Introduction

Most filamentous cyanobacteria that perform N2-fixation during the day develop specialized cells, called heterocysts, that do not evolve O2 photosynthetically. N2-fixation also occurs in unicellular cyanobacteria, and these have to reconcile the incompatible reactions of N2-fixation catalyzed by the O2-sensitive nitrogenase and photosynthetic O2-evolution. Some of the unicellular cyanobacteria separate the two processes by performing N2-fixation in darkness and photosynthetic O2-evolution and CO2-fixation during the day. However, few other unicellular cyanobacteria can fix both CO2 and N2 in light. These cells must be able to protect their nitrogenase from damage by O2, but the mechanisms for this are largely unknown. The subject on unicellular N2-fixing cyanobacteria has extensively been reviewed (Fay 1992; Turner et al. 2001; Berman-Frank et al. 2003; Stal and Zehr 2008) and shall not be repeated here. As discussed in the following, some N2-fixing unicellular cyanobacteria have recently been discovered to have an inactive photosynthetic reaction II, the O2-evolving part of the photosynthetic apparatus in cyanobacteria, eukaryotic algae, and plants. Thus, the nitrogenase of these unicellular cyanobacteria is not exposed to the O2 generated photosynthetically. Such organisms offer fascinating perspectives both in cell metabolism, ecology, evolution, and application and will, therefore, be discussed in the present minireview.

A group of unicellular N2-fixing cyanobacteria of the oceanic picoplankton reveals a new mode of life

Approximately 50% of the global biological N2-fixation proceeds in the oceans of the world (Galloway et al. 2004; Stal 2009). The free-living filamentous Trichodesmium and the heterocystous Richelia intracellularis within the diatoms such as Rhizoselenia were thought, until recently, to be major players for this marine N-input (Zehr et al. 2008). Trichodesmium and Richelia prefer warm, tropical and subtropical areas, whereas in the temperate regions, e.g. in the Baltic Sea, genera such as Nodularia and Aphanizomenon seasonally form the major N2-fixing blooms (Diez et al. 2008).

Additional contributors in tropical and subtropical regions are unicellular cyanobacteria with a diameter between 2 and 8 μm, such as Cyanothece of more coastal and benthic areas and Crocosphaera watsonii (belonging to the unicellular N2-fixing cyanobacterial group B) in the open oceans (Zehr et al. 2001; Mazard et al. 2004). Cells with a diameter <1 μm, thus belonging to the picoplankton, are even more important N2-fixing species (unicellular cyanobacteria group A = UCYN-A) in these areas (Falcon et al. 2004; Langlois et al. 2005; Goebel et al. 2008; Church et al. 2009; DeLong 2010). These cyanobacteria were not previously detected, since they are smaller than the related Cyanothece and Crocosphaera and lack phycoerythrin (Goebel et al. 2008). These UCYN-A organisms were discovered by amplification of the nitrogenase gene (nifH) and its transcripts from oceanic water samples (Zehr et al. 1998, 2001) and also are characterized by their 16S rRNA gene sequences (Zehr et al. 2008). They are found to be small dim cells by flow cytometry coupled with the quantitative polymerase chain reaction (Goebel et al. 2008). It has previously been suggested from nifH sequences that UCYN-A cells are most closely related to those from the marine unicellular cyanobacterium Cyanothece sp. strain ATCC 51142 and from the spheroid bodies of R. gibba (Zehr et al. 1998). However, the phylogenetic tree both of 16S rRNA (Fig. 1a) and the nifH (Fig. 1b) gene indicates that these organisms, though being in the same group, cluster on somewhat different lineages in the phylogram. Despite being identified as unicellular N2-fixing cyanobacteria, the UCYN-A have yet to be cultivated. Using molecular techniques (the quantitative polymerase chain reaction), they have been found to be widely distributed in tropical and subtropical waters and appear to inhabit cooler waters than Trichodesmium or Crocosphaera (Langlois et al. 2005; Church et al. 2008; Moisander et al. 2010).

