Proc. Natl. Acad. Sci. USAVol. 84, pp. 2532-2536, April 1987Neurobiology

Molecular structure of mammalian neuropeptide Y: Analysis bymolecular cloning and computer-aided comparison with crystalstructure of avian homologue

(avian pancreatic polypeptide/molecular modeling)

JANET ALLENt, JIM NOVOTNOt, JOSEPH MARTINt, AND GERHARD HEINRICH*Howard Hughes Medical Institute, Departments of *Medicine and tNeurology and *Molecular and Cellular Research Laboratory, Laboratory for MolecularEndocrinology, Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114

Communicated by Charles F. Stevens, December 30, 1986

ABSTRACT Identification and characterization of thecDNA encoding rat neuropeptide Y revealed the nucleotidesequence coding for a 98-amino acid precursor. The deducedamino acid sequence for rat neuropeptide Y is identical to thehuman peptide and is highly homologous to avian pancreaticpolypeptide. The tertiary structure of avian pancreatic poly-peptide has been previously derived from crystallographic databy Blundell and coworkers. The homology between neuropep-tide Y and avian pancreatic polypeptide preserves all of theresidues essential for the maintenance of the tertiary structure.Thus, it has been possible to compute a three-dimensionalmodel of the mammalian neuropeptide, neuropeptide Y, basedon the known structure of the avian homologue. This modelsuggests that neuropeptide preserves a compact tertiary struc-ture characterized by extensive hydrophobic interactions be-tween an N-terminal polyproline-II-like helix and a C-terminala-helix. The model has been used to identify amino acidsresiding in key positions within this structure and, thereby, todirect future analysis of neuropeptide Y structure-functionrelationships.

Neurons produce a variety of peptides that are secreted fromaxons and dendrites and are collectively known as neuro-peptides. These peptides arise from larger precursors that areprocessed enzymatically to yield the mature neuropeptides(1). Neuropeptides are believed to serve as neurotransmittersand neuromodulators (2) by their interaction with specific cellmembrane receptors after secretion and have been implicatedin the control of behavior and autonomic and motor func-tions.

Understanding the synthesis and secretion of neuropep-tides and their mechanism of action requires detailed knowl-edge of the molecular structure of neuropeptides and neuro-peptide precursors. To gain such knowledge, we have com-bined the methods of molecular cloning of cDNAs comple-mentary to specific mRNAs with computer-aided molecularmodeling to deduce both the primary and tertiary structuresof the mammalian neuropeptide "neuropeptide Y" (NPY).This neuropeptide was originally identified by chemicalmeans and subsequently purified from porcine brain (3, 4).NPY is widely distributed throughout the mammalian central(5, 6) and peripheral nervous systems (7, 8). Central admin-istration has implicated NPY in the control of feeding (9, 10)and in secretion of gonadotrophin-releasing hormone secre-tion (11). Peripheral administration of NPY induces vaso-constriction in many vascular beds (12-14) and potentiatescatecholamine-induced vascular smooth muscle contractionin vitro (15). NPY may play a role in degenerative diseasessuch as Alzheimer's disease (16). The structure of the

precursor of human NPY was recently determined by mo-lecular cloning methods (17).NPY is a member of a larger family that includes the

pancreatic polypeptide (18) and peptide YY (19) (Table 1). Inparticular, NPY is highly homologous in amino acid sequenceto avian pancreatic polypeptide (APP), these two having 19of the total 36 amino acids in common (4). The three-dimen-sional structure of APP was recently solved by x-ray crys-tallography at high resolution (20).We deduced the amino acid sequence of rat NPY from the

nucleotide sequence of the mRNA that encodes the rat NPYprecursor. We then took advantage of the extensive homol-ogy between NPY and APP to construct an approximatethree-dimensional model of rat NPY using the conformation-al search program CONGEN (21).

