www.acsnano.org
Microemulsion-Based Soft Bacteria-Driven Microswimmers for Active Cargo Delivery Ajay Vikram Singh, Zeinab Hosseinidoust, Byung-Wook Park, Oncay Yasa, and Metin Sitti* Physical Intelligence Department, Max Planck Institute for Intelligent Systems, Stuttgart 70569, Germany S Supporting Information *
ABSTRACT: Biohybrid cell-driven microsystems offer unparalleled possibilities for realization of soft microrobots at the micron scale. Here, we introduce a bacteria-driven microswimmer that combines the active locomotion and sensing capabilities of bacteria with the desirable encapsulation and viscoelastic properties of a soft double-micelle microemulsion for active transport and delivery of cargo (e.g., imaging agents, genes, and drugs) to living cells. Quasimonodisperse double emulsions were synthesized with an aqueous core that encapsulated the fluorescence imaging agents, as a proof-of-concept cargo in this study, and an outer oil shell that was functionalized with streptavidin for specific and stable attachment of biotin-conjugated Escherichia coli. Motile bacteria effectively propelled the soft microswimmers across a Transwell membrane, actively delivering imaging agents (i.e., dyes) encapsulated inside of the micelles to a monolayer of cultured MCF7 breast cancer and J744A.1 macrophage cells, which enabled real-time, live-cell imaging of cell organelles, namely mitochondria, endoplasmic reticulum, and Golgi body. This in vitro model demonstrates the proof-of-concept feasibility of the proposed soft microswimmers and offers promise for potential biomedical applications in active and/or targeted transport and delivery of imaging agents, drugs, stem cells, siRNA, and therapeutic genes to live tissue in in vitro disease models (e.g., organ-on-achip devices) and stagnant or low-flow-velocity fluidic regions of the human body. KEYWORDS: microswimmers, targeted drug delivery, double emulsions, golgi tracker, soft robots
M
targeting bacteria to accumulate in and subsequently destroy tumors by using the naturally occurring gradients inside the human body.18,19 Passing the proof-of-concept stage toward realistic biomedical applications, however, requires the design of functional synthetic materials that can work in concert with the biological cells effectively and efficiently to perform the desired functions. For applications in gene and drug delivery and biomedical imaging, microemulsions have been proven to provide many significant advantages.20,21 Their soft and viscoelastic properties allow better interaction with tissue cell membranes,22 they can encapsulate a wide range of hydrophilic and hydrophobic cargo and protect their payload against harsh environments,23 and they allow for the free diffusion of the cargo at oil−water interface into the tissue cells.24,25 Compared with nanoemulsions, microemulsions can carry 3−6 orders of magnitude higher amounts of cargo load, and they are mechanically softer; this means they can significantly deform and squeeze through micropores, conform well to the microscale topography of the
iniaturized soft robots have significant potential to revolutionize biomedical technology and improve the quality of human life.1−4 Despite notable advances, the realization of micron-scale soft robots is hindered by the extreme difficulty of designing functional autonomous systems at such small scales.5−8 Biological cells, however, function autonomously and efficiently at the micro/nanoscale, and thus integrating them into synthetic constructs allows for the exploitation of their complex functionalities such as active locomotion and sensing for designing multifunctional autonomous microsystems. Various biological entities such as muscle cells,9 algae,10 and bacteria11,12 have been integrated with synthetic materials to design functional biohybrid microsystems.13 Bacteria are of particular interest for the development of biohybrid microsystems, such as microswimmers,14,15 microcrawlers,14 and micropumps,16 because they can be genetically engineered relatively easier than eukaryotic cells, because of their robust propulsion and sensing capability in diverse physiological environments and their ability to harness the biochemical energy source, that is, ATP, inside their cell membrane to propel themselves.17 The random trajectory of bacterial swimming can be guided by environmental gradients of nutrients, oxygen, or pH; a property that has enabled tumor© 2017 American Chemical Society
Received: March 26, 2017 Accepted: August 31, 2017 Published: August 31, 2017 9759
DOI: 10.1021/acsnano.7b02082 ACS Nano 2017, 11, 9759−9769
Article
Cite This: ACS Nano 2017, 11, 9759-9769
Article
ACS Nano
Figure 1. Schematic representation of the development process for double emulsion-based soft bacteria-driven microswimmers. (A) Preparation of w/o/w double microemulsions loaded with cargo inside their water cores and decorated with streptavidin on their outer oil shells. (B) Attachment of motile bacteria to the double microemulsion outer oil shells through the specific biotin−streptavidin interaction. (C) Collection of quasi-monodisperse bacteria-driven double microemulsions using Transwell membranes with given pore sizes.
