Facile Access to Fe(III)-Complexing Cyclic Hydroxamic Acids in a Three-Component Format

Cyclic hydroxamic acids can be viewed as effective binders of soluble iron and can therefore be useful moieties for employing in compounds to treat iron overload disease. Alternatively, they are analogs of bacterial siderophores (iron-scavenging metabolites) and can find utility in designing antibiotic constructs for targeted delivery. An earlier described three-component variant of the Castagnoli—Cushman reaction of homophthalic acid (via in situ cyclodehydration to the respective anhydride) was extended to involve hydroxylamine in lieu of the amine component of the reaction. Using hydroxylamine acetate and O-benzylhydroxylamine was key to the success of this transformation due to greater solubility of the reagents in refluxing toluene (compared to hydrochloride salt). The developed protocol was found suitable for multigram-scale syntheses of N-hydroxy- and N-(benzyloxy)tetrahydroisoquinolonic acids. The cyclic hydroxamic acids synthesized in the newly developed format have been tested and shown to be efficient ligands for Fe3+, which makes them suitable candidates for the above-mentioned applications.


Introduction
The ability of cyclic hydroxamic acids (N-hydroxylactams) to chelate metal ions in general-and Fe 3+ in particular-defines their utility in drug design [1]. One principal avenue in this regard is based on the recognition of the similarity of synthetic cyclic hydroxamic acids to bacterial siderophores-special metabolites excreted by bacteria to scavenge and deliver iron for the microorganism's nutritional needs [2]. Conjugating such moieties to antibiotics helps shuttle those inside bacteria and thus overcome resistance of the latter to the antibacterial agent [3]. Besides said application in resistance-free antibiotic design, selective, non-toxic iron chelators are needed as treatments for hereditary iron overload disease [4]. Unfortunately, streamlined and multicomponent methods to prepare N-hydroxylactams are lacking. Among the existing synthetic methods for hydroxamic acid, intramolecular nucleophilic cyclization onto O-protected acyclic hydroxamic acids [5], nitroso moiety insertion in cyclic ketones [6], and ring-closing methathesis of bis-olefinic hydroxamic acids [7] can be mentioned.
Imines 1 are known to condense with α-C-H anhydrides of dicarboxylic acids 2 to give polysubstituted lactams 3 [8]. In the recent literature, this powerful reaction has been regarded as a name reaction-the Castagnoli-Cushman reaction (CCR) [9]-to acknowledge its origination from the research efforts of Neil Castagnoli and Mark Cushman over 40 years ago [10,11]. Homophthalic anhydride (HPA, 4) is one of the most frequently employed anhydrides in this reaction, which normally gives rise to trans-configured tetrahydroisoquinolonic (THIQ) acids 5 ( Figure 1) [12,13]. These have Recently, we successfully replaced the imine component in the CCR of 4 with oximes 6 and obtained, after 24 h reflux in toluene, good to excellent yields of the respective N-hydroxy THIQ acids 7 [17] which are representative of cyclic hydroxamic acids. The forcing conditions required in order to obtain 7 were in sharp contrast with ambient temperature normally sufficient for the preparation of 5. This was justified [17] by the need to re-generate HPA (4) from the initial, rapidly formed Oacylation product 7′ (Scheme 1). Scheme 1. Reaction of oximes 6 with HPA (4) and its mechanistic interpretation [15].
More recently, we employed refluxing toluene for in situ dehydration of homophthalic acid which allowed, for the first time, preparing THIQ acids 5 in a true multicomponent format (i.e., by mixing an amine, an aldehyde and homophthalic acids) without a need for preliminary imine synthesis or the use of hydrolytically prone HPA (Scheme 2) [18]. Scheme 2. Synthesis of THIQ acids 5 in a three-component format from homophthalic acid [18].
These findings encouraged us to investigate the preparation of N-hydroxy THIQ acids 7 from homophthalic acid and oximes and possible applicability in this case of the same three-component format as presented in Scheme 2. Herein, we present the results of these findings and present characterization of compounds 7 with respect to their iron-binding properties, which validates them as potential candidates for iron overload disease treatment or the design of siderophore-based constructs for antibiotic delivery. Recently, we successfully replaced the imine component in the CCR of 4 with oximes 6 and obtained, after 24 h reflux in toluene, good to excellent yields of the respective N-hydroxy THIQ acids 7 [17] which are representative of cyclic hydroxamic acids. The forcing conditions required in order to obtain 7 were in sharp contrast with ambient temperature normally sufficient for the preparation of 5. This was justified [17] by the need to re-generate HPA (4) from the initial, rapidly formed O-acylation product 7 (Scheme 1).
