| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Endocrine Sections of the Departments of Pediatrics (T.O.C., B.K.E.) and Internal Medicine (K.L.I., W.M.P., J.S.), Yale University School of Medicine, New Haven, Connecticut 06520; and CuraGen Corp. (R.S.), New Haven, Connecticut 06511
Address all correspondence and requests for reprints to: Dr. Thomas O. Carpenter, Department of Pediatrics, Yale University School of Medicine, P.O. Box 208064, New Haven, Connecticut 06520-8064. E-mail: thomas.carpenter{at}yale.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Complex mechanisms for regulating total body phosphate (Pi) homeostasis have evolved, but these mechanisms are only partially understood, and the overall regulation of this process is not well characterized. In humans, it is apparent that the kidney is a major site of the regulation of total body Pi homeostasis, and that Pi transport occurs primarily via the NaPi2 class of electrogenically driven Na/Pi symporters in the proximal renal tubule (2). An acquired disorder of Pi homeostasis, oncogenic osteomalacia (OO; also referred to as tumor-induced osteomalacia), has been of interest because of the accompanying potent disruption of renal tubular transport, which is attributed to substances generated by the tumors that cause the syndrome. Recent studies have identified a novel member of the fibroblast growth factor family (FGF23) as a candidate mediator of this syndrome, because cells expressing FGF23, when repetitively injected into nude mice, can result in the characteristic severe phosphaturia (3). An emerging body of evidence has accumulated that implicates an important pathophysiological role for FGF23 in mediation of the OO syndrome (4, 5, 6, 7, 8, 9, 10). These data include the finding that mice overexpressing FGF23 demonstrate a rachitic Pi-wasting phenotype (4, 5), and excess phosphate retention is observed in the FGF23 knockout mouse (6). Furthermore, elevated circulating FGF23 levels, seen during symptomatic OO, correct as the renal Pi wasting resolves, in an acute time frame after tumor removal (7, 8, 9). However, whether FGF23 is the sole direct mediator of Pi transport inhibition is controversial, because evidence for inhibition in in vitro assays of renal tubular Pi transport have yielded variable results (3, 10, 11).
We identified small mesenchymal cell tumors in two patients with the typical presentation of OO (11). Upon removal, correction of biochemical parameters occurred, thereby establishing the tumors as causes of the syndrome. Using patient-derived tumor cell cultures from these individuals, we specifically compared cultures that expressed Pi transport inhibitory activity in vitro with those cultures in which activity was absent. Examination of candidate mediators identified a member of the FGF family, FGF7, as a direct potent inhibitor of renal tubular Pi uptake. That the activity in the cultures was caused by FGF7 was verified by immunoassay of tumor cell-conditioned medium specific to cells expressing the activity. Primers for FGF7 were selected and used to demonstrate the expression of FGF7 message in frozen samples from each of the tumors.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Patient 1, a 48-yr-old Caucasian man, complained of a 1.5-yr history of rib cage and lower extremity bone pain and difficulty walking. A right mandibular tumor had been resected 3 yr earlier. A mass was palpated at the site of the original surgery, and cortical expansion of upper mandible encompassing the area of the previous tumor was confirmed by radiographic examination of the jaw. Laboratory findings included low serum Pi (1.3 mg/dl), a very low renal tubular threshold maximum for Pi (<1.0 mg/dl), and elevated serum alkaline phosphatase activity (133 IU/liter; normal, 30–114). A 1 x 2-cm right midmandibular tumor was removed, with subsequent correction of serum Pi levels and clinical resolution of osteomalacia. Pathological examination revealed a spindle cell tumor with features of myoblasts. Abundant cellularity was present, but with a relatively low mitotic index.