imaging agents, genes,36 stem cells, and RNA. We report cargo delivery by a biohybrid microswimmer composed of double emulsions driven by bacteria that are steered by an external concentration gradient toward the cells of interest. The developed soft biohybrid microswimmers are composed of w/o/w microemulsions conjugated to motile Escherichia coli (E. coli) bacterial cells. We synthesized quasi-monodisperse, biocompatible, and biodegradable double emulsions and loaded their aqueous cores with various organelle-labeling dyes, as the proof-of-concept cargo demonstration in this study. Next, we decorated their outer oil shells with streptavidin for specific and strong attachment of biotin-conjugated E. coli bacteria for efficient fluidic propulsion. The biohybrid microswimmers swam across a barrier of microporous membrane and efficiently delivered organelle-tracking dyes to the living cells, which enabled real-time live-cell imaging of internal cell organelles. In the following text, the phrase microswimmer refers to bacteria plus microemulsions.
cell surface when they come in to contact, and burst after contact, which enables efficient delivery of the given cargo to the cells through diffusion. Moreover, microemulsions are neutrally buoyant, biocompatible, and biodegradable,26 which are essential properties for biomedical applications. Colloidal drug carriers based on oil-in-water (o/w), water-in-oil (w/o), and water-in-oil-in-water (w/o/w) double microemulsions are well established.27,28 Single-emulsion systems, such as liposome-based drug delivery vehicles, have a lower cargo capacity, lower stability, and lower monodispersity.29,30 Contrary to this, stable w/o/w double emulsions have been developed as a biologically relevant alternative to w/o single emulsion for the encapsulation of hydrophilic cargo. They can carry hydrophilic cargos in their aqueous core, while their external oil shell allows for a stable dispersion in aqueous media and enables a better interaction with the tissue cell membranes.31 Additionally, they can be mass-produced in various sizes in a cost-effective manner and with a reasonably narrow size distribution.32 What makes the double microemulsions particularly useful for the developing biohybrid soft microrobots is the fact that the oil−water interface at the outer shell can be decorated with specific molecules to facilitate stable conjugation of biological cells to the emulsions.33 Despite these advantages, there is no prior report of using double emulsion systems driven by motile bacteria to deliver cargo to live cells. Microswimmers (micron-scale swimming robots) have many promising biomedical applications,5,34,35 such as active and/or targeted transport and delivery cargo, for example, drugs,19
RESULTS AND DISCUSSION Synthesis of Double Emulsions for Biohybrid Microrobots. In this work, we developed versatile double emulsions in terms of stability, quasi-monodispersity, and surface functionalization with biologically relevant adhesive molecules to realize biohybrid microrobots. Double emulsions were prepared using water and soybean oil. The preparation steps are presented in Figure 1A. Vegetable oil was chosen over the more popular silicone oil because of its biocompatibility.37,38 9760
DOI: 10.1021/acsnano.7b02082 ACS Nano 2017, 11, 9759−9769
Article
ACS Nano Table 1. Physicochemical Properties of Synthesized Double Microemulsions volumetric ratio of aqueous phase to w/o emulsions
% hydrophilic surfactant (high HLB)
% lipophilic surfactant (low HLB)
HLB of surfactant mixture ratio
4:1 4:1 4:1 4:1 5:1 5:1 c 5:1
1.0 0.5 0.25 0.1 1.0 0.5 0.5
0.1 0.25 0.50 1.0 1.0 0.5 0.5
10.6 6.25 5.5 6.1 16 8 8
a
emulsion diameter (μm)b
PIa
ζ potential (mV)b
± ± ± ± ± ± ±
0.4 0.3 0.4 0.5 0.6 0.4 0.3
+4.3 ± 0.9 +1.7 ± 0.3 +0.7 ± 0.1 +1.2 ± 0.6 +2.9 ± /1.1 +1.5 ± /0.8 +1.1 ± /0.1
7.0 6.4 6.1 5.7 4.9 5.2 5.5
1.0 0.5 0.9 0.7 0.9 0.1 0.3
PI: polydispersity index. bData = mean ± SD (n = 5). cwith streptavidin.
Figure 2. Size distribution and protein adsorption for the synthesized double microemulsions. (A) Fluorescence confocal micrograph overlay with differential interference contrast micrograph, showing thin-shelled w/o/w double microemulsions; inset: a higher magnification image of single microemulsion, showing the shell thickness. Scale bar represents 10 μm. (B) Particle size distribution of w/o/w double microemulsions, as determined by DLS. Data compiled with a bin width of 1. (C) Percent fraction of streptavidin adsorbed at the oil droplet shell as a function of the weight percent (concentration) of the protein in the final aqueous phase. Error bars represent standard deviation from the mean for three independent measurements. Fluorescent microscope images of quasi-monodisperse microemulsions attached to (D) one and (E) two bacteria, stained with SYTO9 dye. 2D swimming trajectories (red line) of the microswimmers propelled by (F) one and (G) two bacteria are shown on phase-contract optical microscope images. Scale bar represents 5 μm.
