Probing the CD Lumenal Loop Region of the D2 Protein of Photosystem II in Synechocystis sp. Strain PCC 6803 by Combinatorial Mutagenesis

Abstract

The CD lumenal loop region of the photosystem II reaction center protein D2 contains residues involved in oxygen evolution. Since detailed structural information about this region is unavailable, an M13-based combinatorial mutagenesis approach was used to investigate structure-function relationships in this vital region of D2 in Synechocystis sp. strain PCC 6803. The CD loop coding region contains close to 100 nucleotides, and for effective mutagenesis, it was subdivided into four regions of seven to eight codons. A gain-of-function selection protocol was employed such that all mutants that were selected contained a functional D2 protein. In this way, conservation patterns of residues along with numbers and types of amino acid substitutions accommodated at each position for each set of mutants would indicate which residues in the CD loop may play important structural and functional roles. Results of this study have substantiated the importance of residues previously studied by site-directed mutagenesis such as Arg180 and His189 and have identified other previously unremarkable residues in the CD loop (such as Ser166, Phe169, and Ala170) that cannot be replaced by many other residues. In addition, the pliability of the CD loop was further tested using deletion and D1-D2 substitution constructs in M13. This showed that the length of the loop was important to its function, and in two cases, D2 could accommodate homologous sequences from D1, which forms a heterodimer with D2 in photosystem II, but not the other way around. This study of the CD loop in D2 provides valuable clues regarding the structural and functional requirements of the region.

Photosystem II (PS II), located in the thylakoid membrane of plants and cyanobacteria, converts light energy to chemical energy in a unique photochemical process that results in the splitting of two molecules of water to produce four reducing equivalents and protons and to release one molecule of oxygen. The oxygen-evolving complex (OEC), which is located on the donor side of PS II, oxidizes water via a complex containing manganese. Each photon of light that oxidizes the primary donor of PS II, P680, extracts one electron from the OEC, and after four positive charges have been accumulated, two water molecules are oxidized and oxygen is released. Cofactors involved with electron transfer in PS II are associated with the reaction center proteins D1 and D2. The electrons extracted from the manganese cluster are transferred to P680⁺ via the intermediate electron carrier, Y_Z (tyrosine 161 of D1). Site-directed mutagenesis studies confirmed that this tyrosyl residue is in redox contact with the primary donor, P680, and the OEC (18). Symmetrically arranged in relation to Y_Z is Y_D, tyrosine 160 of D2. This redox-active Tyr residue forms a dark-stable radical that is not directly involved in the reduction of P680⁺ by the water-splitting complex (5, 32). These functional differences between Y_Z and Y_D may be related to regional structural differences in the protein environment surrounding the radicals and the OEC.

One of the regions of PS II that possibly interacts with the OEC and that contains residues close to Y_Z and Y_D as well as to P680 is the CD lumenal loop region of the D1 and D2 proteins. The CD lumenal loop that is present in both of these proteins connects the transmembrane helix C, containing Y_Z or Y_D, with helix D, which contains His residues that are likely to be ligands to the primary donor P680 and the electron acceptors Q_A and Q_B (reviewed in reference 34). Based on comparison with the reaction center from purple bacteria, the CD loop in the two proteins is expected to contain a helical region that interacts with the lumenal edge of the thylakoid membrane and that may contain a central ligand to accessory chlorophyll in the reaction center.

The CD loop of D1 and D2 has been studied using site-directed mutagenesis, and residues Asp170, Glu189, and His190 in D1 along with Gln164, Arg180, and His189 in D2 have been identified to interact with the OEC, the redox-active tyrosines, and/or P680. Asp170 is thought to be a Mn ligand or to be involved with assembly of the cluster (4, 21). Another D1 residue, Glu189, also seems to play a structural role in stabilizing and assembling the Mn cluster (4), possibly through its involvement in a network of hydrogen bonds that influence both Y_Z and the Mn cluster (7). Alterations in His190 of D1 and His189 of D2 indicate that these residues may serve as proton acceptors upon Tyr oxidation (4, 11, 26, 28). Mutational studies indicate that Arg180 of D2 affects Y_D and P680 function and, based on the purple bacterial structure, may be involved in accessory chlorophyll binding (16). Mutations in Gln164 have been shown to alter the electron paramagnetic resonance (EPR) signal of Y_D (28).

While these residues appear to be functionally important, little is known about the structural requirements in the CD loop and about which other residues in the region are critical for the stable functioning of the donor side of PS II. One approach that has been used to determine functionally and structurally important residues in essential regions of proteins with unknown structure is combinatorial mutagenesis (3, 22).

