Background
The Viridiplantae (literally meaning green plants) include all green algae and embryophyte plants. They represent a monophyletic group of organisms, which display a surprising diversity with respect to their morphology, cell architecture, life histories and reproduction, and their biochemistry. The colonization of the terrestrial habitat by streptophyte algae 450 – 470 million years ago [reviewed in 1] was undoubtedly one of the most important steps in the evolution of life on earth 234, which paved the way for the evolution of the various groups of land plants (embryophytes = bryophytes, pteridophytes and spermatophytes) resulting in our current terrestrial ecosystems 5.
A thorough understanding of the evolution of land plants requires knowledge about the phylogeny of green algae and embryophytes as well as insight into the evolution of plant genomes with special reference to developmental processes. Whereas our knowledge about the phylogeny of the Viridiplantae has greatly increased over the last years, the latter has hardly been addressed to date.
The Viridiplantae are grouped into two divisions: the Chlorophyta and the Streptophyta 6. The Chlorophyta comprise the vast majority of green algae including most scaly green flagellates (e.g. Pyramimonas, Tetraselmis), the Ulvophyceae (e.g. Ulva, Acetabularia), Chlorophyceae (e.g. Chlamydomonas, Volvox) and Trebouxiophyceae (e.g. Chlorella) 78. The Streptophyta include all embryophyte plants and a diverse paraphyletic assemblage of freshwater green algae, the Charales (stoneworts), Coleochaete, the Zygnematophyceae and a few other taxa 9. Currently, the Charales are thought to be the sister group of the embryophytes suggesting that the evolution of true land plants already started with a complex organism 10. Remarkably, only a single scaly green flagellate Mesostigma viride Lauterborn, has been found to belong to the Streptophyta 10111213. The exact phylogenetic position of Mesostigma viride, however, is still controversial 101112141516. Mesostigma has recently attracted much attention as a putative key organism for the understanding of the early evolution of the Streptophyta 17181920.
Two aspects in the evolution of land plants seem to be important in this respect. First, many key evolutionary inventions of plants took already place within the streptophyte algae. According to Graham et al. 21 one can distinguish several major transitions in the evolution of land plants starting with a Mesostigma-like flagellate ancestor: development of a cellulosic cell wall, multicellularity, cytokinesis by a phragmoplast, plasmodesmata, apical meristematic cell and apical cell proliferation leading to branching, asymmetric cell division, cell differentiation, retention of zygotes, heteromorphic life history, and a root meristem. Of these distinguishing features only the latter two evolved not until the embryophytes emerged. Second, the colonization of the terrestrial habitat with its exposure to air, increased solar radiation and life in a desiccating environment led to adaptations of cell architecture, metabolism and body plan to survive in the terrestrial ecosystems 5. The evolutionary history of these adaptations is currently not known. Important questions are: How did the green algal progenitor adapt to the terrestrial habitat? Which genomic changes were associated with this transition? And which of these genes are derived from streptophyte green algae? To gain insight into these questions we have started to analyze ESTs from various streptophyte algal lineages.
Here, we present an analysis of 10,395 ESTs representing 3306 non-redundant expressed genes obtained from Mesostigma viride. We show that the number of genes shared is higher between Mesostigma and the embryophytes than between Mesostigma and Chlamydomonas. Comparison of expressed genes from Mesostigma with the genomes of Arabidopsis, Chlamydomonas, the red alga Cyanidioschyzon, and rice as well as ESTs from Physcomitrella and Porphyra allowed us to identify conserved and derived cellular functions within the different evolutionary lines and to obtain a first insight into the metabolic capabilities of the flagellate ancestor of green plants.
Results
Preparation and characterization of libraries
Total RNA was isolated from an axenic culture of Mesostigma viride during the light phase. The culture contained about 5 % cell division stages. The isolated RNA was used for the construction of 4 different cDNA libraries (Meso 1 – Meso 4). Meso 1 and 2 differed in the size of the cloned inserts. For Meso 3 and 4 full-length enriched cDNA was prepared and normalized prior to cloning. Meso 3 was obtained from the total normalized full-length enriched cDNA, whereas for Meso 4 the normalized full-length enriched cDNA was size-fractionated by gel permeation chromatography to remove small fragments. The basic characteristics of the four libraries are given in Table 1.
<p>Table 1</p>
Mesostigma viride cDNA libraries used.
Name
cDNA
Number of primary clones
Percentage recombinant clones
size of inserts1 (bp)
average size of inserts1 (bp)
number of ESTs sequenced
Meso1
small size fraction
310 000
90
250–2000
706
100
Meso2
large size fraction
295 000
88
600–3200
1142
4954
Meso3
total cDNA normalized
106 000
63
100–1100
650
535
Meso4
large size fraction normalized
304 000
56
400–6600
2025
2403 cDNAs = 4806 ESTs2
1 determined by agarose gel electrophoresis, 2sequenced from 3' and 5'end
Initially, about 100–500 ESTs were sequenced from all libraries and analyzed by BLASTX against the Swissprot and translated Genbank databases. Since the Meso 2 and 4 libraries containing the larger inserts gave more promising results, we subsequently sequenced about 4000 additional ESTs from the Meso 2 and Meso 4 libraries, respectively yielding a total of 10,395 reads (5,527,413 bp). Based on comparison with published sequences from Mesostigma viride the rate of sequencing error was determined to be generally between 1% and 7 % (average 4 %) depending on the quality of the sequence.
