Bloconjugate Chem. 1993, 4, 537-544
597
Site-Directed Double Fluorescent Tagging of Human Renin and Collagenase (MMP-1) Substrate Peptides Using the Periodate Oxidation of N-Terminal Serine. An Apparently General Strategy for Provision of Energy-Transfer Substrates for Proteases Kieran F. Geoghegan,' Michael J. Emery, William H. Martin, Alexander S. McColl, and Gaston 0. Daumy Central Research Division, Pfizer Inc., Eastern Point Road, Groton, Connecticut 06340. Received July 6, 1993'
Periodate in neutral aqueous solution rapidly converts N-terminal Ser or Thr to an a-N-glyoxylyl moiety that can serve as the locus for incorporation of a modifying group [Geoghegan, K. F., and Stroh, J. G. (1992) Bioconjugate Chem. 3, 138-146. Gaertner, H. F. et al. (1992) Bioconjugate Chem. 3, 262-2681. The usefulness of this procedure has been further illuminated in a route to "energy-transfer" substrates for endoproteases. Each such substrate is an oligopeptide cleavable by a proteinase, but modified (usually at its termini) with two chromophores that form an energy donor-acceptor pair. Production of these substrates is an exercise in double site-directed peptide modification. The new route is composed of three steps, beginning from an unprotected peptide in which a sequence recognized by the pertinent enzyme is placed between N-terminal Ser and C-terminal Lys. Lys may not occur elsewhere in the peptide. Periodate oxidation converts the N-terminal Ser to an a-N-glyoxylyl group, which is then allowed to form a hydrazone with the carbohydrazide derivative Lucifer Yellow CH, a hydrophilic fluor with a large Stokes shift (excitation maximum, 425 nm; emission maximum, 525 nm). Finally, the modified peptide is allowed to react with 5-carboxytetramethylrhodaminesuccinimidyl ester. This reaction selectively modifies the eamino group of C-terminal Lys, the only amino group remaining in the peptide. 5-Carboxytetramethylrhodaminestrongly (>go% ) quenches Lucifer Yellow fluorescence by resonance energy transfer in the intact substrate, but enzyme-catalyzedcleavage eliminates the quenching. The resulting increase in fluorescence may be used to follow the hydrolytic reaction. New substrates for human renin and fibroblast collagenase (matrix metalloproteinase-1) have been made to illustrate the procedure. Each was characterized by structural, spectroscopic, and kinetic methods and furnished a continuous fluorescence-based assay for its respective proteinase. It appears that the scheme can be applied to the preparation of comparable substrates for other proteinases.
INTRODUCTION Site-directed modification is the delivery of a modifying group uniquely to a preselected site in a peptide.' Like classical chemical modification, it is effected using a reagent specific for a single type of functional group that may exist in a peptide (I), but consists of applying such a reagent to a peptide containing only one group of the target type. The result is that there can be only one modification event. This is sometimes achieved by designing or redesigning the molecule to harbor only a single instance of one of the reactive groups naturally found in peptides (for example, a cysteine thiol) and making this group the locus of site-directed modification (2, 3). A complementary strategy is to direct the modification to a group that is not naturally present in peptides and whose reactivity distinguishes it from those which are. One approach exploits the very rapid and selective attack of periodate on N-terminal Ser or Thr ( 4 , 5 ) . This reaction converts the N-terminal residue to an a-N-glyoxylyl moiety, OHCCO-, providing an aldehyde function for Abstract published in Advance ACS Abstracts, October 1, 1993.
Abbreviations: "peptide" is often used here in a broad sense meaning 'peptide or protein"; PDMS, plasma desorption mass spectrometry; TFA, trifluoroacetic acid; LY- denotes Lucifer Yellow CH in hydrazone linkage with the a-N-glyoxylylmoiety at the N-terminus of a peptide following periodate oxidation; CTMR- denotes the 5-carboxytetramethylrhodaminemoiety coupled to the C-terminal lysyl residue of the peptide.
follow-up chemistry that introduces a modifying group (6-9). This may be non-peptide or peptide-like in character, and work to date has centered on the use of hydrazide modifiers that form usefully stable hydrazones. Reduction of the hydrazone is a further option in some cases (9). This method can be one element of a strategy for double site-directed modification. There are cases in which a peptide must be modified with two distinct groups at known sites, with the aim of supporting an application based on the interaction of the two modifying groups (10, 11). Alternatively, it may be necessary to introduce two unconnected non-native properties into a peptide by dual modification. We have applied this concept to the provision of "energy-transfer" fluorescent substrates for the continuous assay of proteinases (12-21). In this type of substrate, two chromophoric groups that form a Forster energy donor-acceptor pair are placed at opposite ends of a peptide. The fluorescence emission spectrum of the donor overlaps the absorption spectrum of the acceptor, causing the fluorescence of the donor to be quenched while the substrate remains intact. When the intervening peptide region is cleaved by a proteinase, relief of the strongly distance-dependent quenching provides the means to follow this activity. Here we describe the preparation of new energy-transfer substrates for human renin and fibroblast collagenase (matrix metalloproteinase-1). A peptide substrate of each enzyme was doubly modified with Lucifer Yellow CH and 5-~arboxytetramethylrhodamine, and the resulting fluorogenic substrates were used in continuous assays of their
