PROTEINS: Structure, Function, and Genetics 141-9 (1992)

Alkaline Phosphatase-Somatostatin Hybrid Proteins as Probes for Somatostatin-14 Receptors Hanno T. Langen’ and John W . Taylor’72 ‘Laboratory of Bioorganic Chemistry and Biochemistry, The Rockefeller University, New York, New York 10021, and ‘Department of Chemistry, Rutgers University, Piscataway, New Jersey 08855-0939

ABSTRACT By inserting appropriate pep- tide ligands into surface loops on globular pro- teins, we expect to develop probes for the loca- tion, accessibility, and steric and electrostatic environment of these ligand-binding sites on their membrane-bound receptors. Three resi- dues in a loop on the surface of E. coli alkaline phosphatase were substituted by an 18-residue peptide containing the receptor-binding seg ment of somatostatin-14 without significantly affecting the catalytic properties of the enzyme. This hybrid protein was then used to investi- gate the ligand-binding site of somatostatin re- ceptors. Tryptic cleavage of the hybrid protein within the inserted sequence, and binding of the hybrid protein to antisomatostatin antibod- ies demonstrated the surface accessibility of the guest peptide. Both the wild-type enzyme and the hormone-enzyme hybrid displaced “‘I- labeled somatostatin from rat brain membrane receptors only at high concentrations. How- ever, chemical cationization of the hybrid pro- tein, which again did not disturb the phos- phatase activity, enhanced its receptor-binding potency to a level only 23 times lower than that of somatostatin itself and 280 times higher than that of the cationized wild-type protein. This al- kaline phosphatase/somatostatin hybrid pro- tein appears, therefore, to be a suitable starting point for the development of probes for the steric and electrostatic environment of the ligand-binding site of somatostatin receptors. 0 1992 Wiley-Liss, Inc.

Key words: protein engineering, cassette mu- tagenesis, peptide hormone, molec- ular modeling

INTRODUCTION Understanding how membrane-bound receptors

are located, bound, and activated by their respective ligands, and learning how to control these functions, are important aspects of receptor research where current understanding is greatly limited by the spe- cific structural information available. A few models of such recognition events have been developed for certain systems, including acetylcholine receptor recognition by agonists and antagonists such as a-

0 1992 WILEY-LISS, INC.

bungarotoxin,’ lipoprotein binding by the apoBE receptor,2 and receptor selective binding by the opi- oid pep tide^.^ In addition, some structural informa- tion has been obtained from fluorescence tech- niques, including a specific distance measurement for the EGF-receptor complex4 and the characteriza- tion of bound peptide ligand accessibility for anti- body binding.5

As an alternative approach, we are exploring the use of globular proteins that incorporate genetically engineered, surface-inserted “guest” peptide li- gands6 as multifunctional probes for studying the structural environment of the ligand-binding sites of transmembrane receptors. For example, by deter- mining how the steric bulk of the host protein or the length of the inserted peptide ligand loop influences receptor binding, the accessibility of the binding site could be probed. Alternatively, such hybrids might be used to probe the electrostatic environment of the ligand-binding site by varying the surface potential of the hybrid through mutagenesis of the host pro- tein structure. If tight-binding hybrids can be devel- oped, it should also be possible to attach fluorescent reporter groups to cysteine residues introduced at various points on the hybrid protein surface. Multi- ple distance measurements may then be used to lo- cate the ligand-binding site relative to the cell sur- face by fluorescence energy transfer method^.^

In our initial efforts to develop this guest peptide approach, we are investigating the properties of hy- brid proteins designed by computer modeling to have active segments of the peptide hormoneheuro- transmitter somatostatin7 inserted into a surface loop in the crystal structure of E. coli alkaline phos- phatase (AP).* Somatostatin was originally isolated

*The following abbreviations have been used: AP, alkaline phosphatase; CD, circular dichroism; CAP, cationized alkaline phosphatase; cSAP, cationized somatostatidalkaline phos- phatase hybrid protein; SAP, somatostatidalkaline phos- phatase hybrid protein; SDS, sodium dodecyl sulfate; X-Gal, 5-bromo-4-chloro-3-indolyl P-D-galactopyranoside; X-P, 5- bromo-4-chloro-3-indolyl phosphate (p-toluidine salt).

Received October 24, 1991; revision accepted December 4, 1991.

Address reprint requests to John W. Taylor, Department of Chemistry, Rutgers University, P. 0. Box 939, Piscataway, NJ 08855-0939.

2 H.T. LANGEN AND J.W. TAYLOR

A H-Ala-Gly-Cys-Lys-an-Phe-Phe

I \ T\rp

I LYS

I / HO-Cys-Ser-Thr-Phe-Thr

B (AP91) -Gly-Thr-Glu-Cys-Lys-an-Phe-Phe

I \ s TQ I I I p" S

(-95) -!Cyr-Asp-Val-Cys-Ser-Thr-Phe-Thr

Fig. 1. Amino acid sequences of (A) somatostatin-14 and (B) the extended somatostatin sequence substituted for AP residues 92-94 in the hybrid protein SAP.

from sheep hypothalimus in its 14-residue form (so- matostatin-14) as an inhibitor of growth hormone release. It is also present in a 28-residue form, and both peptides have a wide distribution in the mam- malian central nervous system and periphery, and share multiple pharmacological activities with over- lapping specificities. Recently, a somatostatin recep- tor complex from rat pancreatic membranes has been characterized in a solubilized form that binds both somatostatin-14 and somatostatin-28, has an Wapp of about 400,000 by gel permeation chroma- tography under native conditions, a PI of 4.8, and appears to be G-protein linked.' Structure modifica- tion studies of somatostatin-14 have indicated that the amino acid sequence -Phe-'lkp-Lys-Thr- (resi- dues 7-10, Fig. lA), constrained in a p-turn confor- mation by disulfide or other, nonnatural bridges, is sufficient for tight receptor binding, and that the residues N-terminal or C-terminal to this sequence are not required.'-'' There appeared, therefore, to be a high probability that the requirements for re- ceptor binding by this hormone would not be lost as a consequence of the insertion of this receptor rec- ognition epitope into a loop structure on the surface of a compatible host protein.