Combinatorial mutagenesis is a technique that allows for simultaneous introduction of many random, degenerate mutations in a large protein domain. By coupling the mutagenesis with a selection protocol that screens only for functional proteins, the data obtained can be used to gain information on the structurally and functionally important residues in a selected region (8, 20, 36). Various combinatorial techniques have been successfully used to study structure and function in photosynthetic reaction centers. The residues required for the Q_B binding niche and the region surrounding Y_Z in the D1 protein of Synechocystis sp. strain PCC 6803 have been studied using a PCR-based combinatorial method (12, 13). To generate a set of functional mutants for electron transfer studies, combinatorial cassette mutagenesis was used to randomly mutagenize regions of the L and M subunits near the active-branch monomeric bacteriochlorophyll in Rhodobacter capsulatus (23). Moreover, an M13-based combinatorial method was recently used to study a conserved region in the E loop of the CP47 chlorophyll binding protein in Synechocystis sp. strain PCC 6803 (27).

In this study, we applied combinatorial mutagenesis, also using an M13-based technique, to develop a set of mutants in the CD loop of the D2 protein in Synechocystis sp. strain PCC 6803 to gain insight into structure and function relationships within the CD loop region of this protein. As a transformable, oxygenic prokaryote, Synechocystis sp. strain PCC 6803 is a simple model system for studying PS II structure and function by genetic manipulation (35). Moreover, knowledge gained from this system is directly applicable to eukaryotic plant systems.

For effective mutagenesis, the CD loop was divided into four regions of seven to eight amino acids (see Fig. 1). Within each region, the codons were replaced with random sequences in psbDI (the only D2 gene in the system that was studied) and photoautotrophic mutants were selected. In addition, seven- to eight-residue regions of the CD loop of D1 and D2 were replaced by the homologous regions of D2 and D1, respectively. Since the two proteins have only nine residues in common in the CD loop and yet appear to have structural symmetry, substituting homologous D1-D2 sequences provides insight into the residues that contribute to functional differences between D1 and D2. In addition to development of combinatorial and substitution mutants in each region, the CD loop was further probed by attempting to isolate photoautotrophic deletion mutants for each of the four regions. This comprehensive genetic study of the CD lumenal loop in the D2 protein of Synechocystis sp. strain PCC 6803 identifies important residues not readily determined by primary sequence analysis and provides a unique set of mutants that contribute valuable information on regional primary structures that can support oxygen evolution in the PS II complex.

FIG. 1.

(A) Protein sequence of the CD lumenal loop region of the D1 and D2 proteins in Synechocystis sp. strain PCC 6803 showing the subdivision of each loop into four regions of seven or eight codons (regions numbered 1 to 4 for D1 and 5 to 8 for D2). (B) DNA sequence of psbDI along with an illustration showing three primers that were designed to introduce changes into region 7 in psbDI. Similar sets of deletion, D1-D2 substitution, and combinatorial primers were designed for each region. Nucleotides in the substitution primer shown in italics are the substituted bases from the corresponding region of the psbAII gene. Shown underlined and in boldface is a silent substitution introduced to enhance primer-template annealing. Degenerate primers were synthesized with an equimolar mix of all four nucleotides (shown as N) at the first and second nucleotide positions of each codon and an equimolar mix of guanines plus cytosines (shown as S) at the third (wobble) position.

MATERIALS AND METHODS

Acceptor strains.

The genome of Synechocystis sp. strain PCC 6803 carries two genes that code for the D2 protein, psbDI and psbDII, and three genes that code for the D1 protein, psbAI, psbAII, and psbAIII. To achieve successful mutagenesis and to be able to utilize gain-of-function selection protocols, it was necessary to develop acceptor strains for transformation that lacked all but one psbA and/or psbD gene, and where the remaining copy carried a CD loop deletion. In the D2ΔCD strain, the psbDII gene was replaced by a spectinomycin resistance cassette (33), while the psbDI gene in this strain carries a CD loop deletion resulting in the loss of residues G163 to P195 of D2 and has a kanamycin resistance cassette downstream (33; I. Shalak and W. Vermaas, unpublished data). Therefore, this strain is unable to make functional D2 protein.

The Synechocystis sp. strain PCC 6803 D1ΔCD strain, which is unable to make a functional D1 protein, lacks psbAI and psbAIII and carries a deletion of the region coding for the CD loop in psbAII. The psbAI gene in the D1ΔCD strain was replaced by a spectinomycin resistance cassette, and the psbAIII gene was replaced by a chloramphenicol resistance cassette (6). The CD loop deletion in psbAII was introduced into this psbAI-psbAIII double-deletion strain in the following way. First, the plasmid pAIIEr was constructed by cloning a 2.0-kbp genome fragment carrying most of psbAII (starting at an Eco47III restriction site located 27 bp downstream of the start codon) and ∼900 bp of the downstream region into pUC19. This was followed by insertion of an erythromycin resistance cassette into the StuI site ∼300 bp downstream of psbAII (Fig. 2). The plasmid was then linearized by KpnI, which cuts in psbAII at a site coding for residues in the CD loop, and digested with exonuclease Bal 31. After ligation, the plasmid pAIIErDEL, containing a CD loop deletion in psbAII leading to the loss of residues V157 to G201 of D1, was obtained. This deletion plasmid was then used to transform the psbAI-psbAIII deletion strain; after segregation, the D1ΔCD strain of Synechocystis sp. strain PCC 6803 was obtained. Since both D1ΔCD and D2ΔCD lack functional PS II, they are obligate photoheterotrophs.

