Zebrafish reveals different and conserved features of vertebrate neuroglobin gene structure, expression pattern, and ligand binding.

Neuroglobin has been identified as a respiratory protein that is primarily expressed in the mammalian nervous system. Here we present the first detailed analysis of neuroglobin from a non-mammalian vertebrate, the zebrafish Danio rerio. The zebrafish neuroglobin gene reveals a mammalian-type exon-intron pattern in the coding region (B12.2, E11.0, and G7.0), plus an additional 5'-non-coding exon. Similar to the mammalian neuroglobin, the zebrafish protein displays a hexacoordinate deoxy-binding scheme. Flash photolysis kinetics show the competitive binding on the millisecond timescale of external ligands and the distal histidine, resulting in an oxygen affinity of 1 torr. Western blotting, immune staining, and mRNA in situ hybridization demonstrate neuroglobin expression in the fish central nervous system and the retina but also in the gills. Neurons containing neuroglobin have a widespread distribution in the brain but are also present in the olfactory system. In the fish retina, neuroglobin is mainly present in the inner segments of the photoreceptor cells. In the gills, the chloride cells were identified to express neuroglobin. Neuroglobin appears to be associated with mitochondria-rich cell types and thus oxygen consumption rates, suggesting a myoglobin-like function of this protein in facilitated oxygen diffusion.

Transport and storage of oxygen in vertebrate animals are typically mediated by globins, small proteins that bind O 2 by the means of a porphyrin-coordinated Fe 2ϩ ion (1)(2)(3). The heterotetrameric hemoglobin is present in red blood cells of nearly all vertebrates and transports O 2 in the circulatory system from the respiratory surfaces to the inner organs. The monomeric myoglobin, typically found in the myocytes of cardiac and striated muscles, facilitates intracellular O 2 diffusion to the mitochondria and stores O 2 (3,4) but also functions as an nitric-oxide dioxygenase (5). Neuroglobin (Ngb) 1 and cytoglobin are two recently discovered vertebrate globins (6 -9). Whereas cytoglobin might play a role in collagen synthesis (10,11), the leading hypotheses suggest Ngb to be involved in neuronal oxygen homeostasis (6,(12)(13)(14)(15)(16).
Ngb shares only a few amino acids with vertebrate hemoglobin and myoglobin (Ͻ25% identity) but rather resembles the nerve-specific globins known in some invertebrates (6,13). In fact, phylogenetic analyses suggest an ancient origin of Ngb, probably diverging from the other globins before the Protostomia-Deuterostomia split (6,8). Mouse and human deoxy-Ngb display hexacordinated hemochrome-binding schemes at the Fe 2ϩ (17,18). The proximal histidine can be replaced by an external ligand such as O 2 , resulting in an oxygen affinity (P 50 ) of ϳ1-2 torr, similar to that of myoglobin (6,17).
Initially, Ngb was found to be predominantly expressed in the brains of man and mouse (6), but recent analyses show Ngb to be also present in the peripheral nervous system and some non-neuronal endocrine tissues (19 -21). Particularly high amounts of Ngb were observed in the mammalian retina where it could reach a concentration similar to that of myoglobin in muscle cells (14). This study also suggests that, at the subcellular level, Ngb is localized adjacent to the mitochondria. In cultured neuronal cells from the mouse brain, synthesis of Ngb is enhanced under low oxygen conditions (12), although these results could not been confirmed by in vivo studies (22). The presence of Ngb promotes the survival of cultured neuronal cells at low oxygen levels, and overexpression of Ngb may protect neurons from hypoxic-ischemic injury in vitro and in vivo (12,15).
The available data suggest that Ngb expression levels correlate with high oxygen consumption rates, implying an important role of Ngb in neuronal O 2 supply, although other functions of Ngb are still conceivable (13,23). Whereas Ngb was originally identified in mammalian species (6), it is also present in fishes (24), suggesting a widespread occurrence in vertebrates. The investigation of Ngb function in non-mammalian species is an essential prerequisite for the understanding of its role in the vertebrate metabolism. Here, we have carried out a detailed genetic, biochemical, and expressional analysis of Ngb from the zebrafish Danio rerio.

