Tyrosine sulfation is a common modification of many proteins, and the ability to phosphorylate tyrosine residues is an intrinsic property of many growth-factor receptors. In the present study, we have utilized the peptide hormone CCK8 (cholecystokinin), which occurs naturally in both sulfated and unsulfated forms, as a model to investigate the effect of tyrosine modification on metal-ion binding. The changes in absorbance and fluorescence emission on Fe3+ binding indicated that tyrosine sulfation or phosphorylation increased the stoichiometry from 1 to 2, without greatly affecting the affinity (0.6–2.8 μM at pH 6.5). Measurement of Ca2+ binding with a Ca2+-selective electrode revealed that phosphorylated CCK8 bound two Ca2+ ions. CCK8 and sulfated CCK8 each bound only one Ca2+ ion with lower affinity. Binding of Ca2+, Zn2+ or Bi3+ to phosphorylated CCK8 did not cause any change in absorbance, but substantially increased the change in absorbance on subsequent addition of Fe3+. The results of the present study demonstrate that tyrosine modification may increase the affinity of metal-ion binding to peptides, and imply that metal ions may directly regulate many signalling pathways.
- ferric ion
Modification of tyrosine residues is a critical post-translational modification of many proteins. Tyrosine sulfation occurs in the trans-Golgi network, and is thought to be important in protein secretion [1,2]. A more specific role for tyrosine sulfation in viral entry into cells has been revealed by the observation that sulfation of Tyr10 and Tyr14 of the chemokine receptor CCR5 facilitates interaction with the envelope glycoprotein of HIV type 1 . Tyrosine phosphorylation is a common feature of many intracellular signalling pathways . For example ligand-dependent activation of the tyrosine kinase activity of growth-factor receptors results in autophosphorylation of their cytoplasmic domains. The phosphotyrosine residues are essential elements in the docking sites for proteins that trigger the signalling cascades that are ultimately responsible for cell proliferation. Although Fe3+ (ferric ions) are known to bind avidly to the phosphoserine residues of phosvitin and casein , and to induce aggregation of the hyperphosphorylated tau protein of Alzheimer's disease , the binding of Fe3+ to tyrosine-modified peptides or proteins has not been reported previously.
The octapeptide of the hormone CCK (cholecystokinin; CCK8, DYMGWMDFamide) provides a convenient model system in which to study the effects of tyrosine modification on metal-ion binding. CCK, which was originally isolated as a 33-residue peptide from the mucosa of the gastrointestinal tract, is responsible for gall-bladder contraction and pancreatic enzyme secretion, and also functions as a neurotransmitter in the central nervous system . Truncation of the N-terminal end of CCK33 to CCK8 occurs naturally, and has no effect on immunoreactivity or bioactivity, but sulfation on the sole tyrosine residue greatly increases receptor binding and biological potency . CCK is structurally and functionally related to the gastric peptide hormone gastrin (ZGPWLEEEEEAYGWMDFamide), with which it shares a common amidated C-terminal pentapeptide. The first reported biological activity of gastrin was the stimulation of gastric acid secretion, but gastrin is now also recognized as an important growth factor for the gastric mucosa . Like CCK, gastrins occur in sulfated and unsulfated forms . Although gastrin can also be phosphorylated by the EGF (epidermal growth factor) receptor tyrosine kinase in vitro , there are no reports of phosphorylated gastrin or CCK occurring naturally.
Gastrins bind two Fe3+ , the first to Glu7 and the second to Glu8 and Glu9 . Fe3+ is essential for the biological activity of non-amidated forms of the peptide as a stimulant of cell proliferation and migration. Thus either the substitution E7A, or treatment with the iron chelator desferrioxamine, completely blocked the biological activity of glycine-extended gastrin . In contrast, Fe3+ was not required for the biological activity of amidated gastrin . In the present study we anticipated that the high affinity of gastrin for Fe3+ might be disadvantageous, as the contribution from phosphorylation or sulfation of the tyrosine residue would be less apparent. Because the binding of ferric ions to CCK8 is much weaker than to gastrin, and since CCK8SO4 (sulfated CCK8) is more readily obtainable than sulfated gastrin, we chose to study the effects of tyrosine modification on metal-ion binding using CCK8 as a model system. Although CCK8PO4 (phosphorylated CCK8) does not occur naturally we also examined the binding of metal ions to CCK8PO4 to allow direct comparison with CCK8SO4.
