AbstractsQuantum chemical C N D O /~ calculations for the conformational preference of the side chain of thioridazine as a function of angle indicated the crystallographically determined structure gave the lowest energy. There is also a small region of conformational flexibility within the first 90" of rotation of the side chain. This is commensurate with the results which we had obtained previously for our similar calculations for promazine and its C1 and CF, derivatives, perazine and its C1 and CF, derivatives, and for the hypothetical hitherto unknown N-piperidinopromazine and its C1 and CF, derivatives. The conformational profile of thioridazine resembles that of the perazines. The calculated gross atomic populations on the alkyl nitrogen in thioridazine was within the range we had previously found necessary for neuroleptic activity.Des calculs quanto-chimiques de type C N D O /~ pour la preference conformationnelle de la chalne de c6tk de la thioridazine en fonction de I'angle indiquent que la structure determinee par la cristallographie donne l'energie la plus base. I1 y a aussi une petite region de flexibilite conformationnelle autour de I'angle 90". Ceci est en accord avec des rksultats obtenus par nous-mtmes pour la promazine et ses dkrivCes C1 et CF,, la perazine et ses derivkes CI et CF, et pour la N-piperidinopromazine (inconnue jusqu'ici) et ses derides Cl et CF,. Le profil conformationnel de la thioridazine resemble i celui des perazines. Les populations atomiques brutes calculkes de l'azote alkyle dans la thioridazine se trouve dans les limites nkcessaires pour I'activiti neuroleptique.Quantenchemische c~~o/2-Berechnungen der konformationellen Praferenzen der Seitenkette von Thioridazin als Funktion des Winkels deuten an, dass die kristallographisch bestimmte Struktur die tiefste Energie hat. Es gibt auch ein kleines Gebiet mit konformationeller Flexibilitat urn die ersten 90"-Drehungen der Seitenkette. Dies steht mit den Ergebnissen ahnlicher Berechnungen in Einklang, die wir erhalten haben: Promazin, Perazin und die bisher unbekannte N-piperidinopromazin mit ihren C1-und CF,-Derivaten. Das Konformationsprofil von Thioridazin ist dem der Perazin ahnlich. Die berechneten atomaren Bruttopopulationen des Alkylstickstoffs in Thioridazin befinden sich in dem fur neuroleptische Aktivitat notwendigen Bereich.

No abstract

The substitution of an ally1 group for the methyl group on the nitrogen of morphine changes the pharmacological activity from that of a narcotic agonist to that of a mixed narcotic agonist-antagonist; substitution of an allyl group for the methyl group in oxymorphone changes it from a narcotic agonist to a pure antagonist [ I]. (The molecular structures are depicted in Figure 1.) When we first initiated our quantum chemical investigations on narcotics and narcotic antagonists [2] there was no structural information on the position and conformation of an ally1 group relative to the piperidine moiety in such molecules. Subsequently, at our request Dr. I. Karle graciously determined the crystal structure of naloxone (from a sample which we supplied) [3]. From these atomic coordinates it was possible for us to perform the quantum chemical calculations for naloxone [4-81 at what corresponds to at least one of its most stable minimum energy conformations. We had also calculated similar conformations for nalorphine by both the C N D O /~ and PCILO methods. In our early quantum chemical calculations of nalorphine [9-111 we had used the old crystal structure of morphine [12] and tried various positions of the allyl group relative to the piperidine ring. We noted that there was a discernible sensitivity to the position of the allyl group not only in the calculated total energy but also even in the energy of the highest occupied orbital. While it would have been possible for us to calculate a conformational energy map for nalorphine by the CNDO/2 or INDO methods as we have done for the neuroleptic phenothiazines [4,6,7,[13][14][15], at that time we did not have the sufficient necessary computer time available to this particular praject.The PCILO (Perturbative Configuration Interaction based on Localized Orbitals) method [ 161 developed in the group of Professor B. Pullman has been used most successfully by him and coworkers to calculate the conformational energy maps of a large variety of biological and pharmacological molecules. (Reference [17] presents a comprehensive overview of many of their P C I L .~ results on pharmacological molecules.)Thus, knowing the crystallographic position of the allyl group in naloxone relative to the piperidine moiety, we investigated by PCILO calculations if the same relative position of the allyl group would be the preferred conformation for nalorphine. In the interim a new more accurate crystal structure of morphine was published [18]. We performed the PCILO calculations for nalorphine using Gylbert's crystal structure for morphine replacing the N-methyl group with the allyl group in the same position relative to the piperidine moiety as was found for naloxone [3].Designating the atoms of the allyl side chain of nalorphine (Fig. 1, compound 11) as N--C17-C18=C19r we varied both the angles of N-C17 and CL7-CI8 simultaneously by 30" increments. The resulting map indicated that the position of greatest stability was the one we chose. There was one other small region of stability from 90" to...

Illustrative examples are given of comparative results on chlorpromazine by the C N W / 2 method and by an INDO method. For this we had previously extended the INDO formalism to include d orbitals and parametrization to include compounds with second-row atoms. Results are presented for other antipsychotic drugs such as perazine and trifluoperazine and for the narcotic agent, morphine.

The results of our recent ab-initio SCF calculations on pyrrole and pyrazole using large Gaussian basis sets are compared to those using less rigorous methods: CNDO/2. INW. PCILO extended Hiickel, and ARCANA (a type of iterative extended Hiickel calculation). The ab-initio orbital energies are higher in energy than all those resulting from less rigorous methods. For the two highest occupied orbitals, both of n type, (pyrrole-la, and 2b, ; pyrazole-3a" and 2a") there was excellent agreement between the measured photoelectron spectroscopic ionization potentials and the Koopman's theorem values from the ab-initio results. Thus the orbital energies calculated by the less rigorous methods all lie too low in energy. The CNW/2 and INDO orbital energies also span too large a width for the valence band compared to the ab-initio results.The gross atomic populations from the C N D O /~, INDO and PCILO methods show discrepancies from the ab-initio resultsfor both the heteroatoms and the hydrogens. Total overlap populations calculated from the ab-initio SCF wave functions show large negative overlap populations between nonbonded ring atoms. Using the off-diagonal density matrix elements from C N D O /~ and INDO wave functions as if they were true off-diagonal elements from n o n -z m calculations, scaling them (shown previously to lead to reasonable values for TOPS between bonded atoms) led to TOPS between nonbonded ring atoms that were essentially zero. Wiberg bond indices, by definition only capable of giving positive numbers, also cannot reproduce these negative overlap populations. The extended Hiickel and ARCANA methods, which are n o n -z w methods, both gave negative TOPS between nonbonded ring atoms, although not as large as those which resulted from ab-initio calculations.