Molecules 2019, 24, x 2 of 14 These have a documented utility in medicinal chemistry (in therapeutic areas such as cancer [14], malaria [15], and diabetes [16]). Recently, we successfully replaced the imine component in the CCR of 4 with oximes 6 and obtained, after 24 h reflux in toluene, good to excellent yields of the respective N-hydroxy THIQ acids 7 [17] which are representative of cyclic hydroxamic acids. The forcing conditions required in order to obtain 7 were in sharp contrast with ambient temperature normally sufficient for the preparation of 5. This was justified [17] by the need to re-generate HPA (4) from the initial, rapidly formed Oacylation product 7′ (Scheme 1). Scheme 1. Reaction of oximes 6 with HPA (4) and its mechanistic interpretation [15].
More recently, we employed refluxing toluene for in situ dehydration of homophthalic acid which allowed, for the first time, preparing THIQ acids 5 in a true multicomponent format (i.e., by mixing an amine, an aldehyde and homophthalic acids) without a need for preliminary imine synthesis or the use of hydrolytically prone HPA (Scheme 2) [18]. Scheme 2. Synthesis of THIQ acids 5 in a three-component format from homophthalic acid [18].
These findings encouraged us to investigate the preparation of N-hydroxy THIQ acids 7 from homophthalic acid and oximes and possible applicability in this case of the same three-component format as presented in Scheme 2. Herein, we present the results of these findings and present characterization of compounds 7 with respect to their iron-binding properties, which validates them as potential candidates for iron overload disease treatment or the design of siderophore-based constructs for antibiotic delivery. Scheme 1. Reaction of oximes 6 with HPA (4) and its mechanistic interpretation [15].
More recently, we employed refluxing toluene for in situ dehydration of homophthalic acid which allowed, for the first time, preparing THIQ acids 5 in a true multicomponent format (i.e., by mixing an amine, an aldehyde and homophthalic acids) without a need for preliminary imine synthesis or the use of hydrolytically prone HPA (Scheme 2) [18]. These have a documented utility in medicinal chemistry (in therapeutic areas such as cancer [14], malaria [15], and diabetes [16]). Recently, we successfully replaced the imine component in the CCR of 4 with oximes 6 and obtained, after 24 h reflux in toluene, good to excellent yields of the respective N-hydroxy THIQ acids 7 [17] which are representative of cyclic hydroxamic acids. The forcing conditions required in order to obtain 7 were in sharp contrast with ambient temperature normally sufficient for the preparation of 5. This was justified [17] by the need to re-generate HPA (4) from the initial, rapidly formed Oacylation product 7′ (Scheme 1). Scheme 1. Reaction of oximes 6 with HPA (4) and its mechanistic interpretation [15].
More recently, we employed refluxing toluene for in situ dehydration of homophthalic acid which allowed, for the first time, preparing THIQ acids 5 in a true multicomponent format (i.e., by mixing an amine, an aldehyde and homophthalic acids) without a need for preliminary imine synthesis or the use of hydrolytically prone HPA (Scheme 2) [18]. Scheme 2. Synthesis of THIQ acids 5 in a three-component format from homophthalic acid [18].
These findings encouraged us to investigate the preparation of N-hydroxy THIQ acids 7 from homophthalic acid and oximes and possible applicability in this case of the same three-component format as presented in Scheme 2. Herein, we present the results of these findings and present characterization of compounds 7 with respect to their iron-binding properties, which validates them as potential candidates for iron overload disease treatment or the design of siderophore-based constructs for antibiotic delivery. Scheme 2. Synthesis of THIQ acids 5 in a three-component format from homophthalic acid [18].
These findings encouraged us to investigate the preparation of N-hydroxy THIQ acids 7 from homophthalic acid and oximes and possible applicability in this case of the same three-component format as presented in Scheme 2. Herein, we present the results of these findings and present characterization of compounds 7 with respect to their iron-binding properties, which validates them as potential candidates for iron overload disease treatment or the design of siderophore-based constructs for antibiotic delivery.

Results and Discussion
The initial reaction of oxime 6a with homophthalic acid in toluene proved rather encouraging and yielded, after 16 h reflux and cooling to room temperature, 72% of the desired product 7a [17] (Scheme 3). The reaction proceeded via formation of HPA (4), which was supported by the results of a separate experiment, where homophthalic acid was refluxed in toluene with azeotropic removal of water to form corresponding anhydride in quantitative yield. This experiment can be also regarded as a new method for the preparation of homophthalic anhydride.

