Bitu'himwa et Biophysica Acta. 1{170(I 9ql ) 23-28 ~' 1991 Elsevier Science Publishers B V. 111fi7-4~38/01/$113 /ID()NIS 1116,74838'911~1248S

23

50

BBAPRf) 33t~5x

Influence of polyfluodnation of the phenylalanine ring of angiotensin II on conformation and biological activity Philippe R. Bovy ~, D a n i e l P. G e l m a n ~ J o h n M. M a t s o u k a s ~ a n d G r a h a m J. M o o r e ~ i Sear/e Researd~ and lh, t'elopment. Momanto Life ScJenc¢~ Re~ear(h Ct'nter, ("he~ter]ield. M O (U.S.A. ), 2 Mon ~tlnlo ('o, ,5"t I.ouf ~, MO ~U.S.A.) and ~DepartPrwnt o] . Afedwa[ . . Buv'hemtstO'. . Un. (,r~itv of Ca/gar~' tlealth S~ ,once* ('emre, ('algu~' (Cunada~

(Received 28 March It~JI)

Ke~, words: Pentai]uorophenylalanine: Angi~teosin 11: Conft~rmation by NMR: Rcct:ptor mechanism

[Phe(Fs)Slangiotensin 11 was synthesized by the solid phase method and purified by reverse.phase HPLC. In rat uterus and rabbit aorta bioassays the analogue had 10 and 50%, respectively, of the contractile activity of angiotensin I! and demonstrated antagonist properties. These findings illustrate that inversion of the Phe s ring quadrupole moment in angiotensin II decreases agonist activity and invokes antagonist properties, t H-NMR studies at 400 MHz in DMSO-d 6 demonstrated the presence of cis and trans isomers in the ratio 1:3 due to restricted rotation of the His-Pro bond. Downfleld shifts ef the His C 2 and C4 protons in [Phe(Fs)]ANG !1 compared to ANG !! suggest that the Phe(F5) residue may be involved in a parallel-plate ring pairing interaction with the imidazole group. However heteronuclear NOE studies, carried out by measuring the proton difference spectrum before and after saturation of the fluorine resonances, showed the absence of any NOE enhancement illustrating that electrostatic influences of the Phe(F5) ring occur at relatively long range.

Introduction Thc octapcptide angiotensin II (ANG 11, Asp-ArgVaI-Tyr-lle-Pro-His-Phe) is a potent vasoconstrictor hormone involved in the pathogenesis of hypertension. Precise details of the interaction of ANG 11 with smooth muscle receptors remain to be elucidated. The importance of the Phe ~' ring of ANG II for receptor activation has been known since the di~overy [I] of the antagonist Saralasin (tSar j AlaS]ANG !i). Further investigations showed that antagonists of ANG !1 could be obtained by substituting a large variety of amino acids for the Phe s residue, e.g., lie [2], MePhe [3]. o-Phe [4], suggesting that the aromaticity of the Phc ring in ANG !I together with its relative orientation with respect to the rest of the molecule are important factors in determining receptor stimulation properties. Later it was found that the substitution of cyclohex3'lalanine (Cha), which is nearly isosteric with Phe, but

Abbreviations: NOE. nuclear Overhau~r ct[cct: RP-HPLC, rcversed-pha~ high pressure liquid ¢hromalogr/tph~,. Correslxmdence: G.J. Moore. Department of Biochemislr~. I!mversity of Calgary. Health Sciences Center. 33311 Hospital Dr. N W. Calgary. Alberta. Canoda. T2N 4NI.

lacks the quadrupole moment associatcd with an aromatic ring. produced an antagonist which retained some agonist properties [5]. This finding could illustrate a role for the quadrupole moment of Phe in the expression of agonist efficacy and led to the present investigation on the effects of reversing the quadrulx~le moment of Phe by substituting Phe(Fs). Depending on the nature of the constraints imposed on the interactions of amino acid side-chains at the C-terminus of ANG 11, two Ix~ssible .~utcomes seemed feasible. Either the Phe(F~) ring would .,,te1:cally mimic the orientation of Phe in ANG 11 so that the ring quadrupolc was inverted compared to ANG II or the PhelF~) ring would turn through 90 ° and thereby align -

+

-

+

Benzene (Phe) [Quadr~pole ~trength=-29 ]

Hexafluorobenzene (FsPhe) [Quadrupole st rength=+32 ] Scheme ].

