Identification of phosphorus (P)-responsive genes is important in diagnosing the adequacy of dietary P intake well before clinical symptoms arise. The mRNA abundance of selected genes was determined in the intestine, pyloric ceca, and kidney of rainbow trout fed low-P (LP) or sufficient-P (SP) diet for 2, 5, and 20 days. The LP-to-SP ratio (LP/SP) of mRNA abundance was used to evaluate the difference in gene expression between LP and SP fish, and to compare the response with bone and serum P, which are conventional indicators of P status. The LP/SP of intestinal, cecal, and renal type II sodium-phosphate cotransporter (NaPi-II) mRNA abundance changed from ∼1–2 (day 2) to ∼1.4–4 (day 5) and to ∼2–10 (day 20). The LP/SP of renal NaPi-II, vitamin D 24-hydroxylase, and vitamin D receptor mRNA abundance correlated inversely with serum P on day 5 but not on day 2 and day 20. In another study, differentially expressed genes between LP and SP fish were examined by subtractive hybridization, confirmed by Northern blot, and evaluated by t-test and correlation with serum and bone P concentrations. About 30 genes were identified as dietary P responsive at day 20, including intestinal meprin and cysteinesulfinic acid decarboxylase, renal S100 calcium-binding protein and mitochondrial Pi carrier, and cecal apolipoprotein E, somatomedin B-related protein, and NaPi-II. The LP/SP of mRNA abundance of renal mitochondrial Pi carrier and intestinal cysteinesulfinic acid decarboxylase changed significantly by day 2, and intestinal meprin by day 5. Hence, these genes and NaPi-II are among the earliest steady-response genes capable of predicting P deficiency well before the onset of clinical deficiency.
- dietary regulation
- gene expression
- sodium phosphate
- vitamin D
in mammals, phosphorus (P) plays numerous roles in intermediary metabolism, including protein phosphorylation, generation of high-energy carriers, and blood buffering. P is a key component of numerous intermediary metabolites, genetic materials, phospholipids, and skeletal tissues. Despite these vital roles of P in life, little is known about the effect of P deficiency on gene expression. This is likely due to the fact that P is abundant in natural foods, and dietary P deficiency is virtually impossible in normal humans (19). In fish, however, the incidence of clinical P deficiency is increasing because of environmental statutes mandating aquaculture facilities to reduce P excretion in the effluent water (38), which in turn requires minimizing P in the feeds. Thus management of aquaculture facilities requires frequent monitoring and precise knowledge of the P status of fish. Traditionally, P status of fish has been estimated based on whole body P, bone P, bone ash, and blood P levels (22). However, the sensitivity and accuracy of these indicators are insufficient, especially at an early stage of P deficiency or excess. Thus alternative indicators that can determine or even predict P deficiency or excess with greater sensitivity and precision are needed.
In fish, there have been few studies on mechanisms regulating P homeostasis. Fish small intestine generally lacks villi, and fish kidney has no loop of Henle, so P processing in these organs may be different from that in mammals. However, fish can absorb various minerals from the water via the gills and skin. In fact, fish can meet most or all of their Ca requirement by directly absorbing Ca ions from the water [∼10–150 (freshwater) to 400 (seawater) mg/l Ca], so that dietary Ca deficiency or Ca-P imbalance rarely becomes a problem (34). However, P must be supplied in the food, because P concentrations in natural waters are very low (<0.1 mg/l for fresh and seawater), and P uptake mechanisms from the gill may be inefficient (45). Hence, intestinal absorption of Pi is important, and a Pi transporter that mediates transport of Pi from intestinal and renal tubular lumen type-II sodium-phosphate cotransporter (NaPi-II) has been cloned in some fish species (50). Like that in mammals, intestinal and renal tubular Pi transport in fish is regulated by dietary Pi levels and probably by parathyroid (PTH) and other hormones. NaPi-II mRNA in the trout intestine and kidney increased during chronic dietary P restriction (8); however, Pi transport in the gut seems to occur predominantly in the pyloric ceca (PC) via paracellular diffusion (44). Although fish do not have a PTH gland, PTH/PTH-related protein receptors as well as PTH-related protein and PTH-like proteins that bind to these receptors have been identified in fish species (13). Phosphatonins are a group of phosphate-regulating factors associated with mammalian phosphaturia (37), and a phosphatonin-hydrolyzing endopeptidase (PHEX) gene has also been found in fish (3). However, the roles of PTH-like and PHEX-like proteins in P homeostasis of fish remain to be studied. Calcitonin is another hormone that affects P metabolism in mammals but whose role in P metabolism in fish is not clear. It does inhibit Ca uptake by the gill (49). Stanniocalcin also inhibits gill Ca transport but appears to stimulate renal Pi reabsorption in fish (25). The large variety of factors regulating P metabolism in fish was demonstrated in flounder renal tubule cell cultures where net Pi reabsorption was increased by salmon stanniocalcin, rat prolactin, salmon/flounder somatolactin, bovine PTH, and salmon growth hormone (24, 25). These findings show the similarities and differences between fish and mammalian mechanisms of P homeostasis as well as the paucity of research in fish P metabolism.
