Chemo-bio catalysis using carbon supports: application in H2-driven cofactor recycling

Chemical

Science

View Article Online

View Journal | View Issue

EDGE ARTICLE

Chemo-bio catalysis using carbon supports:

application in H₂-driven cofactor recycling†

Cite this: Chem. Sci., 2021, 12, 8105

a

Xu Zhao,‡^aSarah E. Cleary, ‡^aCeren Zor,^bNicole Grobert,

^bHolly A. Reeve

*

All publication charges for this article

have been paid for by the Royal Society

of Chemistry

a

*

and Kylie A. Vincent

Heterogeneous biocatalytic hydrogenation is an attractive strategy for clean, enantioselective C]X

reduction. This approach relies on enzymes powered by H₂-driven NADH recycling. Commercially

available carbon-supported metal (metal/C) catalysts are investigated here for direct H₂-driven NAD⁺

reduction. Selected metal/C catalysts are then used for H₂oxidation with electrons transferred via the

conductive carbon support material to an adsorbed enzyme for NAD⁺reduction. These chemo-bio

catalysts show improved activity and selectivity for generating bioactive NADH under ambient reaction

conditions compared to metal/C catalysts. The metal/C catalysts and carbon support materials (all

activated carbon or carbon black) are characterised to probe which properties potentially inﬂuence

catalyst activity. The optimised chemo-bio catalysts are then used to supply NADH to an alcohol

dehydrogenase for enantioselective (>99% ee) ketone reductions, leading to high cofactor turnover

numbers and Pd and NAD⁺reductase activities of 441 h^ꢀ1and 2347 h^ꢀ1, respectively. This method

demonstrates a new way of combining chemo- and biocatalysis on carbon supports, highlighted here

for selective hydrogenation reactions.

Received 15th January 2021

Accepted 30th April 2021

DOI: 10.1039/d1sc00295c

rsc.li/chemical-science

scale. Therefore, continual recycling of a catalytic quantity of

NAD⁺/NADH is necessary for these reactions to be industrially

1. Introduction

viable, and current practices rely on enzymatic cofactor recy-

cling driven by superstoichiometric reductants (e.g. glucose)

leading to a build-up of waste.¹¹Clean, highly atom-economical

H₂-driven NADH generation has been reported using whole cell

systems,¹²isolated enzymes,¹³and heterogeneous biocatalytic

strategies.^14,15The enzymes contain an H₂oxidation site and an

NAD⁺reduction site that are in direct electronic contact, either

natively or provided by co-immobilisation on an electronically

conducting carbon support. These methods are selective for

producing bioactive NADH and have been demonstrated with

a range of C]X bond-reducing enzymes.^14,16,17

Organometallic catalysts have also been reported for medi-

ated H₂-driven NAD⁺reduction. Early examples relied on H₂-

driven reduction of electron carriers (e.g. pyruvate, Safranine)

that went on to reduce NAD(P)⁺.^18,19More recently, homoge-

neous organometallic complexes have been used to produce

NADH under H₂without the need for mediators,²⁰though these

complexes are known to face biocompatibility issues when used

in one-pot reactions with enzymes that consume NADH (e.g.

alcohol dehydrogenases).^6,21

The ability to combine metal- and enzyme-catalysed steps in

one-pot reactions simplies synthetic approaches that rely

upon both chemo- and biocatalysis.^1–3Common challenges

associated with chemo-bio systems are reaction condition

compatibility (e.g. solvent, pH, temperature, pressure) and

mutual poisoning of metal and enzyme.^4–7Optimisation of

catalyst preparation and reaction conditions are therefore

required to ensure both metal and enzyme catalysts are simul-

taneously active.⁸Reaction times and processing steps are oen

minimised when both catalysts are immobilised on heteroge-

neous supports,⁹therefore the known benets of chemo-bio

catalysis could be amplied by co-immobilising the two cata-

lysts onto a heterogeneous support.

Many synthetically useful biocatalysed transformations

depend on redox cofactors.¹⁰One of the most common cofac-

tors, NADH, acts as a hydride source for enantio- and chemo-

selective reductions, but its stoichiometric use is prohibitively

expensive when reactions are carried out on a manufacturing

^aDepartment of Chemistry, University of Oxford, Inorganic Chemistry Laboratory,

South Parks Road, Oxford, OX1 3QR, UK. E-mail: holly.reeve@chem.ox.ac.uk; kylie.

vincent@chem.ox.ac.uk

Wang and co-workers have reported platinum nanoparticles

(NPs) supported on carbon or metal oxides for H₂-driven NAD⁺

reduction (Fig. 1a).^22,23Elevated temperatures and pressures

were used to overcome the activation energy for Pt-catalysed

NAD⁺reduction, but side-products resulted (e.g. from over-

reduction of the pyridine ring) in addition to the desired 1,4-

^bDepartment of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK.

E-mail: nicole.grobert@materials.ox.ac.uk

† Electronic supplementary information (ESI) available. See DOI:

10.1039/d1sc00295c

‡ These authors contributed equally.

Chem. Sci., 2021, 12, 8105–8114 | 8105

View Article Online

Chemical Science

Edge Article

Fig. 1 Methods for H₂-driven NADH generation using carbon-supported metal nanoparticles. (a) Both H₂oxidation and NAD⁺reduction must

occur at metal “chemo/C⁰⁰site. (b) Metal-catalysed H₂oxidation is coupled to a co-immobilised NAD⁺reductase for “chemo-bio/C⁰⁰catalysed

NADH generation. (c)(i) Mixed carbon agglomerate “chemo/C+bio/C⁰⁰catalyst involves mixing separate carbon-supported metal and carbon-

supported NAD⁺reductase, which then agglomerate such that electron transfer is still possible between the site-separated catalysts (see later).

(c)(ii) H₂-driven cofactor generation can extend to NADH-dependent biocatalysis (e.g. asymmetric ketone reduction by an alcohol dehydro-

genase, “ADH”).