microemulsions. Treatment intensity and duration were optimized for each layer, namely, the internal w/o phase and the final w/o/w emulsion to maximize stability and the monodispersity of the microemulsions. We fine-tuned the treatment intensity (Table S2), duration of treatment (Table S3), and the loading efficiency of the model cargo, rhodamine B (Table S4). Sonication intensity of 25 W was chosen for the
Most double emulsions reported in the literature have been reported based on silicone oil, therefore main design parameters such as water-to-oil ratio, agitation method and duration, surfactant combinations, and surfactant ratios had to be optimized for our new design. Based on the optimized parameters, presented in Tables S2−S5, a combination of sonication and vortex mixing was used to prepare the double 9761
DOI: 10.1021/acsnano.7b02082 ACS Nano 2017, 11, 9759−9769
Article
ACS Nano
using ImageJ, after image segmentation (clustering) of the videos obtained from fluorescent light microscopy (Figure S2). The hydrodynamic diameter was determined indirectly based on the Brownian motion, and thus a complete agreement between data presented in Figure 2B and Figure S2 is not expected. Various batches of double emulsions were prepared, encapsulating Mito-, Golgi-, and ER-tracker dyes with entrapment efficiencies of 21%, 30%, and 41%, respectively. The standard curves in Figures S3−S5 were used to calculate the encapsulation efficiency of rhodamine B uptake, based on the measured fluorescence intensities. Integration of Bacterial Cells with Microemulsions. The outer oil shell of the double emulsions was functionalized with streptavidin to allow for specific attachment of chemically biotinylated E. coli MG1655 with their cell membrane, not with their flagella, the appendage used for bacterial swimming function. For this purpose, the cargo-loaded w/o emulsions were dispersed into aqueous medium containing different concentrations of streptavidin and 0.1% CTAB. Figure 2C shows the total streptavidin adsorbed to the oil shell as a function of the concentration of streptavidin in the medium. Nearly half of the protein was adsorbed to the microemulsion surface at 0.5 wt % streptavidin concentration. Presence of the high HLB surfactant CTAB is expected to be one of the limiting factors affecting protein adsorption by creating spatial hindrance in the oil phase.42 Protein adsorption at the oil/water interface is a complex process, and therefore it is difficult to identify which characteristics associated with streptavidin are important for the interfacial adsorption at the oil shell. The triglyceride polarity and high viscosity of oil phase could additionally relate the inverse adsorption of protein to the protein concentration.27 The w/o/w double emulsions were subsequently washed and diluted into motility media containing biotinylated E. coli MG1655 (OD600 = 0.5) (Figure 1B). The bacteria were labeled with SYTO9 for facile imaging and tracking. Once attached, the bacteria-emulsion construct was imaged using fluorescence microscopy to confirm the attachment of bacteria to the microemulsions (Figures 2D and 2E). At this stage, the microemulsions were separated into fractions using membranes with various pore sizes, and only the fraction with a size range of 5 to 7 μm was used for active cargo delivery experiments (Figure 1C). The microemulsions in this size range bound to 1−2 bacterial cells, as shown in Figure 2D,E. The size fractionation allowed for more uniform swimming patterns and thus a more controlled investigation in the developmental stage of the system. Propulsion Performance of Bacteria-Driven EmulsionBased Microswimmers. The swimming trajectories of bacteria-propelled microemulsions (microswimmers) are shown in Figure 2F,G. Microemulsions without bacteria exhibited an average speed of 0.5 ± 0.2 μm/s, which can be attributed to the Brownian diffusion.43 Microemulsions attached to a single bacterium swam with an average speed of 6.5 ± 0.8 μm/s in the motility media (Figure 2F and Supplementary Movie S1), whereas microemulsions attached to two bacteria swam at 4.5 ± 0.2 μm/s (Figure 2G and Supplementary Movie S1). The reduction of average speed could be attributed to the canceling of propulsion force vectors by multiple bacteria attaching randomly around the micelle. In the case of single bacterium attached to the emulsion surface, the microswimmer exhibits a random zigzag movement due to the rotation of the particle and the stochastic “run” and “tumble” swimming mode of the attached E. coli, as observed in
preparation of the internal core of the double microemulsions because it led to w/o emulsions that were more stable through the second round of mechanical treatment (Table S2). For the second treatment step, vortex mixing was the method of choice; the optimal duration of vortex mixing was chosen to be onethird of the initial sonication treatment to avoid the distortion of the w/o core (Table S3). This preparation method was used to prepare double emulsions loaded with rhodamine B as the cargo to deliver into the tissue cells. As demonstrated in Table S4, a treatment involving a ratio of vortexing time to sonication time equal to one-third led to a higher encapsulation efficiency of rhodamine B (25%), suggesting that using a shorter vortexing duration likely limited the disruption of the w/o core during the preparation of the double emulsion and results in better retention of the cargo. The standard curve in Figure S1 was used to calculate the encapsulation efficiency of rhodamine B