Because of the metabolic importance of Pi, dietary P restriction should immediately alter the expression of genes involved in P homeostasis. Surprisingly, there is scant information about dietary P responsive genes in animals. Norbis et al. (33) screened a whole full-length cDNA library of rabbit small intestine and identified a cRNA, tentatively named Pi-uptake stimulator (PiUS), that stimulated Pi uptake when injected into Xenopus oocytes. Custer et al. (10) used differential display and found a new gene, named diphor-1, that was upregulated by approximately twofold in dietary P restriction. Cheung et al. (6) used proteomics and found six genes in the proximal tubules of rat kidney that were putatively responsive to dietary P. None of these six genes, however, were genes previously demonstrated to be P responsive in mRNA abundance, suggesting that the expression levels of mRNA and its protein could differ from one another.
In contrast to the scarcity of information on P-responsive genes in animals, P-responsive genes in plants have been relatively well studied. For example, in the shoot of the flowering mustard plant Arabidopsis, expression of 61 genes (out of 8,100 genes in a microarray chip) changed >2.5-fold ∼100 h after P deprivation, whereas 64 genes responded in 4 h. Interestingly, only nine genes were common between the late- and early-response genes (15). Other estimates, using a microarray of 6,172 genes, reported that almost 2,000 Arabidopsis genes responded more than twofold after 72 h of P deprivation (51).
In mammals, several genes, in addition to those mentioned above, are responsive to dietary P restriction. Two key cytochrome hydroxylase enzymes (P450) in vitamin D metabolism, 25-hydroxyvitamin D-1α-hydroxylase (CYP27B1) and 25-hydroxyvitamin D-24-hydroxylase (CYP24), are regulated by dietary P to maintain P homeostasis. Low-P (LP) diets downregulate CYP24 mRNA expression (52, 55), thereby preventing the catabolism of the active form of vitamin D3, 1,25-dihydroxyvitamin D3 (calcitriol), at times of P deficiency. The vitamin D receptor (VDR) binds calcitriol and forms a heterodimer complex with retinoid X receptor (RXR) to activate the vitamin D response element of target genes, including NaPi-II. Thus an LP diet upregulates VDR gene expression (28, 43).
Based on these findings in mammals, we hypothesized that similar responses might occur in fish, namely, that dietary P restriction increases NaPi-II, PiUS, and VDR mRNA abundance and decreases CYP24 mRNA abundance. In the first study, therefore, we examined the mRNA abundance of these genes in known P-responsive tissues [intestine, PC, and kidney (44, 45)] at days 2, 5, and 20 of dietary P restriction. Because, in the first study, only NaPi-II responded markedly to dietary P restriction, we continued our search for other P-responsive genes. In the second study, we screened whole cDNA libraries of the intestine, PC, and kidney obtained from fish fed P-deficient or P-sufficient diet for 20 days. The differentially expressed genes were identified by subtractive hybridization, and ∼30 clones were confirmed by Northern blot to be P responsive. Here, we report a set of dietary P-responsive genes in rainbow trout.
MATERIALS AND METHODS
Fish rearing and feeding.
Rainbow trout, Oncorhynchus mykiss, Donaldson strain (initial average body weight ± SE: 222 ± 11 g, n = 40) were fed either an LP diet or an SP diet for 20 days at 1.5% of their body weight in a single feeding daily. Each fish in a tank was identified by clipping the tip of ventral fins in different shapes, so that the growth rate of each fish could be determined. Because individual fish vary markedly in feeding rate under ad libitum conditions, we applied force feeding to ensure that each fish consumed the same amount. To do this, fish were slightly anesthetized with tricaine methane sulfonate (100 mg/l), and feed pellets were directly introduced into the stomach of each fish using a polished glass tube and a plunger. The feeding itself took less than a few seconds per fish. There was no mortality during the experimental period. All fish were treated according to the guidelines of the Institutional Animal Care and Use Committee of the University of Medicine and Dentistry of New Jersey.
The LP and SP diets contained the following ingredients (dry basis): 70% commercial trout feed (acid-washed, to remove P), 5% wheat gluten, 5% egg albumin, 10% wheat flour, and 10% soybean oil. To this mixture were added 25 g/kg vitamin and mineral premix and 30 g/kg CaCO3. The mixture was divided into two, and 38 g/kg NaH2PO4·H2O was added to the SP diet, and 16.4 g/kg NaCl was added to the LP diet (to balance Na concentration). The dough was thoroughly mixed, cold extruded, air dried, and crumbled to make pellets. The diets had the following analytical compositions (dry basis): 0.27% (LP) or 1.30% (SP) total P, 0.07% (LP) or 0.84% (SP) available P, 1.82% (LP) or 1.83% (SP) total Ca. The available P contents were determined separately following a standard digestibility method using chromic oxide as a marker (45). The available P content of the SP diet was slightly higher than the minimum dietary P requirement of trout (∼0.67%, dry basis) (34). The SP diet served as the control of the LP diet in all measurements.