NADH. Under milder conditions used to supply an alcohol exclusively, the biologically active 1,4-NADH (2). Additionally,

dehydrogenase with 1,4-NADH, only seven cofactor turnovers a mixed carbon agglomerate “chemo/C+bio/C⁰⁰catalyst is

occurred aer 100 h of reaction time, suggesting that further developed in which the metal and enzyme are each immobilised

catalyst development would be benecial.²²

on separate carbons (Fig. 1c(i)). When mixed together, the

Carbon materials (typically graphite, carbon black and acti- agglomerated carbons should still allow electron transfer from

vated carbon) are widely used as supports for metal NPs,^24–26and metal to enzyme via the carbon (see later). Each catalyst is

have emerged as versatile supports for enzymes.^27–29These compared for activity and selectivity for producing biologically

carbon supports are cheap, available in a range of forms, and active 1,4-NADH (2) using a range of characterisation tech-

oﬀer pH robustness, electron transfer, and high surface area niques, and the diﬀerent commercial carbon materials are

compared with other supports for metal and enzyme immobi- analysed and compared. Both the chemo-bio/C and the chemo/

lisation (e.g. silica or alumina).³⁰We therefore wondered if C+bio/C catalyst types are then paired with an NADH-dependent

carbon particles could act as a support for hybridising chemo- alcohol dehydrogenase to promote asymmetric ketone reduc-

and biocatalysis onto a single heterogeneous support.

tions with high selectivity and turnover numbers (Fig. 1c(ii)).

Herein, we investigate a range of catalyst systems for H₂-

driven NAD⁺reduction (Fig. 1). We rst evaluate commercial

metal on carbon “chemo/C” catalysts, then these catalysts are

also used in the preparation of “chemo-bio/C⁰⁰catalysts through

simple adsorption of an NAD⁺reducing enzyme onto the

chemo/C (Fig. 1b). We have previously shown that a conductive

carbon support transfers electrons from a H₂-oxidising enzyme

to a co-immobilised NAD⁺reductase,^15,31so we hypothesised

that the same should be possible from a metal nanoparticle

capable of H₂oxidation to the co-immobilised NAD⁺reductase

enzyme. The NAD⁺reductase contains a avin mononucleotide

active site which takes up two electrons via an electron-relay

chain of iron-sulphur clusters in the protein. A proton from

solution then combines with the two electrons at the avin, and

the resulting hydride is transferred to NAD⁺to selectively form,

2. Results and discussion

2.1 Screening of chemo/C catalysts for H₂-driven NADH

generation

Five commercial metals (rhodium, iridium, ruthenium, plat-

inum, and palladium) immobilised on activated carbon and

carbon black with low metal loading (1–5 wt%) were evaluated

for NADH production under the reaction protocol described in

S1.5.1 (see ESI for method and Table S1† for details on the

carbon material type of each catalyst). The gures of merit used

to evaluate chemo/C catalysts were activity and selectivity for

generating bioactive NADH, while avoiding metal-catalysed

cofactor degradation over time.³²The reaction course was

monitored by taking aliquots for analysis at 30 min, 2 h, and

8106 | Chem. Sci., 2021, 12, 8105–8114

View Article Online

Edge Article

Chemical Science

Fig. 2 Cofactor compositions after H₂-driven NAD⁺reduction by diﬀerent chemo/C catalysts. a. Values were calculated using a combination of

1

quantitative H NMR spectroscopy and UV-visible spectroscopy at three time points (t₁¼ 30 min, t₂¼ 2 h, t₃¼ 17 h). At 17 h, 5–8% relative

standard deviation was determined from reactions performed in triplicate using 3 wt% Pd/C (Fig. S1, ESI†). Reaction conditions: 2 mL of 5 mM

NAD⁺in Tris–HCl (50 mM, pH 8.0) was stirred with 0.32 mg commercial chemo/C under a steady ﬂow of H₂(1 bar) at room temperature. “Blank”

experiments were run as a control in the absence of catalyst, and our previous report showed that carbon black (BP2000) does not catalyse H₂-

driven NAD⁺reduction.¹⁴Further details in ESI (S1.5.1†). (b) Simpliﬁed structures of biologically active and inactive nicotinamide cofactors.

17 h time points (t₁, t₂, and t₃, Fig. 2a). Cofactor composition of other categories. “Missing” cofactor could be due to adsorption

unreacted NAD⁺(1, light grey), biologically active 1,4-NADH (2, onto the carbon material. This adsorption was suggested by

green) and undesired 1,6-NADH (3, black)³³was quantied a control experiment in which a decrease in concentration of 2

1

using H NMR spectroscopy. Simplied cofactor structures are was quantied when stirred with the catalysts under N₂

shown in Fig. 2b. UV-vis spectroscopy and HPLC were used to (Fig. S4†).

verify product composition (ESI Fig. S1–S3 provide analytical

Under these reaction conditions, Rh/C (5 wt%) showed high

details†). The “other” category (dark grey) encompasses cofactor conversion but poor selectivity, while Ir/C (1 wt%) showed

forms that are not easily quantied by NMR spectroscopy (e.g. selectivity for 2, but low activity. Ru/C (5 wt%) was inactive

4,²²5, Fig. S3†) as well as cofactor missing from the sum of all toward NAD⁺reduction. Pt/C (5 wt%) exhibited reasonable

Chem. Sci., 2021, 12, 8105–8114 | 8107

View Article Online

Chemical Science

Edge Article

conversion rates and fairly high selectivity for 2, while Pd/C supports, then tested if they could be coupled with an immo-

(3 wt%) did not produce any detectable amount of the desired bilised NAD⁺reductase where electrons from Pd-catalysed H₂-

product, 2, at any time point.

oxidation could be supplied to the NAD⁺reductase for selective,

To determine if metal wt% inuences NAD⁺reduction enzymatic reduction of 1.