EXPERIMENTAL PROCEDURES
Cloning and Sequencing of D. rerio Ngb cDNA and Gene-D. rerio were kept at 28°C in a freshwater aquarium. Genomic DNA of D. rerio was extracted using the Qiagen DNeasy kit. Several oligonucleotide primers were designed according to the Ngb cDNA sequence (24) and partial data base entries of the zebrafish Ngb gene. Overlapping fragments were amplified using the SAWADY "Mid Range" PCR system (PeqLab). The sequences were obtained after the cloning of the PCR products into the pCR4-TOPO vector (Invitrogen). 5Ј-RACE experiments were carried out with the Invitrogen kit according to the manufacturer's instructions using ϳ1 g of D. rerio total RNA and three nested oligonucleotide primers. Sequences were obtained either directly from the PCR products or after cloning into the pCR4-TOPO vector (Invitrogen). Sequences were analyzed by applying the GeneDoc 2.6 (25), MatInspector V2.2 (26), and BLAST searches (27) using Eukaryotic Promoter Data base, release 76 (28).
Expression and Purification of Recombinant Ngb-The complete D. rerio Ngb cDNA was cloned into the pET3a expression vector (Novagen) using PCR-generated NdeI and BglII restriction sites. The plasmid was transformed into the Escherichia coli BL21(DE3)pLys strain and grown at 25°C in TBY medium (0.5% NaCl, 1% tryptone, 0.5% yeast extract, pH 7.4) containing 100 g/ml ampicillin, 30 g/ml chloramphenicol, and 1 mM ␦-aminolevulinic acid. The culture was induced at A 600 ϭ 0.8 by isopropyl-␤-D-thiogalactopyranoside at a final concentration of 0.4 mM and grown overnight. The bacteria were harvested by a 20-min centrifugation at 4,322 ϫ g and resuspended in 50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 0.5 mM dithiothreitol, 8 g/ml DNase, and 4 g/ml RNase supplemented with Complete TM proteinase inhibitor mixture (Roche Applied Science) and Pefabloc (Roche Applied Science). The cells were broken by three freeze-thaw cycles in fluid nitrogen followed by ultrasonication (10 ϫ 30 s). The sample was incubated for 2 h at 37°C to digest the DNA and RNA. The cell debris was removed by a centrifugation for 1 h at 4°C at 10,000 ϫ g. The supernatant was fractionated by ammonium sulfate precipitation. The reddish 60 -70% ammonium sulfate pellet was dissolved in 5 mM Tris-HCl, pH 8.5, and desalted using an Amicon Ultra filter (Millipore). Further purification of Ngb was achieved by a DEAE ion-exchange column with a gradient of 0 -1 M NaCl in 5 mM Tris-HCl, pH 8.5. Size exclusion chromatography was carried out using a High Load TM 26/60 Superdex TM 75 prep grade (Amersham Biosciences). The final Ngb fractions were analyzed by gel electrophoresis, pooled, concentrated, and stored frozen at Ϫ20°C. Protein concentrations were determined using the Bradford (29) method.
Absorption Spectra and Ligand-binding Kinetics-The spectral forms were determined on a Cary400 or HP8453 spectrometer. The main characteristic differing from human hemoglobin is the deoxy form, which has the enhanced ␣ and Soret band characteristic of the hexacoordinated species. The CO-bound form can be photodissociated to follow the subsequent competitive rebinding of CO and the distal histidine. A 10-ns yttrium aluminium garnet laser at 532 nm was used for photodissociation. A monochromatic detection beam was used consisting of a 50-watt quartz halogen lamp with an interference filter.
