Journal of Experimental Botany, Vol. 59, No. 5, pp. 10231034, 2008 doi:10.

1093/jxb/erm282 Advance Access publication 7 December, 2007

SPECIAL ISSUE RESEARCH PAPER

Proteomic analysis of the cyanobacterium of the Azolla symbiosis: identity, adaptation, and NifH modication
Martin Ekman1, Petter Tollback2 and Birgitta Bergman1,*
1 2

Department of Botany, Stockholm University, SE-106 91 Stockholm, Sweden

Stockholm University Proteomics Facility, Department of Analytical Chemistry, Stockholm University, SE-106 91 Stockholm, Sweden

Received 24 June 2007; Revised 21 October 2007; Accepted 22 October 2007

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

Abstract
Cyanobacteria are able to form stable nitrogen-xing symbioses with diverse eukaryotes. To extend our understanding of adaptations imposed by plant hosts, two-dimensional gel electrophoresis and mass spectrometry (MS) were used for comparative protein expression proling of a cyanobacterium (cyanobiont) dwelling in leaf cavities of the water-fern Azolla liculoides. Homology-based protein identication using peptide mass ngerprinting [matrix-assisted laser desorption ionization-time of ight (MALDI-TOF-MS)], tandem MS analyses, and sequence homology searches resulted in an identication success rate of 79% of proteins analysed in the unsequenced cyanobiont. Compared with a free-living strain, processes related to energy production, nitrogen and carbon metabolism, and stress-related functions were upregulated in the cyanobiont while photosynthesis and metabolic turnover rates were down-regulated, stressing a slow heterotrophic mode of growth, as well as high heterocyst frequencies and nitrogen-xing capacities. The rst molecular data set on the nature of the NifH post-translational modication in cyanobacteria was also obtained: peptide mass spectra of the protein demonstrated the presence of a 300 400 Da protein modication localized to a specic 13 amino acid sequence, within the part of the protein that is ADP-ribosylated in other bacteria and close to the active site of nitrogenase. Furthermore, the distribution of the highest scoring database hits for the

identied proteins points to the possibility of using proteomic data in taxonomy.

Key words: Azolla, cyanobacteria, proteomics, symbiosis, taxonomy. NifH modication,

Introduction Symbiotic associations between the water-fern Azolla and nitrogen-xing cyanobacteria have gained attention through the centuries due to their potential as natural nitrogen fertilizers, especially in rice cultivation (van Hove and Lejeune, 2002). The symbiotic organs of the Azolla plants are the comparatively large cavities that occupy each dorsal leaf of the plant. These cavities are in nature obligately infected by lamentous cyanobacteria (cyanobionts) and bacteria, both held within a mucilaginous sheath (Nierzwicki-Bauer et al., 1989; LechnoYossef and Nierzwicki-Bauer, 2002). The cyanobiont is restricted to the Nostaceae and was originally designated Anabaena azollae (Strasburger, 1873). The taxonomic status is, however, still not claried, and relatedness to the genera Nostoc and Anabaena, as well as to neither of these two genera, has been proposed (Meeks et al., 1988; Canini et al., 1992b; Baker et al., 2005; Svenning et al., 2005). As in other plantcyanobacterial symbioses, the plant host supplies the cyanobiont with xed carbon, while the cyanobiont in return supplies the plant with the nitrogen needed via nitrogen xation (Rai et al., 2000; Lechno-Yossef and Nierzwicki-Bauer, 2002). The benets for the plant are therefore apparent, while less so for the

* To whom correspondence should be addressed. E-mail: Birgitta.Bergman@botan.su.se Abbreviations: ESI, electrospray ionization; FBA, fructose/tagatose bisphosphate aldolase; FBP, fructose-1,6-bisphosphatase; F6P, fructose-6-phosphate; FNR, ferredoxin NADP+ reductase; G6P, glucose-6-phosphate; GPI, glucose-6-phosphate isomerase; GS, glutamine synthetase; MS, mass spectrometry; MALDI-TOF, matrix-assisted laser desorption ionization-time of ight; OPP, oxidative pentose phosphate; 6PGD, 6-phosphogluconate dehydrogenase; PMF, peptide mass ngerprinting; Q-ToF, quadropole-time of ight; TCA, tricarboxylic acid. The Author [2007]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org

1024 Ekman et al.

cyanobiont which by itself has photosynthetic capacities. Attempts have been made to identify mechanisms used by the plant to maintain the cyanobiont as an efcient nitrogen xer, but many aspects are still unknown. Unique to the Azolla symbiosis among cyanobacterial plant symbioses is that a small cyanobiont inoculum is sequestered into the reproductive organ (sporocarp) and vertically transferred from one generation to the next (Peters and Meeks, 1989; Zheng et al., 1990; Perkins and Peters, 1993; Rai et al., 2002). In all other cynobacterialplant symbioses, as well as Frankia and Rhizobium plant symbioses, de novo infection of each individual plantlet is required (Vessey et al., 2005; Bergman et al., 2007). There is also no conrmed report of successful cultivation of the cyanobiont outside the Azolla plants or reconstitution of the symbiosis (Tang et al., 1990), suggesting that some crucial trait needed for survival as a free-living organism may have been lost during co-evolution of the partners, possibly via gene loss (due to redundancy) from the cyanobiont, or gene transfer to the nucleus of the host. Hence, in the context of the endosymbiotic theory, according to which chloroplasts evolved from an ancient monophyletic symbiosis between a cyanobacterium and a pigment-free eukaryote, the cyanobiont of Azolla may be evolving towards becoming a nitrogen-xing organelle. This is also supported by PCR-based DNA ngerprinting which shows that the association is highly specic and that each Azolla species associates with one specic cyanobacterial strain irrespective of geographical origin (Zheng et al., 1999). In order to obtain further insights into adaptations required by a cyanobiont, potentially on its way to develop into a nitrogen-xing organelle, an investigation of the proteome of the cyanobiont in Azolla liculoides was initiated. As the genome of this organism has not yet been sequenced, proteins were identied by a combination of matrix-assisted laser desorption ionization-time of ight mass spectrometry (MALDI-TOF MS), electrospray ionization (ESI), tandem MS (MS/MS), and peptide homology software analyses, an approach which has previously been successfully used on other non-sequenced organisms such as Xenopus laevis (Liska et al., 2004) and Candida magnoliae (Kim et al., 2004). As MS is also suitable for examining protein modications, attempts were also made to identify the modication of NifH, one of the subunits of the nitrogen-xing enzyme nitrogenase and therefore a process (nitrogen xation) of outmost importance for the symbiosis. The information gained is discussed in the context of taxonomy, adaptations, evolution, and codevelopment of the cyanobiont and its host. Materials and methods
Growth conditions and isolation of the cyanobiont Azolla liculoides plants were grown hydroponically in the greenhouse at the Department of Botany, Stockholm University.