Fig. 1
figure 1

aPhylogenetic tree of unicellular cyanobacterial 16S rRNA nucleotide gene sequences. The arrow indicates representative UCYN-A sequences. Sequences were aligned using the SILVA aligner (http://www.arb-silva.de/aligner/) (Pruesse et al. 2007). The tree was constructed in MEGA4 (Tamura et al. 2007) using a Neighbor-Joining method and evolutionary distances were computed using a Jukes-Cantor correction. Bootstrap values were inferred from 1,000 replicates. Due to the short length of EU187793 and the complete deletion of positions containing gaps or missing data, there were 614 positions in the final data set. bPhylogenetic tree of nifH protein sequences from unicellular cyanobacteria. The arrow indicates representative UCYN-A sequences. Protein sequences were aligned using a hidden Markov model from the Pfam database (http://pfam.sanger.ac.uk/) (Finn et al. 2010). The tree was constructed using a Neighbor-Joining method, a Poisson correction was used to compute the evolutionary distances, and a bootstrap test was conducted with hahah 1,000 replicates. There were 108 positions in the final data set

Full size image

Examples of their occurrence in high abundances are cooler, nutrient-poorer areas, e.g. between 14°N und 29°N in a region between 152°W and 170°W of the Pacific Ocean (Church et al. 2008) and along a transect westwards and in parallel to Japan (Kitajima et al. 2009). UCYN-A organisms have recently been reported in particularly high abundances at substantially higher latitudes including in deeper (up to 100 m depth) subsurface waters in the South Pacific Ocean (Moisander et al. 2010).

UCYN-A cyanobacteria express the nifH gene with maximal transcript abundance during daytime (Zehr et al. 2007). They apparently can fix N2 in the light which is unlike several unicellular N2-fixing cyanobacteria that separate photosynthetic O2-evolution and N2-fixation by performing the former process during the day and the latter in darkness (Gallon 2001; Trepel et al. 2009). UCYN-A cells have nitrogenase gene arrangement and composition similar to those of Cyanothece sp. ATCC 51142 and of the spheroid bodies of Rhopalodia gibba (Zehr et al. 2008). A metagenomic analysis of cells concentrated by flow cytometry did not reveal any genes coding for phycocyanin, phycoerythrin or associated linkers which explains that they have not been detected earlier in the oceans by phycoerythrin fluorescence microscopy or flow cytometry. In addition, genes coding for the Calvin-Benson cycle and for photosystem II were not detected on the genomic contig, and target genes of these two photosynthetic part reactions (Rubisco and psbA, respectively) could not be amplified by PCR (Zehr et al. 2008). Thus, UCYN-A cyanobacteria cannot synthesize organic carbon by photosynthesis but are strictly dependent on extracellular sources. Due to the absence of photosystem II, their nitrogenase is not exposed to O2 generated photosythetically. The genomic contig revealed photosystem I genes. UCYN-A cells apparently meet their energy demand by cyclic photophosphorylation. It has not been determined yet which type of thylakoids they possess.

