Synthesis and carbonic anhydrase activating properties of a series of 2-amino-imidazolines structurally related to clonidine1.

The Carbonic Anhydrase (CA, EC 4.2.1.1) activating properties of histamine have been known for a long time. This compound has been extensively modified but only in few instances the imidazole ring has been replaced with other heterocycles. It was envisaged that the imidazoline ring could be a bioisoster of the imidazole moiety. Indeed, we report that clonidine, a 2-aminoimidazoline derivative, was found able to activate several human CA isoforms (hCA I, IV, VA, VII, IX, XII and XIII), with potency in the micromolar range, while it was inactive on hCA II. A series of 2-aminoimidazoline, structurally related to clonidine, was then synthesised and tested on selected hCA isoforms. The compounds were inactive on hCA II while displayed activating properties on hCA I, VA, VII and XIII, with KA values in the micromolar range. Two compounds (10 and 11) showed some preference for the hCA VA or VII isoforms.


Introduction
Carbonic anhydrases (CAs, EC 4.2.1.1) are metallo-enzymes widespread in all life kingdoms. These enzymes catalyse a plethora of reactions, among which the reversible hydration of CO 2 is the most important one 1 . The active site contains the cofactor, a metal ion (usually Zn 2þ ) which coordinates a water molecule responsible, once activated as hydroxide ion, of the nucleophilic attack onto carbon dioxide. Eight genetically different families have been found (a-i); 15 isoforms belonging to the a class have been characterised in humans 1,2 .
CAs have been drug targets since more than 70 years; inhibitors of these enzyme are used for the treatment of oedema, glaucoma and epilepsy but several new therapeutic applications are under study 3 . In recent years, the attention has been focussed also on activators of these enzymes, despite the fact that CA are among the most efficient enzymes known. In fact, genetic deficiencies of several CA isoforms were reported in the last decades (reviewed in Refs. [3,4]), and in principle a loss of function of these enzymes could be treated with CA selective activators (CAAs). In addition, there is evidence that CA activation improves cognitive performance [5][6][7][8][9][10] . However, the influence of CA on these processes is complex since also inhibitors have been found to improve memory deficits in animal models (reviewed in Ref. [11]); these findings point out the need for isoform selective inhibitors or activators to elucidate the role of CA isoforms in cognitive processes. Other possible applications of CAAs could be in the formation of artificial tissues 12 and in CO 2 capture and sequestration processes 13 .
Histamine (HST, Chart 1) was among the first reported activators, whose interaction mode was elucidated by means of X-ray crystallography 14 . The adduct with hCA II revealed a complex network of H-bonds involving the Zn-bound water molecule, His64 and the imidazole ring of the activator, which is located far away from the metal ion, in a region approaching the edge of the active site cavity. X-ray crystallographic studies have later shown that also other activators bind in this area 4 .
As common structural feature, CAAs possess flexible tails decorated with protonable moieties, with pK a values spanning between 6 and 8. The molecule of histamine has been extensively modified, placing substituents on the imidazole C atoms and on the NH 2 group, showing that the latter is not essential, since it can be largely modified to keep or improve potency (reviewed in Ref. [4]). Only in few instances the imidazole ring has been replaced by another heterocycle, such as a thiadiazole ring 15 .
In search for bioisosters of the imidazole moiety, our attention was attracted by the imidazoline ring. This feature is present in a well-known drug, Clonidine (CLO, Chart 1), which is clinically used as an antihypertensive agent being an agonist at the central a2adrenergic receptor, but it is able to interact with other targets, such as the imidazoline binding sites and the hyperpolarizationactivated cyclic nucleotide gated channels 16,17 . Therefore, we decided to measure the potential CA activating properties of this compound, finding that CLO behaves as CAA on several CA isoforms (Table 1). Encouraged by this positive outcome, we synthesised a series of 2-substituted imidazolines (compounds 1-24, Chart 1) and tested their activity on five different hCA isoforms. The ubiquitous cytosolic enzymes CA I and II, the mitochondrial CA VA, which is associated with the glucose homeostasis, 18 the cytosolic CA VII which is particularly abundant in the CNS and has been recently demonstrated to have a protective role against oxidative damage, 19 and the cytosolic CA XIII, which is particularly expressed in the reproductive organs 20,21 were selected.

