Oxygen Radical Scavenger Activity, EPR, NMR, Molecular Mechanics and Extended-Hückel Molecular Orbital Investigation of the Bis(Piroxicam)Copper(II) Complex

The oxygen radical scavenger activity (ORSA) of [CuII(Pir)2] (HPir = Piroxicam = 4-hydroxy -2- methyl -N-2- pyridyl -2H- 1,2-benzothiazine -3- carboxamide 1,1-dioxide) was determined by chemiluminescence of samples obtained by mixing human neutrophils (from healthy subjects) and [CuII(Pir)2(DMF)2] (DMF = N,N -dimethylformammide) in DMSO/GLY/PBS (2:1:2, v/v) solution (DMSO = dimethylsulfoxide, GLY = 1,2,3-propantriol, PBS = Dulbecco’s buffer salt solution). The ratio of the residual radicals, for the HPir (1.02·10−4M) and [CuII(Pir)2(DMF)2] (1.08·10−5M)/HPir (8.01·10−−5M) systems was higher than 12 (not stimulated) [excess of piroxicam was added (Cu/Pir molar ratio ≈1:10) in order to have most of the metal complexed as bischelate]. In contrast, the ratio of residual radicals for the CuCl2 (1.00·10−5M) and [CuII(Pir)2(DMF)2] (1.08·10−5M)/Hpir (8.01·10−5M)system was 5. The [CuII(Pir)2] compound is therefore a stronger radical scavenger than either HPir or CuCl2. A molecular mechanics (MM) analysis of the gas phase structures of neutral HPir, its zwitterionic (HPir+-) and anionic (Pir-) forms, and some CuII-piroxicam complexes based on X-ray structures allowed calculation of force constants. The most stable structure for HPir has a ZZZ conformation similar to that found in the CuII (and CdII complexes) in the solid state as well as in the gas phase. The structure is stabilized by a strong H bond which involves the N(amide)-H and O(enolic) groups. The MM simulation for the [CuII(Pir)2(DMF)2] complex showed that two high repulsive intramolecular contacts exist between a pyridyl hydrogen atom of one Pir- molecule with the O donor of the other ligand. These interactions activate a transition toward a pseudo-tetrahedral geometry, in the case the apical ligands are removed. On refluxing a suspension of [CuII(Pir)2(DMF)2] in acetone a brown microcystalline solid with the Cu(Pir)2·0.5DMF stoichiometry was in fact prepared. 13C spin-lattice relaxation rates of neutral, zwitterionic and anionic piroxicam, in DMSO solution are explained by the thermal equilibrium between the three most stable structures of the three forms, thus confirming the high quality of the force field. The EPR spectrum of [CuII(Pir)2(DMF)2] (DMSO/GLY, 2:1, v/v, 298 and 110 K) agrees with a N2O2+O2 pseudo-octahedral coordination geometry. The EPR spectrum of [CuII(Pir)2·0.5DMF agrees with a pseudo-tetrahedral coordination geometry. The parameters extracted from the room temperature spectra of the solution phases are in agreement with the data reported for powder and frozen solutions. The extended-Hückel calculations on minimum energy structures of [CuII(Pir)2(DMF)2] and [CuII(Pir)2] (square planar) revealed that the HOMOs have a relevant character of dx2−y2. On the other hand the HOMO of a computer generated structure for [CuII(Pir)2] (pseudo-tetrahedral) has a relevant character of dxy atomic orbital. A dxy orbital is better suited to allow a dπ-pπ interaction to the O2- anion. Therefore this work shows that the anti-inflammatory activity of piroxicam could be due in part to the formation of [CuII(Pir)2] chelates, which can exert a SOD-like activity.


