Unveiling the Reaction Pathway of Oxidative Aldehyde Deformylation by a MOF-Based Cytochrome P450 Mimic

Abstract

Understanding the reaction pathway of aldehyde deformylation catalyzed by natural enzymes has shown significance in developing synthetic methodologies and new catalysts in organic, biochemical, and medicinal chemistry. However, unlike other well-rationalized chemical processes catalyzed by cytochrome P450 (Cyt P450) superfamilies, the detailed mechanism of the P450-catalyzed aldehyde deformylation is still controversial. Challenges lie in establishing synthetic models to decipher the reaction pathways, which normally are homogeneous systems for precisely mimicking the structure of the active sites in P450s. Herein, we report a heterogeneous Cyt P450 aromatase mimic based on a porphyrinic metal–organic framework (MOF) PCN-224. Through post-metalation of iron(II) triflate with the porphyrin unit, a five-coordinated Fe^II(Porp) compound could be afforded and isolated inside the resulting PCN-224(Fe) to mimic the heme active site in P450. This MOF-based P450 mimic could efficiently catalyze the oxidative deformylation of aldehydes to the corresponding ketones under room temperature using O₂ as the sole oxidant and triethylamine as the electron source, analogous to the NADPH reductase. The catalyst could be completely recovered after the catalytic reaction without undergoing structural decomposition or compromising its reactivity, representing it as one of the most valid mimics of P450 aromatase from both the structural and functional aspects. A mechanistic study reveals a strong correlation between the catalytic activity and the C^α-H bond dissociation energy of the aldehyde substrates, which, in conjunction with various trapping experiments, confirms an unconventional mechanism initiated by hydrogen atom abstraction.

1. Introduction

The aldehyde deformylation reactions catalyzed by natural enzymes are of significant importance in biological systems and organic transformation processes [1,2,3]. For instance, in biofuel production, cyanobacterial aldehyde decarbonylase (cADO), belonging to the family of ferritin-like nonheme diiron oxygenase, can catalyze the conversion of linear C_n-fatty aldehydes to formate and the corresponding C_n-1 alkanes, which are high-valued hydrocarbon compounds closely related to those in the petroleum and natural gas industries [4]. Another important enzyme, cytochrome P450 (Cyt P450), is a superfamily of hemeproteins found in the majority of organisms and plays crucial roles in the biosynthesis of many endogenous compounds, drug metabolism, and the oxidative degradation of xenobiotics [5,6]. Its enzymatic catalytic function is substantially contingent upon its heme active center with a cysteine residue, which links to the iron center through the sulfhydryl group (-SH) (Scheme 1a) [1,7].

Both P450 and cADO-catalyzed aldehyde deformylation involve the interaction of metal ions with substrates, as well as the activation and transformation of oxygen. However, their specific mechanism is still unclear, especially for P450 aromatase [8,9,10]. One of the most accepted mechanisms involves a three-step oxidation pathway in which P450 aromatase catalyzes androgen in the presence of the cofactor NADPH and O₂ through a series of electron transfer processes (Scheme 1b). The mechanistic insight into the final aldehyde deformylation step is controversial [11]. One insists on a nucleophilic attack pathway of an in situ formed Fe(III)–peroxide species on the carbonyl group center to afford a peroxyhemiacetal adduct, which then decays via homolytic C-C bond cleavage to release formic acid and create an aromatic ring in the resulting estrogen products. Meanwhile, other perspectives suggest the involvement of Compound I (Cpd I) featuring a Fe(IV)–oxo π-cation porphyrin radical and the key oxidative intermediate in the catalytic cycle of Cyt P450 enzymes (frame in Scheme 1b). Figuring out the fundamental insights of P450-catalyzed aldehyde deformylation has significant impacts not only on sexual differentiation, reproductive system development, and the onset and progression of hormone-dependent diseases, including breast cancer and endometrial cancer, but also has profound implications for practical applications in the fields of medicine, the chemical industry, and environmental protection.