selectivity, we chose Pt/C and Pd/C (0.5–60 wt%) for further

investigation based on the commercial availability of these

metals at a range of loadings. The majority of the Pt/C and Pd/C

catalysts were supplied in powder form, however 20 and 60 wt%

Pt/C and 0.5 wt% Pd/C were supplied as mm-cm carbon parti-

cles. In order to achieve reactions with comparable catalyst

handling, the large particles were ground to a paste in water

2.2 Characterisation of selected chemo/C materials

The commercial 0.5–3% Pd/C were composed of diﬀering

carbon types (carbon black or activated carbon, see Table S1; †

diﬀering morphologies are evident in SEM images, Fig. 3 and X-

ray powder diﬀraction patterns, Fig. S6†). We sought to under-

stand the variability of the carbon support materials in more

detail. In particular, we studied the carbon properties that

might impact enzyme adsorption (degree of graphitisation,

surface area, pore diameter, uniformity, elemental surface

composition)^34–36using a variety of techniques as a way to

evaluate the chemo/C catalysts for their potential use in chemo-

bio/C catalysts. A metal-free, highly porous carbon black

nanopowder (BP2000), which has previously been used as

a carbon support for biocatalytic H₂-driven NADH genera-

tion,^14,16was used for comparison.

All of the Pd/C carbons and BP2000 showed low degrees of

graphitisation by Raman spectroscopy (Fig. S7†). Nitrogen

adsorption analysis showed BP2000 has over 50% more surface

area compared with the other carbon supports ($696 m²per g

according to Langmuir model) and the largest average pore size

(5.6 nm, Table S4†). SEM images of the carbon samples show

that BP2000 has the most uniform morphology (Fig. 3a)

whereas the Pd/C samples had some small particles, large

agglomerates and platelets (Fig. 3b–d). Some or all of these

properties likely improve enzyme adsorption: Bradford assays

reveal only 3% enzyme le in the supernatant aer NAD⁺

reductase was immobilised onto BP2000. This value was $20-

fold lower than the Pd/C (Table S5†), demonstrating improved

adsorption on BP2000. Table S5† summarizes EDX elemental

ꢁ

using a mortar and pestle, then le in a 60 C oven overnight

before use. Reactions were again set up following the procedure

described in S1.5.1,† and the same analytical techniques were

used for evaluation of cofactor composition.

When 5–60 wt% Pt/C were used, some preference for 2 was

observed in the rst 2 h, however other cofactor forms were

detected aer 17 h (Fig. 2a). This observation suggests that

onward reaction of 2 with the metal catalysts can lead to by-

products, conrmed by control experiments in which chemo/

C catalysts were stirred in a solution of 2 under H₂resulting

in production of further products (e.g. 3, Fig. S5†). Only trace

conversion was observed when 1 wt% Pt/C was used. For

comparison, Wang et al. reported that 1 wt% Pt/C can reduce 1

under more forcing reaction conditions (10 bar H₂, 37 ^ꢁC, pH 7),

and that this provided a mixture of unknown side products in

addition to 2 and 3.²³

At no loading did the Pd/C catalysts produce a detectable

quantity of 2. The 3 wt% and 10 wt% Pd/C predominately

produced inactive cofactor at every time point, which precludes

these catalysts from use in cofactor recycling. Low metal load-

ings of Pd (0.5–1 wt%) did not appear to act on 1 and did not

generate any inactive cofactor forms. We therefore took these

low Pd content catalysts for characterisation of the carbon

Fig. 3 SEM micrographs showing morphology of carbon supports. (a) BP2000 has uniform morphology with particle sizes in the nm range. (b–d)

Pd/C have small particles as well as large agglomerates and micron size platelets. (e–f) The mix of {BP2000 + 1 wt% Pd/C} shows large

agglomerates and platelets (likely Pd/C) that appear to be coated with small spherical particles (likely BP2000, see additional image in Fig. S9,

ESI†).

8108 | Chem. Sci., 2021, 12, 8105–8114

View Article Online

Edge Article

Chemical Science

analysis of the material surfaces which did not reveal any (Fig. 4a). The formation of 3 as by-product could be from Pd

correlation between elemental composition and protein reacting with 1 or enzyme-produced 2. For the 0.5% Pd chemo-

adsorption, although tuning of surface chemistry through bio/C, selectivity for 2 was observed. In further experiments, the

modication of the support is something that could be explored NAD⁺reductase loading was increased to achieve r_m¼ 1,

in the future.

leading to selective generation of the desired product, 2, with

both 0.5 and 1% Pd chemo-bio/C. In these experiments, no

inactive cofactor was detected, although there is some ‘missing’

cofactor, which could be due to adsorption onto the carbon

support.

2.3 Co-immobilised Pd and NAD⁺reductase “chemo-bio/C”

catalysts

Commercially available Pd/C catalysts (0.5–3 wt%) were selected

to act as both the “chemo” H₂-oxidation site and carbon support

for the preparation of chemo-bio/C catalysts for generating

NADH, and the “bio” catalysts were a co-immobilised NAD⁺

reductase. Palladium was chosen over platinum due to wider

commercial range of <5 wt% metal loadings. The chemo-bio/C

catalysts were straightforwardly prepared on the benchtop by

incubating NAD⁺reductase (various loadings, from R. eutropha,

170 kDa)^14,37with a slurry of Pd/C (100 mg in 5 mL Tris–HCl

buﬀer) under N₂over ice in a stirred round bottom ask for 1 h.