uptake by double emulsions, based on the measured fluorescence intensities. The concept of hydrophilic−lipophilic balance (HLB) was used to choose appropriate surfactants. All surfactants used in optimization experiments are listed in Table S5 along with their corresponding HLB values. A high HLB index indicates a higher solubility in water, whereas a low HLB index indicates a non-polar and/or oil soluble surfactant.39 Various surfactant combinations were investigated, and a range of surfactant ratios were tested for preparation of the stable double emulsions. The combined HLB index of the microemulsions was calculated as the weight-average of the HLB index of the two surfactants. The chosen surfactant pair had a combined optimum HLB index equally distanced from the HLB index of its constituents.40 This result is presented in Table 1 for the chosen surfactants: cetyltrimethylammonium bromide (CTAB) and Merpol A. The lowest possible CTAB concentration in the outer shell was implemented because of the anticipation that this surfactant might negatively affect the conformation of streptavidin integrated into the outer shell (to increase bacterial attachment) or inversely affect viability of bacteria and/or cells that come into contact with the emulsions at later stages. Finally, the optimal volumetric ratio of the outer aqueous phase to the inner w/o emulsion was found to be 4:1 (Table 1). For the w/o core, an oil-to-water ratio of 2:1 was chosen based on the recommendations in the literature.41 The chosen optimal preparation conditions led to a positive ζ potential that helped to create also a slight electrostatic attraction between the negatively charged bacterial cells and the positively charged emulsions. Further, the mutual repulsion between emulsion shells prevented coalescence and stabilized the microemulsions. To further increase the efficiency of bacterial attachment to the micelles, streptavidin was added to the outer oil shell during emulsion preparation (Figure 1B). We did not observe a significant change in ζ potential after the functionalization of the outer shell with streptavidin (+1.5 ± 0.8 versus +1.1 ± /0.1; p-value = 0.08). For all subsequent experiments, the microemulsions were prepared based on the optimum conditions identified at this stage. The microemulsions, prepared using the identified optimal conditions exhibited a mean oil shell thickness of 500 ± 31 nm determined using fluorescence microscopy images of Nile redinfused oil shells (Figure 2A). Figure 2B shows the hydrodynamic diameters of the synthesized microemulsions determined using dynamic light scattering (DLS), ranging between 3 and 10 μm with an average polydispersity index of 0.3. The diameter of the double emulsions was also quantified 9762
DOI: 10.1021/acsnano.7b02082 ACS Nano 2017, 11, 9759−9769
Article
ACS Nano
Table 2. Comparison of Effective Diffusion Constants (Deff) and Mean Speeds (Vmean) for Microemulsions Driven by No, One, and Two Bacteria double emulsion diameter (μm)
Deff for double emulsions with no attached bacteria (m2/s)
Deff for double emulsions with single E. coli (m2/s)
Deff for double emulsions with two E. coli (m2/s)
Vmean for double emulsions with single E. coli (μm/s)
Vmean double emulsions with two E. coli (μm/s)
Vmean for double emulsions with no attached bacteria (μm/s)
5.0 ± 0.1 7.0 ± 0.3
2.5 × 10−14 2.1 × 10−13
8.3 × 10−10 1.6 × 10−10
7.5 × 10−10 7.7 × 10−10
7.7 ± 1.0 6.5 ± 0.8
7.1 ± 0.9 4.5 ± 0.2
0.53 ± 0.2 0.50 ± 0.2
Figure 3. (A) Schematic diagram of the in vitro test setup used to investigate the active cargo transport and delivery of the microswimmers. (B) Scanning electron micrograph from top surface of the Transwell membrane, showing bacteria accumulation around the pores of the membrane. Scale bars represent 20 μm. (C) A single cluster of bacteria around the membrane pores, imaged from bottom. Scale bars represents 20 μm. (D) Results of the MTT assay showing decrease in viability of MCF7 cells as a result of incubation with the microswimmers and their components. Error bars represent standard deviation from the mean for three independent measurements.
our previous studies.44,45 Microswimmers with two attached bacteria showed a smoother swimming trajectory. Further details of the mean speed and effective diffusion of the microswimmers and their associated parameters are given in Table 2. The effective diffusion constant of bacteria-driven and passive microemulsions in Table 2 is one of the essential parameters that determines their cargo delivery efficiency. If the carriers diffuse fast, then they could reach and contact tissue cells in their vicinity faster in high numbers and deliver the cargo in high amounts after the contact. Thus, because the effective diffusion constant of bacteria-driven emulsions is 3−4 orders of magnitude higher than the passive emulsions, their cargo delivery efficiency would be significantly higher than the passive emulsions, widely used in the literature for cargo delivery. Therefore, the effective diffusion constant of the microswimmers should be maximized as much as possible for the optimal cargo delivery by selecting the type of bacteria with high mobility and optimizing the emulsion size and the number and attachment locations of multiple bacteria.45 To assess the stability of bacteria attachment on the microemulsions under the prevailing physicochemical conditions, a fraction of double emulsions with a larger hydrodynamic diameter of 10 μm and 4- to 5-attached bacterial cells per microemulsion were monitored for 4 h. The results for microemulsions with or without streptavidin indicate that