The temperature was maintained at ;30 C and the light varied according to natural daylight with some addition of light from articial sources. The cyanobiont was separated from the plant by the gentle rolling technique (Meeks, 1988). After crushing the plant in a buffer containing 50 mM HEPES pH 7.8 and 1% polyvinylpyrrolidone (PVP), using a roller, the cyanobacterial slurry obtained was ltered through a nylon lter with a pore size of 100 lm. The cyanobacterial slurry was concentrated by centrifugation (1200 g, 10 min), placed on top of 40% Percoll (GE Healthcare, Uppsala, Sweden) in a centrifuge tube, and centrifuged for 5 min at 400 g. The cyanobacteria gathered at the bottom of the tube, while chloroplasts, small-sized plant cell debris, and bacteria remained in the supernatant. The pellet was collected and centrifugation repeated with 80% Percoll for 15 min. This time the cyanobacteria stayed on top of the Percoll while plant starch granules formed a pellet. Approximately 90% pure cyanobacterial preparations were obtained as evidenced by light microscopy. Protein extraction Proteins were extracted as previously described (Ekman et al., 2006) by freezing the samples in liquid nitrogen and grinding with acid-washed sand, followed by sonication in sample buffer. The protein concentration in the samples was determined using the RCDC kit (Bio-Rad, Hercules, CA, USA). 2D gel electrophoresis A 200 lg aliquot of protein was separated as previously described (Ekman et al., 2006), rst by isoelectric focusing using the IPGphore system (GE Healthcare) and immobiline 18 cm gel strips of pH 47, and secondly according to size in 10% acrylamide gels. The gels were stained with SyproRuby (Bio-Rad) according to the manufacturers instructions, and a Typhoon 8600 laser scanner (GE Healthcare) was used to visualize the gels. In-gel digestion and mass spectrometry All protein spots selected from the 2D gels for analysis by MALDITOF MS were subjected to in-gel tryptic digestion, according to Gharahdaghi et al. (1999). In order to examine nitrogenase (NifH) modications, the trypsin was replaced by GluC (V8 proteinase) of the same concentration. For double digests, GluC was added to the already trypsin-digested samples and incubated overnight. MALDI-TOF MS peptide mass spectra were obtained using a Voyager-DE STR mass spectrometer (Applied Biosystems, Foster City, CA, USA). As matrix, a-cyano-4-hydroxy cinnamic acid was used. ESI-MS/MS experiments were carried out using a Micromass Q-ToF (Waters, Milford, MA, USA) as previously described (Ekman et al., 2006). Prior to analysis on the Q-ToF, salts and contaminants were removed from the samples by ZipTip purication and the peptides were eluted in water/acetonitrile (1:1 V:V) with 0.1% formic acid. A 13 ll aliquot of the sample was loaded into a nanospray tip (Proxeon, Odense, Denmark). Data processing MALDI-TOF MS spectra were calibrated using internal trypsin autolysis peptides and the MoverZ software available online from Genomic Solutions (Ann Arbor, MI, USA). Obtained peptide mass lists were compared with sequences present in the NCBInr database using the Mascot (Perkins et al., 1999) search engine, available online from Matrix Science (Boston, MA, USA) with the following settings: taxonomy was set to all organisms; the mass tolerance to 30 ppm; xed modications to alkylation of cysteine by carbamidomethylation; and variable modications to oxidation of methionine.

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

Azolla cyanobiont proteome 1025

MS/MS spectra were processed with MassLynx MaxEnt 3 before database searches in NCBInr, utilizing Mascot. As the Azolla cyanobiont is not represented in the available databases, this strategy relies on fragmentation of conserved parts of the proteins. Hence, additional information could be obtained by homology searches, allowing some individual amino acid substitutions. For this purpose, the peptides were sequenced de novo using the processed MS/MS spectra and the BioLynx peptide sequencing tool (Micromass). One to seven peptides/sample were fragmented, each resulting in 110 suggested sequences. All sequences, with 46 amino acid substitutions allowed, were submitted to MS-Homology from Protein Prospector, available online from the University of California, San Francisco (see http://donatello.ucsf.edu/index.html or http://prospector.ucsf.edu/). Data for which positive identications were obtained by common database search (cf below) were not included in the homology searches. Quantitative protein analysis The relative quantity of all identied proteins of the Azolla cyanobiont was determined using the PDQuest software (Bio-Rad). Spot intensities were normalized to the total staining intensity of the gel in question. The quantication was based on four replica gels: two gels of each of two separate protein extractions. For proteins whose identity had also been determined in the proteome of Nostoc PCC 73102 (Ran et al., 2007), the volume ratio (i.e the ratio between the abundance in the Azolla cyanobiont and the abundance in Nostoc 73102) of the protein quantities relative to those determined in this organism was also calculated.

suggested as appropriate for identifying proteins of unsequenced organisms (Liska and Shevchenko, 2003).
Protein identication

Results and discussion In proteomics, proteins separated by two-dimensional gel electrophoresis are usually identied by peptide mass ngerprinting (PMF), which is a MALDI-TOF MS-based method that relies on matching of peptide masses to corresponding masses derived from theoretical digestions of protein sequences present in a database. Protein identication using this method requires that a large proportion of the protein in question has a sequence that is identical to that of a database entry. For instance, theoretical predictions by Wilkins and Williams (1997) proposed a requirement of 80% sequence identity. Therefore, the use of this method for unsequenced organisms usually results in a low protein identication success rate, e.g. 45% for Xenopus (Liska et al., 2004) and 42% for Zea mays (Chang et al., 2000). However, this rate may be improved by using tandem MS (MS/MS) by which proteins can be identied even if only a small part of the protein is completely conserved and only a few peptides match the database protein sequence (Shevchenko et al., 2001). Furthermore, by using the MS/MS spectra to predict amino acid sequences de novo, and allowing amino acid substitutions, homology searches using these predicted sequences may identify proteins with sequence identities to their closest homologue as low as 65% (Liska et al., 2004). Here, a combination of these three methods, PMF, MS/MS, and homology searches, was used to identify proteins in the Azolla cyanobiont, a strategy previously

In order to identify cellular mechanisms (adaptations) acting in the cyanobiont of the Azolla symbiosis, the cyanobacteria were extracted from the greenhouse-grown (cloned) A. liculoides plants by the gentle rolling technique, followed by protein extraction and separation of the proteins by 2D gel electrophoresis. Approximately 300 protein spots were visualized when using a broad pH interval, 47 (Fig. 1A). The 52 most abundant proteins were selected for analyses by MALDITOF MS (Fig. 1A; Table 1). Of these, 32 spots were signicantly identied (Mowse score >76) (Pappin et al., 1993) by their peptide mass ngerprints, resulting in 27 different proteins, as ve of the proteins resolved into two spots. For another six proteins, the highest scoring hit in the database were cyanobacterial proteins with the expected mass and pI, but with a Mowse score of <76. Five of these were therefore analysed by MS/MS, and for three identities were conrmed; two (az21 and az50) by database searches using the MS/MS spectra and one (az26) via homology search. For the 14 remaining proteins, no cyanobacterial candidate was identied by PMF. Eight of these were analysed by MS/MS and six were identied; four by their MS/MS spectra (az04 and az30 by two peptides, az16 and az41 by one peptide) and two by homology search (az09 and az37). Hence, nine (3+6) of the 13 proteins analysed by MS/MS in combination with homology searches were identied. Some proteins with low peptide concentration, giving weak MALDI-TOF spectra, were not analysed by the less sensitive MS/MS. In total, 11 of the analysed proteins were not identied, giving a success rate for the MALDI-TOF analyses of 62% (32/52), and a success rate for MS/MS analyses in combination with homology searches of 70% (9/13). Since most MALDI-TOF-identied proteins would, if tested, have been identied by MS/MS, MS/MS would appear to be the method of choice when attempting to identify proteins in organisms with non-sequenced genomes, but is also considerably more time-consuming than MALDI-TOF MS.
Proteomics and taxonomy