AP was chosen as the host protein for several rea- sons. The crystal structure of AP has been de- scribed,12-14 and the molecular dimensions of this 94-kDa dimeric enzyme are expected to be appropri- ate for probing the extracellular domains of receptor structures. Also, its phosphatase activity may readily be monitored as a measure of retention of this structure after sequence modification, using col- orimetric assays in solution or, in vivo, on agar plates. AP is a periplasmic protein expressed at high levels in E. colt and readily purified from the iso-

lated periplasm by elution on DEAE-cellulose. Fi- nally, AP is a very stable protein that is not readily denatured by physical or chemical means, is resis- tant to proteolytic degradation, and is known to tol- erate peptide insertions in surface loops without sig- nificant loss of enzyme activity.6

The design, genetic engineering, and functional characterization of our first somatostatidalkaline phosphatase hybrid protein are now described. The results indicate that such hybrid proteins do indeed have the potential to serve as molecular probes suited to exploring the steric and electrostatic envi- ronment of the ligand-binding site of somatostatin receptors.

Molecular Modeling Crystal structure coordinates obtained after fur-

ther refinement of the structure published by So- wadski et a1.l' were provided by E. E. Kim and H. W. Wyckoff, Department of Biochemistry, Yale University, New Haven, CT. A refined structure has recently been described by Kim and Wy~koff. '~, '~ Protein and peptide modeling studies were per- formed on a Stellar GS 1000 Supermini computer running Quanta version 2.1 software (Polygen Corp., Waltham, MA).

Genetic Engineering Restriction enzyme digests, agarose gels, plasmid

preparations, ligations, transfections, and indicator plate assays were all performed essentially as de- scribed by Maniatis et al." The p h d gene used in these studies was isolated from the E . coli chromo- some by Inouye et a1.16 The starting vector consisted of a 2.5-kb HindIII-BarnHI restriction fragment containing the phoA gene that had been subcloned into a pBR322-derived vector having the small AccI DNA fragment (nucleotides 651-2246) removed, as previously reported.6 Oligonucleotides were synthe- sized on a Systec Microsyn 1450A automated DNA synthesizer, using P-cyanoethylphosphoramidite chemistry. l7 These were purified by electrophoresis in 6% polyacrylamide gels containing 7 M urea, fol- lowing the method of Maxam and Gilbert." DNA sequencing was performed locally around the site of the mutations by the dideoxy chain termination method of Saenger et al.," using a synthetic oligo- nucleotide primer and ~ A T P c x ~ ~ S for visualization by autoradiography.

Plasmid vector pHL101, which is suitable for the introduction of multiple DNA inserts coding for bio- active peptides by one-step cassette mutagenesis, was created by oligonucleotide-directed mutagene- sis of the phoA gene in M13 DNA by the high-effi- ciency phosphorothioate method,20s21 using Amer- sham Corp.'~ in vitro mutagenesis kit and following the manufacturer's instructions. The mismatched oligonucleotide used was 5'-TAT GCG CTG AAT

METHODS

HYBRID PROTEINS DESIGNED AS RECEPTOR PROBES 3

AAA GGT ACC CGG G GTC GAC TAT AAA CCG GAC TAC GTC-3'. Correct mutation introduced a frameshift and a KpnISaZI cassette with a new SmaI site a t the desired site of peptide loop inser- tions corresponding to AP (92-94). Analysis for the SmaI site in Rf-DNA purified from 10 different plaques identified 8 positives, and the correct struc- ture and location of the mutation was confirmed by sequencing the single-stranded phage DNA. The mutant gene was then subcloned back into the pBR322 ( - A d ) vector, giving pHL101. The phoA indicator strain, AW1043," was transformed to ampicillin resistance by this plasmid vector.

Cassette mutagenesis of pHLlOl to generate a plasmid for expression of SAP, pHL5, was performed by ligation of the following oligonucleotides into the KpnI-SaZI site in pHL101:

5'-C GAG TGC AAG AAC 'ITC TTC T- 3'-CA TGG CTC ACG TTC 'M'G AAG AAG A-

GG AAG ACC 'ITC ACC TCG TGC G-3' CC TTC TGG AAG TGG AGC ACG CAG CT-5'.

Plasmid preparations containing the desired cas- sette were identified as blue AW1043 colonies with restored phosphatase activity on X-P (Sigma) indi- cator plates, by analysis for the new XmnI restric- tion site, and then by local sequencing of the single- stranded DNA after subcloning into M13 mp18 DNA.

