p-Benzoyl-L-phenylalanine, A New Photoreactive Amino Acid PHOTOLABELING OF CALMODULIN WITH A SYNTHETIC CALMODULIN-BINDING PEPTIDE*

,

A new photoreactive amino acid analog, p-benzoyl-L-phenylalanine, is described. Convenient methods for the preparation of this amino acid and its subsequent incorporation into synthetic peptides by the solidphase technique are outlined. To illustrate its utility, p-benzoyl-L-phenylalanine was substituted in place of tryptophan in a 17-residue calmodulin-binding peptide. The substitution did not measurably affect the affinity of this peptide for calmodulin. When this peptide was photolyzed at 350 nm in a 1:l molar ratio with calmodulin in the presence of 500 p M CaCL, 70% of the calmodulin was derivatized. The specificity of the reaction was investigated by photolysis in the absence of CaClz where little binding occurs; under these conditions little or no photolabeling occurred.
Photoaffinity labeling (1, 2) is a method which has been widely successful for the identification and localization of macromolecular receptors (3-5). Despite the general usefulness and importance of the technique for the labeling of peptide hormone receptors, there are no direct methods for the solid-phase synthesis of peptides containing photoactivatable probes. Photolabile peptides have generally been prepared by derivatization of the parent peptides with "heterobifunctional" cross-linking agents which combine in one molecule both chemically and photochemically reactive groups (3). The major disadvantage of this approach is that the photolabile group can only be introduced at chemically reactive sites present in the peptide, if indeed such sites exist. Furthermore, the covalent modification may limit the peptide's abilities to bind and activate its receptor. An alternate approach involves the de nouo synthesis of peptides containing derivatives of p-azidophenylalanine (6-9) which can form nitrenes on photolysis. Such peptides are generally synthesized by solution rather than solid-phase methods because of the limited chemical stability of aryl azides. Another method involves the solid-phase synthesis of derivatives of p-aminophenylalanine-containing peptides which are subsequently converted to azides by multistep procedures (9). Finally, 3-[ p-(trifluoromethyl)-3H-diazirin-3-yl]-phenylalanine, a car. bene precursor, has very recently been synthesized and may prove useful as a peptide photoaffinity label, although it has not yet been incorporated into peptides (10).
To simplify the synthesis of photoreactive peptides it would be desirable to prepare an amino acid which could be routinely incorporated into peptides by standard solid-phase tech-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed.
niques. Arylketone derivatives of phenylalanine appear to be ideal for this purpose because of their chemical stability and remarkably selective photochemistry. The n -?r* transition of diarylketones can be effected with relatively low energy ultraviolet radiation giving rise to a triplet biradical which preferentially reacts with C-H bonds uersw reaction with water (11,25). Finally, in cases where benzophenone derivatives have been used for photolabeling they have been highly successful. Breslow (12) has photolyzed diarylketones attached to or complexed with hydrocarbons in an attempt to achieve selective functionalization of methylene groups remote from other functionalities. In a similar study, Biro and co-workers (13) demonstrated that arylketones react with C-H bonds with high regio-and stereospecificity when the ketones were photolyzed in crystalline surfactant hosts. 4-Benzoylbenzoylpentagastrin has also been shown to label bovine serum albumin (to which it binds, albeit with poor affinity) upon photolysis (14).
In this paper we describe the synthesis of p-benzoyl-Lphenylalanine (Bpa)', and the incorporation of this residue into a 17-residue calmodulin-binding peptide by the solidphase technique. Previously (15) we described the design, synthesis, and characterization of peptide I, which binds calmodulin (CaM) in a calcium-dependent manner with a 400 PM dissociation constant. Lys-Leu-Xxx-Lys-Lys-Leu-Leu-Lys-Leu-Leu-Lys-Lys-Leu-Leu-Lys-Leu-Gly; peptide I, X x x = Trp and peptide 11, X x x = Bpa. Based on model building (16, 26), we predicted that the tryptophanyl group at position 3 could be replaced by a Bpa residue without substantially changing the affinity of the peptide for CaM. Indeed, we have found that peptide I1 binds to CaM, and forms a covalent adduct upon photolysis.

