Peptidyl Fluorescent Chemosensors for the Detection of Divalent Copper

Anal. Chem. 2003, 75, 1706-1712

Peptidyl Fluorescent Chemosensors for the Detection of Divalent Copper Yujun Zheng,† Xihui Cao,† Jhony Orbulescu,† Veeranjaneyulu Konka,† Fotios M. Andreopoulos,*,‡ Si M. Pham,‡ and Roger M. Leblanc*,†

Department of Chemistry, University of Miami, Coral Gables, Florida 33124-0431, and Department of Surgery, School of Medicine and Department of Biomedical Engineering, University of Miami, Miami, Florida 33136

Fluorescent organic chemosensors for the detection of divalent copper with high selectivity and sensitivity are the subject of intense research in the recent years. Structurally, ionophore and fluorophore are two essential parts determining the resultant performance of the chemosensor. While much work has been focused on designing highly selective ligands, little attention has been paid to the possible influence of ionophore-fluorophore interaction on their properties in metal ion binding. We studied here fluorescent chemosensors based on the GlyHis peptidyl motif and found that the functionality of the chemosensors was greatly influenced by the spatial alignment of the fluorophore in the molecules. In Gly-His-Lys(Dns) (1), the dansyl group is on a side branch and does not interact with copper, while in Dpr(Dns)-His-Lys (2), the dansyl group is also on a side branch but the close placement allows it to directly participate in the binding with copper ions. Therefore, although dansyl can signal the binding event in both molecules, the mechanisms involved are quite different, and this difference resulted in different sensing performance, e.g., the selectivity. Even more strikingly, the dansyl group in Dns-Gly-His-Gly (3) exhibited a profound effect on the molecular complexation. The binding constant decreased, and binding mode was affected since only 1:1 binding was observed while in side-branch-labeled ligands, a 2:1 binding may also be involved. In contrast to those side-chain-labeled ligands, molecule 3 is extremely simple in structure and possesses superior detecting qualities such as selectivity, molecular sensitivity, and applicability in a wide range of pH. Recently, much effort has been placed on exploring fluorescent chemosensors for the detection of copper ions due to their biological and environmental significance.1-3 Copper sensing is usually based on a metal complexation site (ionophore) and a signal transduction unit (fluorophore), which are intramolecularly * Corresponding author. E-mail: [email protected]. † University of Miami, Coral Gables. ‡ University of Miami, Miami. (1) Peptide-based chemosensors: (a) Bhattacharya, S.; Thomas, M. Tetrahedron Lett. 2000, 41, 10313-10317. (b) Torrado, A.; Walkup, G. K.; Imperiali, B. J. Am. Chem. Soc. 1998, 120, 609-610. (c) Zheng, Y.; Huo, Q.; Kele, P.; Andreopoulos, F. M.; Pham, S. M.; Leblanc, R. M. Org. Lett. 2001, 3, 32773280.

