Photosynthesis Research

, Volume 93, Issue 1, pp 27–43

The presence of multidomain linkers determines the bundle-shape structure of the phycobilisome of the cyanobacterium Gloeobacter violaceus PCC 7421

Authors

  • David W. Krogmann
    • Department of BiochemistryPurdue University
  • Bertha Pérez-Gómez
    • Departmento de BioquímicaCentro de Investigación y Estudios Avanzados-IPN
  • Emma Berta Gutiérrez-Cirlos
    • Unidad de Biomedicina FES-Iztacala UNAM
  • Alicia Chagolla-López
    • Cinvestav-Unidad Irapuato
  • Luis González de la Vara
    • Cinvestav-Unidad Irapuato
    • Departmento de BioquímicaCentro de Investigación y Estudios Avanzados-IPN
Original paper

DOI: 10.1007/s11120-007-9133-9

Cite this article as:
Krogmann, D.W., Pérez-Gómez, B., Gutiérrez-Cirlos, E.B. et al. Photosynth Res (2007) 93: 27. doi:10.1007/s11120-007-9133-9

Abstract

The complete genome sequence of Gloeobacter violaceus [Nakamura et al. (2003a, b) DNA Res 10:37–45, 181–201] allows us to understand better the structure of the phycobilisomes (PBS) of this cyanobacterium. Genomic analysis revealed peculiarities in these PBS: the presence of genes for two multidomain linker proteins, a core membrane linker with four repetitive sequences (REP domains), the absence of rod core linkers, two sets of phycocyanin (PC) α and β subunits, two copies of a rod PC associated linker (CpcC), and two rod cap associated linkers (CpcD). Also, there is one ferredoxin–NADP+ oxidoreductase with only two domains. The PBS proteins were investigated by gel electrophoresis, amino acid sequencing and peptide mass fingerprinting (PMF). The two unique multidomain linkers contain three REP domains with high similarity and these were found to be in tandem and were separated by dissimilar Arms. One of these, with a mass of 81 kDa, is found in heavy PBS fragments rich in PC. We propose that it links six PC hexamers in two parallel rows in the rods. The other unique linker has a mass of 91 kDa and is easily released from the heavy fragments of PBS. We propose that this links the rods to the core. The presence of these multidomain linkers could explain the bundle shaped rods of the PBS. The presence of 4 REP domains in the core membrane linker protein (129 kDa) was established by PMF. This core linker may hold together 16 AP trimers of the pentacylindrical core, or alternatively, a tetracylindrical core of the PBS of G. violaceus.

Keywords

CyanobacteriumGloeobacter violaceusPhycobilisomeMultidomain linkers

Abbreviations

AP

Allophycocyanin

apcAB

Gene encoding the α and β subunits of allophycocyanin

apcC

Gene encoding the 8–kDa core linker polypeptide

ApcE

The 129-kDa core membrane linker phycobiliprotein

A.U.

Absorbance units

cpcBA

Genes encoding the α and β subunits of phycocyanin

cpcC1

Gene encoding the PC-rod linker polypeptide

cpcD

Gene encoding the 9-kDa rod linker polypeptide

cpcG

Gene encoding the rod-core linker polypeptide

DCPIP

Dichlorophenol-indophenol

DMF

Dimethyl-formamide

EDTA

Ethylenediamine tetraacetic acid

FNR

Ferredoxin-NADP+ oxidoreductase

FNR-2D

FNR containing only the two catalytic domains

FNR-3D

FNR containing three domains (the CpcD-like domain and two catalytic domains)

F.U.

Fluorescence units

LC8

Core linker polypeptide of 8 kDa

LCM129

LR9, the 9-kDa, phycocyanin-associated rod linker polypeptide

LR33

CpcC, the PC-rod linker polypeptide

CpcG

The rod core linker polypeptide

PAGE

Polyacrylamide gel electrophoresis

PBP(s)

Phycobiliprotein(s)

PBS

Phycobilisome(s)

PC

Phycocyanin

PE

Phycoerythrin

petH

Gene encoding FNR

PMF

Protein mass fingerprinting

PMSF

Phenyl-methylsulfonyl fluoride

REP

Repetitive sequence

SDS

Sodium dodecylsulfate

WCE

Whole-cell extract

Introduction

Phycobilisomes (PBSs) are the major light-harvesting systems in cyanobacteria and red algae. This type of photosynthetic antenna consists of a macro complex composed of light-absorbing proteins named phycobiliproteins (PBPs). The PBPs are composed of α and β subunits and are a brilliantly colored group of disc-shaped proteins bearing covalently attached phycobilins. In addition, linker proteins hold together the PBPs. Both PBPs and linker proteins are organized as a core connected with photosystems and surrounded by several peripheral rods (Sidler 1994; MacColl 1998; Tandeu de Marsac 2003; Grossman 2003; Adir 2005; Liu et al. 2005). G. violaceus is a unicellular cyanobacterium, which lacks thylakoids. It has a lower relative ratio of chlorophyll a with phycobiliproteins than is found in most typical cyanobacteria (Rippka et al. 1974). Underlying the membrane is an electron-dense layer 80-nm thick, which seems to be associated with the PBSs. G. violaceus PBSs differ in their structural organization from those of other cyanobacteria and red algae (Guglielmi et al. 1981): the core is described as amorphous and the rods containing phycocyanin and phycoerythrin are grouped as a bundle. These bundles are organized as two rows of three parallel rod-shaped elements. The rods are 10–12 nm wide and their length varies from 50 to 70 nm. These rods appear to be formed by the coaxial stacking of 7–10 discs, 6-nm thick and 12 nm in diameter (Guglielmi et al. 1981). In addition, G. violaceus PBSs are considerably longer than the peripheral rods of hemidiscoidal PBS. Therefore, the PBSs of G. violaceus have a complex composition since they also possess three types of constitutive PBPs: allophycocyanin (AP), phycocyanin (PC) and phycoerythrin (PE) (Bryant et al. 1981). G. violaceus does not adapt chromatically, and therefore the proportions of the three phycobiliproteins remain unchanged after growing in white, red or green light (Rippka et al. 1974). The genes encoding the phycobiliprotein subunits and the linker proteins of the PBS of G. violaceus have been recently identified in the complete genome of this organism (Nakamura et al. 2003a). The genome contains at least 33 open reading frames (ORFs) encoding polypeptides with similarity to proteins involved in PBS assembly, phycobilin biosynthesis or ligation, and PBS regulation (Nakamura et al. 2003b). The genes may be grouped in five main clusters as seen in Fig. 1. Several unusual features are found in the PBS-related genes. The AP gene family contains a cluster apcEABC. In addition, two genes are found isolated in the genome, a second gene for apcA and an apcD gene (encoding αAPB). The gene encoding βAP-18 is absent from the genome. The apcE gene, encoding the PBS core-membrane linker polypetide (LCM), contains four repetitive sequences (REP domains, Capuano et al. 1991) similar to those found in Anabaena sp. PCC 7120 (Kaneko et al. 2001), Mastigocladus laminosus (Sidler 1994) and Thermosynechococcus (Nakamura et al. 2002). A single gene cluster, designated cpeBA encodes both PE subunits. There are three genes encoding proteins with similarity to the PE linkers: cpeC, cpeD and cpeE making a cluster cpeCDE that is separated from cpeBA as in Fremyella diplosiphon (Cobley et al. 2002). Two distinct gene sets encode PC subunits; these have been designated cpcB1A1 and cpcB2A2. Furthermore, two genes encoding the linker polypeptide associated with PC have been designated cpcC1 and cpcC2 (Gutiérrez-Cirlos et al. 2006). These two genes are also isolated in the genome. Two genes encoding the rod-terminating linkers have been designated cpcD1 and cpcD2, and are located in the same cluster as the PE associated linkers. The cpcG genes, encoding for the core rod linker proteins (Bryant et al. 1991) are absent from the genome (Nakamura et al. 2003a). Finally, the petH gene, which encodes ferredoxin–NADP oxidoreductase (FNR) predicts a protein of only two domains (296 amino acids), and lacks a CpcD-like domain at its N-terminus. FNR is not expected to serve as a component of the PBS (Schluchter and Bryant 1992; van Thor et al. 1999; Gómez-Lojero et al. 2003).
https://static-content.springer.com/image/art%3A10.1007%2Fs11120-007-9133-9/MediaObjects/11120_2007_9133_Fig1_HTML.gif
Fig. 1

