An increasing demand for tissue implants to sustain a healthy and active life for an aging population makes the development of alternatives to transplanted autologous tissue of high interest. This has promoted the development of advanced polymeric biomaterials with properties comparable to those of native tissue and structure favorable for cell integration, for example cellulose synthesized by bacteria of the strain Acetobacter xylinum is emerging as an attractive alternative.1, 2 The random network of nanofibrils forms a hydrogel, which despite its exceptionally high water content (98%) possesses high mechanical strength,3 making it a favorable material for scaffold applications. It also shows good biocompatibility4 with low protein adsorption5 and high permeability for liquids and gases. Cultivation can be made in the presence of molds and porogens of different sizes and shapes, in order to design desirable scaffold geometries at micro- and macro scales.3, 6 Successful results have been reported for its use as urethra-, trachea-, and digestive tract substitutes, as well as meniscus, bone, and skin graft materials.1
Of increasing interest is the development of tubular cellulose scaffolds for its use as artificial blood vessels.3, 7 Cardiovascular diseases form a world-wide health problem and have become the number one cause of death globally.8 Only in the United States, ∼500,000 coronary-artery9 and ∼1,000,000 peripheral-vascular bypass procedures are performed annually.10 The current standard is to use the saphenous vein or the internal mammary artery as grafts.11, 12 However, one-third of the patients lack useable autologous transplants due to inadequate size or quality.13 In addition, patients frequently need multiple-segment grafts or repeated surgery. Thus, there is an urgent need for artificial vascular implants. Advanced efforts are made to supplement the scaffold matrix with cells to obtain a fully functional graft. To facilitate the blood flow through the vessel and reduce the risk for thrombosis, tubular cellulose scaffolds are augmented with a luminal layer of endothelial cells.14 Further functionalization of the cellulose vessels includes seeding them with smooth muscle cells (SMCs),15 which can be stimulated to pulsatile dilatations and contractions of the vessel, further facilitating blood flow with reduced risk of clot formation. In addition, SMCs synthesize large amounts of extracellular matrix components, such as collagen and elastin, promoting long-term adaptation of the vascular implant to varying physiological conditions such as exercise and vascular injuries16 with the hope for life-lasting use of the implant.
The ability of SMCs to adhere to and proliferate on a synthetic scaffold is dependent on the chemical composition of the material used as well as its structure.6, 17, 18, 19, 20 To improve our understanding for how the material properties affect cell integration and proliferation, reliable methods to characterize the morphology of the scaffold matrix and its interaction with living cells are needed. Different microscopy techniques are, for this purpose, used within tissue engineering such as fluorescence microscopy, electron microscopy, and atomic force microscopy. Unfortunately, all are associated with severe limitations, preventing studies of soft, wet materials such as cellulose.21, 22, 23, 24 In many situations only the most superficial sample layer can be probed, providing information of merely topographical character without access to the full three-dimensional morphology below the surface. In addition, the many sample preparation procedures such as labeling, fixation, dehydration, freeze-drying, and material coating, potentially affect both sample structure and composition; hence the sample is studied in a modified state. Many of these limitations are avoided by applying any of the increasingly used nonlinear microscopy techniques, which allow for minimally-invasive, high-resolution, three-dimensional imaging of soft, organic materials and living cells. The spatial resolution of nonlinear techniques (typically 300 nm laterally and 1 μm axially) together with the ability to probe deep into samples make these methods clearly suitable for characterization of tissue engineering constructs often consisting of nano- to micrometer-sized structures arranged in thicker three-dimensional matrices.
Nonlinear microscopy techniques such as second harmonic generation (SHG) and two-photon fluorescence (TPF) have successfully been applied for imaging of scaffold structure, extracellular matrix components, and cells in tissue engineering.21 With this work we show the strength of yet another nonlinear microscopy technique for tissue engineering applications, coherent anti-Stokes Raman scattering (CARS) microscopy, probing vibrations of molecular bonds for image contrast. The high vibrational Raman cross sections of many hydrogen-carbon bonds make the technique suitable for imaging of polymers25, 26 and in particular lipids, the latter by probing CH2 groups in the lipid acyl chains.27, 28 Tissue engineering constructs in many cases consist of a polymer scaffold material seeded with cells for further growth and regeneration. CARS microscopy has the ability to selectively visualize the polymer scaffold but also cell arrangements by probing lipid-rich structures and membranes, making it a useful instrument within tissue engineering. Moreover, the combination of CARS microscopy with other nonlinear methods, such as SHG and TPF, permits simultaneous imaging of multiple sample structures and hence their interaction; the scaffold modules, the cells, and native extracellular matrix components such as collagens.
