Upon the completion of the Arabidopsis thaliana genome project in December 2000, it became possible to identify all genes encoding cytochromes b561 in a plant. Four putative b561 genes were identified in A. thaliana: Artb561-1 to -4 (for Arabidopsis thaliana cyt b561) 7. Genomic sequences possibly encoding cytochromes b561 were also identified in Oryza sativa and Craterostigma plantagineum. Most putative b561-encoding sequences from other plant species, however, are so far only represented by expressed sequence tags (ESTs).

Genomic sequences for cytochromes b561 are also known for a number of animal species, including Drosophila melanogaster, Caenorhabditis elegans, Homo sapiens (human) and Mus musculus (mouse). The genomes of human and mouse each encode three cytochromes b561: one involved in ascorbate regeneration inside chromaffin granules (Hosb561-1, Mumb561-1), one present in the duodenum (Hosb561-2, Mumb561-2) that may function as a ferric reductase in the plasma membrane of duodenal mucosa, and a third that seems to have a ubiquitous distribution in mammalian tissues, but whose physiological function is not known. This is tentatively referred to as 'ubiquitous cytochrome b561' (Hosb561-3, Mumb561-3) (H.A., unpublished work).

All the A. thaliana sequences have a similar organization, with four exons of comparable lengths (Artb561-1 is shown in Figure 1). The intron-exon structure of cytochrome b561 genes identified in O. sativa and C. plantagineum is similar (Figure 1). Mammalian genes coding for the duodenal cytochrome b561 isoform also have a similar organisation as their plant homologues, with four exons and three introns (Figure 1). Interestingly, however, genes encoding the two other mammalian b561 cytochromes have five exons (Figure 1). Cytochrome b561 genes with a different intron-exon structure within one species are also present in C. elegans. One of the C. elegans cytochrome b561 proteins (F55H2.5) is encoded by a gene (Caeb561-1) consisting of only two exons and one intron (Figure 1), while the two other genes in this nematode (F39G3.4 and F39G3.5; Caeb561-2 and -3 respectively) each have four exons. Similarly, two homologous genes have been identified in the fruit fly D. melanogaster (CG1275 and CG8776). The first has two different splicing variants, each with four exons, and the second gene has either five or six exons, also depending on splicing.

The presence of five exons in the mammalian chromaffin granule cytochrome b561 has been considered to support a five transmembrane helix structure for these proteins 19. However, the occurrence of cytochrome b561-encoding genes with very different gene structures indicates that this one-helix-one-exon model is probably not valid for all members of this protein family. Moreover, secondary-structure prediction routines almost invariably indicate the presence of six transmembrane helices for each of the identified putative cytochrome b561 genes. The correlation between predicted transmembrane structures and intron-exon structure is presented in Figure 2 for three different A. thaliana genes (Artb561-1, Artb561-2, Artb561-4), a homolog from rice (Orsb561) and the human chromaffin cytochrome b561 (Hosb561-1). This analysis indicates that individual transmembrane helices may be encoded by different exons. For example, the predicted transmembrane helices 4 in Artb561-2, helix 6 from Artb561-4 and helix 4 in Hosb561-1 are each encoded by exons 3 and 4 (Figure 2). In rice, transmembrane helix 5 is partially encoded by exon 2 and exon 3. A similar situation, in which single transmembrane helices are encoded by more than one exon, was observed for the plasma membrane ATPase (BT002855) and an A. thaliana sodium channel (AY113938) (data not shown). The number of full transmembrane helices encoded by individual exons also varies among the genes. In the different cytochrome b561-encoding genes in A. thaliana, the first exon respectively encodes one, two or three helices (Figure 2).

Alternative splicing of cytochrome b561-encoding genes is observed in D. melanogaster. Different transcripts are found for each of the two putative genes (CG1275 and CG8776). The first gene has two transcripts (Drmb561-1 and -2), each encoded by four exons. Drmb561-1 is only different from Drmb561-2 in that it has 86 extra amino acids at the amino terminus. This extra region (predicted to be cytoplasmic) is absent in all other cytochrome b561 proteins identified thus far. The second gene from D. melanogaster has three different transcripts, encoded by five (Drmb561-3) or six (Drmb561-4 and -5) exons. Again, the only difference is seen at the amino terminus of the encoded proteins. Drmb561-5 is nearly identical to Drmb561-3, with only a difference in 20 amino-terminal amino acids. The sequence of the Drmb561-4 protein corresponds to that of the other two proteins, except for the lack of 80 amino acids at the amino terminus. At the protein level, Drmb561-4 thus proves to be the shortest isoform, lacking the first transmembrane helix. All six helices are present in the other two isoforms.

