Design and Development of Extracellular Matrix Protein‐Based Microcapsules as Tools for Bacteria Investigation

Abstract

The extracellular matrix (ECM) plays an immense role in the homeostasis of tissues and organs, can function as a barrier for infectious agents, but is also exploited by pathogens during infection. Therefore, the development of well‐defined 3D ECM models in the form of microcapsules to elucidate the interactions between ECM components and pathogens in confinement and study disease infectivity is important, albeit challenging. Current limitations are mainly attributed to the lack of biocompatible methods for the production of protein‐based microcapsules. Herein, hollow ECM‐based microcapsules from laminin‐111 or laminin‐111/collagen IV are generated to investigate the behavior of organisms within confined 3D extracellular matrices. Microcapsules are created using water‐in‐oil emulsion droplets stabilized by block copolymer surfactants as templates for the charge‐mediated attraction of laminin or laminin‐collagen proteins to the droplets’ inner periphery, allowing for the formation of modular ECM‐based microcapsules with tunable biophysical and biochemical properties and organism encapsulation. The release of E. coli‐laden ECM‐based protein microcapsules into a physiological environment revealed differences in the dynamic behavior of E. coli depending on the constitution of the surrounding ECM protein matrix. The developed ECM‐based protein microcapsules have the potential to be implemented in several biomedical applications, including the design of in vitro infection models.

Extracellular matrix (ECM) protein‐based microcapsules are established for the very first time using a modular droplet‐based microfluidic technology. They help shed light on different dynamic bacterial behaviors in confined 3D ECM mimics, laying a foundation for the development of more advanced in vitro infection models and the bottom‐up assembly of fully synthetic ECM microenvironments.

1. Introduction

The development of functional bio‐inspired microparticles that contain living organisms is of utmost interest for the engineering of living materials for various biomedical applications.^{[ 1} , ^{2 ]} A wide variety of natural (e.g., agarose,^{[ 3 ]} alginate^{[ 4 ]} and dextran^{[ 5 ]}) or synthetic (e.g., polyvinyl alcohol^{[ 6 ]} and Pluronic^{[ 7 ]}) polymers or inorganic matrices (e.g., porous silica^{[ 8 ]}) have been used for self‐regulated drug delivery devices^{[ 4 ]} or to entrap bacteria^{[ 3} , ^{9 ]} or cells for protective purposes.^{[ 6 ]} However, most of these materials were developed as solid microparticles, which lack the physiological, biophysical and biochemical properties of natural materials as well as well‐defined spatial structures (e.g., core‐shell morphology). Therefore, the development of hollow microcapsules with defined surface architectures and controllable biochemical and biophysical properties suitable for the analysis of the collective behavior of different organisms in confinement has become a topic of interest. Thus far, hollow microcapsules based on albumin proteins or polymers have been made for the delivery of drugs^{[ 10 ]} or probiotics.^{[ 4 ]} Such capsules are created utilizing the assembly potential of proteins at water/solvent interphases or the sequential layer‐by‐layer technique on silicon beads,^{[ 11 ]} two processes that depend on the employment of harsh chemicals to generate the hollow microcapsules.

In order to advance and expand the possibilities, we sought out to develop confined hollow microcapsules consisting of natural extracellular matrix (ECM) proteins. ECM proteins are secreted by cells and form the ECM that serves as a scaffold for cells and is a key determinant in the distribution of resident cells within the host tissue.^{[ 12 ]} Further, the ECM regulates cellular behavior such as migration, proliferation and metastasis of cancerous cells.^{[ 13 ]} Additionally, ECM proteins can both support the adhesion, proliferation and colonization of pathogens^{[ 14 ]} as well as inhibit them.^{[ 15 ]} In several in vitro models ECM protein coatings (especially collagen type IV coatings) have been used to facilitate pathogenic invasion, leading to a greater understanding of the invasion mechanism.^{[ 16} , ^{17 ]} These studies revealed that ECM contact accelerates microbial growth and subsequent biofilm formation.^{[ 15} , ^{18 ]} ECM substrate coatings have played a significant role in elucidating fundamental interactions between bacteria and ECM proteins in 2D, yet they are unable to address the fact that pathogens face 3D in vivo.^{[ 7 ]} Current efforts to create 3D infection models build on decellularized scaffolds that are loaded with various pathogens to observe their antimicrobial efficacy.^{[ 19 ]} However, decellularization and the generation of such 3D ECM scaffolds is a complex and time‐consuming process that requires specialized expertise and is susceptible to batch‐to‐batch variations.^{[ 20 ]} Other strategies focused on the development of 3D double network fibrous scaffolds to provide a desirable microenvironment for cell growth and mimicking the fibrous structure of the ECM.^{[ 21 ]} Nevertheless, most in vitro infection models established today do not specifically assess the involvement of individual ECM components in accelerating the growth of pathogens and biofilm formation.^{[ 22 ]} Therefore, we aimed to address these limitations by establishing modular 3D microenvironments composed of ECM proteins.

Previously it could be demonstrated that confinement effects and matrix (bio)mechanical properties have an important influence on the behavior of various organisms.^{[ 23} , ²⁴ , ^{25 ]} Moreover, the viscoelasticity and degradation kinetics of the matrices influence the proliferation, migration and differentiation of the encapsulated organisms.^{[ 26 ]} More specifically, tuning the viscoelasticity – by making use of dynamic or permanent cross‐linking networks and by varying the number and nature of degradable sequences – has helped elucidate the mechanosensitive range and the responses of eukaryotic organisms.^{[ 27} , ²⁸ , ^{29 ]} What remains to be studied is the influence of key biochemical and biophysical factors on bacterial cell growth and bacterial cell wall extension.

Here we describe the charge‐mediated assembly of bacteria‐laden ECM‐based protein microcapsules by means of modular droplet‐based microfluidics. The developed hollow microcapsules consist solely of ECM proteins, either laminin‐111 or a laminin‐111/collagen IV mixture also known as Matrigel. In the first step, water‐in‐oil droplets were generated for encapsulating the desired pathogens and ECM proteins. Next, ECM proteins were assembled by a charge‐mediated attraction mechanism on the droplets´ inner periphery and polymerized through the presence of specific ions. Upon release of the polymerized ECM‐based hollow microcapsules into physiologically relevant media, the growth and colonization of Escherichia coli (hereafter abbreviated as E. coli) was observed in a timely manner. The developed model system for the investigation of bacterial infections has allowed us to reconstitute two distinct ECM‐dependent natural behavior patterns of E. coli.

