Chemical Engineering Science 239 (2021) 116602 Contents lists available at ScienceDirect Chemical Engineering Science journal homepage: www.elsevier.com/locate/ces Multi-reaction kinetic modeling for the peroxidase–aldolase cascade synthesis of a D-fagomine precursor Gerard Masdeu a,⇑, Zvjezdana Findrik Blažević b, Slavko Kralj c,d, Darko Makovec c, Josep López-Santín a, Gregorio Álvaro a a Department of Chemical, Biological and Environmental Engineering, Universitat Autònoma de Barcelona, Bellaterra, Spain b Faculty of Chemical Engineering and Technology, University of Zagreb, Zagreb, Croatia c Department for Materials Synthesis, Jožef Stefan Institute, Ljubljana, Slovenia d Faculty of Pharmacy, University of Ljubljana, Ljubljana, Slovenia h i g h l i g h t s g r a p h i c a l a b s t r a c t  Multi-enzymatic reaction cascade evaluation for the synthesis of iminocyclitols.  A complex system composed of seven distinct reaction and two enzyme inactivation.  Kinetic modeling for elucidation of reaction mechanism.  Cross-inhibition and inactivation in peroxidase/aldolase reactions.  High improvement in reaction productivity through process intensification. a r t i c l e i n f o a b s t r a c t Article history: The feasibility of a peroxidase–aldolase cascade reaction for the synthesis of therapeutically-valuable Received 20 February 2021 iminocyclitols is discussed herein. A two-enzyme system consisting of chloroperoxidase (CPO) and D- Accepted 19 March 2021 fructose-6-phosphate aldolase (FSA) was evaluated for the synthesis of a D-fagomine precursor Available online 22 March 2021 (preFagomine) from a N-Cbz-3-aminopropanol. An in-depth, systematic, step-by-step kinetic modeling of seven reactions and two inactivation decays was proposed to elucidate the reaction mechanism, pre- Keywords: pare suitable stabilized biocatalysts, and find the optimal conditions for its application. The model Enzymatic reaction cascade described accurately the data and predicted the outcome at different experimental conditions. The inac- Peroxidase Aldolase tivation of FSA caused by CPO was identified as the main bottleneck in the reaction. A two-step reaction D-fagomine approach and the use of immobilized enzymes on magnetic nanoparticle clusters and functionalized Kinetic modeling agarose carriers increased the stability of FSA, with an 1839-fold higher preFagomine formation per mol of enzyme in comparison to a one-pot reaction using soluble enzymes. Ó 2021 Elsevier Ltd. All rights reserved. Abbreviations: CPO, chloroperoxidase; FSA, D-fructose-6-phosphate aldolase; 1. Introduction preFagomine, (3S,4R)-6-[(benzyloxycarbonyl)amino]-5,6-dideoxyhex-2-ulose); DHA, dihydroxyacetone; b-OH, N-Cbz-3-aminopropanol; b-CHO, N-Cbz-3- aminopropanal; b-COOH, N-Cbz-3-aminopropanoic acid; t-BuOOH, t-butyl Iminocyclitols, also referred to as iminosugars, are carbohydrate hydroperoxide; mNC, magnetic nanoparticle clusters; MANA-agarose, analogs in which a nitrogen atom has replaced the endocyclic oxy- Monoaminoethyl-N-aminoethyl agarose; CoIDA-agarose, cobalt-iminodiacetic gen. These compounds are inhibitors of glycosidases with enor- acid-agarose. ⇑ Corresponding author. mous therapeutic potential in many diseases by altering the E-mail address: gemasdeu@gmail.com (G. Masdeu). glycosylation or catabolism of glycoproteins (Calveras et al., https://doi.org/10.1016/j.ces.2021.116602 0009-2509/Ó 2021 Elsevier Ltd. All rights reserved. G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 2009; Gloster et al., 2007; Stütz, 1999; Whalen and Wong, 2006). netic nanoparticle clusters has been previously reported showing One valuable iminocyclitol is D-fagomine (Gómez et al., 2012; increased stability with corresponding higher conversion rates: Hereu et al., 2019; Koyama and Sakamura, 1974; Pearson et al., CPO was immobilized on amino-functionalized agarose (Pešić 2005), which is a nutraceutical that may be used in diabetes cases et al., 2012) and magnetic nanoparticle clusters (mNC) (Masdeu to prevent the digestion and absorption of carbohydrates (Fan et al., 2018), and FSA has been recently attached to both cobalt- et al., 1999; Nojima et al., 1998; Taniguchi et al., 1998). Chemical iminodiacetic acid-agarose (CoIDA-agarose) and mNC (Masdeu syntheses of D-fagomine with carbohydrate precursors as starting et al., 2021). materials usually involve cumbersome protection-deprotection The evaluation of the CPO/FSA reaction will be addressed reactions, achieving moderate isolated yields (Fleet et al., 1987; through (i) kinetic modelling and (ii) process intensification. (i) Goujon et al., 2005; Kumari et al., 2009; Pandey and Kapur, The development of a mathematical kinetic model of the system 2000); the synthesis from amino aldehydes is also described with enables process optimization and give insight into the relationship final yields below 15% (Banba et al., 2001). Alternatively, a D- among the process variables: degradation of peroxide, chemical fagomine precursor (preFagomine, (3S,4R)-6-[(benzyloxycarbo reaction, amino aldehyde oxidation, alcohol oxidation, DHA oxida- nyl)amino]-5,6-dideoxyhex-2-ulose) has been synthesized with tion, aldol addition, and retroaldol addition. (ii) The optimal oper- 89% of isolated yield through stereoselective aldol addition of dihy- ating strategy of a complex system is approximated to maximize droxyacetone (DHA) to N-Cbz-3-aminopropanal (b-CHO) (Fig. 1b) the reaction rates and product yield, while minimizing the forma- (Castillo et al., 2006). The aldol addition was catalyzed by D- tion of side products. In the case of a multi-enzyme system, a fructose-6-phosphate aldolase (FSA, EC 4.1.2.-), a class I aldolase decomposition of the whole system on individual reaction steps from E. coli that is not dependent on phosphorylated donor sub- is required for the model formulation and understanding of each strates. This substrate promiscuity has allowed many syntheses individual reaction separately. of iminocyclitols, which have been assayed against a panel of gly- cosidases (Herna et al., 2019; Sugiyama et al., 2007). Several FSA 2. Materials and methods genetic variants with modified substrate specificity are described elsewhere (Gutierrez et al., 2011; Szekrenyi et al., 2014). 2.1. Materials The chemical synthesis of the protected amino aldehydes is generally achieved with low selectivity and high cost of oxidizing b-OH, b-CHO, N-Cbz-3-aminopropanoic acid (b-COOH), t- agents. Transition metals are commonly used in these reactions, BuOOH (70 wt% in H2O), H2O2 (30 wt% in H2O), 2,4- making the process environmentally unsuitable (Lenoir, 2006). Dinitrophenylhydrazine phosphoric acid solution (DNPH), a- The oxidation of alcohols to aldehydes is hence advantageous to glycerophosphate dehydrogenase–triosephosphate isomerase be performed via an enzymatic process. In previous work, the oxi- (GPD/TPI), fructose-6-phosphate (F6P), D-glyceraldehyde-3- dation of N-Cbz-3-aminopropanol (b-OH). to b-CHO was catalyzed phosphate (G3P), NADH, N-(3-dimethylaminopropyl)-N’- by chloroperoxidase from Caldariomyces fumago (CPO, EC ethylcarbodiimide (CDI), (3-aminopropyl)triethoxysilane, and 1.11.1.10), with nearly full substrate conversion but only 18% ethylenediamine were purchased from Sigma Aldrich. 