Molecular Diversity. 1 (1995) 259-265 ESCOM

259

MOLDIV 045

Identification of cyclized calmodulin antagonists from a phage display random peptide library H e a t h e r H a n s o n Pierce a'*, Nils A d e y c a n d Brian K. K a y a'b ~Curriculum in Genetics and Molecular Biology, and bDepartment of Biology, University of North Carolina at Chapel Hill, Chapel Hill, NC27599, US.A. CMyriad Genetics, Salt Lake City, UT 84108, U.S.A.

Received 21 June 1996 Accepted 8 August 1996 Keywords." BacteriophageM13; Phage display; Conformationallyconstrained peptides; Calmodulin; Antagonists;

Mutagenesis; Alanine-scanning

Summary To isolate peptide ligands that bound calmodulin (CAM) specifically, we screened an M13 phage library displaying cyclized octamer random peptides with immobilized bovine CaM. Isolates were recovered, sequenced, and deduced to express nine independent peptides, five of which contained the sequence TrpGly-Lys (WGK). Four of the nine peptide sequences were synthesized in cyclized, biotinylated form. All of the peptides required Ca 2+ to bind CaM. The cyclized, disulfide-bonded form of one such peptide, SCLRWGKWSNCGS, bound CaM better than its reduced form or an analogue in which the cysteine residues were replaced by serine. The cyclized peptide also exhibited the ability to inhibit CaM-dependent kinase activity. Systematic alanine substitution of residues in this peptide sequence implicate the tryptophan residue as being critical for binding, with other residues contributing to binding to varying degrees. Cloning of ligand targets (COLT) confirmed the specificity of one of the cyclized peptides, yielding full-length and C-terminal CaM clones, in addition to a full-length clone of troponin C, a CaMrelated protein. This study has demonstrated that conformationally constrained peptides isolated from a phage library acted as specific, Ca2+-dependent CaM ligands.

Introduction Combinatorial peptide libraries have become a useful tool over recent years for discovering novel affinity reagents. While the approach to finding novel lead compounds has relied historically on screening vast numbers of organic compounds, which were often expensive and difficult to synthesize, phage display libraries are relatively inexpensive, and provide a readily available and renewable source of structural diversity [1]. To date, the screening of peptide libraries has led to the discovery of monoclonal and polyclonal epitopes [2-5], and peptide ligands for both membrane receptors [6] and cytosolic proteins [7,8]. Calmodulin (CAM) is a cytosolic calcium-binding protein which regulates the activity of many different cellular targets through protein-protein interactions. Proteins regulated by interaction with CaM include cAMP phosphodiesterase, adenyl cyclase [9], phosphorylase kinase *To whom correspondenceshould be addressed. 1381-1991/$ 6.00 + 1.00 9 1996 ESCOM SciencePublishers B.V.

[10], CaM-dependent protein kinase [11] and myosin light-chain kinase [12], to name just a few. The CaMbinding domains of these proteins are generally short ~zhelical peptides, typically ~20 amino acids in length, and share little primary structure [13]. Structural analyses of peptide-CaM complexes [14-17] and mutagenesis studies [18,19[ have shown that the N- and C-terminal lobes of CaM differentially bind and regulate the various CaM targets. Previous studies have identified novel CaM-binding peptides from unconstrained phage display libraries [7,20]. The linear peptides isolated had high affinity for CaM even though they had little sequence similarity to the CaM-binding domains of natural targets. In this work, we examined the interaction of CaM with peptide ligands by screening a phage display library displaying conformationally constrained peptides. By using a constrained library, we hoped to enrich for favorable binding conformations by reducing the entropy of displayed peptide

260 sequences. This reasoning was based on studies of peptide ligands of integrins which showed that disulfide-bridged, cyclic peptides from a phage display library bound the receptors with affinities as much as three orders of magnitude higher than reduced forms of the peptides, or linear peptides with the same binding motif [21,22]. In our study, nine cyclized peptide sequences were isolated which bound to CaM in a calcium-dependent manner. Five of the nine clones isolated displayed peptides with the tripeptide motif, WGK. The W G K motif was also present in adrenocorticotropic hormone, which has been shown to bind CaM in a separate study [23]. Alanine-scanning and random mutagenesis of library-derived peptides have confirmed the WGK tryptophan to be critical to binding. As expected, the cyclic form of one of the peptides bound CaM better than its reduced form or a linear analogue in which the cysteine residues were replaced by serine. Screening of a cDNA expression library with one of the peptides yielded full-length and C-terminal clones of CaM, indicating that it was interacting with the C-terminus of CaM. Some of the peptides were also capable of inhibiting CaM-dependent kinase activity, suggesting that the peptides possessed biological antagonistic activity.

