Bioconjugate Chem. 2000, 11, 161−166
161
Preparation of Thiol-Reactive Cy5 Derivatives from Commercial Cy5 Succinimidyl Ester† Hermann J. Gruber,* Gerald Kada, Bernt Pragl, Christian Riener, Christoph D. Hahn, Gregory S. Harms, Werner Ahrer,‡ Thomas G. Dax,‡ Karin Hohenthanner,‡ and Hans-Gu¨nther Knaus§ Institute of Biophysics and Institute of Chemistry, J. Kepler University, Altenberger Strasse 69, A-4040 LINZ, Austria, and Institute of Biochemical Pharmacology, University of Innsbruck, Peter Mayr Strasse 1, A-6020 Innsbruck, Austria. Received August 13, 1999; Revised Manuscript Received December 2, 1999
The present study offers reliable protocols for the preparation of new thiol-reactive Cy5 derivatives which are urgently needed for single molecule fluorescence microscopy. In a systematic approach, two alternate strategies were found for the extension of commercial amine-reactive Cy5 with thiolreactive end groups. In the two-step method, Cy5 succinimidyl ester was first reacted with ethylenediamine under conditions which gave ∼99% asymmetric “Cy5-amine” and only ∼1% symmetric product with two Cy5 residues. Subsequently, “Cy5-amine” was derivatized with commercial heterobifunctional cross-linkers to introduce thiol-reactive end groups (maleimide or pyridyldithio). Alternatively, commercial Cy5 succinimidyl ester was reacted with a primary amine (MTSEA, methanethiosulfonylethylamine, or PDEA, pyridyldithioethylamine) or a secondary amine (PEM, piperazinylethylmaleimide) to give the corresponding thiol-reactive derivatives in a single step. Results were good for MTSEA, moderate for PEM, and poor for PDEA. An additional drawback of the onestep method was the need for rigorous removal of unreacted Cy5 succinimidyl ester, which would label lysine residues on probe molecules. It is concluded that, except for the Cy5-MTSEA conjugate, the two-step method is much more general, reliable, and easier to follow by the typical biophysicist, biologist, etc., for whose benefit, these procedures are being published. All thiol-reactive Cy5 derivatives showed similar absorption and fluorescence properties as Cy5 succinimidyl ester, and fluorescence was fully retained after binding to thiols on proteins. The kinetics of protein labeling was also examined in order to get an idea of proper labeling conditions.
INTRODUCTION
(SMFM)1
Single molecule fluorescence microscopy depends on fluorochrome labels with a high extinction coefficient, high fluorescence quantum yield, and high resistance to photobleaching. TMR, Cy3, Cy5, and Cy7 have been found to fulfill these requirements (Funatsu et al., 1995; Schmidt et al., 1995; Schu¨tz et al., 1997; La † This article is dedicated to Prof. Heinz Falk on occasion of his 60th birthday. * To whom correspondence should be addressed. Phone: +43 (732) 2468-9271. Fax: +43 (732) 2468-822. E-mail: hermann.
[email protected]. ‡ J. Kepler University. § University of Innsbruck. 1 Abbreviations: BSA, bovine serum albumine; BSA-SH, BSA with g2 mercaptopropionyl residues on lysine residues; Cy5monofunctional dye (or Cy5-mono-O-Su), sodium 1-[5-(N-succinimidyloxycarbonyl)-pent-1-yl]-2-[5-(3,3-dimethyl-1-ethyl-5sulfonato-indolin-2-ylidene)-1,3-pentadien-1-yl]-3,3-dimethyl3H-indolium-5-sulfonate; Cy5 derivatives were abbreviated as shown in Schemes 1-3; DIEA, N,N-diisopropyl-N-ethylamine; DMF, dimethylformamide; DTDP, 4, 4′-dithiodipyridine; 2-ME, 2-mercaptoethanol; MS, mass spectroscopy; MTS, methanethiosulfonyl group; MTSEA, methanethiosulfonylethylamine; NHS, N-hydroxysuccinimide; PDEA, 2-(2-pyridyldithio)ethylamine; PEM, N-[2-(1-piperazinyl)-ethyl]maleimide; PDP, 3-(2-pyridyldithio)-propionyl group; SMCC, O-(N-succinimidyl)-4-(maleimidomethyl)-cyclohexane-1-carboxylate; SPDP, O-(N-succinimidyl)3-(2-pyridyldithio)-propionate; SbTMU, O-(N-succinimidyl)N,N,N′,N′-bis-(tetramethylene)-uronium hexafluorophosphate.