However, experimental studies on enzyme-catalyzed aldehyde deformylation remain challenging due to the extremely short lifetime of the possible intermediates generated during the catalytic cycle, the enzyme stability, the restricted catalytic activity under non-natural conditions, as well as the high cost and limited scalability [11]. Therefore, a wide range of biomimetic models and catalytic systems have been constructed and utilized for research on aldehyde deformylation, showing many intriguing mechanisms, some of which are even competing, i.e., mostly via nucleophilic attack of the metal–peroxo complex on the carbonyl center to form a peroxyhemiacetal adduct, while only a few examples based on manganese(III) peroxo underwent hydrogen atom abstraction from the C^α-position of the aldehydes [3,12,13,14,15,16]. Interestingly, Nam and Fukuzumi proposed the tunneling effect in aldehyde deformylation by a non-heme [Fe^III(OOH)(TMC)]²⁺ complex (TMC = 1,4,8,11-tetramethyl-1,4,8,11-tetraazacyclotetradecane), in which the mechanism was found to switch from a nucleophilic attack mode when the temperature was above 248 K to a hydrogen atom transfer mode when the temperature was lowered [17]. By now, most established biomimetic models for the study of aldehyde deformylation are based on homogeneous coordination metal complexes due to their unique merits, including the precise regulation and mimicking of the structure of the enzyme active site, characterization, and reproducibility.

Over the past few decades, many heterogeneous synthetic models have been meticulously constructed to mimic either the structure of the active center or the function of a certain enzyme, especially those using porous materials such as MOF and COF, which can not only serve as hosts to encapsulate natural enzymes [18,19,20] and nanoenzymes but also mimic the structure and coordination environment of the enzyme by incorporating the tailored functional species into the SBUs (secondary building units) and the organic linkers [21,22,23,24,25,26,27,28]. Lin and coworkers reported a self-adaptive MOF-253 whose adjacent bipyridyl (bpy) linkers could be rotated freely during post-synthetic construction and produce the [(bpy)Fe^III(μ₂-OH)]₂ active sites that mimic the function of the monooxygenases found in nature [29]. This MOF-based monooxygenase mimic exhibits high activity and stability in C–H bond oxidation and alkene epoxidation reactions, with O₂ as the sole oxidant. Such MOF-based models have also been achieved in the mimicking of the [FeFe]-hydrogenases by partially replacing the terephthalate linker in UiO-66 with a [FeFe](bdt)(CO)₆ (bdt = benzenedithiolate) ligand to afford UiO-[FeFe](dcbdc)(CO)₆ for efficient photocatalytic H₂ production [30]. And other important enzymes, such as phosphotriesterase [31], Cyt P450 [32,33], peroxidase [34], nitricoxide reductase [35], and carbonic anhydrase [26,36], among which Cyt P450 mimics using porphyrin-based MOFs, attract great attention since its porphyrin group highly resembles that in the heme center. A variety of biomimetic models have been constructed based on porphyrin MOFs and have been utilized in heme-mimicking catalysis, including hydrocarbon oxidation [37], olefin epoxidation [38], and aldol condensation [38], mostly using either artificial oxidants such as iodosobenzene (PhIO) and H₂O₂ or high temperature and pressure of O₂. Using molecular O₂ under mild conditions to mimic the natural enzyme–catalysis condition is much less probable due to the hindrance of isolating the possible intermediates for characterization. Although heterogeneous P450-aromatase mimics have hardly been achieved, their inherent advantages as recyclable catalysts and their potential ability for efficient catalysis under conditions that mimic natural enzyme catalysis cannot be overshadowed.