The recently obtained metagenomic and metatranscriptomic data sets showed that the genome of UCYN-A is highly conserved (>97% sequence identity) for samples taken from Pacific, Indian and Atlantic Oceans which contrasts strikingly with other marine bacteria such as Pelagibacter or the photosynthetic Prochlorococcus that show a high sequence diversity (Tripp et al. 2010). The assembly of the complete genome (Tripp et al. 2010) indicated that UCYN-A contains complete metabolic pathways such as glycolysis (from glucose-6-P to pyruvate), pentosephosphate pathway, fatty acid biosynthesis, pigment biosynthesis and assimilatory sulfate reduction besides N2-fixation. In contrast, no Calvin cycle or any other of the five known CO2-fixation pathways (Thauer 2007) are present, and they lack the genes coding for enzymes of the tricarboxylic cycle, the biosynthesis of several amino acids such as valine, leucine or isoleucine and of purine (Tripp et al. 2010). They possess transporters for sugars and dicarboxylic acids. It is therefore tempting to assume that they assimilate dicarboxylic acids, unidentified sugars and/or amino acids from ocean waters despite its low nutrient content. Sugars would then be degraded via glycolysis to pyruvate. Also other non-photosynthetic bacteria of the oceans, e. g. the uncultured SAR11 group, apparently utilize monosaccharides by modifications of the glycolytic pathway (Schwalbach et al. 2010). In UCYN-A, the two pyruvate:ferredoxin oxidoreductases described for cyanobacteria (Schmitz et al. 2001) are absent, but the genes coding for all three enzymes of the pyruvate:dehydrogenase complex (pyruvate dehydrogenase; lipoamide acetyltransferase, lipoamide dehydrogenase) have been detected on the genome (unpublished data). Thus, the UCYN-A cells might be able to cleave pyruvate in the presence of coenzyme A to acetyl coenzyme A and CO2, but the fate of the remaining two electrons (of NADH) remains uncertain. They have no bidirectional hydrogenase to dispose of the reducing equivalents (but uptake hydrogenase is present). To get rid of reductant, they must produce a fermentative endproduct such as malate (malate dehydrogenase is present), lactate, formate or any other compound formed in NADH oxidation. Excreted energy-rich compound (malate, lactate) would likely be rapidly consumed by other microorganisms, which could symbiotically supply the missing amino acids to UCYNA. Despite attempts to observe symbiotic partners (Tripp et al. 2010), no organism accompanying UCYN-A has been detected as yet. In addition, genes coding for lactate dehydrogenase or formate deydrogenase are not apparent in the genome. A transfer of the reducing equivalents from NAD(P)H to an enzyme and then directly to O2 is unlikely.

The pennate diatom Rhopalodia gibba and its endosymbiotic spheroid bodies

The UCYN-A cyanobacteria of the picoplankton bear resemblances to previously reported endosymbiotic microorganisms. Diatoms of the family Epithemiceae contain oblong inclusions, the so-called spheroid bodies (Pfitzer 1869, as quoted by Klebahn 1896). There are 1–4 (maximally 16) per cell within the genera Rhopalodia and Epithemia and with Denticula vanheurcki, however not with D. tenuis (Geitler 1977). The latter author also reported that the fission of the spheroid bodies is largely independent of the multiplication of the host and that at least one spheroid body is transferred to the next generation in each case. The spheroid bodies are not pyrenoids, as assumed originally (Klebahn 1896; Fritsch 1945), since they are outside of the chloroplast of the Epithemiaceae. A relationship to unicellular cyanobacteria was inferred from electron microscopic investigations (Drum and Pankratz 1965). Thus, the spheroid bodies were suspected to be peculiar organelles or intracellular organisms. However, any evidence for a relatedness to unicellular cyanobacteria was questioned by Geitler (1977).

The electron microscopic image (Fig. 2a) shows two spheroid bodies in Rhopalodia gibba that resemble coccoid cyanobacteria (Floener and Bothe 1980). The spheroid bodies are enclosed by a thick cell wall and are surrounded by a membrane of the host. As with the Rhizobium–legume and mycorrhiza–plant symbioses, host and spheroid bodies are strictly separated, and often an optically empty space is seen in between them, as already noted by Drum and Pankratz (1965). Thylakoids are discernible and extend from the inner layer (probably the cytoplasmic membrane) radially to the center of the spheroid bodies (Fig. 2b). Almost all cyanobacteria and also the cyanelles of glaucophyta (Löffelhardt et al. 1997; Reyes-Prieto and Bhattacharya 2007) possess concentric thylakoids. However, a radial thylakoid arrangement was described for the cyanobacteria Phormidium retzi and Oscillatoria limosa (Golecki and Drews 1982). Since their cultures grew independently of the addition of combined nitrogen, Drum and Pankratz (1965) speculated that Rhopalodia gibba may meet its N-demands by N2-fixation.

