Local micromechanical properties of decellularized lung scaffolds measured with atomic force microscopy

Acta Biomaterialia 9 (2013) 6852–6859

Contents lists available at SciVerse ScienceDirect

Acta Biomaterialia journal homepage: www.elsevier.com/locate/actabiomat

Local micromechanical properties of decellularized lung scaffolds measured with atomic force microscopy T. Luque a,b,d, E. Melo a,c,d, E. Garreta a,c,d, J. Cortiella e, J. Nichols e, R. Farré a,c,d, D. Navajas a,b,d,⇑ a

Unit of Biophysics and Bioengineering, School of Medicine, University of Barcelona, Barcelona, Spain Institute for Bioengineering of Catalonia, Barcelona, Spain c Institut d’Investigacions Biomèdiques Augus Pi i Sunyer (IDIBAPS), Barcelona, Spain d Ciber Enfermedades Respiratorias (CIBERES), Bunyola, Spain e University of Texas Medical Branch, Galveston, TX, USA b

a r t i c l e

i n f o

Article history: Received 9 October 2012 Received in revised form 21 February 2013 Accepted 26 February 2013 Available online 5 March 2013 Keywords: Biological scaffolds Extracellular matrix mechanics Alveolar mechanics Atomic force microscopy Bioengineered lungs

a b s t r a c t Bioartificial lungs re-engineered from decellularized organ scaffolds are a promising alternative to lung transplantation. Critical features for improving scaffold repopulation depend on the mechanical properties of the cell microenvironment. However, the mechanics of the lung extracellular matrix (ECM) is poorly defined. The local mechanical properties of the ECM were measured in different regions of decellularized rat lung scaffolds with atomic force microscopy. Lungs excised from rats (n = 11) were decellularized with sodium dodecyl sulfate (SDS) and cut into 7 lm thick slices. The complex elastic modulus (G⁄) of lung ECM was measured over a frequency band ranging from 0.1 to 11.45 Hz. Measurements were taken in alveolar wall segments, alveolar wall junctions and pleural regions. The storage modulus (G0 , real part of G⁄) of alveolar ECM was 6 kPa, showing small changes between wall segments and junctions. Pleural regions were threefold stiffer than alveolar walls. G0 of alveolar walls and pleura increased with frequency as a weak power law with exponent 0.05. The loss modulus (G00 , imaginary part of G⁄) was 10fold lower and showed a frequency dependence similar to that of G0 at low frequencies (0.1–1 Hz), but increased more markedly at higher frequencies. Local differences in mechanical properties and topology of the parenchymal site could be relevant mechanical cues for regulating the spatial distribution, differentiation and function of lung cells. Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction Respiratory diseases are a leading cause of death worldwide. Lung transplantation is the only treatment currently available in end-stage severe respiratory diseases, including chronic obstructive pulmonary disease, alpha-1-antitrypsin deficiency, pulmonary fibrosis and pulmonary arterial hypertension [50]. However, lung transplantation achieves only 50% survival at 5 years and is hampered by a severe shortage of donor organs [30,50]. Bioartificial lungs re-engineered from decellularized lung matrix scaffolds are a promising alternative to lung transplantation. Recently, a bioreactor has been used to culture pulmonary epithelium and vascular endothelium cells seeded on acellular lung matrices [31,34]. These biological scaffolds displayed a remarkable repopulation of epithelial and endothelial compartments. Moreover, the re-engineered lungs participated in gas exchange when implanted into rats ⇑ Corresponding author. Address: Unitat de Biofísica i Bioenginyeria, Facultat de Medicina, Universitat de Barcelona, Casanova 143, 08036 Barcelona, Spain. Tel.: +34 934024515. E-mail address: [email protected] (D. Navajas).

