Facile and selective N-alkylation of gentamicin antibiotics via chemoenzymatic synthesis

The rise and spread of antimicrobial resistance has necessitated the development of novel antimicrobials which are effective against drug resistant pathogens. Aminoglycoside antibiotics (AGAs) remain one of our most effective classes of bactericidal drugs. However, they are challenging molecules to selectively modify by chemical synthesis, requiring the use of extensive protection and deprotection steps leading to long, atom- and step-inefficient synthetic routes. Biocatalytic and chemoenzymatic approaches for the generation of AGA derivatives are of interest as they allow access to more concise and sustainable synthetic routes to novel compounds. This work presents a two-step chemoenzymatic route to regioselectively modify the C-6′ position of AGAs. The approach uses a transaminase enzyme to generate an aldehyde on the C-6′ position in the absence of protecting groups, followed by reductive amination to introduce substituents selectively on this position. Seven candidate transaminases were tested for their ability to deaminate a panel of commercially available AGAs. The C-6′ transaminases could deaminate both pseudo di- and trisaccharide AGAs and tolerate the presence or absence of hydroxyl groups on the C-3′- and C-4′-positions. Additionally, sugar substituents on the C-6 hydroxyl were accepted but not on the C-5 hydroxyl. The most promising enzyme, GenB4, was then coupled with a reductive amination step to synthesise eleven novel 6′-gentamicin C1a analogues with conversions of 13–90%. Five of these compounds were active antimicrobials and four of these retained activity against an aminoglycoside-resistant Escherichia coli. This approach allows facile and step-efficient access to novel aminoglycoside compounds under mild reaction conditions and could potentially enable the development of greener, sustainable, and more cost-effective syntheses of novel AGAs.


General molecular biology methods
Molecular biology reagents were obtained from New England Biolabs (NEB) or Thermo Scientific. Spin column purification and gel extraction of DNA fragments was performed using a Monarch PCR and DNA Clean-up kit and DNA Gel Extraction kit using the manufacturer's instructions, respectively. Mini-preps of plasmid DNA were performed using a QIAprep Spin Miniprep Kit (QIAGEN) using the manufacturer's instructions. Synthesis of oligonucleotides for PCR, synthetic genes and Sanger DNA sequencing were all performed by Eurofins Genomics.

General chemistry methods
All compounds were purchased from Sigma-Aldrich, Alfa Aesar or Acros Organics and used without further purification. Freeze drying of samples was performed using a SP VirTis Benchtop Pro with Omnitronics (SP Scientific) with the freeze drier connected to a vacuum pump (Edwards). Centrifugal evaporation was performed using a Jouan RC1022 centrifugal evaporator connected to a Heto cooling trap and vacuum pump. 1 H and 13 C nuclear magnetic resonance (NMR) spectra were recorded at 600 MHz and 150 MHz, respectively on a Bruker Avance III 600 instrument equipped with a helium-cooled cryoprobe. NMR spectra were recorded in deuterium oxide [reference peak: 1 H NMR: HOD (4.79 ppm), 13 C: TFA salt peaks (quartets at 163.7 and 117.0 ppm)]. For 1 H NMR spectra, chemical shifts are quoted to the nearest 0.01 ppm, with signal splittings reported as: broad (br), singlet (s), doublet (d), triplet (t), quartet (q), multiplet (m) or combinations thereof. For 13 C NMR spectra, chemical shifts are reported to the nearest 0.1 ppm. All NMR data were obtained in the UCL Chemistry NMR Facility.
High-resolution ASAP-HESI mass spectra were acquired using an ASAP probe integrated into a HESI ion source connected to a Q-Exactive Plus mass spectrometer. All mass spectral data were obtained in the UCL Chemistry Mass Spectrometry Facility.
KanB, ForB, GenB1, GenB2 and GenB4 genes were digested with NdeI and XhoI, DNA fragments were purified by gel extraction and ligated into pET28a(+) digested with the same enzymes. These constructs had a 6x-His tag at the N-terminus followed by a thrombin cleavage site. The gene encoding PchB was amplified from Paenibacillus chitinolyticus DSM 11030 genomic DNA by PCR using the following conditions: 1x Phusion High-Fidelity PCR master mix with HF buffer (2x stock from NEB), 1 µM forward and reverse primers, 1 µL template DNA, 10% DMSO, up to 50 µL with sterile H2O [Thermocycling conditions -98 °C/5 mins, 30 cycles of (98 °C/10 secs, 72 °C /30 secs, 72 °C/45 secs), 72 °C/7 mins]. Primers are described in Table S2. PCR products were digested with SapI and cloned into pET29a(+)-sacB 1 to give plasmid pQR2763. This construct contained a C-terminal 6x-His tag following the last amino acid. All plasmids were verified by restriction digestion and DNA sequencing.