Fig. 2
figure 2

Longitudinal sections through Rhopalodia gibba. a. The electron microscopic image reveals two spheroid bodies (marked with E) which show a thick cell wall and thylakoids. A sausage-like chloroplast (Ch) is also present. b. Section of a spheroid body showing the radially orientated thylakoids. Part of the neighboring chloroplasts with its thylakoids is seen. W thick wall of the spheroid body. The Rhapolodia cells were fixed with glutaraldehyde—OsO4 and prepared as described by Schnepf and Deichgräber (1978). For further details, see Floener (1982) from where the originals of the photos were taken. The bar represents μm

Full size image

To our knowledge, members of the Epithemiaceae are not available in culture collections currently. These organisms are apparently not very rare in freshwaters. Rhopalodia gibba has been described for the Große Plöner See in Northern Germany (Geitler 1977), and this species and members of the genus Epithemia can readily be isolated from ponds of the botanical gardens of the Universities of Vienna, Austria (Geitler 1977) and Marburg, Germany (H. A. von Stosch, K. Wenderoth, personal communication). In 1978, the doctoral student L. Floener of Cologne isolated R. gibba in Marburg with the help of the late H. A. von Stosch. This diatom can be grown in liquid media on all conventional N-sources (NO3 −, NH4 + or N2 as the only N-source) with essentially the same generation time of 2–3 days (Floener 1982). However, they were strictly epiphytic, meaning that they could be grown only on the surface of 2% agar in liquid culture medium. After 4–6 weeks of growth, they were scratched off with a bent glass rod for physiologic experiments. Although it was impossible to get enough cells for a detailed biochemical analysis, the material was sufficient for performing N2-fixation (C2H2-reduction) experiments (Fig. 3).

Fig. 3
figure 3

C2H2–reduction (N2-fixation) by Rhopalodia gibba The experiment was performed in seven 2-ml Fernbach flasks containing intact cells of R. gibba with 0.3 mg protein in 3 ml nutrient solution, 4% O2 in the gas phase of the vessels and low light intensities at their surface (~3,500 lux). Prior to the assay, the cells had been grown under N2-fixing conditions, scratched off from the agar surface of the culture vessels, centrifuged and taken up in the assay medium. The amount of C2H4 formed was determined by gas chromatography. The original data are from Floener (1982). Lines between two filled squares denotes under N2-fixing conditions (without combined nitrogen in the medium). Lines between two filled circles denotes with nitrate in the medium. Lines between two filled triangles denotes with ammonium ions in the medium. Lines between two filled diamonds denotes in darkness, under N2-fixing conditions

Full size image

The Rhopalodia cultures reduced C2H2 immediately, without any lag phase, continuously and strictly light dependent with a rate of approximately 10 nmol C2H4 formed/h x mg total diatom protein. The activity was distinctly higher when the cells were grown on N2 and not NO3 − or NH4 + as N-source (Fig. 3). Generally, the addition of combined nitrogen to cultures completely blocks nitrogenase synthesis, but here this N-source possibly did not reach the spheroid bodies quantitatively. Maximal C2H2-reduction activity required 2–4% O2 in the gas space of the vessels (Floener and Bothe 1980). Thus, as with several unicellular cyanobacteria (Fay 1992), N2-fixation by R. gibba required reduced O2-tensions (microaerobic conditions). The C2H2-reduction rate was maximal at the low light intensity of 5,000 lux (approximately) at the vessels, but higher light intensities were not inhibitory (Floener 1982). The R. gibba culture was not free from contaminant bacteria. An enrichment culture of the contaminant bacteria did not fix N2 (Floener 1982). The spheroid bodies are permanently incorporated into the cells of the Epithemiaceae, and are not like the kleptoplastids where heterotrophic euglenids temporarily exploit cyanobacteria for performing photosynthesis prior to their digestion (Schnepf et al. 2002). The spheroid bodies also do not resemble any of the other known endosymbioses within protists (Nowack and Melkonian 2010). Besides being cyanobacteria they are totally unrelated to the marine heterocystous cyanobacterium Richelia intracellularis that lives endosymbiotically in the diatoms of the genera Rhizosolenia or Hemiaulus (Zeev et al. 2008).

