Structure-Activity Relationships for Some Diamine, Triamine and Schiff Base Derivatives and Their Copper(II) Complexes

Ethylenediamine (en), putrescine (pu), diethylenetriamine (dien), dipropylenetriamine (dpta), spermidine (spmd) and their CuII compounds as well as the Schiff bases with 2-furaldehyde (dienOO), 2- thiophenecarboxaldehyde (dienSS) and pyrrole-2-carboxaldehyde (dienNN) of dien and that of dpta with 2- thiophenecarboxaldehyde (dptaSS), were prepared and characterised. They were tested against Bacillus substilis, Bacillus cereus, Staphylococcus aureus, Escherichia coli, Proteus vulgaris and Xanthomonas campestris as antibacterial reagents, the highest activity being exhibited by Cu(dptaSS)(NO3)2 complex, which acts as antibiotic. In the antiproliferative tests (vs. T47D,L929 and BHK21/c13 cell lines) the best results were obtained with Cu(dptaSS)2+ and Cu(dienSS)2+. Electronic structure calculations gave for dptaSS and dienSS the higher negative charges on the N atoms. The counter-ions (Br-, NO3- and SO42-) play an important role by modulating the reagent's selectivity versus the bacteria [Gram(+) or Gram(-)], but they have no effect on the antiproliferative activity.


Introduction
Di-and triamines and, in general, polyamines are biologically occurring substances resulting from the metabolism of ornithine, bleomycin, etc. [1]. They are excellent complexating reagents, capable of coordinating to a number of transition metal ions, including Cu II [2,3]. Cyclic amines such as imidazole, histamine, etc. and their co-ordination compounds have been extensively studied [2] but the acyclic amines, despite their occurrence as terminal amines in the metabolism of natural products such as bleomycin [4], have received less attention. So far the growth inhibitory activity of the metalion complexes has been correlated to the atomic mass [5], electronegativity [6], atomic radii [7], the number of unpaired electrons of the metal ion [8], and, in general, to the electronic structure parameters of the transition metal compounds [9]. Some of these correlations, however have been questioned [2b,9] and are no longer generally accepted (vide inffa). Important physicochemical parameters [10] of biofunctional ligands are their lipid solubility, charge distribution, polarisability and steric parameters. Transition metal compounds with such ligands are very potent chemotherapic reagents, the ptlI-compounds taking first place 11]; the CuII compounds take a lower ranking position, but they are also important because of their "plasticity" [12], i.e. they are capable of assuming different shapes with different co-ordination numbers and thus adapt to the substrate 13]. At least two factors are known [6,10] to be important for the growth inhibitory activity of co-ordination compounds: (a) the ease to adopt certain geometry and thus avoid possible steric hindrance during their physiological action; (b) the partition coefficient between lipid and water media; it depends strongly on the charges of the atoms in the active molecule. Ligands, co-ordinated to a metal ion, can modify both factors and thus enhance or lower their growth inhibitory properties. Thus, for example, a ligand can reduce the charge on the metal ion through electron donation and in this way it can ease the permeability of the metal ion into the cell. Electrostatic interaction of a positively charged complex with adjacent negative groups of the biomolecules can confer rigidity to the fluxional ionic adducts which result from such an interaction and modify their biological function.
In this paper we report the synthesis, characterisation and bacterial growth inhibitory properties of some acyclic di-and triamines and their Schiff base derivatives as well as their Cu II compl.exes. An attempt has been made also to trace their antiproliferative activity and to relate their inhibitory and cytostatic properties with their electronic structure parameters.
The following ligands and their Cu II compounds have been studied: Schiffbases of dien with 2-furaldehyde (dienOO), 2-thiophene-carboxaldehyde (dienSS) and pyrrol-2-carboxaldehyde(dienNN) and their Cu II compounds as well as a Schiff Base of dpta with 2-thiophenecarboxaldehyde (dptaSS) and its Cu II compound were also prepared. Some of the CuII compounds have been isolated with different counter-ions (Br', NO3" and SO42").
Several of our ligands and their Cu II compounds have been studied in solution as to the possible equilibria and stability [14][15][16]. For the pu complex several species were found, CuHL, CuLz(OH)+, and at pH 9 precipitation is known to occur [16].