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

As a relatively high percentage (62%) of the proteins were identied by PMF alone, a high percentage of the genome of the Azolla cyanobiont must be highly homologous to genomes present in the database. Indeed, all except two of the proteins identied showed highest homologies to genes/proteins of cyanobacteria of the same morphotype (lamentous and heterocystous; Section IV cyanobacteria; sensu Rippka et al., 1979); the sequenced strains Nostoc punctiforme (i.e. Nostoc ATCC 29133 comparable with

1026 Ekman et al.

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

Fig. 1. Protein patterns of (A) the cyanobiont of Azolla and (B) the symbiotically competent, isolated, and free-living Nostoc PCC 73102. Numbers on gel A refer to Table 1; numbers on gel B refer to proteins in Nostoc PCC 73102 previously identied (see text for details; Ran et al., 2007).

Nostoc PCC 73102; 13), Anabaena variabilis (14), and Nostoc PCC 7120 (10), and the unsequenced Nostoc commune (1) and a Nostocaceae cyanobiont (1). One protein was identied via the genome of the unicellular cyanobacterium Synechosystis PCC 7942 (Table 1). These data show that the cyanobiont of A. liculoides, and perhaps all Azolla cyanobionts, are phylogenetically closely related to other heterocystous cyanobacteria (Rippka et al., 1979) such as Nostoc and Anabaena. Recent studies on 16S rDNA phylogeny of Nostoc and Anabaena and some symbiotic strains, including the cyanobiont of A. liculoides, showed that the three sequenced Nostoc and Anabaena strains mentioned above are closely related, while the Azolla cyanobiont appeared in a separate branch constituting true Anabaena strains (Svenning et al., 2005). The distribution of the highest scoring protein hits found here is in line with such a situation, i.e. the Azolla cyanobiont is approximately equidistant from the sequenced Nostoc/Anabaena strains. However, as none of these sequenced Nostoc/Anabaena strains belongs to the true Anabaena branch, a more detailed analysis of the phylogeny of the Azolla cyanobiont was not possible. The taxonomic uniqueness of the Azolla symbiosis among cyanobacteria in plant symbiosis is evident moreover when comparing proteomic-based phylogenetic data from the cyanobiont of Azolla with the corresponding data of the cyanobiont in the angiosperm Gunnera manicata as the latter are almost exclusively based on the N. punctiforme genome (Ekman et al., 2006). This emphasizes a distinct genetic difference between the permanently associated cyanobiont of Azolla and that of the de novo infection-based Gunnera cyanobiont.

The fact that proteomic-based taxonomic analyses are dependent on which genomes are present in the database is a limitation to this approach. However, given the recent technological development in high-throughput sequencing of genomes and the subsequent fast increase in numbers of sequenced genomes (especially microbial), the use of such proteomics data from unsequenced organisms in phylogenetic analyses has a high potential to become a useful complement to traditional methods. One protein was identied as a non-cyanobacterial protein, the ATP synthase of the liverwort Anthoceros formosae. The reason for this is probably that some proteins from the Azolla plant remained in the extract since ATP synthase is a highly abundant and also highly conserved plant protein, and that the Azolla protein sequence was not present in the database.
Protein proles in the Azolla cyanobiont The proteins identied in the Azolla cyanobiont were restricted to those most highly expressed (Fig. 1), and the majority of these proteins have functions related to fundamental processes in the cyanobacterial cell machinery, such as translation, stress response, ATP synthesis, and carbon and nitrogen metabolism (Table 1). The relative abundances of the SyproRuby-stained proteins identied in the 2D gels of the Azolla cyanobiont were determined using the PDQuest (Bio-Rad) software. Twenty-four (ve of which resolved in two different spots) of the proteins identied in the Azolla cyanobiont (Table 1) were previously identied in an attempt to prole the proteome of Nostoc PCC 73102 (originally isolated from the cycad Macrozamia) when grown under

Table 1. Proteins identied in the cyanobiont of the water-fern Azolla liculoides

Gel IDa Annotation Protein ID MALDI- Matched % MS/MS Homology Theoretical Theoretical Observed Observed Organism TOF peptidesc Sequence no. of searche mol. wt pI mol. wt pI d Mowse coverage peptides scoreb 88 92 94 88 91 92 76 82 95 78 122 125 84 97 111 76 54 88 37 78 80 32 98 78 95 94 127 12/39 11/28 9/21 13 18 17 2 9/17 10/23 8/11 8/18 7/14 8/16 7/15 11/27 13/32 6/9 9/27 12/32 8/24 5/13 6/9 5/23 8/21 8/21 4/18 8/15 7/18 15 16 12 15 1 19 21 12 27 17 12 1 25 40 19 12 17 8 16 18 8 18 22 2 0 Yes No Yes 147 76 68 46 59 78 76 79 72 58 58 54 52 52 37 36 42 45 56 53 49 50 58 65 58 53 52 39 39 39 0 14/32 46 No 32 5.1 30 5.4 Anabaena variabilis 9.10** 4.8 4.9 4.8 4.7 4.9 5.2 5.2 5.6 5.9 5.1 5.1 5.1 5.0 4.9 5.1 5.6 5.2 5.5 5.2 5.2 5.7 5.5 5.5 5.6 5.5 6.3 5.9 5.4 5.5 5.2 150 75 70 48 60 88 75 75 70 61 60 55 50 50 41 39 42 44 59 54 50 49 60 65 60 56 48 41 39 38 4.8 4.8 4.7 4.6 4.9 5.1 5.2 5.5 6.2 5.1 5.2 5.1 5.0 4.9 4.9 5.0 5.3 5.4 5.4 5.4 5.5 5.6 5.6 5.7 5.8 6.3 6.2 5.6 5.4 5.3 Nostoc PCC 7120 Nostoc PCC 7120 Nostoc PCC 73102 0.73 Anabaena variabilis 4.21** Nostoc PCC 73102 Nostoc PCC 73102 Nostoc PCC 73102 Nostoc PCC 7120 Anabaena variabilis Anabaena variabilis Anabaena variabilis Nostoc PCC 7120 Anabaena variabilis Anabaena variabilis Synechococcus elongatus PCC 7942 Nostoc PCC 7120 Nostoc PCC 7120 Nostoc PCC 73102 Anthoceros formosae Nostoc PCC 7120 1.04 Anabaena variabilis Anabaena variabilis Anabaena variabilis Nostoc PCC 73102 Nostoc PCC 73102 Nostoc PCC 73102 Anabaena variabilis Nostoc PCC 73102 Nostoc PCC 73102 Nostoc PCC 73102 2.14* 1.68* 0.68 1.30 1.11 1.87* 1.87* 2.45** 1.70** 1.70** 0.71* Volume ratio cyanobiont/ Nostoc PCC 73102f

az01 az02 az03 az04 az05 az06 az07 az08 az09 az10 az11 az12 az13 az14 az15 az16 az17 az18 az19 az20 az21 az22 az23 az24 az25 az26 az27 az28 az29 az30 az31 az32 az33 az34 az35