Tissue collection and RNA preparation.
At days 2, 5, and 20 on the LP or SP diet, blood as well as tissues of proximal intestine, PC, and kidney were collected 6 h postfeeding from five LP and five SP fish. The proximal intestine is the upper part of the intestine, which is a light-colored and unfolded stretch (∼50% in length of the entire intestine), and is easily distinguishable from the distal intestine, which is a dark-colored stretch with a spiral-folded lining. Samples of PC were distal two-thirds from the intestinal junction. Pieces of proximal intestine, PC, and kidney were briefly rinsed in ice-cold Krebs-Ringer buffer, immediately fixed in RNAlater solution (Ambion, Austin, TX), and stored at −20°C for up to 3 mo. Total RNA was extracted from the RNAlater-stored tissues using Trizol reagent (Invitrogen, Carlsbad, CA) and stored at −80°C until further analysis.
Serum Pi and bone P concentrations of fish fed LP or SP diet were determined to verify the P status of each fish. Blood samples were collected from the caudal peduncle, and the serum was separated within 1 h by centrifugation (12,000 g, 5 min) and stored at −20°C until analysis (45). Bone samples (anterior one-third of the vertebral column) were collected from fish stored at −20°C. Fish carcasses were briefly heated to 80–90°C, muscle tissue was removed, and the vertebrae were washed in warm tap water, dried, defatted (methanol: chloroform = 1:1, vol/vol), and redried to constant weight. The dried bone samples were ashed (550°C, overnight), acid-solubilized (hydrochloric:nitric acids, 1:1, vol/vol), partially neutralized, and analyzed for P (45).
Study 1: responses of putative P-responsive genes.
The cDNA probes of CYP24 and VDR were made by RT-PCR using degenerate PCR primers (Table 1) designed from the consensus sequences of other animal species using the Clustal-W algorithm. Total RNA (4 μg) from proximal intestine, PC, kidney, liver, and gills of rainbow trout as well as rat ileum (positive control) was reverse transcribed using Moloney murine leukemia virus (M-MLV) RT system (Promega, Madison, WI) and oligo-dT20 primer following manufacturer's protocol, and 1.0 μl of this solution was amplified by PCR (Hot Star Taq DNA polymerase, Qiagen, Valencia, CA) at 4.5 mM Mg2+. The PCR condition was 95°C for 15 min, and 15 cycles of 94°C for 30 s, 65°C for 60 s, and 72°C for 60 s with 1°C decrease per cycle in annealing temperature was followed by 30 cycles of 94°C for 30 s, 50°C for 60 s, and 72°C for 60 s at constant annealing temperature. The PCR products of the expected sizes were gel purified, subcloned, and sequenced to verify the identity. Only the renal CYP24 and intestinal VDR were sequenced, because in other animals CYP24 and VDR appear to be conserved sequences among various tissues (1, 26, 46). The sequences were deposited in the GenBank (accession no. AY526907 and AY526906, respectively). Our repeated efforts to generate the cDNA probe of trout CYP27B1 were unsuccessful. NaPi-II and PiUS cDNA probes were prepared previously (45).
The abundance of CYP24 and VDR mRNA in the proximal intestine and kidney at days 2, 5, and 20 was quantified by RT-PCR Southern blot since their low abundance could not be accurately quantified by Northern blots. Total RNA (1 μg) from the proximal intestine and kidney were reverse transcribed (M-MLV RT system, Promega) with oligo-dT20 primer, and 1.0 μl of this solution was PCR amplified (iTaq DNA polymerase, Bio-Rad, Hercules, CA) at 1.5 mM Mg2+ using nested primers (Table 1). The numbers of PCR cycles were 30 (intestine) or 20 (kidney) for CYP24, and 20 (intestine) or 15 (kidney) for VDR. The optimum PCR cycles were determined beforehand using pooled samples. To avoid the risk of reaching the asymptote of amplification plateau, the number of PCR cycles was minimized. To eliminate coamplification of contaminated DNA, the primers were derived from different exons. The PCR products were separated in a 1% agarose gel, and Southern transferred onto nylon membrane (Hybond-NX, Amersham Biosciences, Piscataway, NJ) according to a standard protocol (5).