Complete map of phycobilisome genes from G. violaceus PCC 7421. The numbers below the bars (1–6) give the positions of the five clusters in the genome. Number 1 is located at nt 1,000,000, number 2 is at nt 1,263,000, number 3 is at nt 1,326,000, number 4 is at nt 1,344,000, number 5 is at nt 2,985,000 and number 6 is at nt 3,426,000. The cpcB and cpcA genes are frequently followed by genes encoding PC-linker polypeptides in other cyanobacteria (Tandeau de Marsac et al. 1988). The genes encoding PC-linker polypeptides (dark blue) are clearly scattered in different regions. Also, the novel genes for multidomain linkers of this cyanobacterium (glr1262 and glr2806) are distant from one another

The linker proteins determine the positions of the phycobiliproteins within the phycobilisome structure (Bryant et al. 1991; Glauser et al. 1992). Also, the linker proteins interact either directly or indirectly with the chromophores causing changes in their environment. These changes can modulate the spectral properties of different phycobilisome subassemblies (Gottschalk et al. 1991, 1993; Reuter et al. 1999). The rod linker proteins of G. violaceus, CpcC1, CpcC2, CpeC, CpeD and CpeE, have been identified by N-terminal sequencing (Gutiérrez-Cirlos et al. 2006). Two of them are rod phycocyanin associated linker proteins and three are rod phycoerythrin associated linker proteins. The genome sequence of G. violaceus predicts the presence of unique components in the PBS of this cyanobacterium. In this work, we have been able to confirm the presence of linkers in intact and dissociated PBSs from G. violaceus. Using a biochemical approach, we characterized some of the linker properties, which allow us to suggest their localization within the PBS.

Materials and methods

The sequences of the genome of G. violaceus were retrieved from GenBank, accession number BA000045. The online programs for Blastp analyses (Altschul et al. 1997) were obtained using NCBI and ExPASy Web sites. Alignments of linker polypeptides were made using the ClustalX (Thompson et al. 1997). The percentages of identity and similarity between two sequences were obtained after Blastp analyses. Sequences of PBS proteins from other cyanobacteria were retrieved from Entrez (GenBank NCBI).

Membrane extraction and centrifugation in sucrose gradients

G. violaceus (Rippka et al. 1979) was supplied by the Pasteur culture collection. Cultures were grown in 50 ml of BG 11 medium supplemented with 5 mM Mops (pH 7.8). A light intensity of 5 μmol m−2 s−1 was employed. The cultures were grown for 18–25 days. The cells were washed once with 0.65 M Na+/K+ phosphate buffer (pH 8.0) and suspended in the same buffer. The cells were disrupted once or twice in a French pressure cell (Thermo Spectronic) operated at 20,000 psi. The broken cells were centrifuged for 10 min at 27,000 × g. The supernatant of whole cell extract (WCE) was treated with Triton X-100 (1–1.2% v/v final concentration) for 30 min and centrifuged again for 30 min at 27,000 × g. Part of the supernatant was carefully layered on the sucrose gradient. In an alternative experiment, 240 mg per ml ammonium sulfate was added to the colored supernatant and the mixture was centrifuged at 27,000 × g for 30 min. The precipitate was suspended with 0.65 M Na+/K+ phosphate buffer (pH 8.0). About 2 ml samples were layered onto sucrose gradient tubes. Alternatively, when indicated, the broken cells were centrifuged at 150,000 × g for 1 h to sediment the membranes. The membranes were suspended in the same buffer as above, solubilized with Triton X-100 (1%) and layered onto the sucrose gradient. The sucrose step gradients consisted of 4, 5, 4, and 3 ml of 0.5, 0.75, 1.0 and 2.0 M. sucrose solutions in 0.75 M Na+/K+ phosphate buffer pH 8.0. The gradient was centrifuged in a Beckman Ti 60 rotor at 65,000 × g for 12 h at 20°C or at 225,000 × g for 5 h at 20°C. The gradients were collected from bottom to top in 1 ml aliquots. The sucrose concentration of each fraction was determined with an Abbe refractometer. Alternatively, the color fractions were pooled and each band pool was precipitated with TCA (10% final concentration) for subsequent SDS—or lithium dodecyl sulfate (LDS)—PAGE analysis. Absorption spectra of all fractions were obtained by scanning from 340 nm to 700 nm and the absorbances at 502, 620 and 656 nm were used to calculate the amounts of PE, PC and AP. Fluorescence spectra were recorded with a Hitachi F4500 fluorescence spectrophotometer.

SDS and LiDS-PAGE

SDS-PAGE analyses were performed on 10% or 14% polyacrylamide slab gels (Schägger and von Jagow 1987). LiDS-PAGE was performed on 10% polyacrylamide slab gels (Delepelaire and Chua 1979). Electrophoresis was run at 4°C and at constant voltage for 12 h. For fluorescence visualization under UV light, gels were pre-incubated for 15 min in 20 mM zinc acetate (Raps 1990). The proteins were identified by PMF. The bands corresponding to the proteins stained with Coomassie blue were excised from the gel and sent to the Protein Identification Unit of Cinvestav at Irapuato, México. After in-gel trypsin digestion (using sequence grade trypsin from PROMEGA), proteins were analyzed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI ToF MS). The experimental mass values were compared with those derived from available databases and each subunit was then identified in the same instrument (Ettan MALDI-ToF) or by using the MASCOT program.