2. Results and Discussion

2.1. Charge‐Mediated Formation of Protein Microcapsules by Droplet‐based Microfluidics

A modular high‐throughput droplet‐based microfluidic technology (Figure  1 and Video S1, Supporting Information) was designed and implemented in order to generate hollow microcapsules consisting solely of ECM proteins. Much in resemblance to the procedure for creating droplet‐stabilized giant unilamellar vesicles (dsGUVs),^{[ 30 ]} the formation and stabilization of the hollow protein microcapsules was achieved by applying water‐in‐oil emulsion droplets stabilized by diblock or triblock‐copolymer fluorosurfactants as fundamental scaffolds. In order to achieve a charge‐mediated attraction between the encapsulated proteins and the droplets’ inner periphery, we used a negatively‐charged carboxyl‐terminated perfluoropolyether (Krytox) mixed with PEG‐PFPE surfactant or positively‐charged N⁺(Me)3‐terminated PEG‐PFPE surfactant.

Figure 1.

Schematic representation of the modular droplet‐based microfluidic technology for the charge‐mediated formation of droplet‐based protein capsules (dsProCaps) and the subsequent release of protein capsules (ProCaps). The formation module shows two possible methods for producing empty or cargo‐filled dsProCaps: A and B) A single‐inlet device was used to create empty dsProCaps. C) Utilizing a double‐inlet device, E. coli and proteins were encapsulated in a parallel flow. D) A pico‐injection module allowed for the injection of bacteria into dsProCaps. E) C and D lead to the formation of cargo‐filled dsProCaps. F) During bulk release the oil phase around the protein capsules was removed through the addition of perfluoro‐octanol, thereby producing the final ProCaps. G) The successful formation of dsProCaps or ProCaps can be examined with a glass‐slide observation chamber.

We employed a droplet‐based microfluidic formation module with a single channel for the high‐throughput formation and optimization of the droplet‐stabilized protein microcapsules (dsProCaps) (Figure 1A,B). By encapsulating proteins and the ions required for their polymerization, we were able to precisely control kinetics of their charge‐mediated formation and the size of the dsProCaps. In order to generate bacteria‐laden dsProCaps, two aqueous inlet channels were implemented. This minimizes the interaction between the proteins and bacteria prior to their co‐encapsulation inside the droplets, which could interfere with the charge‐mediated attraction of the protein content to the droplets’ inner periphery (Figure 1C; Figure S1, Supporting Information). Moreover, to avoid any bacteria‐protein interactions before the charge‐mediated formation of the protein layer on the inner periphery, we relied on a sequential approach by means of an automated pico‐injection microfluidic technology^{[ 31 ]} (Figure 1D). Toward this end, bacteria were injected into the preformed, protein‐laden droplets in the presence of polymerization‐supporting ions. This sequential approach provides sufficient time for the proteins’ charge‐mediated attraction to the droplets’ inner periphery and their assembly prior to the injection of the desired cargo. Either module, the double inlet device or the pico‐injection device, can be used to generate cargo‐filled dsProCaps (Figure 1E). Following the formation of the dsProCaps (either empty or bacteria‐laden), we employed a bulk release module to release the established protein microcapsules (ProCaps) from the surfactant shell and the oil phase into physiological conditions (Figure 1F). We used perfluoro‐1‐octanol (PFO) as a droplet‐destabilizing agent, which results in the fusion of droplets at the oil/aqueous interphase. Once the polymer‐based surfactant shell was opened up, the intact microcapsules were released into an aqueous buffer. Glass observation chambers were assembled to observe dsProCap formation and their subsequent release into physiological conditions (Figure 1G).

2.2. Formation of dsProCaps and the Subsequent Release of Protein Capsules

Figure  2 shows the charge‐mediated assembly of dsProCaps employing either Krytox or N⁺(Me)3‐terminated surfactants depicted in either the upper or the lower halves of the droplets in Figures 2A‐1–A‐4, respectively. In the case of Krytox, this negatively charged, polymer‐based surfactant was mixed with neutral PEG‐PFPE surfactants to achieve a negative charge on the droplets’ inner periphery (Figure 2A‐1, upper droplet half). This negative charge attracts free Ca²⁺ ions from the droplet lumen to the inner periphery of the water‐in‐oil droplets, thereby generating a positively charged layer (Figure 2A‐2) that in turn recruits the negatively charged proteins (Figure 2A‐3). Alternatively, we also created protein‐containing droplets stabilized by positively charged N⁺(Me)3‐terminated surfactants (Figure 2A‐1, depicted in the lower droplet half). This approach utilizes the direct electrostatic attraction of negatively charged proteins to the surfactants (Figure 2A‐1). In this case, due to the low diffusivity of proteins and the availability of Ca²⁺ ions in the droplet lumen, protein‐Ca²⁺ complexes were established (Figure 2A‐2) prior to their attraction to the positively charged inner periphery (Figure 2A‐3). It is important to note that the Ca²⁺ ions, besides their function in attracting proteins, are also required for the polymerization of laminin‐111 or Matrigel at 37 °C.^{[ 32 ]}

Figure 2.

The charge‐mediated attraction mechanism for the formation of dsProCaps and the subsequent release of ProCaps. A) The sequence of events leading to the charge‐mediated formation of a protein layer on the droplets’ inner periphery depends on the charge of the surfactant layer. dsProCap production using negatively charged Krytox is shown in the upper halves of the droplets in illustrations 1–4: Ca²⁺ ions are attracted to the negatively charged interface first, leading to the electrostatic attraction of negatively charged proteins. dsProCap production using droplets stabilized by the positively charged surfactant N⁺(Me)3‐PEG‐PFPE is shown in the lower halves of the droplets in illustrations 1–4: Ca²⁺ ions aggregate with the proteins first, after which the negative charge of the proteins causes these aggregates to slowly be attracted to the positively charged interface. The final step in dsProCap production was similar for both types of surfactants: the Ca²⁺ ions mediate the polymerization of the proteins at the droplets’ inner periphery. B) Laminin‐111 or Matrigel dsProCaps stabilized with either Krytox/PEG‐PFPE or N⁺(Me)3‐PEG‐PFPE. The overlays in the upper right‐hand corners show intensity profiles depicting the protein attraction to the droplets’ inner periphery. Scale bars are equal to 50 µm. C) A schematic of the step‐by‐step ProCap bulk release process. 1. The dsProCaps are surrounded by stabilizing surfactants in the oil layer. 2. PFO is diluted in the oil phase. 3. The PFO chains slowly incorporate into the stabilizing oil layer. 4. By adding PBS onto the PFO layer, the stabilizing oil shell entirely dismantles and the polymerized ProCaps are released into the aqueous release media. D) Representative confocal images of laminin‐111‐based or Matrigel‐based ProCaps released from either Krytox/PEG‐PFPE or N⁺(Me)3‐PEG‐PFPE stabilized droplets via bulk release. Scale bars are equal to 50 µm.