Materials
Bovine testes CaM was purchased from Pharmacia, and was homogeneous by SDS-polyacrylamide gel electrophoresis and reversephase HPLC. Peptide I was synthesized and purified as previously

10696
Photolabeling of Calmodulin chloride was removed by co-distillation with carbon tetrachloride (b.p. 69 "C). To the residue containing carbon tetrachloride, was added 5 g of potassium carbonate and 5 g of Woelm basic alumina. The mixture was refluxed for 15 min with stirring, and then filtered hot to give 57 g of crystals after cooling. Recrystallization from ethanol yielded 44.7 g (38%) of colorless crystals (m.p. 95-96 "C (17), m.p. 98 "C NMR 64.60 in CDCl3). The product contained about 2% of 4-methylbenzophenone and 0.7% by-product dichloromethylbenzophenone (NMR 62.45 and 6.75, respectively).
N-Acetyl-a-cyano-p-benzoyl-DL-phenylalanine Ethyl Ester-A stirred mixture of 41.9 g (181 mmol) ofp-chloromethylbenzophenone, 30.9 g (182 mmol) of ethyl acetamidocyanoacetate, 300 ml of acetone, 17.3 g of anhydrous potassium carbonate, and 1.6 g of potassium iodide was refluxed overnight, cooled, and filtered. The solid was washed with acetone, and the combined filtrates were evaporated under reduced pressure. The residue was crystallized from 70 ml of ethanol by cooling to 5 "C. The resulting product (59 g) was dissolved in 900 ml of hot ethanol, treated with activated charcoal, filtered, and diluted with 900 ml of hexane to give 45.6 g (69.4%) of colorless crystals (m.p. 151-152 "C). Thin layer chromatography (Merck F-254 plates, 95/5, v/v, CHCl&HBOH) showed the product to he homogeneous (RF 0.3), and its NMR was consistent with the structure. p-Benzoyl-DL-phenylalanine (o~-Bpa)-A suspension of 45.4 g of the above cyanoester in 188 ml of 8 N hydrochloric acid was heated under nitrogen at 100 "C for 20 h. The mixture was cooled, the solid was collected and washed with 8 N hydrochloric acid, then with ethanol and dried to give 36.1 g (95%) of DL-Bpa/HCl (m.p. 202-204 "C). The crude hydrochloride dissolved in 480 ml of boiling water was filtered hot, the filtrate was diluted with 480 ml of hot water and immediately neutralized (to pH 7) with 120 ml of 1 N sodium hydroxide solution. The mixture was cooled on ice and the resulting fine solid was collected by filtration, washed with water, and vacuumdried to give 31.0 g (92%) of anhydrous p-benzoyl-DL-phenylalanine (m.p. 217 "C decomposes).

CmH15NOs
Calculated C 71.36 H 5.61 N 5.20 Found C 71.10 H 5.50 N 5.11

N-Acetyl-p-benzoyl-DL-phenylalanine-
To a stirred solution of 41.2 g (153 mmol) of anhydrous DL-Bpa in 600 ml of 1 N sodium hydroxide was added 600 g of ice and 52 ml of acetic anhydride. The mixture was stirred for 5 min and was then acidified to pH 3 by the slow addition of 500 ml of 1 N hydrochloric acid. The fine crystalline product was filtered, washed with water and dried. Recrystallization from ethanol/hexane yielded 38.04 g (80%) (m.p. 174-176.5 "C); Amax 258, 332 nm (ezs8 18,000 M" cm", c332 180 M" cm") in isopropyl alcohol.

