0.1 M Hepes (pH 7.5); and PC612, 25% PEG 3,350 in 0.1 M Hepes (pH 7.5).

Data Collection.

Crystals were transferred to the cryoprotectant solution of reservoir plus 15% glycerol and then were flash-cooled in liquid nitrogen and mounted in a Cryostream cooler (Oxford Cryosystems) for data collection: PC645 on beamline 9.2, Stanford Synchrotron Radiation Lightsource; PE555 on beamline 23ID-D, Advanced Photon Source; and PC612 om beamline MX1, Australian Synchrotron (Table S1). Diffraction images for PC645 and PE555 were collected on a MarCCD (Rayonix) detector, and diffraction images for PC612 were collected on an ADSC Q210 (Area Detector Systems Corporation) detector. Data collection was carried out using Blu-Ice (31).

Data Reduction and Structure Determination.

All data were processed using XDS (32) and SCALA [CCP4 (33)]. Phasing, auto building, and refinement were carried out using PHENIX (34). A single β subunit from the structure of PE545 (3) was used as a molecular replacement probe using PHASER (35) as implemented in PHENIX. Manual adjustments were carried out using COOT (36). Structural figures were created using PYMOL (37).

Structure of PC645.

The PC645 structure contains one α₁β.α₂β dimer in the asymmetric unit. The complete α₁ molecule is visible in the electron density map, whereas the last two residues of α₂ (Lys⁶⁹–Lys⁷⁰) are disordered. The C-terminal residue of each β subunit (Ala¹⁷⁷) is disordered. The first 14 residues of the β subunit (helix hX) are absent in the electron density, their position being occupied by an ordered PEG molecule that extends across a crystallographic twofold axis. The only modified residue is Asn72 in the β subunit where the side-chain nitrogen is methylated (5).

Structure of PC612.

The PC612 structure contains a single (αβ)₂ homodimer in the asymmetric unit. Electron density is seen for the complete α subunits and one of the β subunits. In the second β subunit, electron density starts at Asp3 and continues to the C terminus.

Structure of PE555.

The PE555 structure has three copies of the (α₁β).(α₂β) heterodimer in the asymmetric unit. The first heterodimer shows clear electron density for the full α₂ subunit and all but the last residue (Val67) of α₁. The β chain associated with α₂ is complete, whereas the remaining β chain starts at Asp3. The heterodimer structure is nearly symmetric.

The other two copies of the heterodimer appear to be superpositions of an (α₁β).(α₂β) dimer with an (α₂β).(α₁β) dimer (i.e., rotation of 180° about the pseudo-twofold axis). In each α subunit, electron density is seen for both possible C-terminal regions [as seen in the asymmetric (α₁β).(α₂β) structure]. There are five residues within the first 60 that distinguish α₁ from α₂. These residues show density (or lack of density) for both possible α chains, in particular α₁Phe45/α₂Met45.

Sequence Determination for H. virescens PC612, Chroomonas sp. CCMP270, and H. andersenii CCMP644.

Algal cells were grown to exponential stage at 20°, 20 μmol photons⋅m⁻²⋅s⁻¹ on a 12-h light/12-h dark cycle. RNA was isolated using RNAqueous 4 PCR (Ambion) or Total RNA Isolation Reagent (Advanced Biotechnologies). cDNA was generated using SuperScript III Reverse Transcriptase (Invitrogen Life Technologies) at 50 °C with random hexamer/nonamer or oligo (dT) as primers and was used for degenerate PCR. The β subunit degenerate primers were designed based on the alignment of DNA sequences from cryptophytes Guillardia theta and Rhodomonas salina and all the red algal β subunit sequences in GenBank. The α subunit degenerate primer pairs were based on Edman N-terminal sequencing (CCMP270) or on the best partial amino acid sequences derived from electron-density maps. PCR products were cloned into T-vectors, and isolated colonies were selected randomly for sequencing. The resulting sequences were used to design outward-directed PCR primer pairs for cDNA-based inverse PCR according to Huang and Chen (38). For the complete α subunit sequences, 5′ RACE was done with the FirstChoice RLM -RACE kit (Ambion) or ExactSTART kit (Epicentre). For the 3′ end of the β subunit, genomic DNA-based inverse PCR was performed. The assembled sequences were confirmed by PCR from the start codon to beyond the stop codon using specific nondegenerate primers.