CCK8 and CCK8SO4 (89 and 93% pure respectively) were purchased from Research Plus. CCK8PO4 (81% pure) was from Peptide Solutions. All peptides were C-terminally amidated, and the impurities consisted of water and salts.
Absorption spectra of peptides [40 μM in 10 mM sodium acetate (pH 4.0) or 10 mM sodium-Pipes (pH 6.5) containing 100 mM NaCl and 0.005% Tween 20] in the presence of increasing concentrations of Fe3+ were measured against a buffer blank, in 1 ml quartz cuvettes thermostatically maintained at 25 °C, with a Cary 5 spectrophotometer (Varian).
The tryptophan fluorescence of peptides (10 μM in the buffers described above) in the presence of increasing concentrations of Fe3+ was measured in 3 ml quartz cuvettes thermostatically maintained at 25 °C, with a Spex Fluorolog-τ2 spectrofluorimeter (Spex Industries), with the excitation and emission wavelengths set at 290 and 345 nm respectively.
CCK8SO4 was dissolved in 90% H2O/10% 2H2O. CCK8 required the presence of 2H6-DMSO (80% H2O/10% 2H2O/10% 2H6-DMSO) to achieve solubility at 0.23 mM. The pH was adjusted to 4.0 or 6.5 with NaO2H/2HCl, and pH readings are uncorrected for the presence of 2H2O. 1H NMR spectra were recorded at 298 K on Bruker Avance 500 or 600 spectrometers, and referenced to 2,2-dimethyl-2-silapentane-5-sulfonate at 0 p.p.m. via the chemical shift of the H2O resonance at 4.77 p.p.m., as described previously . Sequence-specific 1H NMR resonance assignments were made from two-dimensional NOESY (nuclear Overhauser enhancement spectroscopy), TOCSY and DQF-COSY (double-quantum-filtered correlation spectroscopy) spectra. Two-dimensional spectra were analysed using Sparky 3 (T.D. Goddard and D.G. Kneller, University of California, San Francisco).
Expression of CCK1 and CCK2 receptors in COS-7 cells
COS-7 cells were cultured at 37 °C in 5% CO2 in DMEM (Dulbecco's modified Eagle medium; Gibco) supplemented with 5% FBS (foetal bovine serum) in 75 cm2 flasks (Nunc) until 95% confluent. On day one, the cells were dislodged with 0.25% trypsin/0.02% EDTA and seeded into 100 mm Petri dishes at 7.5×105 cells/10 ml per dish. Cells were transfected on day two using the DEAE-dextran method with 2.5 μg of pRFNEO plasmids encoding the human CCK1 receptor or the human CCK2 receptor, as described previously . After overnight incubation the cells were collected from the Petri dish with trypsin/EDTA, seeded into the wells of a 24-well plate (20000 cells/well) and incubated in standard conditions for 72 h before the binding assay was performed.
Dilutions of ligands were prepared in binding buffer (DMEM with 0.1% BSA, 0.15 mM PMSF and 0.05% Bacitracin). Transfected COS-7 cells were washed twice with PBS and incubated in binding buffer (150 μl/well) containing the ligands under investigation and sulfated [125I]-Bolton and Hunter labelled-CCK8 (50000 cpm/well; Amersham Biosciences) for 90 min at 37 °C on a slowly rotating platform. The binding solution was then aspirated, and the cells were washed once with PBS and dissolved in 0.25 M NaOH (300 μl/well). Radioactivity in the resulting solution was measured in a gamma-counter (LKB Wallac).