Materials and Methods Cyclized random octamer library construction Two oligonucleotides were synthesized to generate the DNA cassette used in the construction of the conformationally constrained library. The first was a degenerate oligonucleotide, 5'-TGACGTCTCGAGTTGT(NNK)8TGTGGATCTAGAAGGATC-3', containing an Xho I site, where N represents an equimolar mixture of A, C, G and T; and K corresponds to an equimolar mixture of G and T. The N N K coding scheme utilizes 32 codons to encode 20 amino acids; the frequency of each amino acid is once (Cys, Asp, Glu, Phe, His, Ile, Lys, Met, Ash, Gln, Trp, Tyr), twice (Ala, Gly, Pro, Val, Thr), or thrice (Leu, Arg, Ser) per codon. The second oligonucleotide, 3'-CCTAGATCTTCCTAG-5', contained an Xba I site, and served as a primer for DNA synthesis. The library was constructed as described previously [24]. The doublestranded DNA segments were digested with Xho I and Xba I, ligated to Xho I+Xba I-digested M13-vector m663 [25], and transfected into XL1-Blue E. coli cells by electroporation. This procedure yielded 1.1 x 108 different recombinants, which were plate-amplified to generate the 'R8C' library. Isolation of CaM-binding phage To isolate peptides which bound CaM, the R8C library was screened by affinity selection as described previously [26]. Briefly, bovine-brain CaM (Sigma Chemical Co., St. Louis, MO) was immobilized on polystyrene microtiter plates (Corning, Corning, NY) in buffer (50 gl of 1 rag/

ml in 100 mM NaHCO3, pH 8.3) overnight at 4 ~ Bovine serum albumin (BSA) at a concentration of 1 mg/ml in 100 mM NaHCO 3 (pH 8.3) was added to the wells and incubated for 1 h at room temperature to block nonspecific binding. Of the R8C library (6 x 1014 pfu/ml) 5 gl was subjected to three rounds of affinity purification. After the third round of purification, the bound phage were eluted with 50 gl 50 mM glycine- HC1 (pH 2.0), 50 gl 20 mM dithiothreitol (DTT), or 50 gl 10 mM O,O'bis(2-aminoethyl)ethyleneglycol N,N,N,N'-tetraacetic acid (EGTA) to enrich for phage clones which bound CaM in a cyclization- or calcium-dependent manner, respectively. Individual plaques were isolated from bacterial lawns. The DNA of isolates was prepared from 3 ml overnight cultures with Wizard Mini-Preps (Promega, Madison, WI). Phage supernatant was reserved and stored at 4 ~ for further analysis. DNA was sequenced on both strands with oligonucleotide primers upstream and downstream of the gene-IIl cloning site of the M13-vector m663. DNA sequences were analyzed with MacVector TM. Confirmation of CaM-binding phage was performed as described previously [26].

Synthetic peptides Synthetic peptides were purchased from Chiron Mimotopes (Clayton, VIC, Australia). All peptides used were acetylated at the N-terminus and biotinylated at the Cterminus. All peptides, with the exception of plD-7 linear, were cyclized by intramolecular disulfide-bonding of the cysteine residues, as confirmed by mass spectroscopy. plD-7 linear was an analogue o f p l D - 7 cyclized, in which the flanking cysteines were replaced by serines. Binding experiments Microtiter wells were coated with 50 ng CaM in 100 mM NaHCO 3 (pH 8.5) overnight at 4 ~ We found that the microtiter wells were saturated with CaM at 50 ng per well. To block nonspecific binding, 1% BSA in 100 mM NaHCO3 (pH 8.5) was added for 1 h at room temperature. The wells were washed with 0.1% Tween-20 in PBS (PBS-Tw20). Biotinylated peptides were complexed to Streptavidin-linked alkaline phosphatase (SA-AP) (Sigma) for 30 rain at room temperature in PBS-Tw20. Complexed peptide in a volume of 25 gl PBS-Tw20 was added to the microtiter well, for a total reaction volume of 50 gl. The microtiter plate was sealed and incubated 2 h at room temperature. The wells were washed with PBS-Tw20. To detect bound peptide, 50 gl of SAP buffer (1 M diethanolamine (pH 9.8), 1.0 mM MgC12, 40 gM ZnSO4) was added to each well, incubated 5 rain at 37 ~ followed by the addition of 6 gl 60 mM p-nitrophenylphosphate. After the reaction had proceeded for approximately 10 rain at room temperature, the optical density of the wells was measured with a microplate spectrophotometer at a wavelength of 405 nm.