lQ43-18Q2/93/29Q~Q537$Q4.QQIQ 0 1993 American Chemical Society
538
Bioconjugate Chem., Vol. 4, No. 6, 1993
respective proteinases. This report draws most attention to the direct and, in our view, relatively simple process by which unprotected peptides were converted to doublymodified derivatives ready for practical use. It appears that the process can be repeated in order to provide comparable substrates for other proteinases. EXPERIMENTAL PROCEDURES
Reagents. Lucifer Yellow CH and 5-carboxytetramethylrhodamine succinimidyl ester were purchased from Molecular Probes, P.O. Box 22010,4849 Pitchford Ave., Eugene, OR 97402. The optical properties of these reagents were taken to be those quoted in the supplier's catalog. The peptides SIHPFHLVIHTK (renin substrate) and SGPLGLRAK (collagenase substrate) were supplied as custom syntheses by Synthecell/Vega Biomolecules Corp., 7101RiverwoodDr., Columbia, MD 21046. Sodium m-periodate was purchased from Aldrich, 1001West Saint Paul Ave., Milwaukee,WI 53233. Periodatesolutions were prepared freshly each day. Recombinant human renin (EC 3.4.23.15) was the highly purified material described (22)or a partly purified preparation derived from the same starting material. Recombinant human fibroblast procollagenase (EC 3.4.24.7) was produced in a baculovirus expression system, purified by immunoaffinity chromatography (23), and activated by treatment with beadimmobilized trypsin (Pierce). Mass Spectrometry. PDMS was performed using a Bio-Ion 20 spectrometer (Applied Biosystems). Prefabricated nitrocellulose-coated targets were soaked with about 0.21 mL of 10% ethanol until the nitrocellulose was fully wet (15-30 s), spun briefly to leave only a thin layer of the wetting solution, and loaded with 15 pL of sample, which was usually dissolved in 0.1 % TFA. The instrument was normally run in positive ion mode with the acceleration potential at 15 kV, but was also used in negative ion mode at -12 kV. Masses were estimated either by using the built-in search program or by placing the twin cursors on each side of the topmost segment of the peak and calculating a centroid. This procedure produced mass results within 0.1% of true values for standards. Chromatography. HPLC was performed on a HewlettPackard 1090 Series I1 system equipped with a diodearray detector or on an LDC analytical gradient system using two spectroMonitor I11 variable single-wavelength detectors. Solvents in all cases were A, 0.1 % (v/v) TFA; and B, 0.085% (v/v) TFA in 80% (v/v) CH3CN. All preparative fractionations were performed using a Vydac C4 column (1.0 X 25 cm; type 214TP1010) operating at a flow rate of 4 mL/min with a linear gradient from 0 to 55% solvent B in 36 min beginning 4 min after injection. In experiments to characterize the products of enzymecatalyzed hydrolysis of the new substrates, the column was either a Vydac C18 column (0.2 X 25 cm; type 218TP52) or an Aquapore ODS cartridge (0.2 x 10 cm) (Applied Biosystems). Each of these columns was used with a flow rate of 0.2 mL/min and a linear gradient from 0 to 55% of solvent B in 35 min beginning 4 min after injection. Peaks were collected by hand. Fluorescence Spectroscopy. Fluorescence emission spectra were collected on a Perkin-Elmer LS 50 luminescence spectrometer. The excitation wavelength was 430 nm and both slits were set to a bandwidth of 4 nm. Conditions for kinetic studies are given below. Preparation of L Y-IHPFHLVIHTK-CTMR (Renin substrate). Step 1: Oxidation of SIHPFHLVIHTK. A 1.6-mgsample of SIHPFHLVIHTK (1.1pmol) dissolved in 0.04 M sodium phosphate buffer, pH 7.0, at a concen-
Geoghegan et al.
tration of 1.9 mg/mL was treated with 2.5 mM NaI04 for about 3 min at room temperature, after which the mixture was fractionated by HPLC. (The peptide had to be dissolved initially at 10 mg/mL in 0.01 % aqueous TFA before being diluted into the phosphate buffer.) The single product peak, expected to be a-N-glyoxylyl-IHPFHLVIHTK, was collected and dried in a centrifugal concentrator. Step 2: Coupling of Lucifer Yellow CH to a-Nglyoxylyl-IHPFHLVIHTK. A 200-pL portion of 25 mM Lucifer Yellow CH in sodium acetate buffer, pH 4.5,25 % CH&N was added to the dried a-N-glyoxylyl-IHPFHLVIHTK (1.5 mg) and the reaction was allowed to proceed at room temperature overnight, after which the presence of some precipitated material (presumed to be a reaction product) was noted. This was solubilized by adding 0.1 % TFA in 25% CH3CN to the incubate, which then was fractionated using the same HPLC procedure as before. The reaction product eluted at 29.2 min, expected to be L Y-IHPFHLVIHTK, was collected and dried. Step 3: Coupling of 5-Carboxytetramethylrhodamine to L Y-IHPFHL VIHTK. The dried L Y-IHPFHLVIHTK was redissolved in 300 pL of 133 mM NaHC03, pH 8.8, and treated with 75 pL of 20 mM 5-carboxytetramethylrhodamine succinimidyl ester in CH3CN (final concentrations: 5-carboxytetramethylrhodaminesuccinimidyl ester, 4 mM; LY-IHPFHLVIHTK, 1 mM). After 1 h incubation at room temperature, the reaction mixture was injected onto the HPLC using the gradient program described above. The product peak, expected to be LYIHPFHLVIHTK-CTMR, was collected at 34.5 min. As this procedure gave only about two-thirds conversion of LY-IHPFHLVIHTK to LY-IHPFHLVIHTK-CTMR, it was later repeated using a final concentration of 9 mM 5-carboxytetramethylrhodaminesuccinimidyl ester. This second experiment gave full conversion of L Y-IHPFHLVIHTK to the doubly-tagged product, and is the one shown in Figure 3 (see Results). Specificity of Renin-Catalyzed Hydrolysis of