MATERIALS AND METHODSIdentification, Subcloning, and Nucleotide Sequence Anal-

ysis of a cDNA Encoding Rat NPY mRNA. A cDNA libraryconstructed in bacteriophage Xgtll from rat hypothalamicmRNA template (R. H. Goodman, Tufts University Schoolof Medicine, Boston) was screened by the plaque-hybridiza-tion method of Benton and Davis (22).A double-stranded oligodeoxynucleotide of 60 base pairs

corresponding to nucleotides 228-287 of a cDNA encodinghumanNPY (17) was used as hybridization probe. 32P-labeleddouble-stranded DNA was prepared enzymatically withEscherichia coli DNA polymerase I (Klenow fragment) fromtwo oligonucleotides (33 and 34 nucleotides) synthesized bythe phosphoramidite method (23) so that the 3' end of theshorter oligonucleotide was complementary to 7 bases at the3' end of the longer one.Plaque hybridization with heat-denatured 32P-labeled dou-

ble-stranded DNA was carried out in 1 M NaCl at 55°C.Phages from specifically hybridizing plaques were purified,and DNA was extracted from pure phages and digested withrestriction endonuclease EcoRI to release cloned cDNA fromthe unique cloning site (EcoRI) of the vector. The cDNAinsert was then ligated into the EcoRI site of the vectorpGem3 (Promega Biotec, Madison, WI), transfected into E.coli c600, and prepared in bulk for mapping of restrictionendonuclease sites. DNA at the sites indicated in Fig. 1 waslabeled with 32P at the 5' or 3' ends by using polynucleotidekinase and the Klenow fragment of E. coli DNA polymeraseI, and the DNA was digested with a second enzyme or thestrands were separated to prepare fragments labeled on asingle strand. The nucleotide sequences of the resulting DNAfragments were then determined by the chemical method of

Abbreviations: NPY, neuropeptide Y; APP, avian pancreatic poly-peptide.

2532

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 84 (1987) 2533

Table 1. Amino acid sequences (in one letter notation) of APP, rat and porcine NPY, peptide YY (PYY), and bovinepancreatic polypeptide (BPP)

APP G P S Q P T Y P G D D A P V E D L I R F Y D N L Q Q Y L N V V T R H RyaRat NPY Y P S K P D N P G E D A P A E D M A R Y Y S A L R H Y I N L I T R Q RyaPorcine NPY Y P S K P D N P G E D A P A E D L A R Y Y S A L R H Y I N L I T R Q RyaPYY Y P A K P E A P G E D A S P E E L S R Y Y A S L R H Y L N L V T R Q RyaBPP A P L E P E Y P G D N A T P E Q M A Q Y A A E L R R Y I N M L T R P Rya

Maxam and Gilbert (24) according to the strategy outlined inFig. 1.

Computational Methods. Crystallographic coordinates ofAPP (20) were obtained from the Brookhaven Protein DataBank (25). Structural manipulations and potential energyevaluations were performed with the program CHARMMversion 16 (26) in explicit hydrogen atom representation. Thatis, the polar hydrogen atoms were explicitly constructed bythe program, while nonpolar hydrogens were combined withthe carbon atoms into "extended" atoms. Conformationalspace of side chains was searched by using the procedureCONGEN (21) with a 30° grid applied to all torsional degreesof freedom with the exception of lysine and arginine sidechains, where the torsional angles were chosen according tothe minimum potential energy criterion. The constrainedenergy minimization protocol was that of Bruccoleri andKarplus (27). Solvent accessibility was computed by usingthe Lee and Richards algorithm (28). Electrostatic energy ofboth the APP and NPY structures was computed from theCoulomb formula with the dielectric constant equal to 50 andwas evaluated to infinity (29). Color-coded molecular graph-ics were produced with the program FRODO as implementedon Evans and Sutherland picture systems 340 (30) andmodified by M. Handschumacher (Massachusetts GeneralHospital, Boston).

RESULTSCharacterization and Nucleotide Sequence Analysis of NPY

cDNAs. At least 8 independent recombinant bacteriophagesof 100,000 screened hybridized specifically to the synthetichybridization probe. Five of these were purified, and DNAwas prepared and analyzed by restriction enzyme digestionand Southern blot hybridization. These analyses revealedthat in 3 phages the EcoRI cloning sites had becomenonfunctional. Therefore, the rat DNA was released from theremaining 2 phages with EcoRI and cloned into the uniqueEcoRI site of the cloning vector pGem3. These subcloneswere designated NPY-5 and NPY-7. Complete maps ofrestriction endonuclease sites of the rat DNA inserts ofNPY-5/7 were established by polyacrylamide gel electropho-resis of digests of NPY-5/7 DNA and by blot hybridizationwith the synthetic oligonucleotides constructed according tothe rat NPY mRNA sequence. The restriction site map of theNPY-5 insert is shown in Fig. 1 and was identical for NPY-7.The nucleotide sequence of the NPY-5 insert was deter-

mined according to the strategy outlined in Fig. 1. The

- IC col -

i< lo1 I 200 I r30O 400 1I50 '