Expression of C-6' TAms
The plasmids were transformed into E. coli BL21(DE3) and a single colony was inoculated into 10 mL Terrific Broth (TB, Merck) containing kanamycin (50 µg/mL) and incubated at 37 °C/250 rpm overnight. Overnight cultures were diluted 100-fold in fresh TB media containing kanamycin (50 µg/mL) and 0.5 M sorbitol and grown at 37 °C/250 rpm until an OD 600 of 0.6 -0.8. They were subsequently induced with 0.5 mM IPTG and expressed at 25 °C/200 rpm overnight. Expression was assessed by SDS-PAGE analysis.

Enzyme purification and storage
Cell pellets were harvested 8000 rpm/10 mins and resuspended in 10 mL lysis buffer (50 mM Na2HPO4 pH 7.4, 0.5 M NaCl, 20 mM imidazole, 0.1 mM PLP). Cell suspensions were sonicated 10 sec ON/10 sec OFF for 10 cycles on ice using a Soniprep 150 and centrifuged 11,000 rpm/45 mins/4 °C. The lysates were purified by immobilised metal ion affinity chromatography via the 6x-His tag using Ni 2+ Sepharose resin columns (Sepharose Fast Flow, GE Healthcare) and a vacuum manifold. Proteins were washed with lysis buffer containing a stepwise gradient of imidazole [20 mM (50 mL), 50 mM (50 mL), 100 mM (30 mL)] prior to elution with lysis buffer containing 500 mM imidazole (10 mL). Protein elution was monitored by a Bradford assay and protein purity was assessed by SDS-PAGE. Purified proteins were precipitated with ammonium sulfate and stored at 4 °C.

AGA C-6' TAm deamination reactions
Prior to deamination screening the enzyme solutions were prepared as follows: An appropriate volume of ammonium sulfate precipitated protein was centrifuged 21,130 x g/30 mins/4 °C and resuspended in 300 μL of the assay buffer (50 mM Tris-HCl pH 8.0). This solution was transferred to a Amicon Ultra 0.5 centrifugal filter (3k MWCO, Merck Millipore) pre-equilibrated with the assay buffer and the solution was centrifuged at 13,000 x g/20 mins/ 4 °C. The concentrated protein solution was diluted with 350 -400 μL of assay buffer and centrifuged once more at 13,000 x g/20 mins/ 4 °C. This step was repeated once again. The protein solution was collected by inverting the filter into an empty collection tube and centrifuging at 1000 x g/5 mins/4 °C. This solution was diluted with assay buffer to the volume required for the assay. The protein concentration was checked by a Bradford assay and then used in the reactions. Upon purification, KanB and ForB tended to aggregate and precipitate out of solution. GenB2 was also obtained in poor protein yields during protein purification. As a result, all three enzymes were tested at lower protein concentrations (ForB and GenB2: 0.1 mg/mL, KanB: 0.07 mg/mL) in deamination assays.