RNA polymerase beta prime subunit Heat shock protein hsp90 family Chaperonin Phosphotransferase system, fructose-specic IIC component Chaperonin GroEL Polynucleotide nucleotidyltransferase Translation elongation Phosphoketolase Transketolase Chaperonin GroEL Chaperonin GroEL ATP synthase alpha ATP synthase beta ATP synthase beta Fructose-1,6-bisphosphatase NADPH:quinone reductase Phosphoglycerate kinase GTPases translation elongation factors Anthoceros ATPase Not identied Glutamine synthetase ADP-glucose pyrophosphorylase Thioredoxin reductase Glucose-6-phosphate isomerase Uncharacterized avoproteins Nitrogenase (NifK) Rubisco 6-Phosphogluconate dehydrogenase Not identied Phosphoribulokinase Fructose/tagatose bisphosphate aldolase Fructose/tagatose bisphosphate aldolase Not identied Not identied Nitrogenase reductase (NifH)

NP_485636 BAB74022 ZP_00107038 ZP_00158087 ZP_00110155 ZP_00106148 ZP_00107087 BAB73549 ZP_00163127.2 ZP_00163108 ZP_00163108 NP_484049 ZP_00159432 ZP_00159433 ZP_00163422 BAB74647 NP_488171 ZP_00107088 BAC55424 P00964 ZP_00158969 ZP_00161577 NP_485093 ZP_00111402 ZP_00112339 ZP_00108159 ZP_00158100 ZP_00109191 ZP_00110670 ZP_00110670

0.93

Azolla cyanobiont proteome 1027

2.51* 0.44* 2.47* 0.40** 0.78 0.78

2 7/12 7/12 17 16

P0A3S1

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

1028 Ekman et al.

Table 1. (Continued) Gel IDa Annotation Protein ID MALDI- Matched % MS/MS Homology Theoretical Theoretical Observed Observed Organism TOF peptidesc Sequence no. of searche mol. wt pI mol. wt pI Mowse coverage peptidesd b score 102 89 100 115 94 88 40 86 81 ZP_00162489 ZP_00159658 68 63 4/10 5/11 32 16 2 11 44 6.1 5.9 40 39 6.4 6.3 Anabaena variabilis 0.73 Anabaena variabilis 16/47 53 1 0 5/6 7/16 8/12 7/12 5/12 6/25 12/30 10/18 20 1 32 24 19 54 8 10 15 No Yes No 32 29 24 23 22 34 34 11 80 99 95 5.1 5.0 4.9 4.9 5.5 6.3 6.3 9.7 5.1 5.4 5.2 31 31 26 22 22 31 31 15 78 90 95 5.3 4.8 4.9 4.8 5.4 6.4 6.6 6.6 5.3 5.3 5.4 Volume ratio cyanobiont/ Nostoc PCC 73102f

az36 az37 az38 az39 az40 az41 az42 az43 az44 az45 az46 az47 az48 az49 az50 az51 az52
a b

Nitrogenase reductase (NifH) (modied) Glutathione S-transferase Not identied Not identied Peroxiredoxin Peroxiredoxin Superoxide dismutase Ferredoxin-NADP(+) reductase Ferredoxin-NADP(+) reductase Phycocyanin Glycyl-tRNA synthetase, beta subunit Chaperonin clpB2 Aconitate hydratase Not identied Carboxysome shell protein Nucleoside-diphosphate-sugar epimerases Not identied

P0A3S1 ZP_00111757 ZP_00108523 BAB76340 AAF25009 CAA51088 CAA51088 AAO31788 ZP_00161122 Q8YM56 ZP_00160026

Anabaena variabilis 3.46* Nostoc PCC 73102 Nostoc PCC 73102 Anabaena variabilis Nostoc commune Nostoc PCC 7120 Nostoc PCC 7120 Nostocaceae cyanobiont AE1 Anabaena variabilis

0.74 2.95** 2.95** 0.68

Nostoc PCC 7120 Anabaena variabilis 0.89

Protein IDs as denoted in Fig. 1. For proteins identied by MALDI-TOF MS, Mowse score, number of matched peptide, and percentage sequence coverage is reported. If the highest scoring database hit was cyanobacterial protein, these are reported, even if the Mowse score was below the signicance limit of 76. c The two numbers indicate matched peptides and all peptides in the spectrum. d For proteins identied by database searches of their MS/MS spectra, the number of peptides matching the protein is reported. e A successful identication by homology search is reported with yes and unsuccessful with no. f Volume ratios are in relation to the same proteins identied in free-living Nostoc PCC 73102 and these are marked with (*) when the difference was signicant at the 95% level and with (**) when signicant at the 99% level (Students t-test). When a protein resolved in two different spots, their volumes were summed.

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

Azolla cyanobiont proteome 1029

free-living diazotrophic conditions (Ran et al., 2007). Sixteen (four of which resolved in two different spots) of these cyanobiont proteins were identied based on sequence similarity to proteins of other cyanobacteria, but their closest homologue in the Nostoc PCC 73102 genome had been identied in this study of the Nostoc PCC 73102 proteome. The abundances of these 24 proteins (plus a modied form of NifH, see below) could therefore be compared with the corresponding proteins in the cultured Nostoc PCC 73102 (Fig. 1B) and the relative volume ratio calculated (Table 1). The discussion below will focus on these proteins and less on those only identied in the cyanobiont (Table 1).
Nitrogen xation and assimilation Two of the most abundant protein spots in the Azolla cyanobiont 2D gels were identied as NifH (az35 and az36; Fig. 1), the small subunit of the nitrogenase complex that functions as a reductase (Fe-protein). As one of these spots (az36) was slightly more acidic and of a higher mass, this may represent a modied, probably inactive, form of NifH. Up to 25% of NifH in the cyanobiont of Azolla may therefore be inactive. However, the active non-modied form of NifH showed a >9 times higher level in the cyanobiont compared with the cultured nitrogen-xing Nostoc PCC 73102. Likewise, NifK (az26), one of the larger subunits of the nitrogenase enzyme complex, was 2.5 times more abundant in the cyanobiont. These ndings reect the higher heterocyst frequency in the Azolla cyanobiont, being ;20%, compared with 510% in free-living cyanobacteria (Adams, 2000). Indeed, N2 xation rates in the Azolla cyanobiont may be 418 times higher than in free-living cyanobacteria (Watanabe, 1982) and increase with heterocyst frequencies (Braun-Howland et al., 1988; Canini et al., 1990). The nitrogen xed into ammonia is rapidly assimilated by glutamine synthetase (GS) into glutamine in free-living cyanobacteria (Flores and Herrero, 2005). As seen in Fig. 1 (Table 1), the total GS protein (az21) levels were similar in the A. liculoides cyanobiont and in the cultured Nostoc PCC 73102. This is in contrast to previous studies reporting cyanobiont GS protein levels in Azolla caroliniana to be 540%, and glnA (encoding GS) transcription levels ;10% of the levels in free-living Nostoc and Anabaena strains (Nierzwicki-Bauer and Haselkorn, 1986; Lee et al., 1988; Peters and Meeks, 1989). Nonetheless, given the high frequency of heterocysts in the cyanobiont of Azolla, and that heterocysts normally have twice the amount of GS protein compared with vegetative cells under free-living conditions (Bergman et al., 1985), the GS protein levels detected still indicate a reduction of the GS level in the cyanobiont heterocysts. Similar reductions have previously been observed in heterocysts of

