**4. The MagneChip platform: application examples**

( ) () () ()

where (*j,k*) are the pixel coordinates of the plan view image; therefore, the changes in accord‐ ance to the reference image are indicated by white pixels. The total number of white pixels is defined as *chamber difference score* (*SC*) used as a marker for describing the changes of the MNP layer arrangement. Therefore, the changes compared with the image of the first cycle (reference) were indicated by white areas during the consecutive cycles of the measurement.

In practice, *SC* values under 2000 reflected to negligible changes. However, *SC* > 3000 indicated serious structural change of the *ec*MNP layer, for instance, the complete breakthrough of a bubble (**Figure 5b**) [22]. Air bubbles usually did not split at the channel entrance, rather passed at one side along the chamber wall. Numerical simulations revealed (**Figure 5c–f**) that the velocity profile became asymmetric due to the bubble passage and the overall mass flow rate through the porous MNP layer significantly decreased (28.6–20.7 μL min−1, roughly 72% of its original value), while the remaining fluid passed through the developed tunnel. The passing bubble could drift away particles which decreased the total mass of the biocatalyst in the reaction chamber. Therefore, the biocatalytic activity of the damaged chamber decreased and

Reliability assessment of the measurements was based mostly on the following parameters:

**1.** Chamber difference score (*SC*)—Over *SC* > 4000 (average), the measurement was declined.

Each of the experiments carried out by the platform was justified based on the above criterion.

**Figure 6.** Time plot of the periodic absorbance change during the cyclic measurement (attempt 1, stable layer). The chip is re-initialized between the reaction steps (reaching zero absorbance) by washing out substrate and product com‐

**2.** Control measurement—Over 5% of error, the measurement was declined.

P = í

*diff*

170 Lab-on-a-Chip Fabrication and Application

the consequent measurements were no longer reliable.

*3.5.3. Reproducibility of cyclic reactions*

pletely [22]. The last measurement served as a control.

,, , , 0 , 0, , , 0 ìïP P -P <

ï P -P ³ î *ref ref seq i*

*ref seq i jk jk jk j k jk jk*

() ()

,

,

#### **4.1. Characterization of the PAL reaction with L-phenylalanine (L-1a) in MagneChip**

*4.1.1. Influence of the substrate flow rate on the biotransformation*

**Figure 7.** Dependence of the reaction rate of L-**1a** conversion to **2a** on the flow rate in MagneChip filled by *ec*MNP. Saturation was reached at 25 μL min−1 [22].

MagneChip was filled with *ec*MNP biocatalyst, and biotransformations of L-**1a** to **2a** (**Scheme 1**) at various flow rates were performed in seven consecutive cycles, while the chip was reinitialized at the end of each cycle and a new substrate flow rate was set between 3.6 and 28.6 μL min−1. The first (reference) measurement was repeated in the last cycle as a control. The negligible difference of specific biocatalytic activity (*U*B) between the reference and control measurements (only 3%) and low *SC* score (SC < 338) indicated that the shear forces did not caused irreversible changes on the biocatalytic activity even at high flow rate (up to 28.6 μL min−1) [22].

The reaction velocity was calculated for each cycles. By increasing the flow rate, the calculat‐ ed reaction velocity increased until reaching saturation at about 25 μL min−1 (**Figure 7**).

#### *4.1.2. Calculation of kinetic parameters*

MagneChip was filled with *ec*MNP biocatalyst, and biotransformations of L-**1a** to **2a** (**Scheme 1**) at various concentrations of L-**1a** (*S*0) were performed in 10 consecutive cycles, while the chip was re-initialized at the end of each cycle and a new substrate concentration was set. It was found that the reaction followed the first-order kinetics up to (*S*0) = 3 mM and saturated roughly at (*S*0) = 20 mM [22].

The linear fitting method proposed by Lilly et al. [15] was applied for the calculation of the kinetic constants of the biotransformation of L-**1a** to **2a** in the MagneChip (**Figure 8**, bottom). The values of the kinetic constants are summarized in **Table 2** [22].

**Figure 8.** (a) Dependency of the substrate concentration on reaction velocity in MagneChip for the transformation of L-**1a** to **2a** by MNP biocatalyst. Saturation concentration was reached at 20 mM. (b) Linear fit based on the Lilly–Horn‐ by model [15] to determine *K*m (resulting in *K*m = 2.5 mM) [22].


