**4. Physical properties of basal lamina gels** *in vitro*

In addition to the biochemical structure of its constituents, the following three physical parameters dictate the behavior of molecules, nanoparticles, or cells within the basal lamina: the microstructure, the mechanical properties, and the permeability of the hydrogel. In biopolymer networks, the microstructure of the system has direct implications on both the viscoelastic properties of the network [51] and its permeability properties [31]. Thus, imaging methods for visualizing biopolymer networks such as the basal lamina are discussed first.

## **4.1. Microstructure of ECM gels**

permeability studies and the correlation of the experimental results with physicochemical principles. Thus, a detailed investigation of the selective permeability properties of a complex biopolymer barrier such as the basal lamina requires a reliable *in vitro* model system, which is available in quantities large enough to conduct systematic tests while reproducing the behavior

A suitable source for the purification of an extracellular matrix that mimics the basal lamina is the Engelbreth–Holm–Swarm sarcoma of mice. This tumor produces, in contrast to healthy tissue, large amounts of ECM with laminin and collagen IV being the main components [3]. Depending on the question asked, individual macromolecular components of the basal lamina may be sufficient to take over the role of the complex biopolymer mixture. For instance, adhesion of cells to solid substrates is promoted similarly well by laminin coatings as by coatings with the multicomponent ECM [40, 41]. For other basal lamina properties such as viscoelasticity and selective permeability, it is crucial that the biological complexity of the

system is maintained so that the full spectrum of basal lamina function is obtained.

The high abundance of ECM in Engelbreth–Holm–Swarm tumor tissue makes it possible to purify reasonable amounts of this multicomponent matrix as required for systematic *in vitro* experiments. A first purification protocol for this ECM was established by Kleinman et al. [42, 43] in the 1980s. The extract is liquid at temperatures between 4°C and approx. 15°C and forms a gel at higher temperatures. In its gel form, the matrix was tested for its biological activity, and it was shown in several studies that the purified ECM successfully promotes the differ‐ entiation of various cell types [44–47]. Cells can be either plated on top of the gel, thus simulating a two-dimensional (2D) environment, or they can be embedded into a 3D ECM matrix. Which configuration is chosen depends on the detailed experimental setup, the cell type used and the biological question. For instance, cell migration experiments can be con‐ ducted both on flat surfaces which have been coated by ECM components and in 3-dimensional

The purification protocol of Kleinman et al. is used by several companies for the commercial production of ECM. Although these commercial ECM variants are extracted according to the same purification protocol, significant differences in the behavior of cells embedded into those gels have recently been described [50]: The migration behavior of leukocyte-like dHL-60 cells in four different commercially available ECM gel variants differed strongly even though the gels were prepared at matching total protein concentrations. Moreover, in one of the ECM gels, life–dead stains demonstrated a significantly increased percentage of nonviable cells. At the same time, for this gel variant, there was an additional band visible when the gel was analyzed by SDS-PAGE. Mass spectrometry showed that this additional band contained laminin fragments which indeed are suspected to be harmful for cells. This result demonstrates the dilemma a researcher is exposed to when working with commercial model systems: On the one hand, the relatively easy availability of the material in reasonable quantities allows for conducting *in vitro* experiments which otherwise would not be possible. On the other hand,

of the *in vivo* basal lamina layer*.*

basal lamina gels [48, 49].

**3. Basal lamina model systems**

208 Composition and Function of the Extracellular Matrix in the Human Body

There are various methods to evaluate the structure of a material, and those methods can be subdivided into the following categories: surface imaging techniques, near-field/contact-based techniques and far-field imaging. Which method is used to resolve the structure depends on the material and on the experimental question: Do I require information on the surface topology or on the inner structure of the material? For biological samples, a fixation is needed for most of the imaging techniques so that the structure does not change over time or during the sample preparation process. One technique which is often used to image biological samples is fluorescence confocal microscopy as this method can visualize the 3D structure of a biopol‐ ymer network. However, most biological samples are not fluorescent by themselves and thus a fluorescent dye has to be used to stain the structure of interest. For many target proteins, commercial antibodies are available to which a fluorescent dye is attached. Before such a staining with antibodies is performed, the samples are typically fixed to ensure that the structure of the biopolymer network is not altered by the antibody application and the following washing step.