GenB4 isomerisation assay and reaction optimisation
A typical GenB4 isomerisation assay slightly modified the conditions from Chen and colleagues 3 , The final reaction conditions were as follows: 50 mM potassium phosphate (KPi) buffer pH 8.0, 2 mM sisomicin, 0.5 mM PLP, 0.2 mg/mL GenB4 enzyme prepared as described above in section 1.5.
The reaction was optimised in terms of enzyme concentration, amine acceptor necessity, reaction pH, temperature and sisomicin concentration. For enzyme concentration, the assay conditions from above were used in the presence of 0.03, 0.06, 0.125 and 0.25 mg/mL GenB4 enzyme in triplicate. The assay with and without 2 mM ⍺-ketoglutarate followed the conditions described with and without 2 mM a-ketoglutarate and 0.125 mg/mL GenB4. The reaction pH was varied from pH 7.0, 7.5, 8.0, 8.5 and 9.0 using KPi buffer and performed in duplicate. The temperature assay was performed in duplicate and reactions were incubated at 20 °C, 30 °C, 40 °C, 50 °C and 60 °C. Finally, the sisomicin concentration assay was performed in duplicate using the conditions described and varying the sisomicin concentration: 1, 2, 5, and 10 mM of sisomicin.
All optimisation reactions were performed on a 200 μL scale. Enzyme concentration and amine acceptor necessity reactions were incubated for 2 hrs and 10 μL samples were taken at 0, 15, 30, 60 and 120 mins of reaction. All other reactions were incubated for 24 hrs and 10 μL samples were taken at 0, 60, 120, 240 and 1440 mins of reaction time. All reaction samples were mixed with an equivalent volume of 0.5% TFA solution and frozen at -20 °C. Prior to analysis by HPAE-IPAD, samples were thawed, centrifuged 1000 x g/4 °C/2 mins and diluted 62.5-fold in MilliQ H2O.

Analysis of GenB4 reaction side product formation
A GenB4 reaction was set-up in triplicate (200 μL reactions) as described in section 1.6. After 1 hr of incubation at 30 °C, the reactions were incubated at 95 °C for 1 hr and a 10 μL sample was taken. The reactions were then further incubated at 30 °C for 23 hrs, and a second 10 μL sample was taken. Samples were processed for HPAE-IPAD analysis as described in section 1.5. For LC-MS analysis, a GenB4 reaction with sisomicin was prepared using the conditions described in 1.6 and incubated for 2 hrs at 35 °C. The reaction was then heated at 95 °C for 30 mins. The pH was reduced to 5.0 with AcOH and 10 equivalents of polymer supported cyanoborohydride (polymer = Amberlite IRA-400) was added and this mixture was incubated at 25 °C/1000 rpm/17 hrs. A sample of the reaction was diluted 25-fold and analysed by LC-MS using the method described in section 1.14.2.

Reductive amination pH assay
GenB4 enzyme solution was prepared as described in section 1.5 and added to a final concentration of 0.3 mg/mL in a reaction (6.5 mL) containing 50 mM NaPi buffer pH 8.0, 0.5 mM PLP and 2.5 mM sisomicin. The reaction was incubated for 1 hr at 30 °C/175 rpm. After 1 hr, an equal volume of cold methanol was added directly to the enzymatic reaction and the S5 mixture was centrifuged 10,000 rpm/4 °C for 5 minutes to remove the protein from the reaction mixture. The supernatant was transferred to a new tube. The reaction was then split into fractions in triplicate and the pH adjusted to 4.0, 5.0, 6.0 and 7.0 with concentrated AcOH, giving 12 fractions in total. Ten equivalents of phenethylamine (3g) were added respectively and the reaction stirred for 1 hr at 25 °C. Ten equivalents of polymer-supported cyanoborohydride was added and the reaction was further incubated at 25 °C for 16 -17 hrs. In a separate experiment, The same reaction procedure was followed, but 20 equivalents of benzylamine was added instead of 10 equivalents of phenethylamine.
After incubation, the reactions containing phenethylamine (3g) were diluted 25-fold and analysed by LC-MS using the method described in section 1.14.2. For the benzylamine (3f) reactions, the reactions were briefly centrifuged, 800 μL of the supernatant was separated from the beads and MeOH from the reaction was removed by centrifugal evaporation for 4 hrs at room temperature. The remaining aqueous phase was made up to 400 μL with 50 mM NaPi buffer pH 8.0 and the pH was adjusted to 10.0 -11.0 with 12 M NaOH. The excess benzylamine was extracted with EtOAc (3 x 400 μL) and the pH of the aqueous phase was re-acidified to pH 7.0 -8.0 with 1 M HCl. These solutions were analysed by analytical HPLC. For conditions see section 1.14.3 and Table S4. Samples were kept at -20 °C prior to analysis.