Protein Isolation and Purification A 100 ml culture of AW1043 cells carrying the

appropriate pBR322-derived plasmid was grown overnight from a single colony to stationary phase in MOPS minimal medium2, supplemented with 10 mM KH,PO,, 4 mg/liter dextrose, 1.5 mg/liter cas- amino acids, 50 mg/liter penicillin G, and 25 mg/ liter kanomycin. This culture was diluted 100-fold into 10 liters MOPS minimal medium with the same supplements, except that the KH,PO, concentration was reduced to 0.1 mM in order to induce expression of the AP gene. The 10 liter culture was maintained at 37°C in a 20-liter Nalgene container standing in a waterbath, and was mixed and aerated using an air line passing through a glass wool plug, a 22-pm fil- ter, and a sintered glass aerator. Induction of phos- phatase activity was followed by the method of Brickman and Be~kwith.,~ Typically, enzyme activ- ity was maximal after 6-7 hr growth, a t which time the cells were harvested by centrifugation.

Periplasmic proteins were collected by osmotic shock, and proteins were then precipitated from the supernatant with ammonium sulfate and purified by elution on DEAE-cellulose' followed by an Ultra- gel ACA 34 (LKB) column eluted with a buffer con- sisting of 10 mM Tris-HC1, pH 8.0, 1 mM MgCl,, and 0.1 mM ZnC1,. Fractions with high phosphatase ac- tivity were pooled and stored at 4°C for direct use in

assays. Analysis of these proteins by electrophoresis on overloaded SDS-polyacrylamide t 10%) gels'' in- dicated single bands corresponding to the correctly processed proteins, with few impurities a t detectable levels. Enzyme concentrations were determined spectrophotometrically at 278 nm using a value of Eo.1% = 0.72 cm-' for wild-type AP2, and an ad- justed value of Eo.'% = 0.78 cm-' for SAP to account for the surface Trp side chain.,' All concentrations refer to the dimeric proteins.

Enzyme Kinetics Enzyme activities were assayed as the rate of hy-

drolysis of p-nitrophenyl phosphate (Sigma 104) to p-nitrophenol in the presence of 1 M T r i ~ . , ~ Reac- tions were monitored by following the increase in absorbance at 410 nm (A€ = 1.62 x lo4 M-' cm-') in solutions of increasing substrate concentration in 1 M Tris-HC1, 10 mM MgCL,, 50 pM ZnCl,, pH 8.0, at 25°C.

Circular Dichroism Studies The circular dichroism (CD) spectra of solutions of

AP and SAP buffered with 10 mM Tris-HC1,l mM MgCl,, 50 pM ZnCl,, pH 8.0, were measured be- tween 190 and 260 nm on an Aviv model 62DS spec- tropolarimeter. The protein concentration was 1 .O pM and the pathlength of the cell was 1.0 mm.

Trypsin Digestion Wild-type AP and SAP (0.3 mg/ml) were incu-

bated with 30 pg/ml trypsin (TPCK-treated, Wor- thington) in 10 mM Tris-HC1,lO mM NaOAc, 1 mM MgCl,, 50 pM ZnCl,, pH 8.0 at 37°C for 24 hr. Sam- ples were then analyzed by electrophoresis on SDS- polyacrylamide (10%) gels,15 or were sequenced by automated Edman degradation in the Protein Se- quencing Facility at Rockefeller University.

Antibody Binding Assays Polyclonal antibodies raised against somato-

statin-14 conjugated to BSA as carrier and [1251- Tyr'1]-somatostatin-14 (2000 Ci/mmol) were pur- chased from Amersham Corp. Approximately 3,000 cpm aliquots of radioligand were incubated together with antiserum (diluted according to the manufac- turer's instructions) and various concentrations of unlabeled somatostatin-14, AP, or SAP as compet- ing ligands, in a total volume of 800 p1 of 50 mM sodium phosphate, pH 7.2, for 24 hr at 4°C. All in- cubations were performed in triplicate. Incubation was terminated by addition to each tube of 500 p1 of a suspension of 1 g of activated charcoal (Sigma) in 50 ml of 0.2% (w/v) dextran solution (Sigma, MW,, = 79,100). Free radioligand, which was absorbed by the charcoal, was pelleted by centrifugation at 10,OOOg for 3 min. Bound (supernatant) and free (pellet) radioligand fractions were then counted in a Beckman y-counter.

4 H.T. LANGEN AND J.W. TAYLOR

Rat Brain Receptor Binding Assays Cerebral cortex tissue isolated from male Sprague-

Dawley rats weighing 200-300 g was homogenized in 10 volumes of ice-cold 10 mM HEPES buffer, pH 7.6, using a Polytron homogenizer. The homogenate was centrifuged at 600g (at raJ for 5 min at 4"C, then the supernatant was centrifuged for 30 min at 30,OOOg and the resulting pellet was resuspended in 10 volumes of the HEPES buffer. The second cen- trifugation and resuspension was repeated once. For binding studies, rat cortical membranes (corre- sponding to approximately 50 pg protein by Bio-Rad protein assay) were incubated in a total volume of 400 p1 at 25°C for 1 hr with approximately 20,000 cpm radioligand ([ 1251-Tyr11]-somatostatin) and in- creasing amounts of unlabeled peptide or hybrid protein in 50 mM HEPES, pH 7.4, 5 mM MgCl,, 2 mg/ml BSA, 20 pg/ml PMSF, and 10 pg/ml bacitra- cin. "he incubations were performed in triplicate and were terminated by centrifugation (10,OOOg for 5 min). The pellets were counted in a Beckman y- counter. Saturable binding was taken to be total binding minus binding in the presence of 1.0 pM somatostatin-14. Nonsaturable binding was 44% of the total.

