Strategies for Transferring Photobiocatalysis to Continuous Flow Exemplified by Photodecarboxylation of Fatty Acids

Abstract

The challenges of light-dependent biocatalytic transformations of lipophilic substrates in aqueous media are manifold. For instance, photolability of the catalyst as well as insufficient light penetration into the reaction vessel may be further exacerbated by a heterogeneously dispersed substrate. Light penetration may be addressed by performing the reaction in continuous flow, which allows two modes of applying the catalyst: (i) heterogeneously, immobilized on a carrier, which requires light-permeable supports, or (ii) homogeneously, dissolved in the reaction mixture. Taking the light-dependent photodecarboxylation of palmitic acid catalyzed by fatty-acid photodecarboxylase from Chlorella variabilis (CvFAP) as a showcase, strategies for the transfer of a photoenzyme-catalyzed reaction into continuous flow were identified. A range of different supports were evaluated for the immobilization of CvFAP, whereby Eupergit C250 L was the carrier of choice. As the photostability of the catalyst was a limiting factor, a homogeneous system was preferred instead of employing the heterogenized enzyme. This implied that photolabile enzymes may preferably be applied in solution if repair mechanisms cannot be provided. Furthermore, when comparing different wavelengths and light intensities, extinction coefficients may be considered to ensure comparable absorption at each wavelength. Employing homogeneous conditions in the CvFAP-catalyzed photodecarboxylation of palmitic acid afforded a space-time yield unsurpassed by any reported batch process (5.7 g·L^–1·h^–1, 26.9 mmol·L^–1·h^–1) for this reaction, demonstrating the advantage of continuous flow in attaining higher productivity of photobiocatalytic processes.

Introduction

Although direct excitation of organic compounds by light to facilitate chemical reactions has been known for more than a century, the recent rise of photoredox catalysis has significantly expanded the scope of photochemistry in the synthesis of organic molecules.¹ Apart from enabling new reaction pathways not accessible by thermal control, the use of photon energy is also considered a sustainable approach.² Akin to photocatalysis, biocatalysis has been shown to enable sustainable reaction methodologies due to its generally mild reaction conditions, use of aqueous media, high stereo- and chemoselectivity, and ever-increasing substrate and reaction scope.³

At the interface of these two fields of catalysis, photobiocatalysis arose, whereby either the enzyme itself requires light (naturally or by promiscuity) or photoinduced electrons are transferred to it.⁴ The scope of photobiocatalytic transformations was significantly expanded following the discovery of fatty-acid photodecarboxylases that catalyze light-dependent decarboxylation of fatty acids (Scheme 1).⁵

Scheme 1. Photodecarboxylation of Fatty Acids Catalyzed by the Photodecarboxylase CvFAP from Chlorella variabilis NC64A.

This class of natural photoenzymes and its most known orthologue from C. variabilis NC64A (CvFAP) has subsequently shown great promise for the production of biofuels,⁶ and apart from its natural substrates, fatty acids, the wild-type enzyme, or its variants have been demonstrated to decarboxylate hydroxy- and amino-substituted fatty acids,⁷ dicarboxylic acids,⁸ α-substituted carboxylic acids,⁹ and the herbicide phosphinothricin.¹⁰ In the enzyme’s reaction mechanism, the decarboxylation event is triggered by a single-electron transfer from the substrate’s carboxylate group to the oxidized flavin cofactor, yielding a terminal alkyl radical and the flavin semiquinone. The following back electron transfer or hydrogen atom transfer are less well understood and might involve the cysteine 432 residue, the arginine 451 residue, or a water molecule within the active site.¹¹ In the absence of a substrate, the excited cofactor was shown to perform single-electron abstractions from amino acids in the active site, which leads to deactivation of the enzyme.¹² Furthermore, the cofactor FAD itself is prone to photodegradation.¹³ Hence the limited photostability of CvFAP must be considered in the development of any process.

