Introduction

Symbiosis is widespread in the microbial world (Moya et al., 2008). Symbiotic interactions can fluctuate between parasitism and mutualism through time (Toft and Andersson, 2010). These interactions are a source of evolutionary innovation through genetic rearrangements that give rise to metabolic capabilities and emergence of syntrophy to exploit resources, stabilizing in a mutualistic relationship (Margulis and Fester, 1991; Moya et al., 2008; Moya and Peretó, 2011).

The phototrophic consortium ‘Chlorochromatium aggregatum’ is a model of symbiotic interactions involving prokaryotes. It was isolated from Lake Dagow in Germany (Fröstl and Overmann, 1998) and is formed by green-sulfur bacteria Chlorobium chlorochromatii (epibionts) that surround a central, colorless and motile β-proteobacterium, phylogenetically related to Comamonadaceae (Kanzler et al., 2005). Chl. chlorochromatii CaD3 belong to the Chlorobia phylum. They are Gram-negative, non-motile, anaerobic photosynthetic and diazotrophic bacteria. These bacteria perform anoxic photosynthesis in the chlorosoma, which contains a photosystem I (PSI) reaction center. Their major photosynthetic pigment is bacteriochlorophyll c, and sulfide is used as electron donor. CO2 is fixed via the reverse tricarboxylic acid cycle (rTCA) (Vogl et al., 2006). In natural conditions they are exclusively found in symbiosis although they can be cultured without their symbiont, which is not the case for the obligate-symbiotic β-proteobacterium (Müller and Overmann, 2011). Previous studies have identified genes that code for virulence factors in Chl. chlorochromatii CaD3 that seem to be relevant in the symbiosis, one of them seems to have been acquired by horizontal gene transfer (Vogl et al., 2008).

A recent study compared the proteome and gene expression of symbiotic and free-living states in Chl. chlorochromatii CaD3 (Wenter et al., 2010). It showed that ∼350 genes of Chl. chlorochromatii CaD3 have differential expression patterns between both states (Wenter et al., 2010). A significant fraction of these correspond to genes involved in nitrogen and amino acid metabolism. Among these there were genes coding for a nitrogen regulator protein PII, a gene coding a glutamine synthetase (GS) enzyme and the nifH gene coding for nitrogenase reductase. These genes seem to be having a regulatory role in nitrogen assimilation between symbiotic and free-living states (Brown and Herbert, 1977; Magasanik, 1982; Forchhamer, 2004; Texeira et al., 2010). In this study, we performed the metabolic network reconstruction of Chl. chlorochromatii CaD3 and an elementary flux mode (EFM) analysis (Schuster et al., 1999; Jevremovic et al., 2011) of the reconstructed amino-acid metabolism sub-network, to better understand the metabolic shift between symbiotic and free-living states.

Materials and methods

Genomes used

The following genomes were downloaded from the Genbank database: Chlorobium chlorochromatii CaD3 (NC_007514.1), Chlorobium tepidum TLS (NC_002932.3), Chlorobium phaeobacteroides BS1 (NC_010831.1), Chlorobium phaeobacteroides DSM 266 (NC_008639.1), Chlorobium luteolum DSM 273 (NC_007512.1), Chlorobium phaeovibrioides DSM 265 (NC_009337.1), Chlorobium limicola DSM 245 (NC_010803.1), Pelodictyon phaeoclathratiforme BU-1 (NC_011060.1), Prosthecochloris aestuarii DSM 271 (NC_011059.1), Chloroherpeton thalassium ATCC 35110 (NC_011026.1), Chlorobaculum parvum NCIB 8327 (NC_011027.1) and Escherichia coli K-12 (MG1655) (NC_000913.2).

Reconstruction of the metabolic capabilities of Chl. chlorochromatii CaD3

We first reconstructed the metabolism of Chl. chlorochromatii CaD3 with Pathway-Tools software (Karp et al., 2010). The Pathway-Tools software inferred all metabolic pathways directly from the Genbank file. Thereby, we obtained a Pathway/Genome Database (PGDB) to be manually curated. As C. tepidum is the most studied species of chlorobia, we also performed the automatic reconstruction of its metabolism with Pathway-Tools software from its annotated genome downloaded from Genbank database (Eisen et al., 2002). The automatic reconstruction was curated with overall experimental information, and metabolic function was assigned only if it could be confirmed in the literature. Then, the metabolic reconstruction of C. tepidum was used to cure the metabolic reconstruction of Chl. chlorochromatii CaD3 (see below).