FIG. 2.

Map of the pAIIEr plasmid used in developing the D1ΔCD strain. The plasmid carries a Synechocystis sp. strain PCC 6803 genomic fragment (shown in black) of ∼2.0 kbp extending from the Eco47III site located 27 bp downstream of the psbAII start codon to the HindIII site ∼900 bp downstream of the psbAII stop codon. The genome fragment is interrupted downstream of psbAII by an erythromycin resistance cassette which was introduced into the StuI site ∼300 bp downstream of the coding sequence. The CD loop deletion introduced into this plasmid to make pAIIErDEL, which was used to transform Synechocystis sp. strain PCC 6803 to develop the D1ΔCD strain, is indicated.

Primer design.

Since the CD loop region is too large to be effectively mutagenized as a whole, the region was divided into four domains of seven or eight amino acid residues each (Fig. 1). Three sets of primers (deletion, D1-D2 substitution, and combinatorial [degenerate] primers) were synthesized for each of the eight regions (Fig. 1). Deletion primers of 40 to 45 nucleotides were designed such that approximately half of the sequence corresponded to the wild-type sequence immediately upstream of the deletion region and the remaining half corresponded to the wild-type sequence immediately downstream of the deletion region.

D1-D2 substitution primers ranged in length from 51 to 59 nucleotides. The substitution primers for each region contained the wild-type sequence for psbAII or psbDI flanked by 13 to 19 nucleotides upstream and downstream of the wild-type sequence of the opposite gene. The substitution sequence in the middle of the primer was modified where possible at the third codon position to improve hybridization to the other gene sequence without altering the amino acid sequence.

Combinatorial oligonucleotide primers ranging from 62 to 69 nucleotides were synthesized with 18 to 24 nucleotides of the upstream and downstream wild-type sequence flanking a 21- to 24-nucleotide random degenerate sequence for each region of the CD loop. The degenerate region of the primers was synthesized with an equimolar mixture of A, T, C, and G at the first and second codon positions and an equimolar mixture of G and C at the third (wobble) position. By using this protocol, two of three stop codons were eliminated. Moreover, this led to a more equalized probability for the incorporation of the various amino acids (9, 10).

Mutagenesis procedure.

The procedure used to generate combinatorial DNA for transformation is based on established protocols for in vitro mutagenesis of single-stranded bacteriophage templates (2, 27, 33). Two M13mp19 bacteriophage constructs, M13mp19/psbAII and M13mp19/psbDI, were used as templates for mutagenesis. M13mp19/psbDI was generated as described in reference 33. M13mp19/psbAII was constructed by cloning a 1.2-kbp EcoRI/XbaI psbAII fragment cut from pAIIEr (Fig. 2) into the polylinker region of M13mp19. These bacteriophage constructs were propagated in Escherichia coli strain CJ236 (dut, ung negative) to allow for incorporation of uracil into the templates as a method for preferentially selecting against the template strand after mutagenesis (14). The procedure used for transformation of the appropriate Synechocystis sp. strain PCC 6803 acceptor strain (D1ΔCD or D2ΔCD) with deletion, D1-D2 substitution, and degenerate combinatorial M13 phage constructs was based on established protocols (27, 31, 37).

To introduce seven- to eight-codon deletions or D1-D2 substitutions, deletion and D1-D2 substitution primers were hybridized to the appropriate uracil-containing M13mp19 template containing the wild-type psbAII or psbDI gene. The second strand was synthesized in vitro using methylated dCTP, which results in a double-stranded, hemimethylated M13 recombinant phage. The recombinant M13 is subsequently nicked with restriction enzyme MspI or Sau3AI, which recognizes hemimethylated DNA and cuts only the nonmethylated wild-type strand (29). Upon introduction of the heteroduplex M13 into E. coli strain DH10B for amplification, the nicked template strand is preferentially destroyed, providing a second selection against the template strand, thus increasing the yield of M13 mutants among the plaques. Single-stranded M13 was then isolated from individual plaques and prepared for sequencing to verify the presence of the deletion or substitution. Subsequently, double-stranded M13 DNA carrying the desired deletions and substitutions was isolated for transformation into the appropriate Synechocystis sp. strain PCC 6803 acceptor strain.

Degenerate combinatorial primers were hybridized to single-stranded uracil-labeled M13mp19/psbDI deletion templates. Following in vitro synthesis with unmethylated dCTP and phage amplification in E. coli, the isolated pools of double-stranded phages were directly transformed into the Synechocystis sp. strain PCC 6803 D1ΔCD acceptor strain. The deletion template strands for each region were unable to sustain photoautotrophic growth, thus obviating the need for further selection at the level of M13 and allowing for direct selection of photoautotrophically competent transformants in the Synechocystis sp. strain PCC 6803 acceptor strain. Colonies of photoautotrophic transformants were visible within 1 to 2 weeks after transformation.

DNA sequencing.