ESTs were assembled using the PHRAP software yielding 3300 contigs with an average size of 769 bases (57 – 4452 bases) after manual curation. Further analysis based on sequence similarity searches revealed that 294 of these contigs were of plastidic, mitochondrial, or possibly bacterial origin (sequences showing the highest similarity to organellar or bacterial genomes, Table 2). These contigs were excluded from the data set. 1315 of the 3006 contigs analyzed (44%) showed significant similarity at the protein level to sequences from the public databases (Table 2). Hence, approximately 56% of the contigs represent either novel sequences with unknown function or untranslated regions of a gene. However, when the 1691 contigs with no significant similarity to known proteins were searched against the Interpro protein motif database, 574 (33.9%) of these contigs contained a recognizable protein motif (Table 2). The most common protein motifs found in all 3006 expressed gene sequences were bipartite nuclear localization signals (IPR001472, 197x), proline-rich regions (IPR000694, 150x) and cytochrome c heme-binding sites (IPR000345, 99x).
<p>Table 2</p>
Summary of Mesostigma viride expressed genes obtained from four cDNA libraries (Meso 1 – Meso 4).
Category
No of contigs
Mitochondrial1
36
Plastidic1
65
Bacterial2
193
Novel
1691
with recognizable protein motif
574
no protein motifs
1117
Similarity
1315
with known function3
574
unknown function4
395
low similarity5
346
Total Contigs
3300
1 sequences showing only similarity to organelle genomes. 2 sequences showing only similarity to bacterial sequences or the highest similarity to bacterial sequences; the origin of these putative bacterial contaminations is currently not clear, as bacteria-free cultures of Mesostigma were used. 3similarity to proteins with a well-defined function (BLAST score >100). 4similarity to conserved proteins with no established function (BLAST score >100). 5low similarity to proteins from a few organisms (BLAST score generally between 100 and 200); might reflect conserved protein domains.
A functional catalogue was assembled using the 3006 Mesostigma contigs and the KOG-database and is presented in Table 3. As expected for an interphase cell, genes in the categories (1) translation, ribosomal structure and biogenesis (168), (2) posttranslational modification, protein turnover, chaperones (101), and (3) energy production and conversion (87) are represented by the largest number of contigs (Table 3). In the following, the assembled contigs are referred to as (expressed) genes.
<p>Table 3</p>
Functional classification of 3006 Mesostigma viride contigs using the KOG system [43] and an expectation threshold of e = 10-7.
Functional Category
No. of Contigs
INFORMATION STORAGE AND PROCESSING
236
[J] Translation, ribosomal structure and biogenesis
168
[A] RNA processing and modification
28
[K] Transcription
26
[L] Replication, recombination and repair
9
[B] Chromatin structure and dynamics
5
CELLULAR PROCESSES AND SIGNALING
209
[D] Cell cycle control, cell division, chromosome partitioning
7
[Y] Nuclear structure
0
[V] Defense mechanisms
2
[T] Signal transduction mechanisms
31
[M] Cell wall/membrane/envelope biogenesis
9
[N + Z] Cytoskeleton and cell motility
18
[W] Extracellular structures
0
[U] Intracellular trafficking, secretion, and vesicular transport
41
[O] Posttranslational modification, protein turnover, chaperones
101
METABOLISM
212
[C] Energy production and conversion
87
[G] Carbohydrate transport and metabolism
35
[E] Amino acid transport and metabolism
24
[F] Nucleotide transport and metabolism
9
[H] Coenzyme transport and metabolism
13
[I] Lipid transport and metabolism
21
[P] Inorganic ion transport and metabolism
16
[Q] Secondary metabolites biosynthesis, transport and catabolism
7
POORLY CHARACTERIZED
158
[R] General function prediction only
55
[S] Function unknown
37
[X] Unnamed protein
66
Unknown
2191
Classification of Mesostigma ESTs according to homologous genes in other organisms
EST data represent only a fraction of all genes of an organism. Thus, comparisons of EST data alone cannot be used to describe unique or shared genes of an organism. For embryophytes, chlorophytes and red algae complete genome sequences of at least one organism exist. This makes it possible to find potential orthologous genes if present. Moreover, the surplus of genes of an organism in respect to a complete genome can be detected in EST data. In tBLASTX analyses of the 1315 expressed genes with similarity to known proteins 90.3 % matched proteins from streptophytes, 76.1 % from chlorophytes and 61 % from rhodophytes, respectively. In addition, 46 genes showed similarity to known proteins, which have not been reported from plants or red algae to date. The overlap of Mesostigma genes with different organisms can be visualized in a Venn diagram (Figure 1). For 211 genes, we detected similar proteins only within the streptophyte but not in the chlorophyte or rhodophyte lineages. Conversely, for 62 genes we detected similar proteins only within the chlorophyte but not in the streptophyte or rhodophyte lineages. Surprisingly, we also found 6 genes which showed significant similarity to rhodophyte proteins but for which we could not detect any similar protein sequences within the Viridiplantae. Removal of BLAST hits with significant but low similarity (see Table 2) reduced the overall numbers to 972 expressed genes, but gave similar results (Figure 1). A complete list of genes showing only similarity to proteins with known functions present in specific subgroups of organisms can be found in supplemental Table 1 [see Additional file 1]. We will discuss important differences below.