To improve and curate the initial metabolic reconstruction we performed a search for orthologous proteins between Chl. chlorochromatii CaD3, C. tepidum and the other species of Chlorobia listed under ‘Genomes used’. We also searched for orthologous proteins between the Chlorobia and E. coli and used BranchClust software to find orthologs between those genomes (Poptsova and Gogarten, 2007). We used the E. coli genome assuming that it is the best curated bacterial genome and is available through EcoCyc database (Keseler et al., 2011). The genomes of the others species of Chlorobia were added to the analysis to improve the performance of BranchClust. For the 12 genomes included in this study, we asked BranchClust to consider as orthologs all monophyletic groups of sequences having at least eight different species represented. For those peptide sequences of Chl. chlorochromatii CaD3 that had no orthologs, we looked for conserved protein domains against the Pfam database (Finn et al., 2010) with HMMER3 software (Johnson et al., 2010) by using an E-value cutoff of 0.01. Based on the orthologous search with BranchClust and domain identification with Pfam, we assigned 382 enzymatic reactions to the metabolism of Chl. chlorochromatii CaD3 (Supplementary Material 1). These reactions were obtained from KEGG (Kanehisa et al., 2012) and BRENDA databases (in their WWW available versions in 2011–2012) (Scheer et al., 2011). Finally the reactions were integrated within YANA software (Schwarz et al., 2005).

EFM analysis of symbiotic and free-living states

We performed an EFM analysis of the amino-acid metabolism in the symbiotic and free-living states of Chl. chlorochromatii CaD3. The EFM algorithm decomposes the metabolic network in EFMs, where each EFM is a minimal set of enzymes operating in steady state with all reactions proceeding in the direction dictated by its thermodynamic restrictions (for theoretical details, see Schuster et al., 1999 and Wagner, 2004) being equivalent to a metabolic pathway that supports the network structure in a specific growth condition. The inferred metabolic network of the amino-acid metabolism consists of 99 enzymatic reactions from the whole reconstruction. These reactions were identified following the metabolic pathway classification of KEGG database. Then, based on the expression profiles published by Wenter et al. (2010), we removed with YANA software 5 and 10 reactions not transcribed in symbiotic and free-living states respectively, thus generating two sub-networks: one with 94 reactions corresponding to the symbiotic state and the other with 89 reactions corresponding to the free-living state (see Supplementary Material 1, 2 and 3).

To identify the EFM of the symbiotic and free-living states, the two sub-networks in METATOOL file format (Schuster et al., 1999; Supplementary Material 1) were analyzed with ElMo-comp-1.0.4 software (Jevremovic et al., 2011). For each sub-network there were two output files: one containing the EFMs found, and the other containing the reactions of each EFM. These output files were analyzed with scripts in python v2.7 to see whether the EFMs in both states (free-living and symbiosis) were the same. We further analyzed the EFMs with ACoM-c software (Pérès et al., 2011). This software groups the EFMs in motifs, which are clusters of EFMs that share at least three reactions. Motifs were visualized with Celldesigner (Funahashi et al., 2007) (Supplementary Material 2). Finally, to prove that the obtained EFM reconstructions were independent of chance, we performed 100 random simulations of the symbiotic and free-living networks as follows: from the whole network of amino-acid metabolism containing 99 reactions, we randomly removed 5 and 10 reactions to simulate symbiotic and free-living states with 94 and 89 reactions each. We repeated this procedure 100 times for each simulated state. The 200 resulting networks were analyzed with ElMo-comp-1.0.4 and the EFMs obtained were compared against each other by a python script. All scripts are available upon request.

Results and discussion

EFM analysis of amino-acid metabolism in free-living and symbiotic states

The automatic metabolic reconstruction with Pathway-Tools software assigned approximately 1000 enzymatic reactions to Chl. chlorochromatii CaD3. From this initial reconstruction, only 382 reactions remained after manual curation (Supplementary Material 1). As mentioned above, several enzymes related to amino acid and nitrogen metabolism change their expression profile between free-living and symbiotic states (Wenter et al., 2010). To better understand the metabolic changes related to both states, we selected those reactions belonging to amino acid metabolism for further EFM analysis. In brief, EFM analysis identifies all unique and non-decomposable steady state flux distributions in a metabolic pathway, which represents the metabolic potential of the network in a specific condition (Klamt and Stelling, 2002). The amino-acid metabolic networks of the free-living and symbiotic states contained 89 and 94 reactions, respectively. These networks overlap by 84 reactions.