24 itself according to the electrostatic influences of quadrupole moment rather than by steric considerations alone, in other words, the substitution of Phe(F5) for Phe could be an isosteric or an isofunctional substitution, depending on the relative orientation of the Phe(F 5) ring compared to the Phe ring in ANG I1. Previous studies have sugt,ested that an electrostatic interaction between the Phe ring and the His ring may be important for receptor stimulation. Thus tH-NMR studies have shown that for ANG I1 in DMSO, the Phe ring is shielded and unable to freely rotate, and that replacement of the Phe ring with lie results in the removal of a weak shielding influence on the His ring [6]. The propped electrostatic interaction between the His and Phe rings in ANG II was later suggested to be a perpendicular-plate interaction, rather than a traditional parallel-plate ring pairing interaction, on the basis of ab initio calculations [7]. Accordingly, inverting the quadrupole moment by substituting Phe(Fs) for Phe could result in 90 ° rotation of either the Phe(F5) ring or the imidazole ring in order to reinstate such an electrostatic interaction. The proposal that the His ring may be rendered essentially immobile by interaction with both the C-terminal carboxylate and the Tyr hydroxyl [8], would imply preferential rotation of the Phe(F5) ring. Other studies, which have suggested an F F~CHO

interaction between the Phe ring and the C-terminal carboxb'late anion in ANG II [9], would al~ favor reorientation of the Phe(E.) ring in order to reaccommodate this ion-quadrupole interaction. In any case, the introduction of Phe(F5) at the C-terminus of ANG II might be expected to be accompanied by changes in the proton NMR spectrum and possibly changes in bi~,logical activity as well. Materials and Methods

L-Phe(E0 was synthesized according to the strategy shown in Fig. 1. The key step in the preparation of

N-t-butyloxycarbonyl-2,3,4,5,6-pentafluoro-( S )-phenylalanine (7) involved the asymmetric hydrogenation of methyl ( Z).2-benzamido-3-(2',Y,4',5',6'-pontafluorophenyl) acrylate (4). This olefin was prepared by the condensation of 2,3,4,5,6-pentafluorobenzaldehyde (1) with 2-phenyl-5(4H)-oxazolone (2), generated in situ from N-benzoylglycine and acetic anhydride and subsequent ring-opening of 3 with methanol in the presence of N,N-dimethyl-4-aminopyridine. In this manner a mixture of the desired (Z) isomer (4) and the undesired (E) isomer is obtained. However, recrystallization from methanol affords pure 4 in 25% overall yield from pentafluorobenzaldehyde. The asymmetric hydro0