Antibody Preparation and Western Blotting-Purified recombinant D. rerio Ngb was used to raise a polyclonal antibody in rabbits. Specific antibodies were affinity-purified from the crude rabbit serum using recombinant D. rerio Ngb coupled to a SulfoLink column (Pierce) according to the manufacturer's instructions. The antibody was stored at 4°C in 50 mM Tris, 100 mM glycine, ϳpH 7.4, supplemented with 0.1% NaN 3 . For Western blotting, selected tissues were homogenized in 1% SDS, 5% ␤-mercaptoethanol, 10% glycerol, and 65 mM Tris, pH 6.8. The samples were heat-denatured at 95°C for 5 min and loaded to a 14% SDS-polyacrylamide gel. Proteins were stained with Coomassie Brilliant Blue R-250. Antibody detection was carried out on protein samples transferred to nitrocellulose membrane for 2 h at 0.8 mA/cm 2 . Nonspecific binding sites were blocked by incubation for 2 h with 2% nonfat dry milk in TBST (10 mM Tris-HCl, pH 7.4, 140 mM NaCl, 0.3% Tween 20). The membranes were incubated with anti-Ngb antibodies diluted 1:2,000 in 2% milk/TBST. The membranes were washed four times for 10 min in TBST and incubated with the goat anti-rabbit antibody coupled with alkaline phosphatase (Dianova). The filters were washed in TBST as above, and detection was carried out with either nitro blue tetrazolium chloride or 5-bromo-4-chloro-3-indolyl-phosphate as substrates.
In Situ Hybridization-The DNA template for generating an Ngb antisense RNA probe was produced by PCR, attaching the T7 RNA polymerase promoter sequence to the 3Ј end of the Ngb cDNA. Digoxigenin-labeled antisense RNA was then generated by in vitro transcription using T7 RNA polymerase (Roche Applied Science). Control sense RNA probe was created using a cDNA carrying the T7 promoter sequence at the 5Ј end. Two specimens of D. rerio were prepared by freezing the animals on dry ice and decapitating them. Frontal cryo-sections (15 m) were mounted onto slides coated with aminosilane and stored at Ϫ80°C. Prior to hybridization, the sections were postfixed for 15 min in 4% formaldehyde/PBS and washed 2 ϫ 5 min in PBS, 0.1% Tween 20 and 5 min in PBS. Hybridization with labeled RNA probes was carried out at 42°C overnight in 50% formamide, 2ϫ SSC. Sense probes and hybridization mixtures completely lacking probe RNA were used as negative controls. Slides were washed for 3 ϫ 5 min at room temperature in 2ϫ SSC, 20 min in 0.1ϫ SSC at 60°C and subsequently treated for 15 min at 37°C with RNase A in 2ϫ SSC. After washing 3 ϫ 5 min with 2ϫ SSC and 5 min in 1ϫ PBS, 0.1% Tween 20, and 0.2% bovine serum albumin, nonspecific antibody binding sites were blocked with 1% blocking reagent (Roche Applied Science), 150 mM NaCl, and 100 mM Tris-HCl, pH 7.4. Specimens were incubated for 30 min at room temperature with an anti-digoxigenin antibody (Roche Applied Science) coupled with alkaline phosphatase (diluted 1:100 in PBS) and washed 2 ϫ 5 min in PBS, 0.1% Tween 20, and 0.2% bovine serum albumin and 5 min in 100 mM Tris/HCl, pH 9.5, 100 mM NaCl, and 50 mM MgCl 2 . Detection was carried out using nitro blue tetrazolium and 5-bromo-4chloro-3-indolyl phosphate. The reaction was stopped by washing in PBS, and preparations were mounted in 50% glycerol/PBS and examined with a Zeiss Axiophot microscope. Neuroanatomical regions were identified by the aid of the brain atlas by Wullimann et al. (30).