<p>Figure 1</p>
Classification of expressed genes from Mesostigma according to the presence of similar proteins in other organisms in a Venn diagram
Classification of expressed genes from Mesostigma according to the presence of similar proteins in other organisms in a Venn diagram. All non-redundant expressed genes were used as a query in (t)blastx similarity searches with the Swissprot, Genbank, Chlamydomonas, Cyanidioschyzon, Porphyra, Physcomitrella, Arabidopsis and Oryza data sets. The outermost circle represents all Mesostigma expressed genes. The inner circles, which are labeled chlorophyte, streptophyte and rhodophyte, represent genes, which have similarity to chlorophyte, streptophyte or rhodophyte sequences, respectively. The areas depicted are not proportional to the gene numbers and the number of Mesostigma expressed genes in each category is written in each segment. Numbers in brackets indicate the number of expressed genes in a category after removal of low similarity hits (see Table 2 for a definition of low similarity hits).
![]()
<p>Additional file 1</p>
Supplemental Table 1
Click here for file
Overall protein similarities between various photoautotrophic organisms
To compare the overall similarity between Mesostigma and various photoautotrophic organisms with completed genomes or large data sets of ESTs, we decided to calculate the average identity of a protein between Mesostigma and the various organisms. To compare Mesostigma genes with the genomes or ESTs from different organisms, we calculated the average identity (AI) between Mesostigma and another organism as the mean value of all pair wise identities of the BLAST-matches for each organism (Table 4).
<p>Table 4</p>
Comparison of the Mesostigma expressed genes with the genomes and ESTs from various organisms. Average identity (AI) of pair wise comparisons of Mesostigma expressed genes with the indicated organismal data set.
Data set (No. of contigs)
Chlamydomonas
Cyanidioschyzon
Porphyra
Thalassiosira
Physcomitrella
Arabidopsis
Oryza
Genome
EST
Genome
EST
Genome
EST
Genome
Genome
All (969)
0.573
0.563
0.522
0.528
0.528
0.590
0.577
0.565
Constrained (314)
0.648
0.653 (n = 244)1)
0.557
0.585 (n = 188)1)
0.569
0.675 (n = 301)1)
0.679
0.671
Evolutionary distance D2)
0.473
0.463
0.658
0.597
0.631
0.425
0.418
0.432
The total data set contains all Mesostigma expressed genes with significant similarity to proteins from other organisms with known or unknown function (see Table 2). The constrained data set contains only Mesostigma expressed genes with significant similarity to proteins in all completely sequenced eukaryotic autotroph organisms.
1) Number of ESTs showing similarity to Mesostigma expressed genes from the constrained data set in a tBLASTX analysis. 2) Evolutionary distances were calculated using the constrained data set and the approximation given by Kimura [28]: D = -ln (1 - p - 0.2 p2), where p is the fraction of amino acid that differs between the two species.
The AI between Mesostigma and Chlamydomonas or the embryophytes are very similar. The highest AI value obtained was for Physcomitrella/Mesostigma followed by Arabidopsis/Mesostigma, Chlamydomonas/ Mesostigma and Oryza/Mesostigma. The full data set includes many proteins, which we detected only in some species using Mesostigma expressed genes as a query. Therefore, we constructed a constrained data set (314 expressed genes, including at least 46 nuclear encoded plastidic, 9 nuclear encoded mitochondrial, and 73 cytosolic ribosomal proteins), containing only Mesostigma genes which gave matches with all completed genomes from photoautotrophic eukaryotic organisms (including the diatom Thalassiosira). This constrained data set represents a conserved core set of nuclear encoded expressed proteins from photoautotrophic eukaryote organisms. We calculated AI values for the constrained data set using complete genomes and the available ESTs of Physcomitrella, Porphyra, and Chlamydomonas. The results are included in Table 4. We obtained the highest AI-values in the constrained data set for the three embryophytes, followed by Chlamydomonas. The similar AI values for the three different embryophytes suggest that the overall evolutionary rate was very similar for the embryophytes investigated, when compared with Mesostigma (see below).