The sequences of the Synechocystis sp. strain PCC 6803 photoautotrophic combinatorial mutants were determined by PCR amplification of the CD loop region of psbDI from genomic DNA preparations of the mutants. Single-stranded M13 deletion and substitution constructs were prepared for sequencing using established protocols (14, 17). The region coding for the CD loop and its flanking regions (both as PCR products and as single-stranded M13 deletion and substitution constructs) were sequenced using an ABI 377 DNA sequencer.

RESULTS

Deletions and D1-D2 substitutions in the CD loop.

Eight deletion and eight substitution constructs were introduced into the appropriate acceptor strains of Synechocystis sp. strain PCC 6803 (Fig. 1). The aim of this part of the project was to answer two basic questions: (i) is the length of the loop region important to its function, and (ii) can a homologous region from D1 function in D2 and vice versa? Photoautotrophic growth could not be restored by transformation with any of the eight deletion constructs, indicating that removal of seven to eight amino acids in the CD loop of either D1 or D2 does not allow for formation of a functional PS II complex. In addition, replacement of any of the four D1 regions with corresponding domains of D2 did not lead to photoautotrophic mutants. This indicates that the primary sequence of the CD loop of D1 cannot accommodate major changes without a loss of PS II function. However, D2 was found to functionally accommodate substitution (Fig. 1) by the homologous region from D1 in regions 6 (Pro171 to Ile178) and 8 (Gly187 to Asn194), whereas regions 5 (Gln164 to Ala170) and 7 (Phe179 to Gln186) could not be replaced by the D1 analogs without a loss of photoautotrophic capacity.

The substitution mutant S6, which carried a homologous D1 sequence in region 6 of the D2 protein, displayed a photoautotrophic growth rate similar to that of wild type (Table 1) over a range of light intensities. These data suggest that region 6 may be a linker region in D2 that has no functional role in redox activity and oxygen evolution.

TABLE 1.

Photoautotrophic doubling times at various light intensities of substitution mutants where corresponding D1 sequences have been incorporated into region 6 (S6) and region 8 (S8) of D2 (Fig. 1)

Synechocystis strain	Doubling time (h) at light intensity (μE m−2 s−1)a
	20	40	75
Wild type	18 ± 2.0	12 ± 1.9	14 ± 1.4
S6	21 ± 0.4	14 ± 2.3	12 ± 0.5
S8	27 ± 4.2	22 ± 5.0	18 ± 2.7

^aValues reported are averages of at least three independent measurements. 

In contrast, the substitution mutant S8, carrying a homologous D1 sequence at residues 187 to 194 of D2, had a significantly lower growth rate than that of the control; the photoautotrophic doubling time of the S8 strain was ∼60 to 80% longer than that of the control (Table 1). The altered D2 region in the S8 mutant includes the functionally important residue His189, which is conserved in D1 (His190) and was therefore not changed in S8. However, the decreased photoautotrophic growth rate of S8 indicates that the surrounding environment is important for PS II function as well.

Combinatorial mutants.

Photoautotrophic combinatorial mutants in the CD loop of D2 were generated in regions 5, 7, and 8. Since the S6 mutant showed a growth rate similar to that of the wild type, we chose to focus our efforts on the remaining three regions in the CD loop of D2. The combinatorial mutants isolated for regions 5, 7, and 8 along with their interesting features will be presented and discussed in the paragraphs below.

Region 5 (Gln164 to Ala170).

The primary sequences of the 45 combinatorial mutants that were obtained for region 5 are listed in Table 2. This region is immediately adjacent to helix C and, in terms of primary sequence, is closest to Y_D. The most highly conserved residues in this set of combinatorial mutants are Ser166 and Phe169. Both positions are 76% conserved and have strict replacement requirements accommodating only six and five different residues, respectively. The 166 position shows a preference for small uncharged but mostly polar residues (Ser, Thr, Asn, and Gly) or Ala or His. Position 169 strongly prefers a hydrophobic residue (Phe, Leu, or Val), with His or Thr being found in a few cases.

TABLE 2.

Amino acid sequence of wild-type and combinatorial mutant strains in region 5 of the CD lumenal loop of the D2 protein^a