Chemistry
All melting points were taken on a B€ uchi apparatus and are uncorrected. NMR spectra were recorded on a Brucker Avance 400 spectrometer (      30 The imidazoline 1a was acylated on one imidazoline-nitrogen atom to obtain 23 as reported by Gomez-San Juan et al. 30

CA activation
An Sx.18Mv-R Applied Photophysics (Oxford, UK) stopped-flow instrument has been used to assay the catalytic activity of various CA isozymes for CO 2 hydration reaction 32 . Phenol red (at a concentration of 0.2 mM) was used as indicator, working at the absorbance maximum of 557 nm, with 10 mM Hepes (pH 7.5, for a-CAs) 33-37 as buffers, 0.1 M NaClO 4 (for maintaining constant ionic strength), following the CA-catalysed CO 2 hydration reaction for a period of 10 s at 25 C. The CO 2 concentrations ranged from 1.7 to 17 mM for the determination of the kinetic parameters and inhibition constants. For each activator at least six traces of the initial 5-10% of the reaction have been used for determining the initial velocity. The uncatalysed rates were determined in the same manner and subtracted from the total observed rates. Stock solutions of activators (at 0.1 mM) were prepared in distilled-deionised water and dilutions up to 1 nM were made thereafter with the assay buffer. Enzyme and activator solutions were pre-incubated together for 15 min prior to assay, in order to allow for the formation of the enzyme-activator complexes. The activation constant (K A ), defined similarly with the inhibition constant K I , can be obtained by considering the classical Michaelis-Menten equation (Equation (1), which has been fitted by non-linear least squares by using PRISM 3: where  (2): where v 0 represents the initial velocity of the enzyme-catalysed reaction in the absence of activator [33][34][35][36][37][38][39] . This type of approach to measure enzyme-ligand interactions is in excellent agreement with recent results from native mass spectrometry measurements 40 .

Results and discussion
3.1. Chemistry 2-Amino imidazolines 2-20 were prepared by reacting primary and secondary amines with 2-(methylthio)-4,5-dihydro-1H-imidazole hydroiodide 1a 23 or its N 1 -methyl derivative 1b 24 using tetrahydrofuran or an excess of amine as solvents (Scheme 1). The final compounds were obtained as hydroiodide salts in different yields. Synthetic details are reported in the experimental section. Reactions with 3-or 4-nitrobenzyl amine or with dibenzyl amine were unsuccessful.

CA activating profile
The stopped-flow method 32 has been used for assaying the CO 2 hydration activity catalysed by different CA isoforms; the results are expressed as K A (activation constant, mM). The activating profile of CLO is reported in Table 1, in comparison with HST. CLO behaved as an activator on several CA isoforms (I, IV, VA, VII, IX, XII and XIII), with K A values in the range 7.8-136 mM, while on CA II it was inactive up to a dose of 200 mM. In particular, the isoforms most sensitive to CLO were CA VII and CA XIII.
The K A values of the synthesised compounds are reported in Table 3. All compounds have been tested as hydroiodide salts, with the exception of 21 (oxalate), 22 and 23 (free bases), and 24 (as HBr salt). As CLO, none of the compounds was active on hCA II at the highest tested concentration, while on the other isoforms the compounds showed K A values mainly in the low-medium range, allowing to derive the following structure-activity relationships.

