Introduction

Evolutionary importance of cyanobacteria

Cyanobacteria form one of the most morphologically and genetically diverse group of Prokaryotes (Waterbury, 1991; Castenholz, 2001) showing cellular and colony differentiation. They are classified in five subsections and Orders, comprising unicellular and filamentous forms (Castenholz, 2001). They represent the basis of the nitrogen cycle, because the capacity to fix atmospheric N2 is found throughout this lineage, making them essential components of past and modern ecosystems (Bergman et al., 1997; Capone et al., 1997; Raymond et al., 2004; Tomitani et al., 2006; Haselkorn, 2007). Molecular phylogenetic studies have made it clear that all photoautotrophic eukaryotes (plants and algae) share a single origin, as well as a common endosymbiotic ancestry, for cyanobacteria-derived chloroplasts (Bhattacharya and Medlin, 1995; Delwiche et al., 1995; Douglas, 1998; Moreira et al., 2000; Martin et al., 2002; Raven and Allen, 2003; Stiller et al., 2003; Hedges et al., 2004; McFadden and van Dooren, 2004; Yoon et al., 2004; Rodriguez-Ezpeleta et al., 2005; Hackett et al., 2007 among others). The work of Bhattacharya and Medlin (1995), Nelissen et al. (1995) and Turner et al. (1999) suggested the chloroplast lineage arose at the onset of diversification of the cyanobacterial lineage. Recent work by Deusche et al. (2008) suggested, after a careful examination of four eukaryotic and nine cyanobacterial genomes, that among cyanobacteria, Nostoc and Anabaena, within Order Nostocales, harbor more genes related to those acquired by eukaryotes. This suggests that the ancestor of the chloroplast could lie within the heterocyst-forming cyanobacteria. Heterocysts are specialized cells for N2 fixation that lack the oxygen-generating photosystem-II (PSII). They consist of a thick isolating cell wall that is less permeable to gases, and heterocysts are connected to adjacent vegetative cells by micro-plasmodesmata, through which organic compounds (for example, sugars, amino acids) may pass. Sugar is required for respiratory reductive power, but most of the required ATP is produced through PSI, which is the only PS remaining in the heterocyst. The ATP is needed to fuel the activity of nitrogenase, the enzymatic complex capable to fix atmospheric N2, which is irreversibly inhibited in the presence of oxygen (Bergman et al., 1997). Tomitani et al. (2006) suggested, on the basis of genetic distances and fossil calibrations, an age ranging from 2450 to 2100 million years ago (MYA) for heterocystous cyanobacteria, which may predate the rise of atmospheric oxygen at about 2300 MYA. However, the work of Deusche et al. (2008) considered whole genomes, but cyanobacterial diversity is poorly represented in genomic studies, thus phylogenetic interpretations may be misleading at present. Recently, Deschamps et al. (2008) provided the first evidence of the existence of starch in bacteria within unicellular, N2-fixing cyanobacteria, belonging to Order Chroococcales. These authors suggested that starch formation would define the genetic make-up of the ancestor of the plant kingdom related to storage polysaccharide metabolism. Unicellular N2-fixing cyanobacteria differ phylogenetically from the heterocyst lineage, and have resolved N2 fixation and oxygen-generating photosynthesis through temporal separation, storing polysaccharides in starch granules during the day to fuel N2 fixation at night (Falcón et al., 2004). The above suggests that the ancestor of chloroplasts had the ability to fix N2, fix CO2 by an oxygen-evolving type-II PS and store starch. The ancient symbiosis metabolic fluxes consisted of the export of ADP-glucose from the cyanobiont to the host, eliminating its ability to store polysaccharides, thus in habilitating its capacity to fuel N2 fixation, demanding import of reduced nitrogen from the host to the cyanobiont (Deschamps et al., 2008).

Materials and methods

To estimate the timing of phylogenetic divergence events, we used a data set including 56 cyanobacterial taxa from all subsections, and nine chloroplasts, which included members of Rhodophyta, Glaucophyta, Chlorophyta and Streptophyta. Phylogenetic relationships were estimated on the basis of nucleotide sequences of 16S rDNA (1255 bp), rbcL (1470 bp) and a concatenated set of these two loci. Bayesian phylogenetic analysis for the individual and combined loci were conducted with MrBayes v3.1.2 (Huelsenbeck and Ronquist, 2001), applying the model with best fit to each data set as identified with the Akaike Information Criterion, implemented in Modeltest (Posada and Crandall, 1998; Posada and Buckley, 2004). Each Bayesian analysis consisted of two independent Markov chain Monte Carlo runs, each formed by four differentially heated chains of 5 × 106 generations, in which a tree was sampled every 200 generations. Phylograms topologically identical to the maximum a posteriori (MAP) topology were recovered using PAUP*4.0b10 (Swofford, 2002), and from these, 100 were randomly selected to conduct dating analyses.