Strain	Amino acid at sequence position:
	164	165	166	167	168	169	170
Wild type	Q	S	S	W	F	F	A
C5-1	H	S	S	M	V	T	A
C5-2	M	T	A	F	F	F	A
C5-3	Q	G	S	M	A	F	G
C5-4	Q	A	S	N	L	F	G
C5-5	H	D	S	Y	L	F	G
C5-6	H	R	T	F	S	F	A
C5-7	S	S	S	F	Y	F	S
C5-8	H	F	S	F	L	F	S
C5-9	L	E	A	Y	V	F	A
C5-10	Q	A	S	L	V	F	C
C5-11	H	S	S	F	D	F	S
C5-12	H	H	S	D	A	F	A
C5-13	V	G	S	N	F	F	G
C5-14	Q	N	S	L	V	L	A
C5-15	G	S	S	Y	Y	F	G
C5-16	H	W	S	F	F	V	S
C5-17	H	Q	S	F	M	F	A
C5-18	V	D	S	D	Y	F	A
C5-19	S	T	S	L	F	F	A
C5-20	H	L	S	Y	G	F	A
C5-21	H	G	S	F	F	H	A
C5-22	S	P	S	S	A	F	G
C5-23	S	P	S	I	F	F	A
C5-24	H	R	S	F	S	F	S
C5-25	L	Q	S	F	F	F	A
C5-26	Q	A	A	F	T	L	A
C5-27	Q	I	N	W	L	F	C
C5-28	V	G	H	T	F	F	A
C5-29	Q	I	G	Y	S	L	C
C5-30	V	D	S	A	C	F	A
C5-31	G	T	S	F	F	F	C
C5-32	A	P	T	F	S	F	A
C5-33	Q	P	N	G	A	L	A
C5-34	S	P	S	D	F	F	A
C5-35	T	G	S	W	L	F	A
C5-36	Q	C	S	R	L	F	A
C5-37	H	G	T	F	D	H	A
C5-38	F	T	S	F	L	F	G
C5-39	Q	R	S	N	S	F	C
C5-40	H	S	G	A	F	V	A
C5-41	H	C	S	L	A	L	G
C5-42	H	A	S	W	R	V	G
C5-43	H	P	S	G	T	F	G
C5-44	S	P	S	L	L	F	S
C5-45	V	N	S	L	I	F	A
Identity (%)b	20	11	76	7	24	76	53
No. of residues accommodatedc	10	16	6	14	13	5	4

^aResidues in the mutant strains that are identical to the wild type are in boldface, as is the wild-type sequence. Wild-type residues that have been studied by site-directed mutagenesis are underlined and italicized. 

^bIdentity percentages represent the percentages of mutant strains that retain the wild-type residue at the indicated position. 

^cThe number of residues accommodated is an indication of the variability that can be tolerated at the indicated position. 

Another interesting residue in this region is Ala170. This residue is conserved in 53% of the mutants, but at this position, only small uncharged but somewhat hydrophilic residues (Cys, Ser, and Gly) can be functionally accommodated. While Gln164 has been shown to alter the EPR signal of oxidized Y_D (28), it is conserved in only 20% of our mutants, and the position can accommodate 10 different residues, Gln, His, Met, Ser, Leu, Val, Gly, Ala, Thr, and Phe. Positions 165 and 167 are the least conserved in region 5, accommodating 16 and 14 different amino acids, respectively, from large to small and from hydrophobic to charged. Interestingly, position 165 is unique in that it is the only position in the CD loop in any of the combinatorial mutants analyzed thus far that has accommodated a Pro residue.

Region 7 (Phe179 to Gln186).

Combinatorial mutants obtained in region 7 have been listed in Table 3. This region makes up part of the proposed lumenal helix of the CD loop and contains Arg180, which was identified by site-directed mutagenesis as interacting with both Y_D and P680 and which could not be replaced by other residues without major functional consequences (16). Surprisingly, this residue is conserved in only 43% of the mutants shown in Table 3, but if Arg is absent at position 180, an Arg appears in position 184. This position is approximately one full turn of the proposed α-helix away from position 180.

TABLE 3.

Amino acid sequence of wild-type and combinatorial mutant strains in region 7 of the CD lumenal loop of the D2 protein^a

Strain	Amino acid at sequence position:
	179	180	181	182	183	184	185	186
Wild type	F	R	F	I	L	F	L	Q
C7-1	F	G	F	V	V	R	F	L
C7-2	F	V	Y	I	V	R	I	F
C7-3	F	R	W	W	F	F	V	L
C7-4	L	R	W	M	L	F	A	H
C7-5	L	V	I	S	L	R	F	I
C7-6	F	R	F	M	L	I	L	A
C7-7	F	G	F	L	V	R	L	L
C7-8	F	R	F	L	I	F	M	I
C7-9	L	V	L	V	V	R	F	I
C7-10	F	R	W	L	L	F	F	L
C7-11	L	T	F	L	L	R	F	L
C7-12	Y	R	F	V	L	F	T	V
C7-13	Y	V	F	L	I	R	F	V
C7-14	F	R	F	A	L	V	F	V
Identity (%)b	64	43	57	7	50	36	14	0
No. of residues accommodatedc	3	4	5	7	4	4	7	7

^aThe wild-type sequence is in boldface, and residues that have been shown to be of importance in the region by earlier site-directed mutagenesis studies are underlined and italicized. Residues in the mutant strains that are identical to wild type are in boldface, and arginine is underlined where it appears at position 180 or 184. 

^bIdentity percentages represent the percentages of mutant strains that retain the wild-type residue at the indicated position. 

^cThe number of residues accommodated is an indication of the variability that can be tolerated at the indicated position. 

The most conserved residue in the region is Phe179, which appears in 64% of the mutants and is replaced only by two residues, Leu and Tyr. The least conserved residue in region 7 is Gln186; in none of the combinatorial mutants of region 7 was a Gln residue found at position 186, and a mild preference for hydrophobic residues was observed.