Reductive amination methanol content assay
GenB4 enzyme solution was prepared as described in section 1.5 and added to a final concentration of 0.1 mg/mL in a reaction containing 50 mM NaPi buffer pH 8.0, 0.5 mM PLP and 2.5 mM sisomicin. The reaction was incubated for 3 hrs at 30 °C/175 rpm. The reaction was then split into 5 x 500 μL equal fractions. An equal volume of cold MeOH was added to one of the fractions to give a final MeOH content of 50%. For the remaining fractions, the volume of water was reduced by rotary evaporation. The volumes were then adjusted to 400 μL, 300 μL, 200 μL and 100 μL and 600 μL, 700 μL, 800 μL and 900 μL of cold MeOH was added respectively to give final methanol contents of 60%, 70%, 80% and 90%. All samples were adjusted to pH 7.0 with AcOH, 15 equivalents of phenethylamine was added and reactions were incubated for 1 hr/25 °C/1000 rpm. Ten equivalents of polymer-supported cyanoborohydride was added and the reaction was further incubated at 25 °C for 16 -17 hrs. After incubation, the reactions were diluted 50-fold and analysed by LC-MS using the method described in section 1.14.2.

Substrate solubility tests
GenB4 enzyme solution was prepared as described in section 1.5 and the reaction with sisomicin was performed in triplicate (3 x 500 μL) as described in section 1.6. After 1 hr of incubation, 10 μL of the reaction was sampled and mixed with an equivalent volume of 0.5% TFA. To the remaining reactions, an equal volume of cold methanol was added, 10 μL of the reaction was sampled and mixed with an equivalent volume of 0.5% TFA. These samples were analysed by HPAE-IPAD using the method for 6'-oxogentamicin C1a, see section 1.14.1.
For gentamicin C1a solubility tests, 9 x 100 μL solutions of 12.5 mM gentamicin C1a in 200 mM NaPi buffer pH 5.0 were prepared in triplicate (27 solutions total). Three solutions were kept at 100 μL. The remaining solutions were diluted with water in triplicate to 200 μL, 300 μL, 400 μL, 500 μL, 600 μL, 700 μL, 800 μL and 900 μL final volumes. To these solutions, cold methanol was added to give a final volume of 1 mL in each solution. These solutions were centrifuged for 2 mins at 10,000 x g/4 °C. Samples were diluted 62.5-fold and analysed by HPAE-IPAD using a 3 mM KOH isocratic method for 20 mins.

Reductive amination conversion assay
GenB4 enzyme solution was prepared as described in section 1.5 and a 7.5 mL reaction was performed as described in section 1.6. After 1 hr, an equal volume of cold methanol was added and the mixture was centrifuged 10000 rpm/4°C for 5 minutes to remove the protein from the reaction mixture. The supernatant was transferred to a new tube. The reaction was split into 14 x 500 μL fractions and adjusted to pH 5.0 with concentrated AcOH. Ten equivalents of amines 3a, 3f -3r were added and reactions were stirred for 1 hr at 25 °C/1000 rpm. Ten equivalents of polymer-supported cyanoborohydride was added and the reaction was further incubated at 25 °C for 17 -18 hrs. Samples were subsequently diluted 25-fold in water and analysed by LC-MS using the method described in section 1.14.2. Amines 3b -3e (Table 2.15) were assayed on a separate day, a larger scale GenB4 reaction (20 mL) was set-up as above and split into 4 x 5 mL fractions. The procedure then followed as described above, 10 equivalents of each amine were added and samples for LC-MS analysis were taken after 18 hrs of incubation with PS-CBH.
For HPLC determination of reaction conversion, the GenB4 reaction was performed, methanol added and the pH adjusted to pH 5.0 as described above. The reaction was split into fractions and 10 equivalents of 4-chloroaniline (3d), 20 equivalents of benzylamine (3f), phenethylamine (3g), furfurylamine (3j) or 25 equivalents of tyramine (3h) were added and reactions were stirred for 1 hr at 25 °C. Ten equivalents of polymer-supported cyanoborohydride was added and the reaction further incubated at 25 °C for 41 hrs. After incubation, the reactions were briefly centrifuged, 800 μL of the supernatant was separated from the beads and the MeOH from the reaction was removed by centrifugal evaporation for 4 -5 hrs at room temperature. The remaining aqueous phase was basified to pH 10.0 -11.0 with 12 M NaOH and the excess amine was extracted with EtOAc (3 x 400 μL). The pH of the aqueous phase was re-acidified to pH 7.0 -8.0 with 1 M HCl and these solutions were analysed by analytical HPLC. Variations in sample volume due to pH adjustments were recorded and accounted for to determine final analyte concentrations prior to HPLC analysis. For HPLC conditions see section 1.14.3 and Table S4. Samples were kept at -20 °C prior to analysis.

