Supporting evidence for the contractile nature of fish branchial pillar cells was provided by demonstrating the presence of actin fibers and a novel four-and-a-half LIM (FHL) protein in which expression is specific for contractile tissues and sensitive to the tension applied to the pillar cell. When eel gill sections were stained with rhodamine-phalloidin, a selective fluorescent probe for fibrous actin, a strong bundle-like staining was observed around collagen columns in pillar cells, suggesting the presence of abundant actin fibers. A cDNA clone encoding a novel member of the actin-binding FHL family, FHL5, was isolated from a subtracted cDNA library of eel gill. Northern analysis revealed that FHL5 mRNA is highly expressed only in gills, heart, and skeletal muscle. In gills, FHL5 was found to be confined to pillar cells by immunohistochemistry. Confocal fluorescence microscopy showed that FHL5 is present in both cytosol and nucleus; within the cytosol, a large portion of FHL5 is colocalized with the phalloidin-positive actin bundles. Furthermore, transfection of myogenic C2C12 cells with FHL5 cDNA demonstrated, in addition to its interaction with actin stress fibers, a nuclear shuttling activity of FHL5. The mRNA and protein levels were found to be elevated on 1) transfer of eels from seawater to freshwater, 2) volume expansion by infusion of isotonic dextran-saline, and 3) constriction of gill vasculature by bolus injection of endothelin-1. These results suggest contractile nature of pillar cells and a role of FHL5 in maintaining the integrity and regulating the dynamics of pillar cells.
- C2C12 myoblast
- zinc finger
- four-and-a-half LIM
fish gills consist of a large number of filaments arranged along the gill arches. The surfaces of the filaments are greatly enlarged by a series of plate-like lamellae. Each lamella is composed of two sheets of epithelia separated by a thin space through which the blood circulates to allow the exchange of respiratory gases. The separation between the epithelial sheets is maintained by pillar cells and collagen bundles. Pillar cells are spool-shaped cells connecting two epithelial sheets of the respiratory lamella in fish gills (50, 51, 64). They are characterized by collagen bundles contained in the infoldings of the plasma membrane and are oriented perpendicularly to the epithelial sheets, which consist of a thin layer of pavement cells and a basal lamina that is continuous with the collagen bundles traversing the pillar cells (49, 64); for the anatomy of gill lamella, see the schematic illustration in Fig. 9D. The membrane-enclosed collagen bundles help prevent ballooning of the lamella. The inner surfaces of the two epithelial sheets are covered with flanges extended from the pillar cells, forming capillary lumen called lacunae. Another feature of the pillar cells is the presence of numerous myofilament-like structures that course through the cytoplasm in an orientation parallel to the collagen bundles. These myofilaments appear to provide additional structural support and exhibit contractile activity along the longitudinal axis of the pillar cells. The pillar cells are therefore thought to be specialized cells playing the following dual roles: 1) prevention of thrombogenesis and metabolic processing of circulating hormones (48, 49) and 2) maintenance of lacunar diameter and regulation of blood flow within the secondary lamella.
The contractile nature of pillar cells had been suggested as early as 1895 by Biétrix (4), as cited by Gillotreaux (18) and Hughes (21), and later was supported by electron microscopic observations of contractile filamentous materials in the pillar cells (3, 22, 47). The filamentous materials were further shown to be rich in myosin by immunofluorescence microscopy (55). However, it is still controversial whether the pillar cells play only a supportive role to maintain the structural integrity of the lamella or they are contractile and actively regulate the pattern of blood flow in the lamella. Contraction of pillar cells has been questioned because evidence has been lacking on the presence of sufficient amounts of actin in the pillar cells (51). Direct evidence for pillar cell contraction was finally obtained in 1998–1999 by Stenslokken et al. (56) and Sundin and Nilsson (57), who demonstrated 1) a significant increase in pillar cell diameter and 2) redistribution of blood flow within the lamella using a vasoconstrictor peptide, endothelin-1 (ET-1; for review, see Refs. 33, 39, and 53), in the rainbow trout Oncorhynchus mykiss (57) and the Atlantic cod Gadus morhua (56). To establish the contractile nature of pillar cells, the presence of actin in an amount comparable to that of myosin should be demonstrable. Here we show that pillar cells contain 1) abundant actin fibers that can be visualized with fluorescence-labeled phalloidin and 2) an actin-interacting protein, FHL5 (four-and-a-half LIM protein 5).