To test whether the observed differences are significant a paired students t-test was performed, and the results are shown in Table 5. Applying a significance level of 0.0072 [0.05/7 Bonferroni adjustment 22] the differences in AI between Mesostigma/Chlamydomonas and Mesostigma/embryophytes are highly significant (Table 5), whereas the differences in AI among the embryophytes are not significant (Table 5). Furthermore, when we varied the numbers of expressed genes used for the calculation of the AI, we observed that when more than 100 ESTs were included the significance of the differences became very stable (Fig. 2A). In addition, to evaluate the consistency of the data set we calculated 8 times the AI for 150 randomly selected expressed genes from the constrained data set. A clear difference between the AI from the various organisms was always observed (Fig. 2B 1 – 8). The expression level of the expressed genes (as revealed by the number ESTs in a contig) had no effect on the differences between the investigated organisms (Fig. 2B, compare 9 and 10), although highly expressed genes are better conserved (Fig. 2B, 9 and 10).
<p>Table 5</p>
Statistical significance of the obtained AI values. A paired students t-test was performed for the constrained data set to test whether the observed differences between the average identity of pair wise comparisons of Mesostigma expressed genes with the indicated organismal data set are significant. Differences are considered significant when p is < 0.0071 (0.05/8 Bonferroni adjustment [22]).
Variable1
No. of genes shared
2
mean3
standard deviation4
t-value
Degrees of freedom
ρ
1
Mesostigma/Chlamydomonas G Mesostigma/Chlamydomonas E
244
0.652992 0.652992
0.149868 0.151255
-0,03107
243
0.975239
2
Mesostigma/Chlamydomonas G Mesostigma/Physcomitrella E
301
0.649934 0.674618
0.149292 0.153641
-3.24578
300
0.001304
3
Mesostigma/Chlamydomonas G Mesostigma/Arabidopsis G
314
0.648057 0.677994
0.148512 0.140940
-4.44025
313
0.000012
4
Mesostigma/Chlamydomonas G Mesostigma./Oryza G
314
0.648057 0.670382
0.148512 0.148384
-3.15371
313
0.001768
5
Mesostigma/Physcomitrella E Mesostigma/Arabidopsis G
302
0.675364 0.681159
0.153933 0.140567
-135783
301
0.175535
6
Mesostigma/Physcomitrella E Mesostigma./Oryza G
302
0.675364 0.673311
0.153933 0.147624
0,43158
301
0.666355
7
Mesostigma/Arabidopsis G Mesostigma./Oryza G
314
0.678730 0.671175
0.141321 0.148813
2.480053
313
0.013660
1 Compared data sets E = ESTs, G = Genome. 2 No. of genes shared between the compared data sets. 3 AI recalculated on the basis of the genes shared between the compared datasets. 4 Standard deviation of 3.
<p>Figure 2</p>
Consistency of the constrained data set used to calculate AI value
Consistency of the constrained data set used to calculate AI values. (A) The figure illustrates the effect of the number of genes included in the AI-values. The significant differences in the AI values are stable when more than 150 genes are included. (B) 150 genes were resampled randomly and the AIs calculated for the indicated organisms (1 – 8). AI values were calculated for the 150 most strongly (9, as revealed by the number of ESTs in a contig) and weakly (10, only single ESTs) expressed genes.
![]()
Two other results are remarkable. First, for the calculation of the AI it is possible to use large EST-data sets instead of genomes. We obtained the same result for Mesostigma/Chlamydomonas genome and for Mesostigma/Chlamydomonas ESTs (AI = 0.653 for both data sets; p = 0,975, Table 5, using 244 expressed genes from Mesostigma). Similarly, when Mesostigma/Physcomitrella ESTs were compared with the Mesostigma/Arabidopsis genome and with the Mesostigma/Oryza genome only small differences were observed (AI = 0,675/0,681; 0,675/0,673 respectively, using 302 expressed genes from Mesostigma, Table 5). Statistical analysis (paired students t-test) showed that the observed differences are not significant. Furthermore, we note that the genome of the diatom Thalassiosira pseudonana shows a similar AI in respect to Mesostigma as the red algal genome and ESTs (Table 4). The difference values of these distantly related genomes represent presumably an upper threshold for reasonable AI value calculations.
Analysis of metabolic pathways
ESTs have been widely used for the identification of metabolic pathways 23. A complete list of all metabolic pathways identified is presented in supplemental Table 2 [see Additional file 2]. Indeed, many ESTs showed similarity to proteins required for photosynthesis (66 expressed genes), nucleotide synthesis (6), nucleotide sugar conversion, the biosynthesis of precursors of scale polysaccharides (6), heme and chlorophyll biosynthesis (6), fatty acid and lipid biosynthesis (9), terpenoid biosynthesis (6), glycolysis (11) and the TCA-cycle including pyruvate dehydrogenase and respiration (12). The biosynthetic pathways for several amino acids were also well represented in our ESTs (21 expressed genes for Ala, Arg, Gly, Ile, Leu, Lys, Pro, Ser, Thr, Trp and Val). However, for several other amino acids (Asn, Asp, Cys, Gln, Glu, His, Met, Phe, Tyr) we did not find a single EST which could be matched to the known biosynthetic pathways.