**Tablee 2.** Kinetic constants in biotransformation of L-**1a** to **2a** with MNP in shake vial and in MagneChip [22]

It was found that the apparent *K*m value was reasonably smaller in MagneChip (2.5 mM) than in shake vial (9.1 mM). Turnover number (*k*cat) and specificity constant (*k*cat*/K*m) were deter‐ mined also for both reaction modes. While in the shake vial, the turnover number was somewhat higher (3.2·× 10−2 s−1) than the in chip (2.8·×·10−2 s−1), the specificity constant turned out to be significantly higher in chip (11.3 s−1 M−1) as compared with the shake vial (3.5 s−1 M−1). This may be attributed to the smaller *K*<sup>m</sup> value in the MagneChip indicating significant contribution of diffusion effects to the higher apparent *K*m value in shake vial.

#### **4.2. Effect of particle size on the enzyme activity**

The reaction velocity was calculated for each cycles. By increasing the flow rate, the calculat‐ ed reaction velocity increased until reaching saturation at about 25 μL min−1 (**Figure 7**).

MagneChip was filled with *ec*MNP biocatalyst, and biotransformations of L-**1a** to **2a** (**Scheme 1**) at various concentrations of L-**1a** (*S*0) were performed in 10 consecutive cycles, while the chip was re-initialized at the end of each cycle and a new substrate concentration was set. It was found that the reaction followed the first-order kinetics up to (*S*0) = 3 mM and saturated roughly

The linear fitting method proposed by Lilly et al. [15] was applied for the calculation of the kinetic constants of the biotransformation of L-**1a** to **2a** in the MagneChip (**Figure 8**, bottom).

**Figure 8.** (a) Dependency of the substrate concentration on reaction velocity in MagneChip for the transformation of L-**1a** to **2a** by MNP biocatalyst. Saturation concentration was reached at 20 mM. (b) Linear fit based on the Lilly–Horn‐

**Kinetic parameter MagneChip Shake vial**

**Tablee 2.** Kinetic constants in biotransformation of L-**1a** to **2a** with MNP in shake vial and in MagneChip [22]

It was found that the apparent *K*m value was reasonably smaller in MagneChip (2.5 mM) than in shake vial (9.1 mM). Turnover number (*k*cat) and specificity constant (*k*cat*/K*m) were deter‐ mined also for both reaction modes. While in the shake vial, the turnover number was somewhat higher (3.2·× 10−2 s−1) than the in chip (2.8·×·10−2 s−1), the specificity constant turned

*K*m (mM) 2.5 9.1 *k*cat (s−1) 2.8 × 10−2 3.2 × 10−2 *k*cat/*K*m (s−1 M−1) 11.3 3.5

The values of the kinetic constants are summarized in **Table 2** [22].

by model [15] to determine *K*m (resulting in *K*m = 2.5 mM) [22].

*4.1.2. Calculation of kinetic parameters*

172 Lab-on-a-Chip Fabrication and Application

at (*S*0) = 20 mM [22].

The accumulated quantity of *ec*MNPs in the reaction chambers was determined by the embedded resonant magnetometer of the MagneChip device [23]. The measurements revealed that the total mass of the accumulated particles was approximately the same for two differ‐ ent particle sizes (m = 241.6 μg, *ec*MNP600, d = 600 nm and m = 248.3 μg, *ec*MNP250, d = 250 nm) [23]. The total particle mass could be only increased using a binary mixture (m = 283.6 μg, MNP250:600) of the particles. This experiment resulted in a significantly higher MNP mass (17%) captured in the magnetic chamber as compared with the chamber capacity filled with MNPs of uniform particle sizes [23].

MagneChip was filled with different sized MNP biocatalysts, and biotransformations of L-**1a** to **2a** (**Scheme 1**) were also performed [23]. Compared with the larger particles (ecMNP600), the total surface area increased both in the *ec*MNP250 (2.5 times) and the mixture cases (2.06 times). Note that differences in biocatalytic activity can be expected only due to changes of trans‐ port limitations as the enzyme to MNP mass ratio was kept to be constant of 15% in both cases.

**Figure 9.** Specific absorbance of cinnamic acid (**2a**) at 295 nm at the chip outlet using MNP600, MNP250, and 1:1 mixture of the two kind of particles in the chip [25].