Control experiments conrmed that NAD⁺reductase has

negligible activity for NAD⁺reduction under H₂without the

presence of metal (Fig. S10†), and we have previously conrmed

that this enzyme exclusively produces 2 in an NMR spectroscopy

study.¹⁶

Noting that the Pd/C catalysts alone did not produce

detectable quantities of 2 at any metal loading (Fig. 2a), the fact

that 2 is observed as a major product in the reactions in which

the NAD⁺reductase is co-immobilised on Pd/C materials to

form the chemo-bio/C catalysts indicates that the NAD⁺reduc-

tase must be responsible for the NAD⁺reduction here. Since the

NAD⁺reductase cannot oxidise H₂itself, these results are

indicative of electron transfer through the carbon from H₂

oxidation at Pd to the co-immobilised NAD⁺reductase enzyme.

We examined more closely the impact of r_mon conversion to

2. Diﬀerent masses of NAD⁺reductase were incubated with

a slurry of 0.5 wt% Pd/C, then the chemo-bio/C catalysts were

assayed for NAD⁺reduction activity by in situ UV-visible spec-

troscopy following the method described in S1.5.2.† Conversion

and enzyme turnover frequency (TOF, mol 2 per mol NAD⁺

reductase per hour) were compared (Fig. 4b and Table 1).

Palladium TOF (mol 2 per mol Pd per hour) was also calculated

under the assumption that each equivalent of NADH generated

requires one equivalent of H₂to be oxidised by Pd. When entry 4

was repeated in triplicate, a relative standard deviation of 7%

was calculated for metal and enzyme TOF, showing good

reproducibility.

The mass ratio of enzyme and carbon is dened as

mass_enzyme

mass_Pd=C

r_m

¼

, which was initially set to 0.1. Chemo-bio/C

catalysts were evaluated for NADH generation activity and

selectivity using ¹H NMR spectroscopy following the procedure

described in S1.5.3† (Fig. 4a).

The 3 wt% Pd chemo-bio/C catalyst produced some biolog-

ically active reduced cofactor, 2, however signicant formation

of side products highlights the activity of the metal at this

loading toward formation of inactive cofactor forms, precluding

use of this chemo-bio/C catalyst for cofactor recycling.

Whereas the 1% Pd/C chemo/C catalyst alone showed no

activity for reduction of 1 (Fig. 2a), on incorporation of NAD⁺

reductase the chemo-bio/C catalyst showed formation of 2

Using a standard 0.1 mM 1, palladium activity improved with

increased enzyme loading (entries 1–5, Table 1) with the highest

Pd TOF (113 h^ꢀ1) calculated at the highest NAD⁺reductase

loading (entry 5). These activities exceed other reported metal-

catalysed H₂-driven^20,22and formate-driven NADH-generation

under ambient conditions by 3–45-fold.^38–40The r_m¼ 0.4

Fig. 4 Reaction progress of chemo-bio/C catalysts using diﬀerent Pd and enzyme loadings. (a) Cofactor compositions after 17 h determined by

¹H NMR spectroscopy. Reaction conditions: 1 mM NAD⁺stirred with chemo-bio/C catalysts formed from 100 mg Pd/C and the corresponding

mass of enzyme following procedure from S1.5.3 (see detail in ESI†). (b) Conversion to 2 over 60 min monitored by absorbance at 340 nm using in

situ UV-visible spectroscopy. Reaction conditions (see further details in S1.5.2†): 0.1 mM NAD⁺mixed with chemo-bio/C catalysts formed from

20 mg 0.5 wt% Pd/C and enzyme loadings ranging from 2–40 mg (giving an r_mrange of 0.1–2).

Chem. Sci., 2021, 12, 8105–8114 | 8109

View Article Online

Chemical Science

Edge Article

Table 1 Catalyst activities in H₂-driven NAD⁺reduction at varying quantities of NAD⁺reductase (immobilised on 0.5 wt% Pd/C, 20 mg) and

varying [1]^a

Entry

r_m

NAD⁺reductase loading (mg)

[1] (mM)

NAD⁺reductase TOF^b(h^ꢀ1

)

Pd TOF (h^ꢀ1

)

1

2

3

4

0.1

0.2

0.4

1

2

1

2

4

8

20

40

20

40

0.1

0.5

5

530

785

979

520

449

877

n.d.^d

143

316

6.6

20

49

65

5

113

218

n.d.^d

17

6^c

7^c

8^e

9^e

0.1

78

a

Reaction conditions follow S1.5.2 with chemo-bio/C catalysts in entries 1–5 formed from 20 mg 0.5 wt% Pd/C and the indicated enzyme loadings.

When entry 4 was repeated in triplicate, a relative standard deviation of 7% was calculated for the TOFs. ^bActivity was measured during the linear

phase of the reaction. ^cEntries 6 and 7 followed S1.5.3 in a stirred round bottomed ask. ^dNot determined. ^eEntries 8–9 were carried out following

ꢀ

ꢁ

mass_enzyme

mass_BP2000þ mass_Pd=C

S1.5.2 inside a glovebox and the reaction was mixed using a magnetic stirrer bead using chemo/C+bio/C catalysts r_m

¼

formed from 10 mg BP2000, 10 mg 1 wt% Pd/C and the indicated enzyme loadings.

ꢀ

ꢁ

catalyst provided the highest NAD⁺reductase activity (979 h^ꢀ1

;

mass_enzyme

mass_BP2000þ mass_Pd=C

carbon r_m

¼

was set to 1 and 2. In

96 nmol min^ꢀ1mg^ꢀ1, entry 3). The catalysts comprising higher

r_mhad lower enzyme activities, possibly due to limitations of

enzyme adsorption (Table S6†), or because the chemo-bio/C

catalyst becomes H₂oxidation rate-limited.

situ UV-visible spectroscopy was used to calculate activity

following the procedure described in S.1.5.2.† Product compo-

sition was determined using ¹H NMR spectroscopy in a separate

experiment where 1 mM NAD⁺was used following the method

The use of 0.5 mM 1 (entry 6) led to higher enzyme activity

(877 h^ꢀ1; 86 nmol min^ꢀ1mg^ꢀ1) and Pd TOF (218 h^ꢀ1). At 5 mM

1 the reaction conversion plateaued and only 6% conversion

was reached aer 5 h (entry 7). The Pd was seemingly inhibited

by the high cofactor concentration.^20,41A control experiment

(Fig. S11†) supports this hypothesis: when the chemo-bio/C was

stirred in a solution of 4 mM 1 under H₂which was then diluted

to 0.5 mM using Tris–HCl buﬀer, the catalyst activity also pla-

teaued rather than duplicate the 0.5 mM conversion rate. In

a cofactor recycling application, lower NAD⁺loadings would be

used in order to keep reaction cost and waste low, hence this

limitation was not considered a setback.

described in S.1.5.3, which conrmed selectivity for

(Fig. S12†).

