The First Dimeric Derivatives of the Glycopeptide Antibiotic Teicoplanin

Various dimeric derivatives of the glycopeptide antibiotic teicoplanin were prepared with the aim of increasing the activity of the parent compound against glycopeptide-resistant bacteria, primarily vancomycin-resistant enterococci. Starting from teicoplanin, four covalent dimers were prepared in two orientations, using an α,ω-bis-isothiocyanate linker. Formation of a dimeric cobalt coordination complex of an N-terminal L-histidyl derivative of teicoplanin pseudoaglycone has been detected and its antibacterial activity evaluated. The Co(III)-induced dimerization of the histidyl derivative was demonstrated by DOSY experiments. Both the covalent and the complex dimeric derivatives showed high activity against VanA teicoplanin-resistant enterococci, but their activity against other tested bacterial strains did not exceed that of the monomeric compounds.


Introduction
Glycopeptide antibiotics vancomycin, teicoplanin, oritavancin, dalbavancin and telavancin have strong therapeutic potential in cases of severe infections caused by Grampositive bacteria, such as multidrug-resistant Staphylococcus aureus, or enterococci [1][2][3][4][5][6]. The antibacterial activity of vancomycin and teicoplanin is based mainly on their binding to the D-alanyl-D-alanine terminal part of the pentapeptide of the lipid II unit of the bacterial peptidoglycan, thus inhibiting the transpeptidation and transglycosylation phase of peptidoglycan biosynthesis [6]. It has long been known that most of the glycopeptide antibiotics, e.g., ristocetin [7] and eremomycin [8], form non-covalent, reversible dimers in aqueous solutions, and strong dimerization is considered to be beneficial for antibacterial effects [9]. Multivalency or chelate effects support this view, since after one site is bound, the second binding occurs effectively in an intramolecular way. Induction of ligand-induced formation of higher glycopeptide oligomers was demonstrated both in crystal state [10] and in aqueous solution [11][12][13]. Studies on vancomycin revealed that the non-covalent dimerization aptitude of the vancomycin molecule can be an important factor of its antibacterial activity, as the dimerization promotes its affinity/co-operativity to the ligands, i.e., D-alanyl-D-alanine part of the bacterial peptidoglycan [10,[14][15][16][17][18]. Motivated by these observations, numerous covalent dimers of vancomycin and eremomycin have been synthesized in order to improve the antibacterial activity of the parent molecule [19][20][21][22][23][24][25][26][27][28][29]. Indeed, some of these covalent dimers showed enhanced activity against resistant bacteria, such as vancomycin-resistant enterococci type VanA.
Teicoplanin (1) is not able to form non-covalent dimers because of its apolar tag, and its covalent dimers are also unknown. Teicoplanin used in the therapy is a mixture of five closely related derivatives, mainly consisting of the A 2 component, differing from each other in the N-acyl substituent of one of the D-glucosamine moieties (Scheme 1). The separation of these components is difficult; therefore, synthetic modifications of the teicoplanin mixture also lead to complicated mixtures. Removal of the variously N-acylated apolar D-glucosamine substituent from 1 by partial hydrolysis results in a single aglycone derivative 2 bearing two sugar units [30]. In recent years, our group has been dealing with systematic synthetic modifications of teicoplanin pseudoaglycone 3, which can be obtained from teicoplanin mixtures as a single compound with removal of mannose and variously N-acylated D-glucosamine moieties [31]. In the framework of these synthetic studies, we introduced lipophilic substituents in compounds 2 and 3, obtaining derivatives that formed nanoaggregates in solution and were highly active against resistant staphylococci and enterococci [31][32][33]. Notably, dimers of neither the pseudoaglycone 2 with two sugar units nor the pseudoaglycone 3 bearing a single GlucNAc residue have been prepared yet. Therefore, we decided to synthesize some of these derivatives in order to study the influence of dimerization on the antibacterial activity in the teicoplanin series. In this work, we report on the synthesis of covalent dimers of 2 together with a study on a cobalt complex dimer obtained from a derivative of 3.