1,1-dime selectivity towards the aldehyde, which could be increased by cou- thyl-4-chloro-3,5-cyclohexadione (MCD) (Fluka), (Hydroxy(polye pling the FSA reaction (Fig. 1a) (Masdeu et al., 2016a). A peroxide thyleneoxy)propyl) triethoxysilane (Gelest), 10% cross-linked agar- (hydrogen peroxide or t-butyl hydroperoxide, t-BuOOH) was used ose beads and high-density metal-chelated supports (20–40 mmol as a co-substrate, instead of the expensive cofactors required by divalent metal ml1) with iminodiacetic acid for metal chelation dehydrogenases (Andersson and Wolfenden, 1982; Rodriguez- (IDA-agarose) (Agarose Bead Technologies) were used. Iron (III) Hinestroza et al., 2017; Shaw and Hager, 1961; Sudar et al., sulfate hydrate, iron (II) sulfate heptahydrate, citric acid, 2015). Peroxides or derivate radical species, however, may cause tetraethoxysilane (TEOS), NH4OH, and DHA were supplied by Alfa a rapid inactivation of CPO probably by heme group destruction Aesar. (Andersson et al., 2000; Ayala et al., 2011; Chamulitrat et al., CPO from C. fumago was purchased from Chirazyme Labs (42 1989; Park and Clark, 2006; Shevelkova and Ryabov, 1996), simi- KDa, 1400 U mg1 protein). Recombinant His-tagged FSA A129S (se- larly to the activity loss of dehydrogenases initiated by the supply lected due to its high reactivity towards DHA) (Castillo et al., of oxygen in the reactions (Sudar et al., 2021). 2010; Sudar et al., 2013b) was produced in high-cell density fed- The aim of the present research is the evaluation of the feasibil- batch Escherichia coli BL21 (DE3) cell cultures by a modified proto- ity of a two-reaction coupling employing CPO and FSA for the syn- col described elsewhere (Calleja et al., 2016). Plasmid pET22-fsaA thesis of preFagomine using b-OH, DHA and peroxide as starting was generously supplied by Dr. Clapés from Biotransformation materials. The elucidation of the reaction mechanism of both reac- and Bioactive Molecules Group (IQAC-CSIC, Barcelona). The tions is crucial for the determination of the initial reaction condi- enzyme was purified by (cobalt-chelated) affinity chromatography tions for its application: the reaction buffer and pH, the type of with a specific activity was 15.0 U mg1protein (257 KDa). peroxide, and the concentration of the substrates. Enzyme stability, side reactions (e.g. the formation of hydroxy peroxy intermediates) 2.2. Preparation of immobilization carriers (Masdeu et al., 2016b) and substrate/product inhibitions may also alter both CPO and FSA catalytic rates (Robertson, 2005). The Functionalized agarose supports were synthesized by previ- immobilization of both enzymes into agarose particles and mag- ously described protocols: Monoaminoethyl-N-aminoethyl Fig. 1. Schematic representation of the enzymatic synthesis of preFagomine: (a) b-OH oxidation catalyzed by CPO using t-BuOOH. (b) DHA aldol addition to b-CHO catalyzed by FSA. 2 G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 (MANA) agarose was prepared from 10% cross-linked agarose gels static interactions. As designed for reaction purposes, both deriva- (Fernandez-Lafuente et al., 1993; Guisán, 1988), and CoIDA- tives had a final activity of 6.25 U mg1 (CPO-MANA-agarose and agarose was prepared from IDA-agarose gels using cobalt chloride mNC-NH2-CPO retained activities of 40.9 and 62.5%, respectively). (Ardao et al., 2006). The preparations of FSA-CoIDA-agarose and mNC-NH2-FSA are mNC was obtained by the self-assembly of primary maghemite fully described in our recent work (Masdeu et al., 2021). FSA (c-Fe2O3) nanoparticles followed by a coating of the magnetic clus- (0.60 U) was immobilized on CoIDA-agarose (50 mg) for 30 min ter with a layer of silica (Kralj et al., 2010; Kralj and Makovec, in 50 mM sodium phosphate buffer with 0.3 M NaCl and 20 mM 2015; Tadic et al., 2014). To functionalize and ensure the colloidal imidazole, at pH 8.0 and 25 °C in a total volume of 1 ml. In parallel, stability of the nanocomposite, silica-coated mNC was grafted with 2.0 U FSA was immobilized onto 50 mg mNC-NH2 in 10 mM phos- a mixed layer of silane molecules: ~83 mol. % (3-aminopropyl)trie phate buffer (pH 5.0), at 25 °C, and stirred at 1000 rpm. After 1- thoxysilane, and ~17 mol. % of the silane-PEG for the additional hour incubation, 5 mM CDI was added for 3 h to form the covalent steric stabilization of suspensions (Kralj et al., 2011; Masdeu bond between FSA and mNC-NH2. Finally, 0.5 M NaCl was used to et al., 2018). Functionalized mNC are fully characterized and also avoid non-covalent interactions. For reaction purposes, both commercially available by Nanos SCI company under trademark derivatives had a final activity of 11.6 mU mg1 (FSA-CoIDA- iNANOvativeTM. agarose and mNC-NH2-FSA retained activities of 96.4 and 29.1%, respectively). 2.3. Enzymatic activity assays Final immobilized derivatives were washed with buffer, the vol- ume was set to 0.3 ml, and they were immediately used for the The activity of CPO was spectrophotometrically determined by reaction. the chlorination of MCD (e278 = 12.2 mM1 cm1) (Toti et al., 2005) to dichlorodimedone (DCD), according to the method of 2.5. HPLC analysis Hager and Morris (Morris and Hager, 1966). The enzymatic assay contained 0.1 M potassium phosphate buffer of pH 2.75, b-OH, b-CHO, b-COOH, t-BuOOH, and preFagomine were quan- 0.16 mM MCD, 20 mM KCl, 2 mM hydrogen peroxide, and 50 mL tified by HPLC (Dionex UltiMate 3000 with variable wavelength of enzyme sample in a total volume of 1 ml (100 mL in 2 ml for detector) analysis using a CORTECS (Waters) C18 + 4.6  150 mm immobilized derivatives). The absorbance at 278 nm was moni- (2.7 lm) column. Reaction samples were dissolved in acetonitrile tored at 25 °C using a Cary 50 Bio UV–Visible Spectrophotometer (MeCN) and analyzed by injecting 15 lL at a flow rate of (Varian) by duplicate measurements. One activity unit (U) of CPO 0.7 ml min1, 30 °C. The solvent system consisted of mobile phase is defined as the amount of enzyme required for the conversion A –0.1% (v/v) TFA in H2O– and B –0.095% (v/v) TFA in MeCN/H2O of 1 lmol of MCD per minute. 4:1 (v/v)–. Samples were eluted using a gradient from 5 to 28.5% The activity of FSA was determined by a three-enzyme reaction B in 0.5 min, and an isocratic elution at 28.5% B over 15 min described in the literature (Schürmann and Sprenger, 2001). The (k = 254 nm). For the quantification of DHA, sample derivatization mixture for the enzymatic assay contained 5 mM F6P, 0.1 mM was required before chromatography. DNPH derivatization was NADH, 50 mM imidazole, 10 U ml1 of GPD-TPI, and 50 mL of the performed following a described protocol (Ferioli et al., 1995) sample with FSA, in a total volume of 1 ml (100 mL in 2 ml volume 30 lL of DHA (0–2 mM) was mixed with 30 lL of a stock solution for immobilized derivatives). FSA activity was monitored by the of DNPH (0.2 M) for 60 min; 180 lL of sodium acetate (3 M, pH 9.0) decrease at A340 due to the conversion of NADH to NAD+ was added to stop the reaction. The derivatized compound was (e340 = 6.2 mM1 cm1) at 30 °C by duplicate measurements. 1 U extracted into an organic phase using 240 lL of MeCN, and the of FSA activity is defined as the amount of enzyme required for HPLC analysis was performed eluting the samples with an isocratic the conversion of 1 lmol of F6P per minute. In the case of FSA method at 50% B for 2 min, followed by a gradient from 50 to 65% B bound to mNC-FSA, the activity was not quantified by absorption over 5 min (k = 260 nm). Prior calibration with standards of known kinetics due to method sensitivity but using the aldol addition of concentration was used for quantitative analysis. The standard DHA to b-CHO monitoring preFagomine by HPLC analysis deviation was calculated from measurements performed by dupli- (Masdeu et al., 2021). The immobilized enzyme (50 mL) was added cate. In addition, the structures of preFagomine and oxidized DHA to 950 mL of the reaction mixture, containing 40 mM b-CHO and (oxDHA) were characterized by HPLC-MS analysis (Figure S1, S2). 