In this work, we report a MOF-based heterogeneous P450 aromatase mimic, i.e., PCN-224(Fe), produced from the coordination of the porphyrin group in PCN-224 with iron(II) triflate through a post-metalation process. The triflate is a weakly bound counter anion that could occupy the axial position as a co-ligand, leaving the vacant sixth position available for O₂ binding with the central Fe²⁺. As a result, this mode structurally mimics the heme prosthetic group in the Cyt P450 family. Under mild conditions, such as room temperature and atmospheric pressure, with triethylamine as the electron source analogous to the NADPH reductase, this P450 mimic exhibits high catalytic efficiency in the oxidative deformylation of aldehydes to generate the corresponding ketones. It also features high stability and reusability without structural decomposition or loss of reactivity, making it one of the most valid mimics of P450 aromatase. In addition, the mechanistic investigations uncover a robust relationship between the catalytic activity with the C^α-H bond dissociation energy of the aldehyde substrates, which, in conjunction with various trapping experiments, confirms an unconventional mechanism initiated by hydrogen atom abstraction.

2. Results and Discussion

2.1. Synthesis and Characterization of MOF-Based Cyt P450 Mimic

PCN-224 is a highly porous and stable framework composed of zirconium-based clusters and porphyrin linkers [39,40]. The robust nature of the Zr₆ clusters in PCN-224 contributes to the chemical and thermal stability of the framework, while the porphyrin units, typically derived from 5,10,15,20-tetrakis(4-carboxyphenyl)porphyrin (H₂TCPP), serve as the structural components of the framework and can coordinate with central metals through a post-synthetic modification (PSM) method to provide potential catalytic active sites. The incorporation of the porphyrin unit with iron(II) triflate makes the resulting PCN-224(Fe) an impeccable structural mimic of the natural Cyt P450 enzyme and a versatile platform for the development of MOF-based enzyme mimics.

The preparation of the catalyst was illustrated in Figure 1a according to the procedures in the literature with a few modifications [41,42], starting with the synthesis of PCN-224 via a typical solvothermal reaction between ZrOCl₂·8H₂O and H₂TCPP, followed by the post-metalation step using Fe(CF₃SO₃)₂ as the metal source in DMF. Iron(II) triflate as the metal source is one of the most different factors compared with the literature, which normally uses ferric halides with strong coordinating anions as the metal sources [43]. A series of materials, denoted as x-PCN-224(Fe)_y, could be afforded by simply adjusting the temperature (x) in the metalation step and the ratio (y) of Fe²⁺ over the porphyrin ligand. Among these, 100-PCN-224 (Fe)_5 was chosen as a representative sample based on its catalysis performance and was characterized by various techniques. Powder X-ray diffraction (PXRD) results confirmed the successful preparation of PCN-224 with high purity and crystallinity, which is in good agreement with the simulated structure (Figure 1b). From the SEM and TEM images, the obtained PCN-224 grows into cubic crystals with an average size distribution of 300–500 nm (Figure S1). The crystal phase structure is retained after Fe(II) incorporation (Figure 1b). The XRD patterns of the x-PCN-224(Fe)_5 and 100-PCN-224(Fe)_y samples are all similar to those of the PCN-224, suggesting the influence of the metalation process on the crystal structure of PCN-224 is not significant (Figure S2).

The successful incorporation of Fe²⁺ with the porphyrin unit was confirmed by Fourier-transform infrared spectroscopy (FT-IR), showing the disappearance of the N-H bond vibration in the 100-PCN-224(Fe)_5, which could be observed as the non-coordinated N centers in both the H₂TCPP and PCN-224, and a strong symmetric Fe-N stretching appeared at approximately 1000 cm⁻¹ in PCN-224(Fe) (Figure 1c) [41,44]. The UV-Vis absorption spectra can also demonstrate the change in the coordination sphere of the porphyrin unit, which possesses a Soret band at around 450 nm and four intense Q-bands between 500 and 700 nm [41]. These unique characters could be detected in PCN-224, clearly indicating the well-maintained porphyrin structure after the formation of the MOF with Zr clusters (Figure 1d). The metalation with Fe²⁺ would relatively weaken the intensity of the Q-bands. In addition, the Q-bands would become smoother, even reducing the number of Q-bands to two or three by increasing the x or y values, revealing the symmetry change of the porphyrin molecular structure due to the coordination with Fe²⁺ (Figure 1d and Figure S3). The incorporation of Fe²⁺ with PCN-224 does not significantly change its cubic morphology and size (Figure 1e and Figure S4). However, the surfaces become rough with detectable particles, especially for x-PCN-224(Fe)_y (x > 100 and/or y > 5), probably due to the formation of a Fe₂O₃ nanoparticle resulting from oxidation of iron(II) triflate under higher temperatures (Figure S5). The EDX mapping images of the 100-PCN-224(Fe)_5 demonstrate the even distribution of C, N, O, Zr, and Fe, suggesting that Fe is uniformly incorporated into the PCN-224 (Figure 1e). The ICP-OES results for 100-PCN-224 (Fe)_5 further confirm the mole ratio of the Zr. Fe is 3.6, close to the theoretical ratio of 4. From these analyses, a complete coordination of Fe²⁺ with the porphyrin unit in 100-PCN-224 (Fe)_5 could be expected. The slightly higher Fe content may derive from the residual iron triflate inside the pore or on the surface of 100-PCN-224 (Fe)_5, which may bind with the SBU Zr₆ clusters or form iron oxide particles due to oxidation.