In the above selection we have two diamines (en and pu) with chains of different length, and three triamines (dien, dpta and spmd), two of which are symmetric (dien and dpta) with chains of different length, and one of them is asymmetric (spmd). This choice offers a chance to study the effect of number of bonding N atoms, length of aliphatic chain, symmetry of the two branches of chains and the modifying effect of the N-substituents. The compounds were characterised by their IR and electronic spectra, magnetic moments and elemental analyses.
The geometric structures were determined by Molecular Mechanics (MMP2) [17][18][19][20][21], which can take into account the -electronic conjugation in some of the ligands. The MM2 input parameters were used and those of Cu should be specifically mentioned: the bond stretching force constants Cu-N were set to be 0.89 mdyn A1. Bending force constants NCuN were set to 0.25 mdyn rad ! for the angles N-Cu-N=90.0 deg, and 0.35 mdyn tad -1 for the remaining angles. The torsional parameters CCNCu were set to zero, since they had little effect on the energy values. These constants compare well with those obtained from normal co-ordinate analysis for similar Cu-compounds [22]. The molecular structures obtained were compared with available crystal structures [23][24][25][26][27]. The electronic structures of the free ligands were examined by the Austin Model (AM1), which is a version of the MNDO method [28]. The parameters used were those included in the MOPAC 6.0 database. The electronic structures of the Cu II compounds were examined by the Extended Huckel method [29], since no parameters for copper are available for the AM method. The parameters used in the EH calculations were those of Murphy and Fitzpatrick [30]. Although the EH results should be viewed with caution as to their absolute values, they are fully credible used in comparing a series of similarly constituted compounds [29b].
2-Thiophenecarboxaldehyde or pyrrole-2-carboxaldehyde or 2-furaldehyde (20 retool) were mixed with 10 mmol diethylenetriamine or dipropylenetriamine. The resulting product was dissolved in methanol (50 ml) and Cu(NO3)2.3H20, CuSO4.5H20 or CuBr2 in methanol (20 ml) were added. The mixture was stirred for an hour and after 12 h of staying the precipitates were isolated by filtration, washed successively with methanol and ether, and dried in vacuum. The compounds were re-crystallised from methanol. They represent uniform blue crystals or blue crystalline powders with different colours. Analytical results together with the measured magnetic moments and electronic spectra are given in Table I. The compounds Cu(en)z(NO3)2, Cu(pu)z(NO3)2, Cu(spmd)(NO3)2 were prepared by evaporation of a Ctl(NO3)z.3H20 solution in water (50 ml) of the appropriate amine amount as described elsewhere [30].  [32] (vide infra, the discussion of the IR spectra).
Electronic spectra, Electronic spectra (Table I) were recorded on a Shimadzu UV 160A spectrophotometer in the 200-800 nm region using aqueous solutions of the compound. The electronic spectra of the Cu II complexes in aqueous solutions exhibit absorption bands at 610-639 nm and 247-276 nm with another band at 295-298 nm for IV, VI and VII. The 615-639 nm band is of low intensity and is assigned to a d-d transition in 4-co-ordinate CuII, while the 247-276 nm band should be assigned either to a r -r* interligand transition in the Schiffbase ligands or to L ---) M charge transfer transition (vide infra, the electronic structure calculations). The 295-298 nm band is not related to the nature of the ligand or counter ion and should be assigned to rt r* interligand transition.
Magnetic moments. Magnetic susceptibilities of powdered samples were recorded at 25 C by the Faraday method with a home-made balance against Hg[Co(SCN)4] as calibrant. Diamagnetic corrections were estimated from Pascal constants. The measured magnetic moments (Table I) were 1.73 BM for I-IV and 1.83 BM for VII, which is consistent with monomeric structures; 1.61 and 1.66 BM were recorded for V and VI, respectively, possibly indicating the presence of metal-metal interaction through magnetic exchange coupling in polymeric structures.