<p>Additional file 2</p>
Supplemental Table 2
Click here for file
All enzymes except one (triose isomerase) of the Calvin cycle are represented by at least one EST. Interestingly, we found several genes coding for subunits of the plastidic GAPDH. In angiosperms the plastidic GAPDH consists of an A2B2 heterotetramer 24. Compared to GAPDH A, which is present in the plastids of all eukaryotic algae, GAPDH B has a C-terminal extension that contains the two conserved cysteine residues, which are required for regulation by the thioredoxin system. To our knowledge, GAPDH B has only been reported from streptophytes. Two genes of Mesostigma showed significant similarity to GAPDH B from angiosperms. We present an alignment of the C-terminus of Mesostigma GAPDH B with the C-terminus of spinach GAPDH B in Figure 3. The two sequences are very similar and the two cysteines required for regulation by the thioredoxin system are conserved in Mesostigma indicating that the activity of plastidic GAPDH came under the control of the thioredoxin system early during the evolution of streptophytes. We found no evidence for a GAPDH B in Chlamydomonas or other chlorophytes. Therefore, the evolution of a GAPDH B might represent a molecular characteristic (synapomorphy) of the streptophytes.
<p>Figure 3</p>
Alignment of the deduced amino acid sequence of the putative GAPDHB (Meso2a42g12) gene from Mesostigma with spinach (P12860) GAPDHB
Alignment of the deduced amino acid sequence of the putative GAPDHB (Meso2a42g12) gene from Mesostigma with spinach (P12860) GAPDHB. The conserved cysteine residues are indicated in red letters. Numbers refer to the amino acid position (spinach) or nucleotide position (Mesostigma).
![]()
A total of 25 expressed genes encode components of the light-harvesting complex. There are some light-harvesting complex proteins, which Mesostigma shares only with the chlorophytes and red algae (e.g. so called fucoxanthin/chlorophyll a-binding proteins). For others, we detected similar proteins only within embryophytes. However, the lhc proteins form a large superfamily and their phylogenetic analysis is beyond the scope of this study.
Several genes encode proteins of the photorespiratory C2-cycle (glycolate phosphatase, peroxisomal glycolate oxidase, a component of the glycine decarboxylase enzyme complex, and a peroxisomal serine-glyoxylate transaminase). As in embryophytes, the NADH required for reduction of hydroxy pyruvate is produced by a peroxisomal NADH malate dehydrogenase.
A glycolate oxidase activity was never detected in chlorophytes by biochemical enzyme assays, but one Chlamydomonas protein is currently annotated as a glycolate oxidase (gene model C_340068, JGI Chlamydomonas reinhardtii v2.0) We therefore performed a phylogenetic analysis for glycolate oxidases and lactate dehydrogenases, which are both members of the same protein superfamily, from embryophytes, Mesostigma, Chlamydomonas, Cyanidioschyzon, Dictyostelium, a few metazoans and some bacteria (Fig. 4). The glycolate oxidases from embryophytes, Mesostigma and Cyanidioschyzon are monophyletic. In contrast, the glycolate oxidase-like sequence from Chlamydomonas clusters with bacterial sequences, which are annotated as lactate dehydrogenase and glycolate oxidases. Therefore, we conclude that, in agreements with the biochemical findings, Chlamydomonas does not contain a plant-type peroxisomal glycolate oxidase.
<p>Figure 4</p>
Phylogenetic tree of glycolate oxidase and glycolate oxidase-like genes
Phylogenetic tree of glycolate oxidase and glycolate oxidase-like genes. The tree shown was derived by Bayesian inference analysis from 402 amino acid positions using a mixed model for amino acid substitutions and a gamma correction for rate variation among sites. The Bayesian inference utilized MRBAYES, Ver. 3.0 * with posterior probabilities derived from 100000 generations and discarding a burnin of 1000. The tree obtained with a parsimony analysis using PHYLIP gave essentially the same topology.
![]()
We did not find evidence for a hexokinase and sucrose biosynthesis in interphase cells of Mesostigma. Several ESTs represent plastidic pyruvate kinase, however, only a single EST coded for the cytosolic isoform. Expressed genes for PEP carboxylase and a cytosolic malate dehydrogenase are present, suggesting that malate may be the major substrate for respiration in the mitochondrion of Mesostigma as in many embryophytes. The plastidic pyruvate kinase probably functions in the generation of acetyl-CoA required to sustain fatty acid synthesis in plastids.