N2-fixation is restricted to prokaryotes, and no chloroplast of eukaryotes has this capability. The relevance of the spheroid bodies is that they are N2-fixing entities within diatoms, thus within eukaryotic cells. The occurrence of thylakoids indicates the relatedness of spheroid bodies to cyanobacteria. However, no typical cyanobacterial pigments such as phycobilins, echinenone or myxoxanthophyll could be extracted, but diatoxanthin and diadinoxanthin of the Bacillariophyceae (Floener 1982) were present. Spheroid bodies are definitively not related to cyanelles of glaucophyta that do not possess chloroplasts. In glaucophyta, only the cyanelles perform photosynthesis. Cyanelles possess concentrically arranged thylakoids and phycobilins. They cannot perform N2-fixation (Floener et al. 1982).

A unicellular, N2-fixing cyanobacterium tightly integrated into a eukaryote could serve as a fascinating model organism for future research and applications. In the past, there have been many attempts with limited success to artificially construct a stable symbiosis between a crop plant and a symbiotic N2-fixing microorganism to make the plant independent of a supply with combined N (Tikhonovich and Provorov 2009). Mainly due to the lack to obtain sufficient amounts of cells, the research on R. gibba had to be abandoned then by the Cologne laboratory, but the results obtained so far were published in a proceedings volume (Floener and Bothe 1980). In addition, cells became smaller and smaller with the generations, since sexual states of R. gibba were not available. The definitive proof that spheroid bodies are cyanobacteria-related could not be obtained since molecular techniques were not available 30 years ago.

In the past few years, U.G. Maier and coworkers at Marburg University, Germany, reisolated R. gibba and used the same culture conditions as Floener and Bothe (1980). They were then able to definitively prove by molecular methods that spheroid bodies possess nitrogenase and are related to free-living cyanobacteria (Prechtl et al. 2004; Kneip et al. 2007, 2008). These authors were able to isolate spheroid bodies and their DNA from R. gibba. Sequencing and the phylogenetic analysis of the 16S rDNA gene revealed a close relatedness of the spheroid bodies to the free-living N2-fixing cyanobacterium Cyanothece sp, ATTC 51142 (Prechtl et al. 2004). Preparations of spheroid bodies were used to detect N2-fixation by use of nifD gene, one of the two genes that code for the MoFe-protein (larger subunit) of nitrogenase. DNA hybridizations proved the presence of the gene in spheroid-DNA. Immunogold-labeling experiments with nitrogenase antibodies showed that the enzyme was expressed in spheroid bodies. Phylogenetic analysis of the 16S rDNA- and nifD genes showed that the branches that separate free-living cyanobacteria and spheroid bodies are very short. Thus, the symbiosis of R. gibba with its endosymbiotic spheroid bodies has evolved rather recently (Kneip et al. 2007). Spheroid bodies possess all genes that are essential for N2-fixation with some genes being more related to those in Cyanothece and others to those in Gloeothece (Kneip et al. 2008). The non-essential fdxN-gene has been changed to a pseudogene by insertion of stop codons. Inactivation by truncation occurs in nifU, another gene which is also not absolutely required for the expression of nitrogenase. A lot of genes coding for cell metabolism enzymes underwent deletion or inactivation by insertion of A and T nucleotides. Examples of this are genes coding for photosynthesis proteins such as petJ, (encoding cytochrome c6), petE (for plastocyanin), psbC und psbD (both of photosystem II). The spheroid bodies apparently retained only the thylakoids, but the loss of photosynthetic pigmentation (Kneip et al. 2008) might indicate that they are unable to perform any part reaction of photosynthesis. This includes photosystem-I dependent cyclic photophosphorylation, since none of the genes of this pathway has been detected in the fosmid library. However, this will finally be proven when the complete sequenced genome will be available. For the generation of ATP and their own carbon skeleton, the spheroid bodies must be strictly dependent on a supply with organic carbon from the host.

The presumed sole task of the spheroid bodies is to supply the host with a product of N2-fixation (NH4 +?). The estimated genome size of the spheroid bodies is 2.6 Mb and is thus less than half than that of Cyanothece sp. CCY0110 (5.8 Mb). Due to the loss of quite a number of genes, the spheroid bodies are non-autonomous and cannot be grown independently of their diatom host. Remarkably, the two DNA-repair genes recA and recF are intact, in contrast to the situation with other tightly incorporated symbiotic systems (Kneip et al. 2008). R. gibba with its unicellular cyanobacteria-related spheroid bodies represents a new model system that might provide insights into early events in the evolution of chloroplasts (Kneip et al. 2008).