streptavidin increased the stability of bacterial attachment to the microemulsions, but no significant decrease in detachment was observed in either case (Table S6). However, when we compared the average number of E. coli attached at each time point in the absence and presence of streptavidin with increasing time from 0 to 4 h using the student t test, we see a significant difference (t test = 0.03; p-value ≤0.05). This indicates stable biohybrid bonding over time. In Vitro Delivery of Imaging Contrast Agents to Cells by the Soft Microswimmers. We created a concentration gradient of glucose across a Transwell membrane, where the bacteria-driven microswimmers then swam up the glucose concentration gradient toward the cancer cells. The control, in this case, is microswimmers randomly swimming in the absence of the concentration gradient. Such gradients are naturally present in the body; for example, tumor environments possess a higher local glucose and a lower oxygen concentration. To investigate the performance of the developed microswimmers for in vitro delivery of imaging agents to a monolayer of cultured MCF7 breast cancer and J744A.1 macrophage cells, the soft microswimmers, composed of E. coli MG1655 bacterial cells and double microemulsions loaded with dyes (e.g., Mito-, ER-, and Golgi-trackers), were introduced into a Transwell (Figure 3A). Here, a Transwell membrane is used for creating quasi-monodisperse emulsion swimmers, as a size filter. The microswimmers actively swam through the Transwell’s bottom 9763
DOI: 10.1021/acsnano.7b02082 ACS Nano 2017, 11, 9759−9769
Article
ACS Nano
Figure 4. (A) Control fluorescent images represent cell cultures treated with Mito-tracker dyes as per the manufacturer’s routine protocols. In the fluorescent images, stained nuclei are shown with blue color (DAPI); mitochondria are shown with red color (Mito-tracker); and overlay images show combined color for the nuclei and mitochondria. (B) Microswimmers, loaded with Mito-tracker imaging dyes, delivered the dyes to the breast cancer cell line MCF7 after overnight incubation with cells. Scale bar represents 40 μm.
Table 3. Comparing the Cell-Organelle Staining Using Active Dye-Loaded Microswimmers-Based Dye Delivery versus Conventional Direct Staining As Positive Control and Passive Dye-Loaded Double Emulsions (without Attached Bacteria) As Negative Control mean intensity for positive control
mean intensity for samples labeled by dyeloaded emulsions without attached bacteria
mean intensity for samples labeled by microswimmers
500 nM
7.0 ± 1.3
2.3 ± 0.9
4.2 ± 1.1
1 μM
4.1 ± 0.3
1.6 ± 0.3
3.4 ± 0.8
50 particles per cell (PPC)
2.8 ± 0.3
1.2 ± 0.7
2.2 ± 0.1
working dye concentration for microswimmers, double emulsions without bacteria, and direct addition to culture Mitotracker ERtracker Golgitracker
random attachment of E. coli to the membrane (Figures S6A,B), which indicates that bacterial congregation around the pores in Figure 3 was most likely driven by the bacteria actively swimming up the glucose gradient. The biohybrid soft microswimmers and their individual components were assessed for potential adverse biotoxicity effects on the viability of the cultured MCF7 cells with an MTT assay, which quantified the mitochondrial activity of the cells and thus indirectly measured their potential toxicity toward the cell line. The results in Figure 3D indicate that the microswimmers had a lower adverse effect on the mitochondrial activity of the cultured cell lines than the free bacteria, possibly because of the decrease in the rate of uptake by the cells compared to free bacteria. Once they crossed the membrane, the microswimmers were further able to swim and deliver the organelle tracker dyes to a monolayer of cultured MCF7 breast cancer and J744A.1 macrophage cell lines. First, the microswimmers carrying organelle tracker dyes (Mito-trackers) were incubated with the MCF7 cell line overnight to investigate their performance for mitochondria-labeling dye delivery, as shown in Figure 4A,B. Next, the same experiment was repeated with the
membrane and reached the cells at the bottom plate, guided and steered by an external glucose gradient. The pore size of the Transwell membrane was selected to be 8 μm so that microswimmers with diameters of 5−7 μm could pass through the membranes with high efficiency (Figure S2). The microemulsions had a lower density than the cell culture media, hence, in the absence of bacterial propulsion, the buoyancy was dominant, and a very small number of passive microemulsions could pass through the membrane and come into contact with the cells at the bottom. Figure 3B,C shows bacterial cells clustered around the pores of the membrane, where the microemulsions could not be visualized using a scanning electron microscope (SEM). As shown in Supplementary Movie S2, the same E. coli clusters were evident in fluorescence time-lapse videos. This movie was recorded from bottom plane of the culture dish and showed an increase in florescence intensity over time, indicating the clustering of microswimmers (containing SYTO9-labeled bacteria) around the pores of the membrane. We verified that the congregation of bacteria around pores is not a passive drying effect during SEM sample preparation by running the same experiment in the absence of a glucose concentration gradient. We observed 9764
DOI: 10.1021/acsnano.7b02082 ACS Nano 2017, 11, 9759−9769
Article
ACS Nano