Introduction
Copper plays important roles in the humans (it is the third most abundant d-block element, after iron and zinc) as well as in all the other living organisms. This metal is present in several types of enzymes which are involved in oxydo-reduction processes, oxygen transport, copper storage, etc. 1 Free radicals as 02", OH', R02" (R H or an organic group), are released in important physiological and pathological phenomena such as cell respiration, pemxidation of membrane lipids, exposure to radiation. 2 Probably the most efficient superoxide scavenger of the enzyme superoxide dismutase (SOD) family is (Cu,Zn)-SOD which has been found in animals, plants and primitive organisms. 3,4 The X-ray structure of bovine erythrocyte (Cu,Zn)-SOD and computer modeling have led to a proposed catalytic mechanism of the 02" dismutation in which the substmte links covalently to the Cu" centers (it is oxidized to 02, first step). 3 Piroxicam (feldene, Pfizer; Scheme 1, HPir) is a powerful anti-inflammatory agent. Scheme 1. Neutral piroxicam, HPir, in the 4,16-EZE conformation.
It has been shown that copper complexes of anti-inflammatory anti-arthritic drugs are more active than their parent compounds. 5 Furthermore, toxicological studies revealed that anti-inflammatory copper complexes are less toxic than inorganic forms of copper. 5 These facts have led to the hypothesis that copper complexes of non-steroidal anti-inflammatory drugs are the active species in vivo. 5,6 On the basis of such arguments, we reasoned that recently prepared and structurally characterized M(ll)-piroxicam (M Fe, Co, Ni, Cu, Zn and Cd) chelates 7 could have (Cu,Zn)-SOD like activity.
Measurements of the reactivity of [Cu"(Pir)2(DMF)2] with 02" and other oxygen radicals, such as HO', HO0", as well as with singlet-02, revealed that, on a molar basis, the Cu" derivative is a far better scavenger than piroxicam.
Electron paramagnetic (EPR, X-band) and nuclear magnetic (1H, 13C NMR) resonance spectroscopies, as well as molecular mechanics (MM) and extended-HiJckel molecular orbital (EHMO) methods, were used to understand the structure of the coordination sphere in solution and the mechanism of (Cu,Zn)-SOD like activity of the Cu"-piroxicam complexes. R. Cini, R. Pogni, R. Basosi, A. Donati, C. Rossi, L. Sabadini, L. Rollo, Metal Based Drugs S. Lorenz#ti, R. Gelli and R. Marcolongo

Results and Discussion
Oxygen radical scavenger activity (ORSA).

CuCI2
The ratio of ICS values for HPir and [Cu"(Pir)2(DMF)2]/HPir, and CuCI2 and [Cu"(Pir)2(DMF)2]/HPir ranges between 12.6 and 19.5 (unstimulated) and between 8.4 and 11.9 (stimulated), and 5.0 " It has to be noted that DMSO/GLY mixtures were used as solvents to test the anti-inflammatory activity the Cu"-salicylate species dermally applied to rats. 8 As the Pir/Cu molar ratio of the solution is ca. 10:1, it can be reasonably assumed that all the metal is II o linked to the Pir" anion in the form of Cu(Pir)2, on the basis of the value of 2 (10.23, 35  Oxygen Radical Scavenger Activity, EPR, NMR, Molecular Mechanics and Extended-Hiickel Molecular Orbital Investigation of the BISiroxicam)Copper (11) Complex (respectively). These data show that ORSA for [CuU(Pir)2(DMF)2] is much higher than ORSA for HPir, and it is higher than ORSA for copper(ll) chloride, too. Furthermore two other experimental facts have to be noted and analyzed. First, the DMSO and GLY solvents have their own ORSA activity (see Table I). The ratio of ICS values for PBS/DMSO/GLY (2/2/1, v/v/v) and PBS mixtures is 3.3.10 .2 (stimulated); however, the ICS values for the solvent system are large enough to allow accurate measurements of scavenger activity.
Second, the stimulating effect of ZYM in the presence of just HPir does not exist (ratio of ICS values 1) whereas it is detectable for the [Cu(Pir)2(DMF)2]/HPir system (the ratio of ICS values ranges from 1.1 to 2.3). It has to be noted that HPir gave inhibition against some activators of receptor-mediated rat leukocyte aggregation. 10 It is reasonable that a type of PMN receptor HPir interaction inhibits also the oxygen radical stimulation activity of ZYM. However such a type of mechanism is much less efficient when copper(ll) species are present. The affinity of copper(ll) species for PMNs is able to delete (at least in part) the inhibitory effect of HPir.