The abundance of NaPi-II and PiUS mRNA in the proximal intestine, PC, and kidney at days 2, 5, and 20 was determined by Northern blot analysis. Total RNA (15 μg) in 10 μl of diethyl pyrocarbonate water was heated (55°C for 10 min), mixed with 12.5 μl of RNA loading buffer, and electrophoresed at 70 V for ∼2 h in a 1% MOPS gel. The equal loading and the RNA quality were verified by 18S/28S rRNA abundance (not shown). Any postadjustment of data based on a housekeeping gene was not conducted since loading was virtually the same. RNA was transferred overnight with 20× SSC onto a nylon membrane (Hybond-NX, Amersham). The RNA of all fish (5 LP and 5 SP) from the same sampling day were compared on the same gel, membrane, and X-ray film to equalize analytical conditions. For PC, NaPi-II mRNA abundance was also examined by RT-PCR-Southern procedure (described above) using intestine-type NaPi-II-specific primers (44).
The membrane-transferred RNA or DNA was ultraviolet cross-linked, and baked (100°C, 1 h). The membrane was prehybridized in formamide prehybridization (FPH) solution (5× SSC, 5× Denhardt solution, 50% formamide, 1% SDS, and 100 μg of yeast tRNA/ml) for 2 h at 42°C, to which cDNA probe labeled with 32P-labeled deoxycytidine triphosphate(Rediprime II DNA labeling system, Amersham) was added and hybridized overnight at 42°C. The membrane was washed once at 42°C with 2× SSC-0.1% SDS, once at 50°C, and three times at 60°C with 0.1× SSC-0.1% SDS for 30 min each, and exposed to X-ray film (Blue Bio film, Denville Scientific) at −80°C for 3 h to 3 days. After exposure, the probe on the membrane was stripped off by boiling in 0.1% SDS (3 min × 2 times) for reprobing.
Study 2: screening of P-responsive genes by subtractive hybridization.
Total RNA of LP or SP fish (n = 5 each) was pooled, and mRNA was extracted using Oligotex mRNA Kit (Qiagen). The cDNA library construction, adaptor ligation, subtraction, and PCR amplification were performed using Clontech PCR-Select cDNA Subtraction Kit (BD Biosciences, Palo Alto, CA) following manufacturer's protocol similar to Diatchenko et al. (11).
Briefly, 2 μg of mRNA (5 fish pooled) was reverse-transcribed (M-MLV RT) with custom oligo-dT primer, and double-stranded cDNA was synthesized (RNase H, DNA polymerase I, DNA ligase) and digested (Rsa-I). The digested, blunt-ended, double-stranded cDNA was divided into two aliquots, each of which was ligated with a unique cDNA adapter. After ligation, excess SP cDNA (without adapters) was added to each of the two adapter-ligated LP cDNA at a ratio of 1:30. The mixtures were denatured at 98°C for 2 min and allowed to hybridize for 8 h at 68°C to eliminate (hybridize) common transcripts (i.e., mRNAs that were present in both LP and SP near equally or in excess in SP over LP). The second hybridization was conducted by mixing the two aliquots, to which freshly denatured SP cDNA (without adapters) was added, and hybridized at 68°C overnight (this represented a subtracted library of upregulated genes). To screen downregulated genes (in P restriction), LP and SP above were switched and followed the same protocol. Nonsubtracted libraries were prepared by directly hybridizing cDNAs having different adaptors. The subtracted libraries enriched with differentially expressed transcripts were amplified by PCR using primers that annealed to the adaptor sequences. The PCR products were ligated into pDrive cloning vector (PCR Cloning Kit, Qiagen) and subcloned (DH5α, Invitrogen, Carlsbad, CA). The cells were plated onto LB agar, which contained ampicillin and X-gal, and incubated at 37°C overnight. Approximately 200 clones from each of six samples (upregulated in intestine, PC, and kidney, and downregulated in intestine, PC, and kidney) were amplified by PCR (colony PCR) and dot blotted onto nylon membrane (Hybond-NX, Amersham) according to Brown (5). The membranes were probed with oligo-dT20 reverse-transcribed, full-length cDNA libraries (from the pooled RNA of five fish). In some cases, the membranes were also probed with nonsubtracted or subtracted cDNA libraries to detect rare transcripts. All probes were labeled with 32P-labeled deoxycytidine triphosphate. Dot blot membranes were hybridized in duplicate (one with LP and another with SP cDNA library) to identify differentially expressed genes.
Positive clones, identified by dot-blot differential screening, were then confirmed by Northern blot. The true positive clones (defined in Statistical analyses) were cultured in LB medium to obtain plasmid DNA, which was then purified (QIAprep Spin Miniprep Kit, Qiagen) and sequenced (ABI BigDye terminator, ABI 3100 Genetic Analyzer). Identities of the genes were examined by aligning deduced amino acid sequences by Blastx (nucleotide-protein translation) algorithm of NCBI. A significance threshold of 1e-3 (E value, NCBI) was applied to identify related genes. Several P-responsive genes were selected out of ∼30 P-responsive genes identified above to examine their responses at day 2 and day 5, as in study 1.