hCA I
The K A value of CLO on this isoform was 76.3 mM. Removal of both chlorine atoms abolished activity, since 22 (R 2 ¼ Ph) was inactive when tested up to a 150 mM concentration. On the contrary, inserting a CH 2 unit between the exocyclic N atom and the Ph ring of 22 improved the activity: in fact, 2 (K A 4.18 mM), 18 times more potent than CLO, was one of the most potent compounds on this isoform among the newly synthesised analogues. The elongation of the methylene chain gave compounds less active than 2; interestingly, a chain formed by 3 CH 2 units was tolerated (5, K A 36.7 mM) while a CH 2 CH 2 chain was not (3, K A >150 mM). Side-chain branching abolished activity (S-4, R-4: K A >150 mM). Methylation of the exocyclic N a atom was tolerated when R 2 was a phenethyl group (7: K A 68.6 mM) but not when R 2 was a benzyl moiety (6: K A >150 mM). Also the methylation of the endocyclic N 1 atom gave contrasting results, since compounds 8 and 12 were more active than their non-methylated analogues 3 and 5, but 11 (R 2 ¼ benzyl) showed a fourfold decrease of activity with respect to 2 (11: K A 16.9; 2: 4.18 mM). Double methylation was productive for the phenethyl analogues (compare 9 with 3, 7 and 8); for the benzyl analogue 10 this modification improved the activity only with respect to 6. The aminoethyl derivative 21 (K A 3.87 mM) was the most potent compound on the hCAI isoform; the addition of a benzyl moiety on the primary amine group abolished activity (19, K A > 100 mM), and also the N 1 -methyl analogue 20 was inactive.
Aromatic substitution on the benzyl moiety did not improve the potency: in fact, while the 4-Cl derivative 13 was equiactive with 11, a 4-OMe (14) or 4-F substituent (15) increased from 2 to 3.5 times the K A values. The same substituents in the meta position reduced to a higher extent or abolished the activity. As far as the sulphur analogues 1a, 23 and 24 are concerned, the small methyl group seemed tolerated, not the bulkier benzyl moiety (24, K A >150 mM). The basicity of the amidine group appeared to be not crucial, since the NH and the N-acetyl derivatives (1a and 23, respectively) were equipotent.

hCA VA
The K A value of CLO on this isoform was 42.6 mM. The removal of both chlorine atoms did not affect activity, since 22 (R 2 ¼ Ph, K A 52.7 mM) was almost equipotent with CLO. Also the insertion of a CH 2 unit between the exocyclic N atom and the Ph ring of 22 did not substantially modified potency (2, K A 45.7 mM). On this isoform, the majority of the compounds showed good activating properties, with potency higher than CLO: the K A values of 24, 23, 3-5, 7-9, 11 and 12 were in the range 3.7-17.2 mM.
The most potent compound was 10 (K A 0.9 mM), a benzyl derivative carrying a methyl group on both N 1 and N a atoms; this compound was 47 times more active than CLO. The removal of the exocyclic N a -Me group decreased 4 times the activity (11, K A 3.7 mM), while the removal of the N 1 -methyl group was more detrimental: as a matter of fact, compounds 6 (K A 40.5 mM) and 2 (K A 45.5 mM) were about 40 times less potent than 10 (K A 0.9 mM). On the contrary, the degree of methylation did not substantially affect the potency of the phenylethyl and phenyl propyl derivatives, since compounds 3, 5, 7, 8 and 12 had K A values in the range 9.9-17.2 mM. Similarly, aromatic substitution on the benzyl moiety slightly decreased the potency without substantial modulation, the K A values of compounds 13-18 being 2-4 times higher than 11. Side-chain branching (compounds S-4 and R-4) improved the activity on this isoform, and a small enantioselectivity was observed: the R-enantiomer was twice more potent as the S-isomer. A benzyl moiety on the terminal amino group of 21 (K A 31.2 mM) increased the activity, as analogue 19 and its N 1 -methyl derivative 20 were about 3 times more potent than the parent compound.
As far as the sulphur derivatives are concerned, the replacement of the N a H moiety of 2 (K A 45.5 mM) with S (24, K A 11.1 mM) brought a fourfold improvement in activity. Acetylation of the N 1 Scheme 2. Synthesis of compounds 22 and 23. nitrogen was also favourable, as 23 was twice more potent than 1a.