In the present work, this capability is exemplified with studies on the integration of human SMCs in cellulose scaffolds using combined CARS and SHG microscopy. The morphology of the scaffold and its impact on the proliferation of SMCs is monitored at sub-micrometer resolution and in full three dimensions. Whereas CARS microscopy visualizes the arrangement of the cells in the scaffold, the repetitively ordered molecular arrangement of the cellulose fibers makes it possible to visualize them with SHG microscopy.23, 29 Thus, by combining the two microscopy modalities, the interaction between cells and cellulose fibers can be assessed in detail and the full scenario of cell integration, proliferation, and differentiation can be followed in the cellulose matrix to obtain detailed knowledge on these for tissue engineering important processes.
Materials and Methods
Smooth Muscle Cell Isolation and Culture
Human smooth muscle cells were isolated from saphenous veins available as excess material from bypass surgery with two anonymous donors; hence no ethical permission was required. Cells were isolated from the tunica media by explant technique. Briefly, the media was separated from the intima and adventitia, the tissue was cut into 1–2 mm2 pieces and placed in six-well polystyrene tissue culture plates. Tissue pieces were covered with cover slips and the wells filled with 2 mL Dulbecco´s Modified Eagle Medium supplemented with 20% fetal calf serum (FCS), penicillin/streptomycin (5U/mL), Na-pyruvate (1 mM), nonessential amino acids (0.1 mM) and L-glutamine (2 mM, Invitrogen, Paisly, UK). Medium, supplements, and solutions were obtained from PAA (Linz, Austria), if not otherwise specified. Cells were grown in humidified incubators at 37°C and with 5% CO2. At confluence, cells were detached with trypsin/EDTA and expanded. Further culturing after the first passage was done in medium containing 10% FCS and the same amount of supplements as before. SMCs were identified by morphological appearance as well as by positive immunostaining and detection of smooth muscle α-actin (SM α-actin). All cultures were negative for mycoplasma.
Preparation of Cellulose Scaffolds with SMCs
The cellulose production followed the concept of Bodin 3 Briefly, cellulose was produced by Acetobacter xylinum fermentation, forming scaffolds of tubular geometry using a configuration with a silicon tube immersed into the culture. A flow of pure oxygen was running through the semi-permeable silicon tube and cellulose was produced at the interface between the bacterial suspension and the tube, where the bacteria have access to oxygen. After a growth period of seven days, the cellulose tubes were purified by alkali treatment, repeatedly washed, and finally autoclaved. Circular subsamples, 0.8 mm in diameter, of cellulose were punched out and placed on microplates, giving access to either the compact, luminal side of the tubular scaffold or the porous, external side. This enabled investigations on the impact of scaffold morphology on the cell integration. The cellulose samples were seeded with cultures of SMCs proceed until passage 4–5 at a density of 10 000 cells per cm2. Cells were subsequently cultured for three and eight days at which time points the specimens were analyzed by CARS and SHG microscopy, requiring no sample preparation. An overview of the logistics of the experiments is given in Fig. 1.
The setup for nonlinear microscopy [see schematic in Fig. 1e] consists of a pulsed, near-infrared laser system combined with an inverted Nikon Eclipse TE2000-E microscope, equipped with a Nikon C1 scanning head. The laser system consists of a high-power Nd:Vanadate laser (HighQ Laser Production GmbH, IC-LASER-10000-ps), pumping an optical parametric oscillator (APE GmbH, Levante Emerald OPO). The pump laser emits pulses with duration of 7–8 ps at a repetition rate of 76 MHz. The output power is split between the fundamental wavelength 1064 nm and the second-harmonic at 532 nm with typical average powers of 500 mW and 4W, respectively. The 532 nm beam is used to pump the Emerald OPO, which generates tunable wavelengths in ranges 690–990 nm (signal beam) and 1150–2300 nm (idler beam). The inherently synchronized OPO and 1064 nm beams are combined using dichroic mirrors and coupled into the microscope through the scanning unit. Temporal overlap of the pulses at the sample is achieved by use of a variable delay line [marked D in Fig. 1e] in the OPO beam path. The 1064 nm beam serves as the Stokes beam in the CARS process and the output from the OPO tuned to 817 nm as the pump/probe beam for excitation of the symmetric stretch vibration of CH2 groups at 2845 cm−1. In addition to the CARS signal generation, the OPO beam induces the SHG process in the sample. The beams illuminate the sample from below (inverted microscope) and are focused using an oil immersion objective (40× Nikon Plan Fluor N.A. 1.3). Typical average laser powers at the sample position are 40 mW for the 817 nm beam and 20 mW for the 1064 nm beam.