100 mM DHA in 50 mM HEPES (pH 8.0). The reaction was left for 90 min at 30 °C, 1000 rpm, and stopped by acidification to pH 2.5. 2.6. Mathematical model and kinetic parameters estimation 2.4. Enzyme immobilization Reaction rates (r) were named according to the corresponding reaction (R) (Eqs. (1)–(19); nomenclature is revised in Supporting Native CPO was immobilized onto MANA-agarose (CPO-MANA- Information, SI section 3.1): R1 is the CPO-catalyzed b-OH oxida- agarose) according to our previous method (Pešić et al., 2012). tion to b-CHO, which can be further oxidized to b-COOH by the Briefly, MANA-agarose (50 mg) was mixed with CPO (764 U) in same enzyme and peroxide (R2). R3 represents the chemical reac- 1 ml of 10 mM sodium phosphate buffer (pH 5.0) at 25 °C for tion aldehyde-peroxide leading to the hydroxy alkyl peroxide 15 min, and CDI (50 mM) was added for covalent immobilization. (HAP) described in previous work (Masdeu et al., 2016a). R4 is After 3 h, 1 M NaCl was incubated for an additional hour to avoid the degradation of t-BuOOH by CPO that renders tert-butyl alcohol the ionic interactions. CPO was also immobilized onto mNC-NH2 (t-BuOH) and a radical (t-BuOO), as previously reported elsewhere (mNC-NH2-CPO) by prior modification of the surface of the enzyme (Chamulitrat et al., 1989). R5 states for the oxidation of DHA cat- (Masdeu et al., 2018). The sugar moieties on the surface of the alyzed by CPO leading to oxDHA. R6 and R7 are the aldol and enzyme (232 U ml1) were oxidized with 25 mM NaIO4 in 0.5 ml retroaldol additions by FSA. of 50 mM phosphate buffer (pH 5.0) for 60 min. 50 mM propylene The enzyme inactivation was simplified to a first-order decay. glycol was added to terminate the reaction. A colloidal suspension kdCPO is the inactivation rate constant of CPO in presence of t- of mNC-NH2 (50 mg) was mixed TM) with oxidized CPO (500 U) in BuOOH; kdFSA1 is the inactivation rate of FSA caused by CPO and 1 ml of 10 mM phosphate buffer (pH 5.0) at 25 °C for 2 h, followed t-BuOOH; kdFSA2 is the FSA inactivation rate caused by CPO, DHA by the addition of NaBH4 (0.1 mg ml1) to reduce the formed Schiff and t-BuOOH. The stoichiometry and mass balances from all reac- base for 30 min. 1 M NaCl was added to avoid non-specific electro- tions are: 3 G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 R1 : bOH þ tBuOOH ! bCHO þ tBuOH ð1Þ simplex and least squares fit regression methods implemented in CPO SCIENTIST software were used for parameter estimation and model R2 : bCHO þ tBuOOH ! bCOOH þ tBuOH ð2Þ simulations. The software gives statistical output for the estima- CPO tion of parameters, as well as goodness-of-fit statistics for the mathematical models. R3 : bCHO þ tBuOOH ! HAP ð3Þ R4 : 3 tBuOOH ! tBuOH þ 2 tBuOO ð4Þ 2.7. Process intensification CPO R5 : DHA þ tBuOOH ! ox DHA þ tBuOH ð5Þ All reactions were carried out in 0.1 M MES buffer (pH 6.5) at CPO 25 °C, 1000 rpm with withdrawn volume below 10%, as follows: (a) The one-pot reaction was carried out in 1 ml at the following R6 : DHA þ bCHO ! preFagomine ð6Þ initial concentrations: 16 mM b-OH, 39.9 mM t-BuOOH, 72.6 mM FSA DHA, 5.0 kU CPO ml1 and 0.2 kU FSA ml1. (b) The initial two- R7 : preFagomine ! DHA þ bCHO ð7Þ step reaction mixture was 1 ml of 5.0 mM b-OH, 10.0 mM t- FSA BuOOH, and 312.5 U CPO ml1 for the b-OH oxidation, and 2 ml of 0.5 mM b-CHO, 20.0 mM DHA, and 0.58 U FSA ml1 for the aldol d½bOH ¼ r1 ð8Þ addition. After the oxidation, the reaction medium was separated dt by a magnetic field (mNC) or filtration (agarose), and ultrafiltration (soluble enzyme, all cases). Finally, it was left under mild agitation d½bCHO ¼ r1  r2  r3  r6 þ r7 ð9Þ for 60 min before starting the second reaction, to degrade all the dt peroxy radicals. d½bCOOH ¼ r2 ð10Þ dt 3. Results d½HAP ¼ r3 ð11Þ A preliminary coupling of the CPO-catalyzed b-OH oxidation dt and the FSA-catalyzed DHA addition using soluble enzymes was d½tBuOOH investigated to choose the proper peroxide and buffer for the cou- ¼ r1  r2  r3  3 r4  r5 ð12Þ pled reaction (SI section 2). Apart from the well-known inactiva- dt tion of CPO caused by the peroxide, the reaction between b-CHO d½tBuOH and the peroxide, and the oxidation of b-CHO to the acid, the main ¼ r1 þ r2 þ r4 þ r5 ð13Þ dt insight from this pre-analysis was: (a) t-BuOOH led to a notably higher alcohol oxidation rate than hydrogen peroxide. (b) The d½tBuOO  selected buffer was 100 mM MES buffer (pH 6.5) with high opera- ¼ 2 r4 ð14Þ dt tional stability of both enzymes. (c) FSA was inactivated by incuba- tion with CPO and t-BuOOH. (d) No preFagomine was obtained in a d½DHA proof-of-concept reaction, probably due to a fast inactivation of ¼ r5  r6 þ r7 ð15Þ dt FSA. (e) Various side reactions were identified on the complex sys- tem: the oxidation of DHA in presence of t-BuOOH and CPO –which d½ ox DHA competes with FSA for the use of DHA–, and the degradation of t- ¼ r5 ð16Þ dt BuOOH catalyzed by CPO. Fig. 2 includes the five side reactions and the inactivation of d½preFagomine both enzymes within the whole reaction. All these undesirable ¼ r6  r7 ð17Þ dt phenomena hindered the simultaneous coupling of these two reac- tions. To overcome this issue, the development of a mathematical dðCPOactivityÞ kinetic model was proposed as the work strategy. In contrast to a ¼ kdCPO  CPOactivity ð18Þ dt factorial design, a kinetic model might be able to relate the direct dependencies and effects of each reaction. To that end, all reaction dðFSAactivityÞ ¼ ðkdFSA1 þ kdFSA2 Þ  FSAactivity ð19Þ and inactivation rates were intended to be individually investi- dt gated to reduce the disturbing effects of the other reactions, such All reactions were carried out in 1 ml of 0.1 M MES buffer (pH as substrate or product inhibitions. 6.5) at 25 °C, and stirred at 1000 rpm. The total withdrawn sam- A separate kinetic model for each enzyme reaction was built. pling volume was always kept below 10%. The concentrations of The model for CPO-catalyzed reactions was investigated by a step- substrates and enzymes are specified in the corresponding text, wise method: (1) Incubation of CPO and t-BuOOH to determine the figures and tables in SI section 3. inactivation of CPO (kdCPO). Here, R4 was also present. (2) Incuba- The estimation of model parameters was performed by progress tion of t-BuOOH and b-CHO to evaluate the chemical reaction (R3) curve analysis using PSE gPROMSÒ Model Builder 5.0.0, which that rendered the hydroxy alkyl peroxide (HAP). (3) Addition of b- allows the simultaneous determination of multiple parameters CHO to step 1 for the interpretation of the oxidation of b-CHO to b- from several experiments. The optimization algorithm of the soft- COOH. kdCPO, r4 and r3 were also considered. (4) Addition of b-OH ware was used to maximize the formation of preFagomine in sim- to previous steps to investigate the alcohol oxidation. Regarding ulated CPO/FSA reactions. Parameter Estimation in gPROMS was the model for FSA-catalyzed reactions, it included two equilibrium based on the Maximum Likelihood formulation which provides reactions: R6 and R7. After its elucidation, both CPO/FSA systems simultaneous estimation of parameters in both the physical model were put together to study the oxidation of DHA by CPO and t- of the process and the variance model. MicroMathÒ ScientistÒ 2.0 BuOOH, and the inactivation of FSA (kdFSA) caused by the products was exclusively used for initial reaction rate analysis. Non-linear from R4 and R5. 