Teicoplanin (1) is not able to form non-covalent dimers because of its apolar tag, and its covalent dimers are also unknown. Teicoplanin used in the therapy is a mixture of five closely related derivatives, mainly consisting of the A2 component, differing from each other in the N-acyl substituent of one of the D-glucosamine moieties (Scheme 1). The separation of these components is difficult; therefore, synthetic modifications of the teicoplanin mixture also lead to complicated mixtures. Removal of the variously N-acylated apolar D-glucosamine substituent from 1 by partial hydrolysis results in a single aglycone derivative 2 bearing two sugar units [30]. In recent years, our group has been dealing with systematic synthetic modifications of teicoplanin pseudoaglycone 3, which can be obtained from teicoplanin mixtures as a single compound with removal of mannose and variously N-acylated D-glucosamine moieties [31]. In the framework of these synthetic studies, we introduced lipophilic substituents in compounds 2 and 3, obtaining derivatives that formed nanoaggregates in solution and were highly active against resistant staphylococci and enterococci [31][32][33]. Notably, dimers of neither the pseudoaglycone 2 with two sugar units nor the pseudoaglycone 3 bearing a single GlucNAc residue have been prepared yet. Therefore, we decided to synthesize some of these derivatives in order to study the influence of dimerization on the antibacterial activity in the teicoplanin series. In this work, we report on the synthesis of covalent dimers of 2 together with a study on a cobalt complex dimer obtained from a derivative of 3. Scheme 1. Conversion of teicoplanin 1 into teicoplanin pseudoaglycones 2 and 3. The sugar units are highlighted in red.

Results and Discussion
Scheme 1. Conversion of teicoplanin 1 into teicoplanin pseudoaglycones 2 and 3. The sugar units are highlighted in red.

Results and Discussion
For covalent dimer formation, simple linker reagent 8 was prepared with two reactive isothiocyanate groups and a lipophilic n-decylamino substituent to give an amphiphilic character to the dimers. Reacting n-decylamine 4 with a chloro derivative of triethylene glycol (5), tertiary amine-diol 6 was obtained. The hydroxyl groups of 6 were transformed to primary amino groups in a sequence of tosylation, nucleophilic substitution with azide and Staudinger reduction to give 7. Transformation [34] of the amino groups in compound 7 resulted in linker reagent 8 (Scheme 2). Linker 8 can tether two glycopeptides through thiourea bonds by reacting its two isothiocyanate functionalities with either the N-terminal amino group of compound 2 (Scheme 3) or an amino group attached to the C-terminus of 2 (Scheme 4). For covalent dimer formation, simple linker reagent 8 was prepared with two reactive isothiocyanate groups and a lipophilic n-decylamino substituent to give an amphiphilic character to the dimers. Reacting n-decylamine 4 with a chloro derivative of triethylene glycol (5), tertiary amine-diol 6 was obtained. The hydroxyl groups of 6 were transformed to primary amino groups in a sequence of tosylation, nucleophilic substitution with azide and Staudinger reduction to give 7. Transformation [34] of the amino groups in compound 7 resulted in linker reagent 8 (Scheme 2). Linker 8 can tether two glycopeptides through thiourea bonds by reacting its two isothiocyanate functionalities with either the N-terminal amino group of compound 2 (Scheme 3) or an amino group attached to the C-terminus of 2 (Scheme 4). Scheme 2. Syntheses of the α,ω-bis-isothiocyanate linker 8.
Using this diisothiocyanate 8, covalent dimers of 2 and its diethylaminopropyl amide 9 were prepared with an N,N-terminal orientation. For this reason, the carboxy-terminus of teicoplanin was converted into diethylaminopropyl amide, similar to the same structural element of the very potent antibacterial dalbavancin, and the N-acyl-glucosamine moiety with different side chains was removed hydrolytically to yield 9 [35]. Additionally, 2 or 9 was reacted with excess 8, and after workup, the obtained monoisothiocyanates 10 and 11 were reacted with an equivalent of 2 or 9 to give the dimers 12 and 13 (Scheme 3). In preliminary attempts, the linking of the two identical monomers was also tried using exactly half equivalent of the linker. Unfortunately, this approach led to sluggish conversion into dimers and reaction mixtures with significantly more components compared to the two-step method with the excess linker, making the already difficult purification even more challenging. The new two-step method gave much better control over the course of the reaction. The major disadvantage is obviously the waste of the excess linker if not recovered by chromatography. Scheme 2. Syntheses of the α,ω-bis-isothiocyanate linker 8.