Densitometric scans were performed with a Pharmacia LKB UltroScan XL and analyzed with the program GSXL version 2.1. The relative molar amounts of protein per band were obtained as described (Gómez-Lojero et al. 1997). The scaling factors used were 1.29 for LCM120; 0.91 for LRC91 (Glr1262) 0.81 for LR81; 0.33 for LR33, 0.30 for LR30, 0.35 for LR35.

N-terminal sequence analysis

The proteins were prepared for N-terminal sequence analyses as follows. The SDS-PAGE was run with 150 μg of protein in each well of a 1.5-mm thick slab of 10%-polyacrylamide gel for high molecular weight proteins and 14% polyacrylamide gel for low molecular weight proteins. The proteins were then transferred to a Trans Blot (Bio Rad PVDF) membrane using the Trans Blot SD (Bio Rad) semi-dry system. The PVDF membrane was stained with Coomassie brilliant blue R-250 and the bands were cut and sent to the Protein Analysis Laboratory at Purdue University for N-terminal sequencing.

Molecular mass estimation of the PBS-substructures

The estimation of molecular mass was performed on sucrose density gradients by an established method (Martin and Ames 1961). Linear gradients, (16 ml), of 0.5–2 M sucrose in 0.75 M Na+/ K+ phosphate, pH 8.0 were prepared and 2 ml of protein solution was layered on each gradient. Centrifugation was performed in a Beckman 60 Ti rotor at 225,000 × g for 5 h at 20°C. About 1 ml fractions were collected with a peristaltic pump from the bottom to the top of the tube. Phycobiliprotein distribution in the gradient was determined by the absorbance at the peak of each protein. The cells were disrupted only once in the French press, treated with Triton X-100 (1%) for 30 min and then centrifuged for 20 min at 27,000 × g. The supernatant was precipitated with ammonium sulfate (40% saturation). The suspended precipitate (2 ml) was layered on the sucrose gradient.

Absorption and fluorescence spectroscopy

Absorption spectra were recorded with an Aminco DW2 UV–visible spectrophotometer with the OLIS DW2 conversion and OLIS software or with a Hitachi U3310 spectrophotometer. The concentrations of phycobiliproteins were estimated using the extinction coefficients for the protomers of G. violaceus εPE502 245, εPE566 456, εPC566 141, εPC620 365, εPC654 51.5 (Bryant et al 1981), εAP566 42, εAP620 172, εAP654 313 (Füglistaller et al. 1987; Ducret et al. 1998), and are expressed in mM−1 cm−1. Pigment concentrations were calculated with the following set of simultaneous equations:

$$ \left[ {{\hbox{AP}}} \right] = \left( {\hbox{A}} \right._{654} - 0.141\left. {{\hbox{A}}_{620} } \right) \div 289;\,\,\,\,[{\hbox{PC}}] = ({\hbox{A}}_{{\hbox{620}}} - 0.55{\hbox{A}}_{{\hbox{654}}} ) \div 337;\,\,\,[{\hbox{PE}}] = {\hbox{A}}_{{\hbox{502}}} \div 245 $$

Fluorescence spectra were recorded with a Hitachi F−4500 fluorescence spectrophotometer.

Results

Identification of PBS subunits

The distribution of the G. violaceus PBPs after sucrose gradient centrifugation was characterized spectroscopically. Three main fractions of PBPs (3, 6, and 11) absorbing at 566 nm were identified (Fig. 2A). The most rapidly sedimenting fraction contained the PBSs. More than 55% of the pigment was found as PBS (fractions 2–4). Figure 2B shows the spectrum of fraction 3 (PBS), from which 1 to 2.15 to 1.7 molar ratios of AP absorbance at 654 nm, PC absorbance at 620 nm and PE absorbance at 502 nm were calculated. The ratios of the PBPs are close to those found in the whole cell extract (1 : 2.2 : 1.66). The middle peak contained a substructure enriched in PE and PC. The spectrum of fraction 6 is shown in Fig. 2C, and exhibits ratios of AP to PC to PE of 1 to 2.3 to 2.6. The slowly sedimenting peak absorbing at 566 nm is congruent with the peak absorbing at 620 nm, with a ratio of PC to PE of 4 and was obtained from the spectrum of fraction 11 of Fig. 2C. The sucrose gradient shown in Fig. 2A was linear from 15% to 42%, and thus allowed us to estimate the relative sizes of some of the resulting substructures. The ratio of the distances traveled from the top to fraction 11 and to fraction 13 is 1.54, which corresponds to a molecular weight ratio of 1.9, R = (MW1/MW2)2/3 (Martin and Ames 1961). We assigned the molecular mass of 226,794 Da to fraction 11 (calculated for PE and PC hexamers) and thus a calculated mass of 119 kDa was obtained for fraction 13, which contains trimers of AP. The spectrum of fraction 5 in Fig. 2C revealed two main substructures: one rich in AP and the other in PC. A mass of 1,500,000 Da was assigned to fraction 5, whose molar ratios of biliproteins, were 1 : 1.15 : 1 for AP to PC to PE. The estimated molecular mass for an intact pentacylindrical core is 2,000,000 Da (Ducret et al. 1998), and for six PC hexamers is 1,451,000 Da. We concluded that fraction 5 contained a partially dissociated pentacylindrical core and a substructure formed by six PC hexamers.
https://static-content.springer.com/image/art%3A10.1007%2Fs11120-007-9133-9/MediaObjects/11120_2007_9133_Fig2_HTML.gif
Fig. 2

Characterization of the PBS and PBS substructures. The WCE was obtained after one pass through the French press, solubilized by Triton X-100 and precipitated by ammonium sulfate. The sucrose gradient (described in Materials and methods) was loaded with the suspended ammonium sulfate precipitate and centrifuged for 5 h at 225,000 x g. Panel A shows the pattern of fractions (bottom to top) obtained from the sucrose gradient. Tubes containing 1 ml fractions were collected. The absorbance of each fraction at 566 nm for PE (fuchsia squares), 620 nm for PC (blue diamonds) and 654 nm for AP (black circles) are indicated. The fractions were assayed for sucrose concentration (triangles) to estimate the molecular weights of the PBS substructures. Panel B: Absorption spectrum of the PBS (fraction 3 of Panel A). Panel C: Absorption spectra of fractions 5, 6, 11, 13, and 14 after the sucrose gradient of Panel A. Panel D: The sucrose gradient showing the bulk colored fractions pooled for SDS-PAGE analyses