dispersed into cytoplasm.49,50 As shown in Supplementary Movie S3, microemulsion drops attached over cell surface may also undergo membrane rupture due to emulsion shell instability arising from the surrounding hyper osmolar cell culture media and thus releasing organelle tracer dye on the cell surface freely diffusing through the J744A.1 membrane and leading to staining the specific organelle. Further improvement of the system could be achieved by functionalizing emulsion shells with organelle specific antibodies and cell recognition sites to facilitate the intracellular trafficking and delivery. To show that bacteria-driven cargos could be delivered across future in vivo barriers, we assessed the shape deformation and squeezing capability of our soft microswimmers using Transwells with membrane pore diameters of 3, 5, and 8 μm, where 3 and 5 μm pores were smaller than the diameter of the microemulsions (around 5−7 μm). We quantified the number of fluorescently stained microswimmer emulsions passing through the membrane pores by measuring the mean fluorescent intensity of swimmers under the membrane region over 30 min. As shown in Figure 5A, only around 18% less
microswimmers carrying and delivering the Mito-, ER-, and Golgi-trackers to the monolayer of J744A.1 macrophage cells to label the mitochondria, endoplasmic reticulum, and Golgi body, respectively (Figure S7). For positive control, we treated both cell cultures with given tracker dyes as per the manufacturer’s routine direct labeling protocols, which relies on delivering a large amount of dyes to the cells via passive diffusion (Figure 4A−C and Figure S7). We also incubated cells with bare microemulsions (without bacterial attachment) as a negative control, and we found fluorescence intensity was significantly less compared with microswimmers (t < 0.03, p ≤ 0.05) and positive control (t = 0.01, p ≤ 0.05). Representative fluorescence images for macrophages are shown in Figures S7 and S8 for each dye, and the results for mean fluorescence intensities, based on 20 frames from each sample, are tabulated in Table 3. Here, as indicated by comparable mean fluorescence intensities calculated from histograms, microswimmers effectively labeled the internal organelles of the live cells, comparable with standard direct staining methods. On the other hand, active dye-loaded microswimmers labeled the cells with much higher efficiency compared to the passive dyeloaded microemulsions without E. coli attachment (see Figures S7 and S8 and Table 3). We did not demonstrate selectivity in the presence of other cells. Such selectivity can be achieved by using genetically engineered bacterial cells that display surface proteins/peptides targeting biomarkers of interest expressed over tumor cells. To understand the mechanism of the dye/cargo delivery to the cells by the soft microswimmers, internalization of the rhodamine B-encapsulated microswimmers with macrophages was recorded as a time-lapse series of fluorescence images for 30 min after the addition of the microswimmers onto the cell monolayer. Supplementary Movie S3 shows the internalization of the microswimmers by macrophages. The movie demonstrates that, after the physical contact with the cell membrane, soft microswimmers conform to the cell surface and collapse/ burst, due to the osmotic pressure from extracellular to intracellular regions.46 The released dye/cargo molecules then diffuse inside the cells efficiently. In the case of macrophages, the bacteria that were attached to the swimmers are also internalized by macrophages. Furthermore, Figures S9 and S10 show time-lapse fluorescence microscopy images of the uptake of microswimmers and passive microemulsions, respectively, by macrophage cells, as indicated by the fluorescence intensity detected inside the cells. Internalization of the microswimmers could be triggered by sensing of bacteria or their secreted product on the surface of macrophage cells.47,48 Supplementary Movie S4 shows the 3D confocal microscope reconstruction of the macrophages subjected to the rhodamine B-encapsulated microswimmers after 2 h and overnight incubation, respectively, confirming the internalization of the delivered cargo. Our team is currently investigating the pathway that could be leading the cascade for the uptake. Rhodamine B-labeled microemulsions incubated with macrophages show uptake and phagocytosis, which might be triggered by sensing of bacteria or their secreted product at the surface of macrophage cells.47,48 Once internalized, the in vitro double-emulsion systems, shown here, exhibited successful mitochondria, endoplasmic reticulum, and Golgi body staining with the delivered fluorescent dyes. The delivery of cell tracers might occur on account of bursting of internalized emulsion drops due to a difference in osmolarity inside drops versus intracellular environment, which facilitates the intracellular free diffusions of the dyes toward organelles
Figure 5. Demonstration of squeezing and deformation capability of rhodamine B-labeled bacteria-driven microemulsions through Transwell membrane pores with 3, 5, and 8 μm diameters. (A) Rhodamine intensity measured by incubating cells with bacteriadriven double emulsions (white columns) and passive (no bacteriaattached) double emulsions (gray columns), as control. (B) Optical microscope image of a soft double emulsion squeezing and passing through a microfluidic channel smaller than its diameter via shape deformation, when high enough flow or pressure is applied. No bacteria were attached to the emulsion surface in this demonstration. Scale bar represents 5 μm.