Using this diisothiocyanate 8, covalent dimers of 2 and its diethylaminopropyl amide 9 were prepared with an N,N-terminal orientation. For this reason, the carboxy-terminus of teicoplanin was converted into diethylaminopropyl amide, similar to the same structural element of the very potent antibacterial dalbavancin, and the N-acyl-glucosamine moiety with different side chains was removed hydrolytically to yield 9 [35]. Additionally, 2 or 9 was reacted with excess 8, and after workup, the obtained monoisothiocyanates 10 and 11 were reacted with an equivalent of 2 or 9 to give the dimers 12 and 13 (Scheme 3). In preliminary attempts, the linking of the two identical monomers was also tried using exactly half equivalent of the linker. Unfortunately, this approach led to sluggish conversion into dimers and reaction mixtures with significantly more components compared to the two-step method with the excess linker, making the already difficult purification even more challenging. The new two-step method gave much better control over the course of the reaction. The major disadvantage is obviously the waste of the excess linker if not recovered by chromatography.
Two more dimers with N,C terminal orientation were obtained from pseudoaglycone 2 (Scheme 4). The amino group of 2 was protected in the form of t-butyl carbamate and subsequently the carboxylic moiety was converted to an aminoethyl amide to produce 14. The primary amino group of 14 allowed its coupling to linker 8 through a C-terminal thiourea linkage to produce monoisothiocyanate derivative 15. Compound 15 was then reacted with either 2 or its diethylamino propylamide derivative 9, resulting in the two dimers 16 and 17 after hydrolytic removal of the Boc group. Two more dimers with N,C terminal orientation were obtained from pseudoaglycone 2 (Scheme 4). The amino group of 2 was protected in the form of t-butyl carbamate and subsequently the carboxylic moiety was converted to an aminoethyl amide to produce 14.
The primary amino group of 14 allowed its coupling to linker 8 through a C-terminal thiourea linkage to produce monoisothiocyanate derivative 15. Compound 15 was then reacted with either 2 or its diethylamino propylamide derivative 9, resulting in the two dimers 16 and 17 after hydrolytic removal of the Boc group. We also used another approach for the formation of the N,N dimer. Meade and coworkers studied the stability and interaction of Schiff base cobalt diammine coordination complexes bearing labile axial ligands with histidine [36,37]. These complexes contain an acetylacetonatoethylenediimine (acacen) [38] equatorial ligand stabilizing the Co 3+ center. They observed a strong and selective binding of histidine or histidine parts of proteins to the complex. This reaction is based on a dissociative axial ligand exchange. We used We also used another approach for the formation of the N,N dimer. Meade and coworkers studied the stability and interaction of Schiff base cobalt diammine coordination complexes bearing labile axial ligands with histidine [36,37]. These complexes contain an acetylacetonatoethylenediimine (acacen) [38] equatorial ligand stabilizing the Co 3+ center. They observed a strong and selective binding of histidine or histidine parts of proteins to the complex. This reaction is based on a dissociative axial ligand exchange. We used this selective coordination complex formation to obtain a dimer from teicoplanin pseudoaglycone 3. Since this compound does not contain a histidine moiety, we prepared a histidyl derivative of 3. For this purpose, from the di-Boc derivative of histidine (18) [39], an N-hydroxysuccinimide ester (19) was obtained. With the use of the latter, the N-terminal amino group of 3 was acylated and the carbamate protecting groups of the intermediate were removed by trifluoroacetic acid treatment to give 20 (Scheme 5). This compound reacted with the Co 3+ acacen Schiff base complex to produce dimeric derivative 21 ( Figure 1).  The structure of the histidyl derivative (20) and the formation of a dimer (21) induced by the cobalt compound was verified by NMR spectroscopy in detail ( Figure 2). First, the 1 H and 13 C resonances of the newly introduced histidyl group were unambiguously identified on the basis of one-(1D) and two-dimensional (2D) correlation experiments ( Figure 2A). The linkage between the pseudoaglycone and histidyl group was confirmed by the throughspace ROESY connectivities observed between x1(H)-His(Hα) and x1(H)-His(Hβ) protons ( Figure 2B) and also by the multiple-bond heteronuclear ( 1 H/ 13 C) correlations detected between the carbonyl (CO) carbon and the x1(H)/His(Hα)/His(Hβ) protons, respectively ( Figure 2C). The