Results and Discussion
The initial reaction of oxime 6a with homophthalic acid in toluene proved rather encouraging and yielded, after 16 h reflux and cooling to room temperature, 72% of the desired product 7a [17] (Scheme 3). The reaction proceeded via formation of HPA (4), which was supported by the results of a separate experiment, where homophthalic acid was refluxed in toluene with azeotropic removal of water to form corresponding anhydride in quantitative yield. This experiment can be also regarded as a new method for the preparation of homophthalic anhydride. An attempt to perform the same reaction by mixing homophthalic acid with p-anisaldehyde and hydroxylamine hydrochloride in presence of pyridine (1 equiv.) did not lead to the formation of 7a. Instead, the only product discernible by 1 H-NMR (Nuclear Magnetic Resonance) analysis of the reaction mixture was tetracyclic adduct formed, presumably, via the earlier described [19] Perkin-Michael domino transformation (Scheme 4).
The observed complication is likely the result of the inefficient formation of the intermediate oxime 6 in the non-polar reaction medium (due to low solubility in it of hydroxylamine hydrochloride) and the ready participation of free p-anisaldehyde in the route leading to tetracyclic adduct. To circumvent this obstacle, we attempted to prepare hydroxylamine acetate and use it in the same transformation. We anticipated that in addition to improved solubility of the acetate salt in toluene, this transformation would not require the use of a base as acetic acid would be conveniently removed from the reaction medium by azeotropic distillation with the solvent. Interestingly, hydroxylamine acetate has been seldom [20] employed for the preparation of oximes despite the obvious advantage of not having to use any base (as is the case with the usual hydroxylamine hydrochloride). The hydrochloride salt of hydroxylamine was converted to its acetate by the action of sodium acetate and reacted with p-anisaldehyde and homophthalic acid in refluxing toluene over 24 h. To our delight, on cooling to r. t., a thick precipitate formed which was isolated by filtration to provide 65% yield of compound 7a. The same reaction was attempted at reflux in benzene and xylenes. While the former conditions led to <15% conversion over 24 h, the latter gave an appreciable amount of N-deoxygenation product. Considering this and the results of the reaction monitoring at An attempt to perform the same reaction by mixing homophthalic acid with p-anisaldehyde and hydroxylamine hydrochloride in presence of pyridine (1 equiv.) did not lead to the formation of 7a. Instead, the only product discernible by 1 H-NMR (Nuclear Magnetic Resonance) analysis of the reaction mixture was tetracyclic adduct formed, presumably, via the earlier described [19] Perkin-Michael domino transformation (Scheme 4).

Results and Discussion
The initial reaction of oxime 6a with homophthalic acid in toluene proved rather encouraging and yielded, after 16 h reflux and cooling to room temperature, 72% of the desired product 7a [17] (Scheme 3). The reaction proceeded via formation of HPA (4), which was supported by the results of a separate experiment, where homophthalic acid was refluxed in toluene with azeotropic removal of water to form corresponding anhydride in quantitative yield. This experiment can be also regarded as a new method for the preparation of homophthalic anhydride. An attempt to perform the same reaction by mixing homophthalic acid with p-anisaldehyde and hydroxylamine hydrochloride in presence of pyridine (1 equiv.) did not lead to the formation of 7a. Instead, the only product discernible by 1 H-NMR (Nuclear Magnetic Resonance) analysis of the reaction mixture was tetracyclic adduct formed, presumably, via the earlier described [19] Perkin-Michael domino transformation (Scheme 4).
The observed complication is likely the result of the inefficient formation of the intermediate oxime 6 in the non-polar reaction medium (due to low solubility in it of hydroxylamine hydrochloride) and the ready participation of free p-anisaldehyde in the route leading to tetracyclic adduct. To circumvent this obstacle, we attempted to prepare hydroxylamine acetate and use it in the same transformation. We anticipated that in addition to improved solubility of the acetate salt in toluene, this transformation would not require the use of a base as acetic acid would be conveniently removed from the reaction medium by azeotropic distillation with the solvent. Interestingly, hydroxylamine acetate has been seldom [20] employed for the preparation of oximes despite the obvious advantage of not having to use any base (as is the case with the usual hydroxylamine hydrochloride). The hydrochloride salt of hydroxylamine was converted to its acetate by the action of sodium acetate and reacted with p-anisaldehyde and homophthalic acid in refluxing toluene over 24 h. To our delight, on cooling to r. t., a thick precipitate formed which was isolated by filtration to provide 65% yield of compound 7a. The same reaction was attempted at reflux in benzene and xylenes. While the former conditions led to <15% conversion over 24 h, the latter gave an appreciable amount of N-deoxygenation product. Considering this and the results of the reaction monitoring at Scheme 4. Attempted three-component preparation of 7a from hydroxylamine hydrochloride (see ref. [19] for more mechanistic insight).
The observed complication is likely the result of the inefficient formation of the intermediate oxime 6 in the non-polar reaction medium (due to low solubility in it of hydroxylamine hydrochloride) and the ready participation of free p-anisaldehyde in the route leading to tetracyclic adduct. To circumvent this obstacle, we attempted to prepare hydroxylamine acetate and use it in the same transformation. We anticipated that in addition to improved solubility of the acetate salt in toluene, this transformation would not require the use of a base as acetic acid would be conveniently removed from the reaction medium by azeotropic distillation with the solvent. Interestingly, hydroxylamine acetate has been seldom [20] employed for the preparation of oximes despite the obvious advantage of not having to use any base (as is the case with the usual hydroxylamine hydrochloride). The hydrochloride salt of hydroxylamine was converted to its acetate by the action of sodium acetate and reacted with p-anisaldehyde and homophthalic acid in refluxing toluene over 24 h. To our delight, on cooling to r. t., a thick precipitate formed which was isolated by filtration to provide 65% yield of compound 7a. The same reaction was attempted at reflux in benzene and xylenes. While the former conditions led to <15% conversion over 24 h, the latter gave an appreciable amount of N-deoxygenation product. Considering this and the results of the reaction monitoring at different time points, 24 h reflux in toluene was considered to be optimal for the preparation of compound 7a. Thus, these conditions were extended to the preparation of a new series of N-hydroxy THIQ acids 7a-m (Scheme 5).
Molecules 2019, 24, x 4 of 14 different time points, 24 h reflux in toluene was considered to be optimal for the preparation of compound 7a. Thus, these conditions were extended to the preparation of a new series of N-hydroxy THIQ acids 7a-m (Scheme 5).