The difference in mean mRNA abundance of a gene between LP and SP fish was analyzed by t-test (5 LP fish vs. 5 SP fish). In addition, the correlation between the mRNA abundance of a putative P-responsive gene and serum or bone P concentration (P status indicators) of each fish was examined by linear regression analysis for a total of 10 fish from both diets. This approach considered both individual and diet-induced variation in mRNA abundance. The genes were considered true positives (P responsive) when the probability value of t-test or regressions was significant (P < 0.05). Because our study was concerned primarily with differential responses to dietary P concentrations, data were presented as ratios of mean mRNA abundance of LP fish to that of SP fish.
Responses of conventional P-responsive indicators.
Percent body weight gain (or loss) from initial weight did not differ significantly between LP and SP fish at any sampling day, indicating that a change in growth rate is one of the slowest responding parameters of P status (Table 2). In contrast, serum Pi concentration between LP and SP fish differed by day 5 (46%; P = 0.002) and by day 20 (98%; P < 0.001). Bone P concentration decreased slightly but significantly by day 20 (P = 0.001) but not at day 5 (P = 0.28).
Responses of putative P-responsive genes.
The subcloned trout renal CYP24 cDNA had 70% sequence identity and the translated protein sequence had 83% homology with chicken CYP24 (AAC41266). The subcloned trout intestinal VDR cDNA had 80% sequence identity and the translated protein sequence had 71% homology with zebrafish VDR (AAF21427). Based on RT-PCR, CYP24 mRNA was found in the intestine, kidney, and liver but not in PC and gills (Fig. 1A). The VDR mRNA was found in all tissues studied, including intestine, PC, kidney, liver, and gills (Fig. 1B).
The LP-to-SP ratio (LP/SP) of CYP24 mRNA abundance in the kidney was 1.85 at day 5 (P = 0.037) but was 0.67 at day 20 (P = 0.025) (Fig. 2). In the intestine, CYP24 mRNA abundance was independent of diet throughout the duration of the experiment (P = 0.07 at day 2). The LP/SP of VDR mRNA abundance in kidney was greatest at day 5 (1.93; P = 0.007) but insignificant from 1.0 at days 2 and 20 (Fig. 2). In the intestine, the LP/SP of VDR mRNA did not differ significantly from 1 at any sampling day. PiUS mRNA LP/SP in the PC was 0.84 on day 20 (P = 0.042) and in the kidney was 0.84 on day 5 (P = 0.008) and 0.89 on day 20 (P = 0.003) (Fig. 2). In the intestine, PiUS mRNA LP/SP did not differ significantly from 1 (0.84; P = 0.08 on day 20).
Dietary P restriction significantly increased the LP/SP of NaPi-II mRNA in the proximal intestine (day 2 to day 20), PC, and kidney (day 5 and day 20) (Fig. 2). The LP/SP of NaPi-II mRNA was greatest in the PC (3.8 at day 5 and 10.1 at day 20) compared with that in the intestine and kidney where the ratio was moderate (∼1.5–3.0). The RT-PCR-Southern analysis using intestine-type NaPi-specific primers gave similar results to Northern blots (44), indicating that the rapid response of PC was due to the intestine-type NaPi-II. In kidney, NaPi-II mRNA abundance was higher at day 5 than at day 20 (P = 0.001).
In the kidney at day 5, serum Pi concentration correlated significantly with LP/SP of CYP24 and NaPi-II mRNA abundance but not with LP/SP of PiUS and VDR mRNA abundance (Fig. 3). In the kidney, the LP/SP of VDR and NaPi-II mRNA abundance was much less variable at days 2 and 20 compared with that at day 5. At day 5, LP/SP of VDR and NaPi-II mRNA abundance correlated with one another (Fig. 4; day 5 correlation, P = 0.0007) but not at day 2 (P = 0.25) and day 20 (P = 0.10).
Screening of P-responsive genes by subtractive hybridization.
Approximately 1,200 clones in total were raised out of the subtracted libraries (up- and downregulated libraries from intestine, PC, and kidney), of which ∼150 candidate positive genes were identified based on the dot-blot differential screening. These candidate genes were verified by Northern blot analysis. Some were identified as redundant clones, some were fish-specific (not P specific) responses, and many were false positives. Ultimately, ∼30 dietary P-responsive genes were identified (Table 3). Many of these genes responded only modestly (LP/SP < 2 or >0.5). A known P-responsive gene (NaPi-II) was identified in PC but not in the intestine and kidney. Some additional clones were identified as marginally P responsive (P < 0.07), including ubiquitin and trehalase in the intestine and perforin 1 and unknown proteins in PC (not shown).
Selected P-responsive genes from day 20 were also examined for their responses at days 2 and 5 (Fig. 5). Meprin 1A in the intestine (P = 0.009) and apolipoprotein (apo) E in PC (P = 0.05) were each responsive at day 5 but not at day 2. In the intestine, cysteine sulfinic acid decarboxylase-like mRNA was upregulated at day 2 (P = 0.0005) and day 5 (P = 0.09). The renal procollagen C-endopeptidase enhancer, renal mitochondrial Pi carrier, and PC natural killer enhancing factor were also responsive at day 2 but not at day 5.