L -B~u and N-Acetyl-p-benzoyl-D-phenylalanine-
A suspension of 6.24 g (20 mmol) of acetyl-DL-Bpa in 1.0 liter of water was stirred and 3.5 ml of 4.7 N NH40H was added to bring the pH to 7.5. The solution was filtered and 200 mg of aspergillus acylase I (Sigma) and 5 drops of toluene were added. The solution was stirred at 37 "C for 18 h and was cooled to 25 "C and filtered. The damp solid product was dissolved in 50 ml of 0.5 N HCI at 70 "C, Celite was added, the suspension was filtered, and the cake was washed with hot water (50 ml). The clear filtrate was brought to pH 7 by the addition of 7.3 ml of 1 N NaOH. The solid was isolated by filtration (see below for filtrate processing), washed with water and a small amount of ethanol, and dried to give 2.36 g (80%) of L-Bpa sesquihydrate (m.p. 178-179 "C, ag5 = 3.0 f 0.8 ", concentration 1.01 g/100 ml of 1 N HC1). The aqueous filtrate from the acylase hydrolysis described above was acidified to pH 2.4 with HCl. The crystals were isolated (2.91 g) and recrystallized from 500 ml of boiling ethyl acetate to give 2.716 g of N-acetyl-p-benzoyl-D-phenylalanine (87%) (m.p. 186.5-187 "C, Synthesis and Purification of Peptide ZZ-Peptide I1 was synthesized by the Merrifield method (18) using the protecting groups and synthetic protocol described for peptide I (15). The Bpa residue was incorporated using a 3-fold excess of the corresponding Boc-protected symmetric anhydride formed by reaction of 6 eq of Boc-L-Bpa with 3 eq of diisopropylcarbodiimide in CH2C12/N,N-dimethylformamide (1/ 1) for 15 min at 0 "C. This was allowed to react with the resin and the coupling reaction was complete within 4 h as determined by the ninhydrin test (19). The peptide was cleaved from the resin by reaction with HF/p-cresol(lO:l) at 0 "C for 60 min. The crude product ( Fig. lA) was purified in a single step by reversed-phase HPLC using a Hamilton PRP-1 semipreparative column (purchased from Pierce), and a gradient of 35-41% aqueous acetonitrile containing 0.1% trifluoroacetic acid, at 0.33%/min and flow rate of 4.0 ml/min. Fractions containing pure peptide were pooled and lyophilized giving chromatographically homogeneous peptide in 19% overall yield based on the loading of the first amino acid on the resin.
Amino acid analysis ( L~u~.~' (81, L Y S~. *~ (7), G I Y ' .~ ( l ) ) , Edman sequence analysis, and analytical reversed phase HPLC (Fig. 1B) showed that the desired peptide had been obtained in homogeneous form. Fast atom bombardment/mass spectroscopy gave the appropriate parent ion (M + H)+ = 2128 indicating that the Bpa residue had been preserved intact through the synthetic and purification procedures. This was confirmed by UV spectroscopy; peptide I1 showed a single maximum at 260 nm ( e = 18,000 M" cm") which is the same M" cm"). Furthermore, the proton NMR spectrum of the peptide as that for Boc-Bpa in aqueous solution (Amax = 260 nm, c = 18,000 dissolved in dimethyl sulfoxide-d6 was consistent with the proposed structure, and qualitatively similar to L-Bpa in the aromatic region. were stored frozen at -10 "C in aqueous solution at a concentration of 2 mg/ml. Under these conditions, it was stable for at least 6 weeks. Photolubeling of Calmodulin-The photoreactions were carried out in polystyrene dishes (Costar 24-well dishes), with the wells uniformly positioned 1-2 cm from the light source (Rayonet Photochemical Reactor, 3500 A lamp, Southern New England Ultraviolet Co., with 9 lamps positioned horizontally). A t the indicated times, samples were removed from the light source and stored at -20 "C until they were analyzed. Details of the HPLC analysis of photoreactions are described in the figure legends.
Miscellaneous-SDS-gel electrophoresis was performed with 12.5% polyacrylamide gels according to Laemmli (20) with 1.0 mM EGTA in all buffers (21). For analysis of noncovalent calmodulin-peptide complexes, samples were electrophoresed on 15.0% polyacrylamide gels in the presence of 4 M urea by the method of Head and Perry (22), with the exception that 0.1 mM CaC12 was added to the gel buffers. Proteins were visualized with Coomassie Blue R-250 (Bio-Rad).
NMR spectra were recorded using either a 360-mHz Nicolet NT or a General Electric QE 300 NMR spectrometer. Amino acid analyses were obtained using a Waters Associates Picotag system. Peptides were synthesized using a Beckman 990B synthesizer.
Myosin light chain kinase and myosin light chain kinase substrate were purified from chicken gizzards and kinase assays conducted as previously described (23,27).