PI1 N< +

0

F" ~ "F F

F" " ~ "~F F

0

i

F

COzCH3

~~~.~-~ f ~ y\F F

0

F ~ CO2cH3 I. [[ NH-C--Ph [P&(D[PAMP)COD]BF.F" ~T,'~'%F 0 F Hz

5

I HCI~zO F /

CO;H

F~j"I~N

H3CI

F" ~

"~F F

BOC20

_ /F .COzH I'~.r~NHBo¢

NaOH

F ~

~'F F

7 Fig. I. Synthesisschemefor pcnlafluoro-L-phen~lalanine.

25 genation [10] of 4 in methanol at room temperature in the presence of one-half mol percent cyclooctadiene1.5[( R , R ) l ,2-ethanediylbis( o-met hoxyphenyl )phenyl. phosphine]rhodium tetrafluorobc, rate [ll] provides methyl N-benzoyl-2,3,4,5,6-pentafluomphenylalaninatc (5) of an undetermined enantiomeric composition in 88% yield. The hydrolysis of 5 was performed using a mixture of 12 N hydrochloric acid and acetic acid at reflux and afforded the hydroehloride salt {6) in quantitative yield. At this point the enantiomeric composition was determined by high pressure liquid chromatography, after derivatization with Marfey's reagent [12] and shown to be 90:10 (S: R, respectively}. The hydrochloride salt was then converted into the desired N-t-butyloxycarbonyl-2,3,4,5,6-pentafluoro-phenyl-(S)alanine (7) in 54% yield. The enantiomeric compGsition was enhanced to 98 : 2 during the recrystallization process and determined by HPLC after treatment of a sample with trifluoroacetic acid and taen Marfey's reagent. The overall yield of 7 from pentafluorobenzaldehyde was 12%, m.p. III°C. The N-t-butyloxycarbonyl-2,3,4,5,6-pentafluoro-phenyl-(S)-alanine was then esterified to a resin serving as insoluble support for solid phase peptide synthesis. Thus, fo[Iowing the Gisin method [13], the protected amino acid 7 was converted to its cesium salt and reacted with a chioromethyl-functionalized polystyrene resin. The peptide [Phe(Fs)S]ANG II (8} was assembled using the standard Bee-amino acid protocols of the Applied Biosystems System Software version 1.30, except for the coupling of Boc-His (BOrn) which required a customized cycle. Removal of the peptide from the resin and simultaneous deprotection of the side-chain functions were achieved by treatment with hydrogen fluoride (0°C, 60 rain). Purification was by preparative, reverse-phase, high performance liquid chromatography on a C m bonded silica gel column. The homogeneous peptide conformed to the theoretically expected structure by Fast Atom Bombardment mass spectroscopy and high field ~H-NMR. On the Beckman 121M amino acid analyzer, Phe(F0 was found to elute at a position immediately preceding Ue and close to the elution time for Met. Rat isolated uterus and rabbit aorta bioassays were carried out as described previously [14,15]. NMR studies were carried out with Varian 3(10 MHz and Bruker 400 MHz spectrometers at ambient temperature in H 2 0 / D 2 0 (10: 1) and in DMSO-d 6. After RP-HPLC purification, the peptide (which is obtained in protonated form (TFA salt}} was converted to the neutral form (acetate salt) by passage through a CM-cellulose column [14]) prior to dissolution in DMSO-d~ at a concentration of 5 mg/0.5 ml. One dimensional ~H and t*F spectra were obtained in DMSO-d 6 and heteronuclear NOE studies were carried out by saturating ortho, meta and para resonances in three separate

8t~ 6on

20-

[~Prt~i IM) Fig. 2 Dose-response curve for ANG II (~} and [PhdFOSlANG It (.) in Ihe rabbit isolated aorla assay

experiments and recording the proton difference spectrum in each case. Results

[Phe(Fs)]ANG II had 9.6 + 6% (N = 2) of the activity of ANG I1 (comparison of ED25 doses of ANG 11 versus analogue) in the rat isolated uterus assay and was able to elicit a maximum reponse in some tissues, but not all. Attempts to construct dose-response curves for this peptide proved futile due to the mixed agonist/antagonist behaviour and the day-to-day variability of tissue responses towards this analogue. Large doses (10 -~ M) of the peptide produced a long-lasting (30 min) inhibitory influence after washing out, similar to the long term 'desensitizing' effect observed for type I antagonists [16]. In the rabbit aorta a~say [Phe(Fs)s] ANG II acted as a partial agonist, producing only 50% of the maximum contraction to ANG II and invoked a desensitizing effect at high doses (Fig. 2). The analogue also antagonized the effects of ANG II in this assay (data not shown), although the antagonist activity could not be quantitated due to the variability revoked by the intrinsic agonist activity. The ~H-NMR spectrum for [Phe(Fs)]ANG 11 in water at acid pH provided analytical data in support of the