4 G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 Fig. 2. Overall reaction mechanism including the identified side-reactions and the inactivation pathways for both enzymes. 3.1. CPO model Table 1 Estimated values of the parameters in Equations (20)–(37). 95% confidence intervals 3.1.1. Degradation of t-BuOOH and CPO inactivation are indicated. r4 (mM min1) and kdCPO (min1) were determined by the incu- Parameter Value Units bation of t-BuOOH and CPO (experimental distribution in Fig- t-BuOOH degradation, CPO inactivation ure S5). A study of the initial reaction rates was discarded due to kcat4 2.07 ± 0.10 lmol  min1  mgCPO1 the peroxide-dependent fast inactivation of CPO. Instead, a pro- k4 51.54 ± 3.40 mM gress curve analysis of each reaction was performed in terms of n 1.43 ± 0.02 – ka 7.04102 ± 0.17102 min1 peroxide concentration and CPO activity over the reaction time. 2 a 1.8310 ± 0.20102 mM  min1 A sigmoidal behavior was observed for r4: chemical reaction aldehyde–peroxide k3i 1.49103 ± 4105 mM1 min1 kcat4  YCPO  ekdCPO t  ½tBuOOHn k3ii 1.25101 ± 5103 min1 r4 ¼ n ð20Þ kir3 2.18101 ± 1.4102 – k4 þ ½tBuOOHn CPO-catalyzed reactions where YCPO corresponds to the CPO mass concentration (mg kcat2 3.42 ± 0.29 lmol  min1  mgCPO1 KM,t-BuOOH 58.83 ± 5.18 mM ml ), kcat4 is the turnover number in R4 (mmol min1 mg1), 1 KM,b-CHO 36.38 ± 3.84 mM whereas the inactivation of CPO (kdCPO) was expressed as a func- kir4a 6.52 ± 0.04 mM tion of r4; i.e. kdCPO is proportional to the formation rate of the per- kcat1 10.88 ± 0.37 lmol  min1  mgCPO1 oxyl radical t-BuOO (r4) (Chamulitrat et al., 1989; Park and Clark, KM,b-OH 309.5 ± 12.0 mM kcat5 54.18 ± 4.11 lmol  min1  mgCPO1 2006): KM,DHA 833.6 ± 71.2 mM kir4b 11.26 ± 0.64 mM ka  r4 kir1 108.2 ± 8.5 mM kdCPO ¼ ð21Þ a þ r4 kir2 38.21 ± 8.08 mM kir5 32.15 ± 0.76 mM FSA-catalyzed reactions where ka is the maximum value of the CPO inactivation rate con- kcat6 19.31 ± 1.00 lmol  min1  mgFSA1 stant (min1), and a is an inactivation parameter (mM min1). This kcat7 13.06 ± 0.69 lmol  min1  mgFSA1 radical was assumed to be degraded in seconds, considering its ’ KM,DHA 9.00 ± 1.70 mM ’ accumulation in the reactor as negligible. The estimated values of KM,b-CHO 20.94 ± 2.25 mM all parameters are indicated in Table 1; Fig. 3 (3b in this case) shows KM,preFagomine 59.68 ± 1.93 mM kir7 3.09101 ± 5.4102 mM some examples of the fitting to all the equations in the text (for all FSA inactivation fittings, see Figure S6). k1 1.96101 ± 7103 mM1 Regarding inhibitions affecting r4 or modulations over kdCPO, it k2 6.70 ± 0.24 mM1 was not checked with b-OH, b-CHO or DHA at this step, as they k02 33.75 ± 4.20 min1 are substrates of the same enzyme and consequently another reac- b 3.45 ± 0.41 mM  min1 c 3.85101 ± 7103 – tion (R1, R2, R5) could interfere with the result. Inhibition/modula- d 3.85102 ± 1.42102 – tion was indeed evaluated with various concentrations of b-COOH and preFagomine, but no effect was observed (Figure S7). (Fig. 3C, S9). Results indicated a classic equilibrium behavior from a reversible reaction: 3.1.2. Chemical reaction aldehyde–peroxide To describe the rate of the chemical aldehyde–peroxide reaction r3 ¼ k3i  ½bCHO  ½tBuOOH  k3ii  ½HAP ð22Þ (r3, mM min1), various concentrations of both compounds were mixed and left for incubation (Figure S8). Each reaction was ana- where k3i is the forward chemical reaction constant (mM min1), 1 lyzed in terms of concentrations of b-CHO and t-BuOOH over time k3ii is the reverse reaction constant (min1). b-OH, b-COOH, t-BuOH, 5 G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 Fig. 3. Stepwise methodology for the kinetic model in the CPO/FSA reaction system. FSA-catalyzed reactions: (a) Initial reaction rates on R6 and R7; CPO-catalyzed reactions: (b) R4 and CPO inactivation caused by t-BuOO, (c) R3, (d) R2, R3, R4 and CPO inactivation, (e) R1, R2, R3, R4 and CPO inactivation, (f) R4, R5 and CPO inactivation. Experimental data: b-OH (s), b-CHO (j), b-COOH (+), t-BuOOH (), DHA (d), CPO activity (N); model: b-OH ( ), b-CHO ( ), b-COOH ( ), DHA ( ), t-BuOOH ( ), CPO activity ( ). DHA, preFagomine were tested as inhibitors for this reaction. Only where KM,X is the Michaelis constant of compound X. The initial inhibition caused by b-COOH was observed (Figure S10), rewriting reactions rates r1, r2 and r5 were analyzed together with KM,t- Eq. (22) with an extra inhibition term (kir3) (Figure S11): BuOOH as a global parameter (Figure S12, Table S2). Regarding the CPO-oxidation of b-CHO into b-COOH, four reac- r3 ¼ k3i  ½bCHO  ½tBuOOH  ð1 þ ½bCOOHÞkir3  k3ii tions were studied in a mixture of CPO, b-CHO, t-BuOOH: degrada-  ½HAP ð23Þ tion of the peroxide (r4), inactivation of CPO (kdCPO), the chemical reaction (r3), and b-CHO oxidation (r2) (Eqs. (20), (21), (23), (25)). Several reactions with different concentrations of aldehyde and 3.1.3. CPO-catalyzed oxidations peroxide (Figure S13) were carried out to determine the actual Many reported reactions catalyzed by CPO exhibit an undeni- value of the pre-estimated r2 parameters (Fig. 3D, S14). An inhibi- able ping-pong mechanism (Casella et al., 1994; Colonna et al., tion mechanism in r4 caused by the amino aldehyde was observed 1999; Dunford, 2010; Kedderis and Hollenberg, 1983). The estima- from the experimental data, adding a new term (kir4a) to Eq. (20): tion of the kinetic parameters in these systems can be hindered by the tight relation between them. To overcome this limitation, a kcat4 o YCPO o e-kdCPO t o ½BuOOHn r4 ¼ n ½bCHO ð27Þ preliminary study of initial reaction rates was carried out to find k4 oð1 þ Þ þ ½BuOOHn kir4a an initial value of each parameter to be used for model calcula- tions, according to: The best fitting results were obtained considering a competitive kdCPO t inhibition of b-CHO to the binding of the peroxide at the active site. kcat1  YCPO  e  ½bOH  ½tBuOOH r1 ¼ The inhibition from other compounds of the reaction system, such KM;bOH  ½tBuOOH þ KM;tBuOOH  ½bOH þ ½tBuOOH  ½bOH as b-COOH, t-BuOH, and preFagomine on r2 was evaluated, but no ð24Þ inhibition was observed in any case (Figure S15). To investigate the CPO-catalyzed oxidation of b-OH into b-CHO, kcat2  YCPO  ekdCPO t  ½bCHO  ½tBuOOH several reactions with different concentrations of t-BuOOH and b- r2 ¼ KM;bCHO  ½tBuOOH þ KM;tBuOOH  ½bCHO þ ½tBuOOH  ½bCHO OH were carried out (Figure S16). Apart from r1 (Eq. (24)), other ð25Þ rates were considered: r2, r3, r4, and kdCPO (Fig. 3E, S17). Regarding inhibitions affecting r1, it was not evaluated with b-COOH and pre- kcat5  YCPO  ekdCPO t  ½DHA  ½tBuOOH Fagomine, since no inhibitory effect was observed for r2. r5 ¼ Reactions with different peroxide and DHA concentrations were KM;DHA  ½tBuOOH þ KM;tBuOOH  ½DHA þ ½tBuOOH  ½DHA carried out (Figure S18) to estimate the kinetic parameters from ð26Þ Eq. (26). r4 and CPO inactivation should be also considered apart 6 G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 from r5 (Fig. 3F, S19). An inhibition mechanism in r4 caused by kdFSA1 ¼ k1 o r4 ð35Þ DHA was observed from the experimental data, adding an inhibi- Regarding the second inactivation (kdFSA2), incubations of FSA tion term