The only cytochrome b561 protein that has so far been purified and sequenced, is the bovine chromaffin granule cytochrome b561 1415. On the basis of sequence similarity to this protein, a large number of putative cytochrome b561 protein sequences have been identified from EST sequences obtained from a large variety of species 7. The similarity between sequences from different organisms is usually not high at the DNA level. However, at the protein level well-conserved features are apparent. In Figure 3 a selection of cytochrome b561 protein sequences from phylogenetically diverse species is aligned. As reported by 17, and as apparent from Figure 1, the conservation in the first and sixth transmembrane helix of the cytochromes b561 is limited to the overall hydrophobicity. This observation suggests that these helices may be primarily involved in protein folding and stability, rather than in the catalytic activity.

In addition to the conservation of the predicted transmembrane structures, four His residues suggested to bind the two heme groups are strictly conserved in all cytochrome b561 homologs. Furthermore, predicted binding sites for MDHA and ascorbate - as suggested by 18 - are conserved. A potential MDHA-binding motif - xYSLHSWxGx - with x being a hydrophobic amino acid in most proteins, is highly conserved. In the Anopheles gambiae (malaria mosquito) b561 protein, Angb561, the Ser is conservatively replaced by Thr. The Tyr in the consensus sequence is not conserved in the proteins from C. elegans - Caeb561-1, -2 and -3 - nor in Artb561-3, but it is present in all Artb561-3 orthologs identified in other plant species (data not shown).

The suggested ascorbate-binding site (ALLVYRVFR in the mammalian chromaffin cytochrome b561 18) is almost perfectly conserved in all mammalian proteins, but the degree of conservation is much lower in plants (Figure 3). The first three or four amino acids of this consensus sequence are generally hydrophobic, and the Tyr is found in all proteins except Artb561-4, in which it is replaced by His. The triplet codons for Tyr and His differ by only one nucleotide, suggesting that a single point mutation may be responsible for this substitution. The His at this position is found in all Artb561-4 orthologs from other plant species (data not shown), and therefore seems a consistent substitution in this isoform of plant cytochromes b561. The Arg from this motif is conservatively replaced by Lys in all available plant sequences. To the best of our knowledge the sequence requirement for ascorbate and MDHA binding has not been experimentally confirmed, and so it is not possible to evaluate whether the observed conservation is sufficient for the functional interaction with the ascorbate-MDHA redox couple.

In addition to the conserved structural features mentioned above, all cytochromes b561 show strict conservation of five Gly, two Pro, one Lys and one Gln (green in Figure 3). A number of aromatic residues are also well conserved (yellow in Figure 3). These have been suggested to take part in the electron transfer between the two heme molecules, possibly by electron tunneling 17.

When all available cytochrome b561 protein sequences - including those derived from ESTs - are aligned with Clustal W 20, and subsequently organized in a phylogenetic tree (Figure 4), two main clusters containing the plant and animal cytochromes b561 can be discriminated. However, the position of the Artb561-3 orthologs in the plant cluster is not supported by high bootstrap values (less than 80%).

A closer examination of the plant group reveals the clustering of sequences from Lycopersicon esculentum (tomato, Lyeb561), L. hirsutum (Lyhb561), Solanum tuberosum (potato, Sotb561), C. plantagineum (resurrection plant, Crpb561), Beta vulgaris (beet, Bevb561-1), Populus tremula (poplar, Potb561), Medicago truncatula (barrel medic, Metb561-1), Oryza sativa (rice, Orsb561), Zea mays (maize, Zemb561), Hordeum vulgare (barley, Hovb561-1), Sorghum bicolor (sorghum, Sobb561), Triticum aestivum (wheat, Trab561-1) and T. monococcum (Trmb561) with the Artb561-1 protein, suggesting that these proteins are orthologs of Artb561-1 from A. thaliana. The localization of a cytochrome b561 from Pinus taeda (loblolly pine, Pitb561) within this cluster is not supported by a bootstrap value above 80% and should be interpreted with caution. Similarly, sequences from T. aestivum (Trab561-2), H. vulgare (Hovb561-2), M. truncatula (Metb561-2), Glycine max (soybean, Glmb561-2), Zinnia elegans (zinnia, Zieb561) and B. vulgaris (Bevb561-2) are clustered with Artb561-2. Orthologs of Artb561-3 have been identified in Mesembryanthemum crystallinum (ice plant, Mecb561-3), M. truncatula (Metb561-3) and Gossypium hirsutum (cotton, Gohb561), and Artb561-4 orthologs in M. truncatula (Metb561-4) and B. vulgaris (Bevb561-4).