FHL5 is a novel member of the FHL family of proteins defined by the presence of four complete sets plus one half set of the LIM motif for protein-protein interaction. In mammals, this family consists of five members (ACT and FHL1–4) in which the characteristic property is a highly tissue-specific expression pattern. Although the expression of activator of cAMP-response element modulator (CREM) in testis (ACT) and FHL4 is restricted to testis (16, 17, 40), FHL1–3 are enriched in striated muscles (8, 17, 19, 41, 43) and involved in the regulation of cytoskeletal protein organization and transcription through their scaffolding and nuclear shuttling activities (6, 9, 44, 52, 59). We therefore further determined 1) whether FHL5 is colocalized with actin fibers in pillar cells, 2) whether it has a nuclear shuttling activity, and 3) whether its mRNA and protein levels undergo changes in response to tensions applied to pillar cells by volume expansion and treatment with ET-1, the receptor for which, and effects, have been documented in the gill lamellar vasculature (31, 56, 57). These results indicate that the pillar cells possess contractile machinery consisting of myosin, actin, and other regulatory components, including FHL5, and these cells play a dynamic role in the regulation of lacunar microcirculation. Such regulation could well be important in balancing respiratory need and passive osmotic movement of water through the surface of the lamella of the gills.
MATERIALS AND METHODS
Cultured Japanese eels, Anguilla japonica, were purchased from a local dealer. They were maintained without feeding in a 1-ton tank containing ∼750 liters of freshwater for 1 wk to acclimate to laboratory conditions. To prepare seawater-adapted eels, they were then transferred to a 0.5-ton tank containing ∼400 liters of seawater and acclimated there for at least 2 wk before use; freshwater controls remained in the freshwater tank during this period. Water in the tank was continuously filtered, circulated, aerated, and thermoregulated at 18 ± 0.5°C.
RNA isolation and construction of subtracted cDNA library.
Eels were separately adapted to seawater and freshwater for 2 wk each. Total RNA was isolated from the seawater- and freshwater-adapted eel gills by the guanidinium thiocyanate/CsCl centrifugation method (7), and poly(A)+ RNA was affinity purified using an oligo(dT)-cellulose mRNA purification kit (Amersham Biosciences). Two types of subtracted cDNA libraries (seawater mRNA-enriched and freshwater mRNA-enriched ones) from seawater and freshwater eel gills were constructed as previously described (36, 37). About 300 individual clones were sequenced from each library, and their expression levels were determined by Northern blot analysis using RNA preparations from freshwater and seawater eel gills to confirm differential expression.
Northern blot analysis.
Total RNAs were isolated from a number of tissues of freshwater and seawater eels and analyzed by Northern blotting essentially as described previously (36, 37) using a [32P]dCTP-labeled eel (e) FHL5 cDNA probe [a partial eFHL5 cDNA clone of 170 bp (nucleotides 311–481) that was isolated from the subtracted cDNA library mentioned above]. For time course analysis of the eFHL5 mRNA levels in the gill, total RNAs were isolated at various time points after transfer of eels from freshwater to seawater and subjected to Northern analysis. Hybridization with an eel β-actin probe (corresponding to nucleotides 206–343 in the rat sequence) was used as control. Quantification was performed with a Bioimage Analyzer (model BAS2000; Fuji Film) and an Imaging Plate (Fuji Film) with an exceptionally wide dynamic range. Error bars in the quantitative figures indicate the least squares SDs (n = 3–6).
cDNA cloning, sequencing, and 5′-RACE.
Construction of an eel gill cDNA library in λZAP II (Stratagene), screening of the library with the above-mentioned 32P-labeled probe for the eFHL5 cDNA clones, and the sequencing of hybridization-positive cDNA clones were all performed as described previously (36, 37). Five positive clones were obtained by screening 3 × 105 recombinant plaques, which covered the 3′-end but not the 5′-end of the eFHL5 mRNA. To obtain a full-length cDNA, the 5′-ends of eFHL cDNAs were amplified using the 5′/3′-RACE (rapid amplification of cDNA ends) kit (Roche Molecular Biochemicals) according to the manufacturer's instruction. The primers used were a specific primer (SP1: 5′-GTACTCCACATTCTTACTGCC-3′), a nested primer (SP2: 5′-CGCACATGATCTTGTTGTCC-3′), and a second nested primer (SP3: 5′-AATTCATCACGTATTTCTTCCC-3′). The nucleotide sequence of eFHL5 cDNA was established by sequencing three cDNA clones and more than five RACE products using a SequiTherm cycle sequencing kit (Epicentre Technologies).
Sequence analyses and comparison.
The deduced amino acid sequence of eFHL5 was compared with the sequences of FHL family members of other species. Protein sequences of mammalian FHL families were obtained from the GenBank sequence database. Protein sequences of FHL family members of the frog (Xenopus laevis and X. tropicalis), chicken (Gallus gallus), fugu (Takifugu rubripes), zebrafish (Danio rerio), medaka (Oryzias latipes), carp (Cyprinus carpio), and rainbow trout (Oncorhynchus mykiss) were deduced from nucleotide sequences of genomic DNA or EST clones obtained by BLAST search of the databases of GenBank (http://www.ncbi.nlm.nih.gov/BLAST), Medical Research Council (MRC, http://fugu.hgmp.mrc.ac.uk), and The DOE Joint Genome Institute (JGI, http://genome.jgi-psf.org/fugu). Identity (percentage of identical amino acids) and similarity (percentage of similar amino acids of the following groupings: STA, NEQK, NHQK, NDEQ, QHRK, MILV, MILF, HY, and FYW) were calculated with Genetyx-Mac software (Genetyx, Tokyo, Japan).