Protein Cationization Cationization was performed by protein reaction

with ethylene diamine and carbodiimide according Kumagai et aLZ6 The pH of the reaction mixture was 6.2. After dialyzing the reaction mixtures, the cationized AP (CAP) and cationized SAP (cSAP) were bound to a cation exchange column (Whatman CM52) and eluted with 0.5 M NaC1. Both the CAP and the cSAP showed no decrease in phosphatase activity, assayed as described above. Titration of the primary amines with fluorescamine showed a 1.74- fold increase in the fluorescent response for CAP and a 1.58-fold increase for cSAP, compared to wild-type AP, indicating that both proteins had been modified to approximately the same extend.

RESULTS AND DISCUSSION Hybrid Protein Design and Preparation

The molecular surface of alkaline phosphatase was scanned using a computer model representation of the X-ray for a surface-accessible loop that would be appropriate for replacement by a partial somatostatin sequence. The loop was se- lected according to the following criteria: (1) inser- tion of the appropriate somatostatin sequence would not significantly perturb the rest of the structure of the carrier protein, (2) the putative active conforma- tion of the hormone, with the Phe-Trp-Lys-Thr se- quence in a p-turn conformation connecting two antiparallel strands of a could be re- tained upon insertion, and (3) the Phe-Trp-Lys-Thr sequence of the somatostatin insert might be pre-

sented in a favorably accessible orientation for in- teractions with other macromolecules in the result- ant hybrid structure. According to these criteria, the AP surface loop corresponding to amino acids AP(92-94) were selected for substitution by the hor- mone-derived structure.

A cyclic somatostatin structure was then de- signed for substitution at the AP site selected. This structure consisted of somatostatin residues 3 -14, the disulfide-bridged active segment of the hor- mone, extended by a tripeptide sequence at both its N- and its C-terminal ends (Fig. 1B). The tripeptide sequences used to extend the somatostatin sequence were selected from extended loops on the surface of globular proteins of known crystal structure that were in the antiparallel p-sheet conformation, in an attempt to favor the same conformation in the inserted loop of the hybrid SAP protein. The exact sequence chosen was derived from actinidin(l70- 177),27 with the central p-turn residues Gly-173 and Gly-174 deleted, so that the tripeptide se- quences used were Gly-Thr-Glu for the N-terminal extension and Val-Asp-Tyr for the C-terminal extension. These sequences were selected from several candidate sequences, in part because they were unrelated to the native somatostatin sequence and predominantly hydrophilic. However, their selection was additionally constrained by consider- ation of the genetic manipulations required to allow construction of a gene coding for the hybrid protein by cassette mutagenesis of a suitable plasmid vector (Fig. 3 and Methods).

By computer modeling, the AP(92-94) loop structure was cut out. Somatostatin-14 residues 3-14 were then arranged to be in the antiparallel p-sheet conformation" about a central p-turn consisting of the Phe-Trp-Lys-Thr sequence pro- posed as the minimum essential receptor-binding epit~pe.~-" The actinidin extension sequences described above were then added in their native actinidin backbone conformation^,^^ with the excep- tion that the "unusual" cp = -166" value observed for Gly-170 was adjusted to the standard p-sheet value of cp = 120". This structure, illustrated in Figure lB, was then docked onto the opened AP loop and connected to AP residues 91 and 95 with trans peptide bonds, after making minor adjust- ments to the +,cp angles of the modeled somatosta- tin-containing loop that did not significantly alter its antiparallel p-sheet conformation. The final modeled structure (Fig. 2) indicated a high poten- tial accessibility for the receptor-binding somatosta- tin epitope, which is extended almost 30 A from the surface of the AP dimer by the antiparallel p-strands. In comparison, the dimensions of AP itself are 97 x 47 x 52 A.12

A gene coding for the chimeric protein was con- structed from phoA in two steps (Fig. 3). Vector PHLlOl was first constructed via oligonucleotide-

HYBRID PROTEINS DESIGNED AS RECEPTOR PROBES 5

Fig. 2. Space-filling model of the alkaline phosphatase dimer13 with loop residues 92-94 of one monomer (green) sub- stituted by the extended somatostatin sequence from Figure 18,

directed mutagenesis of the phoA gene16 in M13 mp 18.15 A frameshift and restriction sites for SuZI and KpnI were introduced during this mutagenesis. Cor- rect ligation of the complementary oligonucleotides coding for the somatostatin sequence into these new restriction sites then removed this frameshift and led to the expression of a mature fusion protein, de- tected as blue colonies on an AP indicator plate. By SDS polyacrylamide gel electrophoresis, we deter- mined that the somatostatin-alkaline phosphatase hybrid (SAP) has the expected molecular weight and was correctly transported to the periplasm and cleaved from its N-terminal signal peptide (not shown). The correct compartmentalization of SAP had the advantage that we could use the standard procedure for purification already worked out for the wild-type carrier protein.6 The yields and purity of wild-type AP and the hybrid protein were similar and were approximately 20 mg purified protein per 1 liter culture.

as described in the text. The native loop structure (red) in the other monomer (yellow) is partially hidden. In SAP, the extended so- matostatin loop is substituted into both monomer units.

Structural Characterization

SAP showed phosphatase activities similar to wild-type AP (Table I), indicating that the host pro- tein structure had not been significantly disturbed by the introduction of the 18-residue extended so- matostatin sequence in place of surface loop residues 92-94. The CD spectra of the wild-type and mutant proteins were also essentially identical (not shown), demonstrating that their secondary structure con- tents are very similar. Titration of SAP with Ell- man's reagent" revealed no free cysteines, indicat- ing that the desired disulfide bridge in the somatostatin insert had probably been formed, al- though this assay was performed without denatur- ant and the presence of inaccessible cysteine resi- dues in the folded structure cannot be ruled out.