Most of the plant cytochrome b561 sequences so far have been obtained from dicotyledonous plants, but for Artb561-1 and -2 some orthologs from monocotyledons are available. These form separate 'sub-clusters' within the clusters of Artb561-1 and -2 respectively, whereas an Artb561-1 ortholog from the gymnosperm P. taeda is more distantly related with these clusters. For Artb561-3 and -4 no orthologs have been identified in monocotyledonous plants so far. Within the animal group, all known chromaffin granule cytochromes b561 - from H. sapiens (human, Hosb561-1), M. musculus (mouse, Mumb561-1), Sus scrofa (pig, Susb561), Bos taurus (bovine, Botb561) and Ovis aries (sheep, Ovab561) are clustered together with the homolog from Xenopus laevis (African clawed frog, Xelb561). The duodenal cytochromes b561 from humans, mouse and rat (Hosb561-2, Mumb561-2 and Ranb561-2) form a separate group, as do the chromaffin granule cytochromes b561 and the third type of mammalian cytochrome b561 ('ubiquitous' cytochrome b561 from humans and mouse, Hosb561-3 and Mumb561-3). The position of the cytochrome b561 protein from Ciona intestinalis (Ciib561; sea squirt, a tunicate) is uncertain, as the bootstrap value is below 80%. The cytochrome b561 homologs identified in C. elegans and the flatworm Dugesia japonica could not reliably be associated with any of the clusters.

The outcome for the phylogenetic tree was essentially the same when T-Coffee 21 or POA 22 software packages were used, which are in some cases considered more reliable than Clustal W 23. Similarly, using only the central cytochrome b561 'core domain' as defined by Ponting 24, instead of the full sequence, the clustering of cytochrome b561 homologs is identical to that presented in Figure 4, and the overall appearance of the resulting tree is not altered (Figure 5).

To examine the selective pressure on cytochrome b561-encoding genes after various gene duplication events, a Li93 analysis was performed using the DAMBE software package 2526, based on pairwise comparisons. The amount of synonymous substitutions (K_s in Table 1) is apparently saturated when comparing Artb561-3 to the three other paralogous genes from A. thaliana, indicating that Artb561-3 may represent an early gene duplication event, and has accumulated a large number of 'silent' substitutions. Nonsynonymous substitutions (that is, codon alterations leading to amino acid replacements, K_a in Table 1) for these genes range from 0.49 to 0.75 (substitutions/position), which is much lower than the K_s values. Artb561-2 and -4 have the lowest K_s (0.47), suggesting that they originated from a more recent gene duplication in A. thaliana. Comparable values are found for the orthologs of Artb561-2 and -4 in B. vulgaris and M. truncatula, indicating that this last gene duplication had occurred well before the radiation within the dicotyledons (data not shown). In mouse, K_s values are equally high (and probably saturated) for the genes encoding the three cytochrome b561 isoforms. K_a values are also comparable to those of the different plant isoforms (Table 1).

Comparison of cytochromes b561 from dicotyledonous species to their orthologs from monocotyledons reveals that K_s and K_a values are generally both around 0.3 to 0.5. Knowing that both clades diversified 130 million years ago 27, the average rate of synonymous and nonsynonymous substitutions is calculated to be around 3 × 10^-9 substitutions per position per year (Table 1). This is comparable to the synonymous evolution rate of other nuclear-encoded plant genes (5-30 × 10^-9 substitutions/position/year 28). Among dicotyledons (using estimated divergence times for different families, derived from Figure 2 in 29) the average synonymous substitution rate is higher (12.63 ± 2.41 × 10^-9 substitutions/position/year), but the nonsynonymous rate is very similar (2.38 ± 1.18 × 10^-9 substitutions/position/year). For the resurrection plant (C. plantagineum), however, which has adapted to life in an extremely dry environment, a particularly high nonsynonymous substitution rate was found as compared to the ortholog from potato (13.25 × 10^-9 substitutions/position/year), whereas the synonymous rate was the same as for the orthologs in other dicots (Table 1).

K_s and K_a were also calculated for animal cytochrome b561 genes. Comparing the gene from the tunicate C. intestinalis (Ciib561), which branched off the vertebrate lineage around 540 million years ago 30, with the chromaffin granule cytochrome b561 genes from vertebrates (X. laevis, H. sapiens and M. musculus) gives a K_s of around 0.65 and a K_a of 0.8. This is remarkably low, considering the large phyletic distance between tunicates and vertebrates, and implies synonymous and nonsynonymous substitution rates of 1-1.5 × 10^-9 substitutions/position/year.

The Gu99 likelihood ratio test in the DIVERGE software package 31 was used to examine whether proteins from different branches in the phylogenetic tree have different functional constraints, and whether they may have functionally diverged. The Theta ML value, indicative for the level of functional divergence between proteins 31, is relatively low (0.213) according to 32, when monocotyledonous and dicotyledonous Artb561-1 orthologs are compared. This suggests that these orthologs are likely to have similar biochemical functions. The Gu99 likelihood assay also indicated that a functional adaptation may have occurred at only five amino acids between these orthologs, supporting their functional similarity. A similar analysis of Artb561-1 and -2 orthologs yielded a theta ML value of 0.533, and 22 residues that may have functionally adapted. These numbers are still rather low 32. However, four of these residues (A, I, S and P) are situated in the predicted ascorbate-binding site (AIISYKSLP in Artb561-1). A similar analysis for the mammalian cyts b561 (chromaffin cyts b561 versus the cluster with the other two mammalian cyts b561) resulted in a very low theta ML of 0.031, and no residues likely to have an altered function.