261

r r 01D-1 r r

SCVK WGK KEFCGS SCWOKY

WGK

SCYE WGK

ECGS LRWCGS

SCLR WGK WSNCGS SCWR WGK YQICGS

#I-6

SCVS

@1-11 01D-22 01E-3

SCIR WGQ NTFCGS

Consensus

SCWQ

WGA

LKLCGS

WGN

LKICGS

SCVR WGQ

LSICGS

WGK

Fig. 1. Amino acid sequence of peptides displayed by CaM-selected phage from R8C library. Residues within the peptides that match the consensus have been aligned. Bold residues are fixed in all library clones; sequencesincluded in syntheticpeptides are underlined.

Mutational analysis' A series of alanine-scanned variants of plD-7 were synthesized by Chiron Mimotopes. Binding activity was assayed as outlined above, by coating the microtiter wells with 500 pg CaM, complexing the peptides to Streptavidin-linked horseradish peroxidase, and using the appropriate detection buffers. In addition, mutational analysis of the 01D-1 coding region was carried out by random mutagenesis of the 01D-1 insert sequence. Oligonucleotides were synthesized corresponding to the insert sequence of 01D-1 with a 3.3% doping of each the other three bases at each nucleotide position. These oligonucleotides were annealed and filled in by chain extension in the presence of dNTPs and Taq DNA polymerase. The double-stranded oligos were digested with Xho I and Xba 1, ligated to Xho I+Xba I-cleaved M13-vector mBAX [25] and transfected into JS5 E. coli cells by electroporation. Isolated plaques were assayed for CaM-binding by ELISA. A select number of isolates with varying degrees of binding activity were sequenced. CaM-dependent kinase inhibition assay The ability of peptides to antagonize calmodulin was determined by an enzyme inhibition assay [11]. RSCderived peptides and a synthetic peptide ( L K K F N A R R K LKGAILTTMLAK) corresponding to the CaM-binding domain of CaMKn (Calbiochem, San Diego, CA) were assayed for their relative ability to inhibit CaM-dependent protein kinase II phosphorylation of syntide 2 (Sigma). A synthetic peptide corresponding to the sequence of a peptide (SCDPRAMGWACGK) that binds an antibody against Smith antigen was used as a negative control. The test reactions consisted of 100 nM CaM, 35 mM Hepes (pH 7.8), 10 mM magnesium acetate, 2 mM CaCI2, 50 mM syntide, 50 gM ATP, 56 nM [~t-32p] ATP (9000 cpm/ fmol), and test peptides (added in 5 gl aliquots). The reactions were incubated on ice before adding 2 gl 360 pg/gl of CaMK~ in enzyme dilution buffer (50 mM Hepes

(pH 7.5), 0.5 mM DTT, 0.1% Tween-20); the reaction was then incubated at 30 ~ for 10 rain. CaMK~ was the generous gift of Dr. A.R. Means (Department of Pharmacology, Duke University, Durham, NC). To quantitate the phosphorylated substrate, 40 gl of each reaction was then spotted onto a p81 phosphocellulose filter (Whatman, Hillsboro, OR). The filters were washed in 75 mM phosphoric acid, rinsed with acetone, air-dried, and counted by scintillation autoradiography.