X-ray photoelectron spectroscopy (XPS) measurements were performed to determine the surface chemical states of the elements. The total survey spectra confirmed the existence of N, O, and C elements in PCN-224 and the addition of Fe into 100-PCN-224(Fe)_5 (Figure 2a). The high-resolution N 1s spectra of PCN-224 can be deconvoluted into pyridinic N (397.98 eV) and pyrrolic N (399.78 eV) [45]. Upon coordination with Fe²⁺, a new peak at 401.18 eV assigned to Fe-N_x could be detected, in addition to the negatively shifted C-NH peak, revealing the successful coordination of iron with the porphyrin ligand (Figure 2b). The C 1s XPS spectra for both PCN-224 and 100-PCN-224(Fe)_5 can be fitted with three peaks around 284.84, 286.54, and 288.94 eV, corresponding to the C-C, C-O-C, and O-C=O of the porphyrin group (Figure 2c) [38,46,47]. For the iron species, two peaks at 710.78 and 724.18 eV in the high-resolution Fe 2p spectrum of 100-PCN-224(Fe)_5 suggested that Fe²⁺ species were dominant in this sample (Figure 2d).

The porous character of the MOF samples was determined by the N₂ adsorption–desorption measurements at 77 K. Both PCN-224 and 100-PCN-224 (Fe)_5 show type IV isotherms (Figure S6) with Brunauer–Emmett–Teller (BET) surface areas of 1572 and 1539 m²·g⁻¹ for PCN-224 and 100-PCN-224 (Fe)_5, respectively. The slightly decreased BET surface area of 100-PCN-224 (Fe)_5 should be ascribed to the introduction of Fe sites occupying a certain space. The above characterizations could support the successful coordination of Fe(II) with the porphyrin unit of PCN-224. The investigation of the O₂ adsorption measurement on 100-PCN-224 (Fe)_5 at 273 K shows a considerable O₂-binding capability on the Fe sites (Figure S7), implying the high possibility of generating an iron–oxygen adduct that functions as a potential oxidant for catalytic reactions.

2.2. Oxidative Aldehyde Deformylation Performance

The oxidative deformylation of aldehydes is one of the uncommon reactions catalyzed by cytochrome P450s, which rationalizes the final step in the aromatization of androgens to estrogens through releasing formaldehyde [47,48]. The as-synthesized 100-PCN-224 (Fe)_5 possesses a heme-like center similar to the active site of P450s and the ability to bind with oxygen, thus being potentially applicable for effectively catalyzing the deformylation of various aldehydes using oxygen as the sole oxidant.