In order to characterize the proteins of the G. violaceus PBS and of its corresponding substructures, the colored bands from the gradients were pooled from the ultracentrifuge tubes and subjected to SDS-PAGE analyses. Figure 3 shows two electrophoretograms of the colored bands obtained from the sucrose gradient shown in Fig. 2 Panel D. The electrophoretogram on the left shows a 10% acrylamide gel, used to identify the high molecular weight components, while on the right electrophoretogram there is a 14% acrylamide gel to identify the low molecular weight components. Lanes I to III in both gels display the proteins in the colored bands obtained from the sucrose gradient shown in Fig. 2D. The apparent molecular masses of the PBS polypeptides determined by SDS-PAGE, the molecular masses predicted from amino acid sequences, and the method used for identifying each protein are presented in Table 1. Fraction I (left electrophoretogram on Fig. 3) contained three linker proteins of 129, 91 and 81 kDa, which correspond to ApcE (LCM129) and to the protein products coded by glr1262 and glr2806 genes, respectively. Fraction II exhibited a higher proportion of the 129 and 81 kDa polypeptides and less of the 91 kDa polypeptide as compared with Fraction I (the quantitative aspects of these electrophoretograms are presented in Table 2). The polypeptides from Gll0415 and Glr0638 are conspicuous in lane I, vary from sample to sample and were identified as porins that are the main proteins of the external membrane. The groups of polypeptides of 35, 33 and 30 kDa have been previously identified as CpeC, CpcC1, CpcC2, CpeD and CpeE (Gutiérrez-Cirlos et al. 2006). Lane III of this gel was used to correlate the quantities of the middle size linkers with their associated PBPs. The 14% acrylamide gel (at the right of the Figure) was used to identify the low molecular weight components. In lane I, a faint band of 8 kDa could be seen at the bottom of the gel. In lane II, two faint bands of 12 and 11 kDa were detected. The intense band of 8 kDa corresponds to ApcC as shown by PMF (Table 1). Fraction III contained more intense subunits of 12 and 11 kDa, which were identified as CpcD1 and CpcD2 by N-terminal sequence analysis (Table 1). Together the results of both 10 and 14% gels led us to conclude that the PBS of G. violaceus contains 11 linker proteins (Table 1). Changes in concentration of other proteins are also shown in Table 2 and are correlated with the percentage of PBPs obtained from the spectra of the bulk fractions shown in Fig. 2D. Data in Table 2 allow us to establish a correlation between the linkers in the different fractions of the sucrose gradient and the amount of each of the PBPs. The high molecular weight linkers LCM129, LRC91 and LR81 are present in the heavy fractions. The middle-size linkers LR35, LR33 and LR30 correlate with PC and PE. As expected LR30 correlates with PE and LR33 with PC. In contrast, LR35 correlates in a more complex manner with two components, one that associates PC with PE, and another that interacts with two PC hexamers. The LC8 correlates with AP. The LR11 and LR12, which are the terminating linkers, are detected in the fraction rich in PE that is the distal part of the rods (Lanes II and III in Fig. 3 and Table 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs11120-007-9133-9/MediaObjects/11120_2007_9133_Fig3_HTML.gif
Fig. 3

LiDS-PAGE analyses of the pooled bands (I–III) of the sucrose gradient shown in Fig. 2, Panel D. Bands IV–VI were also run but the electrophoretograms are not shown. The left Panel shows a 10% acrylamide gel. The electrophoresis was stopped when the PBPs arrived at the bottom edge of the gel for the proper separation of the high molecular weight components. The right Panel shows a 14% acrylamide gel for detection of the low molecular weight components. This electrophoresis was stopped when the bromophenol blue arrived at the bottom edge of the gel. Each of these gels was prepared with duplicates of each of the pooled fractions. One half of the gel was blotted onto a PVDF membrane to yield bands for N-terminal sequencing. The other half was used for Coomassie blue staining and densitometric scanning, to obtain molecular masses, the relative amounts of the proteins and, finally, for PMF. The proteins identified by N-terminal sequencing and by PMF are shown at the left of each electrophoretogram. The gels were scanned and the estimated molecular masses (molecular weight markers were included in each gel, but not shown) are reported in Table 1, the calculated relative amounts of proteins are presented in Table 2

Table 1

Phycobilisome linker proteins and co-migrating porins of G. violaceus as obtained from genomic analysis and classified according to OLN (order locus name)

ORF No

Gene

Predicted Molecular mass (Da)

SDS-PAGE M.w. (kDa)

N-terminal sequence

PMF No

PMF coverage

Proposed function

gll415

 

66,045

63

 

12

24%

PORIN

glr638

 

62,027

59

 

12

31%

PORIN

glr0950

cpcC1

31,043

35

Gutierrez-Cirlos et al. 2006

  

LR31

glr1245

apcE

129,836

120*

 

24

24%

LCM129

glr1248

apcC

7,754

8

 

3

52%

LC7.7

glr1262

 

91,882

87

VLDDQSKAY

  

LRC91

glr1263

cpeC

31,825

35

Gutierrez-Cirlos et al. 2006

  

LR31.8

glr1264

cpeD

28,400

30

Gutierrez-Cirlos et al. 2006

  

LR28.4

glr1265

cpeE

28,354

30

Gutierrez-Cirlos et al. 2006

  

LR28.3

glr1266

cpcD1

8,169

12

SGXVXTGVAX

  

LR8.1

glr1267

cpcD2

7,769

11

MYXNVTAXXX

  

LR7.7

glr2806

 

81,442

81

SATTYDWRKVI

  

LR81

gll3219

cpcC2

30,878

33

Gutierrez-Cirlos et al. 2006

  

LR30.8

The characterizing abbreviations are given at the top of each column. PMF coverage indicates the percentage of the amino acid residues identified in fragments by PMF/total amino acids residues. Letter X in the N-terminal sequences of LR8.1 (Glr1266) and LR7.7 (Glr1267) indicates that the predominant amino acid found is not the one expected from the genome sequence. However, the expected amino acid was present in a lower concentration. Despite the ambiguity in these positions, the remainder of the sequences provided the unambiguous identification of CpcD1 and CpcD2 in the G. violaceus genome

Table 2

Relative amounts of the components of PBS and PBS substructures of G. violaceus isolated from the sucrose density gradients

Subunit

I (10)

I (14)

II (10)

II (14)

III (10)

III (14)

LCM129

4

4

6

7

2

 

LRC91

4

3

2

 

0

 

LR81

5

4

9

7

5

 

LR35

16

15

14

12

14

16

LR33

8.2

8

5

4

3

4

LR30

25

21

25

26

30

34

PBPs

100

100

100

100

100

100

AP

21

21

32

32

18

18

PC

42

42

38

38

40

40

PE

37

37

30

30

42

42

LR12

   

6

 

9

LR11

   

5

 

7

LC8

 

3

 

40

 

10

In order to obtain quantitative values of intensities of the bands from the proteins stained with Coomassie blue, the gels shown in Fig. 3 were scanned by laser densitometry. The area under the curve (A.U. × mm) obtained from the densitometric scan was multiplied by the factor related to its molecular weight and described in Materials and methods. The above result for the PBPs was divided by 12 (the number of subunits per PBP hexamer). By using this number all the values of the linkers were normalized, taking the value for the PBPs as 100. The percentage of each PBP (AP, PC and PE) was obtained by calculating the concentration from absorbance values of the spectra of each of the fractions employed (I–III) from the sucrose gradient. Each of the linker components is listed with its proper linker name and its molecular weight in the superscript. The PBP components are listed as well. Numbers in parentheses indicate the percentage of the acrylamide gel used for the determination