Immunofluorescence Studies-Tissues were fixed for 8 h or overnight in 4% paraformaldehyde in PBS (8 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , 140 mM NaCl, 2.7 mM KCl). Cryosections (12 m) were prepared and placed onto glass slides coated with chromalaun gelatin. Nonspecific binding sides were blocked for 15 min in 2% bovine serum albumin, 0.1% TritonX-100 in PBS. The sections were incubated with anti-Ngb antibodies (1:20 -1:50) in the blocking solution overnight at 4°C. The sections were washed 3 ϫ 8 min in PBS. Incubation with the secondary antibody coupled with Cy3 (Dianova), diluted 1:500 in blocking solution, was carried out for 2 h at room temperature in the dark. The sections were washed 3 ϫ 8 min in PBS, mounted in Elvanol polyvinyl alcohol (Mowiol, Calbiochem), and examined with a Leitz DM RD microscope.

RESULTS
The D. rerio Neuroglobin Gene-The nucleotide sequence of zebrafish Ngb cDNA has been determined previously (24). We designed multiple primers that allowed the amplification of overlapping fragments from the D. rerio Ngb gene. The sequences of the PCR fragments are essentially identical (Ͼ99%) with the sequence of clone CH211-233N24 from the zebrafish genome project (Sanger Institute). Within the Ngb coding region, only at the third nucleotide of codon 153, a silent C3 T transition was observed (Supplemental Fig. S1). Putative mRNA transcription start sites were determined by 5Ј-RACE experiments and deduced from the longest expressed sequence tags (fr71g03.y1, fr74a03.y1, and fr73f03.y1). Counting from the 5Ј most transcription start site to the poly(A) tail, the zebrafish Ngb gene covers a total of 6286 bp ( Fig. 1 and Supplemental Fig. S1). A comparison of the genomic sequences with the cDNA reveals the presence of four introns. The second, third, and fourth introns are positioned within the coding region at exactly the same positions as in the human and mouse neuroglobin genes (6) as follows: B12.2 (i.e. between codon positions 2 and 3 of the 12 th amino acid of the globin helix B); E11.0; and G7.0. An additional intron of 184 bp is located at bp 119 in the 5Ј-non-coding region, whereas the start ATG was found at position 446. Thus, the D. rerio Ngb mRNA covers 2640 bp, including a 5Ј-untranslated region of 261 bp and a 3Ј-untranslated region of 1899 bp.
The expressed sequence tags in the sequence databases show the presence of another transcribed region of unknown function 1.5 kb upstream to the Danio Ngb gene (data not shown), thus delimiting the 5Ј-regulatory sequences of the Ngb gene. This upstream region, putatively encompassing the Ngb promoter, was compared with the Eukaryotic Promoter Data base, which is an annotated collection of eukaryotic PolII promoters with experimentally determined transcription start sites (28). In this analysis, the upstream region revealed significant similarities with various PolII promoter regions. In particular, the promoters of mouse myoglobin and rabbit ␤-hemoglobin gave significant scores (E ϭ 2 ϫ 10 Ϫ13 and E ϭ 2 ϫ 10 Ϫ7 ), indicating the integrity of the predicted transcription start site. Further inspections of the 5Ј-upstream sequences of the zebrafish Ngb gene show that the putative promoter lacks a TATA box, whereas multiple other regulatory sequences such as GATAbinding sites, CCAAT/enhancer-binding protein, and CAAT box are present (Supplemental Table S1). Two putative neuronrestrictive silencer elements (NRSEs) were also identified in the Ngb promoter region (Fig. 1).
The Ngb gene plus the upstream and downstream genomic regions were scanned for the presence of putative hypoxiaresponsive sequence elements (HREs). HREs in hypoxia-regulated genes are defined by the binding motif of the hypoxiainducible transcription factor HIF-1 (5Ј-RCGTG-3Ј) (31). Typically, an HRE includes two HIF-1 motifs (in direct or inverted orientation) or one HIF-1 motif combined with an erythropoietin (EPO) box (32) or a HIF ancillary sequence (33). Five such closely spaced motif combinations are present in the zebrafish Ngb gene region (Fig. 1).