As an additional criterion for assessing the con- formational integrity of the AP host structure in SAP, its susceptibility to proteolysis was compared

6 H.T. LANGEN AND J.W. TAYLOR

H i n d 111 alkaline phosphatase gene

oligonucleotide-directed mutagenesis in M13

in plasmid

Step 1

Step 2

5 ' - C GAG TGC AAG AAC TTC TTC T G U G ACC ACC TCG TGC G - 3' 3 ' - CA TGG CTC ACG TTC TTG AAG AAG AC-C TGG BBG TGG AGC ACG CAG CT - 5'

XnnI (-Gly-Thr-Glu-Cys-Lys-Asn-Phe-Phe-Trp-Lys-Thr-Phe-Thr-Ser-Cys-Val-Asp-Tyr-)

Fig. 3. Engineering of the AP gene to generate the SAP gene via a vector generally suitable for the preparation of hybrid pro- teins by cassette mutagenesis. Step 1 was performed in M13 rnpl8 by oligonucleotide-directed mutagenesis using the high-

TABLE I. Kinetic Parameters of AP and SAP*

V,, (M s-l) K, (MI k,,, (s-? AP 2.72 x 10-7 1.23 x 10-5 53.5 SAP 2.71 x 10-7 2.12 x 10-5 56.4 *Kinetic assays were performed at 25°C in 1.0 M Tris-HC1, 10 mM MgCl,, 0.5 mM ZnCl,, pH 8.0, with p-nitrophenyl phos- phate as substrate.

to that of wild-type AP. Wild-type AP was resistant to cleavage with high amounts of trypsin (lo%, w/ w), as reported previously. Cleavage of only the Arg- 10-Ala-11 bond after 24 hr incubation at 37°C was apparent by microsequencing and analysis on SDS polyacrylamide gels, as reported p r e v i o u ~ l y . ~ ~ After trypsin digestion of the fusion protein under identi- cal conditions, we observed two bands on a SDS poly- acrylamide gel corresponding to the apparent molec- ular weight of 41 and 11 kDa. This indicates that trypsin cleavage of the hybrid protein had occurred in or near to the region of the peptide chain corre- sponding to the guest peptide insert, and that the nicked product was resistant to further degradation. Both wild-type AP and the nicked SAP retained 80% of their phosphatase activities after trypsin diges- tion. These results, therefore, confirmed the confor- mational integrity of the AP host structure in SAP and the overall stability of the hybrid protein. The surface accessibility of the somatostatin insert, which contains two lysine residues that are poten- tial sites for trypsin cleavage, is also indicated by this result. However, sequence analysis of the nicked SAP to confirm that the new trypsin-acces-

efficiency phosphorothioate method.20t21 After subcloning into pBR 322 (-Accl) to obtain cassette mutagenesis vector pHLlOl, step 2 was performed by cassette mutagenesis to create pHL5 having a restored reading frame in the gene now coding for SAP.

sible site is indeed located in this insert has not been performed.

Recognition by Polyclonal Antibodies Polyclonal antibodies raised against somato-

statin-14 were able to bind the alkaline phos- phatase-somatostatin chimeric protein (Fig. 4). In competitive, solution-phase binding assays, SAP displaced [1251-Tyr11]-somatostatin-14 binding to antibodies a t concentrations of the dimeric hybrid protein only about 2- to %fold higher than soma- tostatin-14 itself did. In contrast, no antibody bind- ing by wild-type AP was observed, even at much higher concentrations. This result confirmed the surface location and accessibility of the somatosta- tin sequence in SAP, and provided further evidence that this inserted peptide loop may be unhindered by the AP structure in its binding interactions with other proteins.

Brain Membrane Receptor Binding Binding of SAP to somatostatin receptors in brain

membranes was investigated in competitive binding assays using rat cerebral cortex membrane prepara- tions. In these experiments, the fusion protein was able to displace membrane-bound [1251-Tyrl-so- matostatin-14 (Fig. 5), but its potency was approxi- mately 8,000-fold lower than that of somatostatin-

+In separate experiments (unpublished), a Glu residue has been introduced into the loop of residues between the cysteines of the somatostatin insert in SAP, and nicking of the resultant AP structure a t this site by Staphyloccocus aureus V8 protease has been confirmed by sequence analysis. Both wild-type AP and SAP are resistant to V8 protease digestion as evidenced by electrophoretic analysis in denaturing gels.

HYBRID PROTEINS DESIGNED AS RECEPTOR PROBES 7

60:

50 -

40 -

30 -

20 -

10 -

0 %

h

8 v

TI C m m 0 TI (II

TI C

0

.- -

.- a

a m

\

’ I ’ I . I . b ’ 4

Fig. 4. Binding of proteins to anti-somatostatin-1 4 polyclonal antibodies. Percent of total [125-l-Tyr1i]-somatostatin in the incu- bation mixture that is bound to antisomatostatin polyclonal anti- bodies is plotted as a function of the concentration of the com- peting ligands somatostatin-14 (o), SAP (a), and wild-type AP (1 All error bars are shown and indicate standard deviations in total radioligand bound for each triplicate measurement.

I ..

h

8 Y

m E TI c .-

E 0 c 0 Q) P v)

.-

-20 -1 1 .o -9.0 -7.0 -5.0

Log{[Protein](M)}

Fig. 5. Binding of proteins to somatostatin receptors in rat brain membranes: Percent specific [125-I-Tyr1i]-~~mato~tatin binding is plotted as a function of the concentration of the following competing ligands: somatostatin-1 4 (0); equimolar concentrations of somatostatin-14 and CAP dimer (0); SAP dimer (0); cSAP dimer (m); AP dimer (A); CAP dimer (A). All error bars are shown and indicate the standard deviations in % specific binding for each triplicate measurement.