Scheme 5. Three-component synthesis of N-Hydroxy THIQ acids 7a-m.
An interesting result was obtained in an attempt to involve o-salicylaldehyde in the same reaction. When all the aldehyde starting material was consumed (according to HPLC (High Performance Liquid Chromatography) analysis of the reaction mixture), only a trace amount of anticipated compound was detected. The major product identified and isolated from the reaction mixture was coumarin 8. The same reaction run without hydroxylamine acetate gave identical yield of 8 (67%). It likely that the reaction (described earlier but not interpreted from the mechanistic viewpoint [21]) involves acylation of the phenolic hydroxy group followed by intramolecular Knoevenagel reaction (Scheme 6). Involvement of O-acetyl salicylaldehyde in the same reaction led to de-acetylation and the formation of 8 in 50% yield. The reaction generally worked very well with electron-rich aromatic aldehydes and gave good yields of respective N-hydroxy THIQ acids 7a-k (Scheme 5). Substrates without alkoxy groups (such as tert-butyl-, methoxycarbonyl-, fluoro-, bromo-, and nitro-benzaldehydes,) surprisingly, gave no desired product under these conditions. Increasing the concentration of reactants from 0.1 M to 1 M and the reaction time from 24 to 48 h, allowed us to slightly increase the reaction scope by involving An interesting result was obtained in an attempt to involve o-salicylaldehyde in the same reaction. When all the aldehyde starting material was consumed (according to HPLC (High Performance Liquid Chromatography) analysis of the reaction mixture), only a trace amount of anticipated compound was detected. The major product identified and isolated from the reaction mixture was coumarin 8. The same reaction run without hydroxylamine acetate gave identical yield of 8 (67%). It likely that the reaction (described earlier but not interpreted from the mechanistic viewpoint [21]) involves acylation of the phenolic hydroxy group followed by intramolecular Knoevenagel reaction (Scheme 6). Involvement of O-acetyl salicylaldehyde in the same reaction led to de-acetylation and the formation of 8 in 50% yield.
Molecules 2019, 24, x 4 of 14 different time points, 24 h reflux in toluene was considered to be optimal for the preparation of compound 7a. Thus, these conditions were extended to the preparation of a new series of N-hydroxy THIQ acids 7a-m (Scheme 5).