According to the Beer–Lambert law, the penetration of visible light into a medium decreases exponentially with increasing path length. Consequently, external illumination of common batch reactors, especially at larger volumes, leads to nonhomogeneous irradiation and low reaction rates, as most of the light gets absorbed near the surface of the vessel.¹⁴ Performance of light-dependent reactions in continuous flow, in particular, in microchannels (ID < 1 mm), represents a convenient way of circumventing this issue and has been demonstrated for chemocatalytic applications.^14,15 Although the performance of biocatalytic transformations in continuous flow has been established,¹⁶ examples of flow photobiocatalysis remain scarce and are not of synthetic applicability.¹⁷ Examples include the generation of oxygen by a cyanobacterium for the oxidation of cyclohexene to cyclohexanol¹⁸ and the photochemical cofactor regeneration for the enzymatic reduction of CO₂ to formate.¹⁹

Using light-dependent enzymes in flow offers various advantages, such as the increased light penetration, but goes hand in hand with challenges to be solved. The photoenzyme may be either immobilized and applied on a stationary phase, which requires light-permeable carriers, or applied in solution within the mobile phase. A major hurdle to be addressed is the potential photolability of the enzyme or its cofactor, which could be foreseen to represent a particular difficulty in terms of its immobilization and reuse. Careful optimization of illumination parameters (light intensity and illumination wavelength) and residence time is therefore needed. Additionally, the substrate concentration should be sufficient to achieve significant space-time yields (STY), which may require cosolvents or surfactants to make up for potential solubility issues, and the corresponding additives need to be compatible with the (immobilized) enzyme. In addition to a single experiment that was reported during the revision of this manuscript,²⁰ the outlined issues have not yet been addressed in the context of photoenzymes. Therefore, herein, the photodecarboxylase from C. variabilis (CvFAP) is taken as a showcase for the development of a flow photobiocatalysis methodology, with particular focus on solutions to encountered obstacles and future challenges.

Results and Discussion

Wavelength and Light Intensity Optimization

When it comes to the development of a photoenzyme-catalyzed transformation, the appropriate wavelength and light intensity represent key parameters to be considered. The wavelength dictates the energy provided for the excitation of the catalyst, whereas the light intensity contributes to the rate of catalysis as well as the photoinactivation. The fatty-acid photodecarboxylase (CvFAP) was chosen as a model enzyme and expressed in Escherichia coli using a gene construct lacking the predicted chloroplast-targeting sequence (residues 62–654 of CvFAP) as reported previously.²¹ As photostability of CvFAP is known to be a severe issue (see also Figure S1),^11b,22 hexanoic acid (10 mM) and FAD (50 μM) were supplied to all buffers during purification to avoid deactivation processes initiated by excited flavin in the absence of the substrate (for the effect of different additives on the enzyme stability, see Figure S2). Furthermore, during purification, the enzyme was handled in the dark (or dim white light) at 4 °C or on ice (see the Experimental Section). The activity of the purified enzyme was assayed by means of decarboxylation of palmitic acid under blue light illumination (455 nm) at two light intensities (photon flux densities of 42 and 424 μmol·L^–1·s^–1) and varying enzyme concentrations (Figure 1). Furthermore, as preliminary experiments employing CvFAP-containing cell lysate showed that full conversion of palmitic acid was obtained at 528 nm under the conditions employed (Figure S3), the reaction was also investigated at 528 nm. It turned out that 0.046 mol % catalyst (6 μM CvFAP) was sufficient to reach >99% transformation of 13 mM palmitic acid to pentadecane at 455 nm and 42 μmol·L^–1·s^–1 photon flux density within 24 h and that >99% conversion was also attainable at 528 nm, albeit at a higher enzyme loading (10 μM). A 10-fold increase of light intensity at 455 nm had a negative impact on turnover at all tested concentrations of CvFAP, plausibly due to the accelerated photodegradation of the enzyme. Consequently, the 455 nm wavelength and 42 μmol·L^–1·s^–1 photon flux density were chosen as optimal illumination parameters for the decarboxylation of palmitic acid (for the conversion over time, see Figure S4).

Figure 1.