[34,43]. Cortiella and co-workers [7] reported a first attempt to use whole acellular lung as a biologic scaffold to support the development of engineered lung tissue from murine embryonic stem cells. In constructs produced in whole acellular lungs, these authors found organization of differentiating stem cells into three-dimensional (3-D) structures reminiscent of complex tissues. These works provide proof of the concept of generation of transplantable recellularized lungs as a viable strategy for lung regeneration, either as a part or as the entire organ. However, this approach is currently limited by low repopulation efficiency. A more efficient repopulation of lung scaffolds requires better understanding of the intricate combination of biophysical and biochemical factors that modulate cell engraftment, proliferation and differentiation. There is compelling evidence that cells sense and respond to the mechanical properties of their microenvironment [9,28,32,48]. By seeding cells on a gradient of substrate elasticity, Lo and co-workers [26] showed that cells accumulate on stiffer substrates, in a process called durotaxis. Engler and co-workers [12] observed that embryo-derived cardiomyocytes maintained their spontaneous beating on substrates with elasticity less than or equal to that of

1742-7061/$ - see front matter Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.actbio.2013.02.044

T. Luque et al. / Acta Biomaterialia 9 (2013) 6852–6859

normal heart tissue, but the cells stopped beating on rigid matrices that mechanically mimic a fibrotic scar. Stiffness of the microenvironment also mediates stem cell differentiation. Engler and co-workers [11] cultured naive mesenchymal stem cells on twodimensional synthetic matrices of varying elasticity. They reported that soft matrices mimicking brain are neurogenic, stiffer matrices that mimic muscle are myogenic, and more rigid matrices that mimic collagenous bone are osteogenic. Saha and co-workers [40] found that neuron differentiation is favored on soft matrices mimicking normal brain, whereas differentiation into glia is promoted on harder matrices that typify glial scars. These findings suggest that cell engraftment, proliferation and differentiation in the lung are modulated by the mechanical properties of the cell microenvironment. A further understanding of this cell–microenvironment interplay requires a precise knowledge of the local mechanical properties of the acellular lung extracellular matrix (ECM). Mechanical properties of acellular lung tissue have been probed in air-filled decellularized murine lungs by recording quasi-static pressure–volume curves during inflation and deflation [34] and by measuring respiratory impedance with the forced oscillation technique [46]. This approach does not provide accurate estimation of lung parenchymal ECM, since mechanical measurements performed on air-filled whole lung scaffolds depend on several other factors, including the 3-D architecture of lung parenchyma, the mechanical properties of airways and blood vessels embedded in the alveolar structure, and the surface tension of the alveolar air–liquid interface. The contribution of airways, vessels and surface tension can be avoided by using liquid-filled lung parenchymal strips obtained from peripheral regions of decellularized lungs. By subjecting strips to uniaxial stretching, the elastic modulus and tensile strength of 3-D lung tissue can be measured [34], but the actual mechanical parameters of the alveolar wall cannot be determined. Atomic force microscopy (AFM) provides a direct approach to measuring the local mechanics of alveolar ECM. AFM uses a sharp tip to indent the surface of the sample with nanometer resolution, and simultaneously measure the applied force. The complex elastic modulus (G⁄) of the parenchymal lung ECM can be measured at different frequencies by vertically oscillating the AFM tip. Moreover, ECM mechanics can be locally determined at the microscale at which alveolar cells sense their mechanical microenvironment. In this work, the local mechanical properties of the ECM of different regions of decellularized scaffolds of rat lungs were measured by AFM. Lungs excised from rats were decellularized and cut into thin slices. The complex elastic modulus of the ECM was measured over a wide frequency range by applying low-amplitude indentation oscillations with the AFM tip. Measurements were taken in alveolar wall segments, alveolar wall junctions and pleural regions of the scaffold. The frequency dependence of G⁄ was interpreted in terms of a rheological model defined by a linear superposition of two power laws.

2. Materials and methods 2.1. Lung decellularization The study was carried out on lungs (n = 11) excised from young (8–9 weeks old) adult Sprague–Dawley rats (250–300 g) (Charles River, Wilmington, MA). The animal protocol was approved by the Ethical Committee for Animal Experimentation of the University of Barcelona. The lungs were decellularized following a procedure described previously [7]. The whole trachea, esophagus and lungs were excised from exsanguinated rats and washed with phosphate buffered saline (PBS). The lungs were frozen at 80 °C, thawed in a water bath at 40 °C and frozen again. The