Region 8 (Gly187 to Asn194).

The primary sequences of the 34 combinatorial mutants in region 8 of D2 are listed in Table 4 and clearly substantiate the importance of His189 in this region for sustaining photoautotrophic growth. This residue is 100% conserved in the 34 mutants in this region. However, note that the site-directed mutants H189L, H189Y, and H189Q can grow photoautotrophically (26, 28), suggesting the importance of neighboring residues in supporting the stable functioning of this region of the CD loop.

TABLE 4.

Amino acid sequence of wild-type and combinatorial mutant strains in region 8 of the CD lumenal loop of the D2 protein^a

Strain	Amino acid at sequence position:
	187	188	189	190	191	192	193	194
Wild type	G	F	H	N	W	T	L	N
C8-1	G	A	H	A	V	L	Q	L
C8-2	A	F	H	G	F	I	M	Q
C8-3	S	H	H	G	F	V	L	L
C8-4	S	F	H	N	F	L	L	V
C8-5	S	L	H	S	V	L	S	Q
C8-6	G	Y	H	A	Y	L	L	H
C8-7	G	F	H	G	F	A	D	S
C8-8	S	F	H	R	V	L	A	Q
C8-9	A	I	H	S	F	L	S	M
C8-10	A	S	H	C	I	V	C	L
C8-11	S	T	H	H	L	V	L	L
C8-12	G	I	H	N	F	S	L	Q
C8-13	S	F	H	N	V	A	S	S
C8-14	S	V	H	A	M	T	M	Y
C8-15	S	H	H	A	L	I	S	N
C8-16	V	I	H	G	I	V	F	L
C8-17	S	I	H	S	F	T	S	L
C8-18	S	F	H	G	V	T	A	L
C8-19	G	Y	H	A	F	L	R	S
C8-20	T	S	H	D	L	L	L	A
C8-21	S	L	H	H	L	L	L	T
C8-22	A	L	H	H	L	L	L	T
C8-23	G	M	H	G	I	A	V	S
C8-24	S	Q	H	Q	L	T	L	Q
C8-25	V	F	H	Q	W	S	L	H
C8-26	S	V	H	N	V	M	Q	L
C8-27	S	L	H	R	I	L	N	N
C8-28	V	S	H	N	V	M	Q	L
C8-29	G	I	H	N	Y	V	I	F
C8-30	S	I	H	S	F	T	S	V
C8-31	S	V	H	N	Y	T	M	S
C8-32	A	I	H	N	L	V	S	L
C8-33	G	L	H	G	F	L	L	S
C8-34	A	V	H	D	L	L	V	S
Identity (%)b	24	21	100	24	3	10	29	9
No. of residues accommodatedc	5	11	1	9	7	7	11	10

^aResidues in the mutant strains that are identical to wild type are in boldface, as is the wild-type sequence. Wild-type residues shown to be of importance in the region by earlier site-directed mutagenesis studies are underlined and italicized. 

^bIdentity percentages represent the percentages of mutant strains that retain the wild-type residue at the indicated position. 

^cThe number of residues accommodated is an indication of the variability that can be tolerated at the indicated position. 

Gly187 is proposed to be at the end of the CD lumenal helix (25, 38). In our mutants, it is only 24% conserved, but in our collection of mutants, only five residues (Gly, Ala, Val, Ser, and Thr) were found to function at position 187. This shows a distinct preference for a small, uncharged amino acid at this position, thus supporting the idea that the helix may end at position 187.

Two other residues in the region, Trp191 and Thr192, are not highly conserved in the combinatorial region 8 mutants, but they can be replaced by a limited number of amino acids. The least conserved position in region 8 is Trp191 (conserved in only one of the combinatorial mutants), and this position was found to accommodate either another aromatic residue (Phe or Tyr) or a hydrophobic residue (Val, Ile, Leu, or Met). Thr192 also shows low conservation but stringency in replacement. This position is 10% conserved, and it accommodates only seven different residues being replaced by hydrophobic amino acids (Leu, Ile, Val, Ala, and Met) or by Ser.

DISCUSSION

Combinatorial mutagenesis, which typically uses gain-of-function selection protocols, allows for rapid, efficient identification of structural or functional amino acid residues in proteins of unknown structure (8, 22, 27). In this study, combinatorial mutagenesis was used to probe the CD lumenal loop region of the D2 protein in Synechocystis sp. strain PCC 6803. The CD loop is located on the donor side of PS II, and little detailed structural information is available for this region, as the primary structure similarity to the purple bacterial reaction center is particularly poor in this protein domain. The selection method utilized the concept of functional complementation whereby obligate photoheterotrophic Synechocystis sp. strain PCC 6803 acceptor strains were restored to photoautotrophy by transformation with M13 constructs, resulting in a pool of functional mutants from which structure-function information could be gained.

