Functionalisation of vitamin B12 derivatives with a cobalt β-phenyl ligand boosters antimetabolite activity in bacteria

This study describes the syntheses of four singly- and two doubly-modified vitamin B12 derivatives for generating antimetabolites of Lactobacillus delbrueckii (L. delbrueckii). The two most potent antagonists, a Coβ-phenyl-cobalamin-c,8-lactam and a 10-bromo-Coβ-phenylcobalamin combine a c-lactam or 10-bromo modification at the “eastern” site of the corrin ring with an artificial organometallic phenyl group instead of a cyano ligand at the β-site of the cobalt center. These two doubly-modified B12 antagonists (10 nM) inhibit fully B12-dependent (0.1 nM) growth of L. delbrueckii. In contrast to potent 10-bromo-Coβ-phenylcobalamin, single modified 10-bromo-Coβ-cyanocobalamin lacking the artificial organometallic phenyl ligand does not show any inhibitory effect. These results suggest, that the organometallic β-phenyl ligand at the Co center ultimately steers the metabolic effect of the 10-bromo-analogue.


Introduction
Non-functional analogues of vitamins and vitamin building blocks represent an important class of drugs and drug candidates for treating different classes of diseases ranging from bacterial and fungal infections to human cancer. [1][2][3] In the rst half of the 20th century, prontosil was introduced as the rst commercially available antimicrobial agent and saved millions of lives. 4 The sulfonamide-based drug targets effectively bacterial, but not human biosynthesis of vitamin B 9 (folic acid) explaining its selective therapeutic effect. Modied folic acid derivatives such as methotrexate, trimetrexate or pemetrexed represent other examples of important antibacterial and anticancer agents by inhibiting folic acid-dependent enzymatic transformations. 1,5,6 In contrast to these folate-based drugs, modied cobalamin (vitamin B 12 ) derivatives have not been developed so far to approved antibacterial or anticancer drugs. 7 Nevertheless, some singly-modied vitamin B 12 derivatives showed promising inhibitory effects on B 12 -dependent pathways. 8 For example, B-ring-modied hydroxycobalamin-c,8-lactam (Fig.  1B  le) and b-ligand-modied 4ethylphenylcobalamin (Fig. 1B right) decreased signicantly the activities of liver L-methylmalonyl coenzyme A mutase and methionine synthase as indicated by elevated plasma methylmalonic acid concentration and total homocysteine concentration in rodents. 9,10 In contrast to these important results, bi-functionalised B 12 derivatives combining two instead of a single modication were not explored so far in a systematic fashion.
Herein we report on the antibacterial activity of doubly-modied B 12 derivatives. It is demonstrated in a proof-ofconcept study with L. delbrueckii, that a specic second