Patient 2, a 9-yr-old girl, developed ankle, knee, and hip pain, noticed approximately 6 wk before evaluation. Lower extremity radiographs demonstrated irregularities of the distal femoral growth plates consistent with early rickets; no leg bowing was present. Biochemical evaluation confirmed hypophosphatemia secondary to renal losses and an inappropriately normal circulating 1,25-dihydroxyvitamin D level. Computed tomography of the nasal sinuses revealed a small erosion of the inner table of the left ramus of the mandible, and magnetic resonance imaging identified an abnormal structure consistent with a small tumor at the sigmoid notch of the left mandible. A chondroid tumor 2 cm in diameter was removed surgically; clinical details and tumor pathology were discussed previously (12). In both cases, tumor resection resulted in rapid correction of serum Pi, renal Pi wasting, and circulating 1,25-dihydroxyvitamin D (Table 1
). Resolution of musculo-skeletal pain was evident several weeks later.
|
Tumors were removed in the operating suite; after adequate diagnostic specimens were procured, the remaining tumor was transported to the laboratory in 
MEM (Invitrogen Life Technologies, Inc., Grand Island, NY). Tumor was minced into 3- to 5-mm diameter portions and explanted into two to four petri dishes containing MEM supplemented with penicillin, streptomycin, L-glutamine, ascorbate, and 10% fetal bovine serum, which was changed twice weekly. Cultures derived from these explanted tumor fragments reached confluence in approximately 3 wk. Conditioned medium was collected at each feeding for assay of Pi transport. With patient 1’s tumor, the original explant cultures were used, and with patient 2, cultures were continued for four or five passages.
Pi transport in OK cells
The Pi transport assay is a modification of the method described by Biber et al. (13), using a continuous opossum kidney proximal renal tubular cell line (OK, provided by Dr. J. Cole, University of Missouri, Columbia, MO). OK cells formed confluent monolayers in 5–7 d and demonstrated sodium-dependent Pi uptake at confluence. Just before confluence (4–6 d after seeding), growth medium was changed to serum-free medium for 24 h, then replaced with tumor cell-conditioned medium or unconditioned medium as a control. After an 18-h incubation, the Pi transport rate was determined. After two rinses in Pi-free buffer, transport solution (15 mM HEPES, 1.8 mM CaCl2, 5.4 mM KCl, 0.8 mM MgSO4, 116 mM NaCl, and 26 mM NaHCO3) containing 10 µCi/ml 0.1 mM KH232PO4 (PerkinElmer, Boston, MA) was added, and the mixture was agitated slowly for 5 min at 37 C. The cells were then washed four times with a sodium-free, ice-cold stop solution (15 mM HEPES, 1.8 mM CaCl2, 5.4 mM KCl, 0.8 mM MgSO4, 26 mM choline bicarbonate, and 116 mM choline chloride) and solubilized with 1 ml 0.4 M sodium hydroxide. The radioactivity in 100 µl solubilized cells was quantitated by scintillation counting and normalized to protein content. Data were expressed as a percentage of Pi uptake in OK cells incubated with control medium (unconditioned -MEM; taken to be 100%).
This system demonstrates characteristic dependence of Pi and sodium concentration, as previously described (13). The effect of up to 10% fetal calf serum was negligible. For experiments examining Pi transport inhibitory activity in candidate proteins, FGF7, IGF-binding protein-5 (IGFBP5), CD4, osteoprotegerin, and anti-FGF7 antibody were obtained from R & D Systems, Inc. (Minneapolis, MN). For neutralizing experiments, FGF7 or medium was incubated with matched vehicle or antibody solutions for 2 h before addition, and Pi transport was performed as usual.
For experiments assessing transport of a nonmetabolizable glucose analog and alanine, Pi was replaced with 0.1 mM methyl--D-glucopyranoside (6 µCi/ml) or 0.1 mM L-alanine (2 µCi/ml) in transport medium, respectively, and separate assays were run after preincubation with FGF7, as described above (14).
