Out-of-equilibrium microcompartments for the bottom-up integration of metabolic functions

Self-sustained metabolic pathways in microcompartments are the corner-stone for living systems. From a technological viewpoint, such pathways are a mandatory prerequisite for the reliable design of artificial cells functioning out-of-equilibrium. Here we develop a microfluidic platform for the miniaturization and analysis of metabolic pathways in man-made microcompartments formed of water-in-oil droplets. In a modular approach, we integrate in the microcompartments a nicotinamide adenine dinucleotide (NAD)-dependent enzymatic reaction and a NAD-regeneration module as a minimal metabolism. We show that the microcompartments sustain a metabolically active state until the substrate is fully consumed. Reversibly, the external addition of the substrate reboots the metabolic activity of the microcompartments back to an active state. We therefore control the metabolic state of thousands of independent monodisperse microcompartments, a step of relevance for the construction of large populations of metabolically active artificial cells.

). The initial rates of NADH oxidation were determined in the same experimental setup (Supplementary Fig.   12b).
The oxygen consumption of IMVs was determined using an oxygraph equipped with a Clark-type electrode (Oxytherm, Hansatech) ( Supplementary Fig. 12c). To a solution containing 450 µL buffer (100 mM NaOH-Tricine (pH 8.0), 5 mM MgCl 2 ), 10 µL IMVs stock solution was added and the reaction was initiated upon addition of 40 µL NADH stock solutions, such as the final NADH concentrations were 0, 0.25, 0.5 and 1 mM. Thus, final IMVs concentration was 4.4×10 9 vesicles per mL, corresponding to 50× dilution. In addition, another experiment with 100× dilution (2.2×10 9 vesicles per mL) and 0.5 mM final NADH concentration was performed to probe the influence of IMVs concentration on the oxygen consumption rate. The initial rate was independent on the NADH concentration (0.25, 0.5 and 1 mM) but decreased twice when the IMVs concentration was doubly reduced in accordance with the linear dependence with respect to NADH (Supplementary Fig. 13c).
In the case of 0.5 mM NADH, which was further used in the microfluidic measurements, the oxygen concentration decreased with about 20%. After consumption of the NADH the oxygen concentration was equilibrated to the initial values due to the fact that the measurement cuvette was open to the ambient air ( Supplementary Fig. 12c).

Supplementary Note 4: 2D observatory chamber assembly
A 2D observation chamber was assembled adapting the protocol described in [6] (Supplementary Fig. 15). Glass microscopy slides were used as top and bottom covers (76 x 25 x 1 mm, Marienfeld). Two access holes of 1.5 mm diameter were created in the top glass slide using micro-sandblasting. Both slides were thoroughly cleaned using soap, water, ethanol and acetone and dried at 70 • C. The chamber geometry was cut in a 60 µm-thick double-sided bonding tape (1375, SDAG Adhésifs) using a Graphtech cutting plotter (CE 6000-40). The double-sided bonding template was transferred on the bottom glass slide and the top glass slide was then bonded to seal the system. The chamber was incubated 24 h at 70 • C. Next, two nanoports were attached to the holes using a UV curable glue (Loctite 3526, Henkel). Subsequently, the surface of the 2D chamber was treated using fluoro-silane (Aquapel, Aquapel). Lastly, the chamber was dried under argon, filled with fluorinated oil and sealed until used. The chamber was reused multiple times and cleaned after each experiment by flushing fluorinated oil. This procedure resulted in a chamber having the following dimensions : 35 mm x 10 mm x 60 µm.

Supplementary Note 6: Modelling kinetics
We consider the coupled reaction as described in Figure 3a, Main Text: Glucose-6-phosphate dehydrogenase is a model enzyme for studies on bisubstrate reactions and the protein isolated from Leuconostoc mesenteroides, which can utilize NADH alongside NAD(P)H with comparable rates, has been characterized in detail. Levy and coworkers assigned ordered sequential mechanism with NADH and speculated about isomerization of enzyme [7] but afterwards revised the kinetics to random mechanism based on additional experiments [8]. Since ordered and random mechanisms are indistinguishable with respect to initial velocity expressions [9] and high mechanistic precision would not affect the outcome of the model, we assigned ordered mechanism as in [7] (Eq. 4 in Supplementary Note) and used the available Michaelis and inhibition (complex dissociation) constants.
Regarding the IMVs, the approaches for quantitative formulation of the electron transport chain and oxidative phosphorylation kinetics generally range from non-equilibrium thermodynamics [10] to kinetic models at various approximation levels [11]. The aerobic electron transport chain of E. coli comprises different NADH dehydrogenases and terminal oxidases linked through a quinone pool [12] and in some cases the respiratory complexes have been modeled through irreversible kinetics of Michaelis-Menten type [13]. Provided that we did not have information about the exact protein compostion of the IMVs and did not have sufficient experimental evidences to assign complex kinetics, we arbitrarily considered the IMVs as one enzyme complex, which oxidizes NADH and reduces oxygen and sought for the simplest kinetic expression that would describe the observed experimental behavior in terms of NADH oxidation functionality. The NADH and oxygen reaction rates depended linearly on IMVs concentration ( Figure S11b, S12s, S13c) but there was little dependence on the NADH concentration in the higher range (>250 µM, Figure S11a), which was used in microfluidic experiments. Therefore, based on the assumption that the level of oxygen was sufficiently high as discussed above, we assigned kinetics of Michaelis-Menten type, whereby the rate constant k 2 was defined at the given oxygen concentration. The oxygen transfer rate (OTR) was conventionally expressed (Eq. 8) through the constant k L a, whereby the saturation concentration was taken from oxygen measurements ( Figure S11b). The fitted mass transfer coefficient k L a was rescaled to be consistent with the well plate experiments in Figure 3 through the surface-to-volume ratio a. These assumptions led to the mathematical description: and: The parameter estimation was performed by minimizing the residual sum of squares RSS between simulation and experimental datasets, using the toolbox Copasi [14]. The optimization algorithm evolutionary programming was used to identify an approximation of a parameter set for a suitable global minimum of the RSS [15]. Additionally, the gradient orientated simplex algorithm was applied to certainly reduce the RSS into potential global minimum and optimal parameter set, respectively [16]. Since the RSS around the global minimum quickly exceed the confidence limits, characterized by a F-distribution with 95 percent upper α-critical value, n constants and m data points of measurement RSS(p) ≤ RSS(p * ) 1 + n m − n F 95% n,m−n , the fitted kinetic constants (p) are identifiable and can be considered as reliable in the respect confidence interval [17]. Supplementary