Electrocatalytic Volleyball: Rapid Nanoconfined Nicotinamide Cycling for Organic Synthesis in Electrode Pores

Abstract In living cells, redox chains rely on nanoconfinement using tiny enclosures, such as the mitochondrial matrix or chloroplast stroma, to concentrate enzymes and limit distances that nicotinamide cofactors and other metabolites must diffuse. In a chemical analogue exploiting this principle, nicotinamide adenine dinucleotide phosphate (NADPH) and NADP+ are cycled rapidly between ferredoxin–NADP+ reductase and a second enzyme—the pairs being juxtaposed within the 5–100 nm scale pores of an indium tin oxide electrode. The resulting electrode material, denoted (FNR+E2)@ITO/support, can drive and exploit a potentially large number of enzyme‐catalysed reactions.

Abstract: In living cells,redoxchains rely on nanoconfinement using tiny enclosures,s uch as the mitochondrial matrix or chloroplast stroma, to concentrate enzymes and limit distances that nicotinamide cofactors and other metabolites must diffuse. In ac hemical analogue exploiting this principle,nicotinamide adenine dinucleotide phosphate (NADPH) and NADP + are cycled rapidly between ferredoxin-NADP + reductase and as econd enzyme-the pairs being juxtaposed within the 5-100 nm scale pores of an indium tin oxide electrode.T he resulting electrode material, denoted (FNR + E2)@ITO/support, can drive and exploit ap otentially large number of enzyme-catalysed reactions.
Nature has evolved efficient systems whereby coupled enzyme reactions occur in irregular nanoconfined threedimensional zones-mitochondria and chloroplasts being prime examples. [1] In photosynthesis,s unlight is used to regenerate nicotinamide adenine dinucleotide phosphate (NADPH) and adenosine triphosphate (ATP). [2] Ferredoxin (Fd) transfers two electrons,o ne at at ime,f rom photosystem It of erredoxin NADP + reductase (FNR), which is af lavoenzyme that catalyses the conversion of NADP + to NADPH. [2,3] Fast and efficient recycling occurs within < 100 nm in the chloroplast stroma. [2,3] We discovered recently that FNR binds tightly in the pores of an indium tin oxide (ITO) electrode,w here it exhibits rapid electron transfer, reversible electrocatalysis of NADP + /NADPH interconversion, and catalytic coupling to ad ehydrogenase introduced to the solution. [4] Thee lectrode was constructed by electrophoretic deposition of ITO particles (diameter < 50 nm;S upporting Information, Figure S1) onto ac onductive support ( Figure 1A;F igure S1). Theinterparticle spaces form disordered pores with diameters ranging from 5-100 nm ( Figure 1A;F igure S1), which is similar to chloroplast stroma. [2,3] Ap ristine ITOg lass electrode (that is,w ithout an electrophoretically deposited porous ITOl ayer) showed no discernible FNR binding or electrocatalytic activity.
We now report amassive nanoconfinement effect for twoenzyme electrocatalysis utilising NADP(H) recyclinga nd show that the fundamental principle enabling enzyme cascades to operate so efficiently in living cells extends to the development of electrodes for controllable and selective organic synthesis.W ed emonstrate that electrochemically driven cofactor recycling occurs exclusively between FNR and adehydrogenase enzyme (E2) that is co-entrapped in the pore,l eading also to local enhancement of NADP(H) concentration. Ther esulting material, denoted "(FNR + E2)@ITO/support", is highly electrocatalytic because of the combined action of two enzymes.
Thetight binding of FNR in the ITOpores is quantified by analysing the stable "non-turnover" cyclic voltammetry peaks measured in the absence of NADP(H), which correspond to rapid, reversible 2-electron reduction and oxidation of the enzymesf lavin cofactor, FAD( Figure 1B). [4a] Different amounts of FNR were pre-adsorbed at an ITOe lectrode by placing it in astirred FNR solution (2.5 mm in buffer,pH8.0) for various durations.Electroactive coverages were obtained from charges given by the peak areas ( Figure 1B). As aguide, and based on globular diameter,asingle packed monolayer of FNR on aflat surface would give acoverage of approximately 5pmol cm À2 .I nc ontrast, actual values ranged from 15 to 135 pmol cm À2 ,depending on exposure time.
Each electrode was used to carry out the oxidation of (S)-(+ +)-4-phenyl-2-butanol (hereafter, 4-phenyl-2-butanol) catalysed by ADH and monitored by chronoamperometry at + 0.08 Vv ersus standard hydrogen electrode (SHE; Figure 1C). We measured oxidation rather than reduction to facilitate experiments that otherwise require rigorous anaerobicity to avoid current spikes from trace O 2 when reagents are injected. Upon addition of ADH (final concentration 0.03 mm)t ot he cell solution (3 mL), already containing NADPH (5 mm), the current (which is directly proportional to catalytic rate;that is,current/2F,where F is Faradays constant) rose gradually from zero to reach al imiting value. Ther ight axis here and elsewhere shows the catalytic rate derived from the current using Faradaysc onstant. Importantly,the curve contains information on two different rates: 1) the slope during the approach to the limiting value represents the rate of development of catalytic activity (that is,h ow fast the operational electrocatalyst is assembled); 2) the limiting value represents the optimal steady-state rate of catalytic conversion (moles product/time) that is achieved. Acurrent density of 54 mAcm À2 (geometric electrode surface area) corresponds to ac atalytic conversion rate of 1 mmol product cm À2 h À1 .