Antimicrobial activity screening of C-6' gentamicin C1a derivatives 1.12.1 Minimal inhibitory concentration determination
Minimal inhibitory concentration assays were performed according to the guidelines specified in the literature 4 . Assays were performed in triplicate in sterile 96-well plates. The following wild-type bacteria were streaked onto LB agar: Escherichia coli ATCC 25922, Pseudomonas aeruginosa ATCC 27853, Staphylococcus aureus ATCC 29213, Salmonella typhimurium LT2, Bacillus subtilis RM125 and Klebsiella pneumoniae. Four colonies of each strain were inoculated in Mueller Hinton broth and incubated overnight at 37 °C/250 rpm. The overnight culture was diluted 100-fold and incubated until the cells reached log-phase. At this point the cells were diluted to an OD625 approximately equivalent to a CFU/mL of 1 x 10 8 and then diluted 100-fold to give a concentration of approximately 1 x 10 6 CFU/mL. All dilutions were done in Mueller Hinton broth.
C-6' gentamicin C1a derivatives were dissolved in MilliQ H2O to a final concentration of 2 mg/mL (free base basis) and then diluted to the following concentrations dependent on the bacterium being tested: E. coli and S. typhimurium -4 μg/mL, P. aeruginosa and S. aureus -16 μg/mL, B. subtilis and K. pneumoniae -1 μg/mL in Mueller Hinton broth. These solutions were subsequently serially two-fold diluted 6 times in Mueller Hinton broth in each well of the microtiter plate to give the antibiotic concentrations ranges tested. An equivalent volume of cells were added to each well to a final cell concentration of approximately 5 x 10 5 CFU/mL. All plates contained a negative control which had no cells, a positive control lacking antibiotic and a gentamicin C1a positive control in triplicate to control for interplate variation. Plates were sealed with Breathe-Easier sealing film (Diversified Biotech) and incubated at 37 °C static overnight. After incubation, the OD600 was measured. For MIC determination, the turbidity measurements were background subtracted with the negative control, the turbidity of each well was averaged across replicates and expressed as a percentage turbidity compared to the positive control lacking antibiotic. The antibiotic concentration which caused a greater than 90% reduction in turbidity compared to control lacking antibiotic was taken as the MIC.

Generation of E. coli ATCC 25922 strains carrying resistance plasmids and testing
Chemically competent cells of E. coli ATCC 25922 were transformed with resistance plasmids R26 (80 kb, carrying an aac(3)-I gene), pSa (37 kb, carrying an aac(6')-Ib gene) and pQR1865 (A BglII fragment of pSa which contains the aac(6')-Ib and ant(3") genes). Transformants were screened for the presence of each plasmid by mini-prep and restriction digestion using NotI for R26 and HindIII for pSa and pQR1865. Transformants were also screened for antibiotic resistance on LB media containing 10 μg/mL gentamicin C1a, 50 μg/mL kanamycin or 100 μg/mL streptomycin and grown at 37 °C for 24 hrs.
Antimicrobial activity testing was performed in triplicate against the resistant E. coli strains as described above in section 1.12.1. The antibiotic concentration range tested was 32 μg/mL -0.0625 μg/mL.