Scheme 5. Three-component synthesis of N-Hydroxy THIQ acids 7a-m.
An interesting result was obtained in an attempt to involve o-salicylaldehyde in the same reaction. When all the aldehyde starting material was consumed (according to HPLC (High Performance Liquid Chromatography) analysis of the reaction mixture), only a trace amount of anticipated compound was detected. The major product identified and isolated from the reaction mixture was coumarin 8. The same reaction run without hydroxylamine acetate gave identical yield of 8 (67%). It likely that the reaction (described earlier but not interpreted from the mechanistic viewpoint [21]) involves acylation of the phenolic hydroxy group followed by intramolecular Knoevenagel reaction (Scheme 6). Involvement of O-acetyl salicylaldehyde in the same reaction led to de-acetylation and the formation of 8 in 50% yield. The reaction generally worked very well with electron-rich aromatic aldehydes and gave good yields of respective N-hydroxy THIQ acids 7a-k (Scheme 5). Substrates without alkoxy groups (such as tert-butyl-, methoxycarbonyl-, fluoro-, bromo-, and nitro-benzaldehydes,) surprisingly, gave no desired product under these conditions. Increasing the concentration of reactants from 0.1 M to 1 M and the reaction time from 24 to 48 h, allowed us to slightly increase the reaction scope by involving The reaction generally worked very well with electron-rich aromatic aldehydes and gave good yields of respective N-hydroxy THIQ acids 7a-k (Scheme 5). Substrates without alkoxy groups (such as tert-butyl-, methoxycarbonyl-, fluoro-, bromo-, and nitro-benzaldehydes,) surprisingly, gave no desired product under these conditions. Increasing the concentration of reactants from 0.1 M to 1 M and the reaction time from 24 to 48 h, allowed us to slightly increase the reaction scope by involving 4-(tert-butyl)benzaldehyde and 4-(methoxycarbonyl)benzaldehyde in the developed approach and to prepare corresponding N-hydroxy THIQ acids 7l and 7m in good yields (54 and 51% respectively). However, still the reaction was not applicable to more electron-poor substrates.
The 1 H-and 13 C-NMR analysis of corresponding reaction mixtures showed that in case of electron-poor aldehydes the major reaction products are nitriles (formed by dehydration of oximes). To prevent this side reaction, we replaced hydroxylamine acetate with O-benzylhydroxylamine (Scheme 6). To our delight this allowed to involve previously unreactive electron-poor aldehydes (even 4-fluoro-and 4-nitro-benzaldehydes). Following this new modified protocol (also performed at 1 M concentration of reactants) seven novel N-benzyloxy THIQ acid derivatives 9a-g have been prepared in high and good yields (Scheme 7). For all prepared compounds 7a-m and 9a-g trans-configuration was concluded from 3 J HH coupling constant values (~1-2 Hz) between protons at positions C3 and C4, which is consistent with our previous data [17]. 4-(tert-butyl)benzaldehyde and 4-(methoxycarbonyl)benzaldehyde in the developed approach and to prepare corresponding N-hydroxy THIQ acids 7l and 7m in good yields (54 and 51% respectively). However, still the reaction was not applicable to more electron-poor substrates.
The 1 H-and 13 C-NMR analysis of corresponding reaction mixtures showed that in case of electron-poor aldehydes the major reaction products are nitriles (formed by dehydration of oximes). To prevent this side reaction, we replaced hydroxylamine acetate with O-benzylhydroxylamine (Scheme 6). To our delight this allowed to involve previously unreactive electron-poor aldehydes (even 4-fluoro-and 4-nitro-benzaldehydes). Following this new modified protocol (also performed at 1 M concentration of reactants) seven novel N-benzyloxy THIQ acid derivatives 9a-g have been prepared in high and good yields (Scheme 7). For all prepared compounds 7a-m and 9a-g transconfiguration was concluded from 3 JHH coupling constant values (~1-2 Hz) between protons at positions C3 and C4, which is consistent with our previous data [17]. Additionally, two types of post-modifications have been investigated for the prepared compounds 7 and 9: decarboxylation (Scheme 8) and debenzylation (Scheme 9). Attempts to perform decarboxylation of compound 7a under previously reported conditions [22] lead to formation of complex mixture of products resulting from side reactions of oxidation and dehydration (supported by HRMS (High Resolution Mass Spectrometry) and 1 H-NMR data) (Scheme 8). At the same time, decarboxylation can be smoothly conducted for O-benzylated compound 9b providing compound 10 in 77% yield, thus demonstrating another advantage of O-benzyloxime-based protocol for preparation of N-hydroxy THIQ acids derivatives.

Scheme 8. Decarboxylation of compounds 9a and 9b.
We also have investigated the possibility of removal of benzyl protective group from prepared compounds (Scheme 9). N-Benzyloxy THIQ acids 9b and 10 were successfully converted to corresponding OH-hydroxamic acids 7a and 11 via hydrogenolysis under standard conditions with excellent yields. Surprisingly, attempted O-debenzylation of compound 7c only gave the product of N-dehydroxylation 12 (Scheme 9). Additionally, two types of post-modifications have been investigated for the prepared compounds 7 and 9: decarboxylation (Scheme 8) and debenzylation (Scheme 9). Attempts to perform decarboxylation of compound 7a under previously reported conditions [22] lead to formation of complex mixture of products resulting from side reactions of oxidation and dehydration (supported by HRMS (High Resolution Mass Spectrometry) and 1 H-NMR data) (Scheme 8). At the same time, decarboxylation can be smoothly conducted for O-benzylated compound 9b providing compound 10 in 77% yield, thus demonstrating another advantage of O-benzyloxime-based protocol for preparation of N-hydroxy THIQ acids derivatives.