Dependence of conversion on photon flux density and CvFAP concentration at 528 and 455 nm. Reaction conditions: palmitic acid (13 mM), CvFAP (1–10 μM), Tris·HCl buffer (100 mM, pH 8.5), 30% v/v dimethylsulfoxide (DMSO), 1 mL volume. The reactions were run at 25 °C, 500 rpm, 24 h. Displayed concentrations are mean values calculated from duplicate experiments.

Homogenization of the Reaction Mixture Using Cosolvents, Ionic Liquids, and Surfactants

As it turned out that the above-described reaction mixture (100 mM Tris·HCl at pH 8.5, 30% v/v DMSO) containing 13 mM palmitic acid was prone to substrate precipitation, the solubility of the substrate had to be improved for a transfer to continuous flow. Consequently, a set of six organic cosolvents (THF, n-PrOH, i-PrOH, EtOH, 1,4-dioxane, MeCN) was investigated, whereby 10% v/v tetrahydrofuran (THF) and n-propanol (n-PrOH) allowed the dissolution of palmitic acid at 13 mM concentration (for 50 mM substrate 20% v/v cosolvent was required; Table S1). However, the solvents had a detrimental effect on the enzyme’s photodecarboxylation activity, and in most cases, a significant loss of activity was observed (Table S2). Subsequently, ionic liquids and surfactants were investigated (Figure S5 and Table S3), whereby Ammoeng 102 (1% v/v) enabled the solubilization of 13 mM palmitic acid while leading only to a slight decrease of activity (Table S2). Furthermore, it was observed that heating a solution of palmitic acid or sodium palmitate in Tris·HCl buffer (100 mM, pH 8.5), without additives, to 70 °C for 15 min, led to the formation of a uniform, turbid (colloid-like) dispersion. Although the substrate (13 mM) still precipitated after standing at room temperature, the decarboxylation reaction proceeded under those conditions with comparable conversion to that observed in reactions employing DMSO (30% v/v) as the cosolvent (Table S4).

Immobilization of CvFAP

As a prerequisite for continuous flow using CvFAP on a stationary phase, conditions for the immobilization of the enzyme were investigated. In the case of light-dependent reactions, the enzyme carrier should ideally be transparent and display stability under illumination. For this purpose, a set of commercial solid supports enabling different binding modes for the enzyme was investigated (Table 1).

Table 1. Enzyme Loading and Achieved Conversion of Immobilized CvFAP (Purified or from Cell Lysate) on Various Carriers.

^aImmobilization conditions: carrier (10 mg), purified CvFAP (2 mg·mL^–1), Tris·HCl (100 mM, pH 8.5), total volume 0.5 mL, 600 rpm, 4 °C, 15 h, apart from Dowex 66, which was shaken for 30 min.

^bBead activation: beads (10 mg), EDC·HCl (250 mM, 600 μL) in H₂O and N-hydroxysuccinimide (250 mM, 600 μL) in H₂O, shaken at 25 °C, 900 rpm, 30 min. Immobilization: activated beads (≈10 mg), CvFAP (4 mg·mL^–1, 500 μL) in H₂O, shaken at 25 °C, 900 rpm, 2 h.

^cImmobilization conditions: EziG (10 mg), lyophilized cell lysate (10 mg·mL^–1, 0.36 U) in 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) (100 mM, pH 8), total volume 0.5 mL, 4 °C, 900 rpm, 3 h.

^dImmobilization conditions: EziG (10 mg), purified CvFAP (4 mg·mL^–1) in Tris·HCl (100 mM, pH 8.5), total volume 0.5 mL, 4 °C, 900 rpm, 3 h.

^eRefers to the conversion observed in the photodecarboxylation of palmitic acid with immobilized CvFAP. Photodecarboxylation conditions: Immobilized enzyme (≈10 mg), palmitic acid (13 mM), Tris·HCl buffer (100 mM, pH 8.5, 700 μL total volume), DMSO (30% v/v), total volume 1 mL, 455 nm irradiation (42 μmol L^–1 s^–1), 25 °C, 1 h, 500 rpm.