2

Compared with the co-immobilised chemo-bio/C catalysts,

the chemo/C+bio/C had lower activities per mg enzyme (in

Table 1, compare entries 4 and 8; 5 and 9). The best Pd TOF

(78.0 h^ꢀ1, r_m¼ 2) was lower than the TOF achieved using the

corresponding chemo-bio/C catalyst (113 h^ꢀ1). One explanation

could be due to fouling of the palladium by BP2000, thus less

metal exposure for H₂oxidation. In an alternative preparation

method, the BP2000 and Pd/C were mixed prior to enzyme

adsorption. When r_m¼ 1, there was a longer lag time for

activity, aer which the NAD⁺reductase was comparable

(163 h^ꢀ1; 16 nmol min^ꢀ1mg^ꢀ1, Fig. S13†).

2.4 Mixed carbon agglomerate “chemo/C+bio/C” catalyst

design

The observation of activity with the chemo/C+bio/C catalysts

supports our hypothesis that there is good physical contact

In previous studies, we have observed excellent adsorption of between the two carbon types, facilitating electron transfer

enzymes onto BP2000.²⁹This inspired us to develop a mixed between metal and enzyme. Furthermore, the use of BP2000 for

carbon agglomerate chemo/C+bio/C catalyst composed of NAD⁺the mixed carbon platform in principle enables chemo-bio

reductase on BP2000 that was then mixed with Pd/C, with elec- catalysts to be prepared from chemo/C catalysts that do not

tron transfer expected to occur via the carbon agglomerates themselves eﬃciently adsorb enzymes. It also permits the user

(Fig. 1c(i)). This strategy also enabled us to test if keeping the to adjust metal wt% in order to tune reaction selectivity. Finally,

enzyme and metal spatially separated alleviated any mutual although activity was not improved here, the ability to physically

inhibition. SEM images showed that the small, spherical BP2000 separate the enzyme and metal NP may circumvent mutual

(Fig. 3a) coats the larger Pd/C platelets and agglomerates (Fig. 3c) inactivation in catalysts for which this is an issue.

when the two are mixed as a slurry (Fig. 3e, f and S9†).

Aer mixing one part {1 wt% Pd/C} with one part {BP2000

2.5 Asymmetric ketone reductions by chemo-bio H₂-driven

cofactor recycling

modied with NAD⁺reductase}, we conrmed that the surface

palladium composition corresponded with that of commercial

0.5 wt% Pd/C using SEM/EDX (1.4% and 1.5%, respectively; The chemo-bio catalysts were used to continually supply 2 to an

Table S5†). This enabled us to directly compare the chemo/ (S)-selective alcohol dehydrogenase (ADH-105, Johnson Mat-

C+bio/C catalyst activities to the equivalent 0.5 wt% chemo-bio/ they) for acetophenone (6) reduction (Fig. 1c(ii)). Reactions were

C catalysts (entries 8–9, Table 1). The mass ratio of enzyme to carried out following the procedure described in S.1.5.4† and

8110 | Chem. Sci., 2021, 12, 8105–8114

View Article Online

Edge Article

Chemical Science

analysed for conversion and enantiomeric excess (>99% ee in all merit (summarised in Table 3). When 4 bar H₂was used with

cases) by chiral phase GC-FID at 2 h and 24 h (see Fig. S14†). chemo/C+bio/C, the highest Pd turnover frequency (TOF, 441

Pressurised reactions (4 bar H₂) were performed in a pressure h^ꢀ1) and NAD⁺reductase TOF (2347 h^ꢀ1, not tabulated) were

vessel, and atmospheric H₂(1 bar) was achieved by a steady ow achieved (entry 2), which could be a result of improved H₂

of gas over the reaction headspace.

availability. This NAD⁺reductase TOF is on the same order of

Initial experiments were designed to optimise parameters for magnitude as that achieved when a hydrogenase was used in

high conversion to 7 (Table 2). Chemo-bio/C catalysts formed place of palladium under ambient pressure (3168 h^ꢀ1),¹⁴which

from 1 wt% Pd/C promoted higher conversions compared with suggests that Pd does not signicantly inhibit NAD⁺reductase

those formed from 0.5 wt% Pd/C (entries 1–4). Entry 6 shows activity.

that a comparable conversion was obtained when the analogous

High NAD⁺turnover numbers (TN, mol 7 per mol NAD⁺) up

chemo/C+bio/C catalyst was used (compare to entry 3). We to 384 (entry 3) indicated that the Pd did not appreciably

observed enhanced dispersion of the heterogeneous chemo/ decompose the cofactor. This improves upon existing chemo-

C+bio/C catalyst, which remained as a slurry with agitation, bio NAD(P)⁺TN by >7-fold,^{18,19,22,43–46}and further modications

compared with the chemo-bio/C which settled to the bottom of in substrate concentration may provide turnover numbers

the reaction tube. To avoid limitations due to mixing, we used suitable for commercial application (>1000).⁴⁷The NAD⁺

the chemo/C+bio/C to further explore reaction parameters reductase total turnover numbers (TTN, mol 7 per mol NAD⁺

(entries 6–10). Following this optimisation, we returned to the reductase) up to 14 000 indicated that the immobilised enzyme

chemo-bio/C using the optimal conditions with success (97% was stable to the reaction conditions. High TTN of Pd (up to

conversion of 20 mM 6 under ambient H₂, entry 5).