kir4b: and CPO with different concentrations of t-BuOOH and DHA (Fig- kcat4  YCPO  ekdCPO t  ½tBuOOHn ure S25) were carried out. Again, the inactivation was described r4 ¼ ð28Þ with a linear expression, not related to r4 but to the oxidation of k4  ð1 þ ½bCHO þ ½DHA n kir4a ki Þ þ ½tBuOOHn r4b DHA by CPO (r5) (Figure S26). The best results were obtained considering a competitive inhi- kdFSA2 ¼ k2 o r5 ð36Þ bition of DHA to the binding of the peroxide at the active site. To check for possible interactions of r5 with r1 or r2, five reac- Additional series of experiments were performed to confirm the tions were performed with a fixed concentration of CPO, b-OH, obtained kinetic values for the inactivation of FSA, mixing different and t-BuOOH, but a variable amount of DHA (Figures S20). Three concentrations of b-OH, DHA, t-BuOOH, CPO and FSA (Table S3). inhibition terms were added: two competitive inhibitions of DHA The fittings from these experiments (Figure S27) indicated: (1) to the binding of b-OH or b-CHO to the active site (kir1, kir2), and kdFSA1 was confirmed to be a linear function of r4, (2) kdFSA2 was the competitive inhibition of b-OH to the binding of DHA (kir5): not a linear function of r5, but a hyperbolic function as already described for FSA reactions (Česnik et al., 2019), (3) an increase kcat1  YCPO  ekdCPO t  ½bOH  ½tBuOOH r1 ¼ ½DHA of YFSA acts as a positive modulator effect on FSA activity, lowering KM;bOH  ð1 þ Þ  ½tBuOOH þ KM;tBuOOH  ½bOH þ ½tBuOOH  ½bOH kir1 kdFSA2; (4) (an increase of the DHA concentration leads a higher r5, ð29Þ which produces a more rapid inactivation of FSA. However, DHA is a natural substrate of FSA and may protect the active site of FSA kcat2  YCPO  ekdCPO t  ½bCHO  ½tBuOOH from inactivation: r2 ¼ ½DHA KM;bCHO  ð1 þ kir2 Þ  ½tBuOOH þ KM;tBuOOH  ½bCHO þ ½tBuOOH  ½bCHO 0 ð30Þ k2  r5 kdFSA2 ¼  ð1 þ YFSA Þc  ð1 þ ½DHAÞd ð37Þ b þ r5 kdCPO t kcat5  YCPO  e  ½DHA  ½tBuOOH r5 ¼ KM;DHA  ð1 þ ½bOH kir5 Þ  ½tBuOOH þ KM;tBuOOH  ½DHA þ ½tBuOOH  ½DHA 3.3. Model validation ð31Þ For the validation of all estimated kinetic parameters, several 3.2. FSA model reactions were carried out fixing the concentration of b-OH, t- BuOOH and CPO at high values (47.0 mM, 77.7 mM, 0.714 mg mL1) FSA is a class I aldolase and, therefore, the conserved active to increase the productivity of preFagomine. Since the concentra- lysine in the enzyme active site forms a Schiff base intermediate tions of DHA and FSA are the most sensitive variables to the inac- with DHA (substrate A), which attacks the carbonyl carbon of sub- tivation of FSA, up to seven reactions were carried out with strate B (b-CHO) (Schürmann and Sprenger, 2001). The aldol and different concentrations of these two variables (Figure S28). The retroaldol additions (r6, r7) are then described as an ordered model fitted all the data correctly (Fig. 4, Figure S29), although very sequential mechanism: low concentrations of preFagomine were obtained at each of these conditions. This could be overcome by model exploitation, opti- kcat6  YFSA  ½DHA  ½bCHO mizing the reaction conditions to increase the product yield. r6 ¼ 0 0 0 ð32Þ KM;bCHO  ½DHA þ KM;DHA  KM;bCHO þ ½DHA  ½bCHO Besides, the use of immobilized enzymes could enhance the com- patibility between both enzymes, lowering their inactivation rates. kcat7  YFSA  ½preFagomine r7 ¼ ð33Þ KM;preFagomine þ ½preFagomine 3.4. Model exploitation for process intensification Contrarily to CPO, no inactivation of FSA was expected among 3.4.1. One-pot strategy its reactions; an initial reaction rate analysis can be easily per- The constructed mathematical model was used to find the opti- formed to estimate the parameters. To this end, several reactions mal conditions for the one-pot approach reaction. The variables for were performed varying the concentration of DHA or b-CHO, and optimization were the initial concentrations of all substrates (t- preFagomine, analyzing the reaction rate before 10% of substrate BuOOH, b-OH, DHA) and enzymes (CPO and FSA) (Table S4). The conversion was achieved (Fig. 3a). optimal initial YCPO and YFSA were predicted at the upper bound To check if there is any inhibition between both reactions, var- of the variable range used in the optimization (Table S5); this range ious reactions with different concentrations of b-CHO and DHA was not further increased due to medium viscosity. Considering a were carried out (Figure S21). As it has been already reported else- reaction time of 5 h, the model predicts that a high quantity of both where (Sudar et al., 2013a), product inhibition by the amino alde- enzymes is required to obtain a moderate preFagomine yield of hyde was observed in the retroaldol addition, adding an inhibition 14.6%. The predicted values were experimentally confirmed by a term (Figure S22): coupled reaction at these initial conditions using both soluble kcat7  YFSA  ½preFagomine CPO and FSA, double-checking the applicability of the model and r7 ¼ ð34Þ KM;preFagomine  ð1 þ ½bCHO kir7 Þ þ ½preFagomine the unfeasibility of the coupled system. Some accumulation of the aldehyde was observed at the end of the reaction due to the high instability of FSA (CPO remained active at 66%). No immobi- 3.2.1. Coupled system: FSA inactivation lized derivatives were tested in the one-pot configuration due to The inactivation of FSA in the whole CPO/FSA system was char- a required enzyme load too high to be immobilized on the support acterized for the eventual use of this model for the coupled reac- (considering 5% v/v support/reaction-medium for pseudo- tion, distinguishing two inactivation pathways: kdFSA1 and kdFSA2. homogeneous catalysis). Regarding kdFSA1, FSA was incubated with different concentrations A (semi)continuous addition of peroxide, DHA and/or enzymes of t-BuOOH and CPO (Figure S23). The inactivation was mathemat- was considered as a strategy to overcome the inactivation of FSA. ically expressed as a linear dependence of r4 (Figure S24): Preliminary simulations, however, indicated lower oxidation and 7 G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 Fig. 4. Fitting example of one of the reactions used in the CPO/FSA model validation. Experimental data: b-OH (s), b-CHO (j), b-COOH (+), DHA (d), t-BuOOH (), preFagomine (}), CPO activity (N), FSA activity (4); model simulation: b-OH ( ), b-CHO ( ), b-COOH ( ), DHA ( ), t-BuOOH ( ), preFagomine ( ), CPO activity ( ), FSA activity ( ). aldol addition rates, ruling out this strategy. An alternative opera- at the maximum enzyme load maintaining the 5% v/v support/ tion to one-pot is the separation of both reactions, performing the medium (SI section 3.13). Then the optimal values for the two vari- overall reaction as two consecutive reaction cascade. That avoids ables in the oxidation reaction (initial concentration of b-OH and t- the oxidation of DHA, the inhibition of CPO reactions by DHA, BuOOH) and one variable in the aldol addition reaction (initial con- and the inactivation of FSA. centration of DHA) were predicted. First, the two-step reaction was carried out using both soluble 3.4.2. Two-step reaction CPO and FSA (Fig. 5a). Regarding the oxidation of b-OH, the model The two-step coupled reaction was carried out in two separate accurately predicted the experimental data, leading to a b-CHO vessels with an intermediate step of ultrafiltration to remove CPO yield of 17.9% (expected: 18.4%). The reaction was stopped after from the reaction medium. The filtered mixture was left under 5 h when the measured activity of CPO was negligible; the pseudo mild agitation for 60 min to ensure the degradation of the formed half-life time of the enzyme (CPO-t1/2) was only 18 min. The reac- radical. tion mixture was filtered, and the aldol addition was started after The optimal initial conditions were predicted by a different 60 min. Here, however, the kinetic model was not able to predict optimization procedure: to enable the further use of immobilized the experimental data: the measured preFagomine yield was CPO and FSA in the reaction, the enzyme concentration was fixed 38.4%, whereas the predicted one was 79.3%. The latter value was Fig. 5. CPO and FSA-catalyzed two-step synthesis of preFagomine. (a) Soluble enzymes, (b) mNC-enzymes, (c) agarose-enzymes. Experimental data: b-OH (s), b-CHO (j), b- COOH (+), t-BuOOH (), preFagomine (}), CPO activity (N), FSA activity (4); model simulation: b-OH ( ), b-CHO ( ), b-COOH ( ), t-BuOOH ( ), CPO activity ( ). 