Given the precedential reactions of the metal-porphyrins and other homogeneous nonheme complexes with aldehydes [17,49,50,51], we then investigated the aldehyde deformylation performance of x-PCN-224(Fe)_y using 2-phenylpropionaldehyde (2-PPA) as the model substrate to optimize the reaction conditions (Figure 3A). As shown in Table S1, under room temperature and an oxygen atmosphere with 10 mg of 100-PCN-224(Fe)_5 as the catalyst, 1 mL CH₃CN as the solvent, and 2-PPA (0.075 mmol) as the substrate, acetophenone could be detected as the deformylated product, with a conversion efficiency of 37% in 20 min (Table S1, Entry 1). The screening of various additives, including triethylamine (TEA), acetic acid, and water, revealed TEA as the most efficient additive, showing a conversion of up to 95% under the otherwise same conditions (Table S1, entries 2–4, and Figure S8a). Its optimal usage could be managed at five equivalents to the amount of 2-PPA (Table S1, entries 2 and 5–8, and Figure S8b). When using other solvents instead of acetonitrile, including dichloromethane (DCM), tetrahydrofuran (THF), and trifluoroethanol (TFE), albeit acetophenone could be afforded predominantly, significant decreases in the conversion of 2-PPA were observed (Table S1, entries 2 and 10–12). In conjunction with subsequent mechanistic studies and the results from literature surveys, this could be attributed to the fact that 2-PPA was transformed into a keto-enol tautomer intermediate during the catalytic reaction. The dipolar aprotic solvents are beneficial for the enol form, which is a more active nucleophilic reagent and, thus, could promote the equilibrium shift to the enol type for the subsequent deformylation process [52]. However, in the case of using the TFE with the strongest polarity as the solvent, its acidity can lead to the reaction with the basic TEA, preventing it from participating in the conversion of the reactive intermediates and, thus, reducing the total conversion of the aldehydes. In addition, neither using N₂ instead of O₂ nor without PCN-224(Fe) would yield acetophenone, indicating that the aldehyde deformylation is indeed an aerobic catalytic reaction. Therefore, the optimized reaction conditions are finalized (Figure 3B).

The following experiments were performed under such optimized protocols with specific deviations. The investment of catalytic performances on the x-PCN-224(Fe)_y materials prepared at other metalation temperatures or with different amounts of Fe(CF₃SO₃)₂ reveals the lower conversion behaviors in comparison with that of 100-PCN-224 (Fe)_5 (Figure S9). The reason could be ascribed to the fact that, at lower metalation temperatures or with a smaller amount of iron triflate, the coordination between Fe²⁺ and the porphyrin unit is incomplete. Free porphyrins and coordinatively unsaturated Fe(II) sites result in fewer active catalytic sites and, thus, reduce the catalytic efficiency. On the other hand, higher metalation temperatures or more Fe(CF₃SO₃)₂ can lead to the formation of Fe₂O₃ nanoparticles on the surface or within the pores of PCN-224, leading to the obstruction of mass transfer and reducing catalytic efficiency (Figure S5).

In addition, the stability of the MOF catalyst could reflect its advantage over competing homogeneous catalysts in biomimetic catalysis. The 100-PCN-224 (Fe)_5 could be easily recovered by centrifugation from the reaction mixture and reused consecutively for at least five runs of the oxidative deformylation reaction with no significant decrease in conversion (Figure 3C). PXRD and UV-Vis absorption spectra of the recovered catalyst display similar patterns, with a slightly decreased intensity compared to the fresh one (Figure 3C and Figure S10). Similar morphologies and sizes of the recovered and fresh catalyst from their SEM images also confirmed the high stability of this MOF-based P450s mimic under the standard catalysis conditions (Figure S11).