Although in EFM analysis, the amount of EFMs normally increases according to the number of reactions (Klamt and Stelling, 2002), in this study the EFM analysis of amino-acid metabolism showed that the free-living state network containing 89 reactions has more EFMs (7069 EFMs), than the network in symbiotic state with 94 reactions (5075 EFMs). This is easily explainable because five of the reactions of the symbiotic state are consecutive in two different pathways (three in the biosynthesis of isoleucine and two in the biosynthesis of leucine). In contrast, in the simulated networks, the number of EFMs increased with the number of reactions. The random networks simulating the number of enzymes in the free-living state had a mean of 7500 EFMs, which approximated the 7069 EFMs obtained for the free-living state (Figure 1). In contrast, the random sampling simulating the symbiotic state had a mean of 12 446 EFMs, approximately two standard deviations (σ=4029.2) higher than the EFMs obtained in the symbiotic network (5075 EFMs). These results suggest that there is a low probability of obtaining this number of EFMs in a random configuration of 94 reactions and that the symbiotic state is clearly constraining the structure of the metabolic network of amino acids in Chl. Chlorochromatii CaD3.

Figure 1
figure 1

Histograms showing the distribution of EFMs in the randomly resampled networks of the symbiotic (a) and free-living (b) states of Chl. chlorochromatii CaD3. Asterisks indicate the bin on which the amount of EFMs obtained in symbiosis (a) and free-living (b) states would be located.

Full size image

Experimental data showed that GS was present only in symbiosis (Wenter et al., 2010). Accordingly, the motifs obtained with ACoM-c software showed that the EFMs were clearly different in hotspots of NH3 assimilation between symbiotic and free-living states (Figure 2). In free-living state 5805 EFMs (82%) contained a reaction performed by alanine dehydrogenase (AlaDH) EC:1.4.1.1:

Figure 2
figure 2

Distribution of EFMs in the metabolism of amino acids of Chl. chlorocromatii CaD3 in free-living and symbiotic states. Reactions that contain 1-0 EFMs were not included in this plot. The reactions represented by bars with circles are involved in EFMs of the GS and by bars with stars are involved in EFMs of the AlaDH pathways.

Full size image

Whereas in symbiosis, 3438 EFMs (68%) contained a reaction performed by GS EC:6.3.1.2:

The reaction performed by AlaDH in the symbiotic state was only present in 3258 EFMs (64%). Random simulations of the free-living and symbiotic states gave very different percentages of the number of EFMs containing the reaction performed by AlaDH (Table 1).

Table 1 Number of times the reactions performed by GS and AlaDH were found in free-living and symbiotic states
Full size table

Is nitrogen assimilated via AlaDH in free-living state?

Previous work has shown that N2 fixed to ammonia by the green-sulfur bacteria is assimilated via GS/GOGAT and the GDH (glutamine dehydrogenase) pathways (Brown and Herbert, 1977; Wahlund and Madigan, 1993). Both mechanisms are regulated by nitrogen availability. The GS/GOGAT pathway is a high-affinity mechanism that is active when there is low availability of nitrogen, and in the case of diazotrophic bacteria, when they are fixing nitrogen (Rudnick et al., 1997; Schulz et al., 2001). On the contrary, the assimilation of ammonia by GDH is a mechanism of low affinity that is active when there is high availability of nitrogen (Brown and Herbert, 1977; Rudnick et al., 1997).

Unlike the other green-sulfur bacteria species, Chl. chlorochromatii CaD3 lacks GDH, and expresses GS only when growing in symbiotic state. How does Chl. chlorochromatii CaD3 assimilate nitrogen when growing in a free-living condition? One possibility is by the use of alanine dehydrogenase AlaDH (gene Cag_1878). The alanine dehydrogenase pathway is another N assimilation mechanism. This enzyme has aminase–deaminase activity and is active in conditions of high nitrogen availability. Although this role for AlaDH has not been identified in green-sulfur bacteria, it has been observed in diazotrophic-photosynthetic bacteria that lack GDH such as Rhodopseudomonas capsulata (Johansson and Gest, 1976), R . acidophila (Herbert et al., 1978), Anabaena cylindrica (Rowell and Stewart, 1975), and in non-photosynthetic diazothophic bacteria such as Bradyrhizobium sp. (Allaway et al., 2000). The increase of EFMs containing AlaDH in the metabolic network corresponding to the free-living state supports this possibility.