2,960, entry 3) suggests it was not signicantly impacted by

A control experiment in which 1 wt% Pd/C was used under 4 cofactor poisoning nor enzyme inhibition, both of which are

bar H₂resulted in 3% conversion to racemic 7 (entry 11), which common compatibility issues in homogeneous chemo-bio

aligns with prior reported direct hydrogenation of 6 to racemic 7 cofactor recycling.^20,38High cofactor and Pd turnovers (220

by 10 wt% Pd/C in water.⁴²Since (R)-7 was never detected when and 3,390, respectively) were also achieved during a reaction in

our chemo-bio catalysts were used with ADH, any background which the organic components were extracted into heptane

non-enzymatic reduction of 6 directly by Pd is negligible.

aer 3 h, then fresh 6 in DMSO was added to the reaction and

While we were pleased that both types of chemo-bio catalysts allowed to mix for another 21 h (entry 4).

could be used with ADH-105 to provide high conversions of 6,

4⁰-Chloroacetophenone (8) can undergo Pd-catalysed hydro-

the chemo/C+bio/C catalyst tended give better catalyst gures of dehalogenation to form 6 or 7 (entry 1, Table 3) or can be

Table 2 Optimisation of H₂-driven chemo-bio acetophenone reduction^a

Chemo-bio catalyst

[NAD⁺]

(mM)

[6]

(mM)

Conversion

(%)

Entry

P_H(bar)

Type

Pd wt%

mg mL^ꢀ1

ADH (mg mL^ꢀ1

)

2

1

2

3

4

5

6

7

8

4

1

4

1

4

0.5

0.1

0.5

0.1

0.05

0.1

5

20

5

20

5

Chemo-bio/C, r_m¼ 1

Chemo/C+bio/C, r_m¼ 1

Chemo/C+bio/C, r_m¼ 2

Unmodied Pd/C

0.5

1

0.3

0.8

0.2

0.4

0.8

0.1

0.2

0.4

0.2

0.4

n/a

19

2.4

41

15

97

46

>99

93

5

20

9

10

11

98

96

1

3 (racemic)

a

Reactions were mixed for 24 h in H₂-saturated buﬀer solution (50 mM Tris–HCl, pH 8.0) that contained NAD⁺, 6, and 1 vol% DMSO following

S1.5.4. Conversions were calculated aer extracting reaction mixture into EtOAc using chiral-phase GC-FID. Entries 1–4, 6–8 and 11 : 1 mL

reaction volume mixed in a pressure vessel that was oscillated at 26 rpm. Entries 5, 9–10 : 0.5 mL reaction volume mixed under a ow of H₂by

shaking at 500 rpm. A standard deviation of 1% conversion was calculated when entry 8 was performed in triplicate. Rows in bold represent

experiments also highlighted in Table 3.

Chem. Sci., 2021, 12, 8105–8114 | 8111

View Article Online

Chemical Science

Edge Article

Table 3 Examples showcasing diﬀerent ﬁgures of merit for chemo-bio acetophenone reduction^a

Chemo-bio catalyst

Entry

P_H(bar)

Type

mg mL^ꢀ1

Conversion^b(%)

Pd TOF^c(h^ꢀ1

)

NAD⁺TN^b

TTN Pd^b

TTN NAD⁺reductase^b

2

1

4

1

Chemo-bio/C, r_m¼ 1

Chemo/C+bio/C, r_m¼ 1

Chemo/C+bio/C, r_m¼ 2

0.8

0.4

0.8

97

93

96

55

38

194

186

384

220

780

8400

14 000

8300

9500

2

441

136

215

2640

2960

3390

3^d

4^e

a

Reactions were mixed for 24 h in H₂-saturated buﬀer solution (50 mM Tris–HCl, pH 8.0) that contained 0.1 mM NAD⁺, 20 mM 6, 0.4 mg mL^ꢀ1ADH-

105, and 1 vol% DMSO (see S1.5.4 for more detail). Entries 1 and 3–4: 0.5 mL reaction volume mixed under a ow of H₂by shaking at 500 rpm. Entry

2 : 1 mL reaction volume mixed in a pressure vessel that was oscillated at 26 rpm. A relative standard deviation of 9% was determined for Pd TOF (at

b

2 h) and 1% for all values calculated at 24 h (conversion, TN, TTNs) when entry 2 was performed in triplicate. Calculated aer 24 h of reacting.

c

Mol 7 per mol Pd per h, calculated aer 2 h of reacting. ^d0.05 mM NAD⁺was used. ^eEntry 4: shaken under typical conditions for 3 h, then any 6

and 7 were extracted from the reaction by addition and removal of a heptane layer, then a fresh aliquot of 6 was added such that [6] ¼ 20 mM and

DMSO ¼ 1 vol%.

Table 4 Chemo-bio catalysed reduction of 8^a

Reaction volume

(mL)

Conversion of

8 (%)

Product ratio

9 : 6 : 7

Entry

P_H

Catalyst type, loading (mg mL^ꢀ1

)

2

1

2

3

4

5

1

0.5

1 bar

4 bar

1 bar

1 wt% Pd/C, 0.1

13

31

14

81

91

0 : 1 : 0

37 : 5 : 1

1 : 0 : 0

Chemo-bio/C, 0.4

Chemo/C+bio/C, 0.4

Chemo-bio/C, 0.8

Chemo/C+bio/C, 0.8

a

Reaction followed S1.5.4 using 0.1 mM NAD⁺, 5 mM 8, and 0.4 mg mL^ꢀ1ADH-105. Reactions performed at 4 bar were placed in a pressure vessel

that was charged to 4 bar H₂then oscillated at 30 rpm. Reactions performed at 1 bar were placed in a shaker plate under a steady stream of H₂and

shaken at 500 rpm.