6853

freezing–thawing cycle was carried out with the lungs uninflated (zero transpulmonary pressure) and repeated four times to enhance the formation of intracellular ice crystals that disrupt cellular membranes and cause cell lysis. The lungs were flushed with 1% sodium dodecyl sulfate (SDS) through the trachea using a syringe coupled to a catheter. Flushing was repeated three times per day to remove the disrupted cellular material with the lungs submerged in 1% SDS in continuous agitation for 5 days. Then the lungs were washed with Dulbecco’s PBS (DPBS) with antibiotics (streptomycin (90 lg ml1) and penicillin (50 U ml1)) and antimycotic (amphotericin B (25 lg ml1)) to eliminate the detergent. DPBS was flushed through the trachea three times per day with the lungs submerged in DPBS in continuous agitation for 3 days. The upper lobe of the left decellularized lung was snap frozen with liquid nitrogen in tissue freezing medium (OCT compound, Sakura, Torrance, CA) until use. For AFM measurements, 7-lm-thick lung sections were cut using a cryostat and placed on top of positively charged glass slides (Thermo Fisher Scientific, Waltham, MA). The matrix sections were rinsed, immersed in DPBS and placed on the AFM sample holder. 2.2. AFM measurements Mechanical measurements were performed on three different lung parenchyma regions: segments of alveolar walls; junctions of alveolar walls; and pleural membrane. Measurements were carried out with a custom-built AFM attached to an inverted optical microscope (TE2000, Nikon, Tokyo, Japan), using a previously described method [2,39]. Lung matrix samples were probed with a Si3N4 V-shape Au-coated cantilever with a four-sided pyramidal tip on its apex with a semi-included effective angle (h) of 20° and a nominal spring constant (k) of 0.1 N m1 (MLCT, Bruker, Mannheim, Germany). The cantilever was displaced in 3-D with nanometric resolution by means of piezoactuators coupled to strain gauge sensors (Physik Instrumente, Karlsruhe, Germany) to measure the displacement of the cantilever (z). The defection of the cantilever (d) was measured with a quadrant photodiode (S4349, Hamamatsu, Japan) using the optical lever method. The slope of a d–z curve obtained from a bare region of the coverslip was used to calibrate the relationship between the photodiode signal and cantilever deflection. The force (F) on the cantilever was computed as F = kd. To correct force measurements for the hydrodynamic drag force (Fd) on the cantilever, the cantilever was sinusoidally oscillated (16 Hz, 75 nm amplitude) at different tip– substrate distances (h) [1]. The drag factor b(h) was computed at different tip–substrate distances as b(h) = Fd/s, where s is the relative cantilever-liquid velocity (s ¼ d_  z_ , where dots indicate time derivative). Drag factor data were fitted with a scaled spherical model. At each measurement point, five force–displacement (F–z) curves were first recorded by vertically oscillating the cantilever with triangular displacement at 1 Hz and peak-to-peak amplitude of 5 lm to reach a maximum indentation of 2 lm (approaching velocity = 10 lm s1). The indentation of the sample (d) was computed as d = (z  zc)  (d  doff), where zc is the position of the contact point, and doff is the offset of the photodiode. The position of the contact point was determined from the last recorded F–z curve. Subsequently, the tip was placed at an operating indentation (d0) of 500 nm, and a small amplitude (75 nm) multifrequency oscillation [39] composed of five sine waves (0.1, 0.35, 1.15, 3.55, 11.45 Hz) was applied for 140 s. This multifrequency signal is cyclical, with a period of 20 s. The chosen frequencies were logarithmically distributed over two decades, and each component was non-sum and non-difference of the others to avoid harmonic cross-talk. All sinusoidal components were taken of equal amplitude, except the lowest frequency component, which was taken