.Infrared spectra. -IR spectra were recorded with a Perkin-Elmer 1640 FT-IR spectrophotometer in the 200-4000 cm 1 region using KBr pellets (Table II). They show some bands that are typical of diethylenetriamine or dipropylenetriamine, respectively [22]. The 1665 cm "1 band in the free dien ligand spectrum was assigned to stretching the C=N bond in the -CH=N-groups. This band is shifted to lower frequencies (about 55-60 cm"l) upon co-ordination, proving that the -CH=N group nitrogen atoms are coordinated to Cu II In the Cu II nitrate compounds the 1740 and 1760 cm "1 bands prove the presence of uncoordinated nitrate groups while the 1380 and 1350 cm "1 bands indicate a second unidentate nitrate group. These results are in agreement with molar conductivity measurements which show that the complexes behave as 1:1 electrolytes in solution dissociating according to the scheme" Hence the IR spectra, the magnetic and molar susceptibility measurements suggest that Cu I in the CuLnitrate complexes (L is a tridentate ligand) is in a 4-co-ordinate CuN30 chromophore.
For the Cu(dienSS)(SO4) complex the 1145, 1120, 1045, 1030 and 960 cm "1 bands can be assigned to a bridging sulphate group.  The v(Cu-N) band is spread over three peaks in the 515-530 cm "1 range which indicates that the Cu-N bonds are slightly different in a single compound but do not differ much in the different compounds. The data for compounds I and VI prove that the counter-ion has little effect on the Cu-N(dien) bands and the same holds for II and V, for which substituting NO 3" with Br" brings about a decrease in v(Cu-N) of only 5 cm"l. Compounds with the same counter ion (NO3") form the series: compound respective Cu compounds were (within The IR spectra of the free Schiff-base ligands and those of the II experimental error) the same in the regions where the 5-membered C4H40 C4H4NU and C4H4S rings absorb, thus proving that these substituents do not participate in bonding to CuII.

Biological Tests
Antibacterial Activity The antibacterial activity (Table III)  NaC1, 0.01% MgSO4 and 0.1% yeast extract, pH 7). The compounds were dissolved in distilled water with a 2-fold successive serial dilution from 800 to 25 g ml"1. All cultures were incubated at 28 C. Control tests with no active ingredients were also carried out. c Blanks indicate no action for concentrations lower than 800 lag ml'l; 800-400 tg m1-1 indicate weak activity, 200 lag ml l indicate satisfactory activity and 100 lag ml "1 indicate strong activity.
Several findings emerge from Table III. a/Both ligands and complexes were inactive against B.Substilis (Gram +) and Pr. Vulgaris (Gram -). For this reason they were omitted in Table III. b/The ligands pu, spmd, dpta, unlike their Cu II complexes were inactive against all studied bacteria. c/ While there are marked differences in the bacterial screening tests in MSB with different Cu II compounds, the results from tests in MHB indicate very poor performance in all tests (>800 g mll). This can be attributed to the different compositions of the two broths (vide infra). d/The best results were obtained with Cu(dptaSS)(NO3)2 (100 pg ml"1) against B. cereus (Gram +). It is highly selective, since it is inactive against all the other bacteria. Next come Cu(dien)(NO3)2 and Cu(dienSS)(NO3)2 against X. campestris and their action is also highly selective. e/Moderate success (400 g m1-1) is achieved by Cu(dienOO)(NO3)2 against X. Campestris, and by Cu(dienNN)(NO3)2 against S. aureus and X. campestris; Cu(dienSS)(Br)2 acts selectively against X. campestris and Cu(dienSS)(SO4) against S.aureus and X. campestris.
It can be thus concluded that the number of N atoms (en vs. dien), length of chain (dien vs. dpta) of the polyamines is essential for their antibacterial activity: compounds of tridentate ligands with longer aliphatic chains perform better than compounds of bidentate ligands with shorter chains. The N-substituents in the Schiff bases (dien vs. dienSS, dienOO and dienNN), however, seem to be highly effective since the Cu(dienSS) 2+and Cu(dptaSS)2+-species are particularly active.

Base Derivatives and their Copper(II) Complexes
There are significant differences in activity between the copper complex and the corresponding ligand and that differences are highly selective with respect to the studied bacteria. We take a few salient examples: With respect to X. campestris: dien (400 lag ml"l) vs. Cu(dien) (NO3)2 (200 g ml"1); dienSS (800 gg ml"1) vs. Cu(dienSS) (NO3)2 (200 lag ml'l).
In fact the last example shows that the ligand dptaSS is weakly active, while its complex acts as an antibiotic.