There is currently little information about dietary P-responsive genes in animals. In this study, we chose two strategies to identify P-responsive genes: the first tracked changes in expression of genes known to be responsive to P deficiency in mammals, and the second searched for novel P-responsive genes using subtractive hybridization. Gastrointestinal and renal tissues were chosen because these tissues assume major regulatory roles in dietary P absorption and homeostasis. The P status of each fish was verified based on conventional indicators (serum and bone P) and then compared with the mRNA abundance of candidate P-responsive genes of each fish (for days 5 and 20).
In mammals, NaPi-II is the gene known to be responsive to dietary P restriction. Previously, our laboratory reported that intestinal and renal NaPi-II mRNA abundance in trout was upregulated in chronic (∼7 wk) dietary P-restriction (7, 45). There was, however, no study in fish reporting the time-course changes of NaPi-II mRNA abundance in response to dietary P restriction. In the present study, the LP/SP of renal and PC NaPi-II mRNA abundance in trout increased within 5 days of dietary P restriction, a time course similar to that demonstrated by mammalian kidney (31). At day 2, intestinal NaPi-II mRNA abundance was upregulated in LP fish, suggesting that the intestine may be the initial responder, even though the magnitude of the intestinal response was only approximately threefold by day 20. In contrast, the LP/SP of NaPi mRNA abundance in the PC dramatically increased at day 5 and further increased to 10.1 at day 20. This indicates not only that the NaPi-II mRNA abundance in the PC is a sensitive indicator of P status but also that the mechanism of NaPi-II regulation in PC may be different from that in the intestine. Because two NaPi-II isoforms are present in trout PC (44), we examined NaPi-II abundance in PC by RT-PCR Southern blot using intestine-type NaPi-II-specific primers (data not shown), which confirmed that the intestine-type NaPi-II was responsible for the diet-induced increase of NaPi-II in PC.
In the kidney, the LP/SP of NaPi mRNA abundance was high at day 5 but then decreased from day 5 to day 20 (Fig. 2), a pattern similar to CYP24 and VDR mRNA in the kidney (Fig. 3). These genes are apparently synchronizing each other in response to dietary P, but the response appears to be transient. This transient link between NaPi-II and the calcitriol-related genes (CYP24 and VDR) indicates potential regulation of trout NaPi-II by calcitriol as in mammals. The promoter of human NaPi-II contains a vitamin D response element that binds a 1,25-VDR/RXR heterodimer complex and coregulators to initiate transcription (47). In young rats, intestinal NaPi-II mRNA abundance and Na+-dependent P transport rate increased after calcitriol injection (17, 53). Also, P deprivation increased CYP27B1 mRNA abundance in rat kidney (54) as well as stimulated CYP27B1 activity to increase calcitriol synthesis (9, 36). These results in mammals show the regulatory role of calcitriol on NaPi-II transcription and suggest that the expression of CYP27B1 is critical to NaPi-II regulation. In birds, vascular, but not luminal, perfusion of calcitriol rapidly increased intestinal Pi uptake (32). In fish, however, incubating trout intestine (2) or flounder renal proximal tubule cells (25) in a solution containing calcitriol did not affect net Pi transport. In addition, plasma calcitriol concentration in trout did not change after moderate dietary P restriction (8). These results in fish suggest that the role of vitamin D in P homeostasis may be different in fish and mammals. In mammals, other mechanisms by which dietary P regulates NaPi-II expression also include hypophosphatemic proteins that alter NaPi-II mRNA stability (27, 30), coregulators that alter the rate of NaPi-II translation (30), and a P-response element that increases the rate of NaPi-II transcription (18). Thus NaPi-II expression in trout may also involve such calcitriol-independent mechanisms.
NaPi-II mRNA was successfully isolated from PC by subtractive hybridization. The isolated region was near the 3′ end of the mRNA, where the sequence was unknown in trout but known in flounder, zebrafish, and mammals. It was shown to be upregulated nearly 10-fold (not shown in Fig. 5), confirming the result of Fig. 2that used a different cDNA probe derived from a different mRNA region. Unfortunately, but not surprisingly, NaPi-II was not isolated from the intestinal and renal subtracted cDNA libraries. The present method, according to the manufacturer, requires at least a fivefold difference in mRNA abundance to be subtracted efficiently, suggesting that moderately P-responsive genes (2- to 3-fold) might not be recovered in this study. Most genes confirmed as true positives by Northern blot were only moderately P responsive, indicating that other moderately P-responsive genes (<2-fold) may have been missed.