Ligand-binding Kinetics of Zebrafish Neuroglobin-The Danio Ngb was expressed using the pET3a vector system. The recombinant protein was purified by ammonium sulfate precipitation followed by ion-exchange and size exclusion chromatography (Supplemental Fig. S2). The ferrous deoxy form of Ngb shows a typical hexacoordinated absorption spectrum ( Fig.  2A) as characterized by large amplitudes for the ␣ band (560 nm) and the Soret band (426 nm) (17). The kinetics after CO photodissociation revealed a biphasic form. The rapid phase is competitive binding of CO or the distal histidine residue. For higher CO levels, the rate and relative amplitude of this phase increases. At low CO levels, the histidine binding becomes more competitive and the slow phase becomes more prominent. The slow phase involves the replacement of the histidine by CO and requires several seconds for a complete return to the initial CO-bound state. From the kinetics at various CO levels, one can extract the CO on-rate and the histidine both on-rate and off-rate (Table I). Similarly, competition of oxygen and CO can be used to obtain the rate coefficients for oxygen. Note that two ligand affinities are given in the table. The intrinsic affinity is the ratio of the binding rates for a given ligand. This would be the affinity in the absence of competing ligand. The overall oxygen affinity, taking into account the competing histidine ligand, is calculated to be ϳ1 torr, similar to human NGB without dithiothreitol (34). With dithiothreitol, the disulfide bond is broken and human NGB shows a lower oxygen affinity. This effect was not observed for Danio Ngb.
Detection of Ngb Protein in Zebrafish Tissues-Specific antibodies against recombinant zebrafish Ngb were raised in rabbits and further purified by affinity chromatography. In Western blotting, these antibodies detect the recombinant Ngb at ϳ16 kDa (Fig. 3A). Ngb was also stained by the antibody in extracts from the brain, the total eye, and the gills but not in the muscle or the blood. As already observed in mouse (14), the apparent molecular mass of the native Ngb is slightly higher than that of the recombinant protein. The reason for this discrepancy is still unknown but may be explained by posttranslational modifications. After preadsorption of the antibodies with the antigen, the Ngb band disappeared in Western blots of retinal and brain tissues, demonstrating the specificity of the anti-Ngb antibodies. To investigate whether Ngb may be released from the cells, the total brain was incubated for 2 h in PBS. However, no Ngb could be detected in the supernatant while there was a signal in the brain extracts (Fig. 3B).
Expression Pattern of Danio Ngb by mRNA in Situ Hybridization-To localize the sites of Ngb mRNA expression in zebrafish, frontal cryosections of head regions were analyzed by in situ hybridization (ISH) using an in vitro transcribed antisense RNA probe. Cross-sections from several layers were inspected (Supplemental Fig. S3). A perinuclear signal typical for mRNA hybridization of Ngb was found in the neuronal somata of essentially all of the regions of the Danio brain (Fig. 4) (see also Supplemental Fig. S4 and Supplemental Table S2). Control experiments that included nonspecific RNA (mouse intes-  tine total RNA) as competitor showed identical staining (data not shown). Sections hybridized with an in vitro transcribed sense RNA probe showed no staining (Supplemental Fig. S5), thus demonstrating the specificity of the Ngb labeling. For example, in a section of the posterior mesencephalon (Fig. 4), we found a particularly strong Ngb signal in the periventral zone of the tectum opticum. Additional staining was observed in putative neurons scattered in the white matter of the brain and in neuronal populations such as the preglomerular nucleus, posterior tubular nucleus, the torus longitudinalis, the hypothalamus, and the pituitary gland. An examination of further frontal sections showed the presence of Ngb mRNA in various brain regions (Supplemental Fig. S4 and Supplemental Table  S2), e.g. Ngb mRNA was also detected in brain regions of the visual system, predominantly in parts of the tectum opticum and the torus semicircularis, which are both part of the mesencephalon. ISH signal was also present in two other regions involved in visual signal processing, the area dorsalis telencephali of the telencephalon and the medulla oblongata, a part of the rhombencephalon. The other parts of the Danio brain in which strong ISH signals were observed included parts of the telencephalic area dorsalis telencephali, the diencephalic posterior tuberculum, the hypothalamus, and the synencephalon. In the rhombencephalon, which includes the metencephalon and the myelencephalon, signals could be seen in the formatio reticularis.