The two broth media contain different constituents: MHB contains peptone and starch and MSB contains glucose and phosphate as main constituents. The latter may bind to Cu II and invoke replacement of the ligands of the original complex if the stability constants of CuII with glucose and phosphate are greater than those with the polyamine ligands. Were this the case, all complexes should have the same activity in a given broth. This is exactly the case with MHB but not with MSB. Hence, it may be concluded that the peptone and/or the starch form more stable complexes with CuII, replacing the polyamine ligands. The opposite case holds for the MS broth since presumably the glucose and phosphate complexes of Cu II are less stable than the corresponding polyamine complexes.
Cell Culture and Antiproliferative Activity T47D cells from metastatic pleural effusion of patients were grown in Dulbecco medium plus 10% fetal bovine serum. Mouse fibroblast cells L929 and Baby Hamster Kidney fibroblast were grown in Eagle's Minimal Essential Medium plus 10% fetal bovine serum BHK21/cl3. Antiproliferative activity (Table IV) was evaluated in cells grown on a monolayer. The number of cells was measured by the Trypan Blue method [34]. The ligands show no activity while the complexes behave as growth inhibitors with IDs0 values ranging from 60 to 250 l-tg ml l Group A contains the three Cu(dienSS)-compounds with different counter-ions, while the best performer 2+ in this group s the outsider Cu(dptaSS) -complex. It is thus evident that the counter-ion plays a very modest role and the active species in the first group are Cu(dptaSS) 2+ and Cu(dienSS)2+, both containing the 2thiophenecarboxaldehyde substituent, Cu(dptaSS) 2+ doing slightly better than Cu(dienSS)2+.
The reason for the large differences in activity of the Cu II Schiff-base compounds with different substituents will be examined further by calculations on the molecular and electronic structures which may shed light on the problem.

Molecular and Electronic Structures of the Ligands and the Cu II Compounds
Both the ligands (en, pu, spmd) and their Cu II complexes have been extensively studied (see, for example, the collection in ref. [35]). In our tests the best results were obtained with dpta and dien and Schiff-base dien and dpta and for this reason we shall focus our discussion on the molecular and electronic structures of these ligands and their Cu II complexes. The Cu II dien [23] and dpta [24] complexes were studied before, but their Schiff-base substituted complexes were not.
(a) Comparison of the. Results for the Dien, DienOO, DienNN, DienSS and DptaSS Ligands It is seen from the AHr values in Table V that dpta is the most stable ligand and, among the dien derivatives, dien itself is the most stable one. DienSS ranks last in stability.
Here and everywhere else N A refers to N from the -CH=N-C (N is sp 2 hybridised) or -NH2 (N is sp 3 hybridised) groups; N refers to N from the central C-NH-C group (N is sp 3 hybridised). With the exception of dienSS, MM and AM1 gave practically the same values 3.02-3.98 A for the NA-N distance in dien and substututed diens with no dependence on the substituents. In dpta and dptaSS, the NAN distance depends on the substituent: 3.1 and 3.4 .A, respectively. The NA-N A distances show a very complicated picture: 5.1 A. for dien, 4.6-4.9 A for substituted diens; 4.3 A for dpta and finally 4.4 A for dptaSS. These values probably reflect both the difference in chain length and the repulsion between the bulky N-substituents. or-NH group; N refers to N from the central C-NH-C group; The NANIN A angle is defined by the two NAN lines. The lowest NANIN A angle is obtained for the dptaSS and dpta ligands (88 and 79 deg), the highest NANIN A angle for dien (122 deg). These values correlate with the NAN A distance: the longer the NAN A distance, the larger the NANIN A angle. Further it might be expected that the smaller the NANN A angle, the longer the Cu-N bond should be.
By comparing the charges on the nitrogen atoms it is readily seen (Table VI, the second numbers in the rows refer to the ligands) that the charges on N are almost the same (-0.33 by AM1), while the charge on N A shows a great variation when passing from dien to dienSS, dienOO and dienNN (-0.33,-0.18,-0.17,-0.20); the charge on N A in dien and dpta are practically the same (-0.33 and-0.31, respectively); same trend is observed with dienSS and dptaSS (-0.18, -0.17). The explanation is that N is always sp 3 hybridised and does not participate in r-conjugation with the NA-substituents, while N A is also sp 3 in un-condensed di-and triamines but sp 2 in the Schiff bases, in which electron density is taken away from N A and localised at the 5membered rings, dienOO and dienSS doing slightly better than dienNN. There is no such delocalisation for un-condensed dien and dpta and they show the highest charges on both types of nitrogen atoms.