Several substrates with different C-H bond dissociation energies at the α position of the aldehyde group were selected for the deformylation reaction to investigate their reactivity differences (Figure S12). Under the optimized protocols, similar to the reaction of 100-PCN-224 (Fe)_5 with 2-PPA, all substrates were converted to the corresponding ketones or aldehydes as deformylation products. Interestingly, the catalytic reaction efficiency is found to be closely related to the C^α-H bond dissociation energy (BDE) of the aldehyde group. Those substrates with the relatively lower C^α-H BDEs of the aldehyde group exhibited higher reaction efficiencies (Figure 4 and Figure 5a). Diphenylacetaldehyde with a C^α-H BDE of 75.02 kcal mol⁻¹ could be almost completely converted to benzophenone in 12 min with a turnover frequency (TOF) of 58 h⁻¹. Only 15% of cyclohexanecarboxaldehyde (CCA) with a C^α-H BDE as high as 106.1 kcal mol⁻¹ could be converted to form cyclohexanone, even with a prolonged time of 9 h (Figure 4). However, the rule that the more C-H bonds there are, the better efficiency, which normally happens in C-H activation reactions, seems not to hold in this reaction system. Phenylacetaldehyde has two α-hydrogen atoms, but its conversion efficiency is much less than that of 2-PPA (Figure 4 and Figure 5a). Moreover, those without α-hydrogen atoms could not be deformylated, including benzaldehyde, furfuraldehyde, 5-hydroxymethylfurfural, and 2-phenylpropionic acid (Figure S12, Substrates 6–9). This phenomenon sparked our interest in the mechanistic study of the aldehyde deformylation reaction by the MOF-based P450 mimic.

2.3. Mechanistic Studies

In the catalytic cycle of the P450s and the previous homogeneous biomimetic catalysis system of the aldehyde deformylation reaction, a peroxoiron(III) heme complex has been invoked as an active intermediate, which results from the binding of dioxygen with the heme center as the primary function of P450 monooxygenases. However, the mechanism of the following steps in aldehyde deformylation is still controversial by now, i.e., the nucleophilic attack of the peroxoiron on the carbonyl group to form a peroxo hemiacetal adduct model versus a P450 Compound I formation route via the proton-assisted homolysis of the peroxo hemiacetal intermediate [10,53].

To unravel the mystic reaction pathway in aldehyde deformylation, we first investigated the source of the oxygen in the acetophenone product by carrying out isotope labeling studies. Under an ¹⁸O₂ atmosphere and otherwise the same optimized protocols, the reaction of 100-PCN-224 (Fe)_5 and 2-PPA produced acetophenone containing ¹⁸O predominantly, with only less than 1% of ¹⁶O-containing acetophenone (Figure S13). An ¹⁸O₂ labeling study on other aldehyde substrates displays the same results. This indicates that the oxygen in the deformylated product originates from molecular oxygen. Considering the O₂ adsorption measurement and the previous catalysis performance study that showed no deformylated products in the reaction without a catalyst, the existence of an intermediate generated between the 100-PCN-224 (Fe)_5 catalyst and the O₂ oxidant could be estimated.

To further identify the possible intermediate generated by 100-PCN-224 (Fe)_5 and O₂ during the aldehyde deformylation, in situ diffuse reflectance infrared Fourier-transform spectroscopy (DRIFTS) tests on 100-PCN-224 (Fe)_5 under both argon and oxygen atmospheres were conducted. As shown in Figure 5(Bb), when 100-PCN-224(Fe)_5 was detected in O₂, two new peaks emerged at approximately 805 and 815 cm⁻¹, while two peaks around 985 and 1011 cm⁻¹ were strengthened compared to those measured under argon purging. The reason for the case where the DRIFT spectrum of PCN-224(Fe) in Ar also shows bulging or elevated steps around 985 and 1011 cm⁻¹ could be ascribed to the inevitable exposure to air during the test process. The peak at 985 cm⁻¹ could be attributed to the vibrational absorption of υ_O-O. Notably, this peak is significantly lower than the υ_O-O stretching band at 1195 cm⁻¹ observed for an iron(III)–superoxide species as the Fe-TPP (TPP = tetraphenylporphyrin) and O₂ adduct in the homogeneous systems [54]. Instead, these bands closely align with the O-O vibrations at 801, 984, and 1017 cm⁻¹ for the peroxomanganese(IV) porphyrin complex in PCN-224-Mn^II upon O₂ binding, and the υ_O-O band of manganese peroxide Mn(TPP)O₂ (υ_O-O = 983 cm⁻¹), as well as several other side-on metal–peroxo synthetic models [55,56,57] and in metalloproteins including hemoglobin (815 cm⁻¹) [58]. The band at 805 cm⁻¹ is also consistent with the 806 cm⁻¹ of a side-on peroxide of Fe^IIIOEP(O₂^2-)^- (OEP = octaethylporphyrin) [59]. Therefore, it is convincing that the oxygen binds with the Fe center of PCN-224(Fe) in the form of peroxide. Although in the previously reported case, PCN-224-Fe^II could bind O₂ and give a five-coordinated heme–superoxide complex with sufficient characterizations including single-crystal X-ray diffraction, IR, and Mössbauer spectra [42], it is also pointed out that the binding of PCN-224-Fe^II with O₂ at ambient temperature lacks possibility, which was realized in their latter work on PCN-224-Mn^II, confirming the successful preparation and characterization of the Mn^III–peroxo at ambient temperature [55]. In addition, the TEA additive would provide a reductive condition to drive the conversion of “end-on” Fe^III–superoxo to “side-on” Fe^III–peroxo, similar to that reported in multiple molecular complex models [16]. Therefore, it is rationalized that, under our conditions, an Fe–superoxo intermediate was formed upon O₂ binding to the Fe center and reduced to Fe–peroxo species by the electron source TEA, which is analogous to the NADPH reductase in the P450 catalysis.