The CYP24 mRNA sequence has not been reported previously in any fish species except as ESTs in pufferfish and zebrafish having sequence similarity to CYP24. The presence of CYP24 mRNA in trout intestine, kidney, and liver highlights the importance of these tissues in vitamin D and, most likely, also P metabolism. However, it is not clear to us why only renal but not intestinal CYP24 mRNA abundance would change in response to dietary P and why renal LP/SP would increase on day 5 but then decrease on day 20.
Fish VDR mRNA sequences have been reported in zebrafish and flounder; however, the functional roles have not been studied. In trout, VDR mRNA appears to be widely distributed in a variety of tissues. In flounder, VDR mRNA was reported to be omnipresent but not in the liver (46). In chicken, however, trace amounts of VDR mRNA were reported in liver (26). The VDR is a zinc-finger transcription factor regulating the expression of several genes. The VDR, once liganded with calcitriol, forms a heterodimer complex with RXR, which then binds to vitamin D response element in the promoter region of target genes, and, along with several coregulators, initiates or suppresses their transcription (4, 35). In chickens and rats, an LP diet increased VDR gene expression in the intestine (28, 43). The increased expression, however, was modest in magnitude, brief in duration, and limited only to intestinal tissues (43). In trout, intestinal VDR mRNA abundance was apparently independent of diet at all times, whereas renal VDR mRNA increased significantly at day 5 (∼2-fold) but not at day 2 or day 20. Because this transient pattern seemed synchronous with those of changes in NaPi-II and CYP24 mRNA abundance, signaling mechanisms regulating these three genes may be linked to each other. However, there has been a recent report that dietary P-induced changes in expression of NaPi mRNA and protein as well as Pi transport were observed in VDR-null mice (42).
PiUS, known as inositol-hexakisphosphate kinase, has ATP synthase activity, and transfers Pi from diphosphoinositol pentakisphosphate to ADP to form ATP. The reverse reaction catalyzes formation of diphosphoinositol pentakisphosphate from inositol-hexakisphosphate using ATP (16, 48). Thus diphosphoinositol pentakisphosphate is thought to be a form of energy currency similar to creatine phosphate, but may act as a localized energy source (39, 41). In rats, dietary P restriction (7 day) upregulated intestinal PiUS mRNA abundance (∼2-fold) and Pi uptake rates (∼2-fold) (17). In trout, PiUS mRNA in the intestine and kidney increased slightly with moderate dietary P restriction (45), whereas in the present study PiUS mRNA in PC and kidney decreased slightly with more severe dietary P restriction. The modest response of PiUS to dietary P may imply that its role in P homeostasis is minor or indirect.
Other P-responsive genes.
S100 calcium-binding protein A11 (S100A11) was the most responsive gene identified by subtractive hybridization at day 20. At days 2 and 5, however, S100A11 mRNA abundance did not differ between LP and SP fish (t-test P = 0.08 and 0.12, respectively), suggesting that S100A11 may be an excellent marker of chronic P deficiency, but it is not an acute (immediate) responder to dietary P restriction. S100A11 appears to be involved in endocytosis and phosphorylation of annexin 1 (12), but its role in P metabolism is unknown. Although the amino acid sequence had the highest homology to S100A11 (65%), homology to other S100 proteins was also high (e.g., A13, A1, P). This suggests that the identity of this P-responsive gene may be another homologous S100 protein or that it is a novel S100 protein. The sequence also contained a conserved domain of S100 “intestinal” calcium-binding protein based on Blastx algorithm. However, the homology to calbindins, known intestinal calcium-binding proteins whose expression is inducible by calcitriol, was low (<45%). Obviously, the identity of this P-responsive gene must be determined.
Renal mitochondrial Pi carrier mRNA was upregulated threefold at day 20 of dietary P restriction. The response at day 2 was also significant (P = 0.0005) and almost significant at day 5 (P = 0.10). This protein transfers Pi from cytosol to mitochondrial matrix for use in oxidative phosphorylation. In mammals, chronic P deficiency is known to decrease ATP concentration in skeletal and cardiac muscles as well as kidney (19). Thus an adaptive increase in renal mitochondrial Pi carrier expression implies a vital compensatory response under conditions of P deprivation.
Meprin 1A was another P-responsive gene that was highly upregulated in the intestine and moderately upregulated in PC. The upregulated response in the intestine was already significant by day 5 (P = 0.009). Meprin is a brush-border metalloendopeptidase capable of hydrolyzing a variety of peptide and protein substrates. Another intestinal brush-border peptidase, dipeptidyl-peptidase IV, was also highly upregulated in fish fed LP diet. The role of these peptidases in P metabolism is not known, but it might be interesting to know why intestinal brush-border peptidases increase markedly during P deficiency.
Cysteine sulfinic acid decarboxylase-like mRNA abundance in the intestine and PC increased by P restriction. Other cysteine-related genes, including S-adenosyl-homocysteine hydrolase in the PC and cysteine-rich protein in the intestine also altered the mRNA abundance in 20-day dietary P restriction. It is not clear, however, how these seemingly related responses are linked to P metabolism.