1706 Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

connected through a linking bridge.4,5 Binding of the recognition unit with copper ions causes a fluorescent quenching to the fluorophore as a result of electron or energy transfer.6 Under this framework, researches were focused on investigating novel binding moieties with a superior selectivity toward copper ions. Once such a binding model is established, logically, the rest of the work is to attach a fluorophore to the vicinity of the binding center for signal transduction. This strategy obtains great popularity due to its considerable flexibility of design. On the other hand, inadequate attention has been paid to the possible influence of the added fluorophore unit on the copper-binding process. As such, there seemed to be a tendency that binding and signaling parts are treated independently, but their relationship is just being linked intramolecularly for signaling purposes. In the current (2) (a) Beltramello, M.; Gatos, M.; Mancin, F.; Tecilla, P.; Tonellato, U. Tetrahedron Lett. 2001, 42, 9143-9146. (b) Singh, A.; Yao, Q.; Tong, L.; Still, W. C.; Sames, D. Tetrahedron Lett. 2000, 42, 9601-9605. (c) DeSantis, G.; Fabbrizzi, L.; Licchelli, M.; Mangano, C.; Sacchi, D.; Sardone, N. Inorg. Chim. Acta 1997, 257, 69-76. (d) Corradini, R.; Dossena, A.; Galaverna, G.; Marchelli, R.; Panagia, A.; Sartor, G. J. Org. Chem. 1997, 62, 62836289. (e) Mayr, T.; Wencel, D.; Werner, T. Fresenius J. Anal. Chem. 2001, 371, 44-48. (f) Klein, G.; Kaufmann, D.; Schurch, S.; Reymond, J.-L. Chem. Commun. 2001, 561-562. (g) Kramer, R. Angew. Chem., Int. Ed. 1998, 37, 772-773. (h) Yoon, J.; Ohler, N. E.; Vance, D. H.; Aumiller, W. D.; Czarnik, A. W. Tetrahedron Lett. 1997, 38, 3845-3848. (i) Sasaki, D. Y.; Shnek, D. R.; Pack, D. W.; Arnold, F. H. Angew. Chem., Int. Ed. Engl. 1995, 34, 905-907. (3) Surface-oriented chemosensors: (a) Zheng, Y.; Gatta´s-Asfura, K. M.; Li, C.; Andreopoulos, F. M.; Pham, S. M.; Leblanc R. M. J. Phys. Chem. B 2003, 107, 483-488. (b) Zheng, Y.; Orbulescu, J.; Ji, X.; Andreopoulos, F. M.; Pham, S. M.; Leblanc R. M. J. Am. Chem. Soc., in press. (c) Mayr, T.; Werner, T. Analyst 2002, 127, 248-252. (4) (a) Bargossi, C.; Fiorini, M. C.; Montalti, M.; Prodi, L.; Zaccheroni, N. Coordin. Chem. Rev. 2000, 208, 17-32. (b) Pina, F.; Bernardo, M. A.; GarciaEspana, E. Eur. J. Inorg. Chem. 2000, 2143-2157. (c) Fabbrizzi, L.; Licchelli, M.; Pallavicini, P.; Parodi, L.; Taglietti, A. In Transition Metals in Supramolecular Chemistry, Perspective in Supramolecular Chemistry; Sauvage, J.-P., Ed.; John Wiley & Sons: New York, 1999; Vol. 5, Chapter 3, pp 93-134. (d) Chemosensors of Ion and Molecule Recognition; Desvergne, J. P., Czarnik, A. W., Eds.; NATO ASI Series; Kluwer Academic Publishers: Dordrecht, 1997. (e) Fabbrizzi, L.; Licchelli, M.; Pallavicini, P.; Perotti, A.; Taglietti, A.; Sacchi, D. Chem. Eur. J. 1996, 2, 75-82. (5) Intermolecular strategies were also proposed: (a) Grandini, P.; Mancin, F.; Tecilla, P.; Scrimin, P.; Tonellato, U. Angew. Chem., Int. Ed. 1999, 38, 30613064. (b) Berton, M.; Mancin, F.; Stocchero, G. P.; Tonellato, U. Langmuir 2001 17, 7521-7528. (c) Huo, Q.; Sui, G.; Zheng, Y.; Kele, P.; Leblanc, R. M.; Hasegawa, T.; Nishijo, J.; Umemura, J. Chem. Eur. J. 2001, 7, 47964804. (6) Few fluorescence enhancement examples were reported: (a) Hennrich, G.; Walther, W.; Resch-Genger, U.; Sonnenschein, H. Inorg. Chem. 2001, 40, 641-644. (b) Mitchell, K. A.; Brown, R. G.; Yuan, D.; Chang, S.-C.; Utecht, R. E.; Lewis, D. E. J. Photochem. Photobiol. A: Chem. 1998, 115, 157-161. 10.1021/ac026285a CCC: $25.00