14. Furthermore, wild-type AP was also able to displace radioligand binding in the same high con- centration range required for SAP. Since the bind- ing assays are performed in the presence of BSA at 2 mg/ml, this suggests that the host protein we have chosen, AP, has some specificity for [ 1251-Tyr]-so- matostatin-14 displacement from brain membranes that is independent of the somatostatin insert. This activity might arise from weak interactions of cer- tain unidentified surface structures unique to this

protein with the receptor surface. However, later ex- periments (see below) suggest that the decrease in radioligand binding at high AP concentrations may also represent artefactual radioligand displacement from sites in the membrane preparations that are defined as “nonsaturable” by somatostatin displace- ment.

Initially, these results suggested to us that the binding site for somatostatin on its brain membrane receptors was inaccessible to the receptor binding sequence in SAP, either due to steric hindrence by the host AP structure, or because the inserted so- matostatin sequence was constrained in a conforma- tion entirely incompatible with receptor recognition. However, the small (%fold), but reproducible differ- ence in potency between AP and SAP indicated that the inserted somatostatin structure was still able to make a small contribution to radioligand displace- ment, and that engineering the somatostatin se- quence into the AP surface did not block receptor recognition by the hormone sequence entirely.

We next investigated, through chemical modifica- tion, the possibility that electrostatic forces may be inhibiting recognition of the somatostatin binding site by the hybrid protein. AP and SAP were reacted under identical conditions with ethylene diamine and a water-soluble carbodiimide,26 effectively con- verting negatively charged carboxylic acid side chains to positively charged amines. These “cation- ized” proteins were purified as relatively sharp elu- tion peaks by cation-exchange chromatography, and retained full phosphatase activity. Fluorescarnine assay of the surface-accessible amines indicated that the cationized wild-type AP and the hybrid SAP gave fluorescent signals 1.74 times and 1.58 times that of unmodified wild-type AP, respectively. This demonstrated that both proteins had been modified to a similar extent, corresponding to conversion of approximately 20 carboxylic acid side chains to amines per molecule. The locations and distribu- tions of the sites of modification have not, however, been further characterized.

The cationized proteins were then tested in the receptor binding assay (Fig. 5). Although the un- characterized weak specificity of wild-type AP for radioligand displacement in this assay was slightly enhanced in cationized wild-type AP (CAP), the cat- ionized hybrid protein (cSAP) was now 280 times more potent in displacing membrane-bound radioli- gand than CAP, demonstrating a significant specific interaction of the receptor with the somatostatin in- sert in cSAP. At high concentrations, cSAP dis- placed more membrane-bound radioligand than 1.0 FM somatostatin did, indicating that radioligand was being displaced from both the saturable sites and the nonsaturable sites in the membrane prepa- ration. We also observed this effect using equimolar mixtures of CAP dimer and somatostatin-14. How- ever, the displacement curves for the CAP + so-

8 H.T. LANGEN AND J.W. TAYLOR

matostatin-14 mixture and for somatostatin-14 alone were essentially superimposable up to a t least 20 nM (Fig. 5), a concentration equivalent to the cSAP dimer concentration causing 80% dis- placement of the saturable radioligand binding. We conclude, therefore, that displacement of the nonsat- urable radioligand binding by cSAP does not signif- icantly affect comparisons of receptor binding poten- cies of somatostatin-14 and cSAP at the level of 50% displacement of saturable radioligand binding (IC50 values). These IC,, values were determined from linear regression analysis of log-probit plots to be 0.26 and 6.0 nM, respectively, indicating that cSAP has a receptor binding affinity just 23-fold lower than somatostatin-14.

The relative potencies of SAP in the receptor bind- ing assay before and after cationization are most easily explained in terms of unfavorable interac- tions between the electrostatic fields around SAP and the receptor surface. These unfavorable inter- actions may take the form of a direct repulsive force, since it is well established that membrane surfaces have a net negative ~ h a r g e ~ ' * ~ l and SAP has a net charge of -5 at neutral pH and behaves as an an- ionic macromolecule on ion-exchange resins. In ad- dition, the somatostatin receptor from rat pancreas is an acidic protein with a pZ of approximately 4.8.' Alternatively, such unfavorable electrostatic inter- actions might be attractive in nature, but inhibit the SAP dimer from approaching the hormone binding site in the correct orientation for binding.

Specific differences between wild-type AP and SAP in the sites subjected to the cationization mod- ification could also explain the apparent insert- specific component of receptor binding by cSAP. Similarly, specific conformational changes in the so- matostatin loop as a result of the cationization re- action could explain the enhanced receptor binding potency of cSAP compared to SAP, either directly or through enhanced degradation of cSAP by proteases present in the membrane preparations. However, our preliminary studies of two additional cationized SAP mutants have demonstrated that receptor bind- ing potency decreases with decreasing length of the inserted peptide sequence, even when the somato- statin-related segment of the insert has been kept constant.* This argues against the importance of

*Two additional AP mutants have now been prepared that have AP residues 92-94 substituted with progressively shorter peptide sequences consisting of -Gly-Thr-Glu-Lys-Asn-Phe- Phe-Trp-Lys-Thr-Phe-Thr-Ser-Val-Asp-fir- and -Am-Phe- Phe-Trp-Lys-Thr-Phe-Thr-Ser-. After cationization and purifi- cation as described for SAP, fluorescamine derivatization gave fluorescence responses 1.99-fold and 1.76-fold higher than wild-type AP for these hybrids, respectively. The IC,, concen- trations of these cationized hybrid proteins in the rat brain membrane receptor binding assay were 12 and 130 nM, corre- sponding to potencies approximately 2-fold and 22-fold lower than cSAP, and approximately 140-fold and 13-fold higher than CAP, respectively.

specific differences in cationization or loop confor- mation, and in favor of the importance of long-range electrostatic interactions between SAP and the so- matostatin receptors. A more detailed exploration of these effects through multiple, residue-specific mu- tagenesis of the AP host structure of SAP is also now underway.