For evolutionary analyses, the protein sequences were aligned using Clustal W software (60), and then a phylogenetic tree was constructed by the neighbor-joining method (46) using MEGA software (26) based on Poisson-corrected evolutionary distances (46). Statistical analysis was performed by bootstrap methods (46). Dates of the duplication events of FHL genes were calculated with MEGA software based on the molecular clock hypothesis (46) that assumes equal rates of amino acid substitution during the course of evolution. The divergence time between amphibians and mammals, i.e., 360 million years ago (25), was used as a reference point for dating.
A total of 16 eels were divided into the following two groups: one (183.9 ± 15.4 g, n = 8) for acute infusion of plasma expander into the ventral aorta, and the other (203.0 ± 13.1 g, n = 8) for bolus injection of human ET-1. Eels were anesthetized by immersion for 15 min in 0.1% (wt/vol) tricaine methanesulfonate (Sigma) neutralized with sodium bicarbonate. In all eels, a polyethylene cannula (0.8 mm OD) was inserted in the ventral aorta for drug administration and blood pressure measurement. After surgery, eels were placed in a plastic trough in which aerated freshwater continuously circulated at 18°C. In the case of acute infusion, the cannula in the ventral aorta was connected via a three-way stopcock to an infusion pump and a pressure transducer (DX-300; Nihon Kohden, Tokyo, Japan). The transducer was connected to a polygraph (366 System; NEC, San-ei, Tokyo) and a pen recorder for continuous measurement of arterial pressure. Eels were allowed to recover from anesthesia for at least 18 h postoperatively. The troughs were covered with a black vinyl sheet to minimize visual stress during the experiment.
In one group, isotonic 0.9% NaCl containing 2% dextran (mol wt = ∼7,000; Sigma) was infused in the ventral aortic cannula at a rate of 5 ml/h by an infusion pump. Dextran was supplemented to adjust the colloid osmotic pressure. The infusion rate was determined by taking into account the quantity of blood in eels. A control infusion was made by infusing the same solution at a rate of 0.5 ml/h. The acute infusion caused immediate increases in arterial blood pressure, but the control infusion did not. After 1 h of infusion, eels were anesthetized and gills were removed for total RNA isolation. In the other group, human ET-1 (Peptide Institute, Osaka, Japan) dissolved in 0.9% NaCl solution containing 0.01% Triton X-100 was injected as a bolus at 50 pmol·40 μl−1·kg body wt−1 through the ventral aortic cannula. The cannula was immediately flushed with 60 μl of vehicle (0.9% NaCl). The ET-1 solution was freshly prepared from a stock solution (10−4 M), and the dose was chosen from the dose-response curves reported in the literature (28, 56). Triton X-100 was added to prevent the adsorption of ET-1 to the tube wall. Injection of vehicle alone (100 μl) served as a control. After 25 min, eels were anesthetized as described above, and gills were removed and kept at −80°C until mRNA isolation.
Antibody production, Western blotting, and immunohistochemistry.
An 840-bp fragment encoding eFHL5 (amino acid residues 2–280) was subcloned in a bacterial expression vector pRSET B (Invitrogen), and the construct was transferred into E. coli XL1-Blue. Production and purification of recombinant protein (His6-eFHL5), immunization of rabbits, and affinity purification of polyclonal antibodies on a His6-eFHL5-coupled HiTrap column were performed as previously described (36, 37). The antiserum raised was named “anti-eFHL5.” Western blot analysis and immunohistochemistry were also carried out according to the published methods (38). Samples for Western blot analysis were prepared as follows: eel tissues were homogenized in 9 vol of 0.25 M sucrose containing 1 mM phenylmethylsulfony fluoride, 10 μg/ml leupeptin, and 10 μg/ml pepstatin A at 4°C. Total protein concentrations were determined using a BCA protein assay kit (Pierce Biotechnology), and 15 μg of protein were separated by SDS-PAGE on 12% polyacrylamide gel and subjected to Western blotting.
Confocal immunofluorescence microscopy.