© 2003 American Chemical Society Published on Web 02/27/2003

Chart I. Binding Model of Gly-His and Gly-Gly-His with Cu2+

work, we have demonstrated that different positions of the fluorophore unit (dansyl, Dns) in the sensor molecule may cause drastic disturbance to the interaction of the binding site with metal ions and further lead to very different (deleterious or favorable) changes in the performance of the probe molecule, such as selectivity and sensitivity. The artificial design of fluorescent chemosensors for copper ions, to some extent, can be accomplished by simulating the models in naturally occurring copper-binding proteins.7 Two such typical structural motifs showing high affinity for Cu2+ are GlyGly-His and Gly-His. Gly-Gly-His is a copper-binding motif originating from the amino terminal Cu- and Ni-binding (ATCUN) site.8 Gly-His is the functional component in the well-known growth factor Gly-His-Lys.9 Both of them have the common feature of monomeric multidentate binding, which is formed with the participation of the N-terminal amine, one or two amide nitrogen, and the unprotonated imidazole nitrogen from histidine (Chart 1). Herein, we synthesized and studied fluorescent chemosensors based on the Gly-His motif and, in particular, revealed the unusual effect of intramolecular spatial placement of the fluorophore on the subsequent recognition process. Three modular peptides, sequenced as Gly-His-Lys(Dns) (1), Dpr(Dns)-His-Lys (2), and Dns-Gly-His-Gly (3), respectively, were synthesized. In peptide 1, the fluorophore Dns, attached on the side chain of the lysine residue, is distant from the binding site. In peptide 2, Dns is also on a side branch but more closely coupled to the binding center via a CH2 linkage. In either case, the design attempts not to disturb the modular tridentate binding of the peptides toward Cu2+. It was previously observed that fluorescent sensitivity of the fluorophore toward the metal-binding event is related to the distance between the fluorophore and the recognition sitesthe shorter the linkage, the more sensitive the recognition interaction.1b,10 In this regard, peptide 2 should be more advantageous as a probe than 1 since the two units are more closely correlated with each other in the former ligand. Surprisingly, a quite different view was disclosed in this paper. In peptide 3, the dansyl group was attached on the N-terminal end of the peptide. Since the free N-terminal amine in the Gly-His motif is considered an important requirement for the binding of the peptide with copper ions (indeed, we found that acetyl-Gly-HisLys did not show any binding with Cu2+),8 the N-terminally (7) Sigel, H.; Martin, R. B. Chem. Rev. 1982, 82, 385-426. (8) Harford, C.; Sarkar, B. Acc. Chem. Res. 1997, 30, 123-130. (9) Pickart, L.; Lovejoy, S. In Methods in Enzymology; Barnes, D.; Sirbasku, D. A., Eds.; Academic Press: Orlando, FL, 1987; Vol. 147, pp 314-328. (10) Imperiali, B.; Pearce, D. A.; Sohna, J.-E. S.; Walkup, G.; Torrado, A. Proc. SPIE-Int. Soc. Opt. Eng. 1999, 3858, 135-143.

