Carbapenems as water soluble organocatalysts

Background: Identification of organocatalysts functioning in aqueous environments will provide methods for more sustainable chemical transformations and allow tandem reactions with biocatalysts, like enzymes. Here we examine three water-soluble carbapenem antibiotics (meropenem, doripenem, and ertapenem) as secondary amine organocatalysts in aqueous environments. Methods: The Michael addition of nitromethane to cinnamaldehyde was used as the model reaction. The reactions were monitored by 1H NMR, and the enantioselectivity was determined by chiral HPLC. Results: The effects of buffer components, pH, organic co-solvents and anchoring into a protein scaffold were investigated. Moderate yields of the Michael addition were obtained in buffer alone. The use of methanol as a co-solvent in a ratio of 1:1 increases the yield by 50%. Anchoring of the catalysts into a protein backbone reverses the enatioselectivity of the reaction. Conclusions: Despite only moderate yields and enantioselectivities being obtained, this study lays the foundations for future development of efficient organocatalysis in aqueous environments.


Introduction
The need for more sustainable catalysts in chemical transformations continues to attract significant interest 1 . Iminium ion catalysis facilitated by secondary amines retrieved considerable attention in the early 2000's due to their ability to activate enals for enantioselective nucleophilic addition 2 . Now known as organocatalysts, these small molecules are reasonably cheap, non-toxic, sustainable, and stable (i.e. tolerant to moisture and air) 3 . However, attempts to use organocatalysts in solvents like water as a homogenous system are yet to achieve great success 4 . This is because most organocatalysts have bulky hydrophobic groups that are important for creating chiral environments, but significantly lower their water solubility 5 . Not surprisingly, one of the most well-known water soluble organocatalysts is l-proline (1, Figure 1A) that bears no bulky hydrophobic group. However, at neutral pH in water the proximity of the carboxylate to the amine of l-proline, making the amine more prone to be protonated and consequently inhibiting formation of the substrate iminium ion and leading to poor catalytic activity 6 . In addition, the small chiral substituent in proline often hampers the enantioselectivity of the reactions.
Besides proline, several carbapenems (e.g. meropenem 2a, doripenem 2b, and ertapenem 2c; Figure 1) also have a secondary amine in the form of pyrrolidine. Carbapenems are βlactam antibiotics that inhibit enzymes in bacterial cell wall biosynthesis 7 . Just like proline, carbapenems are completely water soluble. However, the lack of an adjacent carboxylate group in carbapenems implies that these molecules should be able to form an iminium ion readily in aqueous environments for efficient catalysis. On the other hand, substituents around the pyrrolidine ring of carbapenems are larger than that of proline and likely to induce enantioselectivity for the catalysis 8 . In addition, an enhanced chiral environment can be created by anchoring carbapenems into a protein environment. For example, a covalent adduct 3 ( Figure 2) is formed during the metabolism of carbapenems by penicillin binding proteins, and this intermediate can be trapped by mutating the key glutamic acid residue to an alanine residue to prevent the hydrolysis of the ester intermediate 9 .
Here we explored the uses of three carbapenem antibiotics, meropenem 2a, doripenem 2b and ertapenem 2c, as organocatalysts in aqueous systems. The effects of buffer components, pH, and organic co-solvents on reaction yield were investigated. We also tested whether anchoring a carbapenem in a protein scaffold increases the enantioselectivity of the reaction.
For reactions carried out at pH 7.5 and pH 8.0, the buffers were adjusted with 1 M NaOH, and the pH measured using a pH meter. For the co-solvent screen the reactions were performed by first adding 245 µl of the indicated buffer followed by 250 µl of the indicated solvent.
For reactions in buffer alone, the reaction mixtures were extracted with 700 µl of CDCl 3 and analyzed by 1 H NMR spectroscopy. For those reactions with methanol or acetonitrile as the co-solvent, the solvents were evaporated under reduced pressure, and the crude mixture was extracted with 700 µl of CDCl 3 . Reactions containing benzene or chloroform as the co-solvent were performed using the corresponding deuterated solvents and subjected directly to NMR analysis. Product yields were estimated from integration of signals arising from cinnamaldehyde 6, product 7, and side product 8. 1 H NMR spectra were recorded in CDCl 3 or C 6 D 6 on a Bruker Ascend 500 MHz or a Bruker Fourier 300 MHz instrument. Chemical shifts are reported in parts per million (ppm) and are referenced to the residual solvent resonance as the internal standard (CHCl 3 : δ = 7.26 ppm and C 6 H 6 : δ = 7.15 ppm). Spectra were analysed using Bruker TopSpin version 3.5 11 .