Localization of the linker proteins

G. violaceus membranes, free from soluble dissociated PBS substructures, were solubilized with Triton X-100 and applied to a sucrose gradient. A different pattern of PBPs was obtained in this linear gradient. In particular, the blue band between the fuchsia and the violet bands was clearly observed. We characterized both the violet and the blue bands. Figure 4A shows the absorption and the fluorescence excitation spectra of the violet band. The absorption spectrum shows PBS with almost the same absorption by phycocyanin and phycoerythrin. The minor peak at 502 nm corresponds to the phycourobilin, one of the chromophores of PE. The near equality of the maxima in the spectrum indicates the loss of some PE during the first stage of the preparation. The fluorescence excitation spectrum reveals energy transfer from PE and PC to AP since the emission wavelength used is 687 nm, and is indicative of energy transfer to AP, which is the final energy acceptor in the PBS. Inserted in Panel A is a LiDS-PAGE that shows the components of this PBS. Also, the ratios of high molecular weight components of the PBS are inserted beneath the spectra. Figure 4B contains information about the blue band above the PBS. The absorption spectrum shows that PC and AP are the main components of this fraction. The fluorescence excitation spectrum shows less energy transfer with a maximum at 665 nm corresponding to the excitation of the last AP component of the PBS. From the inserted LiDS-PAGE in Panel B, we obtained data about the ratios of the high molecular weight linkers, shown beneath the spectra. The concentration of LRC91 was particularly low in this fraction or null in other experiments. We concluded that this fraction is formed by the core of the PBS and probably by six PC hexamers maintained together as a substructure by LRC81. The above interpretation is in agreement with the molecular weights estimated from Fig. 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs11120-007-9133-9/MediaObjects/11120_2007_9133_Fig4_HTML.gif
Fig. 4

PBS and PBS dissociation products separated by sucrose gradient centrifugation from Triton X-100 solubilized membranes from G. violaceus. Panel A: Absorption (solid line) and fluorescence excitation (dashed line) spectra of the PBS. Panel B: Absorption (solid line) and fluorescence excitation (dashed line) spectra of the blue band equivalent to band II of Fig. 2D. The relative fluorescence units (F.U.) were normalized with the highest value of absorbance. Inserted in panels A and B are the electrophoretograms (10% acrylamide) of the corresponding fractions. Underneath the spectra are the identified high molecular weight linkers. Superscripts indicate the molecular masses predicted by amino acid sequences and confirmed by comparison with the mass of the LCM99 of Synechococcus sp. PCC 7002 (see Fig. 5). The numbers after the equal symbols (=) are the relative stoichiometries of the multidomain linkers obtained from the densitometric analyses of the gels

Figure 5A shows gels of the high molecular mass linker polypeptides of G. violaceus from the violet (I), blue (II) and fuchsia (III) bands obtained from the sucrose gradient like those seen in Fig. 2D and compares them with LCM99 of Synechococcus sp. PCC 7002 (marked as 7002). Panel B shows the same region after treatment with zinc acetate and UV transillumination. Using the known molecular mass of Synechococcus LCM99, as a marker, an apparent molecular mass of 120 kDa for the G. violaceus ApcE was assigned. Figure 5C shows the mass spectrum of a trypsin PMF of G. violaceus ApcE protein. The coverage of the spectrum was 24% with 24 peptides matched for ApcE. Fortunately, the trypsin peptide at the C-terminus was present in the matched peptides (Panel D), indicating the presence of the complete amino acid sequence of ApcE predicted in the genome and thus justifying the LCM129 nomenclature. Finding this peptide in the mass spectrum eliminates the possibility of any post-transcriptional modifications in the C-terminal region of the molecule.
https://static-content.springer.com/image/art%3A10.1007%2Fs11120-007-9133-9/MediaObjects/11120_2007_9133_Fig5_HTML.gif
Fig. 5

Panel A: Coomassie staining of the LiDS-PAGE (10% acrylamide) of the colored bands from the gradient analyzed in Fig. 4 and compared to the PBS of Synechococcus sp. PCC 7002. Panel B: The same gel under UV light after spraying with zinc acetate. Lanes I and I’ have 75 and 125 μg, respectively of the band I (violet); lane II, 100 μg of the band II (blue) and lanes III and III’ 75 and 125 μg of protein, respectively of the band III (fuchsia). The lanes marked 7002 are PBS of Synechococcus sp. PCC 7002. The identity of ApcE was determined by PMF. Panel C: Peptide mass spectrum of ApcE from G. violaceus. After the protein was excised from the gel and digested with trypsin, it was analyzed by MALDI-ToF MS. The number of mass values searched is 36; and the number of mass values matched is 24. The sequence coverage is 24%. Panel D shows the amino acid sequence of ApcE (LCM129). The N-terminal segments came from the alignment with the sequence of the αAP subunit of G. violaceus. Residues in the regions homologous to G. violaceus αAP are shown on light blue. This αAP-like domain is interrupted by a loop. LCM129 shows four REP domains in tandem defined after comparison with CpcC1 of G. violaceus. Residues of REP domains 1–4, are shown on dark blue. Between the REP domains there are Arms connecting them. Residues of the loop and the Arms, 1–5 top to bottom, are shown as black on white. The 24 amino acid sequences of peptides matched by Mascot program are underlined

Characterization of the multidomain linker proteins

The high molecular weight linkers LRC91 and LR81 are multidomain proteins that show characteristic sequences of REP domains. Both linkers have three REP domains and these are shown in Fig. 6A, clustered together in four groups aligned by their similarity to the complete sequence of the CpcC1 (the phycocyanin associated linker) of G. violaceus. The sequence alignment clearly shows the strong similarity of these REP domains to CpcC1 and among themselves. They exhibit a pattern composed of three tandem repeats having a coding capacity of 190 amino acid residues in CpcC1. The amino acid sequence of LRC91 is composed of an N-terminal segment of 87 amino acid residues followed by rod linker-like REP domains. The REP1 domain stretches from amino acids 88 to 286, REP2 from amino acids 353 to 535 and REP3 from amino acids 623 to 823. The spacing sequences between REP domains are known as Arms. The spacing Arms are 67 and 87 amino acid residues long. The amino acid sequence of LR81 is composed of an N-terminal segment of 71 residues, followed by three REP domains and two Arms. The REP domain 1 extends from amino acids 72 to 264, REP domain 2 from amino acids 296 to 493 and REP domain 3 from amino acids 525 to 724. The spacing Arms are very short, both are 32 amino acid residues long (lines 4 and 5 of the first alignment cluster). Note that the Arms of the top cluster from both linkers differ in length and lack similarity to one another. Figure 6 shows the sequences of LRC91and LR81 compared to the rod linker, CpcC1. The amino acid sequences of the first two multi-domain products of 823 and 759 amino acid residues predict polypeptides of masses of 91,780 and 81,350 Da. The molecular mass of ApcE obtained by SDS-PAGE analysis (Fig. 5A, B) was approximately 120 kDa and after trypsin PMF was, more precisely, 129 kDa (Fig. 5C, D). The amino acid sequence of LCM129 is composed of 1,155 amino acids and the predicted molecular mass is 129,630 Da. Figure 5D shows the AP subunit domain of LCM129, interrupted by a loop. This interrupted phycobiliprotein is the final energy acceptor of the PBS. It is attached to the multidomain linker (black on white). This multidomain region has four REP domains each approximately 125 amino acids long, homologous to the C-terminal part of CpcC1. The Arms between these domains are longer (87–125 amino acids long) than those seen in the other multidomain linkers (Fig. 6). Based on the presence of these four REP domains in LCM129, we can suggest that G. violaceus LCM129 might bind four additional trimers in the core of the PBS. Therefore, the PBS of this cyanobacterium might have a pentacylindrical core.
https://static-content.springer.com/image/art%3A10.1007%2Fs11120-007-9133-9/MediaObjects/11120_2007_9133_Fig6_HTML.gif
Fig. 6