FIG. 1. The restriction map of NPY-5 and NPY-7. The arrowsshow the DNA sequencing strategy. An open circle at the end of anarrow indicates a 3' labeled fragment, and a filled circle at the end ofan arrow denotes a fragment labeled at its 5' end.

complete nucleotide sequences of both strands agreed withone another and are shown in Fig. 2.The nucleotide sequence of the NPY-5 insert revealed a

high degree of homology with the reported human NPYmRNA sequence. Therefore, nucleotides 1-68 of NPY-5must represent the 5' untranslated region of rat NPY mRNA;nucleotides 69-365, the translated region; and nucleotides366-539, the 3' untranslated region. The presence of 20adenosine residues at the 3' end preceded by a typical

caagctcattcctcgcagaggcg .. cccagagcagagcacccgcaccccatccgctggctcacccctcggagacgctcgcccgacagcatagtacttgc

-29Met Met Lou Gly Asn Lys Arg

tgcgcagagaccacagcccgcccgcc ATG ATG CTA GGT AAC AAA CGAcgcccagccacgccccgccagcc c-- --- --G ---

______0____signal peptide-20

Met Gly Lou Cys Gly Leu Thr Lou Ala Lou SerATG GGG CTG TGT GGA CTC ACC CTC CCT CTA TCCC-- --- --- -CC --- --- --- --- --C --G ---Leu Ser

-10Lou Lou ValCTG CTC GTG

- ---__

-1 1

Cys Lou Gly Ile Lou Ala Glu Gly Tyr Pro Ser Lys Pro AspTGT TTG GGC ATT CTG GCT GAG CGG TAC CCC TCC AAG CCG GAC--C C-- --T GCG --- --C --- -C- --- --- --- --- --- ---

Ala Ala

10Asn Pro Gly Glu Asp Ala Pro Ala GluAAT CCG GCC GAG GAC GCG CCA GCA GAG--C --- --- --- --- --A --- --G ---

_ _Neuropept1de Y

20Asp Met Ala Arg TyrGAC ATG GCC AGA TAC

- - - - - - - -

30Tyr Ser Ala Leu Arg His Tyr Ile Asn Leu Ile Thr Arg GlnTAC TCC GCT CTG CGA CAC TAC ATC AAT CTC ATC ACC AGA CAG--- --G --G --- --- --- --- --- --C --- --- --- --C ---

42

89

131

173

215

257

*~~~~~~~~- 40Arg Tyr Cly Lys Arg Ser Ser Pro Glu Thr Leu Ile Ser AspAGA TAT CGC AAG AGA TCC AGC CCT GAG ACA CTG ATT TCA GAT 299___ _-- --A --A C-- --- --- --A --- --- --- --- --- --C

50Lou Lou Met Arg Glu Ser Thr Glu AsnCTC TTA ATG AGA GAA AGC ACA GAA AAT.----G --- --- --- --- --- --- ---

60Ala Pro Arg Thr ArgCCC CCC AGA ACA AGG-TT --- --- --T C--Val

69Leu Glu Asp Pro Ser Met Trp EndCTT GAA GAC CCT TCC ATG TGC TCA tgggaaatgaaacttgctctcct--- --- --- --- G-A --- --- --- tgggaaatgagacttgctctctg

Ala

341

388

gacttttcctagtttccccccacatctcatctcatcctgtgaaaccag.. tctgc 440gccttttcctattttcagccca. tatttcatcgtgtaaaacgagaatccac

ctgtcceacccaatgcatgccaccaccaggctggattccg .acccatttcccttgccatccta .ccaatgcatgcagccactgtgctgaattctgcaatgttttcctttg

ttgtcgttgtatatatgtgtgtttaaataaagtatcatgcattcaaaaaaaaaaatcatcattgtatatatgtgtgtttaaataaagtatcatgcattc.

aaaaasaa*..... .....--

494

551

FIG. 2. Complete nucleotide sequence of NPY-5 (nucleotidenumber is shown on the right). For comparison, the nucleotidesequence for the cDNA encoding human NPY is shown below thatfor the rat cDNA. Areas of base deletion for either cDNA are shownby dotted lines; the asterisk indicates glycine for amidation. Thenucleotides comprising the open reading frame for NPY-5 arecapitalized, and only differing bases are demonstrated for the humancDNA. The derived amino acid sequence for the precursor of ratNPY is shown above the open reading frame, and, where differencesoccur in the human sequence, those amino acids are shown below thehuman nucleotide sequence. (CPON, C-terminal flanking peptide ofNPY). The N-terminal tyrosine is labeled 1.