LYIHPFHLVIHTK-CTMR. A 0.02-pg sample of renin was added to 40 pL of LY-IHPFHLVIHTK-CTMR solution (140 pM in 0.02 M Mes, pH 6.5, containing 5% DMSO). The reaction was incubated at 37 "C for 2 h, after which 15 pL was injected onto the HPLC (second of the two systems described above). Separate product peaks containing the Lucifer Yellow and 5-carboxytetramethylrhodamine chromophores, respectively, were eluted earlier than intact substrate, and fractions containing these products were dried and characterized by PDMS. Continuous Fluorescence-Based Assay of Renin Using L Y-IHPFHLVIHTK-CTMR. Renin activity was measured at 22 OC by monitoring the change in fluorescence of a sample containing various amounts of L Y-IHPFHLVIHTK-CTMR and human renin dissolved in 0.05 M sodium phosphate buffer, pH 6.8, containing 0.1 M NaCl and 4% DMSO. The fluorescent substrate was excited at 430 nm and emission was read at 520 nm in a Perkin-Elmer LS-5B luminescence spectrometer. The excitation and emission slits were each set to a bandwidth of 10 nm. As the concentration of L Y-IHPFHLVIHTK-CTMR was increased, internal filtering of Lucifer Yellow fluorescence was also enhanced due to increasing absorbance by 5-carboxytetramethylrhodamine.A correction for this was applied by first collecting enough data to estimate the rate of fluorescence change, and then adding 2 pM CP108,671 (a strong renin inhibitor) in order to suppress further renin-induced fluorescence changes. The extent
Bloconjugate Chem., Voi. 4, No. 6, 1993 539
Note overlap of LY smidon spectrum wlth CTMR a b m p t h spectrum
LY
CTMR
of the internal filtering effect was then gauged by adding Lucifer Yellow CH, so that its concentration was increased by96 nM, and comparing the resulting fluorescencechange to that caused by the same concentration of Lucifer Yellow in the absence of 5-carboxytetramethylrhodamine. Observed rates were then normalized on the basis of this estimate of quenching. Preparation of L Y-GPLGLRAK-CTMR (Collagenase Substrate). Preparation of the collagenase substrate closely followed the method described for the renin substrate, with the following differences. The starting material SGPLGLRAK was directly soluble in the phosphate reaction buffer for the first step, and did not require solubilization in 0.01% TFA; there was no need to add acetonitrile to the reaction mixture for the coupling of Lucifer Yellow CH to a-N-glyoxylyl-GPLGLRAK; there was no precipitate formed in this reaction; and the final product, L Y-GPLGLRAK-CTMR was readily soluble at near millimolar concentration in its assay buffer of 0.05 M Tris, 0.2 M NaC1,0.005 M CaC12,pH 7.7. Intermediates, the product itself, and the products of its collagenasecatalyzed cleavage were characterized by the methods used for the renin substrate, including PDMS. Specificity of Collagenase-CatalyzedHydrolysis of L Y-GPLGLRAK-CTMR. To determine the site of cleavage catalyzed by recombinant collagenase, 200 p L of LY-GPLGLRAK-CTMR solution (20 pM) in 0.05 M Tris, 0.15 M NaC1,0.005 M CaC12, pH 7.5, was treated with 10 pL of activated collagenase solution (nominally about 5 pg of enzyme, although this is probably an overestimate). After 4 h at 37 OC, the substrate initially present in the reaction mixture was shown by HPLC to be >905% cleaved. The two product peaks were collected, dried, and subjected to mass analysis. Continuous Fluorescence-Based Assay of CollagenaseUsing L Y-GPLGLRAK-CTMR. Fluorescencebased rate measurements of collagenase activity were taken at room temperature using a buffer of 0.05 M Tris, 0.15 M NaC1,0.005 M CaC12, pH 7.7, in the Perkin-Elmer LS 50 luminescence spectrometer. The excitation wavelength
was set to 430 nm (slit at 10-nm bandwidth), and fluorescence emission was recorded at 530 nm (5-nm slit).
RESULTS Strategy for Double Site-Directed Tagging of a Renin Substrate Peptide. Human renin selectively liberates angiotensin I from the N-terminal region of angiotensinogen by hydrolyzing the peptide bond LeuloVal". It has no other known enzymatic activity, exerting an extreme specificity due to active center subsites that recognize the substrate a t least across the P4-P1' residues (and probably more extensively, as in mouse renin) (24). Synthetic substrates must match this extended structure, leaving little scope for the inclusion of a non-native chromophore and suggesting that continuous assays for renin are most easily designed using the energy-transfer principle. Two recent reports described such assays (20, 21).
Site-directed labeling requires a unique group that becomes the locus of modification. Double site-directed labeling requires the presence of two such groups with distinct reactivities. Design of the renin substrate consisted of taking a renin-sensitive core sequence, IHPFHLVIH (residues 5-13 of human angiotensinogen), and flanking it with additional N-terminal and C-terminal residues selected to match the intended chemistry. The core was flanked N-terminally by Ser to provide the locus for the first round of modification, consisting of periodate oxidation and subsequent introduction of Lucifer Yellow CH at the resulting a-N-glyoxylyl group. It was flanked C-terminally by Thr-Lye to provide an extra residue of spacing and an amino group for targeted modification by reaction with 5-carboxytetramethylrhodaminesuccinimidyl ester. The complete reaction scheme is summarized in Scheme I. Lucifer Yellow has a broad absorption band centered near 425 nm (E425 of 12 000 M-' cm-l) and a fluorescence emission centered near 525 nm. Complementing this as energy acceptor was the strong chromophore 5-carboxy-
540 Bkconlugete CY”.., Vol. 4, No. 6, lQ93
*
Geodmgan et ai.