attached dansyl would most likely change or diminish the binding functionality of the Gly-His motif. EXPERIMENTAL SECTION The Wang resin and amino acids used for peptide synthesis were purchased from Advanced Chem Tech (Louisville, KY) or Novabiochem (San Diego, CA). All the amino acids were of L-configuration except glycine. Other organic chemicals and solvents were of reagent grade and purchased from Aldrich (St. Louis, MO). Inorganic salts used for fluorescence measurements were analytical grade purity. The water used was purified by a Modulab 2020 water purification system (Continental Water Systems Corp., San Antonio, TX). It has a resistance of 18 MΩ‚ cm and a surface tension of 72.6 mN/m at 20 °C. Peptide synthesis. Peptides 1-3 were synthesized via standard solid-phase 9-fluorenylmethoxycarbonyl (Fmoc) chemistry.11,12 Building blocks used for the synthesis include FmocGly-OH, Fmoc-His(Trt)-OH, Fmoc-Lys(Dde)-OH, Fmoc-Lys(Boc)OH, and Boc-Dpr(Fmoc)-OH. The diisopropylcarbodiimide (DIC) and 1-hydroxylbenzotriazole (HOBt) in situ activation method was used for the coupling reactions. The Fmoc groups were deprotected with 20% piperidine solution in DMF. The Dde group was removed with 2% hydrazine in DMF for 4 min.13 The addition of the dansyl group was performed with 2 equiv of dansyl chloride in the presence of diisopropylethylamine (DIEA) for 30 min. The (11) Fields, G. B.; Noble, R. L. Int. J. Pept. Protein Res. 1990, 35, 161-214. (12) Stewart, J. M. In Methods in Enzymology; Fields, G. B., Ed.; Academic Press: New York, 1997; Vol. 289, pp 44-67. (13) Bycroft, B. W.; Chan, W. C.; Chhabra, S. R.; Hone, N. D. J. Chem. Soc., Chem. Commun. 1993, 778-779.

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

1707

dansylated peptide may further react with excess dansyl chloride to form dually dansylated peptide and thus decrease the product yield. Cleavage of peptides from the resin was conducted with CF3COOH/H2O (95/5, v/v) for 2.5 h (trifluoracetic acid must be handled in a secure hood). Following suction under reduced pressure and removal of TFA with N2 blow-off, crude product was precipitated from cold ether. The solid precipitate was centrifuged and washed with ether and lyophilized under vacuum. Semipreparative RP-HPLC was performed on a Waters 2690 separations module. Two solutions were prepared: (A) 0.1% TFA in water; (B) 0.1% TFA in acetonitrile. Typically a linear gradient (0-30% B within 30 min, flow rate at 2 mL/min) was used. Column conditions: 7.8 × 300 mm, C18 column. Peptide Analysis. The purities of the synthesized peptides were verified by analytical HPLC, 1H NMR, and MS. Analytical HPLC was conducted on a small-scale column (4.6 × 75 mm C18 column). The same gradient as for the semipreparative method was used (flow rate at 1 mL/min). 1H NMR data were taken on a Bruker 300-MHz spectrometer. Low-resolution FAB was recorded on a VG-Trio 2000 mass spectrometer. High-resolution FAB was conducted on a 70-4F instrument in the Mass Spectrometry Laboratory, University of Illinois at UrbanasChampaign. 1: 1H NMR (300 MHz, D2O) δ 8.80 (s, 1H), 8.57 (d, 1H), 8.46 (d, 1H), 8.23 (d, 1H), 7.64 (m, 2H), 7.46 (m, 1H), 7.40 (s, 1H), 4.81 (m, 1H), 4.27 (m, 1H), 3.78 (s, 2H), 3.10-3.21 (m, 2H), 2.87 (s, 6H), 2.85 (m, 2H), 1.75 (m, 1H), 1.60 (m, 1H), 1.39 (m, 4H); FAB-MS 574.245 (MH+, calcd 574.245); UV-visible (in aqueous phosphate buffer at pH 6.8) 215 nm ) 2.9 × 104 L‚mol-1‚cm-1, 246 nm ) 9.8 × 103 L‚mol-1‚cm-1, 327 nm ) 2.8 × 103 L‚mol-1‚cm-1. 2: 1H NMR (300 MHz, D2O) δ 8.63 (d, 1H), 8.52 (s, 1H), 8.37 (d, 1H), 8.24 (d, 1H), 8.00 (d, 1H), 7.84 (m, 2H), 7.23 (s, 1H), 4.21 (m, 1H), 4.09 (m, 1H), 3.40 (s, 6H), 3.28 (m, 2H), 3.13 (m, 2H), 2.80 (t, 2H), 1.77 (m, 1H), 1.65 (m, 1H), 1.50 (m, 2H), 1.33 (m, 2H); FABMS 603.271 (MH+, calcd 603.271). UV-visible (in aqueous phosphate buffer at pH 6.8) 215 nm ) 2.9 × 104 L‚mol-1‚cm-1, 246 nm ) 1.0 × 104 L‚mol-1‚cm-1, 329 nm ) 2.9 × 103 L‚mol-1‚cm-1. 3: 1H NMR (300 MHz, D O) δ 8.62 (d, 1H), 8.50 (s, 1H), 8.38 (d, 2 1H), 8.23 (d, 1H), 7.92 (m, 1H), 7.82 (m, 2H), 7.20 (s, 1H), 4.48 (m, 1H), 3.78 (m, 2H), 3.61 (s, 2H), 3.33 (s, 6H), 3.12-3.01 (m, 3H). FAB-MS: 503.171 (MH+, calcd 503.171); UV-visible (in aqueous phosphate buffer at pH 6.8) 215 nm ) 2.8 × 104 L‚mol-1‚cm-1, 247 nm ) 1.0 × 104 L‚mol-1‚cm-1, 330 nm ) 3.0 × 103 L‚mol-1‚cm-1. Absorption and Fluorescence Measurements. UV-visible spectra were recorded on a Lambda 900 UV/Vis/NIR spectrophotometer (Perkin-Elmer, Inc., Shelton, CT). The fluorescence measurements were performed on a Spex Fluorolog 1680 0.22-m double spectrometer (Spex Industries, Inc., Edison, NJ). In the pH titration experiments, pH values of the solution were adjusted with NaOH and HCl solutions that contained the same concentration of the analytes. RESULTS AND DISCUSSION UV-Visible Spectra. Peptides 1-3 have very similar absorption spectra because all of them have dansyl as the only photosensitive group in the molecules. Three typical absorption peaks (I, ∼215 nm; II, ∼247 nm; III, ∼330 nm) were observed in the UV region (see Experimental Section). Upon addition of Cu2+, the spectral shape remained essentially the same, but each band 1708 Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