Future use of an alternative, monomeric host pro- tein in place of AP will allow us to evaluate the importance of the dimeric structure of SAP in deter- mining its receptor binding potency and other po- tential pharmacological properties. At present, the ability of cSAP to cause receptor activation and sig- nal transduction is unknown. It will be interesting to determine whether such globular proteidpeptide ligand hybrids have the potential to block or pro- mote receptor activation (acting as antagonists or superagonists, respectively), particularly in systems where receptor dimerization is involved in the sig- nal transduction pathway.

CONCLUSIONS AP appears to be an excellent host protein for the

guest peptide insertion approach to designing probes for the ligand binding sites of receptor surfaces. The AP mutant, SAP, provides a starting point for mea- suring both the steric accessibility and the electro- static environment of somatostatin-14 binding sites in rat brain membranes. It is already apparent that the binding sites for somatostatin-14 on its cell sur- face receptors in rat brain are approximately as ac- cessible for binding by cSAP as the active site of trypsin and the peptide binding sites on antiso- matostatin polyclonal antibodies are to SAP. This accessibility is somewhat surprising, since the cell surface receptors are only required to bind a rela- tively small, flexible ligand, and might be expected to have ligand binding sites that are inaccessible to molecules as large as cSAP. In contrast, the ligand- binding sites of antibodies and proteolytic enzymes such as trypsin have probably evolved specifically to allow binding to the surface regions of such globular proteins.

By exploring the effects of altering the length of the hormone insert and the surface potential of the host protein through cassette and oligonucleotide- directed mutagenesis, it should be possible to rede- sign SAP to have a high affinity for somatostatin receptors without the need for chemical modifica- tion, and thereby obtain more detailed structural information concerning somatostatin receptors. The properties of SAP also suggest additional applica- tions of the host-guest hybrid protein design prin- ciple.6 Engineering small bioactive peptides such as somatostatin, that are active in the central nervous system, onto the surface of globular proteins that are transported across the blood-brain barrier might prove to be an effective approach to design- ing new therapeutic agents. Alternatively, prop-

HYBRID PROTEINS DESIGNED AS RECEPTOR PROBES 9

erties of the host protein, such as the phosphatase activity of AP, might be exploited in the design of sensitive markers for identification or visualization of receptors for the guest peptide ligand. The re- quirement for cationization in order to obtain high affinity binding of SAP to somatostatin-14 receptors indicates, however, that characteristics of the host protein can dramatically affect guest ligand inter- actions with receptor macromolecules. Clearly, these effects must be defined or otherwise taken into consideration. For example, our results suggest that the use of phage to search for short peptide sequences that recognize cell-surface recep- tors may be significantly compromised by the char- acteristics of the particular phage protein host that is chosen, or even the exact site on that protein that is selected for peptide insertion.

ACKNOWLEDGMENTS This research was supported by a Postdoctoral

Fellowship from the Swiss National Science Foun- dation awarded to H.T.L., and by USPHS Grant DA04197.

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11

REFERENCES Stroud, R.M., McCarthy, M.P., Shuster, M. Nicotinic ace- tylcholine receptor superfamily of ligand-gated ion chan- nels. Biochemistry 29:11009-11023, 1990. Esser, V., Limbird, L.E., Brown, M.S., Goldstein, J.L., Rus- sell, D.W. Mutational analysis of the ligand binding do- main of the low density lipoprotein receptor. J . Biol. Chem. 263:13282-13290, 1988. Schwyzer, R. Molecular mechanism of opioid receptor se- lection. Biochemistry 25:6335-6342, 1986. Carraway, K.L., 111, Koland, J.G., Cerione, R.A. Location of the epidermal growth factor binding site on the EGF receptor. A resonance energy transfer study. Biochemistry

Sklar, L.A., Fay, S.P., Seligmann, B.E., Freer, R.J., Mu- thukumaraswamy, N., Mueller, H. Fluorescent analysis of the size of a binding pocket of a peptide receptor a t natural abundance. Biochemistry 29:313-316, 1990. Freimuth, P.I., Taylor, J.W., Kaiser, E.T. Introduction of guest peptides into Escherichia coli alkaline phosphatase. J . Biol. Chem. 265896-901, 1990. Reichlin, S. Somatostatin. Part I. N. Engl. J. Med. 309:

Knuhtsen, S., Esteve, J.-P., Cambillau, C., Colas, B., Susini, C., Vaysse, N. Solubilization and characterization of active somatostatin receptors from rat pancreas. J . Biol. Chem. 265:1129-1133, 1990. Veber, D.F., Freidinger, R.M., Schwenk-Perlow, D., Pa- leveda, W.J., Holly, F.W., Strachan, R.G., Nutt, R.F., Ari- son, B.H., Homnick, C., Randall W.C., Glitzer, M.S. Saper- stein, R., Hirschmann, R. A potent cyclic hexapeptide analogue of somatostatin. Nature (London) 29255-58, 1981. Momany, F.A. Conformational energy studies of the growth hormone inhibitor, cyclo(Aha-Cys-Phe-D-Trp-Lys- Thr-Cys). Biochem. Biophys. Res. Commun. 95:61-66, 1980. Knappenberg, M., Michel, A., Scarso, A,, Brison, J., Zanen, J., Hallenga, K., Deschrijver, P., Van Binst, G. The con- formational properties of somatostatin. IV. The conformers contributing to the conformational equilibrium of so- matostatin in aqueous solution as found by semi-empirical energy calculations and high-resolution NMR experi- ments. Biochim. Biophys. Acta 700:229-246, 1982.