cyanobionts in other plantcyanobacterial symbioses, e.g. Gunnera, Anthoceros, lichens, and diatoms (Rai et al., 2000). It is also possible that nitrogen assimilation in the cyanobiont is mainly regulated via the GS activity, which would not be reected in the GS protein abundance. Moreover, it has been shown that although the cyanobiont releases N for the benet of the host, ;60% of the nitrogen xed is retained by the A. caroliniana cyanobiont (Peters and Meeks, 1989), which is a higher percentage than that reported for cyanobacteria in other symbioses (Rai et al., 2000). This would require a substantial GS activity to be maintained given the pronounced upregulation of the nitrogenase enzyme levels of the Azolla cyanobiont detected here (Table 1).
Energy production and conversion

N2 xation is an energetically costly process in terms of reducing equivalents and ATP. In the Azolla cyanobiont, it has been suggested that nitrogenase activity is mainly supported by ATP produced via cyclic photophosphorylation (Peters and Meeks, 1989). Furthermore, high respiratory rates in heterocysts lower the oxygen concentration (Fay, 1992). As both these processes involve ATP synthase (az12az14), the 2-fold higher levels of this protein complex observed in the cyanobiont (Table 1) may be associated with the high levels of nitrogenase and frequency of heterocysts. An important enzyme in the production of reducing equivalents in organisms performing oxygenic photosynthesis is ferredoxin NADP+ reductase (FNR) (az43 and az44; Table 1), which catalyses the terminal step of the photosynthetic electron transport chain. However, it has also been shown that N2-xing non-photosynthetic heterocysts may contain up to 14 times more FNR than the photosynthetic vegetative cells (Razquin et al., 1996) and that FNR has two promoters, one of which is specic for heterocysts (regulated by NtcA) (Valladares et al., 1999). The low rates of photosynthesis in the Azolla cyanobiont (see below) imply that the four times higher levels of this enzyme in symbiosis may be associated with a high symbiotic heterocyst frequency. In symbiosis, FNR activities may therefore mainly be related to electron transfer reactions required to provide reducing equivalents to nitrogenase and respiratory electron transport (Schmetterer, 1994) and also to cyclic electron transport in the heterocysts, as such a function has been observed for this enzyme in chloroplasts (Zhang et al., 2001).
Photosynthesis and CO2 xation The levels of the two enzymes specic for the Calvin cycle, Rubisco (az27) and phosphoribulokinase (az30), were lower in the Azolla cyanobiont than in the cultured Nostoc PCC 73102 (Table 1). These ndings agree with previous data showing a reduction in CO2 xation activity

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

1030 Ekman et al.

(Kaplan and Peters, 1988) and Rubisco mRNA levels (5 7 times lower) (Braun-Howland and Nierzwicki-Bauer, 1990) in the Azolla cyanobiont. The relatively high level of the proteins still persisting is, however, also in accordance with the nding that Azolla cyanobionts removed from the symbiosis x carbon at a rate comparable with that of free-living cyanobacteria (Ray et al., 1979; Kaplan and Peters, 1988). As the cyanobiont is not light limited (Peters and Meeks, 1989), some symbiosisspecic mechanism appears to regulate both the synthesis and activity of photosynthetic components. Fructose-1,6-bisphosphatase (FBP) (az15), fructose/ tagatose bisphosphate aldolase (FBA) (az31az32), transketolase (az09), and phosphoketolase (az08) are other enzymes functioning in CO2 xation. However, since they also participate in catabolic processes, the interpretation of the abundance of these proteins is discussed below.
Carbohydrate uptake and metabolism

With a reduced photosynthetic activity in the cyanobiont, energy requirements will have to be met by carbohydrates supplied by the host plant. Indeed, a phosphotransferase system fructose-specic IIC component (az04) was among the most highly expressed proteins in the 2D gels of the mixotrophically grown cyanobiont of Azolla, and was ;4 times more abundant than its homologue in free-living Nostoc PCC 73102 (Fig. 1A, B; Table 1). It has been suggested that the carbohydrate transferred is mainly sucrose (Kaplan and Peters, 1988), and the transport system identied here is potentially therefore a hexose transporter. 6-Phosphogluconate dehydrogenase (6PGD) (az28) was also strongly up-regulated in the cyanobiont. The same is the case in the Nostoc cyanobiont of G. manicata (Ekman et al., 2006). The activity of this enzyme is specic for the oxidative pentose phosphate (OPP) pathway, which is the major route of carbon catabolism in cyanobacteria and also in providing reductant to nitrogenase (Summers et al., 1995). The up-regulation of 6PGD therefore implies a high demand for reductant by N2 xation, and by the mixotrophic growth mode of the cyanobiont. Given this, the up-regulation of the enzyme glucose-6-phosphate isomerase (GPI) (az24) is probably less associated with the glycolytic conversion of glucose-6-phosphate (G6P) into fructose-6-phosphate (F6P) than with the opposite reaction, conversion of F6P into G6P, the latter subsequently oxidized in the OPP pathway. Furthermore, this enzyme is needed if host-derived sugars, supplied as sucrose or fructose, are to be catabolized via the OPP pathway. That the strong up-regulation of GPI seen is related to carbon ow through the OPP pathway is supported by the fact that when free-living Nostoc PCC 73102 was supplied with fructose, the up-regulation of this protein was less pronounced (;30%, M Ekman et al.,

unpublished results). In free-living cyanobacteria, a higher proportion of the fructose is catabolized via glycolysis and the incomplete tricarboxylic acid (TCA) cycle to meet the higher amount of biosynthetic precursors needed to achieve the higher rates of nitrogen assimilation (Meeks and Elhai, 2002). On the other hand, the level of aconitase (az48) was not signicantly changed in the cyanobiont, which may indicate that glycolysis and the TCA cycle are active to some degree in the cyanobiont. The abundance of this protein was still only a fraction of that of most other proteins identied in the Calvin cycle and the OPP pathway in both organisms (Fig. 1), once again stressing the limited role of the TCA cycle in cyanobacteria. The abundance of FBA, transketolase, and phosphoketolase in the cyanobiont was not signicantly different from the levels in Nostoc PCC 73102, while that of FBP was 30% lower. All four are necessary for both the cyclic OPP pathway and CO2 xation via the Calvin cycle. Given the symbiotic down-regulation of the Calvin cycle, the fact that these enzymes are still expressed at a high level further emphasizes the role of the OPP pathway in symbiosis. However, as regulation of metabolic pathways is often reected in activities rather than enzyme levels, it is also possible that the protein levels would be high even without a highly active OPP pathway. A different FBP isozyme was found to be up-regulated under conditions of nitrogen limitation in N. punctiforme, and was suggested to function mainly in the cyclic OPP pathway of heterocysts (Summers and Meeks, 1996). The reduced levels of the FBP identied here may therefore potentially be associated with a principal function in the Calvin cycle of vegetative cells.
Stress responses