The timing of phylogenetic divergences was estimated with penalized likelihood (Sanderson, 2002), implemented in r8s v1.71 (Sanderson, 2006). The optimal smoothing parameter for each data set was identified through a cross-validation procedure that involved pruning terminal branches. Dating analyses were conducted on the 100 phylograms topologically identical to the MAP tree. The trees were calibrated by fixing the origin of the cyanobacterial lineage at 3500 MYA, based on the age of oldest fossils represented by stromatolites (Schopf and Packer, 1987), and at 2700 MYA, the time at which oxygen-evolving cyanobacteria had likely quite originated due to reports of steranes in carbonaceous shales of northwestern Australia (Brocks et al., 1999). A maximal age constraint of 460 MYA was applied to the crown group of tracheophytes, considering the oldest vascular plant fossil remains (Kenrick and Crane, 1997). Minimal age constraints of 1618 and 1253 MYA were applied to the heterocyst-forming lineage and to the origin of plastids, respectively, derived from a preliminary analysis in which the ages of these lineages were estimated without imposing minimal age constraints. Point estimates of age from each of the 100 phylograms were used to obtain mean and standard deviations of ages of nodes across the tree.

Results and discussion

Ancestry of the chloroplast

Bayesian inferences of phylogenetic relations between cyanobacteria and chloroplasts with 16S rDNA and rbcL genes, plus the concatenated set, produced the following results: (1) chloroplasts constitute a monophyletic lineage and are most closely related to N2-fixing unicellular cyanobacteria and (2) heterocyst-forming cyanobacteria are their sister group (Figure 1). Molecular clock estimates rooting the origin of cyanobacteria at 3500 and 2700 MYA gave intervals of appearance of N2 fixation in cyanobacteria within the late Archaen and early Paleoproterozoic eons, while heterocystous and unicellular N2-fixing cyanobacterial clades must have originated within the Paleoproterozoic (Table 1). The molecular clock estimated age for the heterocystous cyanobacterial clade coincides with the dates suggested by Tomitani et al. (2006) on the basis of genetic distances and fossil calibrations. Dates for the plastid lineage, considering a cyanobacterial phylogeny, occurred in the mid-Proterozoic, in agreement with previous studies (Figure 2 and Table 1).

Figure 1
figure 1

Phylogenetic relationships of cyanobacteria and chloroplasts rooted with (a) Gloeobacter violaceus (rbcL and concatenated set), (b) Chlorobium tepidum and (c) Chloroflexus aurantiacus (16S rDNA). Clades represent chloroplasts, unicellular cyanobacteria with the ability to degrade starch to simple sugars and use these as reductants for N2 fixation, and heterocystous filamentous cyanobacteria. Scales represent genetic distances.

Full size image
Table 1 Dates for the chloroplast lineage and related cyanobacterial clades represented in MYA, calculated for 16S rDNA, rbcL and the concatenated set, showing mean and standard deviations from 100 identical topologies to the maximum a posteriori topology. Results rooting the cyanobacterial lineage at 3500 and 2700 MYA are shown
Full size table
Figure 2
figure 2

Divergence time estimates (MYA) among cyanobacteria and chloroplasts, including 56 cyanobacterial taxa from all subsections, and nine chloroplasts, which include a member of the oldest of the group, Cyanidiales (Rhodophyta), Glaucophyta, Chlorophyta and Streptophyta, all originating from the primary endosymbiosis event. Bayesian phylogenetic analysis showing the maximum a posteriori (MAP) topology, dated with penalized likelihood. Tree calibrated by imposing a fixed origin of the cyanobacterial lineage at 3500 and 2700 MYA, and a maximal age constraint of 460 MYA to the crown group of tracheophytes (indicated by asterisks).

Full size image

After evolution of life, the second major event that transformed the biogeochemistry of the Earth, was oxygen-evolving photosynthesis by cyanobacteria (Dismukes et al., 2001; Kopp et al., 2005; Cavalier-Smith, 2006; Shi and Falkowski, 2008). The Great Oxidation Event (GOE), which establishes the presence of molecular oxygen in the fossil record, and thus of oxygen-producing photoautotrophs, occurred as early as 2450 MYA (Holland, 2002). However, the work of Brocks et al. (1999) showed that steranes were already present in the geological record by 2700 MYA, implying biologically produced molecular oxygen. Microfossils comprising six bacterium morphotypes, including cyanobacteria, have been found in Archaean rocks dating between 3200 and 3500 MYA (Schopf, 2006). Thus, current evidence suggests that the origin of oxygen-producing cyanobacteria may date from as early as, or even earlier than, 3500 MYA, and were likely extant by 2700 MYA. Nevertheless, geological features that require free environmental oxygen, for example, banded iron formations, lateritic paleosols and sulfate deposits, occur shortly before the 2300—2200 MYA global ‘snowball Earth’, but are not present at the ∼2900 MYA Pongola glaciation (Kopp et al., 2005) contradicting the Archaean appearance of oxygenic photosynthesis. Further, it has been argued that the isotopic line of evidence for early >3500 MYA oxygen evolution with ∂13C values attributed to C-fixation, sulfate deposits (∼3450 MYA) and anaerobic methanotrophy (∼2700 Ma), can occur under anaerobic conditions (Hayes, 1994; Canfield et al., 2000; Rosing and Frei, 2004; Kopp et al., 2005). Eigenbrode and Freeman (2006) examined 13C enrichment patterns of the Hamersley Province in Western Australia and suggested that oxygenic photosynthesis must have originated sometime before 2720 MYA. This event eventually triggered the rise of aerobic ecosystems, fueling their expansion from anaerobic settings into the photic zone between 2720 and 2450 MYA. Proterozoic ocean simulations (Fennel et al., 2005) suggest that rise of oxygen was delayed due to feedbacks on the N-cycle. Ammonium, in presence of oxygen, would be biologically converted to nitrate, and denitrification would have rapidly deprived the oceans of fixed inorganic nitrogen, shifting the Proterozoic ocean to a N-depleted state. In this scenario, a free-living cyanobacterium with the capacity to store starch through oxygenic CO2 fixation, plus fix atmospheric N2, would be a very important intracellular acquisition. As can be recounted today from several lines of evidence, this cyanobacterium would have become the chloroplast through endosymbiosis.