Determination of enantioselectivity
The stereoselectivity of products 7 were determined by chiral high-performance liquid chromatography (HPLC) analysis. Enantioenriched samples of (S)-and (R)-nitro products for peak assignment in chiral HPLC measurements were obtained using l-and d-Jørgensen-Hayashi catalysts according to the literature procedure 5 . A racemic sample of the nitro product was obtained using piperidine as the catalyst 12 . The aldehyde was reduced to the alcohol for determining the ratio of two enantiomers (Scheme S1).  The reactions were scaled up 10 times with respect to the above screening conditions for the reactions with meropenem and doripenem in pH 7.0 PBS buffer and with 50% methanol and performed in 10-ml round bottom flasks with continuous stirring at room temperature for 24 h. The protein reactions were scaled up 10 times with respect to the reactions used for screening.
The reaction mixture was then extracted with DCM (10 ml × 3), and the organic fractions were combined, dried over anhydrous Na 2 SO 4 , filtered and concentrated under reduced pressure to ca. 1 ml before purification by preparative TLC (EtOAc: hexane = 25:75). The silica gel was scraped from the plate and stirred in 1% MeOH:DCM (10 ml). The suspension was filtered and evaporated under reduced pressure.
The resulting aldehyde was dissolved in methanol (5 ml), and to this was added ca. 5 equivalents of NaBH 4 . The reaction was stirred overnight at room temperature. The reaction mixture was then neutralized with 1 M HCl (aq) , and then extracted with DCM (10 ml x 3). The organic fractions were combined, dried over Na 2 SO 4 , filtered and concentrated under reduced pressure. The resulting alcohol was purified by preparative TLC (EtOAc:hexane = 35:65). The silica was scraped from the plate and stirred in 1% MeOH:DCM (10 ml). The solution was filtered and evaporated under vacuum.
The purified alcohol was dissolved in 1 ml of 20% 2-propanol in hexane (HPLC grade). 20 µl of each sample was injected onto the HPLC.
The reactions catalyzed by meropenem 2a and doripenem 2b were scaled up ten times from the above conditions and the reaction mixtures were extracted with dichloromethane, dried over Na 2 SO 4 , filtered, and concentrated. The product 7 was purified by preparative TLC (25% EtOAc in hexane). The product was dissolved in 5 ml of methanol, and approx. 5 equivalents of NaBH 4 were added. After overnight stirring at room temperature, the reaction mixture was neutralized with 1 M HCl (aq) , extracted with dichloromethane (3 × 10 ml), dried over Na 2 SO 4 , filtered, and concentrated. The resulting alcohol 9 was purified by preparative TLC (35% EtOAc in hexane) before analysis by chiral HPLC using a Phenomonex Cellulose-Lux1 analytical chiral column held at 20°C (isocratic elution 0.5 ml/min with 25% 2-propanol in n-hexane; detection by absorbance at 210 nm).
To determine the enantioselectivity of the BlaC-carbapenem complexes, the reactions were scaled up 10 times with regards to the small-scale screening reactions. Enantioselectivity was then determined using the above method.
Cloning, expression and purification of BlaC(E166A) The gene encoding for the wild-type Mycobacterium tuberculosis β-lactamase (BlaC) without the 40-amino acid leader sequence was purchased as a double-stranded fragment (GeneArt, Invitrogen; see Supporting Information for the exact DNA sequence). This was cloned into a NdeI and BamHI digested pET28a vector by Gibson assembly to yield the wild type gene with an N-terminal 6 his-tag originating from the vector. Briefly, 25 ng of the linear gene was added to 100 µg of digested plasmid and to this was added 0.5 µl of sterile H 2 O followed by 2.5 µl of Gibson assembly master mix (NEB). The mixture was then incubated at 50°C for 1 hour. The products were transformed into chemically competent MDS42 E. coli cells and grown overnight. Colonies were selected and cultured overnight after which the plasmid was purified by the miniprep method (Qiagen). The construct was confirmed by DNA sequencing.