The EFMs analysis suggests that in symbiosis, ammonia is assimilated via the GS pathway (Figure 2). It is known that GS is a hotspot in nitrogen metabolism, distributing nitrogen in glutamine form to the whole metabolic network (Tyler, 1978; Reitzer, 2003). Likewise, there is a significant ratio of EFMs that goes through the AlaDH pathway. However, in free-living condition, GS is not present and AlaDH is present in a larger number of EFMs (Figure 2). These shifts in the mechanisms of nitrogen assimilation suggest that the availability of nitrogen in Chl. chlorocromatii CaD3 changes between symbiosis and free-living states. Wenter et al. (2010) and Overmann (2010) postulate that this bacterium in symbiosis is in a condition of low nitrogen availability, where the assimilation of ammonia becomes fundamental. The shift in the metabolic networks observed in our analysis supports this hypothesis. In symbiosis, the chlorobia seem to be in a condition of low nitrogen availability and the high-affinity mechanism is active; while in free-living state, where ammonia is available in high concentrations, the chlorobia use the low-affinity mechanism (Figure 3).

Figure 3
figure 3

Mechanism of nitrogen assimilation of Chl. chlorochromatii CaD3. The GS/GOGAT and GDH pathways (modified from the study by Yan (2007)) characteristic of the green-sulfur bacteria are shown in (a); GS/GOGAT (black solid arrow) is assimilating nitrogen in symbiosis, however a gene coding for GDH (gray dashed arrow) is not present in Chl. Chlorochromatii CaD3, instead in (b) the AlaDH (red solid arrow) is assimilating nitrogen in free-living state.

Full size image

Does 2-oxoglutarate mediate N2 fixation in ‘C. aggregatum’?

Several observations indicate that highly specific regulation mechanisms exist between Chl. chlorochromatii CaD3 and the central β-proteobacterium in ‘C. aggregatum’ (Müller and Overmann, 2011). For instance, in ‘Pelochromatium roseum’, a closely related consortium, incorporation of 2-oxoglutarate by the central β-proteobacterium seems to be controlled by the physiological state of the Chlorobia (Glaeser and Overmann, 2003b). This is because incorporation of 2-oxoglutarate is strictly dependent on the simultaneous presence of light and sulfide, which is used by the epibiont as an electron donor in photosynthesis. Besides, addition of 2-oxoglutarate was necessary for the isolation of ‘C. aggregatum’, and this compound is pivotal to culture maintenance where the entire consortium ‘C. aggregatum’ seems to be unable to grow photolithoautotrophically or chemotrophically (Fröstl and Overmann, 1998). However, the nature of the molecules responsible for communication between both bacterial species is still unknown.

The PII protein, a regulatory protein of nitrogen metabolism, is exclusively expressed in symbiosis in Chl. chlorochromatii CaD3 (Wenter et al., 2010). All PII-signaling proteins known so far, can bind up to three molecules of ATP/ADP and 2-oxoglutarate, thereby sensing the intracellular energy and carbon status (Ninfa and Jiang, 2005; Forchhammer 2008, 2010). PII proteins are also central in regulating processes related to nitrogen assimilation (Forchhammer 2010). In E. coli, PII signal protein GlnB is uridylynated (GlnB-UMP) by GlnD. This modification allows the sensing of the cellular nitrogen status directly by GlnD, assessed as the glutamine concentration (Forchhammer, 2007). Also in E. coli, PII protein GlnB controls the expression of nitrogen-regulated genes through the histidine kinase nitrogen regulation protein B (NtrB) and the activity of GS through GS adenylyltransferase GlnE (Forchhammer 2008). The result is that high concentration of 2-oxoglutarate stimulates ammonia assimilation through GS (Ninfa and Jiang, 2005).