Ngb ISH staining was also noticed in the telencephalon, containing the bulbus olfactorius (Fig. 5A). Intense Ngb ISH signal could be detected in the sensory epithelium of the peripheral olfactory organ. Strong Ngb mRNA labeling was also observed in distinct layers of the zebrafish retina (Fig. 5B). The outer and inner nuclear layers, which contain the nuclei, and the ganglion cell layer were heavily stained with the Ngb probe, whereas the outer plexiform layer and the outer segments of the photoreceptor cells appeared to be unstained. The head cross-sections revealed that Ngb is also present in the gills (Fig. 5C). The ISH signal appeared to be limited to the chloride cells that were located mainly at the base and the lateral regions of the secondary lamella.
Immunodetection of Danio Ngb Protein-The presence of the Ngb protein in zebrafish tissues was monitored by indirect immunofluorescence experiments. An affinity purified ␣-Danio Ngb antibody was applied on cryosections from both fixed and unfixed tissues. In agreement with the mRNA ISH data, we observed scattered Ngb immune reactivity of the brain neurons. Particularly strong staining of neurons was observed in the diencephalic lateral hypothalamus (Fig. 6A). In control experiments without first antibody, no staining of the neurons was observed (Fig. 6B). In the retina, no anti-Ngb immune staining was found either in the cells of the retinal pigment epithelium or in the outer segments of the photoreceptors. Bright immune staining was observed in the photoreceptor layer, whereas there was little but detectable immune reaction in the outer and inner plexiform layers and at the ganglion cells (Fig. 6C). No signal was present in the appropriate control experiments (not shown). In the gills, we observed scattered staining of cells that most probably corresponded to the chloride cells of the secondary lamella (Fig. 6D). DISCUSSION Ngb is a recently identified member of the vertebrate globin superfamily (6) that is probably present in all of the vertebrate taxa including fish (24), amphibians, and birds. 2 Although most experiments support the idea of an important role of Ngb in oxygen homeostasis of nerve tissues (6,(12)(13)(14)(15)(16), other functions or additional functions of Ngb are still conceivable (13,16) FIG. 3. Western blot analysis of Ngb protein in selected zebrafish tissues. A, ϳ20 g of total protein form of selected tissue extracts were applied/lane. Ngb was detected in the brain, the gills, and the total eye but not in the muscle or blood. ϳ0.5 g of recombinant Ngb was applied as a positive control. B, detection of Ngb after incubation of the brain in PBS. No Ngb was detected in the supernatant while it was present in the brain tissue. Control, 2 g of purified Ngb. and many open questions remain to be solved. So far, biochemical and molecular studies on Ngb have been carried out exclusively with mammalian species. However, a comparative approach involving non-mammalian species offers a promising tool for the identification of conserved Ngb features and thus for the further understanding of Ngb function.
Conservation and Differences between Fish and Mammalian Ngb Genes-Ngb is a highly conserved protein with substitution rates that are ϳ3-fold lower than those in hemoglobin and myoglobin (6,24,35). This is also true for the gene structure, which contains three conserved introns in the coding regions of fish and mammalian Ngb genes. While the introns in B12.2 and G7.0 are present in all vertebrate and many invertebrate globin genes (13,36), the intron in E11.0 is unique to Ngb. The additional short intron in the 5Ј-untranslated region is most likely homologous to an intron at a similar position in the Tetraodon nigroviridis and Takifugu rubripes genes (data not shown). Such an intron is absent from the mammalian neuroglobins and may have been acquired early in fish evolution before the Ostariophysi (D. rerio) and the Acanthopterygii (T. nigroviridis, T. rubripes) separated ϳ120 million years ago (37,38).