Cloning of ligand targets The 1EXlox 16-day mouse embryo cDNA expression library was obtained from Novagen (Madison, WI). COLT was performed as described previously [27]. Briefly, plD-7 was complexed with SA-AP and incubated with blocked filter lifts of the library overnight at 4 ~ in fresh blocking solution. Bound SA-AP-peptide complexes were detected by incubation in the appropriate detection reagents [27]. Positive plaques were cored, diluted, replated and rescreened twice more, so that positive plaques could be recovered. Plasmids were rescued from the XEXlox phage, and the nucleotide sequences of the inserts were determined by fluorescent dideoxynucleotide sequencing. Results and Discussion In an effort to isolate high-affinity peptide ligands for CaM, we screened a phage display library of peptides with CaM. The random peptides were designed to be conformationally constrained by intramolecular disulfide bonding of flanking cysteine residues. Phage clones expressing CaM-binding peptides were affinity-purified with immobilized bovine-brain CaM. Nine phage clones expressing unique CaM-binding peptides were isolated;

2.5

gl.5 to O 0.5 0 84

pl D-7

pl D-1 p1-11 PEPTIDE

p1-3

Fig. 2. Calcium-dependence of binding by CaM-selected peptides. Binding reactions were done in the presence of 1 gM biotinylated peptide. Binding was measured by ELISA for three buffer conditions: 1 gM CaC12, 1 gM CaC12+ 500 gM EGTA, and 500 gM EGTA + 1 mM CaCI2. Bars represent the average OD at 405 nm for triplicate wells, and error bars represent the standard deviation.

262 A

g

a

4.5 4.

2.5

--B--plD-7 plD-1 p1-11 9 pl-3

2 E

e., it)

o 1.5

/

3.5. 3-

~i'

"E 2.5-

/

o "="

O

z

O

1

1 I ..ii

2-

=

1.5.

0.5

0.5

|

mm w

0

1

5E+2

m

w

5E+3

I

5E+4

1

I

i

5E+5

5E+6

0

pM PEPTIDE

I

125

i

w

I

i

I

250

500

1000

gM DTT

Fig, 3. In vitro binding characteristics of CaM-selected peptide sequences from the R8C library. (A) Relative affinities of CaM-selected peptide sequences from the R8C library. 50 ng of bovine-brain calmodulin was immobilized in each microtiter well, and binding of added biotinylated peptides was measured by ELISA. Points represent the average OD at 405 nm for triplicate wells at the given peptide concentration. Error bars represent the standard deviation. (B) Effect of reducing conditions on binding of a cyclized CaM-selected peptide. Of bovine-brain CaM 50 ng was immobilized in each microtiter well, and binding of added biotinylated peptides (2.5 gM) was evaluated by ELISA. plD-7 linear was an analogue of pl D-7 cyclized in which the cysteine residues were replaced by serine residues. Data points represent the average OD by ELISA for triplicate wells at the indicated concentration of dithiothreitol (DTT); error bars represent the standard deviation. CaM 12.1 is a CaM-binding peptide selected from a linear 26-mer random peptide library (Brennan, manuscript in preparation).

deduced amino acid sequences of the peptides are listed in Fig. 1. The majority of the phage clones contained the tripeptide motif WGK at the same relative position of the peptide. While there was no obvious sequence identity flanking the motif, there was an overall pattern of alternating hydrophobic and charged residues. Protein database searches conducted with the WGK motif yielded only one match, adrenocorticotropic hormone (ACTH), a 39-amino-acid-long peptide involved in corticosteroidogenesis. Previous studies have shown that ACTH can bind CaM in a calcium-dependent manner, and changes in the fluorescence emission spectrum and quantum yield of ACTH, when CaM-bound, showed increased shielding of the single tryptophan, which suggested that it was interacting with hydrophobic regions of CaM [23]. Altogether, these data suggest that the sequence WGK is at least a significant part of the CaM-binding motif in the displayed sequences. To facilitate the characterization of the displayed sequences, biotinylated cyclized peptides were synthesized corresponding to the peptides expressed by CaM-binding phage clones (Fig. 1). We determined whether the libraryderived peptides bound in a calcium-dependent manner by ELISA (Fig. 2). Each of the library-derived peptides bound CaM and not BSA (background) in the presence of calcium, and when an excess of chelating agent (EGTA) was added simultaneously to the binding reac-