8 G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 confirmed by performing an aldol addition reaction using commer- ting of the experimental data, enabling the comprehension of this cial b-CHO at the same reaction conditions (Figure S30) with a pre- highly complex reaction system by identifying the inactivation of Fagomine yield of 81.1%. Therefore, the difference might be caused FSA as the main bottleneck. Our study elucidated the coupled reac- by the inactivation of FSA. Although no inactivation was expected, tion mechanisms, showing the inhibitions among diverse sub- the soluble FSA activity dropped more than 80% in 3 h (FSA- strates and products from the different reactions. t1/2 = 35 min), probably because some inactivating agent was not The kinetic model was used for process intensification, deter- removed from the mixture, and it inactivated the enzyme. Further mining the optimal conditions for the coupling in two reactor con- stabilization was expected by using immobilized enzymes, either figurations: one-pot synthesis or two-consecutive vessels onto nanoparticles or conventional agarose supports. approach. Both strategies were investigated in detail using soluble Four derivatives (nCPO-MANA-agarose, mNC-NH2-CPO, FSA- and immobilized CPO/FSA. The separation of the oxidation and the CoIDA-agarose, mNC-NH2-FSA) were tested at the optimized reac- aldol addition notably reduced the inactivation of FSA driven by tion conditions. However, it should be noted that when immobi- CPO, led to a higher enzyme stability (17-fold increase for CPO, lized enzymes are used, the kinetic parameters can differ from 7-fold for FSA), and incremented the product synthesis. The most the ones found for the soluble catalysts due to mass transfer effects promising results were obtained in the two-step reaction catalyzed or altered inactivation rates due to immobilization (Illanes, 2008). by CPO-MANA-agarose and FSA-CoIDA-agarose, with an 1839-fold The b-OH oxidation catalyzed by mNC-NH2-CPO led to similar increase on the preFagomine production per mol of enzyme in results than the soluble CPO (Fig. 5b). The amino aldehyde yield comparison to the one-pot approach with soluble enzymes. was 19.0% at the same reaction time (5 h), while no improvement in the operational stability of the enzyme was observed (CPO- Declaration of Competing Interest t1/2 = 24 min, 1.3-fold increase). In the aldol addition catalyzed by mNC-NH2-FSA, a lower preFagomine yield (15.0%) was The authors declare that they have no known competing finan- obtained, likely related to the rapid inactivation of FSA through this cial interests or personal relationships that could have appeared immobilization strategy (FSA-t1/2 = 8 min). Regarding the agarose to influence the work reported in this paper. derivatives, the oxidation of CPO on MANA-agarose (Fig. 5c) gave similar results with a b-CHO yield of 20.2%. Although the enzyme Acknowledgments remained active after 5 h (CPO-t1/2  300 min, 17-fold increase), the reaction was stopped because the amino aldehyde concentra- This work was supported by Spanish MINECO (project No. tion reached a plateau. FSA immobilized on CoIDA-agarose led to CTQ2014-53114R), co-financed by European Regional Develop- the best results in the aldol addition reaction, with a preFagomine ment Fund. The Department of Chemical, Biological and Environ- yield of 96.3% (this reaction was left for 4 h since FSA remained mental Engineering of UAB constitutes the Biochemical active, FSA-t1/2 240 min, 7-fold increase). Although diffusional Engineering Unit of the Reference Network in Biotechnology, and limitations in the activity test can hinder its interpretation, FSA the research group 2014SGR452, Generalitat de Catalunya. Finan- stability is truly enhanced by the immobilization of the enzyme cial support from Spanish MINECO for the PhD scholarship of Ger- into CoIDA-agarose. ard Masdeu is acknowledged. The authors acknowledge the With the last catalytic complex formed by CPO-MANA-agarose/ financial support from the Slovenian Research Agency (ARRS) for FSA-CoIDA, the highest final overall yield of the two-step reaction Research Core Funding (P2-0089) and ARRS Project (J1-7302), was obtained: 19.5%. In comparison to the one-pot reaction using and the financial support for the research work in the host institu- soluble enzymes, the results are incredibly promising: (a) in the tions (Jožef Stefan Institute, University of Zagreb) from the Euro- one-pot approach, only 1.82 mmol preFagomine were obtained pean Comission (Erasmus program) and COST-STSM-CM1303. from 16.0 mmol b-OH using 85 mmol CPO and 52 mmol FSA, with The authors thank Dr. Alba Eustaquio and Dr. Maria Jesús Ibarz a total production of 0.026 mmolpreFagomine mol1 1 1 b-OH molCPO molFSA. from Servei d’Anàlisi Química of the Universitat Autònoma de Bar- (b) In the two-step reaction with agarose derivatives, 0.96 mmol celona for their assistance on mass spectrometry analyses. preFagomine were obtained from 5.0 mmol b-OH using 12.8 mmol CPO and 0.31 mmol FSA (concentrations considering the activity loss during the immobilization procedure), with a total production Appendix A. Supplementary material of 47.8 mmolpreFagomine mol1 1 1 b-OH molCPO molFSA. That denotes an 1839-fold increase of the specific preFagomine production of the Supplementary data to this article can be found online at presented immobilized system on the coupled reaction. https://doi.org/10.1016/j.ces.2021.116602. References 4. Conclusions Andersson, L., Wolfenden, R., 1982. A General Method of aplha-Aminoaldehyde Synthesis Using Alcohol Dehydrogenase. Anal. Biochem. 124, 150–157. https:// The feasibility of a multi-enzymatic reaction cascade to synthe- doi.org/10.1016/0003-2697(82)90232-9. size iminocyclitols using amino alcohols as the starting material Andersson, M., Andersson, M.M., Adlercreutz, P., 2000. Stabilisation of Chloroperoxidase Towards Peroxide Dependent Inactivation. Biocatal. has been investigated. Particularly, the coupled CPO-catalyzed b- Biotransformat. 18, 457–469. https://doi.org/10.3109/10242420009015263. OH oxidation and FSA-catalyzed DHA aldol addition that renders Ardao, I., Benaiges, M.D., Caminal, G., Álvaro, G., 2006. One step purification- preFagomine are investigated in detail. A mathematical model for immobilization of fuculose-1-phosphate aldolase, a class II DHAP dependent aldolase, by using metal-chelate supports. Enzyme Microb. Technol. 