formation of dimeric species upon addition of the Co-complex was monitored by Diffusion Ordered Spectroscopy (DOSY) and allowed the measurement of the translational diffusion coefficients of 20 and 21 as 1.79 × 10 −10 m 2 /s and 1.52× 10 −10 m 2 /s, respectively ( Figure 3A). The slightly slower diffusion of the Co-complex suggests the formation of a dimer being in equilibrium with the monomeric form. In addition, the significant broadening of 1 H resonances in Figure 3C (compared to Figure 3B) indicates that the dynamics of this monomer-dimer equilibrium fall in the intermediate regime on The antibacterial activity of the dimeric compounds was evaluated on a panel of Gram-positive bacteria ( Table 1). As can be seen, N,N-terminal dimer 12 displayed very diminished activity in comparison to that of the teicoplanin mixture 1. The formation of a basic amide in 13, similar to dalbavancin, was beneficial to the antibacterial activity, which was still a bit smaller than that of the parent teicoplanin 1. Moderate antibacterial activities were obtained for the two N,C dimers 16 and 17. However, it is remarkable that aglycone 2 and all dimers 12, 13, 16 and 17 displayed relatively high activity against VanA teicoplanin-resistant enterococci. Interestingly, the dimeric derivatives of teicoplanin in general displayed somewhat lower activity against enterococci with the VanB resistance gene and also against nonresistant enterococci ATCC 29212. Since the mechanism of action of glycopeptide antibiotics is based on the inhibition of cell wall peptidoglycan, this phenomenon can be explained by the different structures of the cell walls of these enterococci strains. Similarly, a low MIC value (4 µg/mL) was detected for compound 20, a histidyl derivative of 3. Unfortunately, cobalt complex dimerization of the same derivative did not improve the antibacterial activity of the monomer (see the activity  Table 1). In this case, it has to be noted that complex formation is an equilibrium phenomenon and, consequently, the activity measured for 21 is attributable to the equilibrium mixture of 20 and 21.
through-space ROESY connectivities observed between x1(H)-His(Hα) and x1(H)-His(Hβ) protons ( Figure 2B) and also by the multiple-bond heteronuclear ( 1 H/ 13 C) correlations detected between the carbonyl (CO) carbon and the x1(H)/His(Hα)/His(Hβ) protons, respectively ( Figure 2C). The formation of dimeric species upon addition of the Cocomplex was monitored by Diffusion Ordered Spectroscopy (DOSY) and allowed the measurement of the translational diffusion coefficients of 20 and 21 as 1.79 × 10 −10 m 2 /s and 1.52× 10 −10 m 2 /s, respectively ( Figure 3A). The slightly slower diffusion of the Co-complex suggests the formation of a dimer being in equilibrium with the monomeric form. In addition, the significant broadening of 1 H resonances in Figure 3C (compared to Figure 3B) indicates that the dynamics of this monomer-dimer equilibrium fall in the intermediate regime on the 1 H NMR timescale.    The antibacterial activity of the dimeric compounds was evaluated on a panel of Gram-positive bacteria ( Table 1). As can be seen, N,N-terminal dimer 12 displayed very diminished activity in comparison to that of the teicoplanin mixture 1. The formation of a basic amide in 13, similar to dalbavancin, was beneficial to the antibacterial activity, which was still a bit smaller than that of the parent teicoplanin 1. Moderate antibacterial activities were obtained for the two N,C dimers 16 and 17. However, it is remarkable that aglycone 2 and all dimers 12, 13, 16 and 17 displayed relatively high activity against VanA teicoplanin-resistant enterococci. Interestingly, the dimeric derivatives of teicoplanin in general displayed somewhat lower activity against enterococci with the VanB resistance gene and also against nonresistant enterococci ATCC 29212. Since the mechanism of action of glycopeptide antibiotics is based on the inhibition of cell wall peptidoglycan, this phenomenon can be explained by the different structures of the cell walls of these enterococci strains. Similarly, a low MIC value (4 μg/mL) was detected for compound 20, a histidyl derivative of 3. Unfortunately, cobalt complex dimerization of the same derivative did not improve the antibacterial activity of the monomer (see the activity of dimer 21 in Table 1). In this case, it has to be noted that complex formation is an equilibrium phenomenon and, consequently, the activity measured for 21 is attributable to the equilibrium mixture of 20 and 21.  (5), 3-(diethylamino)-1-propylamine and L-histidine are commercially available. N,N -di-Boc-L-histidine (18) was prepared by the literature procedure [39].