Amino acid sequence alignment of the REP domains of the multidomain proteins LRC91 (Glr1262) and LR81 (Glr2806) and the CpcC1 (Glr0980) linker protein of G. violaceus in Panel A, the comparison is between the complete sequence of CpcC1, (seventh line at the bottom of each cluster), with the three REP domains of LRC91, (first three lines in each cluster) and with the three REP domains of LR81, (fourth to sixth lines) in each cluster. Numbers 1–3 indicate the REP domain number in each linker. The top cluster and the first 15 amino acid residues of the second cluster show the Arms sequences of the three REP domains of each multidomain protein and the CpcD-like domain of the CpcC1 protein (amino acids 1–70 of CpcC1). The black bar above the clustered sequences shows the common domain of the alignments. The sequences shown are derived from the complete genome of G. violaceus (Nakamura et al. 2003a) and the N-terminal extension of CpcC1 from (Gutiérrez-Cirlos et al. 2006) and are aligned using ClustalX with manual adjustments. Identical residues are marked with an asterisk. The period and colon (./:) indicate sequence similarity. Panel B: Sequence alignment of the 40 N-terminal amino acids of the three repeated elements (REP domains 1–3) of the multidomain protein LRC91 (Glr1262) with CpcG linker protein of Synechococcus sp. PCC 7002 and the REP domains 1–3 of the multidomain linker LR81 (Glr2806) with the same CpcG linker protein of Synechococcus sp. PCC 7002

Discussion

G. violaceus is a cyanobacterium that produces PE containing urobilin (Bryant et al. 1981). PE-producing cyanobacteria were classified into three groups based on their ability to adapt physiologically to the color of available light (Bryant 1981; Tandeau de Marsac 1977). In group I, the PBS is independent of the color of light supplied for growth as is the case of G. violaceus (Rippka et al. 1974). In group II, PE production is greater in cells grown in green light than in cells that are grown in red light but the level of PC is not responsive to the color of light, as is the case of Synechococcus marinus WH8102 (Palenik et al. 2003; Six et al. 2005). In group III, PE production increases in green light, and the level of PC decreases, and in red light only PC is present as in F. diplosiphon (Bennett and Bogorad 1973; Tandeau de Marsac et al. 1988). Recently a group IV of chromatic adaptation has been described in some marine Synechococci (Everroud et al. 2006). In group IV the PE proteins are the same in different light conditions; the chromophorylation is affected, the phycourobilin and phycoerythrobilin ratio is modified. It has been stated that in F. diplosiphon, the CpeR is an activator required for the expression of the cpeBA operon (Cobley et al. 2002). Both the nature and order of genes in G. violaceus are different from the two cyanobacterial species belonging to groups II and III in which the genes are clustered in the order cpeESTR and are encoded on the same strand. Figure 1B shows that the G. violaceuscpeR gene follows the operon cpcB1A1 and that the phycoerythrin related genes (cpeBA, cpeY and cpcZ) separate it from genes cpeS and cpeT in the complementary strand. Another peculiarity in the arrangement of the genes that should be noted is the localization of the PC-associated linkers. The cpcC1–2 genes are separated and the cpcD1–2 genes are not in the same operon of cpcBA as in other cyanobacteria (de Lorimier et al. 1990; Kaneko et al. 1996, 2001; Tandeau de Marsac et al. 1988). Instead cpcD1–2 seems to be part of the same operon where the PE-associated linkers are encoded. In this work we present evidence that both the two cpcCs and the two cpcDs of G. violaceus are expressed and their protein products are present in the PBS. In addition, two genes with unknown functions were found in the G. violaceus genome, glr1262 and glr2806 (Nakamura et al. 2003a). The gene glr1262 is situated in the cluster of genes related to PE and to PC (glr1260 to glr1269): pebA-pebB-glr1262-cpeC-cpeD-cpeE-cpcD1-cpcD2-cpcE-cpcF (Fig. 1C). The other novel gene glr2806 is isolated in the genome as indicated in Fig. 1D. In this work we identified the protein products of these genes as multidomain linkers and assigned them a role in the assembly of G. violaceus PBS.

PBSs of the cyanobacterium G. violaceus differ from hemidiscoidal PBSs, in three aspects. First, the PBS appearance under the electron microscope shows the rods of the PBS are grouped as a bundle of 6 rods (Guglielmi et al. 1981) giving the PBS a different shape than the typical fan-like rods in the hemidiscoidal PBS (Bryant et al. 1979; Glazer 1989; Grossman et al. 2001; Adir 2005). Second, is the absence of the regular rod-core linkers encoded by the cpcG genes (Nakamura et al. 2003a, b). Third, is the presence of two multidomain linkers found in silico by Nakamura et al. (2003a, b) unique to this cyanobacterium, and experimentally detected as expressed proteins in this article, which are probably responsible for the unique shape of the PBS and the lack of orthodox linkers.

We have constructed a model that reflects both the pigment and linker protein composition and, which includes the characterization of the three multi-domain linkers, which seem to determine the shape of the PBS of G. violaceus (Fig. 7). To construct the model we have taken into account the overall morphology of the PBS (Guglielmi et al. 1981) and we have assumed that all the rods are homogeneous in PBPs and in linker composition.
https://static-content.springer.com/image/art%3A10.1007%2Fs11120-007-9133-9/MediaObjects/11120_2007_9133_Fig7_HTML.gif
Fig. 7