Cryosections of freshwater eel gills were prepared as described previously (36, 37). Nonspecific reactions were blocked by preincubation with PBS containing 5% normal goat serum in a humidified chamber at 25°C for 1 h. After being washed three times in PBS, sections were incubated with either preimmune serum or primary antibody (anti-eFHL5, 1:1,000 dilution in blocking solution) for 16 h at 4°C. Sections were then rinsed thoroughly in PBS and incubated, for 1 h at 25°C, with a secondary antibody solution containing Alexa 488-conjugated anti-rabbit IgG (0.75 μg/ml; Molecular Probes) for indirect detection of eFHL5 and TOPRO-3 (2 μM; Molecular Probes) for DNA staining. Before being visualized, sections were rinsed in PBS and mounted in fluorescence-mounting medium (Fluoromount-G; Southern Biotechnology Associates). Fluorescence was detected using a Zeiss Axioskop fluorescence microscope equipped with a confocal laser-scanner unit CSU10 (Yokogawa Electronic, Tokyo, Japan). Images were obtained with a high-resolution digital charge-coupled device (CCD) camera (C4742–95; Hamamatsu Photonics) and processed by IPLab software (Scanalytics). The brightness and contrast of the final images were adjusted with Adobe Photoshop software (Adobe Systems).
Cell culture and transfection.
A 1.3-kb BamH I-EcoR I fragment containing the complete coding region of the eFHL5 cDNA was subcloned into pcDNA3 and named pcDNA3-eFHL5. The mouse skeletal muscle cell line C2C12 was grown and maintained as myoblasts in proliferation medium consisting of DMEM (Sigma), 15% FBS, and the antibiotics penicillin (100 U/ml) and streptomycin (100 μg/ml). Cells (0.7–0.9 × 106) were plated on 35-mm plates and transfected with pcDNA3-eFHL5 or pcDNA3 using Lipofectamine Plus Reagent (Life Technologies) and cultured in proliferation medium for 2 days. Cells were then induced to differentiate by switching to a differentiation medium composed of DMEM and 2% horse serum. Differentiation medium changes were performed every 48 h for 12 days.
C2C12 cells in proliferation medium for 2 days and in the differentiation medium for 12 days were rinsed two times with PBS, fixed with 2% formaldehyde in PBS for 15 min at 4°C, rinsed with PBS three times, permeabilized in PBS containing 0.2% Triton X-100 for 15 min, and blocked with 5% FBS in PBS for 30 min at 22°C. Cells and cryosections of freshwater eel gills were first incubated with anti-eFHL5 for 2 h at 25°C, rinsed three times with PBS, and then incubated with anti-rabbit IgG Alexa Fluor 488 for 1 h at room temperature. Nuclei were labeled with Hoechst 33342 (100 ng/ml; Molecular Probes). F-actin was labeled with TRITC-phalloidin (0.4 μM). Cells were examined with an Olympus fluorescence microscope model IX70 and a filter optimized for triple-label experiments. Pictures were taken using a Princeton Instruments-cooled CCD camera (MicroMAX 5 MHz; Roper Scientific) and analyzed using MetaMorph software (Universal Imaging).
Identification and molecular characterization of eel FHL5.
As an extension of our previous attempts to identify genes that are differentially expressed in freshwater and seawater eels (36, 37, 58), we discovered a cDNA clone covering a partial sequence common to members of the FHL family in a freshwater eel gill cDNA library. Preliminary Northern blot analysis indicated that 1) the mRNA level of the FHL clone can be easily detected, even with total RNA preparations indicating that the protein product is also abundantly expressed; and 2) the mRNA levels are markedly elevated in freshwater eel gills (Fig. 1). Furthermore, partial sequence analysis indicated that the eFHL cDNA appears to represent a novel member of the family. We therefore tentatively named it “FHL5” and decided to characterize it.
On Northern blot analysis, a 32P-labeled FHL5 probe hybridized to a 1.7-kb transcript (Fig. 1). To isolate FHL5 cDNA clones of this size, we screened the freshwater eel gill cDNA library previously constructed (36, 37) and obtained five positive clones. The longest clone was of 1,685 bp and contained the apparent 3′-end of the sequence, including a polyadenylation signal (AATAAA) and a poly(A) tail. It was however, not clear whether the clone contained the 5′-noncoding region. We therefore performed 5′-RACE using a cap site-labeled cDNA library to determine the 5′-end of FHL5 mRNA, which yielded an additional sequence of 13 nucleotides, extending the total length of the sequence from 1685 to 1698 nucleotides (Fig. 2). The open reading frame (133 to 972 nucleotides) encodes a protein of 280 amino acid residues with a calculated molecular mass of 32 kDa that contains a single NH2-terminal zinc finger representing a half LIM domain followed by four complete LIM domains (Fig. 3). In Fig. 3, the amino acid sequences of relatively highly conserved NH2- and COOH-termini are shown that can be used for classification of FHL family members. The sequence of the middle region is variable except for the zinc-coordinating Cys and His residues. An interesting feature of the primary structure is the complete conservation of 1) the spacing between the zinc-coordinating Cys and His residues (CX2CX17C/HX2C) among the nine zinc fingers, 2) the two-residue spacing between the double zinc fingers constituting the LIM motif, and 3) the eight-residue spacing among the LIM domains. To illustrate these relations schematically, the zinc-coordinating Cys and His residues are circled in Fig. 2, and their conserved spacing pattern is shown in Fig. 3 (see the numbers on the second zinc finger).