In this study, we used either laminin‐111 or diluted amounts of Matrigel to produce the ECM protein‐based microcapsules. Laminin‐111 is known for its high abundance throughout the human body and its affinity for interacting with various cell types^{[ 33 ]} as well as many pathogens.^{[ 16 ]} Matrigel was used for two reasons. First, Matrigel is widely used as a substrate for the growth and development of cells in 2D and 3D cultures.^{[ 34 ]} Second, it is composed of laminin‐111 (60%), collagen type IV (30%), entactin (8%), and several growth factors such as TGF‐β and IGF‐1.^{[ 34 ]} Entactin serves as a bridging molecule between laminin‐111 and collagen type IV, and supports the structural organization of the Matrigel matrix.^{[ 35 ]} Using Matrigel allows us to investigate the establishment of protein microcapsules made out of a mixture of proteins and the interaction of organisms in a hollow confinement.

Confocal fluorescence microscopy shows the successful attraction of laminin‐111 and Matrigel proteins to the negatively or positively charged droplets’ inner periphery (Figure 2B). In the case of negatively charged dsProCaps, a homogeneous layer of proteins can be observed on the droplets’ inner periphery. The smooth protein organization in the case of the negatively charged droplets can be attributed to a rapid charge‐mediated attraction of Ca²⁺ ions to the droplets’ interface and the consequent diffusion of proteins. In contrast, in the positively charged droplets the driving force for spatial organization is the electrostatic attraction of negatively charged proteins. Therefore, the spatial organization dynamics is affected by partially aggregated protein‐Ca²⁺ ion complexes, leading to a rough appearance of the droplets’ inner periphery (Figure 2A‐3 and 2B). The attraction of proteins to the droplets’ inner periphery without the addition of Ca²⁺ ions is slower; however, a smooth bright protein ring establishes over time (Figure S2, Supporting Information). This emphasizes the generation of protein‐Ca²⁺ complexes which are attracted to the positively charged periphery. Note that in the case of smaller‐sized dsProCaps (10 – 20 µm in diameter) no protein aggregation was detected despite the presence of calcium ions. This observation can be attributed to the shorter diffusion distances (Figure S3, Supporting Information). These results demonstrate the versatility of our microfluidic technology to control the size and type of protein microcapsules depending on the application envisaged.

After the successful implementation of charge‐mediated approaches to yield dsProCaps consisting of various ECM proteins, the next step is the release of the polymerized ProCaps from the stabilizing surfactant shell in the oil phase into physiologically relevant media. This transition is important to nurture the encapsulated cells or organisms and to supply a suitable environment for biomedical applications. To achieve this, a previously described bulk release approach was implemented to transform the dsProCaps into ProCaps.^{[ 30 ]} Toward this end, the de‐emulsifier PFO was mixed with the surrounding oil phase, resulting in the incorporation of its short surfactant molecules into the surfactant layer. This destabilizes the droplet and the polymerized protein microcapsules are released into the previously added PBS (Figure 2C – the top blue layer in the depiction). Figure 2D shows that the released laminin‐111 or Matrigel‐based ProCaps preserve their round shape and feature characteristic protein fiber patterns. However, the ProCaps that were released from the negatively charged dsProCaps do exhibit a smoother protein morphology. This observation can be attributed to the more homogeneous organization of proteins on the inner periphery of the Krytox/PEG‐PFPE stabilized droplets (Figure 2B).

Most charge‐mediated approaches add commercially available Krytox to the surfactant solution in order to achieve a negatively charged water‐oil interface due to the presence of carboxylic acid.^{[ 30} , ^{31 ]} Subsequently, divalent ions are added to the inner buffer to generate a positive charge on the surfactant interface in order to attract negatively charged molecules. However, the implementation of Krytox in the surfactant solution, and the subsequent attraction of divalent ions, might have several undesirable effects on proteins due to a reduction of the pH and not all applications might benefit from the presence of high concentrations of divalent ions. Therefore, the development of a positively charged surfactant allows a straightforward approach for the charge‐mediated assembly of negatively charged proteins.

As confirmed by zeta potential measurements, all ECM‐based ProCaps possess a slightly negative surface charge (Table S1, Supporting Information). Less negative surface potential as observed in case of ECM‐based ProCaps released from the droplets stabilized by the positively charged surfactant can be attributed to the generation of protein‐Ca²⁺ complexes attracted to the positively charged droplet periphery.

To verify that the production process does not affect the biocompatibility of our ECM‐based ProCaps, we co‐cultured human keratinocytes (HaCaT) cells with freshly released ProCaps. The co‐culturing experiments demonstrated that the developed ECM ProCaps are not rejected by mammalian cells since HaCaT cells are able to grow around and on top of the ECM‐based ProCaps establishing a confluent cell monolayer (Figure S4, Supporting Information). In addition, no visible cytotoxic effects of the laminin‐111 ProCaps were observed, regardless of the charge of the surfactants used to stabilize them.

2.3. Generation of Bacteria‐Laden Protein Microcapsules (ProCaps)

After the successful generation of ECM‐based ProCaps, we employed these microcapsules for the real time in vitro observation of bacterial behavior in ECM niches using E. coli as a model organism. A flow‐focusing device featuring two aqueous inlets (Figure 1B) was utilized to minimize the interactions between the E. coli and the proteins prior to their co‐encapsulation inside the droplets. Using this approach, E. coli‐laden dsProCaps consisting of laminin‐111 or Matrigel and stabilized by Krytox/PEG‐PFPE or N⁺(Me)3‐PEG‐PFPE surfactants were successfully generated (Figure  3A). In addition, fluorescent dyes were also co‐encapsulated and confocal microscopy images taken to confirm bacterial viability during the entire dsProCaps formation process (green SYTO‐9 was used to mark viable cells and red propidium iodide was used to mark dead cells). These images revealed that in the case of droplets stabilized by negatively charged surfactants protein attraction to the droplets’ inner periphery was not affected by the co‐encapsulation of bacteria and bacterial viability was unaffected. However, reduced protein attraction was observed in the presence of positively charged surfactants. This can be attributed to the potential competition between negatively charged bacteria and proteins at the droplets’ inner periphery. On average we were able to encapsulate 22 bacteria in each dsProCap (Figure S5, Supporting Information). The average number of E. coli bacteria that can be encapsulated into our dsProCaps is depending on the size of our dsProCaps, the bacteria concentration, and the respective aqueous and oil flow rates.

Figure 3.