Similar to mammalian Ngb (6, 35), the zebrafish gene lacks a TATA box, although several other transcription factor bind-ing sites were present (Supplemental Table S1 and Supplemental Fig. S1). In contrast to the mammalian genes, the putative Danio Ngb promoter region was not associated with a CpG island. The Danio Ngb promoter rather shared significant similarities with experimentally derived promoters from mouse myoglobin (39) and rabbit hemoglobin (40). However, it must remain uncertain whether this is a conserved feature retained from the common globin gene ancestor or has been acquired by convergent evolution. The hypoxia response of Ngb is of particular interest because it may shed light on neuroglobin function and its role in hypoxia-related diseases such as stroke. However, the available data are conflicting. Whereas Sun et al. (12) suggest an up-regulation of Ngb in cultured neuronal mouse cells under oxygen deprivation, Mammen et al. (22) did not find any reaction of Ngb to hypoxia in mouse in vivo. In fact, five putative HREs (31)(32)(33) were present in the Danio Ngb gene region (Fig. 1). However, a comparison of the zebrafish Ngb gene with that of T. nigroviridis and T. rubripes does not reveal any positional conservation of these regulatory elements (35). The actual function of the putative HREs in hypoxia regulation of the zebrafish Ngb genes requires experimental confirmation in future studies. In several mammalian genes, neuron-specific gene expression was mediated by a 21-bp sequence motif, the NRSE that binds to the neuron-restrictive silencer factor (NRSF) (41,42). Putative NRSE motifs have been identified in mouse and human Ngb genes (43), and two such elements were present in the 5Ј-promoter region of the Danio Ngb gene (Supplemental Table S1 and Supplemental Fig. S1). Although the regulatory function of these NRSEs remains to be established in both fish and mammalian Ngb, they possibly explain the largely neuron-specific expression of Ngb.
Fish and Mammalian Ngb Are Hexacoordinated Globins with Similar Ligand-binding Kinetics-Similar to the mammalian Ngbs (17,18) and various other animal and plant hemoglobins (44), deoxygenated zebrafish Ngb displayed a hexacoordinated hemochrome-binding scheme at the Fe 2ϩ of the porphyrin ring. The ligand-binding kinetics of zebrafish Ngb are similar to those of human Ngb, displaying the biphasic form for competitive binding of the internal His with the external ligand. The rate coefficients for Danio are approximately twice the rates of human Ngb, indicating a faster passage in the heme pocket. Because all of the rates are increased by the same factor, the overall observed affinity does not change significantly. As for mammalian Ngb, the Danio Ngb oxygen affinity is ϳ1 torr, suggesting that this parameter has been maintained throughout the evolutionary process. There is a major difference in the dependence on the kinetics on the possible disulfide bond. Although human NGB oxygen affinity decreased to ϳ10 torr (34), Danio Ngb showed little effect. Note that the first Cys in Danio was shifted by two positions relative to the human NGB sequence.
Neuronal Expression of Ngb in the Vertebrate Brain-As in the central nervous system of mammals (6,19,21), Ngb was expressed in brain neurons (Figs. 3-6). Because of the instability of the purified anti-Ngb antibody, we did not carry out a thorough immunohistochemical analysis of the fish brain. Nevertheless, in regions inspected by both methods, ISH and immunofluorescence, the observed patterns matched and demonstrated the presence of Ngb mRNA and protein in the somata of most if not all of the nerve cells. No evidence was found for glial expression of Ngb in fish, which corresponds to the data obtained in mammals (19 -21). Thus, the neuronal expression of Ngb was conserved in gnathostomian vertebrates for ϳ420 million years (37). A novel finding of this study was the observation of strong Ngb expression in the sensory part of the receptor cells of the zebrafish olfactory system (Fig. 5A). Significant Ngb was also present in the following parts of the olfactory tract in the telencephalon and the olfactory cortex. Because of the strong calcification of the rodent olfactory system, similar staining experiments have not been carried out in these species. Thus, it remains to be established whether high Ngb expression also occurs in the olfactory region of the mammals or whether this is a unique feature of the teleost fish.