tions, binding of each peptide was reduced to background levels. When calcium was added back at the beginning of the incubation time, binding was restored; full recovery of binding required a twofold molar excess of calcium versus EGTA. Restored levels of binding were slightly higher than the original levels, because the calcium concentration did not saturate the Ca2+-binding sites of CaM in the original reactions. These results demonstrated that the library-derived peptides bound CaM with a high level of calcium dependency. This is consistent with the observation that, with very few exceptions [28-30], the majority of naturally occurring CaM-binding proteins bind in a calcium-dependent manner. The relative binding strength of the peptides was measured by assaying binding to immobilized bovinebrain CaM in the presence of calcium (Fig. 3A). By far, plD-7 exhibited the strongest binding to CaM, yielding an ELISA signal approximately three times greater than any of the other peptides, p l - l l and pl-3 exhibited similar binding to each other, and pID-1 exhibited the lowest binding of all the peptides tested. In addition, plD-7 produced a comparable ELISA signal to a CaMbinding peptide (CAM12.1) isolated from a library displaying random linear peptides 23 amino acids in length (Fig. 3B). Binding studies conducted to map the sites of interaction of CAM12.1 with CaM showed that the N- or C-terminal halves alone did not bind CaM and that a 17-amino-acid stretch was responsible for the interaction

263 (Brennan, manuscript in preparation). Thus, we have isolated a cyclized peptide which appears to bind CaM with similar properties to a linear peptide nearly three times its length. Direct binding assays were conducted under reducing conditions to verify that the cyclized form of plD-7 was the active binding species (Fig. 3B). At very low concentrations of reducing agent, the binding of plD-7 was reduced to background levels, while CAM12.1, a linear peptide of comparable affinity, maintained its affinity for CaM at even the highest concentrations of reducing agent. To confirm this observation, we synthesized a biotinylated linear species of p 1D-7, replacing the cysteine residues with serines; as expected, the linear analog of plD-7 exhibited much weaker binding to CaM than its cyclized counterpart. These data taken together indicate that cyclization is critical for the binding of plD-7 to CaM. While factors such as stability and transport may contribute to the activity of cyclic CaM-binding peptides in vivo, these data indicate that the superior function of cyclic versus linear CaM-binding peptides in general is due in large part to improved affinity, since the influence of such biological factors is eliminated in an in vitro assay. In order to determine whether peptides identified in this study bound CaM in a manner capable of antagonizing its function, we tested the peptides in a CaM-dependent enzyme inhibition assay. The ability of plD-7 and its linear analog to inhibit CaM-dependent kinase II

(CaMKII) was compared to the CaM-binding domain of CaMKH, which has an IC50 of 52 nM [31]. As shown in Fig. 4, both the CaMK~ CaM-binding peptide and plD-7 were potent inhibitors of CaMKII activity. The linear analog of plD-7 also inhibited CaMK~ activity in a concentration-dependent manner, but not as effectively as the cyclized form of plD-7. Thus, we have demonstrated that a peptide identified in this study is capable of inhibiting CaM-dependent kinase I1 activity with similar potency as the CaM-binding domain of CaMKH. Mutational analysis was conducted to delineate the primary structural requirements for the high affinity of peptides expressed on CaM-binding phage clones. A series of alanine-scanned peptide varianta was synthesized corresponding to plD-7, and direct binding assays were conducted with the peptide variants (Fig. 5) relative to 'wild-type' plD-7. The majority of the variants exhibited reduced binding to CaM with respect to the parent sequence, with the most dramatic reductions exhibited by W3A, K5A, and W6A. In addition, L1A, S7A, and N8A exhibited moderately reduced binding. Unfortunately, no binding data were available for the R2A variant due to solubility problems. Interestingly, G4A exhibited a slightly higher CaM binding signal than the parent sequence, suggesting that strict conservation of the glycine in W G K is not required for binding. It is possible that a small amino acid side chain is optimal at this position in the disulfide-constrained loop, but the affinities of a complete substitution set would have to be determined to conclu-

12000-

9000-

~9

6000-

3000.

0.

r

l'~

eq

i

CaM

CKCB

Negative

plD-7 cyc

plD-7 lin

COMPETITOR CONCENTRATION Fig. 4. Inhibition of CaM-dependent Kinase lI activity by CaM-selected peptides. Reactions were done in the presence of 100 nM CaM. CaM-dependent Kinase II phosphorylationof syntide was measured as the amount of 32pincorporation, expressedas counts per minute (cpm). plD-7 linear was an analogue of plD-7 cyclizedin which the cysteine residues were replaced by serine residues. 'CKCB' is the CaM-binding domain peptidefrom CaM-dependentKinaseII (LKKFNARRKLKGAILTTMLAK),and 'Negative'is a cyclizedpeptide whichbinds an antibody to the Smith antigen (SCDPRAMGWACGK).