E. coli encapsulation in laminin‐111 or Matrigel dsProCaps and ProCaps generated with two differently charged fluorosurfactants. A) Confocal microscopy images of laminin‐111 (left) and Matrigel (right) dsProCaps stabilized either with Krytox/PEG‐PFPE or N⁺(Me)3‐PEG‐PFPE surfactants and loaded with E. coli (living E. coli are dyed green in the images). B) Corresponding laminin‐111 or Matrigel ProCaps containing E. coli directly after release from either Krytox/PEG‐PFPE or N⁺(Me)3‐PEG‐PFPE‐stabilized droplets. C) E. coli growth inside different ProCaps after 2 h incubation. Orthogonal viewpoints were selected for image production. Scale bars equal 50 µm.

We observed that in the first 24 h the E. coli bacteria barely proliferated inside the dsProCaps (Figure S6, Supporting Information). This can be attributed to the low nutrient levels inside the dsProCaps. It has been shown previously that E. coli bacteria enter a quiescent state in nutrient‐limited environments.^{[ 36 ]} To probe viability and potentially track the behavior of E. coli bacteria in the confined ECM‐based ProCaps, the E. coli‐laden ProCaps were released into physiological conditions 24 h after dsProCap production using a bulk release approach. Time‐lapse confocal laser scanning microscopy was used to track this process (Figure 3B; Figure S5, Supporting Information). After 2 h of incubation in a nutrient‐rich environment E. coli proliferation was high and the majority of the bacteria remained retained inside our minimal synthetic ECM‐mimicking ProCaps (Figure 3C). The live/dead cell viability assay confirmed that the E. coli remained highly viable inside the ProCaps and strengthened the hypothesis that E. coli bacteria in dsProCaps are quiescent until they are released into nutrient‐rich environments (Figure 3C and Video S2, Supporting Information). Strikingly, confocal microscopy analysis of the ProCaps revealed an expansion of the ProCaps’ membrane upon E. coli proliferation (Figure 3C). This is in contrast to empty ProCaps, which remained morphologically unchanged over time (Figure S7, Supporting Information). In correlation with observations made by confocal laser scanning microscopy, scanning electron microscopy (SEM) analysis allowed to observe slight differences in the morphology of laminin‐ and Matrigel‐based ProCaps (Figure S8, Supporting Information). The slight differences can be attributed to the higher amount of protein used for the generation of Matrigel‐based ProCaps, leading to more compact and smoother surface morphologies in comparison to laminin‐based ProCaps that tend to be rougher. Moreover, Matrigel contains the bridging molecule entactin to bind collagen type IV and laminin‐111, resulting in a stronger crosslinking as compared to solely Ca²⁺‐mediated crosslinking.

These findings further support the use of the newly developed ECM‐based ProCaps as an advanced and suitable microenvironment to investigate bacterial behavior in vitro. Previous reports have elucidated the importance of adhesin‐mediated interactions between bacteria and the host ECM, interactions that facilitate host cell invasion and the persistence of pathogens adhering in close proximity to the host cell.^{[ 37 ]} In a next step, we investigated in detail the behavior of E. coli in the lumen of the developed ECM ProCaps as well as in contact with ECM‐based ProCaps. Toward this end, the bacterial behavior was analyzed over longer time periods and under static conditions, i.e., without the addition or exchange of nutrient media.

2.4. Bacterial Behavior in ProCaps Consisting of Laminin‐111

We performed confocal time‐lapse experiments to investigate the dynamic behavior of E. coli within the lumen of laminin‐111 ProCaps. Toward this end, we encapsulated E. coli in laminin‐111 ProCaps using a flow‐focusing device consisting of two aqueous inlets and released the assembled bacteria‐laden ProCaps into physiological conditions (Figure 1B). During the initial stages of E. coli growth, the bacteria appeared loosely packed and exhibited their typical rod‐shaped morphology (Figure  4 , t:0 h). Following 3 h after release, the encapsulated E. coli formed tightly organized bacterial communities, in which the growth of the entrapped bacterial colonies was defined by the size of the lumen of the laminin‐111 ProCaps (Figure 4, t:3 h). The colony continued to grow over time until it completely occupied the lumen of the laminin‐111 ProCaps, eventually leading to a distortion of the ProCap “membrane” that separates the interior from the outside environment (Figure 4, t:4 h). Finally, these protruding pod‐like structures ruptured the “membrane” of the laminin‐111 ProCaps, resulting in a bulk release of captured E. coli bacteria into the surrounding media (Video S3, Supporting Information). Importantly, this observation demonstrates that the “membrane” of the laminin‐111 ProCaps was able to expand when the bacterial communities increased their volume upon proliferation. After the eruption, the capsule shell collapses slightly again (Figure 4).

Figure 4.

Observation of E. coli encapsulated in laminin‐111 ProCaps over time. Confocal images of E. coli (living E. coli are dyed green in the images) in laminin‐111‐based ProCaps (stained red) show the growth of contained colonies (marked by an empty arrow head) during the first 4 h of incubation. Soon after, some of the capsules (marked by a filled arrow head) burst under the pressure of the growing E. coli colonies. The ProCaps grew to several times their initial size before the E. coli colonies erupted. During this process the capsule's shell first ruptured and after the bacteria had broken out it collapsed. Scale bars equal 50 µm.

Overall, we demonstrated that our laminin‐111 ProCaps can be implemented as a bioinspired in vitro infection model system to study the growth and outbreak of microorganisms from confined and well‐defined ECM‐based microenvironments. Future work should focus on disease‐causing bacteria. For example, osteomyelitis could be mimicked by encapsulating Staphylococcus aureus in our ECM‐based ProCaps to investigate bacterial susceptibility to antibiotic treatment in a model that mimics the biophysical and biochemical natural microenvironment of the bacteria. It is worth mentioning that microparticles have been used before to encapsulate bacteria in order to preserve the growth of bacterial strains. However, most of the developed systems are composed of non‐human materials^{[ 38 ]} and depend on active degradation of the material to permit bacteria to escape.^{[ 23 ]} Instead, we observed that excessive growth and proliferation of bacteria in the lumen of our laminin‐based ProCaps results in eruption of the capsule's shell and thereby closely resembles the actual in vivo situation.^{[ 39 ]}

2.5. Bacterial Behavior in ProCaps Consisting of Matrigel

Following the analysis of E. coli behavior in ECM‐based ProCaps consisting of laminin‐111, we performed similar confocal time‐lapse experiments in the more complex Matrigel‐based ProCaps. It is important to mention here that due to a higher protein concentration and natural cross‐linkers such as entactin,^{[ 34 ]} the morphology of the obtained Matrigel‐based ProCaps was different from the ProCaps consisting of laminin‐111. Furthermore, Matrigel contains a broad mix of proteoglycans and growth factors.^{[ 34 ]} Matrigel capsules possess a more compact and rounded shape than the filamentous and elongated laminin‐111 ProCaps (Figure 2D).