RNA expression analysis
Lysates were resuspended in diethyl-pyrolidine-carbonate-treated water, and RNA was quantified spectroscopically at 260 nm. After TRIzol extraction, RNA quality was evaluated by spectrophotometry and formaldehyde agarose gel electrophoresis, and RNA yield was estimated by fluorometry with OliGreen (Molecular Probes, Eugene, OR). Polyadenylated RNA was prepared from 100 mg total RNA using oligo(deoxythymidine) [oligo(dT)] magnetic beads (PerSeptive, Cambridge, MA) and quantitated with fluorometry.
RNA preparations were kept culture- and patient-specific, allowing comparison of mRNAs from tumor-derived cultures with transport inhibitory activity to mRNAs from separate cultures derived from the same tumor without this activity, thereby controlling for genetic variation while selecting for the Pi transport inhibitory property.
The method of expression analysis was comprised of three steps: restriction endonuclease digestion, adapter ligation, and PCR amplification (15). The double-stranded cDNA was digested in separate reactions by 96 restriction enzyme pairs, electrophoresed, and compared for signal intensity among samples, then identified via a gene database query (GeneCalling, CuraGen, New Haven, CT).
After double-stranded cDNA synthesis of polyadenylated RNA, cDNA pools were digested with different pairs of restriction enzymes with 6-bp recognition sites. Complementary adapters were ligated to the digested cDNA, and adapter-specific primers were used to direct 20 cycles of PCR. One adapter-specific primer was biotin-labeled, and the other was labeled with the fluorescent dye fluorophore 6-carboxyfluorescein (FAM). After PCR amplification, the biotin-labeled DNA was purified on immobilized streptavidin. Denatured single-stranded DNA fragments were electrophoresed on ultrathin polyacrylamide gels, and FAM-labeled fragments were detected by LASER excitation.
Specifically, contaminating DNA was removed with deoxyribonuclease I (Promega Corp., Madison, WI). Polyadenylated RNA was prepared by fractionation of total RNA with an mRNA purification kit using biotinylated oligo(dT)-streptavidin magnetic beads (MPG, Lincoln Park, NJ), followed by cDNA synthesis via RT of oligo(dT)-primed mRNA (Superscript II, Invitrogen Life Technologies, Inc., Carlsbad, CA) and second-strand synthesis. Terminal Pi was removed with arctic shrimp alkaline phosphatase (Amersham Biosciences, Piscataway, NJ), followed by cDNA purification by phenol-chloroform extraction. The yield of cDNA was quantitated by fluorometry using PicoGreen dye (Molecular Probes, Eugene, OR). Ninety-six enzyme pairs were selected to achieve a representative coverage of most of the possible sequences in a given DNA sample. PCR amplification using specific linkers was carried out, and final DNA products were denatured by heating to 96 C. The DNA products were electrophoresed on ultrathin polyacrylamide gels in 6 mol/liter urea. PCR products were visualized by a FAM label on the product using a multicolor laser excitation imaging system (Niagara, CuraGen, New Haven, CT).
Data obtained from these gels were queried against public and proprietary databases for likely gene matches, employing statistical and mathematical criteria (15), using restriction enzyme pair recognition site information and cDNA fragment size, determined from the migration of labeled fragment on the gels. A candidate is defined as the probability of a cDNA fragment belonging to a known gene. The cDNA fragment data were compiled as a list of likely genes to which that cDNA fragment might belong. If provisional identification of a cDNA fragment could not be obtained by querying databases, the cDNA fragment was designated as belonging to a putative novel gene. Identities of known and novel gene fragments were then independently confirmed.
Immunoassay
Immunoassay of human FGF7 [keratinocyte growth factor (KGF)] was performed using a solid phase ELISA kit (Quantikine, R & D Systems, Inc.), comprised of recombinant human FGF7 and antibodies raised against the recombinant factor. The intraassay coefficient of variation (CV) was less than 4%, and the interassay CV was less than 5.5%. Immunoassay of human FGF23 was performed using a two-site ELISA kit (Immunotopics, Inc., Santa Clemente, CA), with two goat polyclonal antibodies directed to separate epitopes in the C terminus of FGF23. The intraassay CV was less than 5.5%, and the interassay CV was less than 7.5%.