Schematic representation of the proposed model for localization of the multidomain linkers of the phycobilisome of G. violaceus. Light blue, dark blue and purple discs represent trimers of allophycocyanin, hexamers of phycocyanin and hexamers of phycoerythrin, respectively. The single disc in the rods represents hexamers of phycobiliproteins (α6β6). Image A. Six peripheral rods are bound as a bundle to the allophycocyanin core. Three long (seven discs) vertical cylinders in the front and three in the back and beneath it are the ends of five horizontal cylinders (three have four trimers of AP, the basal cylinders and the center in the upper level and two, label H, with only two trimers of AP) With respect to the interactions of the 91 kDa domains, we propose that each polypeptide will link three rods with the core taking the place of LRC that is missing in the G. violaceus genome. We propose that the 81 kDa polypeptide links six discs of hexameric phycocyanin. It is clear that the presence of these multidomain linkers with short Arms will limit the separation of rods and can explain the parallel rods of the bundle. In image B, the PBS shows one cylinder and two half-cylinders (centered) of the second level of the core, which is light blue in color and in the next plane, the six rods are seen in dark blue color. Image C shows the arrangement of the AP horizontal cylinders, two basal cylinders and one complete cylinder form a pyramid centered under the vertical rods and two additional half-cylinders (H) in the second level, 16 AP trimers conform with the two LCM129, which maintain their four REP domains and each REP (◊) is composed of two homologous halves each of which interacts with adjacent trimers (Capuano et al. 1991). Image D is the bottom view of the pentacylindrical core. The Roman number II indicates the Arm II, which connects REP domain 1 with REP domain 2. Image E is the top view of the pentacylindrical core. The Roman numbers indicate Arms III and IV that connect REP domain 2 with REP domain 3 and REP domain 3 with REP domain 4, respectively. Both half-cylinders (H) are centered to give surface to all of the rods. In the original model, those half-cylinders are at the edges of the complete cylinder (Ducret et al. 1998). The core trimer composition is as follows: 1. −(αβ)3AP−LC 2. −(αβ)3AP 3. −(α2β3)APLCM 4. −(α2β3)APαAPB−LC

The core

The core membrane linker phycobiliprotein of G. violaceus has a molecular mass of 129 kDa (LCM129) as shown in Fig. 5. The polypeptide has four REP domains (Fig. 5) as in Anabaena sp. PCC7120 (Ducret et al. 1998), Anabaena variabilis (Isono and Katoh 1987) and Mastigocladus laminosus (Glauser et al. 1992 and Sidler 1994) that are all members of the so-called pentacylindrical subfamily. The structural implication for the presence of four REP domains in LCM129 is a core with 16 AP trimers since there are two LCM129 in the core and each REP domain interacts specifically with two AP trimers. The two basal cylinders are organized by: the PBP domain of LCM129 (contained in trimer 3 of AP Fig. 7D), the Arm1, REP domain 1 and REP domain 2 (Capuano et al. 1991). The REP domain 3 of each LCM is responsible for binding each hemi-cylinder of the upper complete cylinder. It has been argued that in order to assemble the upper cylinder it is necessary to have antiparallel cylinders in the base (Lundell and Glazer 1983). By the same token it has been asserted that the two hemi-cylinders organized by the REP domain 4 remain as hemi-cylinders and each one must be located on both sides of the upper cylinder (Ducret et al. 1998). However, it has been stated that the PBS of cyanobacteria with an LCM of four REP domains could exhibit tetracylindrical core morphology if their LCMs were oriented in parallel (Anderson and Eiserling 1986). The sequence similarity among LCMs of the different cyanobacteria that have a pentacylindrical core is high, in both their domains, and their arms.

The phycobiliproteins

Since we found evidence both in silico and in experiments (Fig. 5) that LCM129, which constitutes the organizer of the core contains four REP domains, we propose (Fig. 7 Panels C, D and E) a core for the PBS of G. violaceus of 16 allophycocyanin trimers. The core structure proposed for this number of allophycocyanin trimers is a pentacylindrical model composed of 3 complete cylinders (4 trimers of AP each) and two hemi-cylinders (H) (two AP trimers) that flank the upper cylinder (Capuano et al. 1991; Sidler 1994). The reconstitution of the core of M. laminosus and Anabaena sp. PCC 7120 PBS supports this core model (Ducret et al. 1998). The three pigments PE, PC and AP are constitutively expressed in G. violaceus. Since there is the equivalent of 8 hexamers of AP in the core, the total number of PCs will be 24 if the PC to AP ratio is 3 and since the PC to PE is 1.33 and the PE to AP ratio is 2.25, there are 18 PEs in six rods. The proportion of the pigments vary between 1 AP per 2.2 to 3.8 PC and 1.7 to 3 per PE, with mean values of 1 to 3 to 2.35, which is close to that predicted by the model. The only reported molar ratio of isolated G. violaceus phycobilisomes is 1 to 3.9 to 2.9 for AP to PC and PE, respectively (Guglielmi et al. 1981) for which the model will be 8 hexamers of AP in the core and the rod modified to 5 hexamers of PC and 4 hexamers of PE per rod. The absorbance ratio at 566 nm : 620 nm of the whole cell extract was found to be 2, which is in the limits of the model.

The rods

We have shown that 11 protein linkers are expressed in the PBS of G. violaceus (Table 1): two for the core, the LCM129 and LC7.7; two novel multi-domain linkers of high molecular weights—91 and 81 kDa; two small linkers (LR8.1 and LR7.7) to cap the rods (this publication) and finally five medium size linkers (Gutiérrez-Cirlos et al. 2006), two PC associated linkers (LR30.8 and LR31) and three PE associated linkers (LR31.8, LR28.4 and LR28.3). Suggestive evidence concerning the location of the various linkers within the phycobilisome is obtained from the examination of the data in Tables 2, 3 and 4. Three PE disks per rod are needed to fit the expressed three PE-linkers; this proportion is compatible with our estimates of the relative content of the PE and PE associated linkers in Table 2 (LR31.8, LR28.4 and LR28.3). Interestingly, sequence similarity was found between rod linker polypeptides themselves and between rod linkers and CpcD1 (rod terminator), in the 70 N-terminal amino acids of CpcCs and in the 70 C-terminal amino acids of CpeC. Also in the short alignment (30 amino acids), similarity was found between CpcD2 and CpeD (Table 3). We have hypothesized that the rod linker that attaches the distal PC has a preserved CpcD-like domain and this is a signal or structure that recognizes a CpcD polypeptide or a CpcD like domain. These cap linkers have affinity with a distal PC in the rod, which is attached by the medium size linker that has a CpcD-like domain. The argument is based on an experimental observation using the enzyme ferredoxin-NADP+ oxidoreductase of three domains (FNR-3D). The FNR-3D has a CpcD-like domain that attaches this enzyme to the distal PC in the phycobilisomes and this position is preferred over free PC disks or proximal disks (Gómez-Lojero et al. 2003; Schluchter and Bryant 1992). We propose that the position for LR31.8 (CpeC) is in the interface between the distal PC disk and the first PE hexamer. The position of the CpcCs is between the third and the fourth PC. The small length showing similarity between CpcD2 and CpeD suggests that this linker position in the distal PE would attract the CpcD rod terminator at the end of the rod. Finally by exclusion, since the C-terminal of CpeE has no similarity to CpcD, we positioned it to attach the first PE on the second PE disk.
Table 3

Amino acid sequence identity and similarity values for related linkers

 

CpcC1

CpcC2

CpcD1

CpcD2

CpeC

CpeD

CpeE

Glr0950

Gll3219

Gsr1266

Gsr1267

Glr1263

Glr1264

Glr1265

CpcC1

 

276 aa

n.s.a.