Whereas the E. coli bacteria in the laminin‐111 ProCaps retained their typical rod‐shaped morphology upon the formation of bacterial colonies, the encapsulated E. coli bacteria in Matrigel ProCaps elongated their shape to become what is referred to as filamentous bacterial cells (Figure  5A). Strikingly, this transformation was independent of the incubation time in nutrient‐limited dsProCaps prior to their release into bacteria culture media (Figure S9, Supporting Information). Moreover, in contrast to the eruptive outbreak of a mass of rod‐shaped E. coli from laminin‐111 ProCaps, in the case of Matrigel ProCaps the elongated filamentous E. coli escaped in succession (Video S4, Supporting Information). This successive outbreak of singular filamentous bacterial cells can be attributed to the greater cross‐linkage between the various proteins in the “membrane” of the Matrigel‐based ProCaps.

Figure 5.

Observation of E. coli encapsulated in Matrigel ProCaps. A) Confocal images of E. coli (living E. coli are dyed green in the images) encapsulated in Matrigel ProCaps (dyed purple) reveal that they grow as filamentous bacteria in the first 4 h and successively escape from the capsules over time. Scale bars equal 50 µm. B) SEM high‐resolution imaging of filamentous E. coli bacteria in the Matrigel capsules. B1) Bacteria were surrounded and covered by a thin protein layer (image taken at 2.95kx magnification). B1a) Some rod‐shaped bacteria were covered with a thin layer of protein (marked by a filled arrow head) while others close by were not (marked by an empty arrow head). The image was taken at 14.61kx magnification. B1b) One long filamentous bacterium was partially covered by proteins (the covered part and the uncovered part are marked by a filled arrow head and an empty arrow head, respectively). The image was taken at 10.47kx magnification. B2) Filamentous bacteria growing out of Matrigel protein capsules (image taken at 1.39kx magnification). B2a) Close up of bacterial filaments in contact with a protein layer (image taken at 5.06kx magnification). B2b) Long twisted filamentous bacteria (image taken at 4.56kx magnification).

We used SEM to verify whether the encapsulation and proliferation of E. coli bacteria in the Matrigel ProCaps results in the development of true filamentous bacteria, rather than chains of bacterial cells (Figure 5B). Confirming the results obtained by confocal microscopy, the scanning electron micrographs showed the formation of elongated filamentous bacteria that appear to break out from the lumen of the Matrigel ProCaps successively. Likewise, the successive outbreak of singular rod‐shaped bacteria was also observed.

As a control experiment, we encapsulated E. coli bacteria in dense Matrigel beads. The Matrigel beads were obtained by increasing the concentration of Matrigel protein and allowing it to polymerize at 37 °C. In contrast to E. coli bacteria encapsulated in the lumen of hollow Matrigel ProCaps, E. coli bacteria trapped in dense Matrigel beads formed smaller bacterial colonies and did not show bacterial filamentation (Figure S10 and Video S5, Supporting Information). In addition, E. coli bacteria were not motile within the Matrigel beads, presumably due to the physical confinement as a consequence of the dense hydrogel network. Moreover, after 4 h of incubation in Matrigel beads bacterial cell viability dropped significantly. This is observable as a color shift in the live/dead dye from green to purple (Video S5, Supporting Information). Our control experiments on 2D protein‐coated surfaces to verify whether the occurrence of bacterial filamentation is solely depending on the type of ECM protein could demonstrate that bacterial filamentation is not only a result of the type of ECM protein (Figure S11, Supporting Information). In contrast to the 2D protein‐coated surfaces, bacterial filamentation was only observed within our 3D Matrigel‐based ProCaps.

Recently, Bhusari et al.^{[ 7 ]} demonstrated that the degree of covalent cross‐linking in the hydrogel matrix impacts the growth behavior of the encapsulated bacteria: the more extensive the chemical cross‐linking of the matrix, the smaller the colony sizes and the slower the elongation rates of the bacteria. Additionally, previous investigations have shown that the morphological change of E. coli bacteria from rod‐shaped to a filamentous shape depends on a combination of factors (for instance environmental stress, starvation, high osmolarity and extreme pH).^{[ 40 ]} Based on these findings, we argue that starvation during the initial stage of encapsulation as well as hindrance of nutrient supply due to higher cross‐linkage in the case of the Matrigel ProCaps are two factors that are responsible for the process of bacterial filamentation and dysregulated bacterial division. Further, we could observe shedding behavior in single filamentous bacteria once they had left the capsule both by confocal microscopy and SEM (Figure S12, Supporting Information). This specific behavior is common for filamentous bacteria in order to induce a secondary infection cycle and for the rod‐shaped infectious bacterial cells to spread.^{[ 41 ]} The underlying process behind bacterial filamentation is still poorly characterized.^{[ 14} , ³⁶ , ⁴¹ , ⁴² , ^{43 ]} Previous research required an in vitro flow system and cell stress to induce bacterial filamentation.^{[ 40} , ⁴¹ , ^{42 ]} However, we were able to closely resemble various stages of uropathogenic tract infections in our bacteria‐laden Matrigel‐based ProCaps. Encapsulated E. coli bacteria were able to form tightly packed and organized bacteria communities, after that many transformed into highly elongated and filamentous bacteria. Additionally, we could also observe via confocal laser scanning microscopy and SEM the typical shedding of single bacteria from the extracellular end of long filamentous bacteria. In the future, it will be investigated whether our Matrigel‐based ProCaps can be applied as an in vitro infection model to shed more light on secondary infections.

In this study, we demonstrate that our Matrigel ProCaps can be applied to generate filamentous E. coli – similar to the E. coli observed in the uropathogenic cascade^{[ 36} , ^{44 ]} – without the presence of host cells or a sophisticated flow system, thus drastically simplifying the experimental conditions. In addition, the newly developed method for the generation of Matrigel ProCaps makes it possible to systematically and precisely tune the biophysical and biochemical properties to best mimic natural bacterial microenvironments and, therefore, can shed light on the factors and mechanisms involved in bacterial filamentation. Future work will focus on pinpointing the presence and involvement of specific genes contributing to bacterial filamentation as well as investigating the effects of porosity, size, mechanical strength, and presence of signaling molecules.