Scales consist mainly of the 2-keto sugar acids 3-deoxy-manno-octulosonic acid (2-keto-3-deoxy-oktonate, kdo), 5OMekdo, 3-deoxy-lyxo-heptulosaric acid, dha) and gal, galA, gul and some minor monosaccharides 25. Expressed genes coding for kdo synthesis, and activation of kdo as CMP-kdo are present. The obtained sequence similar to a CMP-sialA transporter might actually be the CMP-kdo transporter necessary for uptake of CMP-kdo into the Golgi apparatus, as kdo and sialA are structural analogs. Interestingly, kdo-synthase and CMP-kdo-transferase are among the most conserved proteins between Mesostigma and the embryophytes. As in embryophytes 26, galA is synthesized via the UDP-glc dehydrogenase pathway and the myo-inositol oxygenase pathway. We could not detect the latter enzyme in Chlamydomonas or red algae.
Our EST-data support the presence of vitamin B12-biosynthesis and the production of a phosphagen phosphoarginine by arginine kinase in Mesostigma.
Evolution of metabolism and cell structure
259 expressed genes from Mesostigma showed similarity to proteins belonging to various metabolic pathways. A pair-wise comparison of these genes with the genome of Chlamydomonas and the genomes and ESTs of the three embryophytes showed that Mesostigma shares more metabolic genes with the embryophytes than with Chlamydomonas, however, the overall AI is slightly higher with Chlamydomonas than with any embryophyte (AI, Table 6). Statistical analyses showed that the differences in AIs for the total metabolic enzyme data set are not significant (not shown). However, if we calculate AIs for different functional categories separately, we see that metabolic enzymes of the chloroplasts and mitochondria (photosynthesis except the Calvin cycle enzymes, fatty acid synthesis, synthesis of some amino acids, citric acid cycle, and respiration) were generally more conserved between Mesostigma and Chlamydomonas than between Mesostigma and the embryophytes (Table 6). In contrast, proteins of cytosolic pathways (nucleotide metabolism, NDP-sugar metabolism, and glycolysis) in Mesostigma were more similar to embryophyte proteins (Table 6),
<p>Table 6</p>
Comparison of Mesostigma genes related to metabolic functions with Chlamydomonas and three embryophytes. The average identity (AI) of pair-wise comparisons of Mesostigma expressed genes coding for the indicated metabolic function with the ESTs or genome of the given organisms are presented.
Function1
Chlamydomonas
Physcomitrella
Arabidopsis
Oryza
Metabolism (259)
0.613
0.601
0.593
0.594
Plastidic metabolism (107)2
0,595
0,572
0,567
0,572
Mitochondrial metabolism (16)2
0,647
0,593
0,569
0,578
cytosolic metabolism (glycolysis, NDP-sugar metabolism, nucleotide synthesis, 22)2
0,568
0,639
0,641
0,632
1 Numbers in brackets indicate the number of genes in this category.
Genes coding for information storage and processing, and cellular processes and signaling (Table 3) were overall more conserved between Mesostigma and the embryophytes than between Mesostigma and Chlamydomonas. Exceptions to this rule are proteins of the cytoskeleton (Table 7) and proteins involved in protein folding (chaperones, Table 7) and plastidic proteases (not shown), which show higher AI values with Chlamydomonas than with the embryophytes. If the cytoskeletal proteins are removed from the data set, the differences between Mesostigma/Chlamydomonas genome and Mesostigma/embryophytes are statistical significant (p = 0.000109 for Mesostigma/Chlamydomonas versus Mesostigma/Physcomitrella; p = 0.000703 for Mesostigma/Chlamydomonas versus Mesostigma/Arabidopsis, p = 0.006937 for Mesostigma/Chlamydomonas versus Mesostigma/Oryza). Remarkably, the three embryophytes behave differently in our analysis. We obtained higher AI values with Physcomitrella regarding the categories protein folding (chaperones), vesicular transport, transcription, and regulation (Table 7). In contrast, proteins related to DNA structure, replication, cell cycle and RNA-metabolism were more conserved between Mesostigma and the angiosperms Arabidopsis and Oryza than between Mesostigma and Physcomitrella (Table 7).
<p>Table 7</p>
Comparison of Mesostigma genes related to cell structure functions with the genome or ESTs of Chlamydomonas and three embryophytes. The average identity of pair-wise comparisons of Mesostigma expressed genes coding for the indicated cellular functions with the ESTs or genomes of the given organisms are presented.
Function1
Chlamydomonas
Physcomitrella
Arabidopsis
Oryza
Cell structure (201)
0.629
0.641
0.627
0.618
Cytoskeleton (7)
0.751
0.729
0.0.731
0.731
Protein folding/Chaperones (21)
0.685
0.605
0.580
0.581
Cytosolic protein degradation (22)
0.682
0.733
0.722
0.703
Vesicular transport (22)
0.633
0.701
0.680
0.673
Regulation (18)
0.585
0.599
0.548
0.544
DNA structure, replication, cell cycle (21)
0.580
0.616
0.655
0.625
Transcription (16)
0.684
0.675
0.647
0.623
RNA metabolism (23)