P deficiency has been suggested to have several indirect effects on immune functions via vitamin D-mediated pathways (4, 35) and via depression of leukocyte function associated with decreased ATP content (19). Natural killer-enhancing factor (peroxiredoxin) was downregulated in the PC at day 20 of dietary P restriction. In humans, natural killer-enhancing factors possess antioxidant functions to protect protein and DNA from oxidative damage as well as immunoregulatory functions of natural killer activity (40). No study, however, has been reported on the possible link between natural killer cells and dietary P or vitamin D deficiency.
The apo E mRNA appears to be moderately downregulated in PC at days 5 and 20. Dietary P deficiency is well known to increase body fat content in fish (22, 34), but the effect of P deficiency on lipoprotein metabolism has not been studied in fish. In dogs, chronic dietary P restriction increased the levels of serum phospholipids, triglycerides, total cholesterol, and low-density lipoprotein (LDL) cholesterol (20). Serum levels returned to normal after P repletion. The markedly elevated LDL levels in P-deficient dogs was due mainly to an increase in levels of lipoproteins lacking apo B but high in apo E content (20). From this, it can be inferred that decreased apo E mRNA in P-deficient trout may be due to increased serum LDL enriched with apo E and consequential feedback inhibition of apo E de novo production.
Dietary P restriction increases blood collagen concentrations in pigs due to bone resorption (23). In trout kidney, P deficiency was associated with decreased collagen-related gene expression, suggesting a possible feedback regulation of collagen synthesis in a catabolic state. In the present study, several unknown genes, whose sequence did not match any genes of any species, were also detected as P responsive. Future studies should identify and characterize the function of these unknown genes.
A biomarker for P deficiency must be directly linked to P metabolism and not arise from secondary symptoms evolving from chronic P deficiency. In mammals, P deficiency and hypophosphatemia cause various metabolic disorders, including cellular hypoxia, erythrocyte dysfunction, glucose intolerance, necrosis of skeletal muscle, myocardial dysfunction, central nervous system dysfunction, and osteomalacia (14, 19). In fish, chronic dietary P deficiency typically causes decreases in growth (N retention), feed efficiency, appetite, bone strength, bone calcium, bone P, and plasma Pi, as well as increases in body fat and bone deformity (22, 34). Expression of genes not directly related to P metabolism must already be affected by the time these severe clinical symptoms are evident. We used the day 20 samples to minimize nonspecific genetic responses resulting from secondary symptoms of chronic P deficiency. At day 20 of consuming LP diet, growth rate of trout was not yet affected, whereas bone P concentration was just slightly affected. Moreover, after 20 days but not 2 or 5 days of LP consumption, in vitro intestinal Pi uptake rate increased (44). Although serum Pi concentration decreased after only 5 days of LP consumption, the change might not reflect P status of fish because serum Pi is known to be dependent on the immediate dietary P intake, fasting time, and various other factors (19). Thus it is not appropriate to rely on serum Pi alone.
Although day 20 was chosen to screen dietary P-responsive genes, many genes responded in a time-dependent or transient manner. For example, the responses of CYP24, VDR and NaPi-II mRNA in kidney were transient (day 5 > day 20). In rats, time-dependent responses of VDR and CYP27B1 have been reported (21, 43). In addition, Meyer et al. (29) recently reported an array of genes in mouse kidney that responded to 3–5 days of dietary P restriction, and indeed the profile of their genes is very different from ours. This could underscore the importance of examining P-responsive genes at different time intervals to obtain a better picture of the signaling mechanism regulating P homeostasis. Also, not only might P status have transient effects on gene expression, it might also elicit different physiological responses in different tissues and cell types. Identity of these cell types in each tissue should be investigated. Furthermore, the transient responders may have patterns of protein expression different from those of mRNA abundance.
Other differential gene expression methods.
To our knowledge, this is the only study that used the subtraction method to screen P-responsive genes. Other less time-consuming methods were initially considered. Microarray was found to be inappropriate because it was not yet well developed for trout or salmon. Existing salmonid arrays did not contain critical genes in P metabolism (e.g., NaPi-II). Arraying the subtracted library of trout is an expensive option awaiting further development and collaboration. Differential display PCR may be appropriate to identify a few responsive genes, but screening a whole library is a less practical option.
In conclusion, little is known about dietary P-responsive genes in animals. In addition to genes known to be associated with P metabolism, we found in trout dietary P-responsive genes not previously known to be associated with P deficiency. However, it is evident from this and other studies that the profile of early-response genes could be very different from that of late-response genes. Thus different duration/severity of P restriction and different tissues should be examined to obtain a time-dependent profile of P-responsive genes and to increase our understanding of the signaling mechanism involved in dietary P deficiency and homeostasis.
This work was supported by US Department of Agriculture Grants 2001-35102-09881, 2003-35102-13520, and 2004-35206-14154, as well as by National Science Foundation Grant IBN-0235011.
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