Table 1. Shifts of Absorption Peaks and the Wavelength of Visible Peak in UV-Visible Spectra upon the Addition of Cu2+ (pH 6.8) shift (nm)

1 2 3

in peak I

in peak II

in peak III

peak in the visible region upon addition of Cu2+ (nm)

0 0 4

0 0 7

0 3 8

598 598 623

was shifted to a shorter wavelength (Table 1). The value of the shifts is indicative of the degree of the interaction between the fluorophore and the bound Cu2+: the bigger the shift, the stronger the interaction. The least substantial effect of the Cu2+ on the absorption was observed in 1 since no shifts at all were seen. This can be rationalized by considering the fact that Dns is located on the far end of the side chain of lysine residue; thus, complexation has little effect on its absorption. In peptide 2, the sidebranch-labeled Dns is closer to the recognition site and so a slightly bigger shift (3 nm) was observed at peak III. Dns in 3 is attached on a metal-binding nitrogen atom, and the presence of Cu2+ caused very strong shifts to the absorption peaks. Except for UV region, an absorption band appeared in the visible region upon addition of Cu2+ (also shown in Table 1), the d-d transition band. But the position of this peak in 3 is markedly different from the side-branch-labeled ligands. This again indicates the special effect of Dns group on the metal-binding process in 3. Fluorescence Quenching by Cu2+. Fluorescence emission intensity of the three peptides decreased gradually with the addition of copper ions. Figure 1A shows the fluorescence spectra of 3 with varying copper ion concentration. Similar changes were also observed for 1 and 2. However, their binding isotherms showed very different characteristics (Figure 1B). It is seen that the fluorescence intensity of 1 and 2 decreased quickly and linearly with the addition of copper ions until 0.6 equiv of Cu2+ was added. After that point, only a marginal decrease was noticed. This phenomenon indicates that complexes other than of a 1:1 ratio existed in the system. An experiment of continuous variation,14 which is to determine the binding stoichiometry, also showed evidence of the formation of complexes of more than single ratios. In contrast, for 3, a turning point was observed when 1 equiv of Cu2+ was added, indicating a 1:1 binding ratio. These results are consistent with the previous report that N-terminal dansylation induces formation of species with only one ligand bond to copper ions.15 The formation constant of 3-Cu2+ was calculated as 1.87 × 106 M-1 using the method of Connors.16 This value is smaller than that of the nonlabeled Gly-His motif (e.g., 1016 M-1 for the tripeptide Gly-His-Lys17) but is large enough to ensure fast binding kinetics. Obviously the N-terminal dansylation thermodynamically decreased the binding affinity of the peptide with Cu2+. We consider this as good information in terms of sensor (14) Connors, K. A. Binding Constants, the Measurement of Molecular Complex Stability; John Wiley & Sons: New York, 1987; pp 24-28. (15) Antolini, L.; Menabue, L.; Sola, M.; Battaglia, L. P.; Bonamartini Corradi, A. J. Chem. Soc., Dalton Trans. 1986, 1367-1373. (16) Reference 14, pp 339-343. (17) Rainer, M. J. A.; Rode, B. M. Inorg. Chim. Acta 1984, 92, 1-7.