29~8741-8747, 1990.

1495-1563, 1983.

12. Sowadski, J.M., Foster, B.A., Wyckoff, H.W. Refined struc- ture of alkaline phosphatase from E. coli a t 2.8 A resolu- tion. J. Mol. Biol. 150245-272, 1981.

13. Kim, E.E., Wyckoff, H.W. Structure of alkaline phos- phatases. Clin. Chem. Acta 186175-188, 1989.

14. Kim, E.E., Wyckoff, H.W. Reaction mechanism of alkaline phosphatase based on crystal structures. Two-metal ion catalysis. J. Mol. Biol. 218449-464, 1991.

15. Maniatis, T., Fritsch, E.F., Sambrook, J . “Molecular Clon- ing, “ 2nd ed. Vols. 1-111. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 1989.

16. Inouye, H., Michaelis, S., Wright, A., Beckwith, J . Cloning and restriction mapping of the alkaline phosphatase gene (pho A) of Escherichia coli and generation of deletion mu- tants in vitro. J. Bacteriol. 146668-675, 1981.

17. Sonveaux, E. The organic chemistry underlying DNA syn- thesis. Bioorg. Chem. 14274-325, 1986.

18. Maxam, A,, Gilbert, W. Sequencing end-labeled DNA with base-specific chemical cleavages. Methods Enzymol. 65: 499-560,1980.

19. Sanser. F.. Nicklen. S.. Coulson. A.R. DNA seauencine with0 chainkmninating’inhibitors. Proc. Natl. Aiad. ScL U.S.A. 74:5463-5467, 1977.

20. Taylor, J.W., Ott, J., Eckstein, F. The rapid generation of oligonucleotide-directed mutations at high frequency us- ing phosphorothioate-modified DNA. Nucleic Acids Res. 13:8765-8785, 1985.

21. Nakamaya, K.L., Eckstein, F. Inhibition of restriction en- donuclease Nci I cleavage by phosphorothioate groups and its application to oligonucleotide-directed mutagenesis. Nucleic Acids Res. 14:9679-9698, 1986.

22. Neidhardt, F.C., Bloch, P.L., Smith, D.F. Culture medium for enterobacteria. J. Bacteriol. 119:736-747, 1974.

23. Brickman, E., and Beckwith, J . Analysis of the regulation of Escherichia coli alkaline phosphatase synthesis using deletions and $80 transducing phages. J. Mol. Biol. 96:

24. Malamy, M.H., Horecker, B.L. Purification and crystalli- zation of the alkaline phosphatase of Escherichiu coli. Bio- chemistry 3:1893-1897, 1964.

25. Edelhoch, H. Spectroscopic determination of tryptophan and tyrosine in proteins. Biochemistry 6:1948-1954,1967.

26. Kumagai, A.K., Eisenberg, J.B., Partridge, W.M. Absorp- tive-mediated endocytosis of cationized albumin and a p- endorphin-cationized albumin chimeric peptide by isolated brain capillaries. J . Biol. Chem. 262:15214-15219, 1987.

27. Baker, E.N. Structure of actinidin, after refinement at 1.7 A resolution. J . Mol. Biol. 141:441-484, 1980.

28. Ellman, G.L. A colorimetric method for determining low concentrations of mercaptans. Arch. Biochem. Biophys. 74 443-450,1958.

29. Roberts, C.H. Chlebowski, J.F. Trypsin modification of Es- cherichiu coli alkaline phosphatase. J . Biol. Chem. 259: 729-733,1984.

30. London, Y., Vossenberg, F.G.A. Specific interaction of cen- tral nervous system myelin basic protein with lipids. Spe- cific regions of the protein sequence protected from the proteolytic action of trypsin. Biochim. Biophys. Acta 307: 478-490,1973.

31. Verkleij, A.J., Zwaal, R.F.A., Roelofsen, B., Comfurius, P., Kastelijn, D., Van Deenen, L.L.M. The asymmetric distri- bution of phospholipids in the human red cell membrane. A combined study using phospholipases and freeze-etch electron microscopy. Biochim. Biophys. Acta 323:178-193, 1973.

32. Scott, J.K., Smith, G.P. Searching for peptide ligands with an epitope library. Science 249:386-390, 1990.

33. Devlin, J.J., Panganiban, L.C., Devlin, P.E. Random pep- tide libraries: A source of specific protein binding mole- cules. Science 249:404-406, 1990.

34. Cwirla, S.E., Peters, E.A., Barrettt, R.W., Dower, W.J. Peptides on phage: A vast library of peptides for identify- ing ligands. Proc. Natl. Acad. Sci. U.S.A. 875378-6382, 1990.

35. Bass, S., Greene, R., Wells, J.A. Hormone phage: An en- richment method for variant proteins with altered binding properties. Proteins: Struct. Funct. Genet. 8309-314, 1990.

307-316,1975.