Neurobiology: Allen et al.

Proc. Natl. Acad. Sci. USA 84 (1987)

h h r r

FIG. 3. Blot-hybridization analysis of mRNA derived from ahuman pheochromocytoma (lanes h) compared to rat brain (lanes r).Aliquots of RNA (5 g) were probed with a complementary RNAprobe synthesized in the presence of [32P]UTP with the cloned ratcDNA (NPY-5) as template and SP6 polymerase. Hybridization wasperformed overnight at 60°C in 0.6 M NaCl/50% formamide. The blotwas washed in 50% formamide/0.3 M NaCl, and hybridizing RNAwas visualized by autoradiography at -70°C with an intensifyingscreen (Cronex Lightning Plus).

AATAAA polyadenylylation signal suggests that NPY-5contains a complete 3' untranslated region. However, NPY-5may be incomplete at the 5' end. The number of missingnucleotides must be small inasmuch as blot-hybridizationanalysis indicates that rat and human NPY mRNA are similarin size, as are NPY-5 and the reported human NPY cDNA(Fig. 3). Moreover, preliminary analysis of the cloned ratNPY gene also suggests that NPY-5 contains a nearlycomplete 5' untranslated region. Thus, NPY-5 appears tocarry a complete representation of rat NPY mRNA.

Rat NPY mRNA encodes a 98-amino acid precursor of

NPY with a signal sequence of 29 amino acids, followed byNPY (36 amino acids), 3 amino acids (glycine, lysine, andarginine) necessary for posttranslational processing, and theC-terminal flanking peptide of 30 amino acids. The openreading frame that encodes the NPY precursor begins withtwo contiguous initiator condons ATG. This reading frame isfollowed 6 nucleotides downstream by a second open readingframe of 50 codons that is in frame with the first open readingframe. The significance of this second open reading frame isuncertain, although the work of Lomedico and McAndrewsuggests it may be translated (31). The rat and humanprecursors are compared in Fig. 2, as are the nucleotidesequences of the two mRNAs. Rat and human NPY areidentical in amino acid sequence. Remarkable conservationof amino acid sequence is found in the remaining regions ofrat and human NPY precursors. At the nucleotide level thereis less homology, with the exception of the region around thepolyadenylylation signal, where there is a stretch of 38completely conserved nucleotides.

Analysis of Tertiary Structure. A three-dimensional modelof NPY was constructed next based on its sequence homol-ogy with APP. The crystallographic structure of the APPdeposited with the Brookhaven Protein Data Bank is ofnominal resolution 140 pm (1.4 A), with the crystallographicresidual (R factor) 15.6% (20). The peptide in the crystals hasa globular structure with residues 2-8 assuming a left-handedpolyproline-II-like helix that is closely packed through hy-drophobic interactions against an a-helix formed by residues14-32. Unlike pancreatic glucagon, which appears to have aflexible structure in dilute aqueous solutions, this secondarystructure seen in the APP crystals is known to be maintainedin solution (20). The initial potential energy of the crystallo-graphic structure, as evaluated by the program CHARMM,was positive, but upon energy minimization it converged tothe value of -1053 kJ/mol (-252 kcal/mol) with only smalladjustments in atomic positions (root-mean-square 30 pm or0.3 A).

FIG. 4. Stereoscopic view of the APP crystallographic structure (light lines; from ref. 20) and the NPY model (heavy lines). Polypeptidebackbones of the two structures were superimposed by a least-squares procedure (26).

2534 Neurobiology: Allen et al.

Proc. Natl. Acad. Sci. USA 84 (1987) 2535

In constructing the NPY model, coordinates of the back-bone atoms and the side chains conserved between the twosequences were copied directly into the model. The missingside chains were then constructed by using the conforma-tional search procedure CONGEN, which finds positions ofpotential energy minima in the side-chain torsional space.The CONGEN-constructed model was then subjected to thesame energy-minimization protocol as the crystallographicAPP structure and, upon root-mean-square adjustments ofatomic positions of about 30 pm (0.3 A), a structure wasobtained with potential energy -1041 kJ/mol (-249 kcal/mol), comparable to that of the APP (Fig. 4).