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

Some of the most abundant proteins in the 2D gels of the Azolla cyanobiont and Nostoc PCC 73102 were proteins with functions related to protein assembly, modication, and degradation (chaperones; az03, az05, and az10az11), and stress-related proteins, such as superoxide dismutase (az42) and peroxiredoxins (az40 and az41) (Table 1). Some of these have different abundances in the two organisms, possibly reecting the radically different growth conditions experienced in symbiosis. It is also known that specic peroxiredoxins are expressed in rhizobia living in symbiosis with legumes but not when free-living (Dombrecht et al., 2005). Furthermore, the intracellular localization of superoxide dismutase in heterocysts of the Azolla cyanobiont suggested a role in protecting nitrogenase from superoxide radicals generated via respiration (Canini et al., 1992a). Besides experiencing stress when entering and accommodating to a symbiotic lifestyle in a plant, the high heterocyst differentiation and nitrogen xation rate elicited is per se a stress response elicited by N deprivation. Stress proteins were

Azolla cyanobiont proteome 1031

likewise up-regulated in the cyanobiont of G. manicata (Ekman et al. 2006). Down-regulation of polynucleotide nucleotidyltransferase (az06), which is involved in RNA degradation, suggests a reduced RNA turnover rate possibly associated with the lowered cell division rates in cyanobionts in plant symbioses. Also, a translation elongation factor (az18), even if highly expressed in both organisms, occurred in lower relative quantities in the cyanobiont.
Nitrogenase modication

In addition to protein identication, mass spectrometry has also been widely used for identifying post-translational modications of proteins (Larsen and Roepstorff, 2000; Reinders et al., 2004) and, as discussed, a large proportion of the Azolla cyanobiont NifH protein appeared to be modied. The mechanism behind the modication of nitrogenase in cyanobacteria is still unknown and is not due to ADP-ribosylation in the heterocystous cyanobacterium A. variabilis (Durner et al., 1994) as in several other N2-xing bacteria. In the unicellular cyanobacterium Gleothece, NifH was proposed to be modied by several palmitoylations, or palmitoylation in combination with other modications (Gallon et al., 2000), but the reported size difference of >2000 Da between the unmodied and modied protein is considerably larger than that identied here in the cyanobiont of Azolla. Attempts were therefore made to identify the nitrogenase modication by comparing the peptide mass spectra of trypsin-digested unmodied and modied NifH of the Azolla cyanobiont. Theoretically, the peptide carrying the modication would have a higher mass in the modied protein compared with the unmodied protein, and the mass difference would allow identication as well as location of the modication. Indeed, one peptide peak was consistently missing from the spectra of the modied protein (Fig. 2A, B), potentially corresponding to the peptide being modied. However, no new peptide of higher mass carrying the modication was observed. The identity of the disappearing peptide was conrmed by analysing the modied and non-modied protein spectra of NifH of Nostoc PCC 73102, in which a peptide of the same mass disappeared (Fig. 2C, D). The disappearing peptide corresponded to the amino acid sequence CVESGGPEPGVGCAGR of NifH. When Blasting the NifH sequence of the Nostoc PCC 73102 against the NCBInr database, it is clear that this is the most conserved sequence of the protein, being 100% identical in most nitrogenase proteins sequenced. Furthermore, the ADP-ribosylation of Rhodospirillum has been shown to be located on Arg101 (Pope et al., 1985), which corresponds to the arginine at the end of this sequence. Trypsin cleaves at arginine, and a modication of this amino acid would block this cleavage and result in a peptide of ;6000 Da, which is too large to be analysed

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

Fig. 2. MALDI-TOF spectra of trypsin-digested NifH of (A) the Azolla cyanobiont and (C) Nostoc PCC 73102; and the modied NifH of (B) the Azolla cyanobiont and (D) Nostoc PCC 73102. The peptides of mass 1588.7, corresponding to the sequence CVESGGPEPGVGCAGR, are missing from the two modied proteins, as indicated by arrows.

in the reective mode of MALDI-TOF. As a consequence, a second enzyme was used, GluC, which cleaves at glutamic acid (when in bicarbonate buffer). However, again the peptide corresponding to the same part of the protein (amino acid sequence SGGPEPGVGCAGRGIITAINFLEE) in the Azolla cyanobiont and in Nostoc PCC 73102 disappeared in the modied protein, but no other peptide appeared (Fig 3). When the protein was cleaved with both trypsin and GluC, once again a peptide corresponding to the same part of the protein (amino acid sequence SGGPEPGVGCAGR) disappeared (Fig. 4; Table 2). There may be several explanations for the absence of the modied peptide in the mass spectra, even when using GluC, or both enzymes. It has been shown that some modications may interfere with the ionization step in MALDI-TOF MS or, alternatively, the modied protein was not eluted from the gel after in-gel digestion. Attempts were therefore made to elute the whole protein from the gel in order to be able to cleave the protein in solution. This would also allow for determination of the exact difference in mass between the modied and the non-modied protein, thereby identifying a candidate modication whose identity could be conrmed by chemically de-modifying the protein. However, the elution was not successful without using SDS, which is incompatible with MALDI-TOF MS analysis. Currently, optimization of this procedure is under way. However, it

1032 Ekman et al.

is shown for the rst time that the NifH protein modication in cyanobacteria is localized within the 13 amino acid sequence SGGPEPGVGCAGR, which in turn is positioned close to the active site of the enzyme (Georgiadis et al., 1992). Based on migration in the 2D gels, the modication has a mass of 300400 Da (Table 3), making it too small for ADP-ribosylation (541 Da). In bacteria, palmitoylation (238 Da) is usually observed only on lysine and on N-terminal cysteine after cleavage of signal peptide (FindMod tool, Expasy Proteomics Server; available online from the Swiss Institute of Bioinformatics, Basel, Switzerland), making this modication also less likely. It is clear that the NifH modication of the Azolla cyanobiont is located within the part of the protein that is ADP-ribosylated in other bacteria and that it has a mass of 300400 Da, but the identity of the modication remains to be determined.

Conclusions The versatility of the proteomic approach was evident, as it was possible to identify several adaptations to symbiosis in the Azolla cyanobiont, to perform a basic analysis of the taxonomic afliation of the Azolla cyanobiont, and to clarify important aspects of the NifH modication in cyanobacteria. First, the adaptations found reect a metabolism in the cyanobiont largely devoted to production of xed nitrogen (for the benet of the plant) and include the differential expression of a number of key proteins previously not known to be affected in cyanobacterial symbioses. This may in turn be a consequence of the long co-evolution between the plant and the cyanobiont. Secondly, the information obtained with regard to the taxonomic afliation of the Azolla cyanobiont was in line with previous ndings (Svenning et al., 2005), suggesting that proteomics-based taxonomic analyses may become
Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

Fig. 3. MALDI-TOF spectra of GluC-digested NifH of (A) the Azolla cyanobiont and (C) Nostoc PCC 73102; and modied NfH of (B) the Azolla cyanobiont and (D) Nostoc PCC 73102. The peptides of mass 2501.18, corresponding to the sequence SGGPEPGVGCAGRGIITAINFLEE, are missing from the two modied proteins, as indicated by arrows.