0.599
0.619
0.637
0.637
1 Numbers in brackets indicate the number of genes in this category.
Discussion
In this study, we have analyzed about 3000 expressed genes from the scaly green flagellate Mesostigma viride. We compared the expressed genes with the complete genomes from the angiosperms Arabidopsis thaliana and Oryza sativa, the chlorophyte Chlamydomonas reinhardtii, the red alga Cyanidioschyzon merolae and the diatom Thalassiosira pseudonana, as well as the ESTs from the moss Physcomitrella patens, and the red alga Porphyra yezoensis. Altogether, the Mesostigma proteome is more similar to the embryophytes than to Chlamydomonas, although Mesostigma and Chlamydomonas are both flagellate unicells. Mesostigma shares more genes with the embryophytes than with Chlamydomonas, including several enzymes confined to the streptophytes (e.g. GAPDH B, [Cu-Zn] superoxide dismutase), and the average identity of shared proteins is higher between Mesostigma and the embryophytes than between Mesostigma and Chlamydomonas. Therefore, we consider Mesostigma to be a member of the streptophytes, although Mesostigma clearly shares some ancestral characters with chlorophytes. Plastidic (with the exception of the Calvin cycle) and mitochondrial functions e.g. seem to be more conserved between Mesostigma and chlorophytes than between Mesostigma and embryophytes, i.e. these functions are more derived in embryophytes, probably due to adaptation of embryophytes to the terrestrial habitat. In contrast, other cellular functions except for the cytoskeleton are more conserved between Mesostigma and embryophytes than between Mesostigma and Chlamydomonas. Interestingly, in previous phylogenetic analyses plastidic and mitochondrial genes failed to show a clear relationship between Mesostigma and the streptophytes 1415, whereas actin and nuclear-encoded SSU rDNA phylogenies support the notion that Mesostigma is a member of the streptophytes 101112. The different evolutionary rates for different cellular functions observed in this study might explain this discrepancy.
We calculated the average identity (AI) values from automatically generated BLAST output alignments. Automatically derived alignments are prone to errors. However, we believe that our approach is justified for the following reasons: (1) the BLAST alignments cover only the conserved parts of proteins and our calculated AI values indicate that in most alignments more than half of the amino acids are identical enhancing the quality of the automatically produced alignments; (2) although small mistakes may occur, they are insignificant given the high number of amino acids used to calculate the AI. On average the BLAST alignments contained about 150 amino acids and therefore about 45,000 amino acid positions were used in the constrained data set. In large data sets small unbiased errors become irrelevant 27. Our results indicate that at least 100 (better are 150–200) expressed genes have to be used to obtain statistically significant results. It could be argued that our analysis uses only similarity values and no real evolutionary distances. AI values can be easily converted into evolutionary distances using an approximation given by Kimura 28, with the effect that the differences between the various organisms become larger but no changes occur in the order of relatedness (included in Table 4). We conclude that the AI of proteins shared between different organisms represents a reasonable measure of evolutionary relatedness, if sufficiently large data sets are used.
In the following, we briefly discuss some major differences in coding potential observed between the different photosynthetic eukaryotic organisms.
11 of 18 proteins included in supplemental Table 1 [see Additional file 1] which are shared only by Mesostigma and Chlamydomonas are associated with flagellar functions such as axonemal dyneins or components of the IFT (intra-flagellar transport) machinery. Most likely, the angiosperms lost these proteins during evolution together with the ability to produce flagellate cells. The absence of these proteins in the ESTs from the moss Physcomitrella, is presumably due to the fact that ESTs from developing spermatozoids are not available.
Proteins shared by Mesostigma and the embryophytes but not present in chlorophytes perform diverse functions. There are some well known biochemical differences between chlorophytes and streptophytes such as the presence of (Cu-Zn) superoxide dismutase 2930 and glycolate oxidase in streptophytes 3132 but not in chlorophytes. In addition, streptophytes use the DXP and mevalonate pathways for isoprene biosynthesis whereas chlorophytes posses only the DXP pathway 33. For all these functions, we find molecular support in our expressed gene data set except for the mevalonate pathway of isoprene biosynthesis. Two genes matched two different enzymes of the DXP pathway; however, no matches for the MVA pathway were obtained, although the presence of this pathway has been demonstrated biochemically 33. This could be due to the selective expression of one or the other pathway under different environmental conditions.
Remarkably, our list of proteins uniquely shared by Mesostigma and the embryophytes includes several proteins involved in steroid biosynthesis (e.g. a 3-oxo-5-beta-steroid dehydrogenase and a C-4 sterol oxidase), a homeobox protein of the knox family and proteins of the F-box family. The latter protein family underwent a dramatic expansion in the embryophytes (Arabidopsis has more than 700 members of this family).
Our expressed protein data set contains sequences similar to a protein involved in vitamin B-12 metabolism (present in rhodophytes and chlorophytes), an arginine kinase and a ARL6 protein, the latter two are absent in chlorophytes, embryophytes and red algae. It has been shown that arginine kinase is part of the ATP regeneration system in cilia of Paramecium 34. Chlamydomonas lacks arginine kinase and recently Pazour et al. 35 showed that enzymes of the late glycolytic pathway are present in the flagella of Chlamydomonas, suggesting that the ATP required for flagellar function is produced by the glycolytic pathway in Chlamydomonas. The ARL6 protein has been implicated in protein translocation at the rER 36, although its exact function is still not known.