Antimicrobial activity screening on solid media
E. coli ATCC 25922, E. coli ATCC 25922 + pSa, E. coli ATCC 25922 + R26, E. coli ATCC 25922 + pQR1865 were grown in LB overnight at 37 °C/250 rpm, diluted to an OD600 of 0.141 and 100 μL was spread onto LB solid media containing 2 % agar and dried. Using a flamesterilised metal cork borer, wells were created in the media, and 100 μL of test antibiotics (0.1 mg/mL aqueous solutions) were added to each well. Gentamicin C1a was used as a positive control. Plates were incubated upright at 37 °C for 1 day, after which antibiotic activity was assessed for the presence and size of zones of clearing.

Tetraphenylborate precipitation
The method used was adapted from that reported by Katano and co-workers 5 . The residues obtained from work-up of the GenB4 + reductive amination reactions were dissolved in 500 μL of H 2 O and mixed with an equal of volume of 300 mM sodium tetraphenylborate aqueous solution. The white precipitate formed was centrifuged at max. speed for 1.5 mins and the supernatant discarded. The precipitate was washed 3 times with 1 mL of H2O, centrifuging at the max. speed each time, and removing the supernatant. The pellet obtained was dissolved in 1 mL of acetone by vortexing and then an additional 1 mL of acetone was added. 50 μL of 5 M HCl(aq.) solution was added dropwise and a white to off-white precipitate formed after centrifugation. This pellet was washed 3 times with 1 mL of acetone and then dried by rotary evaporation to obtain a crudely purified aminoglycoside product as the HCl salt. S8

Exchange of Amberlite CG50(H + ) resin for CG-50(NH4 + ) form
The method used follows that described by Chen and colleagues 6 . Amberlite CG-50(H + form) resin (Type I, 100 -200 mesh, Sigma-Aldrich) was suspended in 5 times the resin volume of 1 M NH4OH(aq.) solution and shaken at 150 rpm for 4 hrs at room temperature. The resin was then washed with 5 times the resin volume of 1 M HCl(aq.) solution followed by another 5 times volume of 1 M NH4OH(aq.) solution. Between resin acid-base wash steps the resin was washed with deionised water to remove residual NH4OH and HCl until the pH of the solution was near neutral. The resin was collected by vacuum filtration and stored at room temperature in a glass bottle.

Exchange of Amberchrom 1 x 2 (Cl -) resin for 1 x 2 (OH -) form
The method used follows that described by Umezawa and Kondo 7 . Amberchrom 1 x 2 (Clform) resin (200 -400 mesh, Sigma-Aldrich) was transferred to a large glass column using deionised water and then washed with 2 resin volumes of 1% Na2SO4 solution followed by 10 resin volumes of 10 % NaOH(aq.) solution. The resin was then washed with deionised water until the pH of the eluted solution was near neutral. The resin was dried and stored in a glass bottle at room temperature.

High performance anion exchange chromatography with integrated pulsed amperometric detection (HPAE-IPAD)
HPAE-IPAD analysis was performed using a Reagent-Free Ion Chromatography System (ICS) (ICS-5000+ system, Thermo Scientific) fitted with CarboPac PA1 (4 x 250 mm) anion exchange column with a CarboPac PA1 guard column (4 x 50 mm), an Eluent Generator with a KOH 500 cartridge and an electrochemical detector (gold electrode). The flow rate was 1 mL/min and column temperature was 30 °C. The sample tray was cooled to 4 °C. An injection volume of 10 μL was used and samples were prepared in 96-well polypropylene plates. Analytical methods used are described in Table S3.

Liquid chromatography -mass spectrometry
Low resolution liquid-chromatography tandem mass spectrometry (LC-MS/MS) analyses were performed on a Hypersil GOLD C18 column (150 x 2.1 mm, 1.9 μm particle size, Thermo Scientific) fitted on an Accela 600 LC system connected to a Finnigan LTQ Linear ion trap mass spectrometer. Analyses were performed in polypropylene vials (Waters). The mobile phases were A: H2O + 0.1 % pentafluoropropionic acid (PFPA), B: MeCN + 0.1 % PFPA and the gradient method previously reported 8

Analytical high pressure liquid chromatography (HPLC)
Analytical HPLC analysis was performed on an Agilent 1260 Infinity analytical LC system fitted with an C 18 -AR (4.6 x 150 mm, ACE) analytical column and guard column. The injection volume was 10 μL, the detection wavelength was 214 nm and the column temperature was 25 °C. The mobile phases were A: H2O + 0.1 % trifluoroacetic acid, B: MeOH. Compound specific gradient methods are described in Table S4.