264 sively determine the specificity of position four. It is also possible that a broader consensus sequence exists, which is supported by the fact that four of the nine clones isolated contained a W G X sequence; one of these four clones also bound CaM better than a clone containing W G K (Fig. 3A). In parallel, mutational analysis of the ~)ID-1 coding region was carried out by random mutagenesis of the qMD-1 insert sequence for phage display. Individual recombinants were assayed for binding to CaM by ELISA, and a select number of isolates with varying degrees of binding activity were sequenced (data not shown). The most striking result of this analysis was that any amino acid substitution of the central tryptophan of the W G K motif (W3) completely destroyed binding to CaM. This observation, along with the background level binding of the W3A variant of plD-7, further supported the conclusion that the tryptophan of the W G K sequence was essential to the binding of the library-derived peptides. To map the site of interaction of CaM with plD-7, we used the peptide to perform COLT [27]. Briefly, the biotinylated peptide, complexed with Streptavidin-linked alkaline phosphatase, was used to screen a 16-day mouse embryo cDNA expression library. The results of the screen are listed in Table 1. The p l D - 7 peptide selected only CaM and troponin C c D N A clones out of the library; the detection signal was equivalent for all four c D N A clones. The cloning of a C-terminal fragment of CaM indicated that the C-terminus was sufficient for binding of the cyclized peptide. This evidence, along with the small size of the peptide, and the fact that tryptophan is a critical residue for the interaction of the C-terminus of CaM with natural targets, supports the hypothesis that

2.5 2gl,5Igj

O 0.5 0-

TABLE 1 RESULTS OF COLT WITH plD-7 Clone

Identity

Amino acids

I4 16 19 I59

Calmodulin Calmodulin Calmodulin Fast skeletal troponin C

46-149 1-149 1-149 1-160

Cloning of ligand targets (COLT) was performed by screening the )~EXlox 16-daymouse embryo cDNA expression library with biotinyl-

ated plD-7 precomplexed with Streptavidin-linked alkaline phosphatase. Shown are the positive clones obtained alter three rounds of screening. I6 and I9 were distinct clones which differed in the lengths of their 5' and 3' untranslated regions. our cyclized peptides primarily bind the C-terminal portion of CaM.

Conclusions We have constructed a phage display library of cyclized random 8-mer peptides, and identified peptides that bind CaM in a calcium-dependent manner. The isolation of a cyclized peptide whose binding strength is similar to that of a much longer peptide isolated from a linear phage display library has demonstrated that cyclized peptides can yield small, high-affinity ligands. Small, heterocyclic drugs such as trifluoperazine have already been shown to act as potent CaM antagonists [32]. Our observation that cyclization was critical for binding adds to the growing body of research which shows that placing short binding sequences in a cyclized context often enhances binding strength [6,21,22,33-35]. Because of their small size, we suspect that the cyclized peptides are interacting with only one domain of CaM. We propose that plD-7 is primarily interacting with the C-terminus of CaM because a C-terminal fragment of CaM was obtained by COLT, and because tryptophan is a critical residue for the interaction of the C-terminus of CaM with natural targets [36]. In addition, COLT demonstrated the excellent selectivity of the peptides, since only CaM and CaM-related proteins were selected from a library rich with diverse cDNA sequences. Finally, since the cyclized, disulfidebonded peptides bound CaM in a manner capable of antagonizing its function, they should prove to be useful tools for studying the function of CaM in vitro, and in vivo with the introduction of non-reducible cyclic constraints.

Acknowledgements PEPTIDE

Fig. 5. Relative affinities of plD-7 alanine-scanned variant peptides. Of bovine-brain CaM 500 pg was immobilizedin each microtiter well. Bars represent the average OD by ELISA for duplicate wells at 2.5 gM peptide. Residues are numbered with respect to the C-terminal cysteine. Data were not available for the R2A peptide because it was insoluble in all solvents tested.

The authors would like to thank N o a h Hoffman and James Rider for technical advice. We would also like to thank Anthony Means in the Department of Pharmacology at Duke University for the generous gift of CaMK~. Research support was provided by N I H Training Grant # T 3 2 GM07092 and Cytogen Corporation (Princeton, N J).

