Introduction

All living things on Earth maintain their lives by using closed membranes, ranging in size from ∼1 to ∼100 μm. However, it is not yet clear why this scale is so ubiquitous. To gain insight into the effect of cell-sized confinement, many studies have been performed on model cellular systems. For example, gene expression has been monitored in liposomes1,2,3,4,5,6,7,8,9,10. Nomura et al. found that green fluorescent protein (GFP) gene expression is accelerated inside cell-sized giant liposomes (∼5 μm), by using giant liposomes obtained by natural swelling1. Despite this interesting observation, it has been rather difficult to evaluate the confinement effect in a quantitative manner, due to the technical difficulties of encapsulating desired amounts of proteins, DNA and substrates within a size-controlled confinement space. As a related phenomenon, GFP gene expression is known to be promoted in the presence of small phospholipid liposomes in the bulk solution11. Thus, it would be worthwhile to clarify the possible acceleration of protein production in a cell-sized confined space covered by a phospholipid layer.

Recently, cell-sized water-in-oil (W/O) droplets coated with a lipid layer have been shown as a useful system for analyzing the interaction between a lipid membrane and encapsulating bio-macromolecules12,13,14,15,16. It is easy to control the size of such a droplet system from 1 to 200 μm and the system is quite stable independent of the concentration of the entrapped solution and the type of lipid used, which is problematic when using a closed vesicle or liposome17. Furthermore, the droplet can be continuously monitored from soon after encapsulation for more than a day without evaporation18. These features enable us to analyze the size effect, lipid-dependence and time development. In fact, Fiordemondo and Stano indicated that GFP gene expression was enhanced in small droplets coated by lecithin19. Although this report is very interesting, the size dependence on the acceleration of GFP gene expression and its lipid-dependence have not been studied in a systematic manner.

In this study, we encapsulated a GFP gene expression system in cell-sized W/O droplets of different sizes (10−100 μm) and monitored the size-dependence of GFP expression based on the fluorescent intensity per unit volume using a confocal fluorescence microscope. Our results suggest that confinement within a volume that is on the size order of living cells is favorable for the production of protein, i.e., gene expression.

Results

Figure 1A shows confocal images of DOPG droplets entrapping a GFP expression system at 3 h after encapsulation. The fluorescent image (Figure 1A, left) shows the distribution of GFP fluorescence. The merged image (Figure 1A, right) shows that GFP fluorescence was homogeneously distributed in each droplet. Interestingly, the intensity of GFP fluorescence in droplets was strongly dependent on the droplet size. We calibrated the GFP intensity per volume and obtained the GFP concentration CGFP, based on normalization by the value for the bulk solution at 3 h. As shown in Figure 1B, CGFP decreased with an increase in the droplet size. In the case of a small droplet with radius R ∼10 μm (Figure 1B, 1), the value of CGFP was ∼4 times larger than that for a large droplet with a radius ∼45 μm (Figure 1B, 5). On the other hand, CGFP values in droplets with almost the same R ∼15 μm were similar (Figure 1B, 2, 3). We have confirmed that there was no size-dependence of CGFP among droplets under the encapsulation of GFP already expressed in the bulk solution (Supplemental Figure S1). Furthermore, this result indicates that the membrane surface and/or total volume significantly affects GFP expression, since the surface per volume ratio is greater with a smaller droplet. If the lipid membrane contributed to the size-dependence of GFP expression, the lipid species used should affect the relation between the droplet size and expression level. Thus, in the following experiments, we examined the effect of different lipids on the confinement effect.

Figure 1
figure 1

(A) Distribution of GFP fluorescence in DOPG droplets with different sizes, at 3 h after encapsulation. These cross-sectional images show (left) GFP fluorescence, (center) the oil/water interface and (right) the merged image. The radius of the droplets R is (1) ∼10, (2) ∼15, (3) ∼15, (4) ∼25 and (5) ∼45 μm, respectively. (B) Profiles of the GFP concentration, CGFP, along the diameter, where CGFP was evaluated from the GFP fluorescence intensity per unit volume normalized by the value of bulk solution at 3 h. The r. c. stands for relative concentration. Apparent sizes of the droplets are somewhat larger (∼10 μm) owe to the blurring effect in the fluorescent images.

Full size image

Three different species of lipids, anionic DOPG and zwitterionic DOPC and DOPE, were adapted to encapsulate the GFP gene expression system within droplets. The time-courses of GFP concentration per unit volume, CGFP, were monitored for ∼45 h beginning soon after encapsulation. Identical droplets with a radius R = 10−100 μm were monitored (Supplemental Figure S2). As shown in Figure 2, the value of CGFP linearly increased with time up to ∼20 h and reached a saturation point at ∼40 h. Among all three types of lipids, CGFP in a small droplet with R ∼20 μm was higher than that in a large droplet with R > 50 μm. In addition, CGFP at a certain time and the production rate VGFP ( = dCGFP/dt) were both dependent on the type of lipid.

Figure 2
figure 2

Time-courses of the GFP concentration per unit volume, C GFP , in small (radius R ∼ 20 μm; circle) and large ( R > ∼50 μm; square) droplets coated by a lipid layer of (A) DOPG, (B) DOPC, or (C) DOPE, respectively.