*
a
235 nm
- 430 nm
w
TIME (min)
Figure 1. HPLC analysis of the periodate oxidation of SIHPFHLVIHTK. The arrow marks the retention time of SIHPFHLVIHTK prior to modification, and the oxidation product peak marked with an asterisk was collected for mass analysis (Table I). Table I. Predicted and Observed Massea for Intermediates, Substrate, and Products of Ranin-Catalyzed Hydrolysis of the Renin Substrate MH+ theoretical observed SIHPFHLVIHTK 1429.7 1429.5 a-N-crlvonrlvl-IHPFHLVIHTK 1398.7 (unhydrated) 1398.9 -- - _ 1416.7 (hydiated) 1417.1 LY-IHPFHLVIHTK 1826.1 1825.2 LY-IHPFHLVIHTK-CTMR 2238.1 2238.9 LY-IHPFHL 1247.5 1247.3 VIHTK-CTMR 1009.7 1010.2 0 Mass analyses were performed by PDMS (see Materiala and Methods).
tetramethylrhodamine (CWof 63 OOO M-l cm-9, whose absorption spectrum overlapped sufficientlywell with the fluorescence of Lucifer Yellow (see below) to suggest that it could adequately quench Lucifer Yellow fluorescence prior to hydrolysis of the twice-fluorotagged peptide by renin. Periodate Oxidation of SIHPFHLVIHTK. HPLC and PDMS (Table I ) were used to verify the identity and purity of the starting material SIHPFHLVIHTK. Periodate oxidation of this peptide (see Experimental Procedures) gave quantitative conversion to a single product which was recovered by HPLC (Figure 1)and shown by PDMS to be the expected a-N-glyoxylyl-IHPFHLVIHTK (TableI). Characteristically(8),the a-N-glyoxylyl-peptide gave two peaks in PDMS, one corresponding to ita unhydrated form and one corresponding to ita hydrated form (Table I). Conjugation of Lucifer Yellow CH to a-N-glyoxylylIHPFHLVIHTK. The a-N-glyoxylyl-IHPFHLVIHTK was dried and was then allowed to react with Lucifer Yellow CH in sodium acetate buffer, pH 4.5, containing 25 % CH3CN to aid in solubilizing the somewhat hydrophobic peptide. Material presumed to be a reaction product came out of solution during the overnight coupling and was resolubilizedby acidifyingthe reaction mixture with 0.1 % ! TFA in 25% CH3CN. The reaction mixture was then fractionated using the same HPLC method as before, and the result (Figure 2a) again showed complete conversion of the starting material to a single product. The absorption spectrum of the product showed that the new material containedLuciferYellow (Figure 2b). PDMS showed that the product had a mass consistent with the hydrazone of Lucifer Yellow CH with a-N-glyoxylyl-IHPFHLVIHTK (Table I), denoted LY-IHPFHLVIHTK. UV spectralanalysis(not shown) of a comparable adduct (LY-IGSLAK) demonstrated the presence of an absor-
0.0 20
b
i
25
30 35 TIME (min)
40
45
i\
TLL 51.0
I a
0.0220
300 WAVELENGTH 400 (nm) 500
600
Figure 2. Conjugation of Lucifer Yellow CH to a-N-glyoxylylIHPFHLVIHTK. (a) HPLC analysis of the coupling reaction, showing absorption traces at 235 and (-) 430 nm.The arrow marks the retention time of a-N-glyoxylyl-IHPFHLWHTKprior to reaction, and the product peak marked with an asterisk was collected for mass analysis (Table I). (b) Absorption spectrum of the product, showing the presence of the Lucifer Yellow chromophore. (e
a)
bance band centered at 262 nm (CZZ of 13200 M-1 cm-1) that does not exist in the spectrum of a-N-glyoxylylIGSLAK or free Lucifer Yellow CH, and therefore must be a property of the adduct. The position of thia band supported the conclusion that Lucifer Yellow CH added to the peptide as a hydrazone (see Figure 8 of ref a), and not as a Schiff base. This hydrazone can presumably be made in two isomeric forms (synand anti configurations about the hydrazone bond), but the products have not so far been resolved into multiple forms by any of our preparative methods or analyzed (e.g. by NMR) to characterizetheir putative distribution between these two forms. Conjugation of 5-Carboxytetramethylrhodamineto L Y-IHPFHLVIHTK. L Y-IHPFHLVIHTK was dried and then redissolved in 0.13 M NaHCOs for the third and final step of the preparation. For a first attempt, 5-carboxytstramethylrhodaminesuccinimidyl ester dissolved in CHsCN (20 mM) was added to the peptide (final concentration 1mM) at a final concentration of 4 mM. The deaired product L Y-IHPFHLVIHTK-CTMR formed in about 66% yield in 1h and was recovered by HPLC. The incomplete conversion obtained in this trial was attributed to use of insufficient 5-carboxytstramethylrhodamine succinimidyl ester. A separate experiment using 9 mM 5-carboxytstramethylrhodaminesuccinimidyl ester resulted in complete conversion of LY-IHPFHLVIHTK to the desired product (Figure 3a). The product peak was distinguished from several peaks due to residual 5-carboxytetramethylrhodamineby its absorption spectrum (Figure3b),showing local maxima due to Lucifer Yellow (€425 12 OOO M-lcm-l) and 5-carboxytetramethylrhodamine (em 63 ,OOO M-lcm-1) in about a 1 5 ratio of intensity at the wavelengths of these absorptivities, consistent with their presence in 1:l ratio, and then by
Doubk SitaDkected Modlflcatkm
Blooonlugete Chem.. Vol. 4,
No. 6, 1093 641
Table 11. Pmdicted and Observed Masses for Intermediates, Substrate, and Products of Collagenase-Catalyzed Hydrolysis of the collagenase Substrate MH+
theoretical
ObWrVed
SGPLGLRAK a-N-glyoxylyl-GPLGLRAK
0.0 27
30
33
36
TIME (min)
b o.8
A
1
899.1 899.2 868.1 (unhyrated) 868.1 886.2 886.1 (hydrated) L Y-GPLGLRAK 1295.6 1294.9 LY-GPLGLRAK-CTMR 1707.5 1707.0 LY-GPLGb (M - H)(M - H)theor 825.0 obs 824.9 LRAK-CTMR 900.6 899.9 a Methods are the same as those noted in Table I. b LY-GPLG could not be succeaafully analyzed by positive ion PDMS, but waa detected without difficulty when the spectrometer waa operated in negative ion mode.
