The mechanism of cis-trans isomerization of prolyl peptides catalyzed by cyclophilin (CyP) was studied computationally via molecular dynamics (MD) simulations of the transition state (TS) and the cis and trans forms of the ground state (GS), when bound to CyP and when free in aqueous solution. The MD simulations include four enzyme-bound species of tetrapeptide (Suc-Ala-XC([double bond]O)-NPro-Phe-pNA; X = Gly, Trp, Ala, and Leu). In water, the prolyl amide bond is favorably planar with the presence of conformers exhibiting +/-20 degrees twist of the C-N dihedral. In the active site a hydrogen bond between the cis-prolyl amide carbonyl O and the backbone amide N-H of Asn102 retains the 20 degrees twist of the C-N dihedral. The TS structure is characterized by a 90 degrees twist of the amide C-N bond and a more favorable interaction with Asn102 due to the shorter distance between Asn102(HN) and the amide carbonyl O. The conformational change of cis --> TS also involves pyramidalization of the amide N, which results in the formation of a hydrogen bond between the amide N and the guanidino group of Arg55. Both Asn102 and Arg55 are held in the same position in CyP.cis-isomer as in CyP.TS. In the ligand-free CyP the Arg55 guanidino group is highly disorganized and Asn102 is displaced 1 A from the position in the ligand-bound CyP. Thus, the organization of Arg55 and Asn102 occurs upon substrate binding. The geometrical complimentarity of the organized enzyme structure to the TS structure is a result of preferential binding of the proline N and the amide carbonyl of the TS compared to that of GS. However, the N-terminal part (Suc-Ala) becomes repositioned in the TS such that two hydrogen bonds disappear, one hydrogen bond appears and two other hydrogen bonds becomes weaker on the conversion of CyP.cis to CyP.TS. During this conversion, total hydrophobic contact between enzyme and the peptide is preserved. Thus, the interaction energies of GS and TS with enzyme are, as a whole, much alike. This does not support the contention that TS is bound more tightly than GS by K(m)/K(TS) = 10(6) in the cis --> trans reaction. Repositioning of the N-terminal part of the peptide on CyP.TS formation becomes more pronounced when the substrate X residue is changed from Gly < Trp < Ala < Leu. We propose that the larger turning of the N-terminus is responsible for the larger value of the experimentally observed Delta S(++) and Delta H(++), which sum up to little change in Delta G(++). The positioning of the Arg55 and the degree of 20 degrees twist of the amide C-N bond are considered as criteria for Near Attack Conformers (NACs) in cis-trans isomerization. NACs account for approximately 30% of the total GS populations of the cis-isomer. Similar NAC populations were observed with four different substrates. This is consistent with the insensitivity of enzymatic activity to the nature of the X residue. Also, the NAC population in CyP.trans-AAPF was comparable to that in CyP.cis-AAPF, in accord with similar experimentally measured rates of the cis --> trans and trans --> cis reaction in CyP. These NACs, found in CyP.cis and CyP.trans, resemble only one of the four possible TS configurations in the water reaction. The identity of this TS structure (syn/exo) is in accord with experimentally determined KIE values in the enzymatic reaction. However, the geometry of the active site was also complementary to another TS structure (anti/exo) that was not detected in the active site by the same KIE measurements, implying that the geometrical fitness of the TS cannot be a single determining factor for enzymatic reactions.