3. Conclusion

We have developed ECM‐based ProCaps using droplet‐based microfluidics to investigate bacterial cell behavior within confinement. The well‐controlled, charge‐mediated attraction of ECM proteins to either negatively or positively charged surfactant molecules in the presence of the necessary ions resulted in the successful polymerization and development of laminin‐111 and Matrigel‐based ProCaps. Furthermore, we demonstrated the enhanced growth of bacterial colonies in confined microenvironments that provide environmental cues that are normally present in vivo. In addition, the developed ECM‐based ProCaps can be applied to create a local increase in bacterial cell number. Growing E. coli within ProCaps revealed a surprising bulging of the capsule prior to an outbreak of bacteria. This might mimic certain conditions in vivo and should be extended to the study of other bacteria or eukaryotic cells. In addition to future genetic screening of the various factors involved in bacterial filamentation and secondary infection reaction, further investigations will focus on the co‐culture of the developed protein microcapsules with relevant cell lines to shed light on the molecular recognition and defense mechanisms between immune cells and epithelial cells as well as the progression of bacterial infections. We envision that the implememntation of the developed ECM‐based ProCaps will provide much needed insights into molecular pathways involved in health and disease and will enable establishment of advanced in vitro infection models for the development of new treatments. Capitalizing on the strength of the developed microfluidic approach to systematically tune the biophysical and biochemical properties, our novel and simplified ECM mimetics establish a modular and versatile approach to broaden our understanding of cell‐matrix interactions under physiological conditions.

4. Experimental Section

Microfluidic Device Fabrication

Single and double inlet microfluidic devices were drafted with a CAD software QCAD‐pro (Ribbonsoft, Switzerland). The channel height and width were 30 µm each, if not stated otherwise. To achieve a uniform layer of 30 µm in height, a SU8‐3025 negative photoresist (MicroChem, USA) was spin‐coated (Laurell Technologies Corp., USA) at 2600 rpm for 30 s onto silicon wafers (MicroChemicals, Germany). A soft bake (5 min at 65 °C and 15 min at 95 °C on a hot plate (IKA, C‐MAG HS7)) was carried out to remove remaining solvents. To generate microfluidic devices, it was made used of the Tabletop Micro Pattern Generator µPG 101 (Heidelberg Instruments, Germany). The structure of choice was directly exposed into the photoresist‐coated wafer with the following exposure conditions: 50 mW output power of the laser and 20% pixel pulse duration. To remove any unexposed photoresist, the silicon wafers were developed (mr‐DEV 600, MicroChemicals, Germany) and cleaned with isopropanol. To prepare PDMS (polydimethylsiloxane, Sylgard 184, Dow Corning, USA) the oligomer was mixed with the polymerization catalyst at a 9:1 (w/w) ratio, poured over the silicon wafer, degassed in a desiccator, and cured for 2 h at 65 °C. In a final step, the PDMS block with the structure as well as a coverslip (#1, Carl Roth, Germany, 24 × 60 mm) were carefully cleaned with EtOH and activated in an oxygen plasma (PVA TePla 100, Germany). Finally, the activated sides were pressed together and incubated at 65 °C overnight. Directly before using the microfluidic devices, the microchannels were flushed with Sigmacote (Sigma–Aldrich, Germany) to render them hydrophobic, and left for 5 min at 65 °C. Afterward the devices were flushed with pure and filtered HFE‐7500 oil and were stored until further usage. A high‐speed camera was used to obtain images of the dsProCap production and to observe the droplet production process. A Photron FASTCam Mini UX100 was used to record water‐in‐oil droplet production at a resolution of 1280 × 1024 with 4000 fps.

Fluorescent Labelling of Laminin‐111

Laminin‐111 was fluorescently labelled to visualize the encapsulated ECM proteins and detect the generated ProCaps by means of confocal laser scanning microscope (CLSM). First, Laminin‐111 (Laminin from Engelbreth‐Holm‐Swarm murine sarcoma basement membrane, 1–2 mg mL⁻¹ in Tris buffered NaCl, Sigma, L2020) was dialyzed (Slide‐A‐Lyzer™ MINI Dialysis Device, 3.5 K MWCO, 0.5 mL, Thermo Scientific) against sterile 1x PBS at 4 °C. The PBS was exchanged hourly over a period of 6 h and left overnight. The next day, 50 µg of NHS‐Ester DyLight™ 550 (Thermo Scientific, 62 263) was diluted with 500 µL of freshly dialyzed laminin‐111. The vial was shaken for 1 h at 4 °C on a test tube shaker at 600 rpm. To remove any unbound dyes the solution was dialyzed against 1x PBS as described. The labeled protein solutions were aliquoted the next day and stored at −20 °C until further usage.

Synthesis of N⁺(Me)3‐PEG‐PFPE Surfactant

Benzene‐d6, Celite, chloroform‐d, copper(II) sulfate pentahydrate, dichloromethane, diethyl ether, dimethyl sulfoxide‐d6, ethylenediaminetetraacetic acid, hexafluorobenzene, neocuproine, PEG 600, propargyl bromide (80 w/w% in toluene), propargylamine, sodium ascorbate, sodium chloride, sodium hydroxide, p‐toluenesulfonyl chloride, triethylamine and trimethylamine (4.20 m in ethanol) were purchased from Sigma–Aldrich. Dimethyl formamide, ethyl acetate, magnesium sulfate, methanol, oxalyl chloride and tetrahydrofuran were purchased from Merck Millipore. Krytox FSH and Novec 7100 were purchased from Costenoble. Chemicals were used without further purification. NMR spectra were acquired on Bruker Acend 400 (field intensity: 9.4 T, frequency: 400.15 MHz).

PEG 600 ditosylate 1

An aqueous solution (100 mL) of sodium hydroxide (NaOH, 13.3 g, 333 mmol, 4.0 eq.) was cooled down to 0 °C. PEG 600 (50.0 g, 83.3 mmol, 1.0 eq.), dissolved in tetrahydrofuran (THF, 200 mL), was added slowly, so that the temperature of the reaction mixture did not rise above 5 °C. Then, the reaction mixture was allowed to warm up to room temperature and stirred for 1 h at room temperature. After cooling the reaction mixture to 0 °C, a solution of p‐toluenesulfonyl chloride (36.5 g, 192 mmol, 2.3 eq.) in THF (220 mL) was added dropwise so that the temperature did not rise above 5 °C. The reaction mixture was stirred for 18 h, allowing the temperature of the mixture to slowly rise to room temperature. The reaction mixture was separated and the organic layer was collected. The solvent was removed under reduced pressure. The crude product was re‐dissolved in ethyl acetate (900 mL), washed two times with water (100 mL) and once with saturated sodium chloride solution (100 mL). After drying over magnesium sulfate, the solution was filtered and the solvent was removed under reduced pressure. The ditosylated PEG derivative was received as a clear oil (61.1 g, 67.3 mmol, 80.8%).