Phylogenetic relationship to other FHL family members.
A BLAST search of the GenBank protein and genomic DNA databases revealed that the predicted protein sequence (Fig. 2) is most similar to mammalian FHL1 (Figs. 3 and 4) but represents a novel member of the FHL family that does not have a counterpart in mammals (Fig. 4). We therefore named the protein eFHL5. Its sequence identities and similarities to known mammalian FHL protein sequences, including human FHL1 (63% identity, 78% similarity), mouse FHL4 (52%, 71%), human FHL2 (48%, 70%), human FHL3 (43%, 63%), and human ACT (42%, 65%), are relatively low. Although no other members of the FHL family have been cloned from fish species, we were able to identify, through the BLAST search, the following fish FHL sequences (with the indicated identity and similarity to eFHL5): fugu (f) FHL2 (EST, AL835189) (44%, 89%), fFHL3 (scaffold 645) (45%, 83%), fFHL5 (scaffold 802) (75%, 96%), fFHL6 (scaffold 1374) (63%, 93%) and fFHL (scaffold 3225) (46%, 85%), fFHL (scaffold 264) (43%, 83%), and fFHL (scaffold 252) (42%, 82%) in the pufferfish Takifugu rubripes whose genome project has been completed (1) and zebrafish (z) FHL2 (EST, CB355867) (44%, 89%), zFHL3 (EST, BF718175) (45%, 84%), zFHL5 (EST, AF134774) (79%, 97%), zFHL6 (EST, CA474710) (66%, 96%), zFHL (EST, CF997242) (50%, 85%), and zFHL (EST, AW343089) (45%, 84%) in the zebrafish Danio rerio whose genome sequencing is in the final stages (Fig. 4).
A phylogenetic tree of FHL family members in mammals and teleosts is shown in Fig. 4A. The phylogenetic tree and statistical analysis performed by the bootstrap method indicated that the FHL family consists of the following three subfamilies: FHL2/ACT, FHL3, and FHL1/4/5/6. The FHL1/4/5/6 subfamily can further be divided into two groups with bootstrap values of 91 (FHL1/4) and 84 (FHL5/6). Although these two groups appear to be mutually exclusive in mammals (FHL1/4) and teleosts (FHL5/6), they are thought to be paralogs (Fig. 4B) and not orthologs (Fig. 4C), since their separation occurred much earlier than the speciation of teleosts (731 vs. 450 million years ago, Fig. 4B), as estimated by a molecular clock calculation with MEGA software. The lack of the FHL1 and -4 and FHL5 and -6 genes in teleosts and mammals, respectively, and the absence of the FHL4 gene in humans are explained by the birth-and-death evolution theory proposed by Nei et al. (45), in which new genes are generated by gene duplication and some of them stay in the genome whereas the others are either deleted from the genome or inactivated into pseudogenes.
Gill- and striated muscle-specific expression of eFHL5.
To determine the tissue distribution of eFHL5 mRNA and to compare its expression levels in seawater and freshwater eels, we carried out Northern blot analysis using total RNA preparations from various tissues of seawater and freshwater eels, including the gill, kidney, head kidney, heart, liver, posterior intestine, anterior intestine, stomach, and skeletal muscle. A strong signal of ∼1.7 kb was detected in the gill, heart, and skeletal muscle but not in the other tissues examined (Fig. 5). The expression in the heart and skeletal muscle is not surprising since it has been demonstrated, through the characterization of mammalian FHL proteins, that one of their distinctive hallmarks is their muscle-specific distribution. The high level expression in the gill, however, is interesting, since it suggests that the gill is mechanically much more dynamic than has been anticipated.
Salinity- and tension-dependent alterations in expression levels of eFHL5 mRNA.
When Northern blotting patterns of eFHL5 mRNA in freshwater (Fig. 5A) and seawater (Fig. 5B) eel tissues were compared, a significant difference was observed in the gills (Fig. 5C). Namely, the branchial expression level was more than two times higher in freshwater than in seawater eels. In contrast, little or no such difference was seen in the heart and muscle. Figure 6 shows the time course of alterations in eFHL5 mRNA levels in the gill after transfer of eels from freshwater to seawater. The adaptive changes occurred relatively slowly but steadily.