In cyanobacteria and methanogenic Archaea that have an incomplete TCA, because of the lack of 2-oxoglutarate dehydrogenase, the concentration of 2-oxoglutarate is the signal for PII proteins to sense the carbon/nitrogen status of the cell (Forchhammer 2008; Zhang and Zhao, 2008). This is because the cellular concentration of 2-oxoglutarate depends only on the rate of its formation and its consumption for ammonia assimilation, in most cases through the GS-GOGAT cycle. By contrast, in species with a complete TCA cycle, 2-oxoglutarate is an intermediate and its cellular concentration depends only partially on nitrogen assimilation. Therefore, measuring of 2-oxoglutarate concentration by PII indicates the level of available carbon skeletons for ammonia assimilation, and glutamine is used as a primary nitrogen status signal (Forchhammer 2007, 2008). Chl. chlorochromatii CaD3 has a complete reverse rTCA and 2-oxoglutarate is also an intermediate, and its concentration depends partially on the GS-GOGAT cycle.

As mentioned above, PII regulates GS in bacteria by a post-transcriptional modification via adenylation and deadenylation according to nitrogen availability (Magasanik, 1982; Rudnick et al., 1997; Zhang et al., 2005). In cyanobacteria, PII senses the concentration of 2-oxoglutarate, determining its modified state (phosphorylated) in response to nitrogen availability (Forchhammer et al., 1999; Zhang and Zhao, 2008). In a similar way, in Rhodospirillum rubrum, another phototrophic bacterium, the GS pathway reaches its higher ratio of deadenylation when 2-oxoglutarate is present, as the PII protein and 2-oxoglutarate form a complex to allow the deadenylation of GS. Further, 2-oxoglutarate is detected by the PII protein and stimulates ammonia assimilation and N2 fixation (Forchhammer 2004; Jonsson et al., 2007). In Azotobacter vinelandii, 2-oxoglutarate stimulates the transcription of nifA, which is the master regulator of the nif operon, containing all genes involved in N2 fixation (Dixon and Kahn, 2004).

As reviewed above, the sensing by PII of a high concentration of 2-oxoglutarate stimulates N2 fixation and assimilation through the GS-GOGAT cycle. Could it be possible for the β-proteobacterium to use 2-oxoglutarate to stimulate nitrogen assimilation by the epibiont? This hypothesis is attractive because it links previous knowledge on the mechanisms of regulation of N2 fixation and assimilation, with the observation that proteins involved in nitrogen metabolism, like PII and GS, are only expressed in symbiotic state. It also links the chemotactic behavior of ‘C. aggregatum’ towards 2-oxoglutarate with N2 fixation and assimilation. However, this hypothesis has to overcome some observations:

First, the membranes of phototrophic bacteria are impermeable to 2-oxoglutarate, this allows a fine regulation of this important regulator and effector in the assimilation of ammonia and N2 fixation (Vásquez-Bermúdez et al., 2000; Texeira et al., 2010). In addition, none of the known species of green sulfur bacteria utilize 2-oxoglutarate for growth (Glaeser and Overmann, 2003b). Chl. chlorochromatii CaD3 is not the exception; it cannot assimilate 2-oxoglutarate when free-living (Vogl et al., 2006). This limitation could be overcome if there is a transporter protein for 2-oxoglutarate that only expresses in the symbiotic state. Chl. Chlorochromatii CaD3 codes indeed for a homolog of the 2-oxoglutarate transporter in E. coli. The protein coded by the gene Cag_1339 is expressed only in the symbiotic state (Wenter et al., 2010) and is homologous to five transporter proteins in E. coli, among these the 2-oxoglutarate transporter (gene kgtP). However, its best reciprocal hit in the E. coli genome is another predicted transporter coded by the gene yhjE. Therefore, it is not clear if Cag_1339 could function as a 2-oxoglutarate transporter in Chl. chlorochromatii CaD3.

Secondly, 2-oxoglutarate has to travel from the β-proteobacterium to the epibiont. Ultrastructural characterizations of ‘C. aggregatum’ suggest that there is a common periplasmic space between both bacteria (Wanner et al., 2008). This would facilitate the diffusion of 2-oxoglutarate from the central bacterium to the chlorobia. However, recent experiments suggest that the proposed periplasmic space is not a place where molecules can freely diffuse (Müller and Overmann, 2011). Anyway, it has been shown that Chl. chlorochromatii CaD3 when free-living, excretes a variety of amino acids and sugars (Pfannes, 2007), and according to the proposed role of the epibiont in ‘C. aggregatum’, these molecules have to reach the β-proteobacterium, likely through the periplasmic space.