It has been shown that the potential energy in vacuo asused, for example, in the program CHARMM distinguishespoorly between correctly and incorrectly modeled struc-tures, whereas the size and nature of molecular surfaces andsolvent-modified electrostatic energy values have muchgreater diagnostic value (32). Accordingly, the solvent-accessible surfaces and electrostatic energies were computedand compared for the two structures, the APP and NPYmodel (Table 2). It can be seen that the surface characteristicsof the crystallographic APP structure and the NPY model areessentially the same-i.e., 32.9-nm2 total molecular surfaceof APP, 52% of which was hydrophobic, and 32.2-nm2molecular surface of the NPY model, 56% of which washydrophobic atoms. In particular, the nature of the contactsurface between the two helical secondary structures is verysimilar in both peptides, and the NPY model actually seemsto be stabilized by somewhat larger intramolecular contacts(10.2-nm2 contact surface compared to 9.4-nm2 contactsurface of the APP). Likewise, the electrostatic energies ofthe two structures evaluated to infinity with the dielectricconstant of 50 were comparable, and the model appeared tobe stabilized by significantly larger electrostatic forces thanin the parent crystal structure (-16.3 kJ/mol or -3.9kcal/mol, compared to -8.3 kJ/mol or -2.0 kcal/mol in theAPP).

DISCUSSIONWe have cloned and sequenced a cDNA that represents NPYmRNA. Analysis of the coding region revealed that rat andhuman NPY are identical. Moreover, the amino acid se-quences of rat, human, and porcine NPY are highly homol-ogous to APP (Table 1). This remarkable homology preservesall of the residues important for the APP secondary structureand has allowed us to construct a three-dimensional structuremodel of mammalian NPY based on the known crystallo-graphic structure of the avian homologue, APP. This modelsuggests (i) that, consistent with present knowledge of NPYpharmacology and receptor binding, NPY probably possess-es a compact tertiary structure (Fig. 5); and (ii) that severalamino acids residing in key positions should be the focus offuture analyses of NPY structure-function relationships.

Table 2. Comparison of solvent-accessible and proline-I-helix-a-helix contact surfaces in the APP structure and the NPY model

Contact surfaces,nm2 (100 A2)

APP NPY

Total solvent-accessible surface 32.9 32.1Hydrophobic solvent-accessible surface 17.0 17.9

(carbon atoms)Hydrophilic solvent-accessible surface 15.9 14.2

(oxygen/nitrogen atoms)Helix-helix contact surface 9.4 10.2Hydrophobic contact surface 7.3 7.0

By using the deduced amino acid sequence of rat NPY, athree-dimensional model was computer-generated from thecrystal structure of the homologous APP. The method ofhomologous modeling has been criticized (32) and has beenshown to lead to incorrect results, at least in some cases (33,34). However, the small size of the two peptides APP andNPY considered here (36 amino acids), the refined modelingprotocol used here (an exhaustive search of side-chainconformational space), and a high degree of homologybetween APP and NPY that conserves all of the unique traitsof the APP sequence strongly suggest that our NPY modelcaptures correctly the most essential features of its tertiarystructure. The residues most important for the APP tertiarystructure are those mediating helix-helix contacts-namely,the proline residues 2, 5, and 8, and 13 of the polyproline helixand the a-helicAl residues leucine-17, phenylalanine-20, leu-cine-24, tyrosine-27, and valine-30. All of these residues areeither conserved in the NPY molecule or replaced by chem-ically homologous side chains (APP Phe-20 to Tyr-20 in NPY;APP Val-30 to Leu-30 in NPY). Thus, one of the most

FIG. 5. Various aspects of the computer-constructed NPY model(all three panels in the same orientation). (Top) Polypeptide chainbackbone, showing the polyproline-II-like helix (lower backbone)and the C-terminal a-helix (upper backbone) connected by a tightturn. (Middle) Color-coded line representation of the whole model.Yellow, carbon; red, oxygen; blue, nitrogen; green, hydrogen.(Bottom) Space-filling representation of the NPY model. Color-coding of oxygens and nitrogens is as in Middle.

Neurobiology: Allen et al.