We next determined the effects of acute infusion of dextran-saline solution, a volume expander, and intra-arterial injection of ET-1. Acute infusion of dextran-saline in freshwater eels ata rate of 5 ml/h elicited a significant increase in ventral aortic blood pressure, whereas control eels infused at 0.5 ml/h remained unchanged (Fig. 7A). In response to this change, eFHL5 mRNA levels in the gill increased by ∼120% (Fig. 7C). Similarly, intra-arterial injection of ET-1 (50 pmol/kg body wt) provoked an immediate and sustained increase in blood pressure (Fig. 7B). This bolus injection of ET-1 also induced elevated expression (∼90% increment) of eFHL5 mRNA in the gill (Fig. 7D).
Pillar cell localization of eFHL5 revealed by immunohistochemistry.
Rabbit antiserum was produced against a recombinant His-tagged eFHL5. The specificity of the antiserum was ascertained by Western blot analysis using crude extracts of eel gill, heart, and skeletal muscle, in which the immunoreactive protein was detected as a single band of 33 kDa (Fig. 8). The size of the band corresponds very well with the calculated molecular mass of eFHL5 (32.1 kDa). The staining of the 33-kDa band was abolished by preabsorption of the antiserum with the antigen (data not shown). The salinity-dependent expression of eFHL5 seen at the mRNA level (Figs. 1 and 6) was also confirmed at the protein level (Fig. 8).
The cellular location of eFHL5 was determined by immunohistochemistry using freshwater eel gill sections. The anti-eFHL5 antiserum, characterized above, specifically stained pillar cells (Fig. 9, Aa and Ac). The signal could be obtained only with the antiserum or affinity-purified antibody but not with preimmune serum (Fig. 9Ab) or antigen-absorbed antiserum (data not shown). A similar staining pattern was observed with a monoclonal antibody against smooth muscle myosin (Fig. 9Ad), as previously reported by Smith and Chamley-Campbell (55). However, there was a marked contrast in the reaction with vascular smooth muscle cells. In the case of the anti-myosin monoclonal antibody, vascular smooth muscle cells were also strongly stained, whereas no such staining was obtained with anti-eFHL5 (compare Fig. 9, Aa and Ad). These results indicate that, in the gill, the newly identified protein FHL5 is expressed exclusively in the pillar cells.
To determine the subcellular localization of eFHL5 in pillar cells, we performed indirect immunofluorescence microscopy. Confocal imaging of stained freshwater eel gill sections indicated that eFHL5 is present in both the nuclei and cytoplasm of pillar cells (Fig. 10). This is consistent with the recent observations in cultured mammalian cells that FHL proteins are able to translocate between the nucleus and cytoplasm (6, 9, 44, 52, 59).
Visualization of actin fibers in pillar cells with phalloidin.
As mentioned in the Introduction, immunological studies have failed to detect significant levels of immunoreactive actin in the pillar cell. We repeated immunostaining with a variety of anti-actin antibodies, commercially available from Sigma, including monoclonal anti-β-actin (AC15), anti-α-sarcomeric actin (5C5), anti-pan-actin (AC40), anti-α-smooth muscle actin (1A4) antibodies, and polyclonal anti-pan-actin antibodies (residues 20–33, A5060; A2668), but could not obtain definitively positive staining. However, staining of the eel gill sections with phalloidin, a bicyclic heptapeptide toxin that selectively interacts with actin filaments (10), yielded a clear staining pattern, clearly revealing the presence of dense filaments of actin in the pillar cell (Fig. 9, B and C). In cytosol, eFHL5 associates with the actin filaments, which are located close to the collagen column wrapped by the plasma membrane (Fig. 9C). The reason for the poor immunoreactivity of pillar cell actin is unclear, since actins are one of the most highly conserved proteins among species, and the antibodies used here have been demonstrated to cross-react with actins from a variety of species; in fact, they were able to stain actins in eel tissues other than the pillar cell (data not shown).
Interaction of eFHL5 with actin fibers in C2C12 muscle cell line.
Because an in vitro culture of pillar cells has not been established yet, we used, instead, C2C12 myoblasts as a model system for characterizing the nuclear translocation and actin-binding activities of eFHL5. The myogenic cell line C2C12 can be differentiated into contractile multinuclear myotubes expressing characteristic muscle proteins by a lowering of the serum concentration of the culture medium (5, 67). We transfected the cell with a mammalian expression vector containing eFHL5 cDNA and determined the subcellular localization of transiently expressed eFHL5 before and after differentiation. In undifferentiated C2C12 cells, a large portion of eFHL5 was detected in the nucleus, a finding that was confirmed by counterstaining with the DNA probe Hoechst 33342 (Fig. 11, Aa–Ac); the remaining portion of eFHL5, present in the cytoplasm, is associated with actin fibers (Fig. 11, Ad–Af). These signals were not detected within mock-transfected C2C12 cells, and almost all of the transfected cells displayed similar expression patterns (data not shown). On differentiation to muscle cells induced by serum deprivation, eFHL5 translocated from the nucleus to cytoplasm (Fig. 11, Bg–Bi). In this multinucleated state, most of the overexpressed eFHL5 molecules were not associated with aggregated actin bundles (Fig. 11, Bg–Bi). These results suggest that 1) eFHL5 is translocated between the nucleus and cytoplasm to meet the physiological demands of cells and 2) it has the ability to interact with actin and may act as a regulator of actin cytoskeletal dynamics, as is the case with FHL3, which has been shown to inhibit α-actinin-mediated actin bundling (9).