265

References 1 Kay, B.K., Biologically displayed random peptides as reagents in mapping protein-protein interactions, Perspect. Drug Discov. Design, 2 (1995) 251-268. 2 Cwirla, S.E., Peters, E.A., Barrett, R.W. and Dower, W.J., Peptides on phage: A vast library of peptides [or identifying ligands, Proc. Natl. Acad. Sci. USA, 87 (1990) 6378-6382. 3 Kay, B.K., Adey, N.B., He, Y.S., Manfredi, J.R, Mataragnon, A.H. and Fowlkes, D.M, An M13 phage library displaying random 38-amino-aeid peptides as a source of novel sequences with affinity to selected targets, Gene, 128 (1993) 59-65. 4 Motti, C., Nuzzo, M., Meola, A., Galfre, G., Felici, E, Cortese, R., Nicosia, A. and Monaci, E, Recognition by human sera and immunogenicity of HBsAg mimotopes selected fiom an M13 phage display library, Gene, 146 (1994) 191-198. 5 Scott, J.K. and Smith, G.P., Searching for peptide ligands with an epitope library, Science, 249 (1990) 386-390. 6 Koivunen, E., Wang, B. and Ruoslahti, E., Isolation of a highly specific ligand for the alpha-5 beta-1 integrin fJ'om a phage display library, J. Cell Biol., 124 (1994) 373-380. 7 Dedman, J.R., Kaetzel, M.A., Chan, H.C., Nelson, D.J. and Jamieson, G.J., Selection of targeted biological modifiers from a bacteriophage library of random peptides. The identification of novel calmodulin-regulatory peptides, J. Biol. Chem., 268 (1993) 23025-23030. 8 Sparks, A.B., Quilliam, L.A., Thorn, J.M., Der, C.J. and Kay, B.K., Identification and characterization of Src SH3 ligands J~'om phage-displayed random peptide libraries, J. Biol. Chem., 269 (1994) 23853-23856. 9 Munier, H., Blanco, F.J., Precheur, B., Diesis, E., Nieto, J.L., Craescu, C.T. and Barzu, O., Characterization of a synthetic calmodulin-binding peptide derived from Bacillus anthracis adenylate cyclase, J. Biol. Chem., 268 (1993) 1695-170l. 10 Newsholme, R, Angelos, K.L. and Walsh, D.A., High- and intermediate-affi)tity calmodulin-b#~ding domains ofthe alpha- and betasubunits of phosphorylase kinase and their potential rote in phosphory[ation-dependent activation of she hotoenzyme, J. Biol. Chem., 267 (1992) 810-818. 11 Payne, M.E., Fong, Y.L., Ono, T,, Colbran, R.J., Kemp, B.E., Soderling, T.R. and Means, A.R., Calciumcal/modulin-dependent protein kinase 11. Characterization oJ distinct calmodulin binding andinhibito~y domains, J. Biol. Chem., 263 (15) (1988) 7190-7195. 12 Means, A.R., Bagchi, I.C., VanBerkum, M.E and Kemp, B.E., Regulation of smooth-muscle myosin light-chain kinase by calmodulin. [Review], Adv. Exp. Med. Biol., 304 (1991) 11-24. 13 O'Neil, K.T. and DeGrado, W.E, How calmodulin binds' its targets: Sequence-independent recognition of amphiphilic alpha-helices [Review], Trends Biochem. Sci., 15 (1990) 59-64. 14 Chapman, E.R., Alexander, K., Vorherr, ~i:, Carafoli, E. and Storm, D.R., Fluorescence energy-transfer analysis of calmodulinpeptide complexes, Biochemistry, 31 (1992) 12819-12825. 15 Ikura, M., Clore, G.M., Gronenborn, A.M., Zhu, G., Klee, C.B. and Bax, A., Solution structure of a calmodulin-target peptide complex by multidimensional NMR, Science, 256 (1992) 632-638. 16 Meador, W.E., Means, A.R. and Quiocho, F.A., Target enzyme recognition by calmodulin: 2.4/t Structure of a calmodulin-peptide complex, Science, 257 (1992) 1251-1255. 17 Roth, S.M., Schneider, D.M., Strobel, L.A., Van, B.M., Means, A.R. and Wand, A.J., Characterization oJ the secondary structure of calmodulin in complex with a calmodulin-binding domain peptide, Biochemistry, 31 (1992) 1443-1451. 18 Ohya, 3(..and Botstein, D., Diverse essentialfunctions revealed by complementing yeast catmodulin mutants, Science, 263 (1994)963-966.