Measurement of Ca2+ binding
The change in free [Ca2+] during addition of aliquots of CaCl2 to peptides (15–40 μM) in 10 mM sodium-Pipes (pH 6.5), 100 mM NaCl, 0.005% Tween 20 and 0.4% DMSO was measured at 25 °C with a uniPROBE Ca2+-selective electrode (TPS) connected to a Hanna 8521 pH meter (Hanna Instruments), as described by Park et al. . The electrode was first calibrated with solutions of known [Ca2+] in the range 1–1000 μM in the same buffer. The concentration of Ca2+ bound to each peptide was calculated by subtraction of the free [Ca2+] from the total added [Ca2+].
Curve fitting and statistics
Values (expressed as means±S.E.M.) for the independent binding of Fe3+ or Ca2+ to CCK8PO4 were fitted to one-site or two-site ordered models with the program BioEqs [16,17]. Because of the large number of parameters and the limited number of data points, estimates of the equilibrium constants and absorbance ratios for the interaction of Fe3+ with CCK8PO4 in the presence of Ca2+ were obtained by simulation with the program Sigmaplot with the competitive two-site ordered model shown in Figure 6, and the experimentally determined equilibrium constants and absorbance ratios given in Table 1 for the interaction of CCK8PO4 with Fe3+ or Ca2+ alone.
Receptor-binding data were analysed by one-way analysis of variance, followed by Bonferroni's t test. Differences with P values<0.05 were considered significant.
Sedimentation equilibrium analysis of the interaction of CCK8PO4 with Fe3+, and the chemical shifts of CCK8SO4 and CCK8PO4 is provided online as Supplementary Figure S1 (at http://www.BiochemJ.org/bj/416/bj4160077add.htm).
RESULTS AND DISCUSSION
Binding of Fe3+ to tyrosine-modified CCK
The effect of the addition of Fe3+ on the absorption spectrum of CCK4 (the C-terminal tetrapeptide of CCK4), CCK8, CCK8SO4 and CCK8PO4 was first investigated at pH 4.0. This pH value was chosen in order to avoid any problems with precipitation of ferric hydroxides. Although no change in the absorbance of CCK4, CCK8 or CCK8SO4 was observed at 275 nm, the absorbance of CCK8PO4 increased to a maximum of 194% after the addition of 1.77 mol of ferric chloride/mol of peptide (Figure 1A). Because a two-site model, with dissociation constants of 0.22 pM and 0.13 μM, did not adequately fit the non-linearity of the absorbance data (solid line, Figure 1A), the possibility of dimerization was investigated. Measurement of the molecular mass of CCK8PO4 by analytical ultracentrifugation (Supplementary Figure S1) gave very similar values in the presence of EDTA (1326 Da) or Fe3+ (1257 Da), and both values were similar to the theoretical value for the monomer of 1142.7 Da. Dimerization of Fe3+ to form the species Fe2(OH)24+ has also been reported , but speciation plots indicated that no significant amount of the dimeric species would exist at the pH values and [Fe3+] used in the present experiments. Since there was no evidence for peptide or iron dimerization, CCK8PO4 appeared to belong to the growing class of proteins which demonstrate allosteric effects as monomers .
The titration experiments were then repeated at pH 6.5, to assess the contribution to Fe3+ binding of ionization of the modified tyrosine residue. The pKa of the phosphoryl group of phosphotyrosine is 5.9 , and although a value for the pKa of the sulfate group of sulfotyrosine has not been published, electrophoretic data indicate that the pKa is slightly greater than 4.5 . At pH 6.5, no precipitation of iron hydroxides was observed by centrifugation of the sample at the end of the experiment, provided the total Fe3+ concentration did not exceed 100 μM. The absorbance of CCK8PO4 increased on addition of ferric chloride to a maximum of 169% after the addition of 2.28 mol of ferric chloride/mol of peptide (Figure 1B). An ordered two-site model, with dissociation constants of 0.68 μM and 0.77 μM, gave a good fit to the absorbance data. The absorbance of CCK8 and CCK8SO4 also increased on addition of ferric chloride, to maxima of 131 and 136% respectively (Figures 1B and 1D). A two-site model, with dissociation constants of 2.80 μM and 4.69 μM, again gave a good fit to the absorbance data for CCK8SO4. In the case of CCK8, a one-site model, with a dissociation constant of 0.60 μM, gave a better fit to the absorbance data. No evidence of binding of Fe3+ to CCK4 was observed (Figure 1D). Thus an increase in pH from 4.0 to 6.5 enhanced the binding of Fe3+ to both CCK8 and CCK8SO4, consistent with the direct involvement of the deprotonated sulfate or phosphate group in metal-ion binding. At pH 6.5, sulfation or phosphorylation of CCK8 increased the stoichiometry of iron binding from 1 to 2, without greatly affecting the affinity (Table 1).