1H‐NMR (400, CDCl3): δ = 7.74 (d, J = 8.33 Hz, 4 H), 7.29 (d, J = 8.09 Hz, 4H), 4.10 (t, J = 4.83 Hz, 4H), 3.63 (t, J = 4.83 Hz, 4 H), 3.59‐3.52 (m, 44 H), 2.39 (s, 6 H).

PEG 600 diazide 2

PEG 600 ditosylate 1 (82.9 g, 91.1 mmol, 1.0 eq.) was dissolved in dimethylformamide (DMF, 30 mL) and sodium azide (NaN3, 13.0 g, 200 mmol, 2.2 eq.) was added. The reaction mixture was stirred for 90 min at room temperature, followed by 18 h at 50 °C. During this reaction time the mixture became turbid. The reaction mixture was cooled down to room temperature and filtered through Celite. The solvent was removed by co‐evaporation with toluene under reduced pressure. The crude product was re‐suspended in ethyl acetate (800 mL), filtered and washed three times with saturated sodium chloride solution (50 mL). The solution was dried over magnesium sulfate, filtered, and the solvent removed under reduced pressure. The diazide derivative of PEG 600 was received as a yellowish oil (53.98 g, 83.0 mmol, 91.2%).

1H‐NMR (400, CDCl3): δ = 3.69‐3.65 (m, 50 H), 3.39 (t, J = 5.04 MHz, 4 H).

Propargyl PFPE 7000 3

After degassing in a flame‐dried Schlenk flask, PFPE 7000 acid (Krytox FSH, 87.0 g, 12.89 mmol, 1.0 eq.) was dissolved in HFE 7100 (Novec 7100, 150 mL) under nitrogen atmosphere. Oxalyl chloride (3.4 mL, 39.4 mmol, 3.05 eq.) was added dropwise and the mixture then refluxed at 61 °C for 18 h. During this process, the solution became turbid. Any excess of oxalyl chloride or solvent was collected in a liquid nitrogen trap under reduced pressure. The acid chloride removed from the PFPE 7000 was obtained as a slightly turbid, highly viscous liquid. In a second reaction step, a Schlenk flask was equipped with a dropping funnel under nitrogen counter‐flow and the acid chloride obtained in the first reaction step was dissolved in HFE 7100 (90 mL). A solution of propargyl amine (870 µL, 13.5 mmol, 1.05 eq.), triethylamine (2.7 mL, 19.3 mmol, 1.5 eq.) and dried THF (35 mL) was loaded into the funnel and added dropwise. After complete addition, the reaction mixture was stirred for 18 h at room temperature. After removing the solvent and other volatile reagents under reduced pressure, the crude product was redissolved in HFE 7100 and filtered. The solvent was again removed under reduced pressure. The propargyl derivative of PFPE was received as an orange oil (75.7 g, 11.15 mmol, 86.5%). The yield was calculated over both reaction steps. A mixture of C6F6:C6D6 88:12 was used as NMR solvent.

1H‐NMR (400, C6D6): δ = 2.186 (t, J = 2.8 Hz, 1H, H3), 4.135 (t, J = 2.8 Hz, 2H, H1), 6.602 (s, 1H, NH).

N,N,N‐trimethylprop‐2‐yn‐1‐aminium bromide (TAB) 4

Trimethylamine (TMA, 4.20 m in ethanol, 30.7 mL, 129 mmol, 1.11 eq.) was diluted in diethyl ether (30 mL) and cooled down to 0 °C. Propargyl bromide solution (80 w/w% in toluene, 10.0 mL, 116 mmol, 1.0 eq.) was slowly added. The reaction mixture was stored for 24 h at 4 °C. Any unreacted TMA and solvent were removed under reduced pressure. The asymmetric ammonium bromide was received as a white solid (17.1 g, 95.9 mmol, 97.4%).

1H‐NMR (400, DMSO‐d6): δ = 4.47 (d, J = 2.52 Hz, 2 H), 4.08 (t, J = 2.53 Hz, 1 H), 3.16 (s, 9 H).

TAB‐PEG 600 azide 5a

PEG 600 diazide 2 (10 g, 15.4 mmol, 1.0 eq.) and TAB 4 (2.74 g, 15.4 mmol, 1.0 eq.) were dissolved in water (74.0 mL) while sonicating until both substances were completely dissolved. Sodium ascorbate (6.10 g, 30.8 mmol, 2.0 eq.) and, after stirring for 5 min, copper(II) sulfate pentahydrate (3.84 g, 15.4 mmol, 1.0 eq.) were added. The reaction mixture was stirred for 24 h. Next, it was filtered through Celite and washed three times with dichloromethane (each 30 mL). The solvent was removed under reduced pressure. The green, amorph substance consisted of a mixture of one‐end‐functionalized TAB‐PEG 600 azide 5a, dual‐end‐functionalized TAB‐PEG 600‐TAB 5b as well as salts, and was directly used for the next synthesis step without further purification.

1H‐NMR (400, CDCl3): δ = 8.40 (s, 1 H), 6.99‐6.76 (m, 2 H), 4.64‐3.48 (m, 49 H), 3.03 (s, 9 H).

Cationic diblock surfactant TAB‐PEG 600‐PFPE 7000

A solution of propargyl PFPE 7000 3 (20.0 g, 2.84 g, 1.0 eq.) in HFE 7100 (17 mL) was prepared. TAB‐PEG 600 azide 5a (from the previous synthesis step), sodium ascorbate (112 mg, 568 µmol, 0.2 eq.), copper (II) sulfate pentahydrate (70.9 mg, 284 µmol, 0.1 eq.) and neocuproine (94.8 mg, 455 µmol, 0.16 eq.) were dissolved in a mixture of methanol (8 mL) and water (8 mL). Both the aqueous and the HFE 7100 solutions were mixed and stirred under reflux for 48 h. Afterward, the reaction mixture was cooled down to room temperature. An aqueous ethylenediaminetetraacetic acid solution (EDTA, 100 mm, 10 mL) was added and the reaction mixture was transferred into a separation funnel. It was overlaid with methanol (20 mL) and carefully swiveled. When the methanol became saturated and did not change its color anymore, it was replaced. This procedure was repeated until the methanol remained clear. After drying the fluorinated phase over magnesium sulfate, it was first filtered through Celite and then a PTFE syringe filter (0.45 µm). The solvent was removed under reduced pressure. The positively charged diblock surfactant was received as an orange, viscous oil (17.6 g, 1.13 mmol, 79.58%). A mixture of C6F6:C6D6 88:12 was used as the NMR solvent.

1H‐NMR (400, C6D6): δ = 8.29 (br s, 1 H), 7.88 (br s, 1 H), 6.16‐6.08 (m, 2 H), 5.72‐5.64 (m, 2 H), 4.31 (br s, 2 H), 4.25 (br s, 4 H), 4.12‐4.06 (m, 6 H), 3.88‐3.87 (m, 69 H).