^fSingle value. n.d.: not detected; CPG: controlled porosity glass; IMA: immobilized metal affinity; the colored rings refer to surface-coating (orange: semihydrophilic copolymer, red: polyvinyl benzyl chloride).

Ultimately, binding of CvFAP on seven out of ten carriers was observed, although only the enzyme immobilized on Eupergit C250 L and EziG beads displayed activity for the decarboxylation of palmitic acid. Additionally, the controlled porosity glass (CPG) material of the beads was reported to lead to high protein loading and low catalyst leaching.²³ The immobilization on Eupergit C250 L occurs via covalent linkage, while EziG allows selective binding of His-tagged proteins. EziG beads may give the possibility for immobilization of the tagged enzyme ideally directly from the cell lysate, thereby circumventing the need for a separate purification step. The direct immobilization of CvFAP from the cell lysate in Tris·HCl buffer was attempted first with three types of EziG beads (Opal, Amber, Coral) at varying immobilization times (1, 2, 3 h) and at 4 °C (Table S5). Although protein binding occurred, the resulting beads were inactive in the photodecarboxylation reaction. It was considered plausible that the components of the cell lysate in the supernatant (40 mg·mL^–1 lyophilized cell lysate) had a negative impact on the diffusion of the enzyme by virtue of an effect similar to macromolecular crowding.²⁴ The penetration of the enzyme into the pores of the carrier may have also been hampered by hexamer formation induced by the binding of fatty acids from the cell lysate to the active site.^12a,25 In an attempt to diminish the presumed crowding effect, the cell lysate was concentrated with an ultrafiltration device (50 kDa cutoff). As the concentrated lysate exhibited an activity 4 times higher than that of the nonconcentrated lysate (Figure S6), the amount of cell lysate applied for immobilization was reduced four times (from 40 to 10 mg·mL^–1), while still providing the same CvFAP activity. Subsequently, the immobilization of CvFAP from the concentrated lysate batch onto EziG Opal was examined with different buffers (Figure S7). Comparable protein loading (≈8% w/w) was achieved with 3-morpholinopropane-1-sulfonic acid (MOPS) (100 mM, pH 8), HEPES (100 mM, pH 8), and Tris·HCl (100 mM, pH 8.5), while immobilization in potassium phosphate buffer (100 mM, pH 8) and loading buffer (used for the purification of CvFAP) led to significantly lower protein loading (≈4% w/w). Nonetheless, photodecarboxylation of palmitic acid was only observable after immobilization from HEPES (100 mM, pH 8) buffer. The cause of the observed effect has not been elucidated and is highly speculative. Testing immobilization times with different EziG beads revealed comparable enzyme loadings for EziG Opal and EziG Coral, while the activity was only observed using the former as the solid support (Figure S8). On the other hand, the enzyme loading appeared to be lower in the case of EziG Amber, and conversion was found only after 3 h of immobilization. Protein loading appeared to be constant over all three examined time points (within the displayed errors), suggesting that maximal binding had been achieved after 1 h of immobilization already. Determining the activity of the cell lysate before and after immobilization indicated that 26% of activity had been lost from the lysate upon 3 h of immobilization at 4 °C. Elution of the protein bound to EziG beads with the buffer containing high imidazole concentration (500 mM) and subsequent sodium dodecyl sulfate- polyacrylamide gel electrophoresis (SDS-PAGE) analysis of the resulting mixture confirmed that nonselective binding was taking place (Figure S9).

As an alternative strategy for improving the immobilization yield and activity of the heterogenized enzyme, the immobilization of purified CvFAP was evaluated on EziG Opal beads (3 h, 4 °C). Indeed, this approach enabled higher loadings and activity to be reached compared to the immobilization of CvFAP from the cell lysate (Figure 2A). The highest activity was observed after immobilization in Tris·HCl (100 mM, pH 8.5) buffer for 3 h at 4 °C. Tris·HCl (100 mM, pH 8.5) buffer was also used for the subsequent investigation of shorter immobilization times (Figure 2B). A gradual increase in protein loading and activity with time was observed, and the highest activity was attained after 3 h of immobilization. Immobilization at 25 °C revealed constant protein loading over time (Figure 2C). Somewhat higher activity was observed after 1 and 2 h of immobilization, although still comparable to the 0.5 h, considering the corresponding errors.