bacteria-driven microemulsions could squeeze and pass through the 5 μm pores, compared with the case of 8 μm pores. However, we saw much less microswimmers passing through 3 μm pores (around 65% less than the case of 8 μm pores), indicating that reducing the pore diameter by around 50% of the size of the microswimmers significantly reduced their barrier passing capability. Therefore, to be able to pass through 9765
DOI: 10.1021/acsnano.7b02082 ACS Nano 2017, 11, 9759−9769
Article
ACS Nano
microrobotic active or targeted cargo delivery systems can be used for applications such as active or targeted cancer chemotherapy, gene therapy, high-resolution medical imaging, in situ tissue scaffolding and growth.53 Microswimmer optimization, medical imaging, and localization issues for such specific medical uses would be investigated in small animal models in the future. There is certainly a safety risk associated with using microorganisms as therapeutics. The risk, however, is minimized if the microorganism is used in body niches that are not sterile. Oral delivery appears to be the safest route, as demonstrated by the success of probiotics on the market. We have outlined the associated risks in our recent review.17
significantly (50% or more) smaller microbarriers, relative to the microswimmer diameter, the synthesized emulsion size needs to be reduced accordingly. To further verify the shape deformation and squeezing capability of the microemulsions, we passed the larger microemulsions through microfluidic channels that were smaller than the emulsion diameter. We observed the shape deformation of the microemulsions (Figure 5B). The developed soft microswimmers thus offer the efficient transfer of the cargo to the cell membrane, enabled by the soft double emulsions to achieve an easily detectable level of organelle labeling. Use of these microswimmers for delivery of therapeutic and imaging agents can decrease the administered dosage and therefore reduce the potential adverse effects of the cargo on the living tissues. Further investigations are needed to gain a better understanding of the interactions of bacterial cells with the fluid mosaic structures and to assess the stability and functionality of the developed system under varying environmental conditions, relevant to biomedical applications.
MATERIALS AND METHODS Preparation and Characterization of w/o/w Double Emulsions. The double emulsions were prepared using a modified version of a previously reported method.54 Briefly, Millipore water and soybean oil (VWR) were added to a vial with water-to-oil ratio of 1:2 along with the chosen anionic surfactant and mixed by sonication (intensity and duration were optimized). The mixture was then supplemented with water and cationic surfactant of choice (ratios optimized) and mixed by sonication or vortexing. Size distribution and stability of microemulsions were optimized by tweaking water-to-w/o ratio, surfactant pairs, sonication intensity and duration, and the duration of vortexing. The appropriate surfactant pair was chosen based on the calculated composite HLB index as the weight-average values of the respective HLB of each component. The hydrodynamic diameter of the microemulsions was determined using DLS (Mastersizer analyzer, Malvern), and the ζ potential was calculated from measurements of electrophoretic mobility performed by the same equipment. Nikon Eclipse Ti Confocal Microscope with Yokogawa CSU-W1 spinning disk was used for imaging double emulsions loaded with rhodamine B (Ex/Em 553/627). The images were analyzed using ImageJ to calculate the thickness of the outer shell (indicated by fluorescent Nile red dye). Assembly of Double Emulsions and Bacterial Cells. Escherichia coli (E. coli) MG1655 (ATCC700926) was inoculated from glycerol stock onto 0.5% LB-agar plate and after overnight incubation transferred to a motility medium (0.01 M potassium phosphate, 0.067 M sodium chloride, 10−4 M EDTA, 0.01 M glucose, and 0.002% Tween-20, pH 7.0) containing 3 mg of dissolved biotin (Sulfo-NHS-LC-Biotin, Life Technologies) and grown overnight at 37 °C in a shaker incubator. The bacteria were subsequently stained with SYTO9 (Ex/Em 485/498) by mixing the dye (prepared as per instructions from the vendor) and the bacterial suspension 1:1, incubating for 15 min, and gently washing the cells three times with PBS to remove excess SYTO9. The streptavidin-coated double emulsion was added to 1 mL of the overnight bacterial suspension (108 cfu/mL) and incubated at room temperature for 45 min with gentle shaking to allow for biotin−streptavidin binding. Loading and External Functionalization of the Double Emulsions. To load the microemulsions, the fluorescence dye of choice, namely Mito-tracker Red FM (Invitrogen, Ex/Em: 581/644), ER-tracker (Invitrogen, Ex/Em: 587/615), Cell light Golgi-GFP (Invitrogen, Ex/Em: 488/510), rhodamine B (Sigma, Ex/Em: 553/ 627), or Nile-red (Sigma, Ex/Em: 485/525) for measuring oil shell thickness was dissolved in Milli-Q water or vegetable oil at a concentration of 10 μg/mL each. Merpol A was added to this solution at a final concentration of 0.5% (v/v). The dye solution was subsequently added to soybean oil and mixed with probe ultrasonication (Branson Sonifier 250) for 7.5 min. This w/o emulsion was washed to remove excess dye and added to 2.5 volumes of phosphate buffered saline (PBS, pH 7.2) containing 0.1% (v/v) of CTAB and 0.5 mg/mL streptavidin. The mixture was mixed using a vortex for 2.5 min. The durations and concentrations were determined through a rigorous optimization process. Loading efficiency of each dye was determined by breaking up the micelles with addition of 50% (v/v) ethanol and measuring the fluorescence intensity using a Synergy 2