WAVELENGTH (nm)
Figure 3. Conjugation of 5-carboxytetramethylrhodamineto
LY-MPFHLVIHTK. (a)HPLC analysisof the coupling reaction, showing absorption traces at (-) 430 and 540 nm. The arrow marka the retention time of LY-IHPFHLVIHTK prior to reaction, and the product peak marked with an asterisk was collected for mass analysis (Table I). (b) Absorption spectrum of the product, showing the presence of the Lucifer Yellow and 5-carboxytstramethylrhodaminechromophores. (a
a)
nm (Rhodamine) - 540 430 nm (LuciferYellow)
0.51
I
W
1
0.0,
33
I
A I
4
U
I
36
I
I
1
39
TIME (mln)
Figure 4. HPLC analysis of the products formed by renincatalyzed hydrolysis of LY-IHPFHLVIHTK-CTMR. The substrate (peakat 39min) was converted to two earlier-elutedproduct peaks. Absorption traces are shown at (-) 430 (LuciferYellow) and (. .) 540 nm (5-carboxytetramethylrhodamine).See Table I for maee analysis results. PDMS (Table I). The product was dried and stored at -20 "C. It was soluble in DMSO to at least 1.6 mM, and DMSO therefore was selected as the solvent used to deliver it into kinetic experiments. Action of Renin on LY-IHPFHLVIHTK-CTMR HPLC fractionation of an incubate of LY-IHPFHLVMTK-CTMR with purified renin yielded two product peaks containing, respectively, Lucifer Yellow (detected (detected at 430 nm) and 5-carboxytetramethylrhodamine at 540 nm) (Figure 4). PDMS results for these products agreed with predicted MH+ values for LY-IHPFHL and VIHTK-CTMR (Table I). These results showed that human renin cleaved the doubly fluorotagged substrate at the Leu-Val bond that it recognizes in angiotensinogen and set the stage for fluorescence-based studies of renin activity (see below).
500 550 606 650 WAVELENGTH (nm)
Figure 5. Fluorescence emission spectra showing the increase in detectable Lucifer Yellow fluorescence consequent to renincatalyzed hydrolysis of L Y-IHPFHLVIHTK-CTMR.Excitation wavelength was 430 nm. The emission spectra are shown for 4.6 rM solutions of (- -)LY-IHPFHLVIHTK-CTMR (intact substrate), (-) the products of renin-catalyzedhydrolysis of LYIHPFHLVIHTK-CTMR, and (- .) LY-IHPFHLVIHTK (the peptide tagged with only Lucifer Yellow CH, and therefore not fluorescence-quenched). Samples were dissolved in 0.04 M Mes, pH 6.5,with 10% DMSO.
Double Site-Directed Tagging of a Collagenase Substrate Peptide. Vertebrate collagenases are specialized zinc metalloenzymes that cleave strands of the collagen triple helix at Gly-Leu bonds. As this activity is significant in inflammatory disease (25),the enzymes are further important targets of drug-discovery efforts. A fluorogenic substrate was planned around a collagenasesensitive sequence (26)which was flanked by N-terminal Ser and C-terminal Lys, to give SGPLGLRAK as the starting material. This was doubly tagged in the same way as the renin substrate, and the intermediates and final product were characterized likewise (Table 11). Treating the new substrate, L Y-GPLGLRAK-CTMR, with recombinant human collagenase yielded products whose absorption and mass spectra (Table 11)led them to be identified as LY-GPLG and LRAK-CTMR, indicating that cleavage occurred at the Gly-Leu bond. Fluorescence Changes Consequent to Substrate Hydrolysis. The fluorescence emission spectrum of 4.5 pMLY-IHPFHLVIHTK-CTMR was compared with that of the product mixture from its renin-catalyzed hydrolysis (Figure 5). The same experiment was performed for LYGPLGLRAK-CTMR and the products of its collagenasecatalyzed hydrolysis (data not shown, as the result closely resembles that in Figure 5). For both substrates, Lucifer Yellow fluorescence increased by over 10-fold following cleavage, indicating their good potential for successful use in fluorescence-based assays (see below). Excitation at 430 nm, while directed toward the Lucifer
542
Geoghegan et al.
Bioconjugate Chem., Vol. 4, No. 6, 1993
0.50r
0.0
RENIN (pglmL)
0
5
10 15 20 25 COLLAGENASE (pglmL)
b 2.0 e 1.5
h
f
v
1.0 0
I! 9 0.5 0.0
J
1
10
20
30
40
50
60
2
70
using LY-IHPFHLVIHTK-CTMR.(a)Dependenceof initial rate on enzyme concentration; data from experiments using 1 pM substrate. The line is a linear regression plot through the data points; its nonzero y-intercept was considered to reflect experimental error, as no rate was detected in the absence of enzyme. (b)The initial rate of hydrolysis of LY-IHPFHLVIHTK-CTMR was measured using 0.24 pg/mL renin and various substrate concentrations.Each point represents the mean of three observations. Data were fit to the Michaelis-Menten equation (curve), giving K M of 20 pM. Yellow chromophore, also caused some direct excitation of 5-carboxytetramethylrhodamineand generated a minor fluorescence signal centered at 580 nm (Figure 5). This did not interfere with the major signal change from Lucifer Yellow that followed hydrolysis of the peptide midsection. Renin Activity Assay. The initial rate of cleavage of LY-IHPFHLVIHTK-CTMR showed linear dependence on the concentration of added renin (Figure 6a). Under the conditions described in the Experimental Procedures, the reaction followed saturation kinetics with K M equal to 20 pM and V,, equal to 2.37 nM s-l (Figure 6b). This value for K M agreed with results reported elsewhere for the same peptide substrate sequence in its unmodified form (27). Collagenase Activity Assay. In kinetic studies of LYGPLGLRAK-CTMR with recombinant collagenase, the reaction rate exhibited linear dependence on the amount of enzyme added (Figure 7a). Inconsistent results were observed if the reaction buffer was not supplemented with 1p M zinc(II), indicating the need to maintain full loading of the collagenase with the active site metal. Variation of the substrate concentration from 0.1 to 4 pM indicated that substrate binding began to be saturated in this range, with a calculated K M of 3.9 p M (Figure 7b). This result was in the same range as the K M reported for another fluorescent substrate of collagenase, DnpPLGLWAdR-NHz, for which a KMof 7.1 p M was determined at 37 O C and pH 7.7 (17). This substrate has the same sequence as L Y-GPLGLRAK-CTMR in positions P3-Pi. However, the specificity of collagenase is highly
6
8
1
0
SUBSTRATE (pM)
SUBSTRATE (pM)
Figure 6. Data from fluorescence-based assays of renin activity
4
Figure 7. Data from fluorescence-based assays of collagenase activity using L Y-GPLGLRAK-CTMR.Due to the complexity of the collagenase activationprocess, the concentrationof active enzyme in these experiments was not accurately known. (a) Dependence of initial rate on enzyme concentration;data from experiments using 3.5 pM substrate.The line is a linear regression plot through the data points; its nonzero y-intercept was considered to reflect experimentalerror, as no rate was detected in the absence of enzyme. (b) Dependence of rate on substrate concentration.Each point represents a single observation. Data were fit to the Michaelis-Menten equation (curve),giving K M of 3.9 uM.