Proc. Natl. Acad. Sci. USA 84 (1987)

remarkable features of the NPY model that emerges from ourstudy is its extensive and compact tertiary structure, asstabilized by the intramolecular contacts between a polypro-line helix and the a-helix. These two parts of the molecule arebrought together by a tight hairpin fold so that the N and Ctermini of NPY come to spatial proximity (Fig. 5).

Certain structural requirements for the biological andpharmacological effects of NPY have been investigated.Deamidation of the C-terminal tyrosine was found to result incomplete loss of activity in an in vitro ileal preparation (35),although this amide group appears to contribute little to thecrystal structure ofAPP (20) and the tertiary structure of theNPY model. Similarly, deletion of the N-terminal tyrosineresulted in marked reduction in vasoconstrictor potency (36).Interestingly, N-terminal deletion fragments are ineffectiveinhibitors of NPY receptor binding in parallel with decreasedpotency (37). Thus, it appears that both the C- and N-terminalregions ofNPY are required for bioactivity and that receptorbinding and activation may reside in N and C termini,respectively.The N-terminal tyrosine could be involved in NPY recep-

tor binding either directly or indirectly. That is, the solvent-exposed part of this tyrosine side chain may interact directlywith NPY receptor residues, or, alternatively, its buried partmay stabilize and be crucial for integrity of the NPY struc-ture. Inasmuch as APP has a glycine in position 1 and is a poorcompetitive inhibitor of NPY binding, we favor the firstpossibility. The alternative roles of Tyr-1 and the possiblecontributions of other residues to NPY structure, biologicactivity, and receptor binding can now be tested specificallyon the basis of our model. For example, it is already knownby receptor binding kinetics that the related peptide, PYY, isa good competitive ligand for the NPY receptor. These twopeptides possess virtually identical residues to each other intheir N- and C-terminal regions, further emphasizing theimportance of these regions in receptor binding, as suggestedby the model for NPY proposed in this study. The twopeptides diverge in amino acid content within the middleregions; thus, substitutions within this area appear to betolerated without affecting receptor binding. The behavior ofthese naturally occurring homologues, PYY and APP, inreceptor binding studies provides additional information fordefining key amino acid residues.

In this model, substitutions of internal residues in thea-helix (positions 20, 24, 27, and 30) with polar residuesshould destabilize the a-helix/proline-helix interactions,whereas substitutions of external residues (positions 19, 22,and 26) should have no effect on these interactions but rathershould alter the solvent-accessible surface and thereby pos-sible receptor interactions at putative sites distant from the Nand C termini. For example, does replacement of Tyr-1 withglycine significantly lower receptor binding? On the otherhand, the deamidated NPY analogue should bind to thereceptor as well as NPY. These and similar modifications inkey positions when rationalized by computer-aided analysesof the NPY model should serve both to further test thevalidity of the modeling approach and to lead to a moredetailed understanding of the structure and function ofneuropeptides in central and peripheral nervous systems.

We thank Dr. Robert Bruccoleri (Massachusetts General Hospital,Boston) for his help in computer-aided model building and Mr. MarkHandschumacher for his aid with PS 340 computer graphics. We aregrateful to Marie LeSieur for expert assistance and JoAnn Forti andKathy Sullivan for editorial help. J.A. is a recipient of a WellcomeTrust Training Fellowship.

1. Herbert, E. & Uhler, M. (1982) Cell 30, 1-2.

2. Hokfelt, T., Johansson, O., Ljungdahl, A., Lundberg, J. M. &Shultzberg, M. (1980) Nature (London) 284, 515-521.

3. Tatemoto, K., Carlquist, M. & Mutt, V. (1982) Nature (Lon-don) 296, 659-660.

4. Tatemoto, K. (1982) Proc. Natl. Acad. Sci. USA 79, 5485-5489.

5. Allen, Y. S., Adrian, T. E., Allen, J. M., Tatemoto, K., Crow,T. J., Bloom, S. R. & Polak, J. M. (1983) Science 221, 877-879.

6. Chronwall, B. M., DiMaggio, D. A., Massari, V. J., Pickel,V. M., Ruggiero, D. A. & O'Donohue, T. L. (1985) Neurosci-ence 15, 1159-1181.

7. Lundberg, J. M., Terenius, L., Hokfelt, T., Martling, C. R.,Tatemoto, K., Mutt, V., Polak, J., Bloom, S. & Goldstein, M.(1982) Acta. Physiol. Scand. 116, 447-480.