A suitable technique for imaging the surface of a biopolymer material is scanning electron microscopy (SEM). For this technique, the sample surface needs to be electrically conductive. Since this is typically not the case for biological samples, the application of a thin conductive layer, for example gold, is necessary. Depending on the type of SEM used for imaging, the samples are also exposed to a vacuum for imaging; this requires sample fixation and subse‐ quent dehydration as typical additional preparation steps for this imaging method. It is clear that sample preparation steps necessary for both imaging techniques may introduce artifacts, that is, alterations of the microstructure of the biopolymer network. However, as the prepa‐ ration steps for both techniques are different, the obtained pictures are reliable if both imaging methods return comparable structures.

**Figure 3.** Microstructure of four ECM gel variants. Images obtained with immunostaining/confocal microscopy and with a SEM are compared. Both methods show a denser network for the gel variant 2 (ECM2) than for the other three variants. Edited figure with permission from Arends et al. [50]. © 2015 Arends et al. Published under CC BY license.

Example images of basal lamina model systems are shown in **Figure 3**, where the microstruc‐ ture of four different ECM gel variants is compared. For the fluorescent confocal image, the ECM component laminin was stained with a fluorescent antibody and an optical slice with a thickness of 0.9 μm was acquired inside the 3-dimensional gel. Here, the ECM variant in which leukocyte migration was slowed down most (ECM2) shows the lowest porosity. The same difference in the microarchitecture of the ECM gels is obtained when SEM is used for imaging: ECM2 shows the highest density, whereas the other gel variants exhibit a comparable network structure. As the four gel variants have all been reconstituted at identical total protein concentrations, the observed structural difference is most likely due to the higher content of the cross-linking molecule entactin in this ECM variant as detected by SDS-PAGE.

#### **4.2. Viscoelastic properties of ECM gels**

Especially for cell differentiation, the mechanical properties of the ECM play an important role. Using artificial hydrogels such as cross-linked polyacrylamide gels [52], it was shown that cell differentiation can be directed by the stiffness of the substrate. The ECM is a viscoelastic material, that is, its mechanical behavior combines both viscous as well as elastic properties. Those viscoelastic properties can be probed macroscopically with a shear rheometer as illustrated in **Figure 4** as well as microscopically with single-particle tracking microrheology or AFM indentation.

layer, for example gold, is necessary. Depending on the type of SEM used for imaging, the samples are also exposed to a vacuum for imaging; this requires sample fixation and subse‐ quent dehydration as typical additional preparation steps for this imaging method. It is clear that sample preparation steps necessary for both imaging techniques may introduce artifacts, that is, alterations of the microstructure of the biopolymer network. However, as the prepa‐ ration steps for both techniques are different, the obtained pictures are reliable if both imaging

**Figure 3.** Microstructure of four ECM gel variants. Images obtained with immunostaining/confocal microscopy and with a SEM are compared. Both methods show a denser network for the gel variant 2 (ECM2) than for the other three variants. Edited figure with permission from Arends et al. [50]. © 2015 Arends et al. Published under CC BY license.

Example images of basal lamina model systems are shown in **Figure 3**, where the microstruc‐ ture of four different ECM gel variants is compared. For the fluorescent confocal image, the ECM component laminin was stained with a fluorescent antibody and an optical slice with a thickness of 0.9 μm was acquired inside the 3-dimensional gel. Here, the ECM variant in which leukocyte migration was slowed down most (ECM2) shows the lowest porosity. The same difference in the microarchitecture of the ECM gels is obtained when SEM is used for imaging: ECM2 shows the highest density, whereas the other gel variants exhibit a comparable network structure. As the four gel variants have all been reconstituted at identical total protein concentrations, the observed structural difference is most likely due to the higher content of

the cross-linking molecule entactin in this ECM variant as detected by SDS-PAGE.

Especially for cell differentiation, the mechanical properties of the ECM play an important role. Using artificial hydrogels such as cross-linked polyacrylamide gels [52], it was shown that cell differentiation can be directed by the stiffness of the substrate. The ECM is a viscoelastic material, that is, its mechanical behavior combines both viscous as well as elastic properties. Those viscoelastic properties can be probed macroscopically with a shear rheometer as

methods return comparable structures.