There are some typical embryophyte pathways that we failed to detect in Mesostigma, e.g. sucrose metabolism, hexokinase, and enzymes of cellulose biosynthesis. There are no reports about the presence of sucrose metabolism and hexokinase in green algae in the literature, whereas embryophyte-like Ces genes (catalytical subunit of cellulose synthase) have been reported in the streptophyte alga Mesotaenium 37. Although we cannot exclude that Mesostigma lost these genes, we do expect to find theses genes in the genome of Mesostigma.
Evolution of photosynthesis and photorespiration
It is well known that embryophytes and chlorophytes differ in important aspects of photosynthesis and its regulation, and in photorespiration (e.g., presence of GAPDHB, number of enzymes regulated by thioredoxin, glycolate oxidase vs. glycolate dehydrogenase, and presence or absence of (Cu-Zn) superoxide dismutase).
Table 8 summarizes the available information on the regulation of plastidic proteins by the thioredoxin system. The number of thioredoxin-regulated proteins has apparently increased during evolution and Mesostigma in this respect most closely resembles the embryophytes. Similarly, the peroxisomes of Mesostigma have been biochemically characterized as "leaf-type peroxisomes" 38 in full agreement with our EST-data. In contrast, chlorophytes lack glycolate oxidase and photorespiration involves only chloroplast and mitochondrial enzymes 38. Interestingly, red algae possess a peroxisomal glycolate oxidase whereas the other enzymes of the photorespiratory cycle are located in the mitochondrion 32. Thus, it seems likely that at the onset of streptophyte evolution major changes occurred in the regulation of the Calvin cycle and the subcellular organization of photorespiration. What might have been the driving force for these changes? We note that rhodophytes and chlorophytes both presumably evolved in a marine environment [red algae in a coastal benthic habitat, whereas chlorophytes proliferated as marine phytoplankton 39]. Streptophyte algae most likely originated in a freshwater/brackish environment. In contrast to their marine counterparts, they had to deal with much higher light intensities and fluctuating environmental conditions such as salinity and temperature. With higher temperature, the rate of photorespiration increases. The observed changes in regulation of the Calvin cycle and photorespiration might be adaptations to this stress. It is possible that these adaptations to a shallow freshwater/brackish environment prepared streptophytes to colonize the terrestrial habitat later during evolution. In this respect we note that in extant chlorophytes activation of carbon concentrating mechanisms (CCM) is the dominant reaction to compensate for increased photorespiratory losses 38. In contrast, streptophytes are able to channel large amounts of glycolate through the photorespiratory cycle 38. According to Badger and Price 40 CCMs did not evolve until 400 million years ago, long after streptophytes had evolved and the colonization of the terrestrial habitat by streptophyte algae took place. Therefore during the palaeozoic era with reduced CO2- and increased O2-levels 40 streptophyte algae might have had an advantage over chlorophyte algae allowing them to colonize the terrestrial habitat during that time.
<p>Table 8</p>
Regulation of plastidic enzymes by the thioredoxin system. Proteins similar to embryophyte plastidic thioredoxin-regulated proteins were identified in the genomes of Cyanidioschyzon, Chlamydomonas, and the ESTs of Mesostigma using the BLASTP or BLASTX algorithms. A putative thioredoxin-regulated orthologue as revealed by the conserved cysteine residues is indicated with +. An asterisk indicates putative cyanobacterial/plastidic proteins, which do not contain the conserved cysteines required for thioredoxin-regulation. Missing enzymes are indicated with -.
Cyanobacteria
Rhodophytes Cyanidioschyzon
Chlorophytes Chlamydomonas
Mesostigma
Embryophytes
PRK
+
+
+
+
+
SDPase
*
+
+
+
+
G6PDH
*
+
+
n.d.
+
FBPase
*
*1)
+
+
+
γ-ATPase
*
*
+
+
+
GABDHB
-
-
-
+
+
NADP-MDH
-
-
+2)
n.d.
+
Rubisco activase
-3)
-
*
*
(+)4)
n.d. not detected in Mesostigma. 1) In Galdieria (Cyanidioschyzon) 2 (1) of the 3 conserved cysteines occurring in the Viridiplantae are present [48]. 2) Chlorophyte NADP-malate dehydrogenase possesses a C- and N-terminal extension like the embryophyte enzyme, however only the C-terminal cysteines of the embryophyte enzyme are conserved [49, 50]. 3) A few cyanobacteria contain an unusual rubisco activase. Only the central AAA+ domain shows similarity to plant rubisco activases, whereas the N and C terminal domain are very different [51]. 4) Many angiosperms contain two forms of rubisco activase. Only the long form is regulated by the thioredoxin system [52].