Is the cell metabolism of the unicellular UCYN-A cyanobacteria related to that of the acetate photoassimilating green alga Chlamydobotrys?

The cell metabolism of the UCYN-A cyanobacteria of the picoplankton, indeed, shows some analogous features to that of the acetate photoassimilating green alga Chlamydobotrys sp. of the Volvocales, an organism taxonomically very distant from cyanobacteria. A series of papers in the early 1960s established that Chlamydobotrys is capable of a particular form of photoheterotrophy that came to be known as photoassimilation of acetate (Pringsheim and Wiessner 1960, 1961; Wiessner and Gaffron 1964). Chlamydobotrys grows on acetate as the sole carbon source in the light but not in darkness. Under aerobic conditions, the rate of carbon assimilation is about 25% higher than under O2-exclusion (Wiessner 1962). The carbon assimilated into the cell material mainly comes from the methyl group of acetate (Wiessner 1962). The rate of CO2-fixation is at best very low and not sufficient to meet the organic C-demands of the cells (Wiessner and Gaffron 1964). Photosystem II seems to participate only in the regulation (redox poising) of the photosynthetic electron flow and does not serve as principal source of reductant in linear photosynthetic electron transport (Mende and Wiessner 1983). Energy (ATP) is generated mainly by cyclic photophosphorylation around photosystem I. In contrast to UCYN-A, Chlamydobotrys does not perform N2-fixation.

The metabolic pathway leading from acetate to carbohydrates has never been entirely elucidated in Chlamydobotrys. Cyclic photophosphorylation seems to be necessary to activate acetate, perhaps to acetyl coenzyme A or acetyl phosphate, and then enzymes of the glyoxylate cycle are apparently involved in the incorporation of the methyl-group of acetate into the carbon of the cells (Wiessner 1962). From pyruvate on, gluconeogenesis might generate monosaccharides. Growth on acetate does not require the disposal of excess of reductant, and Chlamydobotrys is not dependent on the co-culture with another organism.

It remains to be shown whether acetate or any other chemically simple organic carbon source is sufficiently available in the oligotrophic regions of the open oceans to sufficiently allow growth of the N2-fixing picoplankton. However, some abundant marine cyanobacteria such as Prochlorococcus release significant amounts of carboxylic acids into media (9 to 45 μg C L−1 in the form of acetic acid; Bertilsson and Jones 2003, 2005). The low population density (generally less than 1,000 cells ml−1) and small size of UCYN-A (approximately 0.75 μm) imply a maximum carbon demand for biomass around 38 ng C L−1, or three orders of magnitude less than that which could theoretically be provided by Prochlorococcus. This estimation is based on the assumption that the cell density of UCYN-A is 1.03 g cm−1, its dry weight equals a third of wet weight, and C demand for biomass is half of dry weight. The division times of UCYN-A and Prochlorococcus are likely to be similar, although the growth rates of UCYN-A are not yet known. It is currently speculative to assume that UCYN-A utilizes acetate as Chlamydobotrys does and to what extent the pathways of acetate assimilation are similar in Chlamydobotrys and UCYN-A. This cyanobacterium possesses an acetate kinase also and could use the ATP formed by cyclic photophosphorylation to activate acetate. Genes coding for enzymes of the glyoxylate cycle are, however, absent in UCYN-A. Thus, the photoassimilation pathway of acetate in UCYN-A, if occurring, must be other than in Chlamydobotrys.

Concluding remarks

As heterocysts of filamentous cyanobacteria, UCYN-A organisms of the marine picoplankton and spheroid bodies of R. gibba lost photosystem II activity. The inability to evolve O2 photosynthetically allows them to perform N2-fixation. As with heterocysts, the catabolism of organic carbon is not fully understood in UCYN-A and spheroid bodies. Photoassimilation of acetate, as in the alga Chlamydobotrys, is only one of the possibilities that may happen in their carbon metabolism. Nature may hide further microorganisms with entirely new modes of life.