dependent on the substrate structure, and this was noted during selection of the core sequence of LY-GPLGLRAKCTMR. The choice of Arg at Pz’was inspired by indications that it would strongly enhance substrate binding relative to many other amino acids that might be placed at that position (26),which seemed to be the case. However, the control experiment to show that Arg actually had this effect was not done. DISCUSSION Selective proteolytic action continues to be recognized as a central mechanism of biological control. Elucidation of its details has created a growing need for direct and continuous assays of highly specific proteinases, which is often best met with intramolecularly quenched “energytransfer” substrates. The usual strategy, and the one followed here, is to place the fluorescence donor and acceptor groups on opposing sides of a peptide substrate sequence. A variation is to employ a chromophoric amino acid as both a specificity determinant and a component of the energy relay (17,28). As this application calls for adding different nonpeptide groups to opposite ends of a peptide, it is especially opportune to use chemistry with intrinsic specificity for the N-terminus. Periodate oxidation of N-terminal Ser or Thr is an exceptionally fast reaction that proceeds best at neutral pH (5,8). It is also highly selective when a low molar excess of periodate over peptide is used, as the
Double SltaDirected Modification
desired oxidation is faster than a number of the potentially competing reactions (8). The N-terminal oxidations of both SIHPFHLVIHTK and SGPLGLRAK (Scheme I) were accomplished without complications because neither peptide contained any residue with periodate-sensitivity comparable to that of N-terminal Ser (29). HPLC analysis indicated quantitative formation of the desired products, as illustrated for the renin substrate SIHPFHLVIHTK (Figure 1). Lucifer Yellow CH readily formed a hydrazone with a-N-glyoxylyLIHPFHLVIHTKin the second step of this preparation (Figure 21, consistent with earlier experience of comparable reactions (8). The dye itself was freely soluble in water (0.25 M can be achieved using the dilithium salt), but the relative hydrophobicity of the renin substrate peptide necessitated the addition of acetonitrile to the coupling buffer. No organic cosolvent was required in the coupling reaction with the more hydrophilic collagenase substrate. The third and final step was modification of the t-amino group of C-terminal Lys with 5-carboxytetramethylrhodamine succinimidyl ester. When a mixture of 5- and 6-carboxytetramethylrhodaminewas employed in trial experiments using the peptide SIGSLAK (not shown), products derived from the two dye isomers were separated via reversed-phase HPLC to give very complex chromatograms. Although the use of purified 5-carboxytetramethylrhodamine succinimidyl ester mostly eliminated this problem, different forms of the dye (presumably active ester and free acid) were retained by the HPLC column and gave multiple 540-nm peaks that could have complicated product identification (Figure 3a). However, judicious selection of detector wavelengths or, preferably, use of a diode-array detector allowed the desired product peak to be recognized by the presence of spectral features due to both Lucifer Yellow and 5-carboxytetramethylrhodamine (Figure 3b). HPLC indicated >90% conversion of starting material to desired product in each of the three steps (Figures 1-31, Product identifications were based on the combination of mass spectrometry (Tables I and 11) and absorption spectroscopy (Figures 2b and 3b). The same methods were used to show that renin cleaved LY-IHPFHLVIHTKCTMR at the Leu-Val peptide bond (Figure 4, Table I), and that collagenase cleaved L Y-GPLGLRAK-CTMR at the Gly-Leu bond (Table 11). In each case, this was the expected cleavage site. The present route to fluorogenic substrates (and other doubly-tagged peptides) may be broadly useful. One restriction is that the starting peptide must possess only two amino groups, one in the N-terminal Ser or Thr and the other in C-terminal L Y S ,in ~ order to preserve site specificity in the final step of modification. This is a problem only when Lys must occur in the sequence recognized by the proteinase. In such cases, enzymecatalyzed fragment coupling (7) might represent an interesting route to the provision of a substrate with the required sequence. In this preparation, the fluorescence donor was the first group of the energy-transfer pair to be coupled. As indicated by Wang et al. (301,this can be a significant point to consider in developing synthetic routes to 2 The Lys residue need not be at the C-terminusof the peptide, but must be to the C-terminal side of the scissile peptide bond. It might be useful to increase the water solubility of substrates containing hydrophobic peptide sequences by placing an Arg residue (for example) C-terminal to the Lys intended for modification.