Synthetic Operations
As described under "Experimental Procedures," DL-Bpa was synthesized in multigram quantities by a three-step synthetic procedure starting from commercially available 4-meth- Conversion of DL-Bpa to the acetyl derivative, and resolution by hydrolysis with aspergillus acylase I gave L-Bpa, which spontaneously precipitated from solution, and Ac-D-Bpa which remained soluble. The L-isomer was converted to its at-butyloxycarbonyl-protected derivative, and incorporated into peptide I1 by the standard solid-phase technique. No special precautions were found to be necessary in handling either Bpa-derivatives or Bpa-containing peptides, and they were found to be stable indefinitely when stored at -10 "C.

Noncoualent Complex Formation between CaM and
Peptide IZ Determination of the Stoichiometry of Binding-Polyacrylamide gel electrophoresis was previously used to show that peptide I forms a calcium-dependent 1:l complex with calmodulin that is stable in the presence of 4 M urea (Fig. 2,  lunes 1-5, and Ref. 15). Similarly, when CaM is incubated with 0.5 eq of peptide I1 in 0.5 mM CaC12, a new band with an intensity approximately equal to the calmodulin band was observed (Fig. 2, lune 7). When the peptide/CaM ratio was raised to (1:1), the CaM band disappeared and the band due to the complex increased in intensity (lane 8). No new bands were observed when the peptide/CaM ratio was 21 or 3:l (lunes 9 and 10). In contrast, when the incubation and electrophoresis were carried out in the absence of calcium and with 1 mM EGTA added, no band attributable to a complex was formed, even when the peptide was in %fold excess over calmodulin (data not shown).
Binding Affinity of Peptide I Z for CaM-In order to verify that the substitution of L-Bpa for tryptophan in peptide I1 did not significantly alter the affinity of the peptide for calmodulin, the ability of peptides I and I1 to compete for calmodulin binding to the target enzyme myosin light chain kinase was assayed. We have recently described the use of this competition assay to obtain reasonable estimates of the dissociation constants of peptides for calmodulin (23) . Fig. 3 shows the activation of myosin light chain kinase induced by calmodulin in the absence and presence of 140 nM peptide I or peptide 11. The activation curves in the presence of either peptide are identical within experimental error, and are markedly different than in the absence of inhibitory peptide. Previously it was shown that peptide I binds to CaM with a 0.4 (15) to 0.7 (23) nM dissociation constant. The identity of the curves for peptides I and I1 indicates that peptide I1 also binds calmodulin with a subnanomolar dissociation constant.
Photolabeling of CaM Optimization of Conditions-In an initial attempt to optimize the conditions for photolabeling, CaM was irradiated for 6 min with 350-nm light at room temperature in the presence or absence of peptide I1 in various buffers, and the products separated by HPLC (Fig. 4). Without 350-nm radiation, a mixture of CaM and peptide I1 eluted as 2 distinct peaks with retention times of 13 and 16.5 min (panel A ) . A standard of CaM elutes as a single peak a t 16.5 min (panel B). Thus, under the HPLC conditions employed, noncovalent complexes between CaM and peptide I1 are dissociated. In the absence of peptide 11, irradiation of CaM had no effect on retention time, or peak height (not shown). When an equimolar mixture of CaM and peptide I1 was irradiated in phosphate-buffered saline (PBS, 20 mM sodium phosphate, 0.15 M NaCl, pH 7.0) containing 0.5 mM CaC12, the peak due to CaM decreased by 50% and a new broad peak with a retention time of 17.4 min appeared (panel C). Incubation of peptide I1 and CaM in PBS without CaC12 and with 1 mM EDTA resulted in a 25% decrease in the CaM peak indicating that less labeling had occurred. The yield of the presumed adduct at 17.4 min was very low, and several minor peaks were present (panel D). Previously we showed that the positively charged peptide I could interact in a nonspecific calciumindependent manner with CaM which is an acidic protein, and that this nonspecific binding is greatly reduced in the presence of 4 M urea (15). To test whether similar nonspecific interactions were giving rise to the small amount of calciumindependent labeling, the photolysis reactions were repeated in the presence of 4 M urea. In the presence of CaC12, the addition of urea yielded the same decrease of the CaM peak of approximately 50% as in the absence of urea (panel E),