Detection of FGF7 message in tumors
A 50-mg fragment of frozen tumor from each patient was homogenized in TRIzol with an S25N-10G dispersing tool (IKA Works, Inc., Wilmington, NC). Total mRNA from HF3 human dermal fibroblasts and HK9 human keratinocytes (provided by the cell culture core facility of Yale Skin Diseases Research Center) were used as positive and negative controls, respectively. Total RNA from cell monolayers of these cell lines was extracted directly from 100 x 20-mm cell culture dishes, with TRIzol (Invitrogen Life Technologies, Inc.). RNA quantity and quality were determined using an UltraSpec 2000 (Amersham Biosciences, Piscataway, NJ) at wavelengths of 260 and 280 nm. An aliquot of each total RNA isolate was run out on a 1.0% agarose gel with ethidium bromide stain as an additional quality check. Residual DNA in each RNA isolate was removed by deoxyribonuclease I treatment (MessageClean, GenHunter Corp., Nashville, TN). RNA isolates were purified further using the RNeasy Protect Mini-kit (Qiagen, Inc., Valencia, CA). Furthermore, PCR amplification of RNA that had not been treated with reverse transcriptase was performed, and no evidence of amplifiable DNA was detected.
Gene-specific primers for human glyceraldehyde phosphate dehydrogenase (hGAPDH), and human FGF7 were created using GCG software (Accelerys, Inc., San Diego, CA). The human FGF7B sequences were 5'-gctctacagatcatgctttcac-3' and 5'-cactcttatatcccctccttcc-3', and the hGAPDH sequences were 5'-catgagaagtatgacaacagcc-3' and 5'-tgagtccttccacgatacc-3'. The expected product sizes were 160 and 115 bp, respectively. The 160-bp products were sequence-verified as FGF7-derived. RT of the RNA isolates was performed using a Superscript II Reverse Transcriptase kit (Invitrogen Life Technologies, Inc.) and gene-specific primers. PCR was performed on the cDNAs using AmpliTaq DNA polymerase and a GeneAmp Kit (PerkinElmer, Branchburg, NJ). The PCR products were analyzed on a 3.0% agarose gel with ethidium bromide stain.
Statistical analysis
Pi transport activity was determined in triplicate; significant differences between two groups were determined by unpaired two-tailed t test, and differences in multiple group comparisons were determined by ANOVA and Fisher’s least square difference (Systat, version 8.0 for Windows, SPSS, Inc., Chicago, IL). P < 0.05 was considered significant.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
After surgical removal and procurement of adequate diagnostic specimens, we cultured the residual tumor tissue by explanting fragments into separate petri dishes, each of which developed into a semiconfluent culture. Four cultures were established from tumor 1, and two from tumor 2. Culture medium was conditioned by exposure for 3 d, harvested, and tested weekly for effects on Pi transport. Medium conditioned by exposure to two cultures from patient 1’s tumor consistently inhibited Pi transport to approximately 40% of that in OK cells incubated with nonconditioned medium (Fig. 1, A and C). In contrast, a third culture produced only transient inhibitory activity on d 15, after which no inhibitory activity was evident (Fig. 1B). One culture maintained a moderate degree of inhibitory activity, but with greater variability than the others (Fig. 1D). Two cultures were derived from patient 2’s tumor, one of which maintained inhibitory activity throughout five passages of the cells (68% of control) and one of which did not, losing activity in the third passage, and followed through a fourth passage with no activity (103% of control). RNA was prepared from all of these cultures. Medium harvested at the time of RNA preparation demonstrated continued secretion of inhibitory activity in the active cultures.
|
The comprehensive differential gene expression analysis surveyed over 4000 mRNA fragments, representing approximately 2000 genes. This analysis revealed mRNAs derived from only eight genes to be significantly more abundant in cell lines with Pi transport inhibitory activity than in cell lines without this activity (Table 2). A comparative profile of increased expression of FGF7 from cultures secreting inhibitory activity is shown in Fig. 2.