4-(tert-butyl)benzaldehyde and 4-(methoxycarbonyl)benzaldehyde in the developed approach and
to prepare corresponding N-hydroxy THIQ acids 7l and 7m in good yields (54 and 51% respectively). However, still the reaction was not applicable to more electron-poor substrates. The 1 H-and 13 C-NMR analysis of corresponding reaction mixtures showed that in case of electron-poor aldehydes the major reaction products are nitriles (formed by dehydration of oximes). To prevent this side reaction, we replaced hydroxylamine acetate with O-benzylhydroxylamine (Scheme 6). To our delight this allowed to involve previously unreactive electron-poor aldehydes (even 4-fluoro-and 4-nitro-benzaldehydes). Following this new modified protocol (also performed at 1 M concentration of reactants) seven novel N-benzyloxy THIQ acid derivatives 9a-g have been prepared in high and good yields (Scheme 7). For all prepared compounds 7a-m and 9a-g transconfiguration was concluded from 3 JHH coupling constant values (~1-2 Hz) between protons at positions C3 and C4, which is consistent with our previous data [17]. Additionally, two types of post-modifications have been investigated for the prepared compounds 7 and 9: decarboxylation (Scheme 8) and debenzylation (Scheme 9). Attempts to perform decarboxylation of compound 7a under previously reported conditions [22] lead to formation of complex mixture of products resulting from side reactions of oxidation and dehydration (supported by HRMS (High Resolution Mass Spectrometry) and 1 H-NMR data) (Scheme 8). At the same time, decarboxylation can be smoothly conducted for O-benzylated compound 9b providing compound 10 in 77% yield, thus demonstrating another advantage of O-benzyloxime-based protocol for preparation of N-hydroxy THIQ acids derivatives.