The difference LUMO-HOMO for the studied ligands roughly correlates with the first UV bands (compare Tables and VI) and for this reason these bands may be assigned tentatively to interligand electron transitions, as suggested when discussing the electronic spectra (vide supra).

Base Derivatives and their Copper(II) Complexes (b) Comparison of the results for the Cu u Co-ordination Compounds
It is seen from the AHf results (Table VII) that the most stable species is Cu(dpta) 2+ (AHf -772 kcal mol "l) with Cu(dien) 2+ (AHf= -716 kcal moll) coming next and the most unstable is Cu(dienSS) 2+ (AHf -228 kcal mol ").
The variations of the charge on the Cu II atom are (Table VI) Cu(dpta) and Cu(dptaSS) respectively. The lowest charge on the Cu atom is in the Cu(dienSS) 2+ unit and it is almost 2-3 times lower in comparison with the Cu charge in the other units (see Table VI). The high negative charge on N A indicates that the Cu-N bonds in Cu(dienSS) are the most ionic ones, which favours higher water-lipid partition coefficients and defines a better growth inhibitory activity [6,10]. It should be noted that there are less electrons on the aldimine nitrogen (qN=0.01) than on the nitrogens which take part in the conjugation with the N-substituents [qN=-0.63 in Cu(dptaSS)2+] and this is exactly the opposite as compared with the free ligands the trend is reversed.  (compare Tables and VI), and since they consists mainly of Cu d-AO they may tentatively assigned to d-d transitions (vide supra).

Correlation between Biological Activity and Electronic Structure
The results from the antibacterial tests suggest that Cu(dptaSS)(NO3): is the best antibacterial agent against B. cereus. Cu(dien)(NO3): and Cu(dienSS)(NO3)2 act best against X..campestris. The antibacterial activity depends on the counter-ion and for this reason it might be suggested that the active species is CuLX, where X is the respective counter-ion included in the first co-ordination sphere.
In the antiproliferative studies, the Cu(dptaSS) 2+ and Cu(dienSS) 2+ species have the highest activity, thus emphasising that the presence of 2-thiophenecarboxaldehyde is essential. In the case of Cu(dienSS) 2+ the counter-ion plays no effect and it may be thus suggested that the species penetrating the cells are CuL, stripped of that ion. The species Cu(dptaSS) 2/ has the highest positive charge on copper and the highest negative charge on N, which favours its partitioning between the water and lipid phases during the tests. The lowest LUMO makes the species an excellent electron acceptor in further reactions with redox partners. In contrast, the charge on Cu in Cu(dienSS) is the lowest and charges on N are also low. The thermodynamic stability seems to be a minor factor. It may thus be concluded that different factors influence the activities of Cu(dptaSS) :+and Cu(dienSS)2/-species. The common feature of the two species is a high HOMO (-11.06 and-11.56 eV, respectively).

Conclusions
The Cu II compounds with triamines tested in the present study are almost uniform as to their molecular structures, the predominant group being a Cu atom surrounded by 3N and one X atom in an almost planar arrangement; however, they show subtle variations as to their electronic structures. They have revealed themselves as selective reagents with respect to some Gram-positive and Gram-negative bacteria, the selectivity being modulated by the counter ions such as NO3, SO42, etc., which can co-ordinate to CuII.
The ionicity of the Cu-N bonds in the tested compounds seems to correlate with their growth inhibitory action the high positive charge on the Cu ion, and the high negative charge on the N atoms improve the antibacterial activity, probably due to better water-lipid partitioning ratios [6,10]. While the species active as bacteriostatic reagents include the counter-ions which complete the co-ordination sphere of CuII up to 4, the species active as cytostatic reagents seem to be stripped of these counter-ions and may be co-ordinatively unsaturated. The site presented at the CuN3 unit and its electronic and molecular parameters may thus be expected to be the determining factor for the interaction with the cyto-substrates.