Subsequently, to establish whether the mechanism proceeds through a nucleophilic attack on the carbonyl group, 2-methyl-2-phenylpropionaldehyde (2-Me-2-PPA) was chosen as a probe since it contains a methyl group on the α-carbon position of 2-PPA, which could exclude the influence of the α-hydrogen-related mechanism (Equation (1) in Figure 5(Bc)). However, no deformylation products were observed by GC-MS (Figure S14). In addition to the results that other aldehydes without a C^α-H bond gave no deformylation product (Equation (1) in Figure 5(Bc)), the reaction pathway of oxidative aldehyde deformylation by 100-PCN-224 (Fe)_5 via a nucleophilic attack on the carbonyl group could be ruled out. We also verified that the presence of an aldehyde carbonyl group is essential, as neither benzoyl chloride nor 2-phenylpropionic acid could be converted (Equation (2)).

Then, we further investigated the possible pathway initiated by an initial hydrogen atom abstraction, alternatively with precedent studies in homogeneous systems using carbon tetrabromide (CBr₄), which is known as a radical-trapping agent either through substitution or radical-coupling reactions [12]. Under the standard catalysis conditions, the addition of excess CBr₄ resulted in the formation of the α-brominated product of 2-PPA, i.e., 2-phenyl-2-bromo-propionaldehyde, which indeed results from the α-hydrogen atom abstraction of 2-PPA to generate an α-H-missing radical of 2-PPA, followed by a brominating process (Equation (3) in Figure 5(Bc) and Figure S15). Consequently, from those mechanistic studies, it is clear that this P450 mimic-catalyzed aldehyde deformylation begins with an α-hydrogen atom abstraction by the peroxoiron(III) species formed upon O₂ binding with the ferric porphyrin unit in the MOF.

After the α-hydrogen atom abstraction, the peroxoiron(III) species transforms into an iron(III)–hydroperoxo form, which could also be the active intermediate for the aldehyde deformylation. However, we investigated the reactivities for 100-PCN-224(Fe)_5 using H₂O₂ as the oxidant, which potentially could generate hydroperoxo species (Scheme S1). Only 8.4% of the 2-PPA could be deformylated in 30 min with an inferior selectivity of acetophenone (54.8%). However, in the presence of the TEA additive, both the conversion and the product selectivity increased to 36.8% and 88.9%, which could be further enhanced to 90% and 98.2%, respectively, by prolonging the reaction time to 12 h (Table S2). These results largely prevent the possibility of Fe-OOH species as the active intermediate in the catalytic cycle for aldehyde deformylation. Instead, the iron hydroperoxide species may lose its H atom and revert back to the iron peroxo, reacting with the nucleophilic enol isomer at the olefinic carbon to form a hemiacetal species.