The E166A mutation was introduced by site-directed mutagenesis PCR using PrimeStar HS DNA polymerase (Clonetech) with primers CTGGATGCAGAAGCACCGGAACTGAATC (forward) and GATTCAGTTCCGGTGCTTCTGCATCCAG (reverse). PCR was performed over 33 cycles. The initial temperature was 95°C and held for 3 minutes. Each cycle comprised of 10 seconds for denaturation at 98°C, followed by annealing for 5 seconds at 60°C and finally an extension of 6 minutes and 30 seconds at 72°C. The final extension was 10 minutes. The mutant construct was confirmed by DNA sequencing (Eurofins, Genomics) using the T7 promoter primer (TAATACGACTCACTATAGG).
Plasmid pET28a BlaC(E166A) was then transformed into chemically competent BL21(DE3) cells and grown on LB agar plates supplemented with kanamycin (37.5 µg/ml). One colony from the plate was picked to inoculate a 10 ml LB starter culture containing kanamycin (37.5 µg/ml). After overnight, the starter culture was diluted into 1 L of fresh LB media containing kanamycin (37.5 µg/ml). The cells were grown at 37°C until they reached an OD 600 of 0.8, and IPTG was added to reach a final concentration of 0.5 mM. The cells were then incubated overnight at 20°C. The cultures were harvested and the dry pellet was stored at -20°C.
To purify the enzyme, the cells were lysed by sonication in sodium phosphate buffer (35 ml per litre pellet, 50 mM NaPi, 100 mM NaCl, 10 mM imidazole, pH 8.0) with either 5 mg meropenem or doripenem added to the re-suspended cells. The solids were separated by centrifugation at 20,000 rpm for 30 minutes. The supernatant was loaded on to a Ni-NTA column equilibrated with the above buffer. The column was washed twice with wash buffer (25 ml, 50 mM NaPi, 100 mM NaCl, 25 mM imidazole, pH 8.0) and then the protein was eluted in phosphate buffer (10 ml, 50 mM NaPi, 100 mM NaCl, 300 mM imidazole, pH 8.0). 1 mg of doripenem or meropenem was added to the elution fraction and incubated for 30 minutes at room temperature. Fast protein liquid chromatography (FPLC) was performed on an ÄKTA purifier (GE Healthcare) system at room temperature using a ProteoSEC size exclusion column (Generon, SEC-3-70-100 ml, 26 mm ID, 60 cm length, 3-70 kDa HR resin). Protein elution was monitored by UV absorbance at 280 nm and the elution buffer was 50 mM sodium phosphate buffer 100 mM NaCl pH 7.0. Fractions containing BlaCcarbapenem complexes 3 were combined and concentrated using a 10 kDa cutoff centrifugal concentrator. Expression and purification were confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Figure 3). SDS-PAGE was performed using self-casted 12% acrylamide gels and run for 50 minutes at 200 V. SDS-PAGE gels were stained using Coomassie Brilliant Blue. Unstained protein molecular weight marker (Thermo Scientific) was run alongside the samples.
Protein-carbapenem complexes were confirmed by Liquid chromatography mass spectrometry (LC-MS). LC-MS was performed on a Waters Synapt G2-Si quadrupole time of flight mass spectrometer coupled to a Waters Acquity H-Class UPLC system. The column was an Acquity UPLC protein BEH C4 (300 Å 1.7 µm × 2.1 mm × 100 mm) operated in reverse phase and held at 60°C. The gradient employed was 95% A to 35% A over 50 minutes, where A is H 2 O with 0.1% HCO 2 H and B is acetonitrile (ACN) with 0.1% HCO 2 H. Data was collected in positive ionization mode and analysed using the Waters MassLynx software version 4.1. Deconvolution of protein charged states was obtained using the maximum entropy 1 processing software.