70 aa

178 aa

159 aa

149 aa

Glr0950

    

70 aa

  

CpcC2

67/78

 

n.s.a.

68 aa

159 aa

149 aa

159 aa

Gll3219

    

69 aa

  

CpcD1

n.s.a.

n.s.a.

 

41 aa

n.s.a.

34 aa

n.s.a.

Gsr1266

       

CpcD2

52/66

45/61

41/55

 

70 aa

n.s.a.

n.s.a.

Gsr1267

       

CpeC

49/64

55/66

n.s.a.

52/69

 

288 aa

264 aa

Glr1263

52/66

50/70

     

CpeD

59/73

57/69

41/58

n.s.a.

44/56

 

242 aa

Glr1264

       

CpeE

57/68

59/73

n.s.a.

n.s.a.

39/56

55/71

 

Glr1265

       

The first horizontal line at the top and the first vertical column at the left hand margin of the matrix identify the seven linker proteins to be compared. The numbers followed by aa are the numbers of amino acids aligned in the comparison. The numbers shown as paired in a ratio are the % identity above and % similarity below. The n.s.a. means no significant alignment. All sequences are from the complete genome of G. violaceus (Nakamura et al. 2003a). The CpcC sequences include the N-terminal extensions (Gutiérrez-Cirlos et al. 2006)

Multidomain linkers: 81 kDa multidomain protein

To understand the composition and localization of the proteins of the G. violaceus PBS, we analyzed the colored bands obtained from the Triton X 100 released PBS components and fractionated by sucrose gradient centrifugation (Fig. 2). These bands include the PBS and PBS substructures. Data derived from the heavy blue band (Fig. 2D), with an estimated molecular mass between 1500 and 2000 kDa showed that this band is rich in AP and PC (fraction 5 Fig. 2A, C spectrum 5 and band II Fig. 2D). The presence of core linkers LCM129 and the LC7.9 and the presence of an 81 kDa protein in higher concentration (Fig. 3 lane II and Table 2) allows us to conclude that this band contained the core substructure and a rod substructure of 6 PC hexamers organized by the 81 kDa protein linker (LR81).

The rod core linker (LRC) and the 91 kDa multidomain protein

To get further insight into the localization of the different linkers, we measured the fluorescence excitation spectra of fraction I (PBS) and fraction II (heavy blue band). There was no significant transfer of energy between PC and AP in this fraction (Fig. 4B). In contrast energy transfer does occur in the intact PBS (Fig. 4A). The absence of the 91 kDa proteins from all substructures suggests that the product of glr1262 gene binds the rods to the core (LRC) and that this binding is released during the dissociation of the PBS. In addition we know from the literature that no sequence similarity was found between rod linkers and the rod core linker polypeptides in the 40 N-terminal amino acids. However, all characterized rod-core linker polypeptides are highly similar in their N-terminal amino acid sequences (Glauser et al. 1992). For example, if the REP domains of 91 kDa protein are aligned to the first 40 amino acids of CpcG of Synechococcus sp. PCC 7002, a 45% similarity is found while a similar alignment with the polypeptide of 81 kDa yields only 30% similarity (Fig. 6 Panel B).

The structure as a bundle of rods and the multidomain linkers

The REP domains of these multidomain linkers have two relevant features: an unusual length and the high sequence identity and similarity to REP domains of the same polypeptide. The sequence length of the REP domains (around 190 amino acids, Fig. 6) is enough to allow them to join hexamers rather than trimers. In contrast, the length of the core membrane linker (LCM) REP domains is around 125 amino acids (Fig. 5D) and it binds trimers (Capuano et al. 1991). The shortness of these Arms will determine the space between different domains and will tie together the rods in the bundle structure of the G. violaceus PBS. In addition, the high identity and similarity between REP domains of the same polypeptide (see Table 4, bold numbers) indicate there is interaction between REP domains and so the PBPs must be identical and this also can be expected from the modification and recognition of the PBPs in parallel homogenous disks. LR81 has the most homogeneous REP domains, followed in homogeneity by the REP domains of LRC91. Both are clearly different from LCM129, which has heterogeneous REP domains, as might be expected, since the position of these REP domains in the core is in contact with heterogeneous trimers of AP (Fig. 7). Thus the heterogeneity is derived from the different α subunits or by the linker attached to each trimer.
Table 4

Amino acid sequence identity and similarity values of the REP domains of multidomain linkers

 

LCM129

LCM129

LCM129

LCM129

LRC91

LRC91

LRC91

LR81

LR81

LR81

REP-1

REP-2

REP-3

REP-4

REP-1

REP-2

REP-3

REP-1

REP-2

REP-3

LCM129

REP-1

100

         

LCM129

REP-2

38/59

100

        

LCM129

REP-3

46/67

35/58

100

       

LCM129

REP-4

47/64

40/57

42/57

100

      

LRC91

REP-1

39/56

33/60

34/54

37/58

100

     

LRC91

REP-2

46/50

37/59

42/58

40/55

69/81

100

    

LRC91

REP-3

38/52

36/58

38/57

37/55

60/68

78/84

100

   

LR81

REP-1

42/58

31/58

35/51

38/58

61/71

62/73

64/74

100

  

LR81

REP-2

42/58

31/58

35/51

39/59

65/75

65/77

58/73

89/92

100

 

LR81

REP-3

43/58

27/48

37/52

40/60

68/75

65/77

60/73

89/92

83/87

100

Comparison of the amino acid sequences of the linker REP domains in G. violaceus phycobilisomes. The first horizontal line at the top and the first vertical column at the left hand margin of the matrix identify the 10 REP domains to be compared. The numbers shown as paired in a ratio are the % identity above and % similarity below. All sequences are from the complete genome of G. violaceus. After the alignment of the LCM129, LRC91 and LR81 with CpcC1 of G. violaceus, the REP domains were defined. There are four REP domains for LCM129, three for LRC91 and for LR81. Each one was used in a Blastp comparison to the other sequences to obtain the values of identity and similarity presented in the Table

We have built a comprehensive model for the arrangement of pigments and proteins in the PBS light-harvesting complex of the cyanobacterium G. violaceus. Our work shows that the complete genome sequences predict differences in the composition of large macromolecular pigment protein assemblies such as the PBS, and that these changes are amenable to characterization by combined spectroscopic and biochemical analyses. Also we have demonstrated that subtle variations in the repetitive sequences of the linker components may give rise to large overall changes in PBS morphology.

Acknowledgements

We are greatly indebted to Diego González-Halphen for his editorial assistance and stimulating discussions. We also thank Mary Bower of Purdue University for her expert work in N-terminal sequencing, Jorge Zarco Mendoza for his technical assistance and Cristina Díaz-Trueba for her secretarial assistance. We also appreciate critical reading of the manuscript by Loretta Krogmann and Michele Krogmann.

Grant SEP-CONACYT-México supported this study. Reference Number 46416-Q.

Copyright information

© Springer Science+Business Media B.V. 2007