Jolly Blue has not been tested yet). So far SMFM has been used in cell-free systems (Schmidt et al., 1996; Schu¨tz et al., 1998; Trabesinger et al., 1998; Vale et al., 1996). However, the present challenge of SMFM is the imaging of single molecules in live cells, the main obstacle being high autofluorescence background. Fortunately, autofluorescence is very low at excitation wavelengths of >600 nm (Aubin, 1979; Benson et al., 1979; Marelius, 1995; Wolfbeis and Leiner, 1985). Thus, Cy5 (λex ) 630650 nm) is the most promising label for SMFM on cells by one-photon excitation and, as a matter of fact, a Cy5labeled phospholipid was the first single molecule visualized in live cells (Schu¨tz et al., 1999). Further SMFM studies on cells will require other Cy5-labeled probe molecules, but this is limited by the fact that commercial Cy5 dye can only be used for statistical labeling of proteins on lysine residues. Although statistically labeled antibodies (and other probes) may be good for highsensitivity detection, the single molecule information is lost because the number of Cy5 per individual antibody molecule greatly varies and multiple Cy5 labels on one antibody exhibit strong quenching (e.g., by 75% at an average number of 4 Cy5 labels/IgG, unpublished observation). In addition, many probe molecules lose function when labeled with amine-reactive reagents (see Discussion); in these cases, expression and SH-specific labeling of cysteine mutants is imperative. As the popularity of SMFM increases (see special issue in Science, March 12, 1999), it is obvious that thiol-
10.1021/bc990107f CCC: $19.00 © 2000 American Chemical Society Published on Web 02/16/2000
162 Bioconjugate Chem., Vol. 11, No. 2, 2000
Gruber et al.
Table 1. Rf Valuesa and Spectral Propertiesb of Cy5 Derivatives Cy5 derivative
RfI
RfII
RfIII
Cy5-COOH Cy5-CO-NHS Cy5-MTSEA Cy5-amine Cy5-PDP Cy5-mal Cy5-mal-BSAd
0.09 0.30 0.25 0.11 0.35 0.31
0.17 0.37 0.28 0.10 0.25 0.24
0.27 0.40 0.31 0.08 0.36 0.25
absorption fluorescence maximum (nm) yieldc 647 647 648 648 648 648 653
Scheme 1. Conversion of the Cy5-COOH Impurity in Cy5-Monofunctional Dye into the Corresponding Succinimidyl Ester
nd t1.0 1.4 1.1 1.7 0.7 1.7
a R values were determined after spotting from the reaction f mixture, i.e., from DMF solution. To some degree these Rf values depended on the time between spotting and TLC plate development. b In comparison to Cy5-monofunctional dye. Determined in buffer A, except for Cy5-MTSEA which was measured in buffer C. c Cy5 was excited at 630 nm (5 nm slit) and emission was measured at 670 nm (5 nm slit). d Cy5-mal-BSA refers to a ratio of 0.097 ( 0.03 Cy5 labels per BSA molecule.