210 Composition and Function of the Extracellular Matrix in the Human Body

**4.2. Viscoelastic properties of ECM gels**

In macroscopic shear rheology, the sample is placed between two plates, a stationary bottom plate and a rotating top plate for shear stress application. The bottom plate can be heated or cooled depending on the desired temperature conditions. The top plates are available in various diameters and shapes and are chosen according to the sample properties and the quantity to be measured. For determining viscoelastic properties, the top plate is typically oscillated at different frequencies, either at a fixed strain or at a fixed torque. Small torques during such a measurement ensure that the material response is quantified in the linear response regime, where Hooke's law holds, that is, where the ratio of stress and strain is independent from the amplitude of the applied force.

**Figure 4.** The viscoelastic behavior of ECM gels can be quantified by shear rheology. The sample is placed between a stationary bottom plate and a measuring plate, and then, an oscillating shear stress is induced. The temporal delay (phase shift) of the material response is measured. For a purely elastic material, the phase shift is 0°; in contrast, a phase shift of 90° is obtained for a purely viscous substance. For a viscoelastic material, the phase shift can assume any value between 0° and 90° (adapted from an illustration by Stefan Grumbein). A typical gelation curve for a basal lami‐ na gel at a concentration of 8.3 mg/mL is shown at the bottom right and was acquired using a plate–plate geometry oscillating at a frequency of 1 Hz. For the first 200 s, the measurement was performed at 4°C. At this temperature, the ECM is in its liquid state, thus the loss modulus (open circles) dominates over the storage modulus (full circles). When the temperature is increased to 37°C, gelation is initiated and the storage modulus dominates. After a few minutes, a plateau value is reached. A typical frequency spectrum after gelation is shown as inset. The storage modulus domi‐ nates over the loss modulus over four decades of frequencies.

The viscoelastic properties of the ECM can then be described by the storage modulus *G*' and the loss modulus *G*". Here, *G*' is a measure for the elastic properties and *G*" for the viscous properties of the ECM gel. At low temperatures around 4°C, the ECM is in a liquid state. Here, its viscous properties dominate and the loss modulus is larger than the storage modulus. When the ECM gel is heated to room temperature or above, a gelation process is initiated which results in an increased storage modulus: Within a few minutes after the temperature increase is applied, the storage modulus starts to dominate over the loss modulus and then increases further until it reaches a plateau value (see **Figure 4**). In general, there are several parameters determining the absolute value of the plateau elasticity: The higher the concentration of proteins/polymers the higher is typically the storage modulus [53]. In addition to the concen‐ tration of protein, also the type of polymer/polymer interaction plays a role. The storage modulus of an entangled solution is usually lower than for a cross-linked network. In the study conducted in [50], it was shown that the amount of the cross-linking molecule entactin influences the network stiffness: The higher the concentration of entactin the higher the storage modulus and thus the elasticity of the formed matrix. Typical values for the elastic modulus obtained for ECM gels at a total protein concentration of 3.5 mg/mL are in the range of 1-10 Pa which is very soft and lies in the range of moduli reported to induce neuron-like differentiation of stem cells [52].

In general, the viscoelastic properties of a biopolymer network may depend strongly on the probing frequency [51], especially if the network constituents are only entangled. For crosslinked systems, however, a pronounced plateau in the frequency-dependent shear moduli is expected, and exactly such behavior is also observed for ECM gels (**Figure 4**).

The absolute values of the viscoelastic parameters obtained with macrorheology may not necessarily reflect the local stiffness of a biopolymer network. Thus, microrheological techni‐ ques such as bead microrheology [54] or AFM nanoindentation [55] have been introduced and already applied to other biopolymer systems such as cytoskeletal networks [56–58] or cartilage [59]. With those nano-/microscopic techniques, it is also possible to spatially map the mechan‐ ical properties of native basement membranes [60, 61], which might give insights important for cellular processes such as differentiation or migration.

#### **4.3. Permeability of ECM gels**

One of the major tasks of the basal lamina is to act as a molecular filter. Here, the exclusion of particles or molecules according to their size is one of the simplest mechanisms for establishing permeability: A mesh size smaller or in the order of the particle diameter will prevent the entrance of particles into the network; conversely, if particles have already entered the network, they will be efficiently trapped within the biopolymer matrix. However, this filter mechanism is not very sophisticated as it cannot differentiate between objects of the same size. Thus, a second filter mechanism based on binding interactions between diffusing particles/ molecules and the basal lamina constituents has been put forward to contribute to the selective permeability properties of biopolymer hydrogels such as the basal lamina [31]. With the ECM model system discussed above, the physicochemical principles governing the high selectivity of basal lamina gels can be studied systematically.