39, 22–27. better understanding of the reaction kinetics of the whole system https://doi.org/10.1016/j.enzmictec.2005.09.001. was developed by an in-depth study of each individual reaction Ayala, M., Batista, C.V., Vazquez-Duhalt, R., 2011. Heme destruction, the main separately. This is crucial for avoiding undesired interactions molecular event during the peroxide-mediated inactivation of chloroperoxidase from Caldariomyces fumago. J. Biol. Inorg. Chem. 16, 63–68. https://doi.org/ among reactions which could hinder the analysis of the results. 10.1007/s00775-010-0702-6. Up to 7 reactions were investigated, estimating their kinetic Banba, Y., Abe, C., Nemoto, H., Kato, A., Adachi, I., Takahata, H., 2001. Asymmetric parameters: degradation of peroxide, chemical reaction, amino synthesis of fagomine and its congeners. Tetrahedron Asymmetry 12, 817–819. aldehyde oxidation, alcohol oxidation, DHA oxidation, aldol addi- https://doi.org/10.1016/S0957-4166(01)00136-7. Calleja, D., Kavanagh, J., de Mas, C., López-Santín, J., 2016. Simulation and prediction tion, and retroaldol addition. The inactivation of both enzymes of protein production in fed-batch E. coli cultures: An engineering approach. was also described. The kinetic model was validated with great fit- Biotechnol. Bioeng. 113, 772–782. https://doi.org/10.1002/bit.25842. 9 G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 Calveras, J., Egido-Gabás, M., Gómez, L., Casas, J., Parella, T., Joglar, J., Bujons, J., Koyama, M., Sakamura, S., 1974. The Structure of a New Piperidine Derivative from Clapés, P., 2009. Dihydroxyacetone phosphate aldolase catalyzed synthesis of Buckwheat Seeds (Fagopyrum esculentum Moench). Agric. Biol. Chem. 38, structurally diverse polyhydroxylated pyrrolidine derivatives and evaluation of 1111–1112. https://doi.org/10.1080/00021369.1974.10861295. their glycosidase inhibitory properties. Chem. - A Eur. J. 15, 7310–7328. https:// Kralj, S., Drofenik, M., Makovec, D., 2011. Controlled surface functionalization of doi.org/10.1002/chem.200900838. silica-coated magnetic nanoparticles with terminal amino and carboxyl groups. Casella, L., Poli, S., Gullotti, M., Selvaggini, C., Beringhelli, T., Marchesini, A., 1994. J. Nanoparticle Res. 13, 2829–2841. https://doi.org/10.1007/s11051-010-0171- The chloroperoxidase-catalyzed oxidation of phenols. Mechanism, selectivity, 4. and characterization of enzyme-substrate complexes. Biochemistry 33, 6377– Kralj, S., Makovec, D., 2015. Magnetic Assembly of Superparamagnetic Iron Oxide 6386. https://doi.org/10.1021/bi00187a001. Nanoparticle Clusters into Nanochains and Nanobundles. ACS Nano 9, 9700– Castillo, J.A., Calveras, J., Casas, J., Mitjans, M., Vinardell, M.P., Parella, T., Inoue, T., 9707. https://doi.org/10.1021/acsnano.5b02328. Sprenger, G. a., Joglar, J., Clapés, P., 2006. Fructose-6-phosphate aldolase in Kralj, S., Makovec, D., Čampelj, S., Drofenik, M., 2010. Producing ultra-thin silica organic synthesis: Preparation of D-fagomine, N-alkylated derivatives, and coatings on iron-oxide nanoparticles to improve their surface reactivity. J. preliminary biological assays. Org. Lett. 8, 6067–6070. https://doi.org/10.1021/ Magn. Magn. Mater. 322, 1847–1853. https://doi.org/10.1016/j. ol0625482 jmmm.2009.12.038. Castillo, J.A., Guérard-Hélaine, C., Gutiérrez, M., Garrabou, X., Sancelme, M., Kumari, N., Reddy, B.G., Vankar, Y.D., 2009. Efficient and stereodivergent syntheses Schürmann, M., Inoue, T., Hélaine, V., Charmantray, F., Gefflaut, T., Hecquet, L., of D- and L-fagomines and their analogues. European J. Org. Chem. 160–169. Joglar, J., Clapés, P., Sprenger, G.A., Lemaire, M., 2010. A mutant D-fructose-6- https://doi.org/10.1002/ejoc.200800796. phosphate aldolase (Ala129Ser) with improved affinity towards Lenoir, D., 2006. Selective oxidation of organic compounds-sustainable catalytic dihydroxyacetone for the synthesis of polyhydroxylated compounds. Adv. reactions with oxygen and without transition metals?. Angew. Chem. Int. Ed. Synth. Catal. 352, 1039–1046. https://doi.org/10.1002/adsc.200900772. 45, 3206–3210. https://doi.org/10.1002/anie.200502702. Česnik, M., Sudar, M., Roldan, R., Hernandez, K., Parella, T., Clapés, P., Charnock, S., Masdeu, G., Kralj, S., Pajk, S., López-Santín, J., Makovec, D., Álvaro, G., 2018. Hybrid Vasić-Rački, Ð., Findrik Blažević, Z., 2019. Model-based optimization of the chloroperoxidase-magnetic nanoparticle clusters: Effect of functionalization on enzymatic aldol addition of propanal to formaldehyde: A first step towards biocatalyst performance. J. Chem. Technol. Biotechnol. 93, 233–245. https://doi. enzymatic synthesis of 3-hydroxybutyric acid. Chem. Eng. Res. Des. 150, 140– org/10.1002/jctb.5345. 152. https://doi.org/10.1016/j.cherd.2019.06.025. Masdeu, G., Pérez-Trujillo, M., López-Santín, J., Álvaro, G., 2016a. Chloroperoxidase- Chamulitrat, W., Takahashi, N., Mason, R.P., 1989. Peroxyl, alkoxyl, and carbon- catalyzed amino alcohol oxidation: Substrate specificity and novel strategy for centered radical formation from organic hydroperoxides by chloroperoxidase. J. the synthesis of N-Cbz-3-aminopropanal. Process Biochem. 51, 1204–1211. Biol. Chem. 264, 7889–7899. https://doi.org/10.1016/S0021-9258(18)83127-1. https://doi.org/10.1016/j.procbio.2016.05.022. Colonna, S., Gaggero, N., Richelmi, C., Pasta, P., 1999. Recent biotechnological Masdeu, G., Pérez-Trujillo, M., López-Santín, J., Álvaro, G., 2016b. Data on the developments in the use of peroxidases. Trends Biotechnol. 17, 163–168. identification and characterization of by-products from N-Cbz-3- https://doi.org/10.1016/S0167-7799(98)01288-8. aminopropanal and t-BuOOH/H2O2 chemical reaction in chloroperoxidase- Dunford, H.B., 2010. Peroxidases and Catalases: Biochemistry, Biophysics, catalyzed oxidations. Data Br. 8, 659–665. https://doi.org/10.1016/j. Biotechnology and Physiology. John Wiley & Sons, NJ, USA. https://doi.org/ dib.2016.06.028. 10.1002/cbic.200000397 Masdeu, G., Vázquez, L.M., López-Santín, J., Caminal, G., Kralj, S., Makovec, D., Fan, J.Q., Ishii, S., Asano, N., Suzuki, Y., 1999. Accelerated transport and maturation Álvaro, G., Guillén, M., 2021. Synthesis of a precursor of D-fagomine by of lysosomal alpha-galactosidase A in Fabry lymphoblasts by an enzyme immobilized fructose-6-phosphate aldolase. PLoS One. https://doi.org/10.1371/ inhibitor. Nat. Med. 5, 112–115. https://doi.org/10.1038/4801. journal.pone.0250513. Ferioli, V., Vezzalini, E., Rustichelli, C., Gamberini, G., 1995. High-Performance Morris, D.R., Hager, L.P., 1966. Chloroperoxidase I. Isolation and Properties of the Liquid Chromatography of Dihydroxyacetone as its bis-2,4- crystalline Glycoprotein. J. Biol. Chem. 241, 1763–1768. https://doi.org/ Dinitrophenylhydrazone Derivative. Chromatographia 41, 61–65. https://doi. 10.1016/S0021-9258(18)96701-3. org/10.1007/BF02688001. Nojima, H., Kimura, I., Chen, F.J., Sugihara, Y., Haruno, M., Kato, a, Asano, N., 1998. Fernandez-Lafuente, R., Rosell, C.M., Rodriguez, V., Santana, C., Soler, G., Bastida, A., Antihyperglycemic effects of N-containing sugars from Xanthocercis Guisan, J.M., 1993. Preparation of activated supports containing low pK amino zambesiaca, Morus bombycis, Aglaonema treubii, and Castanospermum groups. A new tool for protein immobilization via the carboxyl coupling australe in streptozotocin-diabetic mice. J. Nat. Prod. 61, 397–400. https://doi. method. Enzyme Microb. Technol. 