reduced by ADH-105 to give (S)-9 (entries 2–5, Table 4; see Demonstrated here for NAD⁺recycling, each component is

Fig. S15† for GC-FID spectra). Under conditions described in simultaneously active and selective for its desired reaction. We

S1.5.4 (using 5 mM 8), the chemo-bio/C catalyst had poor che- show that the activity of metal and enzyme catalysts can be

moselectivity when 4 bar H₂was used (entry 2, Table 4), but the tuned by varying the ratio of catalysts for two half reactions,

chemo/C+bio/C catalyst provided (S)-9, exclusively (>99% ee, with non-desirable metal activity prevented at low metal/high

entry 3). Under ambient conditions and increased chemo-bio enzyme loadings and mild conditions. Highly selective hydro-

catalyst loading (see Table S7† for optimisation), chemo-bio/C genation reactions were achieved with low waste production

and chemo/C+bio/C catalysts each gave (S)-9 as the only relative to product mass (E factor: 2.4–3.0, see eqn S1 in ESI† for

product in 81% and 91% conversions, respectively (>99% ee, details).⁴⁸The methods demonstrated here have the potential to

entries 4–5). This method is therefore tolerant of labile groups be expanded for combining other metal and enzyme reaction

when used under ambient conditions, circumventing the steps in synthetically useful ways.

chemo- and enantioselectivity issues encountered when

unmodied Pd/C was used.

Author contributions

X. Z., S. E. C., C. Z., N. G., H. A. R., and K. A. V. all helped to

3. Conclusions

design experiments. X. Z., S. E. C., and C. Z. conducted experi-

The high metal and enzyme TOFs and TTNs achieved in this ments. X. Z., S. E. C., and H. A. R. analysed data relating to

work highlight the use of carbon supports as a general meth- catalyst activity and selectivity. C. Z. and N. G. analysed data

odology for designing dual-active chemo-bio catalysts. relating to catalyst characterisation. All authors discussed the

8112 | Chem. Sci., 2021, 12, 8105–8114

View Article Online

Edge Article

Chemical Science

results and commented on the manuscript. N. G., H. A. R., and 16 J. S. Rowbotham, M. A. Ramirez, O. Lenz, H. A. Reeve and

K. A. V. supervised the project.

K. A. Vincent, Nat. Commun., 2020, 11, 1454.

17 L. A. Thompson, J. S. Rowbotham, J. H. Nicholson,

M. A. Ramirez, C. Zor, H. A. Reeve, N. Grobert and

K. A. Vincent, ChemCatChem, 2020, 12, 3913–3918.

18 O. Abril and G. M. Whitesides, J. Am. Chem. Soc., 1982, 104,

1552–1554.

Conﬂicts of interest

There are no conicts to declare.

19 S. Bhaduri, P. Mathur, P. Payra and K. Sharma, J. Am. Chem.

Soc., 1998, 120, 12127–12128.

20 Y. Maenaka, T. Suenobu and S. Fukuzumi, J. Am. Chem. Soc.,

2012, 134, 367–374.

Acknowledgements

Research by S. E. C., H. A. R. and K. A. V. was supported

nancially by an award from the Engineering and Physical

Sciences Research Council (EPSRC, IB Catalyst EP/N013514/

1). N. G. would like to thank the Royal Society. We are grateful to

Dr Miguel A. Ramirez (University of Oxford) and Dr Oliver Lenz

(Technical University, Berlin) for providing NAD⁺reductase and

Dr Beatriz Dominguez (Johnson Matthey) for supplying ADH-

105. BET measurements were performed by Colin Johnston

(Oxford Materials Characterisation Service). Thanks to Dr Jack

S. Rowbotham (University of Oxford) for helpful discussions

about the chemo-bio catalyst.

¨

21 V. Kohler, Y. M. Wilson, M. Durrenberger, D. Ghislieri,

¨

E. Churakova, T. Quinto, L. Knorr, D. Haussinger,

F. Hollmann, N. J. Turner and T. R. Ward, Nat. Chem.,

2013, 5, 93–99.

22 X. Wang and H. H. P. Yiu, ACS Catal., 2016, 6, 1880–1886.

23 T. Saba, J. Li, J. W. H. Burnett, R. F. Howe,

P. N. Kechagiopoulos and X. Wang, ACS Catal., 2021, 11,

283–289.

24 E. Auer, A. Freund, J. Pietsch and T. Tacke, Appl. Catal., A,

1998, 173, 259–271.

´

25 E. Perez-Mayoral, V. Calvino-Casilda and E. Soriano, Catal.

Sci. Technol., 2014, 6, 1265–1291.

26 U. Petek, F. Ruiz-Zepeda, M. Bele and M. Gaberscek,

References

ˇˇ

¨

1 M. Honig, P. Sondermann, N. J. Turner and E. M. Carreira,

Catalysts, 2019, 9, 134.

Angew. Chem., Int. Ed., 2017, 56, 8942–8973.

2 C. A. Denard, H. Huang, M. J. Bartlett, L. Lu, Y. Tan, H. Zhao

27 L. Botta, B. M. Bizzarri, M. Crucianelli and R. Saladino, J.

Mater. Chem. B, 2017, 5, 6490–6510.

and J. F. Hartwig, Angew. Chem., Int. Ed., 2014, 53, 465–469. 28 M. Adeel, M. Bilal, T. Rasheed, A. Sharma and H. M. N. Iqbal,

3 T. J. Doyon and A. R. H. Narayan, Synlett, 2020, 31, 230–236.

Int. J. Biol. Macromol., 2018, 120, 1430–1440.

4 F. Dumeignil, M. Guehl, A. Alexandra Gimbernat, M. Capron, 29 L. A. Thompson, J. S. Rowbotham, H. A. Reeve, C. Zor,

N. Lopes Ferreira, R. Froidevaux, J.-S. Girardon,

R. Wojcieszak, P. Dhulster and D. Delcroix, Catal. Sci.

Technol., 2018, 8, 5708–5734.