Results of the measurement using variously sized *ec*MNPs as biocatalysts are summarized in **Figure 9**. In fact, the *ec*MNP600-filled chambers yielded the lowestfinal concentration of product as indicated by the lowest specific absorbance (AU = 1.07, at 295 nm) at the chip outlet. Filling the chip by *ec*MNP250 resulted in an increase of the measured absorbance by 46% (AU = 1.56, at 295 nm). Because the chambers contained the same filling mass (m = 241.6 μg for *ec*MNP600 and m = 248.3 μg for*ec*MNP250) and therefore the same enzyme amount, the difference between the MNP250- and the MNP600-filled reactors can only be attributed to other factors, for exam‐ ple, to the differences in total surface area [23].

The major difference can also stem from the remarkably smaller average microchannel diameters between the particles within the *ec*MNP250-filled chamber as compared to the *ec*MNP600-filled one. This can result in shortened diffusion path and therefore better mass transport [23]. An additional 40% increment was achieved using the 1:1 particle mixture, which was obviously resulted as a synergy of the higher enzyme content (17%) due to the higher chamber capacity and enhanced transport phenomena due to the small average microchan‐ nel diameter [25].

#### **4.3. Testing multiple substrates in MagneChip**

Substrate screening experiments were performed with a single *ec*MNP-loading in the chip passing the solutions of the different substrates (**Scheme 1**: L-**1a** and *rac*-**1b-f**) through the same chip according to a predefined sequence [22]. The intensive washing procedure between the individual tests with various substrates ensured complete removal of any substrate or product from the preceding cycle (reaction). In the first cycle, the ammonia elimination was meas‐ ured from L-**1a** (the natural substrate of PAL). This reaction was chosen as reference for comparison to the other elimination reactions of PAL from the further substrates (*rac*-**1b-f**). The difference between the initial and final (control) measurement with L-**1a** was found to be only 1.5%, while the *SC* score remained below 2000. Surprisingly, in the MagneChip device, higher biocatalytic activities (*U*B) were observed with four of the unnatural substrates (*rac*-**1b,c,e,f**), than with the natural substrate L-phenylalanine L-**1a** (**Figure 10**).

**Figure 10.** Comparison of the specific biocatalytic activity of *Pc*PAL immobilized on MNPs with substrates L-**1a** and rac-**1b-f** in MagneChip system [(S) = 20 mM, flow rate: 48.6 μL min−1] [22]. \*Control measurement.

Noteworthy, all the four unnatural substrates (*rac*-**1b,c,e,f**) which were transformed by the MNP biocatalyst with higher specific biocatalytic activity (*U*B) than that of L-phenylalanine L-**1a** contained slightly more electron-withdrawing aromatic moieties than the phenyl group. This difference from the productivity ranks observed with homogenous *Pc*PAL so far may be due to the reduced contribution of the reverse reaction (equilibrium effect) to the apparent forward reaction rates in the continuous-flow system at high flow rates [22].

#### **4.4. Characterization of an enzyme reaction with a novel substrate**

By a reaction performed in the MagneChip device, it was first demonstrated that PAL can catalyze the ammonia elimination from the acyclic DL-propargylglycine (PG) to yield (*E*) pent-2-ene-4-ynoate, indicating new opportunities to extend the MIO-enzyme toolbox toward acyclic substrates. Deamination of PG, being acyclic, cannot involve a Friedel–Crafts-type attack at an aromatic ring [18].

MagneChip, filled by PAL-*ec*MNPs, was used for the microscale biotransformation of DLpropargylglycine in sodium carbonate-buffered D2O. The device enabled to detect the formation of (*E*)-pent-2-en-4-ynoate at 242 nm and to produce measurable quantities of the product for recording <sup>1</sup> H-NMR spectra without any work-up. Besides the significant in‐ crease of the UV-signal at 242 nm (up to A = 1.2) in the in-line UV-cell (**Figure 11**), the appearance of olefin hydrogen signals in the 1 H-NMR spectrum of the reaction mixture [at δ = 6.34 (*d*) and 6.85 (*d*) ppm] indicated unambiguously the formation of (*E*)-pent-2-en-4-ynoate. On the other hand, emergence of the UV signal at 274 nm during the process indicated the formation of further by-product(s) apart from (*E*)-pent-2-en-4-ynoate (**Figure 11**).

**Figure 11.** Ammonia elimination from DL-propargylglycine in MagneChip filled with PAL immobilized on MNPs and equipped with in-line UV–Vis detector (reaction in D2O at pD 8.8, 37°C) [18]. The progress of the reaction was followed by full UV-spectra.