6854

T. Luque et al. / Acta Biomaterialia 9 (2013) 6852–6859

of twofold amplitude. In addition, the relative phase of sinusoidal components was selected to minimize the peak-to-peak amplitude of the multifrequency oscillation. AFM measurements were obtained in a representative alveolar segment, alveolar junction and pleural region of 11 decellularized rat lungs (Fig. 2). In both the pleural regions and the alveolar segments, measurements were obtained at five different locations separated laterally by 3 lm. At the alveolar wall junctions, four measurements separated by 3 lm were performed around the center of the junction. 2.3. Imaging AFM images of acellular alveolar septa were obtained using a commercial device (Dimension V, Bruker, Mannheim, Germany) with the same tips as those used for mechanical measurements. Images of lung parenchymal slices adhered to positively charged glass slides were recorded in contact mode (operating force 1 nN). Height and deflection images were processed with WSxM software (Nanotec, Madrid, Spain). Scanning electron microscopy (SEM) images were obtained using a DSM 940A microscope (Zeiss, Oberkochen, Germany) with an acceleration of 15 kV. The samples were gently washed with DPBS to remove residual tissue freezing medium. The slices were dried by maintaining them in ambient air for several hours. Dehydration was completed when the samples were placed in a coating chamber, where they were coated with a layer of 14.4 nm of Au using a sputter coater (SC510, Fisons Instruments, San Carlos, CA).

cycle. The loss tangent (g) was also computed, defined as g = G00 (f)/G0 (f), which is an index of solid-like (<1) or liquid-like (>1) behavior of the sample. In each probed region, G⁄ was taken as the average of values obtained from the four or five measurement points. These average values of G⁄ computed at each location of each sample were used for the model fitting and statistical analysis.

2.5. Modeling G⁄(f) data were fitted in the complex plane (SigmaPlot, Systat Software, San Jose, CA) to a linear superposition of two power laws [8,21].

G ðf Þ ¼ Aðif Þa þ Bðif Þ3=4

ð4Þ

The fit was carried out after normalizing f to 1 Hz. This model assumes a low-frequency regime governed by a weak power law, followed by a high-frequency power law regime with an exponent of 3/4. The coefficients A and B can be interpreted as an index of matrix stiffness of the low- and high-frequency regimes, respectively. The exponent of the first power law is related to its loss tangent as g = tan(ap/2). Analogously, the loss tangent of the second power law is g = tan(3p/8), which corresponds to a predominant liquid-like behavior (g = 2.41). The crossover frequency (fc) is defined as the frequency at which G0 (f) = G00 (f) and g = 1. In each region, model parameters were computed as the average of the fittings obtained from the different measurement points.

2.4. Data processing 2.6. Statistics Force–displacement curves were analyzed with the pyramidal Hertz model [2]

F¼

3Etan h 2 d 4ð1  v 2 Þ

ð1Þ

where E is the Young’s modulus of the sample, and m is the Poisson’s ratio, assumed to be 0.5. The pyramidal Hertz model can be expressed in terms of z and d as [38]

d ¼ doff þ

3Etan h ½z  zc  ðd  doff Þ2 4ð1  v 2 Þk

ð2Þ

Statistical analysis was performed with SigmaPlot (Systat Software, San Jose, CA). Differences in G⁄ between matrix regions of the 11 lung scaffolds (three independent locations in 11 independent samples) were analyzed applying unpaired t-tests to values of G0 and G00 measured at mid frequency (1.15 Hz) of the measured range. Differences between parameters of the two power law model of the 11 lung scaffolds were compared using one-way analysis of variance and post hoc pairwise multiple comparison tests (Holm–Sidak method). A p value <0.05 was considered significant.

Non-linear least-squares fit (Matlab, The MathWorks, Natick, MA) was used to estimate E, zc and doff from the loading branch of the d–z curve for a maximum indentation of 1 lm. To assess the dependence of E on indentation, the displacement range of the fitting was progressively increased up to a maximum indentation of 2 lm. The complex shear modulus (G⁄) was computed from multifrequency oscillation data. For low-amplitude oscillations around an operating indentation d0 (computed using the average zc and doff obtained with Eq. (2) from the five d–z curves) and using G = E/ [2(1 + m)], Eq. (1) can be expressed in the frequency domain as [2]

G ðf Þ ¼

  1  v Fðf Þ  ifbðhÞ 3d0 tan h dðf Þ

ð3Þ

where f is frequency, and F(f) and d(f) are the frequency spectra of force and indentation, respectively. The term ifb(h), i being the imaginary unit, accounts for the correction of the viscous drag of the cantilever. The value of b(h) corresponding to the average thickness of the matrix sample (7 lm) was used. G⁄(f) data were separated into real and imaginary parts G⁄(f) = G0 (f) + iG00 (f), where G0 (f) is the elastic modulus that accounts for the elastic energy stored and recovered by the sample per cycle of oscillation, and G00 (f) is the loss modulus, which is a measure of the energy dissipated per

Fig. 1. Photographic image of a decellularized rat lung.