Syntheses and characterisation of modied vitamin B 12 derivatives
Starting from vitamin B 12 (1; Fig. 1A) we prepared two novel bifunctionalized Cbls (4 and 7; Scheme 1) using established synthetic protocols. 11 In particular, we combined in the analogues either a c,8-lactam 11 (Scheme 1 le) or C10-Br 12 (Scheme 1 right) modication at the corrin ring with a b-phenyl ligand at the Co-center (Scheme 1, bottom line). These combinations were purposefully chosen because the selected transformations (i) induced biological effects in previous studies when incorporated into the Cbl scaffold ( Fig. 1B) and (ii) provide analogs in high purity under mild conditions. 9,10 Syntheses, isolation, and characterization of the two novel doubly-modied derivatives 4 and 7 and four singly-modied Cbls (2, 3, 5 and 6) are described in more detail below and in the Experimental section. In this section, we outline only the preparation and characterization of unprecedented compounds Co b -phenyl-Cbl-c,8-lactam (4; Scheme 2 top) and 10-bromo-Co bphenylcobalamin (7; Scheme 2 bottom).
Vitamin B 12 analogue 4 was synthesized in two steps starting from commercially available B 12 in a total isolated yield of 27%. First, c,8-lactam 2 was prepared under basic conditions at 100 C following a procedure of Todd et al. 11 Aer reduction of its Co III center with NaBH 4 (10 equiv.) and subsequent treatment with diphenyl iodonium chloride (2 equiv.) in H 2 O, 13 the organometallic target 4 was obtained. The occurrence of a pseudo molecular ion at m/z ¼ 1404.56 ([M + H] + , m/z calc : 1404.60 for C 68 H 92 CoN 13 O 14 P + ) in the ESI-MS spectrum of 4 indicated successful arylation at the Co center of 2, supported by the observation of a hypsochromically shied g-band (Dl ¼ 18 nm) with diminished intensity (Dlog 3 ¼ 0.22) in the UV/vis spectrum. This spectral behavior is typical for organometallic Cbls featuring a Co III -C bond (Fig. S13 †). 13 For the synthesis of 10-bromo-Co b -phenylcobalamin (7; Scheme 2 bottom), we considered that Cbls bearing good leaving groups are prone towards reducing agents. Therefore, 10-bromo-Co b -cyanocobalamin (5) is not compatible with arylation conditions using NaBH 4 . 14 Having this in mind, we brominated the previously described Co b -phenylcobalamin (8) 13 with N-bromosuccinimide (NBS) at its C10 position according to a method of Wagner. 14 ESI-MS analysis conrmed successful formation of 7 by the presence of its adduct ion peak at m/z ¼ 1485.53 ([M + H] + ) and the characteristic bromine isotopic pattern. Bromination at position C10 of 7 was further proven by the absence of the signal of the proton at C10 in the 1 H-NMR spectrum and the absorption spectrum of 7 exhibited a characteristic redshi of the abband to 537 nm. 12

Bacterial growth assays
The inhibitory potential of the small library of four singly-, and two doubly modied B 12 derivatives (2-7) was assessed with bacterial growth assays using L. delbrueckii.
These Gram-positive bacteria are ideal for such proof-ofconcept studies since they possess ribonucleotide reductase (RNR) as the only Cbl-dependent enzyme (see Fig. 2 for details). [15][16][17] Small concentrations of B 12 (0.1 nM) in the medium are sufficient to support growth of L. delbrueckii (positive control; Fig. 3).
Notably, when exogenous B 12 was absent in the assay medium (negative control), substantial residual growth (r.g.; 80 AE 1%) was still observed suggesting contamination of the B 12free medium with little amounts of Cbl. [17][18][19] In competition Scheme 1 Cascade of multiple chemical modifications of Cbls and their physiologic effect (promotion of growth or activity as antimetabolite; r.g. ¼ residual bacterial growth compared to a B 12 -only control group) on a L. delbrueckii bacterial culture (the strength of the effect is indicated with colors; green: growth promoting; orange: medium inhibition of growth; red: strong inhibition of growth). The scheme does not indicate the course of chemical reactions. Charges omitted for clarity. assays with a 100-fold excess of analogs 2-7 (10 nM) over B 12 , c,8-lactam modied CNCbl 2 showed 22% inhibition (r.g. ¼ 78 AE 1%; Fig. 3 le) of B 12 -triggered growth. Bi-functionalization of analog 2 with an additional organometallic b-phenyl ligand (i.e., analog 4) strengthened further the antimicrobial activity (r.g. ¼ 66 AE 4%; Fig. 3 le). Inhibition in this competition assay was stronger than residual growth in the absence of exogeneous B 12 (i.e., negative control) suggesting that total B 12 in the medium was effectively outcompeted by the presence of analogue 4 (Fig. 3 le). In contrast to single modied c,8-lactam 2, single-modied 10-BrCNCbl (5) had no evident effects on B 12 -dependent growth (r.g. ¼ 100 AE 1%; Fig. 3 right). This biological lethargy changed drastically upon further modication with a bphenyl group at the Co III center. Bi-functionalized 7 with a 10-Br modication was a comparably strong inhibitor (r.g. ¼ 62 AE 3%) as bi-functionalized 4 with a c,8-lactam group (Fig. 3). The importance of the b-phenyl functionality at the Co III center of inhibitors 4 and 7 for triggering the antimetabolic effect was further supported by studying the corresponding Co b -aqua derivatives 3 and 6. These analogues lacking the organometallic phenyl ligand had either negligible (r.g. Fig. 3-le) or even growth-promoting effects (r.g. (6) ¼ 129 AE 1%, Fig. 3  right).
The latter result suggests, that aqua derivative 7 is still recognized, internalized, and metabolized by the microorganism to a growth promoting, enzymatically active organometallic AdoCbl cofactor (Fig. 3). In contrast, this reductive biological transformation is apparently not possible for 2, 4 and 7 containing s-donating cyano or phenyl ligands in addition to a strucurally altered Cbl scaffold. 20 Although we propose herein inhibition of RNR with inhibitors 2, 4 and 7 (Fig. 3), the actual biological target(s) and modes of action of these B 12 analogues have still to be unraveled in future biological studies. So far, our proof-of-concept studies clearly demonstrate that the second, ligand-centered modication of the analogues with a b-phenyl ligand signicantly steers antimetabolite activity. Replacement of this ligand with a weakly coordinating aqua ligand effected complete reversal of the antimetabolite activity of both bifunctionalized derivatives.