In support of the original idea of Biétrix (4) in 1895 and the recent demonstration by Sundin and Nilsson (57) that the pillar cell in the gill is a contractile muscle-like cell, we demonstrated the presence, in the eel pillar cell, of actin fiber and its binding protein FHL5. In the actual course of this research, we first identified FHL5 in a freshwater mRNA-enriched eel gill cDNA library. Its abundant expression in the pillar cells, revealed by Northern blotting and immunohistochemistry, prompted us to examine the presence of actin in pillar cells, since many of the FHL family members had been demonstrated to occur mainly in muscles and to interact with actin, as described in detail below. Thus we succeeded in demonstrating the presence of large amounts of actin and its binding protein FHL5, providing strong evidence for the contractile nature of the pillar cells. Fluorescence microscopy further revealed that actin is organized into fibers that align in parallel to collagen columns and perpendicularly to the two sheets of the basal lamina (Fig. 9, Bf, C, and D). This orientation, as well as the well-organized arrangement of actin fibers, is also consistent with their role in pillar cell contraction.
FHL5 identified here is a member of the recently emerging FHL family that belongs to the LIM protein superfamily and exhibits a muscle-specific pattern of expression. A database search revealed the presence of its orthologs in fugu and zebrafish but not in mammals, suggesting a unique role in fish. Phylogenetic analysis of the known members of the FHL family, including ACT and FHL1–5, indicated that fish FHL5 is most closely related to mammalian FHL1 but that they diverged before the speciation of fish and mammals. This fish-specific nature of FHL5 appears to be consistent with its expression in the pillar cell, for which there is no counterpart in mammals.
Although the mechanisms that regulate FHL5 expression in the pillar cell remain to be clarified, the enhanced expression in freshwater and under mechanical stress (Fig. 7, C and D) suggests that FHL5 levels are likely to be regulated by the mitogen-activated protein (MAP) kinase cascade, which has been demonstrated to be involved in a variety of stress responses (11, 24), including mechanical strain-induced muscle cell differentiation and the induction of differentiation marker proteins (12, 61, 66). Furthermore, activities of MAP kinases have also been shown to be elevated in the gill of the euryhaline killifish Fundulus heteroclitus during hyposmotic stress (23). Changes in the expression levels have also been reported for mammalian FHL1, which is most highly expressed in skeletal muscle, with intermediate or lower expression in a wide range of tissues, including heart, lung, kidney, and brain (8, 17). Its expression is upregulated in the cultured mouse skeletal muscle cell line C2C12 during differentiation in vitro (41), in rat hypertrophied skeletal muscle (32), in human heart in hypertrophic cardiomyopathy caused by transverse aortic constriction (30), and in mouse heart with dilated cardiomyopathy resulting from targeted deletion of muscle LIM protein (8). In contrast, FHL1 mRNA levels were found to be lowered in human ischemic dilated cardiomyopathy (68).
Another interesting feature of the FHL family is that FHL proteins localize within both the nucleus and the cytoplasm. In the cytoplasm, FHL proteins have been shown to be localized to cytoskeletons such as actin stress fibers (6, 9, 52, 65), focal adhesions (6, 29, 65), and the Z-discs (9, 27, 29) of cultured cells and muscle tissues. Recent reports indicate that FHL3 inhibits α-actinin-mediated bundling of actin fibers in vitro (9), and FHL2 mediates the targeting of the metabolic enzymes creatine kinase, adenylate kinase, and phosphofructokinase to the elastic filament protein titin in cardiomyocytes (27). In the nucleus, several FHL family members have been demonstrated to bind directly to DNA-binding transcriptional factors and to function as transcriptional coregulators. For example, ACT acts as a coactivator of CREM- and CREB-mediated transcription in a CREB-binding protein- and phosphorylation-independent manner (16). FHL2 functions as a coactivator of CREB (17), androgen receptor (43), Wilm’s tumor suppressor-1 (14), and activator protein-1 (42). It can also act as a corepressor for promyelocytic leukemia zinc finger protein (35). FHL3 acts as a coactivator of CREB (17) and acts as corepressor for basic Krüppel-like factor/Krüppel-like factor 3 (62). FHL1 and FHL4 lack the ability to modulate the transcription mediated by CREB and CREM (17), and their binding partner within the nucleus is still unknown. Although FHL proteins do not contain a recognizable nuclear localization signal, they may be transferred to the nucleus by binding proteins that contain a nuclear import signal. The nuclear localization of FHL proteins is regulated, at least in part, by signals mediated by integrins (52) and Rho GTPases (44).