Results and discussion
Here we chose the Michael addition of nitromethane 5 to cinnamaldehyde 6 (Scheme 1) as the model reaction to test whether meropenem 2a, doripenem 2b, and ertapenem 2c can function as catalysts. We first investigated different buffer conditions (Table 1). In the presence of 20% catalyst, all four buffers produced similar yields with each catalyst, but significantly higher yields were observed for meropenem 2a and doripenem 2b in comparison to ertapenem 2c. It is noteworthy that no product was formed in non-buffered conditions (i.e. pure water), suggesting the need for a controlled pH throughout the catalytic cycle 13 . On the other hand, the presence of NaCl in PBS buffer had minimal impact on product yield. Since PBS is a common buffer used to mimic biological environments, this buffer was chosen for further investigations.

Modification of pH
We then shifted our attention to modify the reaction pH (Table 2). PBS adjusted to the relevant pH was employed. Although the conversion of the starting material 6 seems to increase at the elevated pH, the yield of the product did not. We found that this was due to formation of the side product 8 by 1,2-addition of nitromethane to cinnamaldehyde 6. As side product formation was least prominent at pH 7.0, we determined that this is the preferred pH for the reaction.

Optimization of yield
To optimize the reaction yield, we decided to explore the effects of different organic solvents as a co-solvent (Table 3). Four solvents were chosen and tested in a 1:1 ratio to the PBS buffer   or if the concentration of the BlaC-carbapenem complex 3 exceeded 500 µM in PBS. Thus, we were not able to test the conditions giving the highest yields (i.e. 1:1 = PBS: MeOH), and the reactions were carried out at diluted concentration in PBS (Table 5). Under the new conditions, we were able to maintain the ratio of BlaC-carbapenem complexes (20 mol% 3) to cinnamaldehyde 6 and negligible precipitation of protein complexes was observed during the time course of the reaction. Under the new (diluted) condition, as expected, lower yields were obtained with either meropenem 2a or doripenem 2b compared to the previous conditions ( Table 1, Table 5). While no product was obtained in the presence of BlaC only, some products were obtained in the presence of BlaC-meropenem 3a or BlaC-doripenem 3bcomplexes, although there was no significant change in product yield compared to the carbapenems 2 alone (Table 5). Notably, inversion of enantioselectivity was observed when the catalyst was anchored in the protein scaffold, although the enantioselectivity is still low. Raw data for all experiments conducted in this study, including the supporting information, are available on figshare 15 .

Conclusions
Here we investigated the ability of carbapenem antibiotics 2a-c to catalyze the Michael addition of nitromethane to cinnamaldehyde 6 in aqueous environments. While ertapenem 2c is not effective, catalysis by meropenem 2a or doripenem 2b provides the product 7 with moderate yields and enantioselectivities.
The results suggest the carbapenems 2a-c can form iminium ions in situ, facilitating the nucleophilic addition of nitromethane for Michael addition in water. The charged iminium intermediate may be stabilized by polar protic solvents (e.g. water, methanol). This is consistent with the observation that higher yields were obtained when using methanol as a co-solvent. On the other hand, polar aprotic solvents (e.g. acetonitrile) had little effect. Addition of non-miscible solvents (e.g. benzene, chloroform) into the buffer system was not beneficial for catalysis, most likely due to the preferential partition of the starting materials into the organic phase, whereas the catalyst remains in the aqueous phase, hindering catalysis.
It does not seem that the substitution around the active nitrogen is enough to induce high enantioselectivity onto the product at 25°C. Attack from either the re or si face of the iminium ion seems to be equally possible for meropenem 2a and doripenem 2b.  (10 mM, pH 7.0). These solvents are either miscible (methanol and acetonitrile) or non-miscible (benzene and chloroform) with the buffer. Addition of methanol was found to significantly increase the product yield, whereas acetonitrile had negligible effects, and the use of the non-miscible solvents totally abolished the product formation.
As reasonable yields could be obtained using either meropenem 2a or doripenem 2b as the catalyst, chiral HPLC was employed to identify the enantioselectivity under these conditions. Reactions were performed on a larger scale to facilitate product purification and isolation by preparative TLC. The aldehyde functionality in the product 7 was reduced by treatment with NaBH 4 to afford the corresponding alcohol 9 before analysis by chiral HPLC (see Supporting Information) 5 . In general, no significant difference in the enantioselectivity was observed for either meropenem 2a or doripenem 2b (Table 4), and the addition of methanol also had little effect. In fact, enantiomeric excess is low in all cases, with the best result of R:S = 41:59 observed with meropenem 3a in the mixture of PBS and methanol.