Because the immunofluorescence signal in the brain was only clearly visible at high magnifications, its total concentration in the brain appeared to be low. Nevertheless, the fish retina and the gills showed bright fluorescence and thus may contain higher amounts of Ngb.
High Concentrations of Ngb in the Photoreceptor Inner Segments-The mouse retina has been observed so far to be the highest Ngb-expressing tissue in mouse with the Ngb protein reaching a concentration similar to that of myoglobin in the striated muscle (ϳ100 M) (14). The ISH pattern of Ngb mRNA in the fish retina was similar to that in mouse with the exception that we observed some ISH signals in the fish inner plexiform layer (Fig. 5B). Although there was, as in mouse, detectable amounts of Ngb protein present in the ganglion cell and the plexiform layers, most Ngb in the zebrafish retina actually appeared to be located in the inner segments of the photoreceptor cells (Fig. 6B). If the idea of an oxygen transport function of Ngb (6,14,16) was correct, it may be explained by the high energy requirements of the fish cones and rods. Whereas upon light adaptation, the pupil diameter of the mammalian retina decreased, most fish did not show a similar response (45). Rather, the cone cells of the fish retina displayed light-induced shortening of the myoid, the neck-region located between the ellispoid and the perinuclear region of the inner segments (46). Consequently, a high concentration of mitochondria in the ellipsoid region has been observed. Thus, the observation of the presence of apparently high Ngb concentrations in the inner segments agrees with previous observation in the mouse retina that suggests an association of Ngb with mitochondria (14).
Wittenberg and Wittenberg (47) observe an extracellular hemeprotein in the choroid blood from perfused retina of two basal teleost fish species (bowfin and bluefish). The hemochrome absorption spectra of these globins were similar to those of mammalian and fish Ngb. Thus, we had speculated that this heme protein may correspond to Ngb (14,24). However, neither the Western blotting experiments (Fig. 3) nor the immunofluorescence data (Fig. 6) provided any evidence that Ngb may be found in the blood or the extracellular space. Therefore, it remains uncertain whether the retinal heme protein identified in the bowfin and the bluefish (47) corresponds to Ngb and, even if it does, whether it represents as proposed by those authors an intracellular globin that has been artificially released into the blood during the preparation procedure.
Ngb in the Chloride Cells Suggests Another Association of Ngb with Mitochondria-With the exception of the expression in some endocrine tissues (19,20), Ngb has been assumed to be an exclusively neuronal protein. Here we show by immunofluorescence studies, ISH, and Western blotting that Ngb was also present in the chloride cells of zebrafish gills. Fish gills perform a variety of physiological functions including respiratory gas exchange, ion exchange, and excretion (48,49) and are considered as high energy-demanding tissues (50,51). In freshwater fish, the chloride cells are the principal site of Ca 2ϩ , Cl Ϫ , and possibly also Na ϩ uptake (52) and are known to be rich in mitochondria (52,53). Again, this observation suggests an association of Ngb with the mitochondria (14) and is in line with the hypothesis that Ngb is involved in cellular oxygen consumption.
Implications for Vertebrate Ngb Function-The phylogenetic origin of Ngb dates back to before the time when the Protostomia and Deuterostomia diverged, suggesting a conserved function of this protein in the metabolism of the animal (6,13,14,24). The Ngb gene structure, oxygen-binding kinetics, and expression patterns are globally similar in mammals and fish. Differences in the retinal expression pattern as well as the presence of Ngb in the fish gills may be easily explained by different oxygen requirements within these tissues. It should be stressed that, similar to mammals, fish Ngb concentration appears to be correlated with an abundant presence of mitochondria in the cells, although there is no evidence that Ngb is located within these organelles. Therefore, our results support the hypothesis of an important role of Ngb in oxygen-linked metabolism (14,16). Whether Ngb has a myoglobin-like function in supplying oxygen to the respiratory chain by facilitated diffusion from the cell membrane to the mitochondria (3,4) or whether it actually sustains the energy metabolism of the cell via another, a still unknown mechanism (13,16), remains to be established.