Figure 1. Relationship of fluorescence of the ligands to the concentration of Cu2+ (pH 7.0). Left: fluorescent changes of 3 (1.0 µM) with the addition of Cu2+ (from top to bottom, 0, 0.2, 0.4, 0.6, 0.8, 1.0, 1.5, 2.0, 3.0, and 4.0 µM, respectively). Right: fluorescent binding isotherms of 1-3 with Cu2+ (The concentration of each peptide is 1.0 µM. Fluorescence intensity has been normalized.). Excitation is selected at 340 nm.

design because the reduced binding affinity will help differentiate the binding of the ligand with its target ion from other metal ions. As shown later, the selectivity is indeed clearly seen. Usual 1 and Unusual 2. Both 1 and 2 are side-branchlabeled ligands. Which one is better to detect Cu2+? From common sense, 2 should be the better choice because the fluorophore is more closely correlated with the recognition site and, thus, detects the binding event more efficiently. We present here the titration curves of fluorescent intensities of the two peptides versus pH of the aqueous solution (Figure 2). Peptide 1 is strongly fluorescent in the range of pH 4-10 (Figure 2A, curve a). In stronger alkaline conditions (pH >10), the fluorescent intensity appeared to decrease because the emission maximum shifted to shorter wavelength. When pH was lower than 4, the fluorescence intensity of 1 decreased sharply and disappeared completely below pH 2. The phenomenon under this acidic pH is related to the protonation of the dimethylamino group (pKa ≈ 4), which prevents the charge transfer between the amine and naphthyl ring thus leading to the quenching of fluorescence.18,19 The fluorescence of peptide 1 in the presence of Cu2+ showed a different style of change with pH (Figure 2A, curve b). The pronounced difference was observed in the range of pH >5.5 in which the fluorescence intensity remained at a very low level (e.g., the ratio relative to free peptide 1 is 20% at pH 7). Clearly, the binding of Cu2+ caused this quenching. When pH was less than 5.5, the fluorescence intensity appeared to increase, and at pH 4, it reached the same level as that of the free peptide. This means that the 1-Cu2+ complex dissociates into free ligand and ion, and thus, Cu2+ causes no quenching to the fluorescence of 1. Again, when pH was below 4, the dimethylamino group in Dns was protonated and fluorescence gradually disappeared. This Cu2+induced quenching provides a starting point for the peptide to be used as a fluorescent chemosensor. (18) (a) Kimura, E.; Koike, T. Chem. Soc. Rev. 1998, 27, 179-184. (b) Aoki, S.; Kawatani, H.; Goto, T.; Kimura, E.; Shiro, M. J. Am. Chem. Soc. 2001, 123, 1123-1132. (c) Koike, T.; Watanabe, T.; Aoki, S.; Shiro, M. J. Am. Chem. Soc. 1996, 118, 12696-12703. (19) Lowe, M. P.; Parker, D. Inorg. Chim. Acta 2001, 317, 163-173.