To probe the interactions between particles and the ECM, single-particle tracking (SPT) can be employed. In contrast to SPT used for microrheology [62], the diameter of the particles embedded into the ECM should be small compared to the mesh size of the gel. Only then one can be sure that the particle motion is not geometrically restricted by the network microarch‐ itecture—which demonstrates the importance of obtaining structural information on the system prior to commencing SPT experiments. In SPT measurements, the diffusive movement of particles within the gel is recorded via light microscopy and every single particle is evaluated separately. The trajectory of motion of each particle, in particular the x- and y-position, is extracted from recorded movies for every frame of the movie—typically over a time course of several seconds up to a minute (depending on the temporal resolution of the image acquisition process, **Figure 5**). These data are then used to calculate the mean squared displacement (*MSD*) of every particle according to the following

is applied, the storage modulus starts to dominate over the loss modulus and then increases further until it reaches a plateau value (see **Figure 4**). In general, there are several parameters determining the absolute value of the plateau elasticity: The higher the concentration of proteins/polymers the higher is typically the storage modulus [53]. In addition to the concen‐ tration of protein, also the type of polymer/polymer interaction plays a role. The storage modulus of an entangled solution is usually lower than for a cross-linked network. In the study conducted in [50], it was shown that the amount of the cross-linking molecule entactin influences the network stiffness: The higher the concentration of entactin the higher the storage modulus and thus the elasticity of the formed matrix. Typical values for the elastic modulus obtained for ECM gels at a total protein concentration of 3.5 mg/mL are in the range of 1-10 Pa which is very soft and lies in the range of moduli reported to induce neuron-like differentiation

In general, the viscoelastic properties of a biopolymer network may depend strongly on the probing frequency [51], especially if the network constituents are only entangled. For crosslinked systems, however, a pronounced plateau in the frequency-dependent shear moduli is

The absolute values of the viscoelastic parameters obtained with macrorheology may not necessarily reflect the local stiffness of a biopolymer network. Thus, microrheological techni‐ ques such as bead microrheology [54] or AFM nanoindentation [55] have been introduced and already applied to other biopolymer systems such as cytoskeletal networks [56–58] or cartilage [59]. With those nano-/microscopic techniques, it is also possible to spatially map the mechan‐ ical properties of native basement membranes [60, 61], which might give insights important

One of the major tasks of the basal lamina is to act as a molecular filter. Here, the exclusion of particles or molecules according to their size is one of the simplest mechanisms for establishing permeability: A mesh size smaller or in the order of the particle diameter will prevent the entrance of particles into the network; conversely, if particles have already entered the network, they will be efficiently trapped within the biopolymer matrix. However, this filter mechanism is not very sophisticated as it cannot differentiate between objects of the same size. Thus, a second filter mechanism based on binding interactions between diffusing particles/ molecules and the basal lamina constituents has been put forward to contribute to the selective permeability properties of biopolymer hydrogels such as the basal lamina [31]. With the ECM model system discussed above, the physicochemical principles governing the high selectivity

To probe the interactions between particles and the ECM, single-particle tracking (SPT) can be employed. In contrast to SPT used for microrheology [62], the diameter of the particles embedded into the ECM should be small compared to the mesh size of the gel. Only then one can be sure that the particle motion is not geometrically restricted by the network microarch‐ itecture—which demonstrates the importance of obtaining structural information on the system prior to commencing SPT experiments. In SPT measurements, the diffusive movement

expected, and exactly such behavior is also observed for ECM gels (**Figure 4**).

for cellular processes such as differentiation or migration.

212 Composition and Function of the Extracellular Matrix in the Human Body

of basal lamina gels can be studied systematically.

**4.3. Permeability of ECM gels**

of stem cells [52].

$$MSD(\tau) = \frac{1}{N} \sum\_{l=1}^{N} [\bar{r}\left(i\Delta t + \tau\right) - \bar{r}\left(i\Delta t\right)]^2\tag{1}$$

Here, *N* denotes the total number of recorded frames, *r* ⃑ (*t*) is the position of the particle at time *t,* and τ denotes the time interval between two particle positions within a given trajectory. For diffusive processes, the *MSD* typically grows with time as a power law τα, with the exponent α characterizing the type of diffusive motion: One can distinguish sub-diffusive (α<1), normal diffusive (α=1), or superdiffusive behavior (α>1), the latter of which is typically linked to active transport phenomena or liquid flow.