To follow the oxygen transportation pathway and discern the role of the oxygen-containing iron species, the ¹⁸O labeling and the potential high-valent metal–oxo radical-trapping experiments are important. Que et al. used triphenylphosphine (PPh₃) as a trapping reagent to confirm the necessity of the generation of the high-valent [Fe^IV(O)(TMC)]²⁺ (TMC = tetramethylcyclam) active species, which resulted in a decreased yield of acetophenone and the formation of Ph₃PO as a product [15]. Similarly, when we performed the trapping experiments without adding 2-PPA, the production of Ph₃PO could not be detected. Meanwhile, under standard reaction conditions, the addition of PPh₃ could significantly reduce the conversion of 2-PPA to 55.1% and simultaneously produce Ph₃PO (Equations (4) and (5), Figures S16 and S17). The ¹⁸O-labeling experiment was also conducted here to confirm that the oxygen in the acetophenone and Ph₃PO products originated from O₂ and was passed down through its binding at the Fe center in PCN-224(Fe). These results indicate that the formation of the Fe(IV)–oxo porphyrin species was intercepted by PPh₃ during the 2-PPA deformylation catalyzed by 100-PCN-224(Fe)_5, which, in other words, suggests the existence of the high-valent iron–oxo species in the catalytic cycle under normal conditions.

Based on the comprehensive characterization, reaction tests, and literature reviews, we propose the following mechanism initiated by an iron–peroxide complex abstracting the α-H of the aldehyde. As illustrated in Figure 5C, the reaction cycle of 100-PCN-224 (Fe)_5 catalyzing 2-PPA deformylation starts from the dormant state PCN-224-Fe^II. Using triethylamine as an electron sacrificial agent to activate oxygen, the PCN-224-Fe^IIIO₂ intermediate containing the active ferric peroxide units could be readily formed, which then abstracts a hydrogen atom from the α-carbon of the aldehyde group in 2-PPA (denoted as “0” in Figure 5C) to generate the PCN-224-Fe^IIIOOH species. Meanwhile, substrate 0 loses a hydrogen atom and converts into the radical form 1. The hydrogen atom re-shuttles between the PCN-224-Fe^IIIOOH species and the substrate. The latter eventually transforms to the enol isomer (2), accompanied by the former reverting to the iron peroxide state PCN-224-Fe^IIIO₂. The enol isomer 2 immediately undergoes nucleophilic attack by the active species PCN-224-Fe^IIIO₂ at the olefinic carbon, leading to the formation of the peroxide hemiacetal species 3 that subsequently undergoes O-O bond homolyses to yield a high-valent iron–oxo species PCN-224-Fe^IV=O and the epoxide 4. The PCN-224-Fe^IV=O species abstracts a hydrogen atom from 4 to generate the PCN-224-Fe^III-OH species, with 4 converting to the radical form 5. Finally, compound 5 combines with the PCN-224-Fe^IIIOH species, releasing a formic acid molecule and yielding acetophenone as the deformylation product 6.

3. Conclusions

In summary, we report the decipherment of the reaction pathway of oxidative aldehyde deformylation using a heterogeneous cytochrome P450 mimic, i.e., PCN-224(Fe), for the first time. The as-synthesized 100-PCN-224(Fe)_5 is capable of catalyzing the deformylation reaction of aldehydes with a broad substrate scope at room temperature using atmospheric O₂ as the sole oxidant and TEA as the electron source. The conversion of 2-PPA could reach 95% within 20 min and be well-maintained for at least five reaction cycles by using the recovered catalyst, representing both high catalytic activity and reusability. In contrast to many well-defined homogeneous biomimetic models based on Fe complexes, the study found that the catalytic efficiencies are closely related to the C^α-H bond dissociation energy of the aldehyde substrates. In situ IR measurement confirms the formation of an iron(III)–peroxo intermediate during the catalytic cycle. Taking together the various trapping and isotopic labeling results, an unconventional reaction pathway initiated by the Fe–peroxo species abstracting the α-hydrogen atom of aldehydes could be rationalized. This research may provide new insights into the biological deformylation process and a strategy for the development of heterogeneous biomimetic materials for aerobic catalytic transformations.