The effect of the addition of Fe3+ on the tryptophan fluorescence of CCK8, CCK8SO4 and CCK8PO4 was also monitored. Analysis of the quenching of tryptophan fluorescence on addition of ferric chloride at pH 4.0 had previously revealed that glycine-extended gastrin bound two Fe3+ with an apparent dissociation constant of 0.6 μM . Although the fluorescence of CCK8 or CCK8SO4 was quenched by Fe3+, the quenching fitted the Stern–Volmer relationship (Figures 2A and 2B), and is therefore likely to be the result of random collisions rather than complex formation . In contrast, quenching of the fluorescence of CCK8PO4 by ferric ions deviated from the Stern–Volmer relationship (Figures 2A and 2B). After allowance for the collisional component of quenching, the fluorescence data for CCK8PO4 were reasonably well fitted (Figures 2C and 2D) by the two-site model with the affinity constants determined from the absorbance data as described in the previous paragraph and given in Table 1. Thus the fluorescence data are also consistent with the conclusion that CCK8PO4 binds two Fe3+ with micromolar affinity.
The possibility must be considered that the observed effects of Fe3+ on the absorbance and fluorescence of CCK peptides may be an artefact caused by precipitation of ferric hydroxides from solution, followed by passive adsorption to the peptide. The following arguments strongly suggest that such artefacts were avoided. First, a series of absorbance experiments was performed with 40 μM CCK peptides at pH 4.0, at which pH a 1 mM FeCl3 solution is stable indefinitely. Secondly, no precipitation was observed in the absorbance experiments at pH 6.5 when the concentration of Fe3+ was less than 100 μM. These experimental observations were confirmed at both pH values by speciation plots which did not indicate any of the polymeric species [Fe2(OH)2]4+or [Fe3(OH)4]5+, formation of which precedes precipitation . Thirdly, the experiments were internally controlled. Thus, at pH 6.5, no increase in absorbance was seen with CCK4, and the observed increase differed between CCK8, CCK8SO4 and CCK8PO4. It seems improbable that passive adsorption should differ so significantly between four such closely related peptides. At pH 4.0 an increase in absorbance was seen only with CCK8PO4, and not with CCK4, CCK8 or CCK8SO4. If passive adsorption were the explanation of the observed increases, then CCK8 and CCK8SO4 would have to adsorb iron differently at the two pH values. Finally, attempts were made to eliminate the possibility of precipitation by the addition of Fe3+ as ferric citrate rather than FeCl3. However, citrate binds Fe3+ more tightly than any of the CCK peptides, as shown by the relative association constants (Table 1), and consequently no increase in absorbance was observed for CCK8, CCK8SO4 or CCK8PO4 during titration with ferric citrate. We conclude that the most likely explanation for the observed changes in absorbance and fluorescence of CCK peptides on addition of Fe3+ is the formation of the Fe–CCK complexes described above.
In order to define the ligands involved in Fe3+ binding, the effect of Fe3+ on the NMR spectra of CCK8, CCK8SO4 and CCK8PO4 was investigated. Assignments of the spectrum of CCK8SO4 at pH 6.5 and 298 K (Supplementary Table S1 at http://www.BiochemJ.org/bj/416/bj4160077add.htm) were in general agreement with the data of Fournié-Zaluski et al. . Assignments of the spectrum of CCK8PO4 at pH 6.5 and 298 K (Supplementary Table S2 at http://www.BiochemJ.org/bj/416/bj4160077add.htm) have not been reported previously. At pH 4.0, addition of 1 mol/mol FeCl3 to CCK8 (results not shown) or CCK8PO4 (Figure 3A) resulted in a significant reduction of peptide signal. A precipitate was observed in the NMR tube, and no changes in chemical shifts of the peptide remaining in solution were observed. These results indicated the formation of an insoluble Fe3+–peptide complex. Addition of 1 mol/mol of FeCl3 to CCK8SO4 resulted in a loss of approx. 50% of the signal intensity (Figure 3C). However, in this case, the chemical shifts of the peptide remaining in solution after addition of FeCl3 showed small but significant downfield movements, in particular for the CHβ resonances of both Asp1 and Asp7, consistent with an interaction between these residues and ferric ions. The largest changes observed were 0.04 p.p.m. for the Asp7 CHβ and CHβ′ resonances, twice that of the Asp1 peaks.