reactive Cy5 derivatives will be in great demand soon. Since the presentation of the Cy5-phospholipid study (Schu¨tz et al., 1999), we have received many requests of how to synthesize the thiol-reactive Cy5 labels because they cannot be purchased. It is the goal of this report to present our methods of preparation of thiol-reactive Cy5. EXPERIMENTAL PROCEDURES2
Materials. Analytical-grade materials were used as long as they were commercially available. Cy5-monofunctional dye and PD-10 columns were obtained from Amersham Pharmacia Biotech. MTSEA‚HBr was purchased from Toronto Research Chemicals. DTDP, DTT, 2-ME, SMCC, and SPDP were obtained from Sigma. SbTMU was supplied by Fluka. Purified BSA (fatty acidfree) was obtained from Behring Werke AG, Germany. PDEA was purchased from BIAcore. PEM was supplied by Dojindo Laboratories. Buffers. Buffer A contained 100 mM NaCl, 50 mM NaH2PO4, and 1 mM EDTA (pH 7.5 adjusted with NaOH). Buffer B contained 150 mM NaCl, 5 mM citric acid, and 1 mM EDTA (pH 5.5 adjusted with NaOH). Buffer C contained 100 mM NaCl and 20 mM sodium acetate (pH 4.5 adjusted with HCl). Methods. General Comments. All operations were performed under dim light, and flasks containing Cy5 dye were wrapped with aluminum foil. Oxidation of primary amines (ethylenediamine, Cy5-amine, MTSEA) was suppressed by performing experiments in an argon atmosphere, especially when deprotonating with DIEA. Unless specified otherwise, all operations were at room temperature (20-25 °C). The molar amounts of all Cy5 derivatives were determined by diluting defined aliquots from the reaction mixtures or from the dissolved products to about 1-4 µM in buffer A and recording the UV-vis absorption spectrum. Molar concentrations were calculated from absorbance at λmax (Table 1), assuming the same molar extinction coefficient ( ) 250 000 M-1 cm-1) as published for Cy5-monofunctional dye (Mujumdar et al., 1993). No significant difference in absorption was seen whether buffer or methanol was used as solvent. 2 The Experimental Procedures in this article describe the synthesis of Cy5-mal by the two-step procedure (Scheme 2) and the synthesis of Cy5-MTSEA by the one-step procedure (Scheme 3). More detailed descriptions of all syntheses and of covalent labeling of BSA-SH with thiol-reactive Cy5 derivatives can be found in the Supporting Information.
Scheme 2. Two-Step Procedure for the Synthesis of Thiol-Reactive Cy5 Derivatives
TLC. Merck plastic sheets (silica gel 60) without fluorescent indicator were used. Eluents I, II, and III contained 70 parts of chloroform, 30 parts of methanol, and 4 parts of concentrated ammonia or water or acetic acid, respectively. Cy5 derivatives were visible without staining, and other components were stained with iodine vapor. Complete Conversion of Cy5-Monofunctional Dye into Succinimidyl Ester (Scheme 1). Commercial Cy5-monofunctional dye (4.3 µmol, a mixture of Cy5-CONHS with 10-50% Cy5-COOH) was dissolved in DMF (2 mL), and SbTMU (7.2 mg, 18 µmol) was added at once with stirring (Bannwarth and Knorr, 1991). DIEA (50 µL of a 5% solution in DMF, 15 µmol) was added, and the mixture was stirred for 25 min. TLC indicated quantitative conversion of Cy5-COOH into Cy5-CO-NHS (see Table 1). Synthesis of