TIME (Minutes)
whereas without CaC1, and with 1 mM EDTA we observed less than a 10% decrease in the CaM peak and no new peaks were detected (panel 3'). Therefore, all further experiments were carried out in the presence of 4 M urea/PBS, containing 0.5 mM CaC12.
Time Course for Photolabeling-CaM was photolyzed with 1 eq of peptide I1 in PBS buffer containing 4 M urea and 0.5 mM CaC12, and aliquots removed at various times for analysis by HPLC (Fig. 5) or SDS-polyacrylamide gel electrophoresis (Fig. 6). The time course for the loss of CaM reached an asymptotic limit of 30% residual CaM a t long exposure times, with a half-time of 4 min. The curve describing the appearance of peak I1 followed the disappearance of CaM, and also had a half-time of 4 min. The area of peak I1 at long exposure times corresponded to 50 f 15% of the initial area of the calmodulin peak before photolysis. Thus, photolysis of CaM with 1 eq of peptide I1 leads to 70% labeling, and approximately 50% of the initial calmodulin can be isolated as a covalent complex  eluting in peak 11. Increasing the molar ratio of peptide to CaM by sequential additions of fresh peptide solution resulted in a greater reduction in the CaM peak, but the chromatogram became more complex, presumably due to the formation of multivalent complexes.
The photolabeling reaction could be inhibited by addition of the non-photolabile peptide I, which competes with peptide I1 for binding to CaM. When peptide I1 and CaM in a 1:l ratio were photolyzed for 10 min in the presence of a &fold excess of peptide I, a 20% decrease in the level of CaM was observed compared to a 70% reduction obtained in the absence of added peptide I.
Determination of the Stoichiometry of the CaMIPeptide 1 1 Photoadduct-To further characterize the material eluting in peak 11, a small amount of this material was purified by HPLC. A total of 9.2 nmol of CaM and 7.5 nmol of peptide I1 were photolyzed in PBS, 4 M urea, 0.5 M CaC12 for 10 min, and the products chromatographed under the conditions described in the legend to Fig. 4. Peak I1 was collected and lyophilized; the resulting product gave a single major peak on analytical HPLC with a small amount of calmodulin (10% of the adduct peak) as the only detectable impurity. The amino acid analysis of this material showed a large increase in leucine and lysine content over that of calmodulin alone, and was consistent with the incorporation of a single molecule of peptide I1 for each molecule of CaM. The compositional analysis for pure, unreacted CaM gave L e~~.~~-L y s~.~~ (expected LeuQ-Lys7) and for the photoadduct yielded Leu'6.22-Lys'4.52 which compares favorably to the values of Leu17-Lys14 expected for a 1:l CaM-peptide I1 complex.

DISCUSSION
p-Benzoylphenylalanine is a photoreactive amino acid which can easily be incorporated into a peptide by solid-phase synthesis. This amino acid was previously synthesized by R. Galardy (24), although the present work describes the first example of the incorporation of Bpa into peptides and its use as a photolabel. Substitution of Trp for Bpa in peptide I gave rise to a peptide which specifically labeled CaM in a calciumdependent manner. Previously, we showed that the indole ring of the tryptophan in peptide I was directly involved in forming the CaM-peptide complex, and that it was held in a rigid, hydrophobic environment at the CaM/peptide interface (15). When Bpa was substituted for Trp, the peptide still formed a non-covalent complex of similar affinity; photolysis of this complex resulted in photolabeling of CaM with a 70% yield. Photolysis for longer periods failed to improve the yield, and the starting peptide disappeared from the chromatogram suggesting that a portion of the peptide either reacted intramolecularly or with solvent. The yield of labeled CaM could be increased by adding additional peptide, although this led to a complex product mixture resulting from a second low affinity peptide-binding site on CaM (15).
The ability to place Bpa anywhere in a peptide sequence provides an excellent method for mapping the binding sites of peptide receptors. By varying the position of Bpa in the sequence of various CaM-binding peptides, it should be possible to identify the residues involved in binding peptides and target enzymes.