19 VanBerkum, M.F. and Means, A.R., Three amino acid substitutions in domain I of calmodulin prevent the activation of chicken smooth-muscle myosin light-chain kinase, J. Biol. Chem., 266 (1991) 21488-21495. 20 Adey, N.B. and Kay, B.K., Identification of calmodulin-binding peptide consensus sequences from a phage-displayed random peptide library, Gene, 169 (1996) 133-134. 21 Koivunen, E., Wang, B. and Ruoslahti, E., Phage libraries displaying cyelic peptides with different ring sizes'." Ligand specificities of the RGD-directed integrins, Biotechnology, 13 (1995) 265 270. 22 O'Neil, K.T., Hoess, R.H., Jackson, S.A., Ramachandran, N.S., Mousa, S.A. and DeGrado, WE, Identification of novel peptide antagonists for GPIlb/llIa J~om a conformationally constrained phage peptide library, Proteins, 14 (1992) 509-515. 23 Malencik, D.A. and Anderson, S.R., Binding of simple peptides; hormones', and neurotransmitters by calmodulin, Biochemistry, 21 (1982) 3480-3486. 24 Adey, N.B., Sparks, A.B. and Beasley, J.R., Construction of phage random peptide libraries, In Kay, B.K., Winter, J. and McCafferty, J. (Eds.) Phage Display of Peptides and Protein Domains: A Laboratory Manual, Academic Press, San Diego, CA, in press. 25 Fowlkes, D.M., Adams, M.D., Fowler, V.A. and Kay, B.K., Multipurpose vectors"Jbr peptide expression on the MI3 viral surface, Biotechniques, 13 (1992) 422-428. 26 Sparks, A.B., Adey, N.B., Quilliam, L.A., Thorn, J.M. and Kay, B.K., Sereening phage-displayed random peptide libraries for SH3 ligands, Methods Enzymol., 255 (1995) 498-509. 27 Sparks, A.B., Hoffman, N.G., McConnell, S.J., Fowlkes, D.M. and Kay, B.K., Cloning of ligand targets: Systematic isolation of SH3 domain-containingproteins, Nat. Biotechnol., 14 (1996) 741-744. 28 Apel, E.D., Byford, M.E, Au, D., Walsh, K.A. and Storm, D.R., Identification of the protein kinase C phosphorylation site in neuromodulin, Biochemistry, 29 (1990) 2330-2335. 29 Baudier, J., Deloulme, J.C., Van, D.A,, Black, D. and Matthes, H.W, Purification and characterization of a brain-specific protein kinase C substrate, neurogranin (p17). Identification of a consensus amino acid sequence between neurogranin and neuromodulin (GAP43) that corresponds" to the protein kinase C phosphorylation site and the ealmodulin-binding domain, J. Biol. Chem., 266 (1991) 22%237. 30 Urbauer, J.L., Short, J.H., Dow, L.K. and Wand, AJ., Structural analysis of a novel interaction by calmodulin." High-affinity binding of a peptide in the absence of calcium, Biochemistry, 34 (1995) 8099 8109. 31 James, R, Vorherr, T. and Carafoli, E., Calmodulin-binding domains. Just two .faced or multi-faceted?, Trends Biochem. Sci., 20 (1995) 38-42. 32 Vandonselaar, M., Hickie, R.A., Quail, J.W. and Delbaere, L.T., Trifiuoperazine-induced conformational change in Cae§ Nat. Struct. Biol., 1 (1994) 795-801. 33 Giebel, L.B., Cass, R.T., Milligan, D.L., Young, D.C., Arze, R. and Johnson, C.R., Screening of cyclic peptide phage libraries identifies' ligands that bind Streptavidin with high affinities, Biochemistry, 34 (1995) 15430 15435. 34 Koivunen, E., Gay, D.A. and Ruoslahti, E., Selection ofpeptides binding to the alpha-5 beta-1 integrin Jrom phage display library, J. Biol. Chem., 268 (1993) 20205-20210. 35 McLafferty, M.A., Kent, R.B., Ladner, R.C. and Markland, W, M13 bacteriophage displaying disulfide-constrained mieroproteins, Gene, 128 (1993) 29-36. 36 Clore, G.M., Bax, A., Ikura, M. and Gronenborn, A.M., Structure o[ calmodulin-taJ~et peptide complexes, Curt. Opin. Struct. Biol., 3 (1993) 838-845.