Molecular Mechanics
Neutral Piroxicam, HPir. Rotations around the C(3)-C(14), C(14)-N(16) and N(16)-C(2') bond axes (Scheme 1) of the solid state structure, 11a produced a total of 43 reliable independent conformations. Structure (ETot 39.52 kcal/mol) has a EZE 12 conformation (Scheme 1) and nicely superimposes with the experimental structure (RMS 0.089 . , for S00, and N atoms). A Hbond interaction links the O(15) and O(17) atoms. The conformation of structure II (38.38) is EZZ. The ETot(I)-ETot(ll) difference (1.14 kcal/mol) is mostly due to the repulsive van der Weals interaction between O(15) and H(3') in I. III (37.82, ZZE) and IV (Figure la, 36.70, ZZZ) are stabilized by the strong H-bond which links the N(16) and O(17) atoms. V (37.60, ZZZ) has the methyl group cis to the (S)O2 oxygen atoms; it is destabilized in comparison with IV because the repulsive interaction between O(15) and the methyl goup hydrogen atoms. It has to be pointed out that the most stable structure for neutral HPir, zwitterionic HPir +" and anionic Pir" has a ZZZ conformation, whereas the solid state structures of neutral HPir have a EZE conformation. 11 This discrepancy may be explained by the intermolecular N(16)...O(1) hydrogen bonds which stabilizes the solid state structure.
The analysis of the NMR data (see below) is in good agreement with the MM results for the HPir, HPir +" and Pit" molecules. This prompted us to apply the force field used for the Pit" moieties, to the [Cu(Pir)2(DMF)2] molecule (see below). Furthermore, the present NMR investigation and a recently published NMR work on piroxicam 14 suggest that both in chloroform solution (preponderance of neutral HPir, and low solvation effects) and in polar solvents such as DMF, DMSO, DMSO/H20 (high solvation effects, and presence of both neutral HPir and zwitterionic HPir+'), the ZZZ conformation is prominent. This type of ligand conformation facilitates the N202 chelation of the Cu 2+ cation, in the solution phase.
[Cu(Pir)2(DMF)2]. The energy minimized structure of the complex molecule with an equatorial N202 coordination set (Figure 2, 77.56, ZZZ) nicely superimposes with the X-ray structure 7 (RMS 0.08 , , based on Cu, S, N, and O atoms). On the basis of the EPR data (see below) it is argued that the brown crystalline powder Cu=(Pir)2.0.5DMF consists mostly of distorted tetrahedml Cu(Pir)2 NMR spectroscopy.
The experimental carbon spin-lattice relaxation rates for a 0.15M piroxicam solution in DMSO and the dipolar contribution to the relaxation (see Experimental) are reported in Table I1.
" The reduction of the CuLcomplex to Cu species (see reaction of anhydrous CuCI 2 with acetone to produce CuCI)15 is excluded by: (1) the existence of an unpaired electron located on the metal, as detected by EPR spectroscopy (see below); (2) the fast reversibility to the green species when treated with DMF or DMSO. 4? Vol. 2, No. 1, 1995 Oxygen Radical Scavenger Activity, EPR, NMR, Molecular Mechanics and Extended-Hilckel In order to analyse the experimental carbon relaxation rotes on the basis of the equilibria between stable conformations the theoretical spin-lattice relaxation rate R1T of the most significant quaternary carbons of pimxicam was computed by taking into account all the intramolecular protoncarbon dipolar interactions within a cut off distance of 3.5 and by using the R1T ];piR1 relationship (Pi is the fractional population of each conformation and R1 is the theoretical contribution related to a specific interaction). The dipolar theoretical terms related to each 1H.13C interaction for five quaternary carbons of pimxicam as well as the proton-carbon distances of the nuclei involved in the dipolar interactions are reported in Table III. proton-carbon interaction modulated by a x c 1.9 x 10"1. 7 Dipolar proton-carbon interaction modulated by a x c 9 x 10"11. 8 Contribution due to the zwitterionic form (33% abundance).