Fig. 4. MALDI-TOF spectra of GluCtrypsin double-digested NifH of (A) the Azolla cyanobiont and (C) Nostoc PCC 73102; and modied NifH of (B) the Azolla cyanobiont and (D) Nostoc PCC 73102. The peptides of mass 1200.563, corresponding to the sequence SGGPEPGVGCAGR, are missing from the two modied proteins, indicated by arrows in (B) and (D).

Table 2. Peptides obtained after cleavage of NifH of the cyanobiont of Azolla with trypsin, GluC, or both in combination
Enzyme Trypsin GluC (bicarbonate) Trypsin+GluC Cleaves at C-terminal side of K or R C-terminal side of E C-terminal side of K or R and also E Modied peptide sequence CVESGGPEPGVGCAGR SGGPEPGVGCAGRGIITAINFLEE 1. SGGPEPGVGCAGR 2. GIITAINFLEE Modied peptide mass 1588.69 2401.18 1200.54 1219.66

Azolla cyanobiont proteome 1033 Table 3. Mass differences between the two NifH proteins of Nostoc PCC 73102 estimated by PDQuest software from four different 2D gels
Gel no. 1 2 3 4 Mass NifH (kDa) 33.6 32.2 32.7 33.1 Mass NifH modied (kDa) 34.0 32.6 33.1 34.4 Mass difference (kDa) 0.4 0.4 0.4 0.3

a useful complement to traditional taxonomic methods. Thirdly, the rst molecular data set was identied on the nature of the NifH modication in cyanobacteria, a molecule of 300400 Da located at a 13 amino acid sequence positioned close to the active site of nitrogenase. It is also clear that although an organism with a non-sequenced genome was analysed, the present approach generated highly valuable information.

Acknowledgements
Funding from the Swedish Research Council (to BB) and the European Science Foundation CYANOFIX Programme (to ME) is gratefully acknowledged.

References
Adams DG. 2000. Heterocyst formation in cyanobacteria. Current Opinion in Microbiology 3, 618624. Baker JA, Entsch B, McKay DB. 2005. The cyanobiont in an Azolla fern is neither Anabaena nor Nostoc. FEMS Microbiology Letters 229, 4347. Bergman B, Lindblad P, Pettersson A, Renstrom E, Tiberg E. 1985. Immunogold localization of glutamine synthetase in a nitrogen xing cyanobacterium, Anabaena cylindrica. Planta 166, 329334. Bergman B, Rasmussen U, Rai AN. 2007. Cyanobacterial associations. In: Elmerich C, Newton WE, eds. Associative and endophytic nitrogen-xing bacteria and cyanobacterial associations. Dordrecht, The Netherlands: Kluwer Academic Publishers, Vol. 5, 257301. Braun-Howland EB, Lindblad P, Nierzwickibauer SA, Bergman B. 1988. Dinitrogen reductase in the cyanobacterial cyanobionts of 3 Azolla specieslocalization and sequence of appearance during heterocyst differentiation. Planta 176, 319322. Braun-Howland EB, Nierzwicki-Bauer SA. 1990. Azolla Anabaena symbiosis: biochemistry, physiology, ultrastructure, and molecular biology. In: Rai AN, ed. CRS handbook of symbiotic cyanobacteria. Boca Raton, Florida: CRC Press, 65117. Canini A, Bergman B, Civitareale P, Rotilio G, Grilli Caiola M. 1992a. Localization of iron-superoxide dismutase in the cyanobiont of Azolla liculoides. Lam. Protoplasma 169, 18. Canini A, Grilli Caiola M, Civitareale P, Galiazzo F. 1992b. Superoxide dismutase in symbiotic, free-living and wild Anabaena and Nostoc (Nostocales, Cyanophyta). Phycologia 31, 225230. Canini A, Grilli Caiola M, Mascini M. 1990. Ammonium content, nitrogenase activity and heterocyst frequency within the leaf cavities of Azolla liculoides Lam. FEMS Microbiology Letters 71, 205210.

Chang WWP, Huang L, Shen M, Webster C, Burlingame AL, Roberts JKM. 2000. Patterns of protein synthesis and tolerance of anoxia in root tips of maize seedlings acclimated to a lowoxygen environment, and identication of proteins by mass spectrometry. Plant Physiology 122, 295317. Dombrecht B, Heusdens C, Beullens S, Verreth C, Mulkers E, Proost P, Vanderleyden J, Michiel J. 2005. Defence of Rhizobium etli bacteroids against oxidative stress involves a complexly regulated atypical 2-cys peroxiredoxin. Molecular Microbiology 55, 12071221. Durner J, Bohm J, Hilz H, Boger P. 1994. Posttranslational modication of nitrogenase. Differences between the purple bacterium Rhodospirillum rubrum and the cyanobacterium Anabaena variabilis. European Journal of Biochemistry 220, 125130. Ekman M, Tollback P, Klint J, Bergman B. 2006. Protein expression proles in an endosymbiotic cyanobacterium revealed by a proteomic approach. Molecular PlantMicrobe Interactions 19, 12511261. Fay P. 1992. Oxygen relations of nitrogen xation in cyanobacteria. Microbiology Reviews 56, 340373. Flores E, Herrero A. 2005. Nitrogen assimilation and nitrogen control in cyanobacteria. Biochemical Society Transactions 33, 164167. Gallon JR, Cheng J, Dougherty LJ, Gallon VA, Hilz H, Pederson DM, Richards HM, Ruggeberg S, Smith CJ. 2000. A novel covalent modication of nitrogenase in a cyanobacterium. FEBS Letters 468, 231233. Georgiadis MM, Komiya H, Chakrabarti P, Woo D, Kornuc JJ, Rees DC. 1992. Crystallographic structure of the nitrogenase iron protein from Azotobacter vinelandii. Science 257, 16531659. Gharahdaghi F, Weinberg CR, Meagher DA, Imai BS, Mische SM. 1999. Mass spectrometric identication of proteins from silver-stained polyacrylamide gel: a method for the removal of silver ions to enhance sensitivity. Electrophoresis 20, 601605. Kaplan D, Peters GA. 1988. Interaction of carbon metabolism in the AzollaAnabaena symbiosis. Symbiosis 6, 5368. Kim H-J, Lee D-Y, Lee D-H, Park Y-C, Kweon D-H, Ryu Y-W, Seo J-H. 2004. Strategic proteome analysis of Candida magnoliae with an unsequenced genome. Proteomics 4, 35883599. Larsen MR, Roepstorff P. 2000. Mass spectrometric identication of proteins and characterization of their post-translational modications in proteome analysis. Journal of Analytical Chemistry 366, 677690. Lechno-Yossef S, Nierzwicki-Bauer SA. 2002. AzollaAnabaena symbiosis. In: Rai AN, Bergman B, Rasmussen U, eds. Cyanobacteria in symbiosis. Dordrecht, The Netherlands: Kluwer Academic Publishers, 153178. Lee KY, Joseph CM, Meeks JC. 1988. Glutamine synthetase specic activity and protein concentration in symbiotic Anabaena associated with Azolla caroliniana. Antonie Van Leeuwenhoek 54, 345355. Liska AJ, Shevchenko A. 2003. Expanding the organismal scope of proteomics: cross-species protein identication by mass spectrometry and its implications. Proteomics 3, 1928. Liska AJ, Popov AV, Sunyaev S, Coughlin P, Habermann B, Shevchenko A, Bork P, Karsenti E, Shevchenko A. 2004. Homology-based functional proteomics by mass spectrometry: application to the Xenopus microtubule-associated proteome. Proteomics 4, 27072721. Meeks JC. 1988. Symbiotic associations. Methods in Enzymology 167, 113121. Meeks JC, Elhai J. 2002. Regulation of cellular differentiation in lamentous cyanobacteria in free-living and plant-associated symbiotic growth states. Microbiology and Molecular Biology Reviews 66, 94121.