We interpret the development of catalytic activity as being ac onsequence of the second enzyme (E2) entering the ITO pores and binding close to FNR, and the complementary enzyme partners executing nanoconfined cofactor recycling with am assively enhanced catalytic rate.P roduction of 4-phenyl-2-butanone was confirmed by 1 HNMR spectroscopy ( Figure S2). Theessentially exponential current growth suggests af irst-order process dependent on the number of adsorption sites available to incoming ADH molecules.T he electrode with the lowest FNR coverage gave the highest initial rate of ADH adsorption but attained the lowest maximum level. Thus,l ower amounts of pre-adsorbed FNR limit the final catalytic current (activity) but present less resistance to incoming ADH molecules.F igure 1D presents experiments in which the quantity of pre-adsorbed FNR was held roughly constant, and three different levels of ADH were introduced. From the magnified view shown in Figure 1E,i ti sc lear that introducing ADH to the cell does not cause an immediate increase in current, as would be expected were ADH to contribute to the catalytic activity while in solution. Themaximum current and rate of rise both increase with ADH concentration in an on-linear manner (a 10-fold increase yielding less than af ivefold increase in maximum current). Theo rder of addition was then reversed;t hat is, FNR was introduced to ITOt hat had been pre-exposed to ADH. Unlike FNR, ADH is not an electron-transfer enzyme, so we could not quantify its adsorption by cyclic voltammetry. Instead, increasing amounts of ADH were preloaded at ITO/ graphite by varying the incubation time between 0.3 and 150 min. Each electrode was then rinsed thoroughly before placing it in acell solution containing substrate and NADPH  Ther ate of increase was greatest for the experiment in which ADH had been exposed to ITOf or the shortest time (that is,0.3 min), suggesting that FNR adsorbs more rapidly if less ADH is already present in the pores.L ong ADH preloading times gave lower maximum current but higher stability.After each experiment, the electrode was rinsed and placed in af resh solution devoid of substrates.C yclic voltammetry verified that the amount of adsorbed FNR increases with decreasing ADH pre-adsorption. Figure 2B presents studies in which the FNR concentration was varied and the preloaded ADH level was kept as uniform as possible by dropcasting for 30 min in each case. Them aximum current and rate of binding of FNR both increase non-linearly with FNR concentration between 0.03 and 0.3 mm.T he current for 0.003 mm FNR was barely visible, while 1.3 mm FNR yielded the most rapid increase but gave the greatest instability.
To establish how tightly each component is trapped in the ITOp ores,a ne xperiment was carried out in which the cell solution was replaced during the reaction (Figure 3). An FNR@ITO/graphite electrode was made by dropcasting FNR (1 mm, 5 mL) for 5min and subsequently rinsing thoroughly with pure water. Thee lectrode (electroactive FNR coverage 60 pmol cm À2 )w as placed in the cell solution containing 4-phenyl-2-butanol (20 mm)a nd NADPH (5 mm), and an oxidising potential (+ 0.08 Vv s. SHE) was applied. After injecting ADH (final concentration 0.3 mm)t he catalytic current gradually increased to reach ahigh steady-state level, as expected. Thee lectrode was then removed and stored in av ial containing the original cell solution. After thoroughly rinsing with purified water, the cell was recharged with fresh buffer containing only 4-phenyl-2-butanol (20 mm). The electrode was removed from storage,r insed thoroughly with purified water, and returned to the cell. Upon resuming measurement, the current decreased immediately,a s expected, but not to zero.I nstead ar esidual current was observed, which decreased over the course of 30 min. Readdition of NADPH produced an immediate current increase to av alue that, despite the disturbance to the electrode,w as about 90 %o ft hat obtained before the experiment had been interrupted. Ther esults confirmed that FNR and ADH are not required in solution to maintain binding on the electrode ( Figure S3) and demonstrated that only NADPH is needed to restore ah igh rate of catalytic conversion. Ther esults showed also that some NADP(H) is retained in the pores,a nd thereby undergoes many recycles before it is lost to solution. Its partial retention and concentration above the solution level was confirmed with cyclic  voltammetry of NADP + alone at aF NR@ITO/graphite electrode ( Figure S4). Plots of current versus (scan rate) 1/2 (corrected for the FADs ignal) consistently showed as mall upward deviation from linearity (standard evidence for as urface excess), even as the peaks broadened at higher scan rate. [9] Ap reformed (FNR + ADH)@ITO/glass electrode (that is,w ith no enzyme in solution) was used to demonstrate the reversibility and catalytic bias displayed by the (FNR + ADH) pair at pH 8.0. Theq uasi-reversible cyclic voltammetry of the NADP + /NADPH couple transformed into reversible electrocatalysis when substrates were added (Figure 4).