Bioconjugate Chem., Voi. 4, No. 6, 1993 543
fluorescent substrates. Assuming that the final product is more likely to be contaminated with a residue of the second reagent to be coupled than with the first, it is preferable that this second reagent be the fluorescence acceptor. This is because residual acceptor harms the assay only if it is present at sufficiently high levels to introduce internal optical filtering due to its absorbance, while residual fluorescence donor creates a background of unquenched fluorescence in the assay. This is especially true if a highly fluorescent donor with a superior quantum yield is used. Lucifer Yellow does not meet this description, being a relatively weak chromophore ( e m s of 12 000 M-l cm-l) and fluorophore compared to excellent fluors such as fluorescein or 5-carboxytetramethylrhodamine (each with tmax> 60 000 M-' cm-l). However, the present method could be adapted to allow use of other fluorophores and chromophores. The pair of Lucifer Yellow with 5-carboxytetramethylrhodamine was selected for the present work because both reagents were commercially available and had complementary optical and chemical properties. An added important benefit of this pair is the relatively hydrophilic character of its components, which may be less restrictive of substrate solubility in aqueous buffers than some alternatives. I t was also important to find a donor-acceptor pair that operates in the visible region. High-wavelength donor-acceptor pairs are preferred in the screening of pharmaceutical sample banks for new inhibitors, because compound files may be rich in UVabsorbing compounds that interfere with assays based on energy-transfer in the ultraviolet. A general problem in assays of this type is the effect known as internal or inner filtering, the blocking of fluorescent emission due to absorbance by the quenching chromophore. The extent of this effect is a function of the molar absorptivity of this chromophore at the relevant wavelength, its concentration, and the cuvette light path for emission. It is not connected to properties of the enzyme. With the present pair of chromophores under the present conditions of assay, internal filtering is significant even at a substrate concentration of 2 pM (where absorbance at 520 nm due to 5-carboxytetramethylrhodamine is0.06), and so a correction for this was routinely applied to all rate measurements (see Experimental Procedures). It may appear from Scheme I that the order of steps 2 and 3 could be reversed if desired. On trying this with the collagenase substrate, and treating a-N-glyoxylyl-GPLGLRAK with 5-carboxytetramethylrhodamine succinimidyl ester at pH 8.8, we detected oligomerized conjugates of 5-carboxytetramethylrhodamine with the oxidized peptide (data not shown). Condensation of the a-N-glyoxylyl moiety with the Arg residue, neither of which is involved in the reaction with 5-carboxytetramethylrhodamine succinimidyl ester, was suspected to be the basis of this result, but this was not clearly established. This limited venture suggested that the N-terminal chemistry, which creates but then suppresses the aldehyde function, should be completed before modification of C-terminal Lys is attempted at pH >8. Hydrazones are potentially subject to hydrolysis when they are removed from the equilibrium conditions in which they are formed. This potential lability has not yet been quantified for the various types of hydrazone that can be formed after modifying N-terminal Ser, but should not be ignored when using this type of bioconjugate (8). In particular, such hydrazones are expected to exhibit an increased equilibrium degree of dissociation with decreas-
544
Bloconjugate Chem., Vol. 4, No. 6, 1993
ing initial concentration. This matter requires more direct experimental evaluation in the future. In response to this concern, Gaertner et al. (9)showed that hydrazones formed by carboxylic hydrazides can be reduced with cyanoborohydride. We confirmed this result, also noting that borohydride gives faster reductions (K. Geoghegan and M. Emery, unpublished). The present substrates were used successfully to perform enzyme assays at and below the micromolar level, which is their intended function, but it remains important when using these and comparable conjugatesto bear their potential pH-dependent (8)lability in mind and check it when possible. While the provision of fluorogenic substrates is an interesting application of peptide modification chemistry, it has been used here principally as a test case. The question has been whether chemistry originating with the selective periodate oxidation of N-terminal Ser provides stable adducts that perform well in a significant biochemical application. While there continues to be incomplete knowledgeregarding the stability of the hydrazones formed under the present reaction scheme, it has now been made clear that they are decidedly useful compounds. The impressively clean and simple nature of the process by which these compounds are made further recommends it for consideration in cases requiring N-terminal derivatization of peptides or proteins. LITERATURE CITED (1) Means, G. E., andFeeney,R.E. (1971)ChemicalModification