Quantum Chemical Calculations.

The initial conformations of the phycobilins were extracted from the Protein Data Bank file with the covalently bound cysteine residue. The cysteine residues were capped with an acetyl and N-methyl amino group. Each of the tetrapyrrole nitrogens was protonated (resulting in a +1 charge), and the two solvent-exposed carbocyclic acid chains were deprotonated (resulting in a −2 charge) with an overall molecular charge of −1 on all phycobilins. The molecular charge also is consistent with polarizable continuum pK_a calculations implemented using the universal solvation model designated for solvation model density (39). Hydrogen atoms were added and optimized using b3lyp/cc-pvtz, followed by a bond-length optimization with dihedral angles restrained using the Gaussian 09 software package (Gaussian, Inc.).

The phycobilin transition density from the S₀ to the S₁ state was obtained from a configuration interaction singles (CIS)/cc-pvtz calculation, again using the Gaussian 09 software package. Transition density cubes were inspected visually to ensure the proper excited state was probed. The gas-phase couplings were computed from the transition densities using the methodology outlined by Krueger et al. (40). No scaling of the couplings was used to correct for the overall overestimation of the transition dipole moments by the CIS because they were overestimated by only ∼9% compared with the experimental value of 2.34 eÅ (41).

2D Electronic Spectroscopy.

A 5-kHz Ti:sapphire amplified laser system produces 150-fs pulses centered at 800 nm with ∼0.6 mJ. About 0.15 mJ seeds a noncollinear optical parametric amplifier (NOPA), producing pulses with a bandwidth of 60 nm (spectral intensity FWHM) centered in the Vis range, nearly free from angular dispersion. A combination of a folded 4-f grating compressor and a single-prism prism compressor is used to compress the pulse from the NOPA. Pulse compression is determined by measuring the nonresonant third-order response from methanol in the sample position. The pulse duration, central wavelength, and bandwidth used for each protein sample are summarized in Table S6. The beam is attenuated by the combination of a broadband half-waveplate and a 0.7-mm-thick wire-grid polarizer before entering the four-wave-mixing setup (20).

A spherical mirror with a 50-cm focal length focuses the beam on a 2D cross-hatched phase mask (UV fused silica substrate). The four first-order diffraction beams, each about 12% of the input power, are arranged in the BOXCARS configuration and are directed by a small steering mirror toward the mirror with a 50-cm focal length. The steering mirror allows us to use the large spherical mirror at an angle of exactly 0°. The large spherical mirror collimates and makes parallel the four beams, which then pass above, below, and to the sides of the steering mirror. Three of the beams traverse antiparallel pairs of 1°, uncoated UV fused silica glass wedges for pulse delays. One wedge from each pair is mounted on a computer-controlled delay stage allowing a step accuracy of ∼850 zeptoseconds. The pulse that serves as the final excitation field interaction is chopped at a frequency of 25 Hz; the chopper also triggers the detector, which acquires signal for 20 ms (100 laser shots). The fourth beam, the local oscillator (LO), is attenuated by 10⁴ and interacts with the sample ∼250 fs before the final excitation field, thus limiting its use to a reference field. The four pulses then encounter a second spherical mirror used at 0° but with a focal length of 20 cm. The pulses focus and cross in the sample plane (beam waist diameter ∼50 μm) after encountering another small steering mirror. The signal and LO are collimated by a curved mirror with a 20-cm focal length and are directed to an imaging spectrometer (f = 16.3 cm) coupled to a CCD detector with 1,024 pixels in the dispersed dimension. The spectrometer is calibrated using an Hg/Ar lamp. The phase stability of the apparatus is λ/350 short term (5 min) and λ/200 long term (2 h). All 2D ES experiments are performed with pulse energies of ≤5 nJ. The coherence time (τ₁) was scanned in intervals of 0.2 fs (0.15 fs) from −45 to 45 fs. The waiting time (τ₂) ranged from 0–400 fs in 5-fs steps for PC645 and PC612 measurements; the PE555 measurement was monitored at 10-fs intervals. The measurements were conducted at 298 K, and the sample was flowed using a peristaltic pump at a flow rate that guaranteed a fresh spot at each pulse.

Protein samples were stored at −75 °C until required and then were thawed and diluted in a suitable buffer to the appropriate OD for spectroscopic measurements (OD_{λ max} <0.4).