integrity of the peptide and showed the presence of an (average} extended chain conformeds) (Fig. 3) as expected 18]. The I H-NMR spectrum of [Phe(Fs)IANG 11 (neatral form) in DMSO-d,, (Fig. 4A) demonstrated the presence of cis and trans isomers (ratio 1:3) due to restricted rotation at the ttis-Pro bond+ based on knowledge from previous NMR studies of ANG I1 analogues [17-+19]. When the resonances of the aromatic protons in the dominant trans isomer of [PhdF~) ANG ll] are compared to the same protons in ANG !! (Table I), a slight downfield shift of the Tyr rnera and ortho protons is observed together with a marked downfield shift of the His C: and C+ protons. Apparently the presence of the Phe(F5) residue invokes a deshielding effect on the His side-chain in the, dominant trans isomer.

26 [F~elF:)B]Ari~ 1~

i~ tw~ spectr~l

H=His,Tyr,Arg.Va],lle,&sp, Phe(F$)

HCHis,Tyr,Phe(Fs) H~Pro,Arg

~

J

ti'

, |

I

MaYa1,1le,Pro,Ar 9 H~Pro,Aro,Ile

~HArg,His,Tyr Val,I;e,Phe(F5) i

I

I

]

Fig, 3. ~It-NMR spectrum of [Phe(F~)S]ANG It in H _,O/D,O (9: I ) pH 3.

The ~gF-NMR spectrum showed resonances for the ortho, meta and para fluorine atoms (Fig. 4B) and also the presence of 1.5-2.0 equivalents of TFA. The TFA is presumably associated with the strongly basic Arg side-chain and with the N-terminal amino group. Apparently the profound differences in the NMR spectra of HPLC-purified angiotensin peptides observed in DMSO before and after treatment of peptides with CM-ccllulose, [8], is due solely to conversion of the TFA salt of the His imidazole group to an acetate salt

A

and reflects the protonation status of the His6 residue (cationic and neutral, respectively). Thus, before and after CM-cellulose chromatography, respectively, lyophilized angiotensin peptides are obtained in the forms His'H+TFA - and His'H+CH3CO~ , respectively which, upon dissolution in dry DMSO, become His. H ++ TFA- and His + CH 3CO~ H +, respectively. This situation does not apply to protic solvents [27]. The presence of cis and trans isomers was not visualized in the ~gF-NMR spectrum, suggesting that

[Pt.e(FsI~]M~,H ~n

19Fte~s~ectrum

IH e,~ ~pe 5 A). Similarly, recent conformational studies on [Sar~]ANG 11 have not demonstrated an NOE between Phe and His [18]. One of the difficulties associated with NOE experiments involving saturation of aromatic rings resonances is the availabilit) of alternative relaxation pathways through vicinal atoms, which often makes it difficult to detect anything other than short range ( < 4 ,,~l Ovcrhauser effects [20]. Previous findings based on 2D-ROESY [18] and NOE studies [19] are supportive of an ion-dipole interaction between the C-terminal carboxylate and the imidazole ring [22] and also an ion-quadrupole interaction b e ~ e e n the C-terminal carboxylate and the Phe ring in ANG 11 [9] (Fig. 5A). Electrostatic interaction between the His and Phe rings has also been implieu [6]. Normally an ion-quadrupole (CO~ :Phc) bond represents a stronger interaction than a dipole-quadrupole bond (His: Phe) at equivalent range [7] and therefore

TABLE I Biological aclMtics and IH-NMR c ~cm~cal shi|t8 for angiotcnsm analogues. Agonist acl '~

Chemical Shill "

Ic/)

Hi,.tidine C",

T) ro'. ne C~

rncla trarz~,

A N G It (Phe(Fs)X]ANG It lilt:SlANG It t.

101! Ifl ' < fl01 ~

7.4,~ 7.71 a 7 611

62";7 6.97 '~ 6.92

" Chemical shifts tppm) ate relative to TMS in DMSO-G. Rat ulcrus hioas:~y. • Afllagonist v, ilh slow rc,~er~al prolx~rlics l l 6 ] 'J Chemical shills are for the dominanl trgn~ i~mer: resonances ~or the ~!s i ~ e r From Rcf h.

6 9~ 7if3 7 (Jl

ortho cJs

tran~

ct~

6~5

6.5'0 66.3 661

b.51

could not be Jdenldied vdrh an~ cerlainty (Fig, 4)

28 the primary influence on the orientation of the ['he ring should derive from the C-terminal carboxylalc rather than the His ring. Consequently, the main influence of substituting the Phe(F~) ring for the Phe ring should be rotation of the Phe(F~) ring through 00 ° in order to reestablish effective interaction with thc Ctcrmina! anion (Fig. 5). Furthermore, to account for the deshiclding influence of the Phe(F~) ring on the His ring (Table 1), a parallel-plate interaction with the His ring might also be p r e ~ n t (Fig. 5B), thoughbeit at relatively iong range ( > 5 A) according to the absence of heteronuclear Overhauser effects. Chemical shift differences of 0.1-0.2 ppm (Table I) ~enerally reflect proton to ring distances of about 6 A 121), which is beyond the usual NOE range of _