Figure 2.

Immobilization of purified CvFAP onto EziG Opal in batch. (A) Buffers tested: MOPS (100 mM, pH 8), HEPES (100 mM, pH 8), KPi (100 mM, pH 8), and Tris·HCl (100 mM, pH 8.5). (B) Evaluation of immobilization time in Tris·HCl buffer at 4 °C. (C) Evaluation of immobilization time in Tris·HCl buffer at 25 °C. Immobilization conditions: EziG (10 mg), purified CvFAP (4 mg mL^–1) in the corresponding buffer, total volume 0.5 mL, 4 °C or 25 °C, 900 rpm, 3 h unless indicated otherwise. Activity assay: Immobilized enzyme (10 mg), palmitic acid (13 mM), Tris·HCl buffer (100 mM, pH 8.5, 700 μL total volume), DMSO (30% v/v), total volume 1 mL, 455 nm irradiation (42 μmol·L^–1·s^–1), 25 °C, 1 h, 500 rpm. The displayed errors correspond to standard deviations of triplicate determinations.

As a result of this study, it turned out that direct immobilization using EziG from the lysate is less suitable. Consequently, the photoenzyme immobilized on EziG Opal and Eupergit C250 L was used for the flow experiments.

Application of Immobilized CvFAP in Flow

After CvFAP immobilization, the heterogenized catalyst was applied to the photodecarboxylation reaction in flow. The corresponding flow setup involved a previously reported custom-made illumination device (Figure 3).²⁶ A packed-bed reactor, employing a fluorinated ethylene propylene (FEP) tube (OD 1/8″, ID 0.062″) connected to PEEK reducing unions (1/8″ to 1/16″) and equipped with a stainless steel Frit-in-a-Ferrule (0.5 μm porosity) (Figure 3A) was placed into a photoreactor chamber before illumination (Figure 3B,C).

Figure 3.

Experimental setup of flow photodecarboxylation under heterogeneous conditions. (A) Packed-bed reactor setup. (B) Overhead view of the reactor during illumination. (C) Frontal view of the complete experimental setup with the visible photoreactor chamber during illumination.

Investigating the EziG beads first, the tube was packed with dry EziG Opal beads and immobilization of purified CvFAP was performed in flow. Prior to the photodecarboxylation step, the beads were washed with the substrate for the purpose of enzyme stabilization. In this step, adsorption of palmitic acid on EziG Opal was found, indicating that equilibration of the beads with the substrate solution is necessary (Figure S10). Furthermore, a distinctly yellow flow-through suggested desorption of CvFAP from the beads. The experiments suggested that this leaching was facilitated by the detergent Ammoeng 102 (Figure S11) maybe due to the binding of polyethylene glycol chains of Ammoeng 102 to the carrier. Trials to circumvent desorption, e.g., by the addition of magnesium chloride (300 mM) to the reaction mixture, led to reduced leaching compared to the mixture containing only Ammoeng 102, however, still not below the level observed with the buffer alone. Performing the experiments in the absence of Ammoeng 102, neither the substrate nor the product were recovered from the resulting flow-throughs; washing of the beads with ethyl acetate revealed that both were retained on the beads.