Immunofluorescence microscopy reveals that eFHL5, like the other FHL family members mentioned above (6, 9, 44, 52, 59), localizes both in the cytoplasm and in the nucleus (Figs. 10 and 11). However, it does not contain any consensus sequences for nuclear export or import, suggesting the carrier protein-mediated translocation mentioned above (2, 9, 34, 52). In the nucleus, eFHL5 is expected to act as a transcriptional coregulator, as has been demonstrated for other FHL proteins (13, 14, 16, 17, 35, 43, 44, 62). The structural feature of the FHL proteins, namely the presence of multiple LIM domains for protein-protein interaction, makes them ideal scaffold proteins that can stabilize multiprotein transcriptional complexes in the nucleus and actin fibers and accessory proteins in the cytoplasm. Further insight into the mode of action of eFHL5 and regulation of the contractile machinery of the pillar cell will be obtained along with the identification of the partner proteins of eFHL5.
In fish, endothelin seems to be the most potent and gill-specific vasoconstrictor (20, 49, 56), and the pillar cell is one of endothelin's major targets, as demonstrated by the contractile response to ET-1 (56, 57). Although the precise location of receptors is still unknown, radioligand binding assay demonstrated the presence of relatively abundant 125I-ET-binding sites in the gill (15, 31), and the pattern of distribution of 125I-ET-1-binding sites, determined by autoradiography (31), is reminiscent of pillar cell localization of the receptor sites. Our demonstration of ET-1-induced changes in the eFHL5 mRNA levels is consistent with the presence of hormone-sensitive contractile machinery (actin-myosin filaments with accessory proteins) that can contract pillar cells and redistribute lamellar blood flow so as to meet the moment-to-moment oxygen needs. We also demonstrated volume expansion-induced changes in the FHL5 mRNA levels, suggesting an involvement of a stretch-activated signaling mechanism. These subtle changes of the accessory component, FHL5, in response to tensions applied to or generated by pillar cells, appear to be consistent with what one would expect, since specific changes in physical forces may require constant remodeling of actin bundling. Although endothelin is a key regulator of pillar cell contraction, its source remains to be determined, namely it is at this time not known whether endothelin is synthesized in the pillar cells and acts in an autocrine/paracrine fashion as in the mammalian system. Determination of the mRNA levels under the experimental conditions employed here would provide useful information on the mechanism of hormonal regulation of blood flow through gill lamella. Endothelin-induced intracellular signaling leading to pillar cell contraction and to altered expression of the components of the contractile machinery, including FHL5, should also be clarified to enable a better understanding of the contractile nature of the pillar cells.
We have demonstrated the presence of abundant actin fibers in pillar cells by phalloidin staining, but were unable to immunostain them, as has previously been the experience of other researchers. This poor immunoreactivity of the pillar cell actin fibers suggests that they are tightly surrounded by other proteins that block the access of antibodies, but not that of small molecules such as phalloidin, to the actin fibers in the core region of the structure. The presence of pillar cell-specific actin with a highly diversified amino acid sequence is unlikely, since database searches revealed only highly conserved actin sequences in the pufferfish and zebrafish genomes. For example, Venkatesh et al. (63) have identified nine pufferfish actin genes that are highly conserved, and three of which have been shown to be expressed relatively highly in the gill. Our preliminary results that pillar cells contain high levels of immunoreactive actinin, a mediator of actin bundling, also support the presence of large amounts of actin fibers in the pillar cells. From the location and orientation of the actin fiber-containing structures visualized with phalloidin (Fig. 9, B and C), they are expected to be anchored to the plasma membranes surrounding the collagen columns. The FHL5 identified here may regulate the contractility and mechanical stability of the actin fiber-containing structures by acting as a scaffold protein that can interact with a number of proteins, including actin. FHL5 may also control the biosynthesis of components of the contractile machinery of the pillar cells by acting as a nuclear shuttling protein with a transcriptional activity. Finally, FHL5 may further function as a transmitter of proliferation and differentiation signals generated by the interaction of integrins with extracellular matrix proteins (e.g., collagen columns), as has recently been demonstrated for FHL2 and FHL3 in striated muscles (54).
This work was supported by Grants-in-Aid for Scientific Research (14104002) from the Ministry of Education, Culture, Sport, Science and Technology of Japan (MEXT) and the 21st Century Center of Excellence Program of MEXT. A. C. Mistry was supported by Postdoctoral Fellowship for Foreign Researchers from the Japan Society for the Promotion of Science.
We thank Dr. Malcolm Forster (University of Canterbury) and Dr. Gert Flik (University of Nijmegen) for drawing our attention to molecular characterization of pillar cells and Setsuko Sato for secretarial assistance. We thank Pacific Edit for review of the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 the American Physiological Society