of Proteins, Holden-Day, San Francisco. (2) Wingfield, P., Graber, P., Shaw, A. R., Gronenborn, A. M., Clore, G. M., and MacDonald, H. R. (1989) Preparation, characterization and application of interleukin-la mutant proteins with surface-accessible cysteine residues. Eur. J. Biochem. 179, 565-571. (3) Chollet, A,, Bonnefoy, J.-Y., and Odermatt, N. (1990) Preparation, application and biological characterization of interleukin-la mutant protein biotinylated a t a single site. J . Immunol. Meth. 127, 179-185. (4) Nicolet, B. H., and Shinn, L. A. (1939)The Action of Periodic Acid on a-Amino Alcohols. J. Am. Chem. SOC.61, 1615. ( 5 ) Dixon, H. B. F., and Fields, R. (1972) Specific Modification of NHz-Terminal Residues by Transamination. Methods Enzymol. 25,409-419. (6) Maassen, A., Thielen, T. P. G. M., and Moller, W. (1983) Synthesis and Application of Two Reagents for the Introduction of Sulfhydryl Groups into Proteins. Eur. J . Biochem. 134, 327-330. (7) Offord, R. E. (1990) Chemical Approaches to Protein Engineering. Protein Design and the Development of New Therapeutics and Vaccines. (J.B. Hook and G. Poste, Eds.) pp 253-282, Plenum Publishing Corp., New York. (8) Geoghegan, K. F., and Stroh, J. G. (1992) Site-Directed Conjugation of Nonpeptide Groups to Peptides and Proteins via Periodate Oxidation of a 2-Amino alcohol. Application to Modification at N-Terminal Serine. Bioconjugate Chem. 3, 138-146. (9) Gaertner, H. F., Rose, K., Cotton, R., Timms, D., Camble, R., and Offord, R. E. (1992)Construction of Protein Analogues by Site-Specific Condensation of Unprotected Fragments. Bioconjugate Chem. 3,262-268. (10) Stryer, L., and Haugland, R. P. (1967) Energy Transfer: A Spectroscopic Ruler. Proc. Natl. Acad. Sci. U.S.A.58, 719726. (11) Stryer, L. (1978) Fluorescence Energy Transfer as a Spectroscopic Ruler. Annu. Rev. Biochem. 47, 819-846 (12) Latt, S. A., Auld, D. S., and Vallee, B. L. (1972)Fluorescence Determination of Carboxypeptidase A Activity Based on Electronic Energy Transfer. Anal. Biochem. 50, 56-62. (13) Carmel, A., Zur, M., Yaron, A., and Katchalski, E. (1973) Use of Substrates with Fluorescent Donor and Acceptor
Geoghegan et al. Chromophores for the Kinetic Assay of Hydrolases. FEBS Lett. 30, 11-14. (14) Yaron, A., Carmel, A., and Katchalski-Katzir, E. (1979) Intramolecularly Quenched Substrates for Hydrolytic Enzymes. Anal. Biochem. 95, 228-235. (15) Nishino, N., and Powers, J. C. (1980)Pseudomonas aeruginosa Elastase: Development of a New Substrate, Inhibitors, and an Affinity Ligand. J. Biol. Chem. 255, 3482-3486. (16) Deyrup, C., and Dunn, B. M. (1983) A New Substrate for Porcine Pepsin Possessing Cryptic Fluorescence Properties. Anal. Biochem. 129,502-512. (17) Stack, M. S., and Gray, R. D. (1989) Comparison of Vertebrate Collagenase and Gelatinase Using a New Fluorogenic Substrate Peptide. J . Biol. Chem. 264, 4277-4281. (18) Geoghegan, K. F., Spencer, R. W., Danley, D. E., Contillo, L. G., Jr., and Andrews, G. C. (1990) Fluorescence-based continuous assay for the aspartyl protease of human immunodeficiency virus-1. FEBS Lett. 262, 119-122. (19) Matayoshi, E. D., Wang, G. T., Krafft, G. A., and Erickson, J. (1990)Novel Fluorogenic Substrates for Assaying Retroviral Proteases by Resonance Energy Transfer. Science 247,954957. (20) Oliveira, M. C. F., Hirata, I. Y., Chagas, J. R., Boschcov, P., Gomes, R. A. S., Figueiredo, A. F. S., and Juliano, L. (1992) Intramolecularly Quenched Fluorogenic Peptide Substrates for Human Renin. Anal. Biochem. 203, 39-46. (21) Wang, G. T., Chung, C. C., Holzman, T. F., and Krafft, G. A. (1993)A Continuous Fluorescence Assay of Renin Activity. Anal. Biochem. 210, 351-359. (22) Geoghegan, K. F., Lanzetti, A. J., Ammirati, M. J., Danley, D. E., O’Connor, B. A., and Hobart, P. M. (1991) Simple Procedure for Recovery of Crystallizable Human Renin from Mammalian Cell-Conditioned Medium. Structure and Function of the Aspartyl Proteases (B. M. Dunn, Ed.) pp. 379-381, Plenum Press, New York. (23) Birkedal-Hansen, B., Moore, W. I. G., Taylor, R. E.,Bhown, A. S., and Birkedal-Hansen, H. (1988) Monoclonal Antibodies to Human Fibroblast Procollagenase. Inhibition of Enzymatic Activity, Affinity Purification of the Enzyme, and Evidence for Clustering of Epitopes in the NHz-Terminal End of the Activated Enzyme. Biochemistry 27, 675143758, (24) Dhanaraj, V., Dealwis, C., Frazao, C.,Badasso, M., Sibanda, B. L., Tickle, I. J., Cooper, J. B., Driessen, H. P. C., Newman, M., Aguilar, C., Wood, S. P., Blundell, T. L., Hobart, P. M., Geoghegan, K. F., Ammirati, M. J.,Lanzetti, A. J., Danley, D. E., O’Connor,B. A,, and Hoover, D. J. (1992) X-Ray Analyses of Peptide Inhibitor Complexes Define the Structural Basis of Specificity for Human and Mouse Renins. Nature 357,466472. (25) Nagase, H. and Woessner, J. F., Jr. (1993) Role of Endogenous Proteinases in the Degradation of Cartilage Matrix. Joint Cartilage Degradation. Basic and Chemical Aspects (J. F. Woessner and D. Howell, Eds.) pp 159-185, Marcel Dekker, New York. (26) Netzel-Arnett, S., Fields, G., Birkedal-Hansen, H., and Van Wart, H. E. (1991)Sequence Specificities of Human Fibroblast and Neutrophil Collagenases. J . Biol. Chem. 266, 67476755. (27) Cumin, F., Le-Nguyen,D., Castro, B., Menard, J.,andCorvol, P. (1987)Comparative enzymatic studies of human renin acting on pure natural or synthetic substrates. Biochim. Biophys. Acta 913, 10-19. (28) Pohl, J., Davinic, S., Blaha, I., Strop, P., and Kostka, V. (1987) Chromophoric and Fluorophoric Peptide Substrates Cleaved through the Dipeptidyl Carboxypeptidase Activity of Cathepsin B. Anal. Biochem. 165, 96-101. (29) Clamp, J. R., and Hough, L. (1965)The Periodate Oxidation of Amino Acids with Reference to Studies on Glycoproteins. Biochem. J . 94, 17-24. (30) Wang, G. T., Matayoshi, E., Huffaker, H. J., and Krafft, G. A. (1990) Design and Synthesis of New Fluorogenic HIV Protease Substrates Based on Resonance Energy Transfer. Tetrahedron Lett. 31, 6493-6496.