The generated signals can either be collected with a lens mounted in the forward, transmission direction or by the objective in epi-mode (back-reflected). All data presented in this work were collected in epi-direction and the signals were registered by single-photon counting photomultiplier tubes (Hamamatsu PMC 100–1 for SHG and Hamamatsu PMC 100–20 for CARS). A dichroic beam splitter separates the 409 nm SHG signal and the 663 nm CARS signal to enable simultaneous detection using two different detectors. Band-pass filters (2 Chroma HQ400/40 for SHG and 2 Chroma HQ665/65+2 Chroma HQ650/50 for CARS) are used to isolate the signals by efficient suppression of scattered excitation light.
Two-dimensional microscopy images are recorded point-wise by scanning the laser beams over the sample and three-dimensional imaging (z-stacks) is achieved by vertical translation of the objective for measurements in different horizontal planes.
For microscopy measurements cellulose scaffolds seeded with SMCs were placed in Petri dishes with a 0.17 mm thick glass bottom (Fig. 1). A small volume of medium was added to the sample to prevent drying. Visual information on the arrangement of cellulose fibers and SMCs was acquired as single images as well as z-stacks with fields of view ranging from 40×40 to 200×200 μm. Image acquisition times were 20 or 60 seconds per image.
Results and Discussion
Nonlinear Microscopy of Human Smooth Muscle Cells and Cellulose Scaffold
Figure 2a shows an ensemble of SMCs established on a cellulose scaffold of compact architecture after eight days of growth visualized by means of conventional differential interference contrast (DIC) microscopy. The cell shapes and inner structures can be clearly distinguished, whereas the cellulose fibers are invisible. As a consequence, the image lacks information on the actual contact and interaction between SMCs and the supporting cellulose scaffold. However, this information can be obtained from the nonlinear microscopy images shown in Figs. 2b, 2c, 2d, 2e, where the fibrillar cellulose material and cells are visualized by means of SHG and CARS microscopy, respectively. The fibrous structure of the cellulose surrounding the SMCs is visualized by SHG microscopy in Fig. 2b and the two SMCs grown on the cellulose are shown by CARS microscopy in Fig. 2c. Both cells are elongated and the nucleus in the left cell appears as a darker central region. Small intracellular structures generate high CARS signal at 2845 cm−1, which is characteristic for lipid-rich organelles. Intracellular lipid deposits as well as micrometer-sized, membrane-rich bodies, possibly representing autophagic vacuoles have been identified in SMCs using electron microscopy30 and both of these structures could give high CARS signals. By comparing the CARS and SHG images, it can be observed that SMCs also generate a weak SHG signal in the SHG image. This is in agreement with the composition of SMCs, containing the SHG-generating structural protein myosin.31 In contrast to skeletal muscle tissue, the arrangement is less ordered without the striation pattern, therefore generating a low SHG signal. The overlay image and the three-dimensional rendering shown in Figs. 2d and 2e both show that the cells are immersed into the cellulose matrix. The optical sectioning capabilities of nonlinear microscopy gives valuable three-dimensional information as shown in Fig. 2e, from which the migration depth can be determined to approximately 10 μm after eight days of growth on a compact cellulose matrix.