|
|
Six of the identified mRNAs encode secreted proteins, four of which we directly tested for Pi transport inhibitory activity in vitro. FGF7 significantly inhibited Pi transport (Fig. 3A; P = 0.029). In contrast, there was no significant Pi transport inhibition with CD4 (Fig. 3B), IGFBP5 (Fig. 3C), or osteoprotegerin (Fig. 3D). A dose-ranging study demonstrated increasing inhibition at doses between 30–1000 ng/ml (Fig. 4A). A comparison of the FGF7 dose response to that of PTH, the best known physiological regulator of renal Pi transport, suggests that FGF7 inhibits Pi uptake in vitro at doses comparable to those of human PTH-(1–34) (Fig. 4B).
|
|
In contrast to the absence of an IGFBP5 effect, FGF7-induced Pi transport inhibitory activity was abrogated by coincubation with antibodies to FGF7 in a dose-dependent manner (P = 0.02). In the presence of 200 ng/ml FGF7, coincubation with increasing concentrations of antibody progressively attenuated the inhibitory effect of FGF7 (Fig. 5A); when a maximal inhibitory concentration of FGF7 (1000 ng/ml) was used, antibody at a concentration of 20 µg/ml attenuated, but did not fully neutralize, the FGF7 effect.
|
). Comparison of effects of FGF7 on renal transport of other solutes
The effects of FGF7 on renal transport of alanine and methylglucopyranoside, a nonmetabolizable glucose analog, were tested to determine the relative effects on these renal solutes in comparison with the effect on Pi transport. Figure 6 illustrates the typical dose-dependent response of Pi transport (Fig. 6A), the less potent effects on methylglucopyranoside transport (Fig. 6B), and the lack of effect on alanine transport (Fig. 6C). Thus, there was relative specificity for the inhibitory effect of FGF7 on Pi transport.
|
We then demonstrated that the concentration of FGF7 in active conditioned medium was markedly greater than that in conditioned medium without transport inhibitory activity or that in control medium not exposed to cells (Fig. 7; P < 0.0001). Finally, FGF23 was of only low abundance in conditioned medium from cultures with (15 pg/ml) and without (16 pg/ml) Pi transport inhibitory activity, indicating that Pi transport inhibitory activity in active conditioned medium could not be attributed to increased FGF23 concentration, but, rather, to the marked accumulation of FGF7.
|
PCR was used to confirm FGF7 expression in original tumor tissue from each patient. FGF7 mRNA was expressed in original tumor from each of the two patients and in human dermal fibroblasts (HF3 cells), whereas human dermal keratinocytes (HK9 cells) did not express FGF7. GAPDH was expressed in all tumor fragments and both cell lines (Fig. 8).
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
FGF7, also known as KGF, is a 28-kDa member of the FGF family, and has been demonstrated to be an active stimulus in wound healing and highly expressed in keratinocytes (25). FGF7 is expressed in other epithelial tissues, including gastrointestinal epithelium, transitional urothelial cells, and type II pneumocytes (26). Its expression is increased in the setting of inflammatory bowel disease (27), has been proposed as a potential tumor marker (28), and has been shown to protect epithelial cells from toxicity from reactive oxygen derivatives (25). FGF7 has been considered an agent to treat oral mucositis (26). Transport-mediating functions of this factor have not been described. FGF7 is believed to selectively activate the FGFR2b receptor, as does FGF10 (29), whereas FGF23 has been shown to bind to FGFR2c and FGFR3c (30). A recent clinical description of a family with hypophosphatemia and skeletal abnormalities evident in members with an activating mutation of FGFR1c, has suggested that this receptor may also mediate renal phosphate wasting (31). Thus, it appears that activation of three different FGF receptors are able to mediate the inhibitory effects of FGFs on Pi transport in renal epithelium.