To evaluate the sensing specificity of peptide 1 toward Cu2+ in the measured pH range, we tested the influence of several miscellaneous transition metal ions (Fe2+, Fe3+, Zn2+, Co2+, and Ni2+, each 10-5 M) on the fluorescence of peptide 1 in the same pH range (Figure 2A, curve c). Obviously, the quenching ability of these miscellaneous ions is much weaker than Cu2+. The specificity of 1 toward Cu2+ can be seen by comparing the three titration curves presented in Figure 2A. In the pH range of 6.411.0, Cu2+ caused a stronger quenching (∼2 times) than the other metal ions; therefore, though not ideal, the specificity is clearly seen. The best pH range in which peptide 1 can be used for the detection of Cu2+ is pH 5.0-6.4. In this range, Cu2+ binds with 1 and displays the largest quenching to its fluorescence, whereas the miscellaneous metal ions present a minimal effect. The fluorescence intensity of peptide 2 versus pH exhibited a marked distinction from 1 (Figure 2B, curve a). A big difference is that the fluorescence intensity steadily increased as pH went higher. To explain this phenomenon, we consider the fact that the amino groups from the peptidyl backbone gradually become deprotonated (neutral state) as pH increases, which brings more hydrophobic nature to the molecule. It is known that the dansyl group is highly sensitive to its microenvironment: its emission intensity becomes stronger and exhibits a hypsochromic shift in hydrophobic media.20,21 Since Dns in peptide 2 is closely associated with the backbone of the peptide (only via a CH2 linkage), this hydrophobic change in microenvironment is sensed by the fluorophore, leading to the enhancement of fluorescence emission. On the other hand, the dansyl group in peptide 1 is on the far end of the side chain of the lysine residue and therefore its fluorescence intensity is less apparently affected by the protonation states of the peptidyl backbone. Cu2+ quenched the fluorescence of 2 at pH >5 (Figure 2B, curve b) which is similar to that of the 1-Cu2+ system. However, the result of pH titration of the solution containing 2 and the (20) Hoenes, G.; Hauser, M.; Pfleiderer, G. Photochem. Photobiol. 1986, 43, 133137. (21) Li, Y.-H.; Chan, L.-M.; Tyer, L.; Moody, R. T.; Himel, C. M.; Hercules, D. M. J. Am. Chem. Soc. 1975, 97, 3118-3126.

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

1709

Figure 2. Effect of pH on the fluorescence of peptides 1 (A, C) and 2 (B, D). (A) and (B) show the intensity at 550 nm; (C) and (D) show the wavelength of maximum emission. (a) Free peptide, (b) peptide and Cu2+, and (c) peptide and miscellaneous metal ions. Concentration of each peptide is 1.0 µM. Concentration of Cu2+ is 1.0 µM. The miscellaneous ions contain Fe2+, Fe3+, Zn2+, Co2+, and Ni2+, and each has a concentration of 10 µM. Excitation is selected at 340 nm.

miscellaneous ions brings out an issue of selectivity (Figure 2B, curve c). We see here the titration curve is almost entirely overlapped with that of peptide 2 with Cu2+. This means that 2 is nonspecific for Cu2+ in the whole range of pH. The question is, why can peptide 1 differentiate Cu2+ from the miscellaneous ions, but 2 cannot? Observation of the effect of pH on the wavelength of emission maximum (λmax) gives insight to the particular molecular mechanism in the fluorescent quenching of 2 (Figure 2C and D). The emission of peptide 1 kept its λmax at 548 nm in the wide range of pH 9. However, compared with the free peptide, this shift is strikingly large (23 nm when pH is changed from 9 to 12). One explanation is that the nitrogen atom of the sulfonamide group on the end of the side chain of the lysine residue, after being deprotonated, is a strong electron donor and may come into 1710 Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