Cy5-Amine (Scheme 2). Ethylenediamine dihydrochloride (62 mg, 470 µmol) was dissolved in water (0.8 mL), and 4 mL of DMF was added (pH ∼66.5). Cy5-CO-NHS (4.3 µmol, see above) was subsequently added dropwise (pH ∼7-7.5). The coupling reaction was started by dropwise addition of DIEA (400 µL of a 5% solution in DMF, 240 µmol). After 30 min of stirring, the reaction mixture was taken to dryness. Cy5-amine was purified by low-pressure reversedphase chromatography on Merck LiChroprep RP-18 gel (bed size 1.5 × 3.5 cm, flow 2 mL/min). Solvent A
Thiol-Reactive Cy5 Derivatives from Cy5 Succinimidyl Ester
consisted of water containing 0.1% acetic acid, and solvent B was a 5/2 mixture (v/v) of ethanol and 1-propanol (Feldhoff, 1991). The crude Cy5-amine was loaded in 2.5% solvent B. After washing with 12 mL quantities each of 5, 10, and 15 solvent B, the product was eluted with 20% solvent B. Removal of solvent gave 4.2 µmol of Cy5-amine, which was pure by TLC (see Table 1). MS (electrospray, negative mode, 80% methanol in water) m/z (%) ) 697 (21, M), 375 (16), 348 (100, M/2), 255 (19). MS (electrospray, positive mode, 80% methanol in water) m/z (%) ) 699 (27, M), 487 (13), 369 (100), 329 (20), 257 (16). Synthesis of Cy5-mal (Scheme 2). Cy5-amine (4.2 µmol) was dissolved in DMF (1 mL) and added to dry SMCC (27 mg, 80 µmol) with stirring. Another 0.5 mL portion of DMF was used to transfer all Cy5-amine into the reaction mixture. After addition of DIEA (5 µL neat phase, 57 µmol) the mixture was stirred for 50 min, and the solvent was removed. The dry residue was dissolved in 0.4 mL of methanol and 1 mL of chloroform and diluted with 8 mL of chloroform. This solution was extracted with 2 × 20 mL of water containing 0.1% acetic acid and 6 × 20 mL of water. Residual dissolved chloroform in the combined aqueous layer was expelled with nitrogen bubbling, and after addition of solvent B (final contents 2.5%), the solution was loaded on the same LiChroprep column as used for Cy5-amine (see above). After washing with 5, 10, 15, 20, and 25% solvent B, the product was eluted at 30% solvent B. Evaporation yielded 3.3 µmol of Cy5-mal, which was pure by TLC. MS (electrospray, negative mode, 80% methanol in water) m/z (%) ) 917 (11, M), 488 (21), 458 (100, M/2), 255 (18), 227 (11). Cy5-mal was further purified by HPLC on a Vydac C-18 column (no. 218TP510, 250 × 10 mm, 2 mL/min, 0.1% acetic acid instead of trifluoroacetic acid in all eluents). When applying a 30 min gradient from 5 to 50%, acetonitrile pure Cy5-mal was eluted at 20-20.5 min after the start of the gradient. The solvent was removed, and the structure of Cy5-mal was confirmed by 1H NMR (see below). In parallel, a small aliquot was purified on an analytical Vydac C-18 column (no,. 218TP54, 250 × 4 mm, 1 mL/min), using the same elution protocol as above. Pure Cy5-mal was eluted 16.5 min after the start of the gradient, as detected by UV-vis absorption and by direct analysis of the eluent with electrospray MS. Only two m/e peaks were detected which exactly met expectation for the monoanion (917) and dianion (458) of Cy5-mal. 1H NMR (CD OD, 500 MHz): δ 8.34 (2H, m, β,β′ 3 protons of the methin bridge) 7.91-7.93 (4H, m, positions 4 and 6 of the indolin residues), 7.37 (2H, t, Jortho ) Jpara ) 7.5 Hz, position 7 of the indolin residue), 6.81 (2H, s, CH)CH of maleimid), 6.70 (1H, m, γ proton of the methin bridge), 6.37 (2H, t, J ) 13.2 Hz, R and R′ protons of the methin bridge), 4.19 (2H, q, J ) 6.8 Hz, CH2 of the N-ethyl group), 4.15 (2H, t, J ) 7.1 Hz, N-CH2-CH2-CH2CH2-CH2-CO), 3.27-3.33 (d, J ≈ 8 Hz, CH2 between cyclohexane and maleimide, seen in 1H-1H-COSY, hidden under CHD2-OD in the 1-d-spectrum), 3.24 (4H, s, CONH-CH2-CH2-NH-CO), 2.21 (2H, t, 7.2 Hz, N-CH2-CH2CH2-CH2-CH2-CO), 2.12 (1H, t, J ≈ 11 Hz from axialaxial coupling with two axial neighbors, the coupling to the two equatorial neighbors is not well resolved, position 1 of cyclohexane in 100% axial conformation), 2.05 (∼10 H, s, acetic acid from reversed-phase chromatography), 1.80-1.88 (4H, m, N-CH2-CH2-CH2-CH2-CH2-CO and equatorial protons in position 2 and 6 of cyclohexane), 1.78 (12 H, s, four methyl groups of Cy5 residue), 1.601.75 (5H, m, N-CH2-CH2-CH2-CH2-CH2-CO, one axial proton in position 4 of cyclohexane, two equatorial protons in position 3 and 5 of cyclohexane), 1.35-1.51
Bioconjugate Chem., Vol. 11, No. 2, 2000 163 Scheme 3. Single-Step Conversions of Cy5-Monofunctional Dye into Thiol-Reactive Derivatives
(7H, m, N-CH2-CH2-CH2-CH2-CH2-CO, CH3 of the Nethyl group, two axial protons in position 2 and 6 of cyclohexane), 1.01 (2H, q, Jgem ≈ Jaxial-axial ≈ 12 Hz, one geminal and two axial neighbors, the coupling to the one equatorial neighbor is not well resolved, axial protons in position 3 and 5 of cyclohexane). All NMR signal assignments were confirmed by COSY spectra of Cy5-mal, of commercial Cy5-NHS (in D2O, slightly different ppm values but the same patterns were observed) and of SMCC (in CDCl3). Synthesis of Cy5-MTSEA (Schemes 1 and 3). A modified activation protocol was used to maintain a high Cy5 dye concentration despite working on a small scale. Cy5-monofunctional dye (1.9 µmol) was dissolved in DMF (1 mL) containing SbTMU (8.4 mg/mL). Neat DIEA (5 µL, 29 µmol) was added with a Hamilton syringe. After 20 min of stirring in the dark, the TLC showed quantitative conversion of the Cy5-COOH fraction into Cy5-CONHS (see Table 1). Then, MTSEA‚HBr (25 mg, 150 µmol) was added at once, more neat DIEA (20 µL, 110 µmol) was injected, and the mixture was stirred for exactly 5 min. TLC indicated ∼100% conversion of Cy-CO-NHS into Cy5-MTSEA. The reaction mixture was frozen in liquid nitrogen, and the solvent removed at 1-10 Pa. The dry residue was dissolved in 15% solvent B and quickly loaded on the LiChroprep column (see above). After washing with 20 and 25% solvent B, the product was eluted with 30% solvent B. Removal of solvent gave 1.3 µmol of Cy5-MTSEA, which was pure by TLC (see Table 1). MS (electrospray, negative mode, 80% methanol in water): m/z (%) ) 792 (100, M), 396 (91, M/2), 388 (82), 265 (20). Cy5-MTSEA was further purified by HPLC either on a semipreparative or an analytical Vydac C-18 column, as described for Cy5-mal (see above). Only two peaks were detected in electrospray MS after HPLC, corresponding to the monoanion (792) and the dianion (396) of Cy5-MTSEA. 