Our results propose the existence of oxygen-evolving cyanobacteria back to the Archaean ∼2700—2500 MYA. Our results coincide with the conclusion of Eigenbrode and Freeman (2006) that the origin of oxygenic photosynthesis must have remained contrived to microbial communities, which led a transition away from purely anaerobic metabolism, fueling atmospheric oxygenation. The delay between the appearance of oxygen-evolving photosynthesis and accumulation of oxygen in Earth's atmosphere must have been of several hundred million years, as suggested by geochemical evidence (Fennel et al., 2005).

Cyanobacterial N2 fixation and climate

The molecular clock dates of appearance of N2 fixation in cyanobacteria correspond to the late Archaean and the early Proterozoic eons at ∼3000—2500 MYA, coinciding with those estimated by Shi and Falkowski (2008).

Current global rates of N2 fixation are estimated to be much smaller than global denitrification. The balance between both processes during glacial/interglacial periods has an effect on the amount of nitrate in the ocean, influencing the rate of carbon sequestration, which is controlled by iron availability (Michaels et al., 2001). A feedback system that controls carbon sequestration dynamics due to N2 fixation/denitrification rates has been proposed, coupled with iron availability and climate on millennium time scales.

Most of the N2 fixation (∼80%) in today's oceans is attributed to Trichodesmium spp. These are colonial, filamentous, non-heterocystous cyanobacteria with specialized cells for N2 fixation (Capone et al., 1997; Michaels et al., 2001). Molecular clock estimated the dates of appearance of Trichodesmium to range between 775 and 504 MYA. The 700- to 500-MYA time interval is associated to the Pan African period and it represents in the fossil record the onset of the Cambrian explosion. The increase in biodiversity within the Ediacaran and Cambrian periods is presumed to have been triggered by the split of the supercontinent Rodinia (1100—750 MYA), which preceded Pangea (Maruyama and Santosh, 2008).

Our results suggest that whereas cyanobacteria such as Trichodesmium could be responsible for major changes in the Earth's climate during the last ∼700 MYA through their global influence on the C and N cycles, other biogeochemically relevant bacteria, such as heterocyst-forming and unicellular, N2-fixing cyanobacteria, were possibly determinant in Earth's functioning during the last 2500 MYA, making them fundamental players in the global C and N cycles. Unicellular cyanobacteria related to the chloroplast line of descent have been acknowledged as important players in oceanic N2 fixation (Falcón et al., 2004; Montoya et al., 2004). Recently, members of this clade have been reported to lack genes for the oxygen-evolving PSII and C-fixation, with implications on their evolutionary history and influence on the global C and N cycles (Zehr et al., 2008).

The dates of the split between Trichodesmium erythraeum, now present in the Red Sea, and Trichodesmium havanum, from the Caribbean Sea, range between 214 and 94 MYA. The above suggests that these species of Trichodesmium shared a genetic pool in the Sea of Thetys and diverged with the split of Pangea. This result again places the temporality of modern biogeochemically relevant cyanobacteria and suggests how global arrangement of emerged continents and oceanic regions had an important role in biogeography and biogeochemistry.

The closest living relatives of the plastid lineage are fundamental components of past and modern oceanic ecosystems. Their double capacity to fix N2 through starch formation had a pivotal role in the instauration of the primary symbiosis. Our study suggests early time points of appearance of the plastid lineage and its sister clades, as well as of the cyanobacterial capacity to fix atmospheric C and N. We conclude that small, gradual changes must have operated during the millennia after the advent of biogeochemically important cyanobacteria, throughout the history of life on Earth. The existence of different biogeochemically important metabolisms, such as oxygen-evolving photosynthesis and N2 fixation, eventually changed the redox chemistry of the planet.