8. Allen, J. M., Polak, J. M., Rodrigo, J., Darcy, K. & Bloom,S. R. (1985) Cardiovasc. Res. 19, 570-577.

9. Stanley, B. G. & Leibowitz, S. F. (1984) Life Sci. 35, 2635-2642.

10. Clark, J. T., Kalra, P. S., Crowley, W. R. & Kalra, S. P.(1984) Endocrinology 115, 427-429.

11. McDonald, J. K., Lumpkin, M. D., Sampson, W. K. & Mc-Cann, S. M. (1985) Proc. Natl. Acad. Sci. USA 82, 561-564.

12. Lundberg, J. M. & Tatemoto, K. (1982) Acta. Physiol. Scand.116, 393-402.

13. Allen, J. M., Bircham, P. M. M., Edwards, A. V., Tatemoto,K. & Bloom, S. R. (1983) Regul. Peptides 6, 247-254.

14. Allen, J. M., Raine, A. E. G., Ledingham, J. G. G. & Bloom,S. R. (1985) Clin. Sci. 68, 373-377.

15. Ekblad, E., Edvinsson, L., Wahlestedt, C., Uddman, R.,Hakanson, R. & Sundler, F. (1984) Regul. Peptides 8, 225-235.

16. Allen, J. M., Ferrier, I. N., Roberts, G. W., Cross, A. J.,Adrian, T. E., Crow, T. J. & Bloom, S. R. (1984) J. Neurol.Sci. 64, 325-331.

17. Minth, C. D., Bloom, S. R., Polak, J. M. & Dixon, J. E.(1984) Proc. Natl. Acad. Sci. USA 81, 4577-4581.

18. Leiter, A. B., Montminy, M. R., Jamieson, E. & Goodman,R. M. (1985) J. Biol. Chem. 260, 13013-13017.

19. Tatemoto, K. (1982) Proc. Natl. Acad. Sci. USA 79, 2514-2518.

20. Glover, I., Maneef, I., Pitts, J., Wood, S., Moss, D., Tickle, I.& Blundell, T. (1983) Biopolymers 22, 293-304.

21. Bruccoleri, R. E. & Karplus, M. (1987) Biopolymers 26,137-168.

22. Benton, W. D. & Davis, R. (1977) Science 196, 180-182.23. Beaucage, S. L. & Caruthers, M. H. (1981) Tetrahedron Lett.

22, 1859-1862.24. Maxam, A. M. & Gilbert, W. (1980) Methods Enzymol. 65,

499-560.25. Bernstein, F. C., Koetzle, T. F., Williams, G. J. B., Meyer,

E. F., Brice, M. D., Rodgers, J. R., Kennard, O., Shimanou-chi, T. & Tasumi, M. (1977) J. Mol. Biol. 112, 535-542.

26. Brooks, B., Bruccoleri, R. E., Olafson, B. D., States, D. J.,Swaminathan, S. & Karplus, M. (1983) J. Comput. Chem. 4,187-217.

27. Bruccoleri, R. E. & Karplus, M. (1985) J. Comput. Chem. 7,165-175.

28. Lee, B. K. & Richards, R. M. (1971) J. Mol. Biol. 55, 379-400.29. Warshel, A., Russell, S. T. & Churg, A. K. (1986) Proc. Natl.

Acad. Sci. USA 81, 4785-4789.30. Jones, T. A. (1978) J. Appl. Cryst. 11, 268-272.31. Lomedico, P. T. & McAndrew, S. J. (1982) Nature (London)

229, 221-226.32. Novotny, J., Bruccoleri, R. E. & Karplus, M. (1984) J. Mol.

Biol. 177, 787-818.33. Delbaere, L. T. J., Brayer, G. D. & James, M. N. G. (1979)

Nature (London) 279, 165-168.34. Read, R. J., Brayer, G. D. & James, M. N. G. (1984) Bio-

chemistry 23, 6570-6575.35. Allen, J. M., Hughes, J. & Bloom, S. R. (1987) Dig. Dis. Sci.,

in press.36. Rioux, F., Bachelard, H., Martel, J. C. & St. Pierre, S. (1986)

Peptides 7, 27-31.37. Martel, J. C., St. Pierre, S. & Quirion, R. (1986) Peptides 7,

55-60.

2536 Neurobiology: Allen et al.