Since EziG Opal turned out not to be suitable, further experiments were performed using Eupergit C250 L, as the covalent binding mode of the enzyme to the carrier was expected to prevent leaching. The immobilization of purified CvFAP onto Eupergit C250 L (80 mg) was performed in flow (60 min residence time) leading to a loading of 2.1% w/w at room temperature, which was comparable to the loading observed previously in a batch at 4 °C (Table 1, Entry 5). To ascertain that the enzyme stays on the beads, the activity of the flow-through after washing the beads with the substrate solution [13 mM palmitic acid in Tris·HCl (100 mM, pH 8.5; Ammoeng 102, 1% v/v)] was assayed and no decarboxylation activity was found as soon as washing was completed (it was also colorless). Additionally, during the equilibration phase that was performed in the dark, no product formation was observed. After immobilization, the enzyme was illuminated at 455 or 528 nm at a photon flux of 2.7 μE·s^–1 and a residence time of 5 min (Figure 4A). The flow-through was collected in 0.5 mL fractions (2.5 mL in total) and analyzed for conversion. Two important observations were made. First, the conversion differed already in the first fraction: a 2.5-fold higher conversion was observed at 455 nm compared to 528 nm. Second, the conversion decreased over the fractions, whereby the decrease of conversion was more pronounced at 455 than at 528 nm. After five fractions, 31% of the initial activity retained for illumination at 455 nm, whereas 80% of the remaining activity was found at 528 nm.

Figure 4.

Photodecarboxylation of palmitic acid catalyzed by CvFAP under heterogeneous conditions, immobilized on Eupergit C250 L (2.1% w/w). (A) Irradiation with 455 and 528 nm light sources (2.7 μE·s^–1 photon flux) at 5 min residence time. (B) Irradiation with 455 and 528 nm light sources at 1.0 and 2.7 μE·s^–1 photon flux, respectively, at 5 min residence time. (C) Variation of residence time at 455 nm (1.0 μE·s^–1). Purified CvFAP (mass equivalent to 5% of the weight of beads) was passed through the reactor packed with Eupergit C250 L (≈80 mg) at a rate of 3.75 μL·min^–1 (60 min residence time). Palmitic acid (13 mM) in Tris·HCl (100 mM, pH 8.5; Ammoeng 102, 1% v/v) was pumped through the packed-bed reactor in fractions (0.5 mL) at a rate of 11.25–45 μL·min^–1 (depending on the desired residence time) at room temperature.

The observed difference in conversion at two different wavelengths in the first fraction may be attributed to a different rate of photodegradation at different wavelengths, or simply due to a varied activity at different wavelengths. As it was previously shown that the apparent quantum yield of CvFAP in principle follows its absorption spectrum,²⁶ the difference in photon absorption due to varied extinction coefficients at different wavelengths was investigated to compare the illumination at different wavelengths. Measuring the ratio of the extinction coefficients at 455 and 528 nm ε(455 nm)/ε(528 nm) revealed a value of 2.8 (Figure S12). When the light intensity at 455 nm wavelength was adjusted by this factor, the observed conversions coincided with the conversion measured at 528 nm, indicating that the previously observed difference in the apparent quantum yield was caused by the difference in the number of absorbed photons at these wavelengths, rather than a different stability (Figure 4B). This underlines that consideration of the extinction coefficients is a prerequisite for adequate comparison of different wavelengths of photochemical reactions. Additionally, the adjustment of the photon flux according to the respective extinction coefficients may allow the use of any wavelength from the absorption spectrum of the enzyme–cofactor complex while maintaining comparable catalytic activity. Next, the effect of residence time on conversion over time was examined. Photodecarboxylation of palmitic acid was performed at 455 nm (1.0 μE·s^–1 photon flux) and eight fractions (0.5 mL) were collected at 5, 10, and 20 min residence times (Figure 4C). Although the highest values of conversion were observed at 20 min of residence time, the highest overall product formation was obtained at 10 min of residence time (7 μmol, TON = 326), followed by 20 min (6 μmol, TON = 282) and 5 min (5.4 μmol, TON = 251). Conversely, the highest space-time yield, under the examined conditions, was obtained after 5 min of residence time (2 mmol·L^–1·h^–1), followed by 10 min (1.3 mmol·L^–1·h^–1) and 20 min (1 mmol·L^–1·h^–1). In principle, the application of immobilized catalysts in continuous flow should be beneficial over batch, as issues connected to mass transfer into the immobilization matrix are overcome. However, the activity of the enzyme, displayed as conversion, dropped rapidly during the course of the experiment, presumably due to photoinactivation as it was investigated in Figure S1. This makes the application of the enzymes over a longer time scale challenging and indicates that this photoenzyme is not suitable to be used on a stationary phase, even though the enzyme is continuously supplied with the substrate. This observation may call into question the applicability of a heterogenized photoenzyme displaying intrinsic photolability if there is no way of stabilization or reactivation. Consequently, photodecarboxylation was investigated in homogeneous flow, where the constant supply of fresh enzyme enables more consistent activity over time.