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010

1034 Ekman et al.

Meeks JC, Joseph CM, Haselkorn R. 1988. Organization of the nif genes in cyanobacteria in symbiotic association with Azolla and Anthoceros. Archives of Microbiology 15, 6171. Nierzwicki-Bauer SA, Aulnger H, Braun-Howland EB. 1989. Ultrastructural characterization of an inner envelope that connes Azolla endosymbionts to the leaf cavity periphery. Canadian Journal of Botany 67, 27112719. Nierzwicki-Bauer SA, Haselkorn R. 1986. Differences in mRNA levels in Anabaena living freely and in symbiotic association with Azolla. EMBO Journal 5, 2935. Pappin DJ, Hojrup P, Bleasby AJ. 1993. Rapid identication of proteins by peptide-mass ngerprinting. Current Biology 3, 327332. Perkins DN, Pappin DJ, Creasy DM, Cottrell JS. 1999. Probability-based protein identication by searching sequence databases using mass spectrometry data. Electrophoresis 20, 35513567. Perkins SK, Peters GA. 1993. The AzollaAnabaena symbiosis: endophyte continuity in the Azolla life cycle is facilitated by epidermal trichomes. I. Partitioning of the endophytic Anabaena into developing sporocarps. New Phytologist 123, 5364. Peters GA, Meeks JC. 1989. The AzollaAnabaena symbiosis; basic biology. Annual Review of Plant Physiology 40, 193210. Pope MR, Murrel SA, Ludden PW. 1985. Covalent modication of the iron protein of nitrogenase from Rhodospirillum rubrum by adenosine diphosphoribosylation of a specic arginine residue. Proceedings of the National Academy of Sciences, USA 82, 31733177. Rai AN, Bergman B, Rasmussen U, eds. 2002. Cyanobacteria in symbiosis. Dordrecht, The Netherlands: Kluwer Academic Publishers. Rai AN, Soderback E, Bergman B. 2000. Cyanobacteriumplant symbioses. New Phytologist 147, 449481. Ran L, Huang F, Ekman M, Klint J, Bergman B. 2007. Proteomic analyses of the photoauto- and diazotrophically grown cyanobacterium Nostoc sp. PCC 73102. Microbiology 153, 608618. Ray TB, Mayne BC, Peters GA, Toia Jr RE. 1979. Azolla Anabaena relationship. VIII. Photosynthetic characterization of the association and individual partners. Plant Physiology 64, 791795. Razquin P, Fillat MF, Scmitz S, Stricker O, Bhome H, GomezMoreno C, Peleato ML. 1996. Expression of ferredoxin:NADP+ reductase in heterocysts from Anabaena sp. Biochemical Journal 316, 157160. Reinders J, Lewandrowski U, Moebius L, Wagner Y, Sickmann A. 2004. Challenges in mass spectrometry-based proteomics. Proteomics 4, 36863703. Rippka R, Deruelles J, Waterbury JB, Herdman M, Steiner RY. 1979. Genetic assignments, strain histories and properties of pure cultures of cyanobacteria. Journal of General Microbiology 111, 161. Schmetterer G. 1994. Cyanobacterial respiration. In: Bryant DA, ed. The molecular biology of cyanobacteria. Dordrecht, The Netherlands: Kluwer Academic Publishers, 409435. Shevchenko A, Sunyaev S, Loboda A, Shevchenko A, Bork P, Ens W, Standing KG. 2001. Charting the proteomes of organisms with unsequenced genomes by MALDI-quadrupole time-of-ight mass spectrometry and Blast homology searching. Analytical Chemistry 73, 19171926. Strasburger E. 1873. Uber Azolla. Jena: Herman Dabis Verlag. Summers ML, Meeks JC. 1996. Transcriptional regulation of zwf, encoding glucose-6-phosphate dehydrogenase, from the cyanobacterium Nostoc punctiforme strain ATCC 29133. Molecular Microbiology 22, 473480. Summers ML, Wallis JG, Campbell EL, Meeks JC. 1995. Genetic evidence of a major role for glucose-6-phosphate dehydrogenase in nitrogen xation and dark growth of the cyanobacterium Nostoc sp. strain ATCC 2913. Journal of Bacteriology 177, 61846194. Svenning MM, Eriksson T, Rasmussen U. 2005. Phylogeny of symbiotic cyanobacteria within the genus Nostoc based on 16S rDNA sequence analyses. Archives of Microbiology 183, 1926. Tang LF, Watanabe I, Liu CC. 1990. Limited multiplication of symbiotic cyanobacteria of Azolla spp. on articial media. Applied and Environmental Microbiology 56, 36233626. Valladares A, Muro-Pastor AM, Fillat MF, Herrero A, Flores E. 1999. Constitutive and nitrogen-regulated promoters of the petH gene encoding ferredoxin:NADP+ reductase in the heterocyst-forming cyanobacterium Anabaena sp. FEBS Letters 449, 159164. van Hove C, Lejeune A. 2002. Applied aspects of Azolla Anabaena symbiosis. In: Rai AN, Bergman B, Rasmussen U, eds. Cyanobacteria in symbiosis. Dordrecht, The Netherlands: Kluwer Academic Publishers, 117135. Vessey JK, Pawlowski K, Bergman B. 2005. Root-based N2xing symbioses: legumes, actinorhizal plants. Parasponia sp. and cycads. Plant and Soil 274, 5178. Watanabe I. 1982. AzollaAnabaena symbiosisits physiology and use in tropical agriculture. In: Dommergues YR, Diems HD, eds. Microbiology of tropical soils and plant productivity. The Hague, The Netherlands: Martinus Nijhoff, 169. Wilkins R, Williams L. 1997. Cross species protein identication using amino acid composition, peptide mass ngerprinting, isoelectric point and molecular mass: a theoretical evaluation. Journal of Theoretical Biology 186, 715. Zhang H, Whitelegge JP, Cramer WA. 2001. Ferredoxin:NADP+ oxidoreductase is a subunit of the chloroplast cytochrome b6f complex. Journal of Biological Chemistry 276, 3815938165. Zheng WW, Lin YH, Lin YH, Lu PJ, Liu CC. 1990. Electron microscopic observation of symbiotic relationship between Azolla and Anabaena during the megaspore germination and sporeling development of Azolla. Acta Botanica Sinica 32, 514520. Zheng WW, Nilsson M, Bergman B, Rasmussen U. 1999. Genetic diversity and classication of cyanobacteria in different Azolla species by the use of PCR ngerprinting. Theoretical and Applied Genetics 99, 11871193.

Downloaded from jxb.oxfordjournals.org by guest on October 8, 2010