CONCLUSIONS The proposed bacteria-driven emulsions-based microswimmers demonstrated the feasibility and promise of soft, biocompatible, and active carriers to deliver cargo with high efficiency to tissue cells. Due to their active propulsion, they had 3−4 orders of magnitude larger effective diffusion constant, which means they could reach and contact the cells in their vicinity significantly faster and more efficiently than the emulsions diffusing passively with no propulsion. Moreover, they had a highly efficient cargo delivery mechanism, close to routine direct staining methods, due to their soft emulsion bodies, where the mechanically soft swimmers are believed to conform to the tissue cell surfaces well when they contacted them and collapse to deliver the cargo inside their cores to the cells in high amounts through diffusion. Potential future in vivo active or targeted cargo transport and delivery applications of these soft, biocompatible, biodegradable, and neutrally buoyant microswimmers can be implemented in organ-on-a-chip devices51 and stagnant or low-flow-velocity fluidic regions of the human body, such as inside the cerebrospinal fluid, eye, urinary tract, bile duct, and capillaries of the vascular system. As demonstrated in the porous membrane experiments, soft emulsion-based microswimmers can squeeze and pass through hollow microtubular structures, such as small capillaries or bile duct branches, and microporous, heterogeneous regions inside complex biological fluids, robustly with no mechanical damage (as long as the diameter of the pores/tubes is not smaller than around 50% of the emulsion diameter). Similarly, in our vascular system, soft 6−8 μm diameter red blood cells can pass through capillaries by deformation.52 Depending on the microporosity of these environments, the diameter of our quasi-monodisperse microswimmers can be tuned to be from hundreds of nanometers up to tens of microns. Swimming next to the tissue cell regions, the cargo would be delivered by mechanical contact and collapsing of the emulsions at the cell surface (Supplementary Movie S3) so that the cargo molecules can diffuse inside the tissue cells. Such in vivo demonstrations would require medical imaging and localization of microswimmers and their physiological environment and optimization of the microswimmers for their motility, immunoresponse, biodegradability, size, cargo amount and type, and (magnetic and taxis-based) steering capability, in their given physiological operation environment and desired medical cargo delivery objective. As potential specific medical applications, such 9766
DOI: 10.1021/acsnano.7b02082 ACS Nano 2017, 11, 9759−9769
Article
ACS Nano
Microscopy and Image Analysis. The trajectory of the soft microswimmers was monitored far from the chamber walls at 20 frames per second (fps) using a Zeiss Axio Observer microscope with EC Plan Neofluar 63×/1.25 Ph3 oil immersion objective and a CCD camera (Andor Scientific, Belfast, UK). The images were postprocessed using an in-house MATLAB (Mathworks) script to obtain the 2D microswimmer trajectories. The trajectories were adjusted for any background fluid flow by subtracting the displacements caused by fluid drift from the microemulsions displacements. The fluid drift velocity was computed from the average velocity of all non-motile bacteria, typically around 6−10 per frame. The instantaneous 2D speeds were measured by first smoothing each trajectory using a fiveframe moving average and then computing the speeds from adjacent frames of the smoothed trajectory. Mean square displacement (MSD) of the swimming trajectories was calculated by
plate reader (Biotek). Loading efficiency was then calculated as the weight percentage of the dye inside the microemulsions to the total dye added to the initial aqueous phase. The measured intensity was converted into concentration by using a standard curve. The standard curve was compiled for each dye by measuring the fluorescence intensity of serial dilutions of each dye (Figures S3−S5). Quantification of Mito-Tracker Release Rate, Streptavidin Loading Density, and Dye Uptake by Cells. Water-in-oil emulsions loaded with Mito-tracker were filtered with Transwell membranes (0.4 μm pore size) to remove non-emulsified water and oil phase. The emulsified phase trapped in the membrane was then diluted into complete low-glucose Dulbecco’s modified Eagle’s medium cell culture media, and release of the dye was monitored with time using a Synergy 2 plate reader in fluorescence mode. To quantify the amount of streptavidin loaded onto the outer shell of the double emulsion, the w/o/w microemulsion was filtered through Transwell membrane (0.4 μm pore size) and resuspended in PBS. Streptavidin release was monitored with time and quantified using the Lowry colorimetric method (Micro BCA Protein Assay Reagent Kit, Pierce). Dye uptake in control cells stained with conventional methods were quantified measuring cell fluorescence using ImageJ and compared with the dye delivery by microswimmers to MCF7 cells with the following formula for corrected total cell fluorescence (CTCF):
MSD(Δt ) = ⟨[xi(t + Δt ) − xi(t )]2 ⟩
(1)
where Δt is the displacement, xi is the ith position, and t is the time. In addition, for each channel, images of 10−20 random microscope fields (0.1 mm × 0.1 mm) were captured and used to analyze and quantify the intensity of stained cells with ImageJ. For each single and double bacteria-attached double emulsions, at least 30 videos were captured and analyzed for measuring their mean speeds and trajectories. Statistical Analysis. All experiments were performed in triplicates. Quantitative data were expressed as a mean with ± standard deviation. Statistically significant difference between samples and control was evaluated using one-way analysis of variance (ANOVA), followed by post hoc tukey’s test for pairwise mean comparison in GraphPad Prism 6, and p-value