Photodecarboxylation in Flow Using a Homogeneous Reaction Mixture

Fluorinated ethylene propylene (FEP) tubes (OD 1/16″, ID 0.03″) were looped through the illuminated chambers four times to allow illumination over a length of 14.5 cm (264 μL volume) and photodecarboxylation mixtures were pumped through the system using a syringe pump (Figure 5).

Figure 5.

Continuous flow setup employing a custom-made illumination device and FEP tubing (OD 1/16″). (A) Complete setup including the illumination device and a syringe pump. (B) Frontal view of the illumination device. (C) Overhead view of the illuminated chamber equipped with FEP tubing.

The reaction mixture containing palmitic acid (13 mM) and purified CvFAP (6 μM) in Tris·HCl (100 mM, pH 8.5; Ammoeng 102, 1% v/v) was illuminated under flow conditions at 455 and 528 nm at varying light intensities and 10 min of residence time (Figure 6A). At 455 nm, a steady increase in conversion was observed when increasing the photon flux until a maximum was reached at 10.7 μE·s^–1. The subsequent drop in conversion may have arisen from enzyme photodegradation. At 528 nm, the maximum was reached at 5.4 μE·s^–1; however, the observed value was nearly 3-fold lower than the one achieved at 455 nm.

Figure 6.

Decarboxylation of palmitic acid catalyzed by purified CvFAP in FEP tubing (OD 1/16″, ID 0.03″, 264 μL reactor volume). (A) Dependence of conversion on photon flux density at 6 μM CvFAP concentration and 10 min residence time. (B) Dependence of conversion on residence time at 10.7 μE·s^–1 photon flux and 6 μM CvFAP concentration. (C) Dependence of conversion on CvFAP concentration at 10.7 μE·s^–1 photon flux and 10 min residence time. (D) Decarboxylation of palmitic acid under homogeneous flow conditions in continuous operation (60 μM CvFAP, 10.7 μE·s^–1 photon flux, 10 min residence time). Reaction conditions: the mixture of palmitic acid (13 mM) and CvFAP (varying concentrations) in Tris·HCl (100 mM, pH 8.5; Ammoeng 102, 1% v/v) was pumped through the tube at a flow rate of 8.8–26.4 μL·min^–1, depending on the desired residence time. The illumination was performed at 455 or 528 nm at room temperature. The conversion was determined from calibration curves. The photon flux was calculated from radiant power specified by the producer as the sum of values of individual light-emitting diodes (LEDs).

Increasing the residence time from 10 to 20 min at 455 nm (10.7 μE·s^–1 photon flux) had no observable influence on conversion (Figure 6B). Consequently, the residence time of 10 min was chosen as optimal and the effect of enzyme loading on conversion was examined. The concentration of CvFAP was varied between 12 and 90 μM, whereby maximal product formation was reached at 60 μM concentration (65% conversion, Figure 6C). The resulting continuous flow process was calculated to have a space-time yield (STY) of 5.7 g·L^–1·h^–1 (26.9 mmol·L^–1·h^–1), which represents the highest pentadecane production rate of any CvFAP-catalyzed process to date.^6c Lastly, to demonstrate that the reported substrate scope of the enzyme under batch conditions can be transferred to flow, six additional saturated fatty acids that are reported to be accepted by CvFAP^21,27 were tested in the developed continuous process (Table 2).