The effect of Fe3+ on the NMR spectra of CCK8, CCK8SO4 and CCK8PO4 at pH 6.5 was also investigated. To avoid precipitation of ferric hydroxides at the higher iron concentrations required for NMR titration experiments, Fe3+ was added as ferric citrate. Although there was no change in the spectrum of CCK8 on the addition of ferric citrate, with CCK8PO4 there was a general broadening of all peaks in the spectrum (Figure 3B). In the case of CCK8SO4, in addition to the broadening, there were also small incremental downfield shifts in the Asp1 CHβ and CHβ′ resonances (Figure 3D), to a maximum of 0.02 p.p.m. at 10 mol/mol of added Fe3+. Parallel changes in chemical shifts to a maximum of 0.01 p.p.m. were observed for the Asp7 CHβ and CHβ′ resonances, but the chemical shifts of other resonances were unaffected by addition of ferric citrate, consistent with selective interaction between Asp1 and Asp7 and Fe3+. The magnitude of the NMR shifts was more consistent with a distant interaction with the metal ion than with direct ligation. Interestingly, the absence of any effect of Fe3+ on specific resonances of CCK8PO4 suggested that, in contrast with CCK8SO4, the metal ion does not interact with the sidechains of the two aspartate residues, and hence presumably interacts directly with the phosphate group.
Effect of Fe3+ on CCK receptor binding
The role of Fe3+ in the binding of CCK8 and its modified derivatives to either human CCK1 or CCK2 receptors was then examined. Although sulfation of CCK8 increased its affinity for both the human CCK1 (Figure 4A) and CCK2 (Figure 4B) receptors, phosphorylation of CCK8 had the opposite effect. The IC50 values for the binding of CCK8, CCK8SO4 and CCK8PO4 to the CCK1 receptor were 35±17 nM, 7.1±0.7 nM and >1 μM respectively, and the corresponding values for binding to the CCK2 receptor were 6.9±2.0 nM, 2.1±0.7 nM and 43±7 nM respectively. The IC50 values for the binding of CCK8 and CCK8SO4 were similar to the values previously reported for the CCK1 and CCK2 receptors. The observation that the iron chelator desferrioxamine had no significant effect on the binding of [125I]-Bolton and Hunter labelled-CCK8SO4 to either the human CCK1 receptor or the CCK2 receptor (Figure 4C) indicated that Fe3+ was not essential for binding of CCK8SO4 to either receptor. This observation is in agreement with previous reports that the sulfate group of CCK8SO4 interacts directly with Met195 and Arg197 of the CCK1 receptor [24,25], and with either Arg57 and Tyr61 , or His207 and Arg208 , of the CCK2 receptor. We conclude that enhancement by metal binding is not the explanation for the previous observation that sulfation of CCK8 increases CCK1 receptor binding and biological activity .
Binding of other metal ions to tyrosine-modified CCK
To determine whether or not CCK8 and related peptides bound Ca2+ as well as Fe3+, the changes in free [Ca2+] during addition of aliquots of calcium chloride to CCK8, CCK8PO4 or CCK8SO4 were measured with a Ca2+-selective electrode. A pH of 6.5 was chosen to maximize the chance of detecting an interaction. The binding curves (Figure 5A) indicated that CCK8 and CCK8SO4 each bound 1 mol of Ca2+/mol of peptide, with dissociation constants of 83 and 870 μM respectively. For CCK8PO4, the data were better fitted by a two-site (Kd1=46 μM, Kd2=580 μM) than a one-site model (Kd=24 μM). With either model the affinity of CCK8PO4 for Ca2+ was greater than the affinity of CCK8, and both peptides had higher affinity for Ca2+ than CCK8SO4.