The CD loop was investigated from three perspectives aimed at addressing these questions: (i) is the length of the CD loop region important to its function, (ii) can a homologous region from D1 function in D2 and vice versa, and (iii) which residues in the CD loop of D2 can be functionally replaced? The first two questions were answered by experiments in which Synechocystis sp. strain PCC 6803 acceptor strains were transformed with deletion and substitution M13 constructs. Since no deletion constructs led to restoration of photoautotrophy, no major (seven- to eight-residue) deletions can be introduced without a loss of function and/or stable assembly of PS II. Results of transformation with substitution constructs indicate that the sequence of the CD loop in D1 is subject to much stricter requirements than that of D2, since no D2 sequence could be functionally accommodated in D1, whereas introduction of two of the four D1 regions into D2 led to functional PS II centers.

Frequency of functionally active sequence combinations.

Each CD loop region studied in this report displayed different degrees of difficulty in isolation of photoautotrophic combinatorial mutants. In region 5, a single transformation experiment yielded all 45 mutants, while the 34 mutants in region 8 were isolated from two transformation experiments. Region 7 was the most difficult region in which to obtain transformants, requiring four separate transformation experiments to yield only 14 photoautotrophic mutants.

Furthermore, the rather strict structural requirements imposed by the selection for stable PS II function were apparent, since all combinatorial mutants in D2 retained at least one, and usually multiple, wild-type residues in the combinatorial region. However, it is clear that we have by no means exhausted the number of possible sequence combinations that can be functionally accommodated, as no sequences that strongly resembled wild type were picked up in our analysis. Therefore, the frequency with which functionally active sequence combinations occur can only be approximated.

One method of approximation is to do an estimation of the number of functionally competent sequences based on the assumption that most combinations of those residues that were found at a particular position in any of the combinatorial mutants yield photoautotrophic mutants. A multiplication of the number of different residues that can be accommodated at each position yields values of 10⁶ for regions 5 and 8. The value for region 7 is an order of magnitude lower (10⁵); a contributing factor is likely to be the smaller number of mutants in this case. Comparing this to the total complexity expected for each domain (2.56 × 10¹⁰ different combinations for the eight-residue domains [regions 7 and 8]; 1.28 × 10⁹ for region 5, which is a seven-residue domain), the probability of a sequence being functionally active is on the order of 10⁻³ for region 5, 10⁻⁴ for region 8, and 10⁻⁵ for region 7. Of course, these numbers assume that all combinations of amino acids that can be accommodated at a position will lead to a photoautotrophic phenotype regardless of the identity of other allowed residues in the combinatorial region. This assumption does not consider the effects of sequence context on the functionality of any one position, a factor that may account for the 100% conservation of His189 in our mutants in light of the existence of photoautotrophic site-directed mutants at this position. Thus, while these probability numbers are likely to be overestimates, they do confirm the observation mentioned previously that region 5 showed the least difficulty in obtaining transformants followed by region 8, with region 7 being the least flexible of the CD loop regions tested.

Another approximation approach is to compare the number of unique photoautotrophic transformants obtained with the number of different M13s with which Synechocystis sp. strain PCC 6803 was transformed. The facts that (i) M13 was amplified about 10⁵-fold between E. coli transformation and phage harvesting-DNA isolation 4 h later and (ii) the frequency of Synechocystis sp. strain PCC 6803 transformation was such that about 1 out of 10⁴ to 10⁵ M13 DNA molecules led to a transformant under control conditions imply that in principle all M13 sequences that initially were present in transformed E. coli will be represented in the collection of Synechocystis sp. strain PCC 6803 transformants. Thus, the number of unique photoautotrophic transformants per experiment, divided by the number of different combinatorial M13 phages used for the transformation of the Synechocystis sp. strain PCC 6803 acceptor strain approximates the probability of an M13 clone leading to restoration of photoautotrophic growth. As the number of different combinatorial phages per experiment was estimated to be 4 × 10⁵ (based on 2 × 10⁶ transformants per experiment without M13 amplification and a 20% probability that the combinatorial strand rather than the deletion strand is amplified), this yields a frequency of 10⁻⁴ for region 5, 10⁻⁵ for region 8, and 10⁻⁶ for region 7. These numbers are comparable within an order of magnitude with the frequencies calculated using the first approach, which may have yielded an overestimate of the probabilities.

The N-terminal part of the CD loop.

In the D2 protein, the CD loop region immediately adjacent to helix C has never been assigned a critical role in PS II function. A surprising observation in this study was that nonetheless three residues in region 5 (Gln164 to Ala170), Ser166, Phe169, and Ala170, remained highly conserved in the combinatorial mutants. This observation may support a previous suggestion that the backbone of Phe169 may hydrogen bond with Y_D (38). The high conservation and the limited number of residues that can be functionally accommodated at these three positions imply that this region may interact with other proteins not considered by the models and/or that PS II and the bacterial reaction center are quite different in this region. The latter would not be surprising since there is poor primary sequence conservation between the two; moreover, the bacterial reaction center lacks an OEC. While this region of the loop may have fewer structural constraints than the other regions, as evidenced by the relative ease by which photoautotrophic mutants were isolated, nonetheless, the strong preference for specific residues in part of region 5 seems to imply a stringent structural requirement around these residues. Interestingly, the apparently important residues in the region at positions 166, 169, and 170 are spaced as if they were in a α-helical arrangement (i.e., one turn of the helix apart). This spacing may be coincidental or may indicate that the CD helix in the D2 protein of PS II is much longer than that in purple bacteria.