Although Chl. chlorochromatii CaD3 does not assimilate 2-oxoglutarate when free-living, ‘C. aggregatum’ shows chemotactic behavior towards 2-oxoglutarate (Fröstl and Overmann, 1998). Several lines of evidence suggest that the β-proteobacterium and not the chlorobia, assimilates this metabolite from the medium (Fröstl and Overmann, 1998; Overmann et al., 1998; Glaeser and Overmann, 2003b; Vogl et al., 2006). However, it is not known whether the central bacterium transfers organic carbon to the epibiont (Müller and Overmann, 2011). Previous studies analyzed the Δ13C ratio (that is, the difference between the δ13C values of biomarkers and the δ13C of ambient) to elucidate the carbon source used by the epibionts in ‘P. roseum’ (Glaeser and Overmann, 2003a). It was found that the Δ13C ratio of the chlorobia in ‘P. roseum’ is consistent with photoautotrophic growth (Glaeser and Overmann, 2003a). As Chl. chlorochromatii CaD3 has a complete rTCA, 2-oxoglutarate can be used to build piruvate in addition to glutamine (Madigan et al., 2009). Therefore, if 2-oxoglutarate enters the chlorobia from the β-proteobacterium it would end up forming biomass, contradicting the Δ13C ratio observations in ‘P. roseum’, but in accordance with the fact that the entire consortium requires of 2-oxoglutarate supplementation to grow (Fröstl and Overmann, 1998).

If the theoretical difficulties described above can be overcome, we propose that the β-proteobacterium diffuses 2-oxoglutarate to the Chlorobia via their shared periplasm (Wanner et al., 2008). Figure 4 summarizes what this study proposes, where in free-living state Chl. chlorochromatii CaD3 can assimilate ammonia via the AlaDH, where ammonia is a low-affinity substrate, and perhaps also by the GOGAT pathway because this enzyme can use ammonia instead of glutamine as a low-affinity substrate (Mäntsälä and Zalkin, 1976; Adachi and Suzuki, 1977; Matsuoka and Kimura, 1986). In symbiosis, the increase in the intracellular concentration of 2-oxoglutarate produces a regulatory shift in nitrogen assimilation, stimulating the PII and NifA proteins and increasing GS activity. Chl. chlorochromatii CaD3 in symbiosis has high rates of N2 fixation and GS activity (Wenter et al., 2010). We postulate that this regulatory shift is induced by the ‘injection’ of 2-oxoglutarate by the β-proteobacterium, and part of the glutamine synthesized by Chl. chlorochromatii CaD3 is diffused to the β-proteobacterium through the periplasmic tubules. The above is supported by experimental evidence that shows the expression of an ABC transporter of amino acids by Chl. chlorochromatii CaD3 only in symbiosis coded by gene Cag_0853, and directly influenced by the β-proteobacterium (Wenter et al., 2010).

Figure 4
figure 4

Diagram showing the proposed mode of nitrogen assimilation in symbiosis and free-living states. (a) Chl. chlorochromatii CaD3 in free-living state is assimilating ammonia by alanine dehydrogenase (AlaDH) and likely by glutamate synthetase (GOGAT) pathways. (b) Chl. chlorochromatii CaD3 in symbiotic state where 2-oxoglutarate (2-OG) is assimilated by the β-proteobacterium and diffused to Chl. chlorochromatii CaD3 through the shared periplasm that is, periplasmic tubules (Wanner et al., 2008). 2-oxoglutarate interacts with the PII and NifA proteins, stimulating the expression of the nif operon (Op. nif) for N2 fixation, and transformation of NH3 to glutamine (Gln) by GS, a portion of which is transported to the β-proteobacterium. Cag_1339 is shown as the probable gene coding for a 2-oxoglutarate permease and Cag_0853 as probable gene coding for a glutamine transporter, both only expressed in symbiosis (Wenter et al., 2010).

Full size image

Last but not the least, Chl. chlorochromatii CaD3 like cyanobacteria lack GlnD homologs (Arcondéguy et al., 2001). However, differing from cyanobacteria that have only GlnB, Chl. chlorochromatti CaD3 has three PII proteins. One of them (Cag_1998) is coded next to an ammonium transporter, similar to GlnK in E. coli. The other two PII proteins are coded next to each other in an operon containing the nifH gene that codes for nitrogenase reductase. Chl. chlorochromatii CaD3 also lacks the gene pphA (sll1771 in Synechocystis sp. PCC 6803) responsible for the phosphorylation of PII in cyanobacteria (Forchhamer, 2004). Therefore, it is likely that regulation of nitrogen assimilation in the epibiont differs in the details from what has been described so far.