To determine whether or not the binding of other metal ions to CCK8PO4 enhanced the binding of Fe3+, the effects of prior addition of Ca2+, Zn2+ or Bi3+ on the changes in absorption spectrum of CCK8PO4 in response to Fe3+ were investigated. None of the above metal ions caused any increase in the absorption maximum at 275 nm for CCK8PO4 (Figures 5B and 5C) or CCK8SO4 (results not shown) at pH 6.5. For CCK8PO4 with added Ca2+ (Figure 5B), Zn2+ (Figure 5C) or Bi3+ (results not shown), subsequent addition of Fe3+ caused an increase in absorption significantly greater (3.5-fold) than the 1.5-fold increase observed in the presence of Fe3+ alone. Although further work will be required to define the precise mechanism of this effect, the data was fitted with reasonable accuracy by the competitive two-site ordered mechanism shown in Figure 6. The data is consistent with the conclusion that the two Fe3+ binding sites can also bind Ca2+, Zn2+ or Bi3+ ions without any effect on peptide absorption, but that binding of Ca2+, Zn2+ or Bi3+ to the first site enhances not only the changes in absorption on binding of a Fe3+ to the second site, but also the affinity of Fe3+ for the second site. In the case of Ca2+ the enhancement of absorption and affinity for the subsequent binding of Fe3+ are 2.5-fold and 500-fold respectively. In contrast, for CCK8SO4 with added Ca2+ or Zn2+, subsequent addition of Fe3+ caused an increase in absorption less than that observed in their absence.
Our observation that phosphorylation or sulfation of the sole tyrosine residue of CCK8 creates an additional binding site(s) for Fe3+ is the first report that tyrosine modification enhances metal-ion binding to a peptide. Phosphorylation of CCK8 also creates an additional binding site for Ca2+. Although the Ca2+ concentration in living cells rarely exceeds 10 μM, whereas the affinity for binding of Ca2+ to CCK8PO4 is 46 μM, once a Fe3+ is bound, this value reduces to 5.9 μM (Figure 6). Hence, in the presence of other metal ions, the Ca2+-binding site on CCK8PO4 could be greater than 50% occupied. It should also be borne in mind that binding of Ca2+ to other phosphorylated peptides may well be tighter than to CCK8. Similarly, although there is no change in absorbance on binding of Ca2+, Zn2+ or Bi3+ to CCK8PO4, their presence enhances the interaction of the peptide with Fe3+. Finally, although metal binding does not appear to influence binding of CCK8SO4 to either the human CCK1 or CCK2 receptors, elucidation of the general biological significance of the observation that tyrosine modification enhances metal-ion binding will require further investigation of other peptides and proteins.
The results of the present study may have significant biological implications in several different fields. For example binding of a metal ion to a phosphotyrosine residue in the cytosolic domain of a growth factor receptor might interfere with the binding of an SH2 domain of a downstream target molecule. Such interference might permit direct regulation of intracellular signalling pathways in response to alteration of the intracellular concentration of a particular metal ion such as Ca2+. Although Fe3+ are unlikely to be present in the reducing environment within cells, binding of a Fe3+ or other metal ion to a sulfotyrosine residue in the extracellular domain of a viral receptor might interfere with the binding of the viral envelope glycoproteins. Hence competitive binding by metal ions might be a useful strategy for prevention of viral infection of target cells.
This work was supported in part by grants from the National Health and Medical Research Council of Australia (400062, 454322; to G. S. B.) and the National Institutes of Health (5RO1GM065926-05; to G. S. B. and R. S. N.).
Abbreviations: CCK, cholecystokinin; CCK8PO4, phosphorylated CCK8; CCK8SO4, sulfated CCK8; DMEM, Dulbecco's modified Eagle's medium
- © The Authors Journal compilation © 2008 Biochemical Society