Semi-preparative HPLC
Semi-preparative HPLC was performed on an Agilent 1260 Infinity II preparative LC system fitted with an Agilent ZORBAX 300SB-C18 (250 x 9.4 mm, 5 μm particle size) column. The mobile phases were A: H2O + 0.1 % trifluoroacetic acid, B: MeOH or MeCN. Preparative separations were performed at room temperature with a flow rate of 4.2 mL/min. Samples were prepared in polypropylene vials (Waters) and fractions were collected in 13 x 100 mm polypropylene tubes (Greiner Bio-One). Compound specific gradient methods used for preparative LC are described in Table S5.

General synthesis of 6'-gentamicin C1a derivatives
GenB4 enzyme solution was prepared as described in section 1.5 and added to a final concentration of 0.3 -0.5 mg/mL in a reaction containing 50 mM NaPi buffer pH 8.0, 0.5 mM PLP and 2.5 mM sisomicin. The reaction was incubated for 1 hr at 30 °C/175 rpm. After 1 hr, an equal volume of cold MeOH was added and the mixture was centrifuged 10,000 rpm/4°C for 5 minutes to remove the protein from the reaction mixture. The supernatant was transferred to a new Falcon tube and the pH adjusted to 5.0 with concentrated AcOH. 10 -25 Equivalents of amine were added, and the reaction stirred for 1 hr at 25 °C. Ten equivalents of polymersupported cyanoborohydride was added and the reaction further incubated at 25 °C for 17 -48 hrs. Reaction progress was monitored by LC-MS as described in section 1.14.2. After incubation, the beads were separated from the reaction mixture and the reaction solvent was removed by rotary evaporation. The remaining residue was washed with EtOAc (3 x 1 mL) to remove excess amine. The compounds were then further purified as specified below for each compound. For NMR and mass spectrometry spectra see Appendices.
Prior to activity testing all compounds were converted to hydrochloride salts by dissolution in 0.2 M HCl solution and freeze-drying. Gentamicin C1a free base obtained from Carbosynth was also converted to a HCl salt by this same process. The products were hygroscopic and were stored under argon at -20 °C.

6'-N-(Phenyl)-gentamicin C1a (4a)
Ten equivalents of aniline (3a) were added to the reaction and the reductive amination was incubated for 17 hrs. The product was purified by semi-preparative HPLC to give 4a as a pale yellow crystalline solid.

6'-N-(2-Chlorophenyl)-gentamicin C1a (4b)
Ten equivalents of 2-chloroaniline (3b) were added to the reaction and the reductive amination was incubated for 43 hrs. The product was purified by semi-preparative HPLC to give 4b as a yellow crystalline solid.

6'-N-(3-Chlorophenyl)-gentamicin C1a (4c)
Ten equivalents of 3-chloroaniline (3c) were added to the reaction and the reductive amination was incubated for 18 hrs. The product was purified by semi-preparative HPLC to give 4c as a yellow crystalline solid.

6'-N-(4-Chlorophenyl)-gentamicin C1a (4d)
Ten equivalents of 4-chloroaniline (3d) were added to the reaction and the reductive amination was incubated for 18 hrs. The product was purified by semi-preparative HPLC to give 4d as a yellow crystalline solid.

6'-N-(Pyridin-3-ylamino)-gentamicin C1a (4e)
20 equivalents of 3-aminopyridine (3e) were added to the reaction and the reductive amination was incubated for 43 hrs. The product was purified by semi-preparative HPLC to give 4e as a yellow to orange crystalline solid.

6'-N-(Benzyl)-gentamicin C1a (4f)
20 equivalents of benzylamine (3f) were added to the reaction and the reductive amination was incubated for 48 hrs. The product was purified by semi-preparative HPLC to give 4f as a white to off-white crystalline solid.