Generation of Negatively and Positively Charged dsProCaps

Water‐in‐oil droplets with a charged water‐oil interphase served as a scaffold for the attraction of ECM proteins. To produce such stable water‐in‐oil droplets, a flow‐focusing droplet production device with a single inlet was used. To generate negatively charged dsProCaps, 5 w.% Perfluoro‐polyether‐poly(ethylene)glycol (PFPE‐PEG) block‐copolymer fluorosurfactants (008 PEG‐based fluorosurfactants, Ran Biotechnologies, Inc., USA) and 8 mm Krytox (157FSH, Chemours, 680‐272‐0) were dissolved in HFE‐7500 oil (3M m, USA). To produce positively charged dsProCaps, 5 mm of the custom‐made N⁺(Me)3‐PEG‐PFPE surfactant were dissolved in HFE‐7500 and used to generate simple positively charged dsProCaps. The aqueous phase for the differently charged dsProCaps was identical. It consisted of a mixture of the protein of choice (laminin‐111 or Matrigel) and 10 mm CaCl₂. 1.17 µm laminin‐111 or 2.4 µm Matrigel were mixed with 10 µm CaCl₂. In order to produce Matrigel beads, the concentration was increased to 20 µm. For producing 30–40 µmm droplets in diameter, the flow rates were set to 600 µl h⁻¹ for the oil phase and 300 µl h⁻¹ for the aqueous phase on the syringe pumps (Standard Infuse/Withdraw Pump 11 Elite Programmable Syringe Pump, Harvard Instruments). Small dsProCaps were produced by emulsifying 300 nm laminin‐111 protein or 200 nm Matrigel protein with 5 mm of the custom‐made N⁺(Me)3‐PEG‐PFPE surfactant dissolved in HFE‐7500 using an ULTRA‐TURRAX dispersing machine at 15 000 rpm for 40 s. Independent of size, all dsProCaps were collected in Eppendorf tubes and incubated for several hours at 37 °C to start the polymerization process.

Bacteria Culture Conditions

To grow E. coli, the bacterial strain was expanded in 25 g L⁻¹ Luria Bertani (LB, AppliChem A0954) broth supplemented with 50 µL mL⁻¹ kanamycin sulfate (Roth, T832.2). To start the culture, E. coli were scratched from the surface of a frozen glycerol stock and dissolved in 20 mL full LB media. The solution was incubated for 6 h at 37 °C at 250 rpm. Afterward, 15 µL of the expanded bacteria culture were transferred to a fresh glass beaker with 20 mL full LB media and left for expansion overnight under identical incubation conditions.

Encapsulation of E. coli into dsProCaps

To encapsulate bacteria inside dsProCaps a flow‐focusing microfluidic device with two inlets for aqueous solutions was used. The same surfactants as described for the generation of empty dsProCaps were used. Two phases needed to be prepared in advance. A high number of E. coli (OD 1.7 & 600 nm ) were resuspended in LB media with 50 µg mL⁻¹ kanamycin and collected in a 0.5 mL syringe. Since a double‐inlet device was used, the protein phases were slightly changed to: a) 1.17 µm laminin‐111 with 10% labelled laminin‐11 and 20 mm CaCl₂ in PBS and b) 2.4 µm Matrigel with 10% labelled laminin‐11 and 20 mm CaCl₂ in PBS. Bacteria‐laden dsProCaps were produced by using the following flow rates: 800 µL h⁻¹ for the oil phase, 200 µL h⁻¹ for the bacteria, and 300 µL h⁻¹ for the protein flow rate. All experiments were performed at least three times.

Release of Microcapsules into Physiologically Relevant Media

A bulk release approach was implemented to remove the stabilizing oil shell surrounding the protein capsules. 5 µL of empty or loaded dsProCaps were loaded into fresh Eppendorf tubes and covered with 20 µL 1x PBS. To break the surfactant shell, 20 µL of PFO (Perfluoro‐Octanol, Sigma, 370533‐25G) was added dropwise to the protein‐PBS mixture and the tilted Eppendorf tubes slowly rotated, resulting in microcapsule release into the aqueous phase. The PBS phase containing the ProCaps was collected and examined with a glass observation chamber under a confocal microscope (5% Laser power, 600 V gain, LSM900, Zeiss). All experiments were performed at least three times.

Zeta Potential Measurements

The zeta potential of the Laminin‐ and Matrigel‐based ProCaps (established using either positively or negatively charged surfactants) was measured by suspending the released ECM‐based ProCaps into a 1 mm KCl solution of pH 7. The dispersant viscosity, reflex index (RI) and dielectric constant of 1 mm KCl were set to 0.8882 cP, 1.330 and 79, respectively. Zeta potential measurements of the ECM‐based ProCaps were performed with a Malvern Panalytical Zetasizer Nano ZS in DTS1070 folded capillary Zeta Cells at 25 °C with a V_max of 25 V and 2 min equilibration time. The RI of the ECM protein‐based microcapsules were set to be 1.45. Three consecutive measurements with 100 runs each were evaluated.

Confocal and Scanning Electron Microscopy

To validate the bacterial filamentous shape with scanning electron microscopy, the samples were first fixed with 3% Glutaraldehyde in PBS, pH 7.3 for 16 h at 4 °C. Afterward the samples were treated with an ascending ethanol and acetone series from 30% to 100% (each step lasting for 1 min at 25 °C). The fixed samples were dried with a critical point dryer in 100% ethanol for 2 h at 1 bar and 28 °C. To be able to visualize the samples with the SEM, the samples were sputtered with a 10 nmm gold layer. Then the samples were loaded in the SEM and images were taken with the SE2 detector at 4 kV using the magnifications given in the figure legends. Further, confocal images were taken with a Zeiss LSM 900 confocal laser scanning microscope with 5% laser power and 600 V gain.

Statistical Analysis

All experimental data were arranged in Microsoft Excel Software and further analyzed in GraphPad Prism Software. These data were plotted as mean ± standard deviation (SD). To evaluate the correlation and obtain p‐values an unpaired t‐test and a one‐way ANOVA was performed.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Supplemental Video 1

Supplemental Video 2

Supplemental Video 3

Supplemental Video 4

Supplemental Video 5

Pashapour S., Seneca S., Schröter M., Frischknecht F., Platzman I., Spatz J., Design and Development of Extracellular Matrix Protein‐Based Microcapsules as Tools for Bacteria Investigation. Adv. Healthcare Mater. 2023, 12, 2202789. 10.1002/adhm.202202789

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.;