Such SPT experiments revealed that both positively and negatively charged microparticles were efficiently immobilized in the ECM gel, whereas PEGylated (and thus only weakly charged) polystyrene particles of identical size were able to diffuse almost freely within the gel [63]. Equivalent results were obtained with liposome particles and suggested that free diffusion within the ECM matrix is only possible as long as the particle surface charge (as quantified by the zeta potential) lies within a window ranging from intermediate negative charge to low positive charge. Enzymatic digestion of the ECM component HS entailed a mobilization of positively charged particles. This finding suggested that the polyanionic HS chains present in the perlecan complex critically contribute to the selective properties of the ECM gel—likely through trapping of positively charged objects by means of electrostatic binding.

The notion that electrostatic binding interactions contribute to particle trapping in ECM gels was confirmed by experiments conducted at elevated ionic strength of the hydrogel buffer. Increased salt concentrations lead to charge screening effects by the formation of a layer of counter ions around the surface of charged objects such as particles or hydrogel polymers. As a consequence, the strength of electrostatic interactions at a given separation distance between two objects is reduced—a process which is described by the Debye–Hückel theory [64]. At physiological concentrations of KCl, both positively and negatively charged polystyrene microparticles are immobilized in ECM gels. However, when the KCl concentration is increased, a fraction of the particles becomes mobile [63, 65]. This mobilization does not have to be permanent as individual particles can dynamically switch between a freely diffusing and bound state over time, and—while in the bound state—also between a weakly and strongly bound configuration. As shown in **Figure 5**, the degree of particle mobilization depends both on the ion concentration and valency which is consistent with the Debye–Hückel theory. However, particle mobilization efficiency seems also to depend on the particular ion species as identical concentrations of the divalent ions Mg2+ and Ca2+ lead to different experimental outcome [65]. This ion-specific effect suggests that, in addition to electrostatic forces, also hydrophobic interactions are likely to contribute to the selective filtering properties of the basal lamina.

Systematic permeability studies with artificial particles were very helpful to unravel the physical mechanisms which are responsible for the trapping of solutes in the basal lamina. However, most compounds which encounter the basal lamina layer under physiological conditions are small molecules rather than microparticles. To investigate the selective prop‐ erties of the basal lamina toward small molecules, a microfluidic setup (**Figure 6**) was recently introduced [66]. Here, customized peptides with tailored amino acid sequences and thus different net charges were used as diffusion probes. To ensure optimal comparability, the molecular weight of those oligopeptides was kept constant. The penetration behavior of those peptides into an ECM gel was visualized by fluorescent microscopy, and similar to the SPT experiments discussed above also the behavior of those molecules critically depended on their charge. Positively charged peptides accumulated at the gel/buffer interface, whereas nega‐

**Figure 5.** Single-particle tracking experiments can locally map the permeability of ECM gels. *Upper panel*: Exemplary trajectories and the corresponding MSD curves of a particle showing free diffusion and a particle showing retarded diffusion. For freely diffusing particles, the dependence of the MSD on time is linear and a diffusion coefficient *D* can be calculated according to the formula shown in the graph. For the calculation of *D,* only the first 10% of the MSD data is used to avoid errors arising from statistical uncertainties. *Lower panel*: Trajectories of a particle which transiently switches between a diffusing and a bound state and of a particle which alternates between a strongly and a weakly bound configuration. The states of motion can be distinguished based on the fluctuation amplitude of the particle. Both trajectories were obtained at a salt concentration of 1 M KCl. The histogram shows that in ECM gels, the fraction of mobile particles depends on the concentration, valency, and detailed species of the ions used. Adapted with permis‐ sion from Arends et al. [65]. Copyright ©2013 American Chemical Society.

tively charged peptides did not. Moreover, when the net charge of the positively charged peptides was increased, the accumulation propensity of the molecules at the gel interface was increased as well. Of course, such an artificial microfluidic setup does not reproduce the complex situation of the basal lamina interface found *in vivo.* However, peptide injection tests in the connective tissue of living mice demonstrated a similar charge-selective accumulation behavior at the basal lamina layer of blood vessels as observed on-chip with the simplified ECM/buffer interface. This underscores the great potential basal lamina model systems and biophysical characterization methods hold for gaining a better insight into the mechanistic principles that establish the complex properties of the basal lamina.