The combination of CARS and SHG microscopy permits detailed visualization of the contact between cells and scaffold as exemplified by the images of Fig. 3. The close-up in Fig. 3b shows cellular filopodia interspersed between the cellulose fibers. This signifies that a natural and close interaction has been formed between the SMCs and the cellulose fibers, as the launch of filopodia requires that adhesion points on the biopolymer surface have been established. Filopodia are also visualized in Fig. 4, where a three-dimensional rendering shows that they extend on top of the cellulose matrix [Fig. 4b, indicated by row of arrows] while the majority of the SMC is submerged [Fig. 4b, single arrow]. In addition, the upper-most cellulose layer [Fig. 4a] here exhibits an irregular wavelike pattern that differs from the more homogeneous arrangement seen in deeper layers of the stack, Fig. 4c. One possible explanation for this pattern could be that the proliferation of filopodia induces local mechanical stresses in the cellulose, causing rearrangement of the material. Scaffold material contractions due to growth of SMCs have been reported for polyglycolic acid (PGA) scaffolds seeded with SMCs after three weeks of growth.17
The applied excitation power is an important parameter when using nonlinear microscopy for characterization of samples as presented in this paper and merits some further discussion. The combined CARS and SHG measurements were carried out using picosecond excitation. Fu 32 have investigated sample damage mechanisms during picosecond CARS microscopy on a carcinoma cell line and observed no photo-induced damage using a total peak power at the sample of 120 W and an intensity of approximately 0.2 TW/cm2. In addition, they investigated dense spinal tissue, for which a peak power of 450 W did not induce sample damage. In our experiments the peak power was 110 W, corresponding to an intensity of 0.1 TW/cm2, i.e., similar to the lower values in their study. This, together with the fact that no signs of sample damage were observed in our images, assures that the powers used were safe. For more sensitive samples excitation with femtosecond pulses could be an option, allowing for lower average powers. Femtosecond excitation is favorable for SHG microscopy, confirmed by a comparison with picosecond excitation illustrating that similar signal-to-noise ratio and image quality is obtained at 60 times lower average power.33 However, for CARS microscopy, femtosecond excitation is less suitable since it provides significantly lower contrast due to the broad spectral line-profile of femtosecond pulses, resulting in increased excitation of the nonresonant background signal. This drawback of femtosecond CARS has been thoroughly analyzed34 and experimentally demonstrated.33 Thus, for simultaneous CARS and SHG microscopy, picosecond excitation is preferably applied, restricting the excitation powers and intensities to values indicated by Fu 32
Influence of Scaffold Morphology on Cell Migration
Figure 5 shows CARS and SHG overlay images measured on SMCs grown on cellulose scaffolds of different architecture for three days; compact [Figs. 5a and 5b] and porous [Figs. 5c and 5d]. In contrast to the eight-day samples in Figs. 2, 3, 4, we note that the cells exhibit a rounded shape and have not adopted the elongated shape characteristic of developed SMCs.35 This could explain, in general, the higher cell densities observed in the three-day samples compared to the eight-day samples, as the lateral cell migration has not yet been able to take place. However, the difference in local cell densities may also be attributed to seeding conditions.
The images of the compact scaffold in Figs. 5a and 5b are measured at the same lateral position in the sample but at different depths separated by 20 μm. It illustrates that a confluent multiple-cell layer [Fig. 5a] is already formed after three days’ growth on top of the compact cellulose scaffold, here found ∼20 μm below the cells [Fig. 5b]. Repeated measurements at different positions confirmed a limited integration into the compact cellulose matrix. Figures 5c and 5d were measured on the surface of porous, outer side of cellulose scaffold showing cells as well as cellulose. Thus, the cells have been able to migrate into the porous cellulose matrix, again already after three days’ growth. The close-up image in Fig. 5d clearly visualizes the intimate contact established between the SMCs and individual cellulose fibers. For both types of scaffold architectures the cells are arranged in clusters with multiple layers and their growth appears to be constricted by the surrounding cellulose matrix and neighboring cells. In agreement with our observations, studies of SMCs grown on cellulose scaffolds using electron microscopy showed a cell migration of 5–10 μm for the compact material, whereas an integration of up to 40 μm could be observed for the porous type.15
CARS and SHG microscopy reveals a tight interaction between SMCs and cellulose fibers, in agreement with the good biocompatibility observed for biosynthesized cellulose scaffolds. The degree of cell integration as observed in the three-dimensional images is to a large extent dependent on the architecture of the cellulose matrix, which can be designed to achieve a distribution of cells desirable for the functionality of the scaffold. Further systematic investigations of the impact of the hierarchical structure of the cellulose material on the cell integration and proliferation are required to fully characterize this relationship. This work takes full advantage of the abilities of nonlinear microscopy to provide three-dimensional, sub-micrometer resolution images of the structural and chemical composition of soft matter, enabling unique insights in the cell — scaffold interaction mechanisms of strong interest for tissue engineering applications. Thus, nonlinear microscopy has a strong potential to become an important instrument within the field of Tissue Engineering.
The authors gratefully acknowledge Dr. Henrik Bäckdahl for production of the bacterial cellulose tubes. The financial support from the Swedish Research Council and the SSF/VR/Vinnova Biosynthetic Blood Vessels (BBV) research program is also gratefully acknowledged.