A good agreement between experimental and theoretical values was usually found (some discrepancy was observed for the C(3) carbon). It has to be pointed out that the contribution of the zwitterionic HPir +" form (33%)14 was also considered in the calculation of R1T for C(2'). A similar approach brought also to a good agreement between observed R1DD and computed R1T values for Pir'. Furthermore, the calculated atomic charges of both HPir and Pir" (Extended-Huckel method (see below) applied to the most stable structures) describes with accuracy the chemical shift properties observed for the two species (see supplementary material available from the authors).
These results confirm that the previously proposed equilibrium between neutral HPir and zwitterionic HPir +" 14 and the present MM analysis are basically correct.

EPR spectroscopy
The EPR spectrum of [Cu'(Pir)2(DMF)2]/HPir in DMSO/GLY (110 K, Figure 3a, Table IV) shows the typical pattern of pseudo-octahedral (tetragonal elongation) copper(ll) complexes in which three of the expected four hyperfine transitions in the parallel region (low-field range) are visible whereas the fourth component (MI +3/2) is masked by the overlap with A_L features in the high-field end of the spectrum.  Table IV. The room temperature spectrum of slow tumbling [Cu(Pir)2(DMF)2]/HPir in DMSO/GLY, ( Figure  3b, Table IV) is typical of a quasi-immobilized species for the high viscosity of the solution ('c 833 xl0 "12 sec). The EPR spectra of the Cu=-complex used for ORSA studies were simulated by using computer programs for: a) rigid limit, and b) slow motion conditions and taking into account the simultaneous presence of the two copper isotopes (63Cu (69%) and 65Cu (31%); nuclear spin 3/2; ratio of nuclear moments for 63Cu and 65Cu, 1.07). Since the geometry of the coordination sphere is crucial for defining ORSA of metal complexes, the EPR spectrum of the Cu"(Pir)2.0.5DMF powder (Table IV) was also deeply investigated. It clearly reveals a distorted tetrahedral geometry. 16

Extended-Hckel Molecular Orbital Analysis and Cu==-O2" Interaction
The HOMO in the [Cu=(Pir)2(DMF)2], [Cu(Pir)2(DMF)] (square pyramidal), and [Cu"(Pir)2] (square planar) molecules has a relatively high Cu(dx2.y 2) character (Table V, main contributions from donor atom orbitals: O15(py), Nl'(px)). These results are in agreement with previously reported theoretical investigations on Cu-complexes.17 The eventual formation of a CuL02" bonding interaction is highly hindered by the presence of the apical (amide) ligands. Furthermore the interaction of the metal complexes with the 02" ion is also not permitted by the symmetry of the HOMOs of the two species.
For example, the 02" anion can approach the metal center of the square planar [Cu"(Pir)2] at the apical sites, but owing to the high py and Pz character of the oxygen orbitals (x is the direction of the superoxide 0-0 bond axis, Scheme 2a, Table V) Table V) has some Cu(dxy) character (main contributions from ligand atom orbitals: O15, O17, C3, and N2). The O2" ion and the complex molecule could thus interact via a p=-d= overlap. A distorted tetrahedral geometry could be a convenient compromise for a suitable Cu-O2" bond and weak O2"'"Pir" repulsions. Some distortion from pure tetrahedral geometry for a d 9 configuration is also demanded by the Jahn-Teller theorem. 17b It has to be pointed out that the coordination geometry of Cu" in (Cu,Zn)-SOD, as obtained from X-Ray diffraction, is pseudo-tetrahedral, 3 and computer simulated docking investigations showed that the O2" ion and the CN" enzyme inhibitor fitted one and both the available sites around the metal center, respectively. 3 We infer that the anti-inflammatory activity of piroxicam passes through the 02" (and other oxygen radicals) scavenger activity of its Cull-complexes.