Table 2. Substrate Scope of CvFAP-Catalyzed Decarboxylation of Fatty Acids under Homogeneous Flow Conditions^a.

fatty acid (C:D)b	conversion (%)	STY (g·L–1·h–1)
10:0	37	2.0
11:0	20	1.2
12:0	24	1.6
14:0	56	4.2
15:0	58	4.8
16:0	65	5.7
18:0	30	2.9

^aReaction conditions: The mixture of fatty acid (13 mM) and CvFAP (60 μM) in Tris·HCl (100 mM, pH 8.5; Ammoeng 102, 1% v/v) was pumped through the tube at a flow rate of 26.4 μL·min^–1 (10 min residence time), 455 nm illumination (10.7 μE·s^–1 photon flux), and room temperature. The conversion was determined from calibration curves.

^bC:D = (carbon atoms):(double bonds).

Although a steady decrease of conversion was observed with declining carbon chain length, as observed in previous reports,²⁸ the decarboxylation of decanoic acid still proceeded with 37% conversion (STY: 15.3 mmol·L^–1·h^–1, 2.0 g·L^–1·h^–1). Favorably, octadecanoic acid was also found to be soluble under optimized conditions and was accepted as a substrate. Further improvement of productivity would likely be attainable using variants of CvFAP.²⁹ To examine if the developed homogeneous flow procedure retains productivity under continuous operation, the decarboxylation of 8 mL of the palmitic acid mixture (13 mM) was performed and only minor changes in conversion were observed over time (Figure 6D).

Alternatively, a continuous flow decarboxylation protocol in neat Tris·HCl (100 mM, pH 8.5) buffer was developed, whereby substrate concentration was lowered to 3 mM to avoid precipitation. Under optimal conditions, a pentadecane production rate of 0.5 g·L^–1·h^–1 (2.5 mmol·L^–1·h^–1) was achieved.

Conclusions

Investigating continuous flow methodologies employing a photoenzyme has led to important conclusions and identified future challenges. Fatty-acid photodecarboxylase from C. variabilis (CvFAP) was used as a showcase for the transfer of photoenzyme-catalyzed transformation from batch to flow. Thereby, several challenges had to be solved: (i) identifying a suitable carrier for immobilization, (ii) avoiding enzyme leaching, (iii) evaluating homogeneous versus heterogeneous conditions, (iv) enabling light to reach the enzyme, (v) ensuring a dissolved substrate, and importantly, (vi) identifying illumination conditions and residence times that allow maximizing productivity over photodegradation. The use of Ammoeng 102 as a surfactant (1% v/v) enabled the dissolution of palmitic acid in Tris·HCl (100 mM, pH 8.5) buffer at 13 mM concentration, while still maintaining high enzyme activity. This setup was successfully used for the continuous flow experiments using the photoenzyme immobilized on Eupergit C250 L as the stationary phase. On the other hand, the surfactant was not compatible with EziG, leading to leaching of the bound enzyme. Furthermore, for optimization and a precise comparison of photoreactions at different wavelengths, the extinction coefficients at these wavelengths need to be considered and light intensities adapted accordingly. When doing so for the reaction with CvFAP at 455 and 528 nm, comparable conversions were observed at the two wavelengths, indicating similar quantum yields, while when using the same light intensity for both wavelengths, a clear difference in conversion was observed. Nevertheless, for CvFAP, a rapid decrease of conversion was found within hours due to photodegradation, which may be a general issue for photoenzymes. If no stabilization or regeneration of the photoenzyme can be ensured, continuous flow using homogeneous conditions might be the preferred option. Despite the low photostability of CvFAP, optimization of the photodecarboxylation of palmitic acid under homogeneous flow conditions resulted in a space-time yield of 5.7 g·L^–1·h^–1 (26.9 mmol·L^–1·h^–1), representing the highest pentadecane productivity of any CvFAP-catalyzed process to date. Overall, the presented results contribute to the immobilization of photoenzymes and pave the way to broader application of photobiocatalysis in flow.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscatal.2c04444.

• Experimental methods, supplementary figures and tables, chromatographic data and calibration curves, and gene sequence (PDF)

The authors declare no competing financial interest.