Scheme 8. Decarboxylation of compounds 9a and 9b.
We also have investigated the possibility of removal of benzyl protective group from prepared compounds (Scheme 9). N-Benzyloxy THIQ acids 9b and 10 were successfully converted to corresponding OH-hydroxamic acids 7a and 11 via hydrogenolysis under standard conditions with excellent yields. Surprisingly, attempted O-debenzylation of compound 7c only gave the product of N-dehydroxylation 12 (Scheme 9). 4-(tert-butyl)benzaldehyde and 4-(methoxycarbonyl)benzaldehyde in the developed approach and to prepare corresponding N-hydroxy THIQ acids 7l and 7m in good yields (54 and 51% respectively). However, still the reaction was not applicable to more electron-poor substrates. The 1 H-and 13 C-NMR analysis of corresponding reaction mixtures showed that in case of electron-poor aldehydes the major reaction products are nitriles (formed by dehydration of oximes). To prevent this side reaction, we replaced hydroxylamine acetate with O-benzylhydroxylamine (Scheme 6). To our delight this allowed to involve previously unreactive electron-poor aldehydes (even 4-fluoro-and 4-nitro-benzaldehydes). Following this new modified protocol (also performed at 1 M concentration of reactants) seven novel N-benzyloxy THIQ acid derivatives 9a-g have been prepared in high and good yields (Scheme 7). For all prepared compounds 7a-m and 9a-g transconfiguration was concluded from 3 JHH coupling constant values (~1-2 Hz) between protons at positions C3 and C4, which is consistent with our previous data [17]. Additionally, two types of post-modifications have been investigated for the prepared compounds 7 and 9: decarboxylation (Scheme 8) and debenzylation (Scheme 9). Attempts to perform decarboxylation of compound 7a under previously reported conditions [22] lead to formation of complex mixture of products resulting from side reactions of oxidation and dehydration (supported by HRMS (High Resolution Mass Spectrometry) and 1 H-NMR data) (Scheme 8). At the same time, decarboxylation can be smoothly conducted for O-benzylated compound 9b providing compound 10 in 77% yield, thus demonstrating another advantage of O-benzyloxime-based protocol for preparation of N-hydroxy THIQ acids derivatives. We also have investigated the possibility of removal of benzyl protective group from prepared compounds (Scheme 9). N-Benzyloxy THIQ acids 9b and 10 were successfully converted to corresponding OH-hydroxamic acids 7a and 11 via hydrogenolysis under standard conditions with excellent yields. Surprisingly, attempted O-debenzylation of compound 7c only gave the product of N-dehydroxylation 12 (Scheme 9). Scheme 9. Debenzylation of compounds 9b, 10, and 7c.
We also have investigated the possibility of removal of benzyl protective group from prepared compounds (Scheme 9). N-Benzyloxy THIQ acids 9b and 10 were successfully converted to corresponding OH-hydroxamic acids 7a and 11 via hydrogenolysis under standard conditions with excellent yields. Surprisingly, attempted O-debenzylation of compound 7c only gave the product of N-dehydroxylation 12 (Scheme 9).
Compounds 7 synthesized using the developed procedure are direct analogs of bacterial siderophores [23] and, therefore, can be potentially regarded as new agents for iron overload disease therapy and as precursors for the design of "Trojan horse" antibiotics [24] against drug resistant bacteria. Therefore five selected compounds-7d,f-h,j which represent different substitution patterns in 3-aryl moiety (4 types of di-substituted and one tri-substituted), were tested for iron(III) binding properties using mole-ratio method to prove this assumption (Figure 2a,b). Upon addition of ferric nitrate to solution of ligand 7 in aqueous methanol a color change from colorless to purple was observed. This corresponds to formation of Fe(III)-7 complex, which is characterized by a new absorption band with maximum around 500 nm (Figure 2b). All ligands 7 in contrast to their complexes with iron do not absorb in the visible region. Representative example of UV-Vis spectrum of free ligand 7 and its changes upon addition of Fe 3+ ions is shown on Figure 2a for compound 7d. These spectra for other tested compounds 7 are reported in ESI (Electronic Supporting Information, Figures S1-S3). In all cases two isosbestic points were observed around 470 and 600 nm at C Fe = 0.5-5 × 10 −4 M. Compounds 7 synthesized using the developed procedure are direct analogs of bacterial siderophores [23] and, therefore, can be potentially regarded as new agents for iron overload disease therapy and as precursors for the design of "Trojan horse" antibiotics [24] against drug resistant bacteria. Therefore five selected compounds-7d,f-h,j which represent different substitution patterns in 3-aryl moiety (4 types of di-substituted and one tri-substituted), were tested for iron(III) binding properties using mole-ratio method to prove this assumption (Figure 2a,b). Upon addition of ferric nitrate to solution of ligand 7 in aqueous methanol a color change from colorless to purple was observed. This corresponds to formation of Fe(III)-7 complex, which is characterized by a new absorption band with maximum around 500 nm (Figure 2b). All ligands 7 in contrast to their complexes with iron do not absorb in the visible region. Representative example of UV-Vis spectrum of free ligand 7 and its changes upon addition of Fe 3+ ions is shown on Figure 2a for compound 7d. These spectra for other tested compounds 7 are reported inESI (Electronic Supporting Information, Figures S1-S3). In all cases two isosbestic points were observed around 470 and 600 nm at CFe = 0.5-5 × 10 −4 M. Applying the mole-ratio method allowed us to determine not only the stoichiometry of observed complexes but also the associated binding constants [17,25]. Absorbance at characteristic wavelength was plotted as function of [Fe 3+ ]/[ligand] ratio to give the curves shown in Figure 3a. The binding process involves two equilibria-the first one corresponding to 1:1 complex, and the second one to 1:2 complex. Stepwise formation constants K1 and K2 were estimated in the range of 10 5 −10 7 M −1 and 10 4 −10 5 M −1 , respectively. These values were calculated using computer nonlinear curve-fitting of the absorbance values taken from experimental mole-ratio plots on Figure 3a to previously derived Equations (1)-(3) describing formation of ML2 complex (See Material and Methods Section for details). The example of such fitting (compound 7d) is presented on Figure 3b. The respective plots with curve-fitting for other tested compounds 7f-h,j are presented in ESI (Figures S4-S8) The results on spectrophotometric investigation of iron(III) binding properties for compounds 7 are summarized in Table 1. The obtained Kf values (~10 10 -10 11 M −1 ) were found to be in accordance with our previous results [17] for similar compounds. No significant dependence of Kf values on substitution pattern was found. Applying the mole-ratio method allowed us to determine not only the stoichiometry of observed complexes but also the associated binding constants [17,25]. Absorbance at characteristic wavelength was plotted as function of [Fe 3+ ]/[ligand] ratio to give the curves shown in Figure 3a. The binding process involves two equilibria-the first one corresponding to 1:1 complex, and the second one to 1:2 complex. Stepwise formation constants K 1 and K 2 were estimated in the range of 10 5 −10 7 M −1 and 10 4 −10 5 M −1 , respectively. These values were calculated using computer nonlinear curve-fitting of the absorbance values taken from experimental mole-ratio plots on Figure 3a to previously derived Equations (1)-(3) describing formation of ML 2 complex (See Material and Methods Section for details). The example of such fitting (compound 7d) is presented on Figure 3b. The respective plots with curve-fitting for other tested compounds 7f-h,j are presented in ESI (Figures S4-S8) The results on spectrophotometric investigation of iron(III) binding properties for compounds 7 are summarized in Table 1. The obtained Kf values (~10 10 -10 11 M −1 ) were found to be in accordance with our previous results [17] for similar compounds. No significant dependence of K f values on substitution pattern was found.