Owing to the local high concentration of pimxicam in the cell (during the drug supply) and to the low amount of free Cu 2+ ion, the presence of [Cu"(Pir)2] (pseudo-tetrahedral coordination geometry) as the active species is highly probable.

Experimental Part
Materials.
X-band EPR spectra were obtained with a Bruker 200D SRC X-band spectrometer using a high sensitivity Bruker ER 4108 TMH cavity. Microwave frequencies were measured through a XL Microwave Model 3120 counter. Magnetic field was calibrated with a MJ-140 magnetometer by Jagmar (Poland). The temperature was controlled by using a Bruker variable temperature unit ER 4111 VT. The spectrometer was interfaced with a PSI2 Technical Instruments Hardware computer and the data acquired using the EPR data system CS-EPR produced by Stelar Inc. (Mede, Italy).
The spectra were simulated through the COSMOS (low motion), 18 and CUSIMNE (rigid limit)19 programs, implemented on a Compaq Deskpro 486/50L personal computer with a 8-MByte memory and a 50-MHz clock. NMR Spectroscopy. A Varian XL-200 spectrometer was used for recording 1H and 13C spectra.
Carbon spin-lattice relaxation rates were obtained by using the inversion recovery (r-x-n/2-t)n pulse sequence. R1 values were calculated by computer-fitting of the relaxation curves. NOE values were determined through the equation: NOE (Iz-Io)/lo (Iz and I0 represent the peak intensities measured under continuous and gated proton decoupling, respectively). A 5% experimental error was estimated for R1 and NOE measurements. The fractional dipolar contribution to the carbon spin-lattice relaxation rates, 7 DD, was determined by comparing the experimental NOE with the theoretical expected value for 13C nuclei totally relaxed by 1H-13C dipolar interaction. The dipolar contribution to the experimental carbon-spin lattice relaxation rate was calculated by using the following equation" R1DD xDD.Rlexp. The correlation times modulating the C-H magnetic interactions were computed from the R1DD values by using standard C-H distances (1.08 .) and the equations reported in Ref 20. All NMR chemical shift values were referred to internal DMSO-d6.
Oxygen Radical Scavenger Activity. A stock solution of the [Cu(Pir)2(DMF)2] complex was prepared by dissolving 9.4 mg of the complex (1.08.10 "2 mmol) and 26.5 mg of HPir (8.01.10 -2 mmol) in 10 mL of DMSO. 0.1 mL of the stock solution was mixed with 4 mL of DMSO, 2 mL of GLY and with PBS up to a total volume of 10 mL (20C); final analytical concentrations: CCu 1.08. 10 -5 M, CPir 10.17"10 -5 M. R. Cini, R. Pogni, R. Basosi, A. Donati, C. Rossi, L. Sabadini, L. Rollo, Metal Based Drugs S. Lorenzini, R. Gelli and R. Marcolongo A stock solution of HPir was obtained by mixing 26.5 mg of the drug (8.01.10 -2 mmol) with 10 mk of DMSO. 0.1 mL of the stock solution was diluted to 10 mL with the DMSO/GLY/PBS solvent system by following the procedure above reported for the solution of [Cu==(Pir)2(DMF)2]; final analytical concentration:  A stock solution of CuCI2.2H20 was prepared by dissolving 4.26 mg of the compound (2.50.10 -2 mmol) in 25 mL of DMSO. 0.1 mL of the stock solution was mixed with DMSO, GLY and PBS by using the same procedure listed above for [Cu=(Pir)2(DMF)2] and HPir; final analytical concentration: CCu 1.00.10 -5 M.