CGFP of the bulk solution is shown as a control (triangle). These data fit linear lines from 0 to 20 h and CGFP at 3 h in the bulk solution is taken as unity for all of the graphs.

Full size image

The time-courses of GFP expression levels in the bulk solution were also monitored by encapsulating the solution in PC droplets at each time point (black lines in Figure 2, Supplemental Figure S1). To avoid effects due to the membrane surface, CGFP was calibrated within ∼30 min after the encapsulation. In contrast to the expression inside droplets, GFP expression levels in the bulk at 25 h were quite low. For a PG droplet with R ∼27 μm, CGFP was more than 15 times higher than that in the bulk (Figure 2 A). The rate of acceleration, VGFP, in the bulk at <20 h was much less than that in droplets. These results suggest that a lipid membrane facilitates effective gene expression.

To clarify the accelerating effect of the membrane surface, CGFP at 25 h was plotted against 1/R in Figure 3A. The data are fitted by the equation CGFP = Cv + Cs(1/R), where Cv and Cs are variable constants related to the inner volume and membrane surface, respectively. Consequently, a linear correlation between CGFP and 1/R was observed with each lipid, unlike with the bulk solution (black line in Figure 3A). The slopes of the fitted line, Cs, were different among the three types of lipids and were approximately 1870 for PG, 1220 for PC, 370 for PE (i.e., PG > PC > PE). On the other hand, the intercept coefficients of the fitted lines in each lipid, Cv, were almost the same at ∼1, which means that CGFP in a large droplet with R ≫ 100 μm is closer to that in the bulk solution. These results clearly showed that the cell-sized confinement space surrounding the membrane surface accelerates GFP gene expression.

Figure 3
figure 3

The size-dependence of (A) the GFP concentration per unit volume CGFP at 25 h and (B) the production rate VGFP from 0 to 20 h, evaluated from the slope of linear increase on CGFP.

The lipids are DOPG (square), DOPC (triangle), DOPE (circle) and the bulk solution as a control (diamond).

Full size image

Finally, the GFP production rate, VGFP ( = dCGFP/dt), was plotted against 1/R in Figure 3B, where, the values of VGFP were deduced from the slopes of GFP expression levels per unit time up to ∼20 h (Figure 2). Figure 3B indicated that VGFP was linearly correlated with 1/R, i.e., VGFP = Vv + Vs(1/R), where Vv and Vs are the kinetics from the contributions of the volume and membrane surface, respectively. As a result, Vs was dependent on the lipid species in the order PG > PC > PE, whereas Vv was quite similar to that in the bulk (∼0.02 h−1). This result also suggests that the membrane surface strongly affects gene expression.

Taken together, our quantitative analysis clearly shows that the membrane surface in a cell-sized system accelerates GFP gene expression. It should be noted that this acceleration was not attributed from oxygen solved in oil, because the oil was held under an atmosphere of nitrogen gas before and during the preparation.

Discussion

Based on the experimental results shown in Figure 3, the GFP concentration per unit volume, CGFP and the production rate, VGFP, are both proportional to the reciprocal of the droplet size, 1/R. Thus, the reaction kinetics is characterized as 0th-order, being confirmed from the fact that VGFP remains almost constant for at least on the order of a day (Figure 2). 0th-order is a typical kinetics for the enzymatic reactions. As the next, we discuss the contributions of the kinetics from the bulk and surface. We denote N as a total amount of the enzymatic molecules, which catalyze the production of GFP under a constant speed in the presence of excess amount of substrates. To gain the essential insight on the reaction kinetics, we discuss the mechanism under a simple framework, i.e., the enzyme is a single species working under a large amount of substrates, i.e., N ∼ CGFP·4πR3/3. When N0 enzyme molecules are encapsulated in a spherical droplet with a radius R, they are partitioned into two groups related to the inner volume Nv and membrane surface Ns, as follows,

where α is an undetermined multiplier. Here, we assume fast equilibration on the partition. From homogeneous distribution of GFP within the droplet as noted in Figure 1, the production rate of GFP in the droplet is computed from a volume integral of VGFP, i.e., the product of total volume of the droplet, 4πR3/3. Since we assumed a 0th-order reaction, the production rate is also expressed by the sum of two different kinetics related to the inner volume a and the membrane surface b, as follow

By substituting eqs. (1) and (2) into eq. (3), we can obtain the following relation,

When the production rate on the membrane surface is more dominant than that in the inner volume, (i.e. b ≫ a) and the volume is conserved (i.e., 4πR3/3 = constant), the GFP production rate per unit volume, VGFP, can be expressed as

When the ratio R/α is larger than 1 (R/α ≫1), we can obtain the relation VGFP ∝ 1/R. Homogeneous distribution of GFP within the droplet (Figure 1) supports this assumption. Furthermore, the number of substrates N is also proportional to 1/R from the linear relationship between VGFP and N in eq. (3). In the case of droplets with large R in eq. (4), VGFP is only determined by the rate of production in the bulk, a, which also corresponds to the experimental result shown in Figure 3B. Thus, we can theoretically explain the dependence of the GFP concentration CGFP and the production rate VGFP on 1/R, under the above-mentioned assumptions including the conditions of b ≫ a.