NMR experiments for structure elucidation ( 13 C-1 H HSQC, HMBC (using 80 ms for n J CH evolution), HSQC-TOCSY (with 80 ms TOCSY mixing time) and 1 H-1 H ROESY (with 60 ms mixing time) were carried out on a Bruker Avance Neo 700 MHz spectrometer equipped with a Prodigy TCI cryoprobe. The temperature was set to 298 K, and chemical shifts were referenced to the solvent signals. A sample of teicoplanin pseudoaglycone 3 (4.7 mg) was dissolved in 550 µL DMSO-d 6 , while the histidyl derivative 20 (10 mg) was dissolved in a mixture of 400 µL DMSO-d 6 and 150 µL MeOD.
Diffusion Ordered SpectroscopY (DOSY) measurements were carried out at 298 K using a 500 MHz Bruker Avance II spectrometer equipped with TXI ( 1 H/ 13 C/ 15 N triple resonance) probe. Diffusion coefficients were determined with the stimulated spin echo LED (Longitudinal-Eddy-current Delay) sequence (ledbpgp2s) of the Bruker pulse sequence library. The length of the diffusion delay (50 ms) and gradient pulses (5 ms) were optimized by short, one-dimensional (1D) experiments. The gradient strength was increased in 32 increments, linearly varying between 5% and 95% of its maximum value. The maximum gradient strength of the probe is 57.7 G/cm. The number of scans acquired for each increment was 16. Histidyl derivative teicoplanin pseudoaglycone (2.07 mg) was dissolved in 25 µL DMSO-d 6 and 475 µL D 2 O mixture for measurement. Then, 4 µL of Co-complex from a stock solution (5.11 mg/100 µL) was added to the sample and the measurement was repeated with the same experimental setup. The dosy module of Topspin 2.1. was used for data processing and generation of 2D DOSY plots.
The 1 H NMR (360 and 400 MHz), 13 C NMR (90 and 100 MHz) and 2D NMR spectra were recorded with Bruker DRX-360 and Bruker DRX-400 spectrometers at 25 • C. Chemical shifts are referenced to Me 4 Si (0.00 ppm for 1 H) and to the solvent residual signals. The spectra were evaluated using MestReNova and TopSpin software. MALDI-TOF MS measurements were carried out with a Bruker Autoflex Speed mass spectrometer equipped with a time-of-flight (TOF) mass analyzer. In all cases, 19 kV (ion source voltage 1) and 16.65 kV (ion source voltage 2) were used. For reflectron mode, 21 kV and 9.55 kV were applied as reflector voltage 1 and reflector voltage 2, respectively. A solid-phase laser (355 nm, ≥100 µJ/pulse) operating at 500 Hz was applied to produce laser desorption and 3000 shots were summed. 2,5-Dihydroxybenzoic acid (DHB) was used as the matrix and F 3 CCOONa as the cationizing agent in DMF. A MicroTOF-Q type Qq-TOF MS instrument (Bruker Daltonik, Bremen, Germany) was used for the ESI-MS measurements. The instrument was equipped with an electrospray ion source where the spray voltage was 4 kV. N 2 was utilized as the drying gas. The drying temperature was 180 • C and the flow rate was 4.0 L/min. The mass spectra were recorded by means of a digitizer at a sampling rate of 2 GHz. The spectra were evaluated with DataAnalysis 3.