General Considerations
All commercial reagents were used without further purification. NMR spectra were acquired using Bruker Avance III spectrometer (Billerica, MA, USA) ( 1 H: 400.13 MHz; 13 С: 100.61 MHz; chemical shifts are reported as parts per million (δ, ppm); solvents-DMSO-d6 or CDCl3, the residual solvent peaks were used as internal standards: 2.50 ppm for 1 H and 39.52 ppm for 13 C (DMSO-d6) or 7.26 ppm for 1 H and 77.16 ppm for 13 C (CDCl3); multiplicities are abbreviated as follows: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad; coupling constants, J, are reported in Hz. Mass spectra were acquired using Bruker microTOF spectrometer (ionization by electrospray, positive ions detection; Billerica, MA, USA). Melting points were determined in open capillary tubes on Stuart SMP50 Automatic Melting Point Apparatus (Stone, UK). Homophthalic acid, hydroxylamine hydrochloride, O-benzylhydroxylamine, aldehydes, and Fe(NO3)3•9H2O were obtained from commercial sources. All reactions were performed in air, unless otherwise noted. Analytical data obtained for compounds 7a, 7b, 7k and 7m have been consistent with those reported in our previous work [17].

General Considerations
All commercial reagents were used without further purification. NMR spectra were acquired using Bruker Avance III spectrometer (Billerica, MA, USA) ( 1 H: 400.13 MHz; 13 C: 100.61 MHz; chemical shifts are reported as parts per million (δ, ppm); solvents-DMSO-d 6 or CDCl 3 , the residual solvent peaks were used as internal standards: 2.50 ppm for 1 H and 39.52 ppm for 13 C (DMSO-d 6 ) or 7.26 ppm for 1 H and 77.16 ppm for 13 C (CDCl 3 ); multiplicities are abbreviated as follows: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad; coupling constants, J, are reported in Hz. Mass spectra were acquired using Bruker microTOF spectrometer (ionization by electrospray, positive ions detection; Billerica, MA, USA). Melting points were determined in open capillary tubes on Stuart SMP50 Automatic Melting Point Apparatus (Stone, UK). Homophthalic acid, hydroxylamine hydrochloride, O-benzylhydroxylamine, aldehydes, and Fe(NO 3 ) 3 ·9H 2 O were obtained from commercial sources. All reactions were performed in air, unless otherwise noted. Analytical data obtained for compounds 7a, 7b, 7k and 7m have been consistent with those reported in our previous work [17].

Hydroxylamine Acetate
Hydroxylamine hydrochloride (10.0 g, 144 mmol) was dissolved in deionized water (5 mL). The solution was added, on stirring, to a solution of sodium acetate (11.8 g, 144 mmol) in water (5 mL). The resulting clear solution was concentrated to dryness. The solid residue was suspended in anhydrous methanol (20 mL) and filtered to remove sodium chloride. The filtrate was concentrated in vacuo to give hydroxylamine acetate (13.

General Procedure for the Preparation of N-Hydroxy THIQ Acids 7l,m
A mixture of homophthalic acid (3.6 g, 20 mmol), hydroxylamine acetate (1.86 g, 20 mmol) and the respective aldehyde (20 mmol) in toluene (20 mL) was placed in a pre-heated oil bath and refluxed for 48 h in a flask equipped a Dean-Stark distilling trap. A thick precipitate which formed on cooling the reaction mixture to −20 • C was isolated by filtration, washed with hexane, and crystallized from aqueous methanol to provide pure title compounds.
Experimental curves were fitted to equation 3 corresponding to 1:2 metal-to-ligand complex formation using nonlinear curve-fitting performed in ThordarsonFittingProgram [25]. The program is based on the iterative adjustment of calculated values of absorbance (A) to observed values using Equation (3) previously derived [25] from Equations (1) and (2), where K 1 and K 2 are stepwise formation constants; ε ∆ML = ε ML − ε L and ε ∆ML2 = ε ML2 − ε ML, where ε i are molar absorptivities of corresponding species; C L and C M are analytical concentrations of ligand and Fe 3+ respectively and L is free ligand concentration.

Conclusions
In summary, we have developed a practically convenient, three-component approach to N-hydroxytetrahydroisoquinoline (THIQ) acids via a variant of the Castagnoli-Cushman reaction involving in situ cyclodehydration of homophthalic acid with concomitant formation of an oxime in refluxing toluene. Using hydroxylamine acetate or O-benzylhydroxylamine in lieu of the hydroxylamine hydrochloride typically employed to prepare oximes was key to the success of the reaction. For prepared N-bezyloxy THIQ acids decarboxylation and debenzylation reactions were additionally investigated. Five selected cyclic hydroxamic acid compounds produced in the course of this study have been profiled and confirmed as ligands for Fe 3+ . Thus, a new, practically simple and flexible approach to potential iron overload disease treatments or analogs of bacterial siderophores for antibiotic delivery has been developed.