1H NMR (CD OD, 500 MHz): δ 8.32 (2H, m, β,β′ 3 protons of the methin bridge), 7.86-7.93 (4H, m, positions 4 and 6 of the indolin residues), 7.34 (2H, t, Jortho ≈ Jpara ≈ 7 Hz, position 7 of the indolin residues), 6.69 (1H, t, J ) 12.4 Hz, γ proton of the methin bridge), 6.34 (2H, t, J ) 14.6 Hz, R and R′ protons of the methin bridge), 4.18 (2H, q, J ) 7.2 Hz, CH2 of the N-ethyl group), 4.13 (2H, t, J ) 7.3 Hz, N-CH2-CH2-CH2-CH2-CH2-CO), 3.51 (2H, t, J ) 6.5 Hz, CO-NH-CH2-CH2-SO2), 3.43 (3H, s, SO2CH3), 3.32 (t, J ≈ 7 Hz, CO-NH-CH2-CH2-SO2, seen in 1 H-1H-COSY, hidden under CHD2-OD in the 1-d-spectrum), 2.22 (2H, t, 7.2 Hz, N-CH2-CH2-CH2-CH2-CH2-CO), 2.02 (∼1H, s, acetic acid from reversed-phase chromatography), 1.99 (∼1H, s, acetonitrile from reversed-phase chromatography), 1.84 (2H, quintet, J ) 7.4 Hz, N-CH2-
164 Bioconjugate Chem., Vol. 11, No. 2, 2000
CH2-CH2-CH2-CH2-CO), 1.76 (12 H, s, four methyl groups of Cy5 residue), 1.71 (2H, quintet, J ) 7.5 Hz, N-CH2CH2-CH2-CH2-CH2-CO), 1.46 (2H, quintet, J ) 7.7 Hz, N-CH2-CH2-CH2-CH2-CH2-CO), 1.40 (3H, t, J ) 7.2 Hz, CH3 of the N-ethyl group). RESULTS
Complete Activation of Cy5-Monofunctional Dye. The fraction of Cy5 succinimidyl ester in commercial Cy5monofunctional dye has improved from ∼50% in the past (Gruber et al., 1997) to >90% during the past year. Nevertheless, complete conversion of the residual Cy5COOH contents into Cy5-CO-NHS (Scheme 1) was advantageous for the purpose of this study because unreacted Cy5-COOH would have coeluted with Cy5-amine (Scheme 2) or with Cy5-MTSEA (Scheme 3) from the preparative RP-18 column and many small-scale HPLC runs would have been necessary to remove Cy5-COOH from the products. In situ activation of Cy5-COOH (Scheme 1) with SbTMU (Bannwarth and Knorr, 1991) was ideal because (i) 100% activation was easily achieved, (ii) the reaction conditions (DMF/DIEA) were the same as in all subsequent steps, and (iii) excess of SbTMU did not interfere with all further reactions, thus activation could be done in situ. Synthesis of Cy5-mal and Cy5-PDP by the TwoStep Protocol. Cy5 is supplied as amine-reactive NHSester only, whereas numerous heterobifunctional crosslinkers with a thiol-reactive function on one end and with a NHS-ester on the other are commercially available. Therefore, it was an obvious idea to use a simple diamine (like NH2-CH2-CH2-NH2) for the linking of Cy5 with a heterobifunctional cross-linker, as shown in Scheme 2. The first requirement was the synthesis of Cy5-amine, in pure form and high yield. Addition of Cy5-CO-NHS to a DMF solution of ethylenediamine, however, resulted in a 70:30 mixture of Cy5-amine and Cy5 dimer, even if a 103-fold excess of ethylenediamine in DMF was vigorously stirred and dilute Cy5-CO-NHS was added very slowly (results not shown). The identity of the Cy5 dimer was not rigorously proven, but from very tight binding to the reversed-phase column (elution at 50% “organic solvent”), as well as to an anion-exchange column [as described in Gruber et al. (1997); not shown], it was concluded that it must be the ethylenediamine derivative with two Cy5 residues. Formation of this undesired byproduct was effectively suppressed by complete mixing of Cy5-CO-NHS with the dihydrochloride of ethylenediamine, followed by gradual deprotonation with DIEA. With this improvement >99% of Cy5-CO-NHS were converted into asymmetric Cy5-amine and