Improving enantioselectivity
To improve the enantioselectivity, we anchored meropenem 2a or doripenem 2b into the M. tuberculosis β-lactamase BlaC. This enzyme is known to metabolize carbapenems ( Figure 2A). However, mutation of Glu166 to Ala can prevent hydrolysis of the ester intermediate 3 14 , enabling stable anchoring of the antibiotics to BlaC. Recombinant BlaC(E166A) containing a N-terminal His-tag was produced in Escherichia coli BL21 (DE3). Cells were lysed in the presence of meropenem 2a or doripenem 2b, and the protein complex was purified by nickel affinity and size exclusion chromatography (Figure 3). Formation of the protein complexes 3a and 3b with covalently bound ester intermediate were confirmed by mass spectrometry (Figure 4).
The Michael reaction was then performed with the BlaCcarbapenem complexes 3a and 3b. Attempts to perform reactions with BlaC-carbapenem 3a or 3b using the experimental conditions in Table 4 were not successful as the enzyme complexes precipitated in buffers containing > 10% (v/v) methanol The crystal structure of the BlaC-carbapenem complex ( Figure 2) suggests that access to the amine by the substrates is still possible when meropenem 2a is bound to BlaC. Indeed, anchoring the carbapenems 2a and 2b into the protein produced similar yields to the catalysts alone. However, an inversion of stereoselectivity was observed for the BlaC-carbapenem complexes 3, suggesting that the protein environment surrounding the catalyst determines the enantioselectivity. The low enantioselectivity may be due to the sub-optimal local environment provided by the wild-type protein, and engineering of this pocket by mutagenesis may increase both reaction yields and enantioselectivities, as demonstrated previously in artificial metalloenzymes 16 .
In this study, we have shown that water-soluble carbapenem antibiotics 2a-c can be repurposed to catalyze Michael addition. Unlike most organocatalysts used in aqueous systems 17 , these catalysts do not require emulsion or biphasic systems. Separation of the catalysts from the product can be achieved easily by extraction without the need for chromatography. Identification of organocatalysts functioning in aqueous environments will provide more sustainable approaches for chemical transformations and allow tandem reactions with biocatalysts, such as enzymes. However, there exists room to improve the system reported here, as only moderate yields and enantioselectivities were obtained. Nonetheless, this study lays the foundations toward developing efficient iminium catalysis in  water and provides new strategies of anchoring small molecule catalysts into chiral protein environments.

Data availability
The H NMR spectra, chiral HPLC chromatograms and full protein LC-MS chromatograms, and mass spectra datasets generated during and/or analyzed during the current study are available in the figshare repository: https://doi.org/10.6084/ m9.figshare.6973880 15 .

Grant information
We are grateful for support from the Cardiff School of Chemistry, Leverhulme Trust (RPG-2017-195 to L.Y.P.L.) and Wellcome Trust (202056 to L.Y.P.L., and 200730 to Y.H.T.). We also thank EPSRC (EP/L027240/1) for acquisition of Waters Synapt G2-Si LC-MS system.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Supplementary material
Supporting information. Further results supporting the conclusions of this article. Included are spectra for Michael addition reactions, chiral HPLC chromatograms, relevant DNA and amino acid sequences and LC-MS analysis of BlaC/BlaC-carbapenem complexes.
Click here to access the data