15, 546–550. https://doi.org/10.1016/0141- org/10.1021/np970277l 0229(93)90016-u. Pandey, G., Kapur, M., 2000. A general strategy towards the synthesis of 1-N- Fleet, G.W.J., Fellows, L.E., Smith, P.W., 1987. Synthesis of deoxymannojirimycin iminosugar type glycosidase inhibitors: demonstration by the synthesis of d- as fagomine deoxynojirimycin 2-acetamido-1,5-imino-1,2,5-trideoxy-D-mannitol well as l-glucose type iminosugars (isofagomines). Tetrahedron Lett. 41, 8821– 2-acetamido-1,5-imino-1,2,5-trideoxy-D-glucitol 2S,3R,4R,5R- 8824. https://doi.org/10.1016/S0040-4039(00)01553-7. trihydroxypipecolic acid and 2S,3R,4R,5S-trihydroxypipecolic acid from Park, J.-B., Clark, D.S., 2006. Deactivation Mechanisms of Chloroperoxidase During methyl 3-O-benzyl. Tetrahedron 43, 979–990. https://doi.org/10.1016/S0040- Biotransformations. Biotechnol. Bioeng. 93, 1190–1195. https://doi.org/ 4020(01)90035-3. 10.1002/bit. Gloster, T.M., Meloncelli, P., Stick, R.V., Zechel, D., Vasella, A., Davies, G.J., 2007. Pearson, M.S.M., Mathé-Allainmat, M., Fargeas, V., Lebreton, J., 2005. Recent Glycosidase inhibition: An assessment of the binding of 18 putative transition- advances in the total synthesis of piperidine azasugars. European J. Org. state mimics. J. Am. Chem. Soc. 129, 2345–2354. https://doi.org/ Chem. 2159–2191. https://doi.org/10.1002/ejoc.200400823. 10.1021/ja066961g. Pešić, M., López, C., Álvaro, G., López-Santín, J., 2012. A novel immobilized Gómez, L., Molinar-Toribio, E., Calvo-Torras, M.Á., Adelantado, C., Juan, M.E., Planas, chloroperoxidase biocatalyst with improved stability for the oxidation of J.M., Cañas, X., Lozano, C., Pumarola, S., Clapés, P., Torres, J.L., 2012. d-Fagomine amino alcohols to amino aldehydes. J. Mol. Catal. B Enzym. 84, 144–151. https:// lowers postprandial blood glucose and modulates bacterial adhesion. Br. J. Nutr. doi.org/10.1016/j.molcatb.2012.04.010. 107, 1739–1746. https://doi.org/10.1017/S0007114511005009. Robertson, J.G., 2005. Mechanistic basis of enzyme-targeted drugs. Biochemistry 44, Goujon, J.Y., Gueyrard, D., Compain, P., Martin, O.R., Ikeda, K., Kato, A., Asano, N., 5561–5571. https://doi.org/10.1021/bi050247e. 2005. General synthesis and biological evaluation of a-1-C-substituted Rodriguez-Hinestroza, R.A., Lopez, C., Lopez-Santin, J., Kane, C., Dolors Benaiges, M., derivatives of fagomine (2-deoxynojirimycin-a-C-glycosides). Bioorganic Med. Tzedakis, T., 2017. HLADH-catalyzed synthesis of beta-amino acids, assisted by Chem. 13, 2313–2324. https://doi.org/10.1016/j.bmc.2004.12.043. continuous electrochemical regeneration of NAD+ in a filter press microreactor. Guisán, J., 1988. Aldehyde-agarose gels as activated supports for immobilization- Chem. Eng. Sci. 158, 196–207. https://doi.org/10.1016/j.ces.2016.10.010. stabilization of enzymes. Enzyme Microb. Technol. 10, 375–382. https://doi.org/ Schürmann, M., Sprenger, G.A., 2001. Fructose-6-phosphate Aldolase is a Novel 10.1016/0141-0229(88)90018-X. Class I Aldolase from Escherichia coli and is Related to a Novel Group of Gutierrez, M., Parella, T., Joglar, J., Bujons, J., Clapés, P., 2011. Structure-guided Bacterial Transaldolases. J. Biol. Chem. 276, 11055–11061. https://doi.org/ redesign of D-fructose-6-phosphate aldolase from E. coli: remarkable activity 10.1074/jbc.M008061200. and selectivity towards acceptor substrates by two-point mutation. Chem. Shaw, P.D., Hager, L.P., 1961. VI. Chloroperoxidase: a component of the beta- Commun. 47, 5762–5764. https://doi.org/10.1039/c1cc11069a. ketoadipate chlorinase system. J. Biol. Chem. 236, 1626–1630. https://doi.org/ Hereu, M., Ramos-Romero, S., Marín-Valls, R., Amézqueta, S., Miralles-Pérez, B., 10.1016/S0021-9258(19)63275-8. Romeu, M., Méndez, L., Medina, I., Torres, J.L., 2019. Combined buckwheat d- Shevelkova, A.N., Ryabov, A.D., 1996. Irreversible inactivation of Caldariomyces fagomine and fish omega-3 pufas stabilize the populations of gut prevotella and fumago chloroperoxidase by hydrogen peroxide. A kinetic study in chloride and bacteroides while reducing weight gain in rats. Nutrients 11. https://doi.org/ bromide system. Biochem. Mol. Biol Int 39, 665–670. https://doi.org/10.1080/ 10.3390/nu11112606. 15216549600201731. Herna, K., Bujons, J., Parella, T., Rolda, R., 2019. Aldolase-Catalyzed Asymmetric Stütz, A.E., 1999. Iminosugars as Glycosidase Inhibitors. Wiley-VCH, Weinheim, Synthesis of N-Heterocycles by Addition of Simple Aliphatic Nucleophiles to Germany. 10.1002/3527601740.fmatter. Aminoaldehydes. Adv. Synth. Catal. 361, 2673–2687. https://doi.org/10.1002/ Sudar, M., Česnik, M., Clapés, P., Pohl, M., Vasić-Rački, Ð., Findrik Blažević, Z., 2021. A adsc.201801530. cascade reaction for the synthesis of D-fagomine precursor revisited: Kinetic Illanes, A., 2008. Enzyme biocatalysis: Principles and applications, Enzyme insight and understanding of the system. N. Biotechnol. 63, 19–28. https://doi. Biocatalysis: Principles and Applications. https://doi.org/10.1007/978-1-4020- org/10.1016/j.nbt.2021.02.004. 8361-7 Sudar, M., Findrik, Z., Lozano, C., 2013a. Mathematical model for aldol addition Kedderis, G.L., Hollenberg, P.F., 1983. Steady state kinetics of chloroperoxidase- catalyzed by two d-fructose-6-phosphate aldolases variants overexpressed in catalyzed N-demethylation reactions. J. Biol. Chem. 258, 12413–12419. https:// E. coli. J. Biotechnol. 167, 191–200. https://doi.org/10.1016/j. doi.org/10.1016/S0021-9258(17)44191-3. jbiotec.2013.07.008. 10 G. Masdeu, Z. Findrik Blažević, S. Kralj et al. Chemical Engineering Science 239 (2021) 116602 Sudar, M., Findrik, Z., Vasić-Rački, D., Clapés, P., Lozano, C., 2013b. Aldol addition of Additions of Glycolaldehyde. Chem. - A Eur. J. 20, 12572–12583. https://doi. dihydroxyacetone to N-Cbz-3-aminopropanal catalyzed by two aldolases org/10.1002/chem.201403281. variants in microreactors. Enzyme Microb. Technol. 53, 38–45. https://doi.org/ Tadic, M., Kralj, S., Jagodic, M., Hanzel, D., Makovec, D., 2014. Magnetic properties of 10.1016/j.enzmictec.2013.03.013. novel superparamagnetic iron oxide nanoclusters and their peculiarity under Sudar, M., Findrik, Z., Vasić-Rački, Ð., Soler, A., Clapés, P., 2015. A new concept for annealing treatment. Appl. Surf. Sci. 322, 255–264. https://doi.org/10.1016/j. production of (3S,4R)-6-[(benzyloxycarbonyl)amino]-5,6-dideoxyhex-2-ulose, apsusc.2014.09.181. a precursor of D -fagomine. RSC Adv. 5, 69819–69828. https://doi.org/10.1039/ Taniguchi, S., Asano, N., Tomino, F., Miwa, I., 1998. Potentiation of glucose-induced C5RA14414K. insulin secretion by fagomine, a pseudo- sugar isolated from mulberry leaves. Sugiyama, M., Hong, Z., Liang, P.-H., Dean, S.M., Whalen, L.J., Greenberg, W.A., Wong, Horm. Metab. Res. 30, 679–683. https://doi.org/10.1055/s-2007-978957. C.-H., 2007. D-Fructose-6-phosphate aldolase-catalyzed one-pot synthesis of Toti, P., Petri, A., Gambicorti, T., Osman, A.M., Bauer, C., 2005. Kinetic and stability iminocyclitols. J. Am. Chem. Soc. 129, 14811–14817. https://doi.org/ studies on the chloroperoxidase complexes in presence of tert-butyl 10.1021/ja073911i. hydroperoxide. Biophys. Chem. 113, 105–113. https://doi.org/10.1016/j. Szekrenyi, A., Soler, A., Garrabou, X., Guérard-Hélaine, C., Parella, T., Joglar, J., bpc.2004.08.003. Lemaire, M., Bujons, J., Clapés, P., 2014. Engineering the Donor Selectivity of D- Whalen, L.J., Wong, C.H., 2006. Enzymes in organic synthesis: aldolase-mediated Fructose-6-Phosphate Aldolase for Biocatalytic Asymmetric Cross-Aldol synthesis of iminocyclitols and novel heterocycles. Aldrichimica Acta 39, 63–71. https://doi.org/10.1002/chin.200729227. 11