Conclusions
We have synthesized and tested antibacterial activity of four single-and two novel bi-functionalised B 12 analogues. B 12dependent growth studies with L. delbrueckii showed strikingly that doubly modied Co b -phenyl-cobalamin-c,8-lactam and 10bromo-Co b -phenylcobalamin were the most potent antagonists. Of note, inhibition was even stronger than residual growth in the absence of exogenous B 12 . Moreover, these studies demonstrated strikingly, that the second, b-axial modication signicantly steers the metabolic effect. In particular, the incorporation of an organometallic b-phenyl ligand at the cobalt center either empowered the inhibitory potential of CNCbl-c,8-lactam or more interestingly, induced antimetabolic activity in an erstwhile innocent 10-brominated Cbl analogue.

General
Chemicals were of reagent grade quality or better and obtained from Sigma-Aldrich, ACROS Organics, Merck or Fluka and used without further purication unless otherwise indicated. Vitamin B 12 was obtained from Sigma-Aldrich or received as a generous gi from DSM Nutritional Products AG (Basel/ Switzerland) and Prof. Em. Bernhard Jaun (ETH Zurich, Switzerland). All solvents were of reagent, analytical, HPLC or LC-MS grade, respectively, and obtained from commercial suppliers. Bi-distilled H 2 O was used in all reactions. H 2 O from a Milli-Q (Merck-Millipore) water purication system was used for UV/ vis spectroscopy, mass spectrometry and when indicated. Reactions were carried out under N 2 (g) or Ar (g) in oven-dried (100 C) glass equipment and monitored for completion by analysing a small sample (aer suitable workup) by LC-MS. Evaporation of the solvents in vacuo was done with the rotary evaporator (Büchi) at the given bath temperature and pressure. SepPak® RP-18 cartridges (Waters) were applied for solid phase extraction. The compounds were dissolved in H 2 O, transferred to the adsorbent, washed with H 2 O or the indicated aq. soln, followed by H 2 O, and eluted with CH 3 OH.

Spectroscopy
UV/vis spectra. Cary 50 Scan spectrophotometer (Varian) or Specord 250 Plus (Analytik Jena) using 1 cm quartz cuvettes (Hellma Analytics); l max (log 3) in nm. Both 1 H-and 13 C-NMR spectra were carried out at 298 K in D 2 O or CD 3 OD and at 500 MHz or 126 MHz, respectively. The 1 H-NMR spectra were performed in an AVANCE NEO 500 MHz spectrometer (Bruker) using a 5 mm-z-gradient RT-BBI probehead; d in ppm relative to HDO (d 4.79; corresponds to TMS (d 0.00)) or CHD 2 OD (d 3.31; corresponds to TMS (d 0.00)), J in Hz. Spectra in D 2 O were presaturated. The 13 C-NMR spectra were performed in an AVANCE NEO 500 MHz spectrometer (Bruker) using a 5 mm zgradient CP-BBO probehead; d in ppm relative to CD 3 OD (d 49.0; corresponds to TMS (d 0.0)), J in Hz.