6'-N-(Phenethyl)-gentamicin C1a (4g)
20 equivalents of phenethylamine (3g) were added to the reaction and the reductive amination was incubated for 41 hrs. The product was purified by tetraphenylborate precipitation and semi-preparative HPLC to give 4g as a white crystalline solid.

6'-N-(4-hydroxyphenethyl)-gentamicin C1a (4h)
25 equivalents of tyramine (3h) were added to the reaction and the reductive amination was incubated for 48 hrs. 2-propanol was used to remove excess amine instead of EtOAc. The product was purified by tetraphenylborate precipitation and semi-preparative HPLC to give 4h as a white to off-white crystalline solid.

Supplementary Discussion
The E. coli ATCC 25922 host strain appeared to have 4 naturally occurring plasmids, as six bands were observed at 1000 -1500, 2000, 3000, 4000 and two above 10,000 bp in the undigested lane of the control strain ( Figure S10). The top bands (orange asterisk) could be open circle (OC) and super-coiled (SC) forms of a medium copy number plasmid, while the faint bands around 3000 -4000 could be OC and SC forms of a plasmid which is present at a low copy number per cell. The two other plasmids appear to have higher copy numbers. Upon digestion with HindIII, the same bands were observed apart from a shift of the band at ~ 2000 bp to ~ 3000 bp (suggesting that the band at 2000 bp is the SC form of a low molecular weight plasmid), disappearance of the top band above 10,000 bp and two additional faint bands at 2000 and 2500 bp ( Figure S10). For the NotI digest of E. coli ATCC 25922 plasmids, no difference was observed compared to the undigested plasmid lane ( Figure S10). The genome of E. coli ATCC 25922 has been sequenced and it was shown that the organism contains two plasmids, one at 48 kb and another at 24 kb 9 . The 48 kb plasmid is not cut by HindIII and is linearised by NotI. Conversely, the 24 kb plasmid is not cut by NotI but is cut by HindIII at 3 sites, producing bands of 20.9 kb, 2.6 kb and 0.6 kb in length. The bands observed above 10000 bp and ~ 2500 bp in the HindIII digest may correspond to the 24 kb plasmid. The band at ~ 3000 bp may be the linearised form of the small plasmid which is observed around 2000 bp. Similar band patterns were observed for all other lanes for the pSa and R26 clones ( Figure  S10). In contrast to pQR1865, the low copy number of pSa and R26 made it difficult to determine if the plasmids were present by gel electrophoresis analysis alone. S17  Table S1: Sources of C-6' TAm sequence data used in this study.

Oligo Name Sequence
PchB fw AAAGCTCTTCGATGAAACAGGATATGGTTGAAAATTCGGAAAAGC PchB rev AAAGCTCTTCGGTGACCCCCACCGATCAGATTATCGAG     Table S4: Compound-specific conditions for analytical HPLC. All gradients were run for 16 mins in methanol followed by a 60 % methanol wash step for 3 mins and reequilibration at the starting mobile phase percentage for 5 mins. a = retention times are shown as an average value for the method, however due to the ion-pairing retention mechanism, the retention times of the compounds drifted substantially over the course of the run. Rt = retention time. Figure Figure S5: HPAE-IPAD analysis of side product formation in GenB4 reaction with sisomicin. HPAE-IPAD chromatograms are displayed which show the 1f peak at t0 is completely converted to 2e after 1 hr. After 24 hrs, 2e is completely converted to the side product. Alternatively, incubation for 10 minutes to denature the GenB4 enzyme also leads to side product formation. G4 = GenB4, G4 denat. = GenB4 enzyme which is denatured by heat treatment.       From left to right -10 μg/mL gentamicin C1a, 50 μg/mL kanamycin, 10 μg/mL streptomycin. "-" = E. coli ATCC 25992 host strain, pSa BglII = pQR1865. (B) Same as for (A) only for the R26 clones. Mild resistance of the host strain to streptomycin is evident, which is increased when the strain is cloned with the resistance plasmids. All plates contain re-streaks of individual colonies grown on the selection plates after transformation. "1 hr" and "ON" refers the length of the recovery period at 37 °C post transformation, which was either 1 hr or overnight.