The region around residue 180 in D2.

The most noteworthy feature of region 7 is the conservation pattern of Arg180. In the mutants that lack Arg at position 180, an Arg residue always appears at position 184. Since this region of the CD loop has been proposed to be at the center of the lumenal helix, the shifting of Arg from 180 to 184 corresponds to one turn of the helix downstream. This suggests that the functional side groups of Arg retain the same relative position in the protein at either position 180 or 184. If an Arg residue is present at position 184, the residue at position 180 (Gly, Val, or Thr) generally is rather small; on the other hand, if Arg180 is present, at position 184 a large hydrophobic residue (Phe) is strongly preferred. It is likely that these residues are involved in proper positioning of the Arg residue.

Arg180 has been studied by site-directed mutagenesis, and of the mutants that were analyzed, only R180Q can sustain photoautotrophic growth, albeit at marginal rates (16). Arg180 mutants show altered Y_D^ox EPR signals as well as altered kinetics of charge recombination between Q_A⁻ and the donor side (16). The suggestion that Arg180 may function as a critical bifunctional residue interacting with Y_D and providing a ligand to chlorophyll (16) and/or modifying P680 redox characteristics (16) is supported by the pattern observed in our combinatorial mutants.

Of the three regions studied here, the smallest number of photoautotrophic transformants were isolated in region 7, and for those isolated, there appeared to be less flexibility in the number of residues that could be accommodated at each position. In the set of combinatorial mutants encompassing this region, not more than seven different amino acid residues were found at any given position, whereas regions 5 and 8 contain several positions that can accommodate 10 or more different residues. In addition, this region shows a stronger global preference for aromatic residues and large hydrophobic residues, probably reflecting the strict requirements imposed by the proposed α-helical structure of the region.

The α-helical structure of region 7 may also explain why tryptophan occurs in this set of mutants at a higher frequency than that in the other CD loop mutant sets. Trp is incorporated into region 7 mutants in four instances even though Trp is absent from the wild-type sequence of the region and the frequency of Trp incorporation is expected to be lower than most other residues since it is encoded by only one codon. While the region 5 set of mutants also has four tryptophan residues, three are conserved at the wild-type position, leaving only one tryptophan accommodated outside of this position in 45 mutants. In region 8, only one mutant contains a Trp, and it is at the wild-type position. The Trp content has been found to be higher on average in transmembrane proteins and preferentially located near the ends of α-helices (reviewed in references 1 and 24). Trp has the capability of forming hydrogen bonds, making this residue suited for stabilization of helices at boundaries between lipid and polar environments (reviewed in references 15 and 24). These properties may account for the increased presence of Trp in this set of mutants because region 7 makes up a large part of the proposed CD loop helix and the helix is believed to interact with the membrane (25).

The C-terminal domain of the CD loop.

The absolute conservation of His189 in the region 8 combinatorial mutants was unexpected, since site-directed mutations at this position can be accommodated. Many of the combinatorial mutants grow slowly and appeared on plates weeks after transformation of the D2ΔCD strain (the wild type comes up within a week), arguing against a bias toward more photoautotrophically competent strains as an explanation for the conservation of His189. Instead, the finding implies that the presence of an altered environment around position 189 (as occurs in the combinatorial mutations) may decrease the ability of the region as a whole to accommodate a proton originating from Y_D, thereby increasing the stringency of the requirement of His at position 189.

Close examination of the photoautotrophic single-site mutants in His189, H189L, H189Y, and H189Q hints at the importance of the relationship between hydrogen bonding of Y_D and His189 and efficient operation of the OEC. Of the three single mutants, H189Y shows the least impairment (26, 28), which may fit with the fact that the replacing tyrosine has hydrogen-bonding capabilities (reviewed in reference 24). If the neighboring wild-type residues are able to support limited photoautotrophic growth in the absence of His189, then the potential candidates should have the ability to partially substitute for His189 in its hydrogen binding-acceptor role with Y_D. One potential candidate for this role from the wild-type sequence in region 8 is Trp191. Other than Trp191 serving as a potential hydrogen bond partner, a recent model of PS II proposed that Trp191 might play a structural role in the region by providing ring-stacking forces to the chlorophyll special pair P680 as well (38).

The results of this study demonstrate that combinatorial mutagenesis is an efficient method for obtaining a substantial pool of different functional mutants with amino acid changes in a large domain of a protein with unknown structure. Additionally, sequence comparisons of sets of combinatorial mutants can provide insight into otherwise unremarkable primary sequence elements that may have structural and/or functional roles in the protein, thus opening new avenues for research. One example of this is the analysis of several combinatorial mutants in region 7, in which introduction of a Trp residue at position 181 of the D2 protein led to a large quenching of variable fluorescence (30).