hCA VII
The K A value of CLO on this isoform was 8.4 mM. All the tested compounds showed activation properties on this isoform, with K A values between 0.9 and 91.6 mM. The least potent compound was the primary amine 21 (K A 91.6 mM), whose activity was however improved by adding a benzyl group on the terminal NH 2 moiety (19, K A 41.7 mM) and a methyl group on the endocyclic N atom (20, K A 29.0 mM). The removal of chlorine atoms of CLO reduced 4 times the activity (22, R 2 ¼ Ph, K A 32.6 mM) while the separation of the phenyl and N a H moieties by means of a CH 2 unit did not substantially modified potency (3, K A 35.2 mM). On the contrary, the potency increased by elongating the chain from 1 to 3 CH 2 units (3, K A 35.2 mM; 5, K A 11.4 mM) and by adding a methyl group on the N a H moiety: with the latter modification the potency of 3 (K A 18.9 mM) and of 2 (K A 35.2 mM) were increased 3 (6, K A 11.0 mM) and 8 times (7, K A 2.4 mM), respectively. Side-chain branching did not substantially affect activity, since S-4 and R-4 were equipotent with 2. Methylation on the endocyclic N atom was the most effective modification in this set of molecules: as a matter of fact, with this structural change the K A value of 2 (K A 35.2 mM) was reduced 39 times, and 11 (K A 0.9 mM) resulted in the most potent compound on this isoform. The same modification was also effective on the phenylpropyl derivative 5 (K A 11.4 mM), whose activity was increased 4 times (12, K A 3.1 mM). The double methylation on the N 1 and N a atoms gave potent compounds (10, K A 6.5 mM and 9, K A 2.6 mM) even if the K A values are, respectively, 7 and 3 times lower than that of 11. Aromatic substitution on the benzyl moiety of 11 was detrimental for activity, as compounds 13-18 were 19-33 times less potent than 11. As far as the sulphur analogues 1a, 23 and 24 are concerned, their K A values were in the range 30.9-46.7 mM, not better that the other tested 2-aminoimidazoline derivatives. Attempts to crystallise adducts of 7, 11 and 12 with hCA VII are ongoing.

hCA XIII
This is the isoform most sensitive to CLO among those studied (K A 7.8 mM). As it happened on the hCA I isoform, the removal of both chlorine atoms, to give 22, abolished activity. Several other compounds resulted inactive when tested at concentrations up to 100 mM, i.e. the sulphur derivatives, the polar aminoethyl derivative 21, and all the compounds having both the N 1 and N a atoms as secondary amines, with the exception of the lipophilic phenylpropyl derivative 5 (K A 24.3 mM). The activity of 21 could be restored by adding a benzyl moiety on the primary amino-group (19, K A 20.1 mM) and a methyl group on the endocyclic N atom (20, K A 16.3 mM). Also the methylation of the phenethyl analogue 3 on the N a atom re-established activity, giving 7 (K A 6.5 mM) which resulted the most potent compound of the series on this isoform. Methylation on the endocyclic N 1 atom gave compounds 8, 11 and 12 whose potency ranged from 10.9 to 31.0 mM, the most potent being the derivative carrying a phenylpropylamino side chain (12). Methylation on both N 1 and N a atoms or aromatic substitution on the benzyl moiety did not improve activity.
As far as selectivity is concerned, the two compounds showing submicromolar K A values displayed also interesting selectivity profiles: 10 was more active on hCA VA with respect to hCA I (33 times), II (>100 times), VII (7 times), and XIII (19 times), while 11 showed a preference for hCA VII over hCA I (19 times), II (>100 times), VA (4 times), and XII (21 times).

Conclusions
We have synthesised a series of 2-aminoimidazolines, structurally related to Clonidine, and tested them on five different hCA isoforms (I, II, VA, VII and XIII). As the lead compound, none of the newly synthesised molecules was active on the ubiquitously expressed CA II; on the contrary, the compounds showed activity in the micromolar range on the other tested CA isoforms. Structure-activity relationships were derived, which were different on the various isoforms, suggesting that it could be possible, in this class of compounds, to find molecules, selective for a particular CA isoform. Indeed, from these preliminary modifications it has been possible to find two compounds, 10 and 11, with a promising preference towards, respectively, CA VA and VII. Work is underway to improve both potency and selectivity, in order to find new pharmacological tool to activate specific CA isoforms in pathologies characterised by their loss of functionality.