The present study shows that protein expression is accelerated in a smaller confinement space, due to the membrane surface. Detailed molecular mechanism of the acceleration is not clear at present, because more than 100 factors organize the translational system20,21. Nevertheless, this result reminds us the importance of characteristic cellular size. Additionally, it can help us to realize the importance of membrane structures in cells. Organelles, such as mitochondria or the Golgi apparatus, increase the surface area by stacking membrane-bound structures. These are understood to hold a large number of membrane proteins22. Our results provide another reason for this increased surface area, i.e., the facilitation of protein expression on the membrane surface.

With regard to the lipid-dependence of GFP expression, we focus on the membrane surface potential under the presence of multivalent cations. PG is an anionic lipid and PC is electrically neutral. Although a PE membrane is slightly negative in water, the membrane potential becomes positive in the presence of divalent cations13,23,24. Therefore, the order of the lipid-dependence (PG > PC > PE) might originate from the membrane surface potential under physiological conditions: PE (positive) > PC (neutral) > PG (negative).

Furthermore, it is known that the cellular membrane has an asymmetric composition between the outer and inner leaflets25. It is known that, in many living cells including mammalian cells, PG and PE are found mainly in the inner leaflet and PC is in the outer leaflet26. The precise composition also varies among the organs25. Living things may tune this lipid composition to control the gene expression system desirable in each organ.

We have reported a remarkable acceleration of GFP gene expression in a cell-sized droplet coated with a lipid layer. The GFP concentration per unit volume and the production rate were both inversely proportional to the radius of the droplets. The degree of the size-dependence differed among lipid species that comprised the membrane surface and fell in the following order: PG > PC > PE. These results demonstrate that gene expression in a cell-sized droplet is accelerated in a cell-sized system due to the surrounding membrane surface. In this study, we showed marked acceleration of GFP gene expression by adapting Rabbit reticulocyte lysate as a cell-free translation system. Further experiments with various kinds of expression systems would have of scientific significance toward a further understanding on the underlying physic-chemical propertied on the cell-sized confinement.

Methods

Preparation of droplets encapsulating a GFP gene expression system

The water-in-oil (W/O) droplet coated by a lipid layer was prepared as described previously13,16. Negatively charged lipid, 1,2-dioleoyl-sn-glycero-3-phosphatidylglycerol (DOPG) and zwitterionic lipids, 1,2-dioleoyl-sn-glycero-3-phosphatidylcholine (DOPC) and 1,2-dioleoyl-sn-glycero-3-phosphatidylethanolamine (DOPE), were purchased from Wako Pure Chemical Industries (Osaka, Japan). A dry film of lipids was made on the bottom of a glass tube. Mineral oil (Nacalai Tesque, Kyoto, Japan) was added to the lipid film prior to sonication for ∼60 min at 50°C. To avoid oxidation27,28, the mineral oil was held under an atmosphere of nitrogen gas before preparation. The final lipid concentration in the oil was 0.5−1 mM. This procedure resulted in dispersed lipids in oil. To obtain W/O droplets encapsulating a GFP gene expression system, a 5−10 vol % aqueous solution of the GFP gene expression system was added to the lipid/oil solution and emulsification was performed by pipetting. As a GFP gene expression system, we used 40 μl of TNT rabbit coupled reticulocyte lysate system (Promega, Madison, WI), which included 1 μg plasmid encoding GFP (more than 200 molecules per 1 pl) (pCMV6-AC-GFP; Origene, Rockville, MD). GFP expression was performed at the optimal condition in reference to the experimental manual available from the manufacture. It is expected that diffusion of GFP (40 μm2/sec29) from surface to inner volume of droplet is much faster than the reaction (lower than 30 times per sec in 1 pL volume). The radius of the resulting W/O droplets ranged from 5 to 200 μm.

Microscopic observation

Bright-field and fluorescence microscopy images were obtained with an inverted confocal laser scanning microscope (LSM510; Carl Zeiss, Jena, Germany). GFP was excited with an argon laser (488 nm) and images were obtained through a 505 nm long-pass filter (Chroma, Rockingham, VT). All images show equatorial sections of the droplets. The physical size of pinhole aperture was fixed at ∼20 μm to obtain an intensity that was high enough for analyses. The GFP intensity per unit volume, IvGFP, was obtained from the center of the cross-section image showing droplets with a relatively large radius, R >10 μm, to avoid artifacts due to curvature effects. As a control data, IvGFP of GFP already expressed in the bulk solution was also monitored under the encapsulated in PC droplet. To avoid effects due to the membrane surface, the value of the bulk solution was calibrated at each time point within ∼30 min after the encapsulation. We refer to the IvGFP normalized by the control data at 3 h as the GFP concentration CGFP (r. c., relative concentration). The time-course of CGFP in the droplets was continuously monitored beginning soon after encapsulation for ∼45 h at 30°C. We then analyzed the temporal differentiation of CGFP and obtained the production rate VGFP ( = dCGFP/dt).