Bacterial growth assays
Culture medium was prepared by dissolving MRS broth (Difco) for Lactobacilli (5.5 g) in Milli-Q H 2 O (100 mL) and subsequently ltered through a sterile 2.0 mm lter. The culture medium (14 mL) was inoculated with Lactobacillus delbrueckii subsp. Lactis, DSM 20355 from a micro-ring culture (previously stored at −70 C). The closed tubes were incubated at 30 C for 24 h. Aerwards a small sample (0.5 mL) was taken out and OD 680 nm was determined and typically yielded values around 1.0 aer 24 h. A second culture was inoculated by addition of the inoculate (200 mL) to fresh MRS broth (14 mL), followed by incubation at 30 C for 24 h, resulting in OD 680 nm ¼ 1.5 prior to the conduction of the assay. The bacterial culture was centrifuged (5000 rpm/5 min), and the remaining pellet was suspended in H 2 O ([NaCl] ¼ 0.9%, 14.0 mL) and incubated at 37 C for 30 min. The resulting suspension was centrifuged again (5000 rpm/5 min), followed by two washing steps in H 2 O ([NaCl] ¼ 0.9%, 14.0 mL) to remove remaining traces of the growth medium. Aerwards, H 2 O ([NaCl] ¼ 0.9%, 5.0 mL) was added to the pellet and the bacterial suspension was stored at 37 C. Vitamin B 12 assay medium (Sigma-Aldrich, 41.5 g) was dissolved in Milli-Q H 2 O (500 mL) and the mixt. was heated to 40 C under stirring until everything was dissolved, before Tween® 80 (1.0 mL) was added and everything was thoroughly homogenized. The pH was adjusted to 6.0 by addition of H 2 O ([NaOH] ¼ 0.5 M) and the medium was ltered through a sterile 2.0 mm lter. Subsequently, sterilized bacterial assay glass tubes (Fisher Scientic, total volume: 7.0 mL) were lled with of B 12 assay medium (6.5 mL), a sterile stock solution of vitamin B 12 in Milli-Q H 2 O (6.5 mL, 1.0 nM), except for the negative control, and a sterile soln of the respective test compound in H 2 O (6.5 mL, 100 nM to 0.1 mM), except for the positive control. All tubes were nally inoculated with 40 mL of the bacterial suspension, tightly closed, and incubated at 37 C for 28-60 h. OD 680 nm was monitored photometrically every 2-8 h (aer through mixing of the tubes), until saturation of growth was detected. All assays were performed in triplicates and average values of OD 680 nm (AE2s) were obtained and plotted vs. time (t) in h (hours) to obtain growth curves. Residual growth values (r.g.) were estimated aer 26 h and are given in% relative to the growth of the positive control group.

Experimental procedures
Co b -cyanocobalamin-c,8-lactam (2). B-Ring lactam formation in Co b -cyanocobalamin (1) was performed according to lit. 11 Briey, 1 (100 mg, 73.8 mmol, 1.0 equiv.) was added to a soln of NaOH (1.0 M in H 2 O, 20 mL) and the soln was heated to 100 C for 10 min. The reaction mixture was adjusted to pH 8.0 by addition of NaHCO 3 and the crude product was extracted using SPE. Purication via preparative HPLC (method B) and subsequent lyophilization afforded 2 (46. 6   . Assignments were made in comparison with data from lit. 21,22 Co b -aquacobalamin-c,8-lactam acetate (3). A soln of 2 (5.0 mg, 3.7 mmol) in H 2 O (1.0 mL) was degassed by purging with N 2 (g) for 15 min, before a soln of NaBH 4 (1.5 mg, 40 mmol, 11 equiv.) in H 2 O (0.1 mL) was added. The resulting mixt. was stirred at 23 C for 10 min, until a color change to dark violet occurred. Subsequently, AgOAc (1.2 mg, 7.2 mmol, 1.9 equiv.) was added, resulting in formation of a white precipitate. The precipitate was ltered off and a gentle stream of air was passed through the remaining soln, upon which it turned red. LC-MS analysis (method 2) revealed formation of the aquo complex 3 as the sole product. The product was extracted using SPE, washed with an aq. soln of NH 4 OAc (0.1 M, 10 mL) and eluted with CH 3 OH. The solvent was evaporated in vacuo (200 mbar, 40 C) and the residue was re-dissolved in H 2 O (1.5 mL) and