N. Grobert and K. A. Vincent, in Methods in Enzymology,

ed. Challa V. Kumar, Academic Press, 2020, vol. 630, pp.

303–325.

5 K. J. Kilpin and P. J. Dyson, Chem. Sci., 2013, 4, 1410–1419.

30 E. Lam and J. H. T. Luong, ACS Catal., 2014, 4, 3393–3410.

6 M. Poizat, I. W. C. E. Arends and F. Hollmann, J. Mol. Catal. 31 H. A. Reeve, P. A. Ash, H. Park, A. Huang, M. Posidias,

B: Enzym., 2010, 63, 149–156.

C. Tomlinson, O. Lenz and K. A. Vincent, Biochem. J., 2017,

474, 215–230.

¨

7 J. Pauly, H. Groger and A. V. Patel, ChemCatChem, 2019, 11,

1503–1509.

32 T. Saba, J. W. H. Burnett, J. Li, P. N. Kechagiopoulos and

X. Wang, Chem. Commun., 2020, 56, 1231–1234.

8 M. Cortes-Clerget, N. Akporji, J. Zhou, F. Gao, P. Guo,

´

M. Parmentier, F. Gallou, J.-Y. Berthon and B. H. Lipshutz, 33 F. Martınez-Pen

Nat. Commun., 2019, 10, 2169. A. M. Pizarro, Inorg. Chem., 2018, 57, 5657–5668.

9 F. Rudroﬀ, M. D. Mihovilovic, H. Groger, R. Snajdrova, 34 M. Mahmoudi, I. Lynch, M. Reza Ejtehadi, M. P. Monopoli,

̃a, S. Infante-Tadeo, A. Habtemariam and

¨

H. Iding and U. T. Bornscheuer, Nat. Catal., 2018, 1, 12–22.

F. Baldelli Bombelli and S. Laurent, Chem. Rev., 2011, 111,

5610–5637.

´

10 L. Selles Vidal, C. L. Kelly, P. M. Mordaka and J. T. Heap,

¨

Biochim. Biophys. Acta, Proteins Proteomics, 2018, 1866, 327– 35 A. E. Nel, L. Madler, D. Velegol, T. Xia, E. M. V. Hoek,

347.

P. Somasundaran, F. Klaessig, V. Castranova and

M. Thompson, Nat. Mater., 2009, 8, 543–557.

36 J. Quinson, R. Hidalgo, P. A. Ash, F. Dillon, N. Grobert and

K. A. Vincent, Faraday Discuss., 2014, 172, 473–496.

11 J. P. Adams, M. J. B. Brown, A. Diaz-Rodriguez, R. C. Lloyd

and G. Roiban, Adv. Synth. Catal., 2019, 361, 2421–2432.

12 T. H. Lonsdale, L. Lauterbach, S. Honda Malca, B. M. Nestl,

B. Hauer and O. Lenz, Chem. Commun., 2015, 51, 16173– 37 T. Burgdorf, A. L. De Lacey and B. Friedrich, J. Bacteriol.,

16175.

2002, 184, 6280–6288.

13 L. Lauterbach, O. Lenz and K. A. Vincent, FEBS J., 2013, 280, 38 J. Canivet, G. Suss-Fink and P. Stepnicka, Eur. J. Inorg. Chem.,

3058–3068.

2007, 2007, 4736–4742.

14 H. A. Reeve, L. Lauterbach, O. Lenz and K. A. Vincent, 39 R. Ruppert, S. Herrmann and E. Steckhan, J. Chem. Soc.,

ChemCatChem, 2015, 7, 3480–3487. Chem. Commun., 1988, 0, 1150–1151.

15 H. A. Reeve, L. Lauterbach, P. A. Ash, O. Lenz and 40 M. de Torres, J. Dimroth, I. W. C. E. Arends, J. Keilitz and

ˇ

¨

K. A. Vincent, Chem. Commun., 2012, 48, 1589–1591.

F. Hollmann, Molecules, 2012, 17, 9835–9841.

Chem. Sci., 2021, 12, 8105–8114 | 8113

View Article Online

Chemical Science

Edge Article

41 M. Aizawa, R. W. Coughlin and M. Charles, Biochim. Biophys. 45 M. M. Grau, M. Poizat, I. W. C. E. Arends and F. Hollmann,

Acta, 1975, 385, 362–370. Appl. Organomet. Chem., 2010, 24, 380–385.

42 N. Hiyoshi, O. Sato, A. Yamaguchi and M. Shirai, Chem. 46 K. A. Brown, M. B. Wilker, M. Boehm, H. Hamby, G. Dukovic

Commun., 2011, 47, 11546–11548. and P. W. King, ACS Catal., 2016, 6, 2201–2204.

43 J. Lutz, F. Hollmann, T. V. Ho, A. Schnyder, R. H. Fish and 47 H. Zhao and W. A. van der Donk, Curr. Opin. Biotechnol.,

A. Schmid, J. Organomet. Chem., 2004, 689, 4783–4790. 2003, 14, 583–589.

€

44 Y. Okamoto, V. Ko and T. R. Ward, J. Am. Chem. Soc., 2016, 48 R. A. Sheldon, Green Chem., 2017, 19, 18–43.

138, 5781–5784.

8114 | Chem. Sci., 2021, 12, 8105–8114

Article Doi

Chemo-bio catalysis using carbon supports: application in H2-driven cofactor recycling

DOI: 10.1039/d1sc00295c

Source and publish data:

Authors:

Article abstract of DOI:10.1039/d1sc00295c

Full text of DOI:10.1039/d1sc00295c

Products guided by the article

R&D Labs maybe for 7095-08-1

Relevant to this article

Hot Product