Ar atio of 20 mm alcohol/1 mm ketone was required to equalise oxidation and reduction currents,w hereas equal concentrations of alcohol and ketone (5 mm)s howed no oxidation at pH 8.5 or below ( Figure S5). Ther esults confirmed not only that (FNR + ADH)@ITOi sa"stand-alone" electrode,b ut also that the system is strongly biased toward alcohol production-ketone being aproduct inhibitor.
Thee xperiments shown in Figure 1w ere extended to three other dehydrogenases-each operating in the reducing direction. Figure 5c ompares the development of catalysis observed for ADH with that obtained when comparable concentrations of RedAm, ME, and (S)-IRED were introduced, all other components being present (Supporting Information) and the FNR@ITO/graphite electrode being prepared identically in each case.T he rate of activity development is enzyme-dependent;t hat is,R edAm @ ADH % ME @ (S)-IRED.A fter reaching am aximum value, the current decreased slowly for each enzyme ( Figure S6). Notably,the activity that was so rapidly attained with RedAm began to decrease slowly after 30 min compared to > 2hfor ADH and ME. In contrast, the activity of (S)-IRED,w hich developed only very slowly after initiation, remained comparably constant after 6h( Figure 5; Figure S6).
Thed ata provide aq ualitative framework on which ad etailed picture will eventually emerge.T he results showed no obvious trend with molecular mass or oligomeric state since the two comparable and fastest adsorbing enzymes, RedAm and ADH, are dimeric and tetrameric at 62 and 172 kDa, respectively.T he high affinity with which FNR (monomeric,3 9kDa) binds to ITOw as evident from observations that aF NR@ITO electrode remains active for NADP + /NADPH cycling for several days when as econd enzyme is absent. [4] Control experiments before and after 20 h at + 0.08 VorÀ0.44 Vs howed no significant changes apart from some physical cracking ( Figure S7). Exposure to highly reducing conditions are known to chemically degrade ITO electrodes,but we do not expect this to occur within our range of operating potentials. [10] Thes urface of ITO( isoelectric point % 6 [11] )w ill be negatively charged at pH 8. In the chloroplast stroma, Fd and FNR form ac omplex in which their redox centres are aligned approximately 6 apart for fast electron transfer. Thebinding is mainly electrostatic;the negative surface of Fd interacts with ap ositively charged patch on FNR. [12] Al ogical proposal for the electrode interaction is that FNR uses this patch to bind to the negatively charged pore walls of ITO. Thei soelectric points of FNR, RedAm, and ADH are higher than ITO( Figure S8), whereas those of ME and slowly binding (S)-IRED are lower.
Thel imiting currents in Figure 5r eflect the inherent catalytic activities of E2 and the balances between activities of FNR and E2 factored for relative coverage.Atthe substrate levels used, (S)-IRED is the least active enzyme,a sjudged also by solution assays used to test the activity of each batch. Guideline (literature) values of turnover frequency k cat and   Table S1 (apart from FNR, K M values are for target substrate).
Ther ate of development of catalytic activity reflects the rate at which the incoming enzyme enters and binds close to the incumbent enzyme,t hereby closing al ocal NADP(H) cycle.C ertain combinations exhibit unstable catalysis. Figure 2s hows that instability is associated with the lowest preloaded level of ADH (panel A) and the highest concentration of incoming FNR (panel B), and Figure 5s hows that RedAm gives al ess stable response despite binding most rapidly.T hese results suggest that instability may stem from aggressive displacement of the incumbent enzyme by the incoming enzyme.
Theb asis for the massive nanoconfinement effect is explained with as imple calculation. Within an assumed reaction volume of 10 10 10 nm 3 (1 10 À21 Lo r1zeptoliter) the concentration of each component of the minimal functional catalytic unit comprised of 1FNR, 1E2, and 1NADP(H) would be 1.6 mm :t he more mobile component, NADP(H) (D = 4.2 10 À6 cm 2 s À1 ), [4a] requires only approximately 0.1 mstotraverse this space.Although we cannot map the 3D occupancyofpores in terms of such minimal units,itis instructive to consider each 10 10 nm face (totalling 10 12 units cm À2 )a sa ne ffective target area on ah ypothetical flat electrode,i nw hich case the conversion rate for ADHcatalysed alcohol oxidation shown in Figure 5w ould correspond to asteady-state turnover frequency (per catalytic unit) in the order of 125 s À1 .Alower density of surface units would be compensated for by the (diminishing) participation of units located deeper in the pores."Minimal" unit is meant literally; improved catalytic rates may well require multiples of FNR or E2 (above 1:1). An alternative calculation based on participation of all the bound FNR (ca. 100 pmol cm À2 )inthe same experiment gives an empirical turnover frequency of 2s À1 , which is ap ractical value that must also represent the lower per-unit limit. Thee lectrochemical nanoreactor system has important implications for technology,w here cofactor regeneration is am aturing field. [13] Optimised in terms of turnover rates, stability,a nd amounts/ratios of enzymes,t he (FNR + E2)@ITO/support material can be scaled up and exploited as an inexpensive "plug-in" electrode to drive,i nteractively, ap otentially unlimited number of organic reactions depending on the identity of E2.