The oxygen free radical (produced by human PMNs) scavenger activity of [Cu"(Pir)2(DMF)2]/HPir, HPir and CuCI2 in DMSO/GLY/PBS (see above) was measured through the chemiluminescence technique, by using a chemiluminometer Berthold Multi-Biolumat LB 9505C. Preparation of PMN. PMN were separated from blood by using the following procedure. 5 mL of blood were mixed with 3.5 mL of PMP in a 10 mL test tube. The mixture was centrifugated at 450+500g (30 mins, 20_2C). The PMN contained in the lower ring were rinsed with a PBS solution. The residual erythrocytes were then distroyed by adding a solution of ammonium chloride (0.83 g/100 mL of water). The purified PMN were then added to the sample solution. Preparation of the LUM solution. 2 mg of LUM were dissolved in NaOH 0.01M (10 mL). The proper amount of the stock solution was added to the sample solution in order to get a concentration of 1.10 -4 M. Opsonization of the cells by Zymosan. 5 mg of ZYM were mixed with 4.5 mL of PBS solution and with 0.5 mL of plasm (obtained during the separation of PMN from blood through centrifugation, see above; plasm is the lightest fraction). The mixture was incubated at 37C for 30 mins and then centrifugated at 900g for 10 mins. The pellet of ZYM was rinsed twice with PBS and then suspended in PBS (0.5 mL). Chemiluminescence measurements. The samples containing the PMNs (106/mL), LUM, the drugs and eventually the stimulator (ZYM) in the DMSO/GLY/PBS solvent were tested (at 37C) through the chemiluminometer and the intesity of the light emitted was recorded for a period of 40 mins.
The integrated signal was computed via the computer program Berthold LB 9505 C Version 4.07. Molecular Mechanics. The strain energies of the free ligand and the metal-complex molecules were computed as the sums: ETot Eb + Ee + E + Enb + Ehb; E'Tot ETot + E (Eb, Ee, E, Enb, Ehb e E are the bond length deformation, the valence angle deformation, the torsional angle deformation, the non bonding interaction, the hydrogen bond interaction and the electrostatic contributions, respectively). The fome field used was MM2. 21 Modification and extension of the fome field was necessary in order to take into account the interactions between the metal center and the donor atoms. The proper fome field parameters were obtained via a trial and error procedure which brought to an excellent agreement between calculated and observed structures. The new fome field parameters used in this study are reported in Table VI. They are in good agreement with those previously used in molecular mechanics studies on other metal complexes with N and/or O donor atoms. 22 All the torsional angles A-B-C-D with A or D Cu were assigned to(sional terms equal 0; those with' B or C Cu were fixed at the experimental solid state values. The total energy (ETot or E'Tot) was minimized with the block diagonal matrix Newton-Raphson method until the root mean square value (RMS) of the first derivative vector was less than 0.01 kJ//. The starting structures were those found for the solid state via single crystal X-ray diffraction for HPir (Ref. 11a, see also 11b) and its zwitterionic form, 13 and for the [Cu(Pir)2(DMF)2] complex. 7 The calculations were carried out by using the MacmModel (MMOD) package version 3.023 implemented on a VAX 6600 computer with a graphic ouput on an Evans & Sutherland PS390.
The atomic-charge calculations were performed by using the extended-HiJckel method via the ICONC&INPUTC 24 program (see below). List of selected atomic charges are available as supplementary material.
As the E contribution did not produce any appreciable improvement in the MM analysis, all the calculations reported in this work do not take into account the E term. This approximation is often applied for calculations of metal-complex molecules.22a, b Extended-I-ckel Calculations.-The molecular orbital calculations were performed through ICONC&INPUTC 24 program implemented on a VAX 6600 computer. The parameters used were those reported in Table VII The self-consistent charge iteration calculation on the Cu atom was performed on the complex molecule. The VSIE (valence state ionization energy) parameters for the Cu atom are those included in ICONC&INPUTC. 24 The geometry was kept fixed for all the calculations. Selected bond distances and angles for the more stable structure used for the calculations are reported in the supplementary material.
In order to simplify the MO analysis of the complex molecules the C(5), (3(6), C (7), and C(8) atoms of pimxicam (Scheme 1) were removed; the valences of C(9) and (3(10) (linked each other by a double bond) were saturated by H atoms. DMF molecules were replaced by formamide molecoules.