In contrast to the experimental approaches used to identify FGF23, sFRP4, and matrix extracellular phosphoglycoprotein, our studies selected for tumor cell products that directly inhibit renal tubular Pi transport in vitro. Of several candidate proteins, FGF7 demonstrated potent inhibitory activity of renal tubular Pi transport in an established in vitro system. That FGF7 was responsible for inhibitory activity was demonstrated by performing experiments with recombinant human FGF7 and using neutralizing antibody to FGF7 to block the activity. Renal tubular Pi wasting appears to occur in response to a variety of factors, and this work together with that of previous investigators would suggest that FGF receptor signaling is critical to the pathogenesis of the disease. One possibility is that a variety of FGF receptor agonists may be potential mediators of these intriguing disorders, and that a variety of factors may modulate the expression of the final direct actors in these pathways. Such factors may prove useful in the treatment of disorders of elevated Pi, such as hyperphosphatemia associated with renal disease.
|
Footnotes |
|---|
Abbreviations: ADHR, Autosomal dominant hypophosphatemic rickets; CV, coefficient of variation; FAM, 6-carboxyfluorescein; FGF, fibroblast growth factor; hGAPDH, human glyceraldehyde phosphate dehydrogenase; IGFBP5, IGF-binding protein-5; oligo(dT), oligo(deoxythymidine); KGF, keratinocyte growth factor; OO, oncogenic osteomalacia; Pi, phosphate; XLH, X-linked hypophosphatemia.
This work was supported by the NIH-supported Yale Core Center for Musculoskeletal Diseases (Physiology and Molecular Cores; Grant P30-AR-46032), individual NIH awards (K24-HD-01288 to T.O.C. and R01-DK-45228 to K.L.I.), and the Cell Culture Core of the Yale Skin Diseases Research Center (Grant AR41942; R.E. Tigelaar, Program Director).
Received February 26, 2004.
Accepted November 4, 2004.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
1i collagen promoter exhibit growth retardation, osteomalacia, and disturbed phosphate homeostasis. Endocrinology 145:3087–3094[CrossRef][Medline]
This article has been cited by other articles:
 |
|
 |
|
  S. Liu, W. Tang, J. Fang, J. Ren, H. Li, Z. Xiao, and L. D. Quarles Novel Regulators of Fgf23 Expression and Mineralization in Hyp Bone Mol. Endocrinol., September 1, 2009; 23(9): 1505 - 1518. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Berndt and R. Kumar Novel Mechanisms in the Regulation of Phosphorus Homeostasis Physiology, February 1, 2009; 24(1): 17 - 25. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Liu and L. D. Quarles How Fibroblast Growth Factor 23 Works J. Am. Soc. Nephrol., June 1, 2007; 18(6): 1637 - 1647. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. A. Imel, M. Peacock, P. Pitukcheewanont, H. J. Heller, L. M. Ward, D. Shulman, M. Kassem, P. Rackoff, M. Zimering, A. Dalkin, et al. Sensitivity of Fibroblast Growth Factor 23 Measurements in Tumor-Induced Osteomalacia J. Clin. Endocrinol. Metab., June 1, 2006; 91(6): 2055 - 2061. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. A. Imel and M. J. Econs Fibroblast Growth Factor 23: Roles in Health and Disease J. Am. Soc. Nephrol., September 1, 2005; 16(9): 2565 - 2575. [Full Text] [PDF]
|
|
|
|
 |
|
 T. Berndt and R. Kumar The Phosphatonins and the Regulation of Phosphorus Homeostasis IBMS BoneKEy, June 1, 2005; 2(6): 5 - 16. [Full Text] [PDF]
|
|
|
|
|
|
E. Seeman and G. J. Strewler Clinical and Basic Research Papers - February 2005 Selections IBMS BoneKEy, March 1, 2005; 2(3): 1 - 5. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
160 bp) was expressed in the positive control (lane 5) and in each of the two original tumors (lanes 3 and 4), but not in the negative control (lane 1). Lower panel, Lanes as in upper panel; hGAPDH (