interaction with copper ions. Likewise, curve c in Figure 2C also showed a hypsochromic shift at pH >10 due to the same source of origin. The λmax of peptide 2 gradually decreased as pH increased (Figure 2D, curve a). As mentioned, an increasing hydrophobic character comes to the peptide in this pH-increasing process, leading to the hypsochromic shift and enhancement in fluorescence intensity. The big hypsochromic shift (12 nm) that occurred from pH 9 to 11 is related to the deprotonation of the sulfonamide group, as mentioned earlier. In contrast, λmax of the 2-Cu2+ solution presented a very unusual behavior (Figure 2D, curve b). Two unexpected phenomena were observed: (a) When pH was higher than 6, no further hypsochromic shift was observed. (b) From pH 4 to 6, λmax changed from 553 to 502 nm, a hypsochromic shift of 51 nm. This observation undoubtedly shows that the pKa value of the sulfonamide group becomes 5 (approximately), instead of being its typical value of 10. A direct conclusion is that Cu2+ drastically facilitated the deprotonation of the sulfonamide group, and vice versa, the deprotonated nitrogen atom of sulfonamide directly participated in the formation of the complex. We propose such a binding model as illustrated in Chart 2. Herein, the binding interaction with Cu2+ occurred with the participation of the imidazole nitrogen on histidine residue, the peptide amide nitrogen, the pseudo-glycine amine, and the dansyl sulfonamide nitrogen. The CPK model of 2 also supports this working scheme.

Figure 3. Effect of pH on the fluorescence intensity (A) and λmax (B) of peptide 3. (a) Free 3, (b) 3 and Cu2+, and (c) 3 and miscellaneous metal ions. Concentration of 3 is kept constant at 1.0 µM. Concentration of Cu2+ is 1.0 µM. The miscellaneous ions contain Fe2+, Fe3+, Zn2+, Co2+, and Ni2+, and each has a concentration of 1.0 µM. Excitation wavelength is 340 nm.

Figure 4. Fluorescent selectivity of ligands 1-3 toward copper ions (phosphate buffer, pH 7.0). (a) Free ligand (4.0 µM), (b) ligand and miscellaneous ions, and (c) ligand, miscellaneous ions, and copper (4.0 µM). The miscellaneous ions contain Fe2+, Fe3+, Zn2+, Co2+, and Ni2+, and each has a concentration of 4.0 µM. Excitation wavelength is 340 nm.

Chart 2. Proposed Structure of Peptide 2 Binding with Cu2+

On the other hand, the CPK model proves that the interaction between Dns and the metal-binding site in 1 could not be as strong as that in peptide 2, because in the former molecule, spatial restriction prevents a full approach of Dns to the metal-binding site. For the solution of 2 with miscellaneous metal ions, the change of λmax versus pH is totally the same as that of the 2-Cu2+ complex solution (Figure 2D, curve c). Obviously, the miscellaneous ions also strongly bind with 2 due to the assistance of the Dns group. This explains why 2 is not able to distinguish Cu2+ from the miscellaneous ions. Characteristics of 3 Binding with Cu2+. We measured the fluorescence intensity of 3 with pH of the aqueous solution in the presence and absence of metal ions (Figure 3A). The basic

features of fluorescent changes of the free ligand are similar to peptides 1 and 2; and Cu2+ also caused a fluorescent quenching at pH >5.5. Surprisingly, in comparison to the side-branch-labeled peptides, the fluorescence of 3 was affected very little by the miscellaneous transition metal ions. Curve c in Figure 3A showed that the miscellaneous ions (containing Fe2+, Fe3+, Zn2+, Co2+, and Ni2+; each ion is of 1 equiv relative to the ligand) only marginally affected the molecular fluorescence in a wide range of pH