T. Luque et al. / Acta Biomaterialia 9 (2013) 6852–6859

Fig. 2. A section of decellularized lung matrix probed with an AFM tip. In each lung, scaffold measurements were taken in a representative pleural region (P), alveolar wall segment (S) and alveolar junction (J). In each region, measurements were obtained at four or five positions separated by 3 lm.

3. Results Decellularized lungs maintained their 3-D structure (Fig. 1). Fig. 2 displays a phase contrast image of a subpleural section of lung ECM probed by an AFM cantilever. The sections (7.0 ± 3.4 lm thick, mean ± SD) of decellularized lung scaffolds preserved the alveolar architecture. The pseudohexagonal alveolar network is bounded by the pleural membrane, which exhibits a similar structure, but is thicker than the alveolar walls. Attachment of the tissue slides to positively charged coverslips adequately immobilized the samples for AFM measurements. Fig. 3 shows an AFM image of decellularized lung matrix. The height image records the vertical displacement of the piezo scanner and yields a smooth topographical image of the sample. The deflection image is computed as the difference between the target and actual scanning forces, and highlights sudden changes in height or stiffness. Acellular alveolar walls have a thickness of a few microns and depict a very dense structure with a relatively smooth topography. SEM images also show a compact material, but some longitudinal structures become apparent (Fig. 4). Local mechanical properties of matrix samples were probed at the pleura and at alveolar wall segments and junctions (Fig. 2). At each measurement point, F–z curves were first recorded (Fig. 5). The force applied by the AFM tip to indent the matrix was higher during loading than during unloading, indicating that the lung matrix has viscoelastic behavior. After the recording of force–displacement curves, the samples were subjected to

Fig. 3. AFM image of an acellular alveolar wall septum. Image field is 10  10 lm: (left) height image; color bar is the scale of height over the substrate (range 0– 2.2 lm); (right) deflection image.

6855

Fig. 4. SEM image of an acellular alveolar wall septum.

low-amplitude multifrequency oscillations (Fig. 5), which contain five sinusoidal components ranging from 0.1 to 11.45 Hz. This multifrequency excitation made it possible to measure G⁄ simultaneously over a wide frequency range. Fig. 6 shows the dependence of Young’s modulus on indentation computed from force–displacement curves recorded in alveolar wall segments. At shallow indentations E displayed large variations, mainly as a result of uncertainties of the contact point. E reached a constant value for indentations deeper than 0.3 lm, reflecting fairly linear mechanical behavior. Moreover, this plateau

Fig. 5. AFM measurement of the complex shear modulus of the lung matrix. Top: example of force–displacement (F–z) curve in approaching (solid line) and withdrawing (dashed line) taken in an alveolar wall segment. The arrow is the contact point. Bottom: a cycle of the small amplitude multifrequency oscillation applied around an operating indentation of 0.5 lm.

6856

T. Luque et al. / Acta Biomaterialia 9 (2013) 6852–6859 Table 1 Parameters (mean ± SD, n = 11) of the two power law model G⁄ = A(if)a + B(if)3/4 fitted to data measured in different regions of decellularized lung matrix (fc is the crossover frequency, where G0 (f) = G00 (f), and R2 is the coefficient of determination).

A (kPa)

a B (kPa) fc (Hz) R2

Alveolar segments

Alveolar junctions

Pleura

5.59 ± 3.39* 0.050 ± 0.009 0.045 ± 0.032 354 ± 77 0.998 ± 0.002

6.79 ± 3.88* 0.048 ± 0.009 0.049 ± 0.024 370 ± 70 0.996 ± 0.011

15.76 ± 13.70 0.054 ± 0.019 0.129 ± 0.136 445 ± 374 0.996 ± 0.006

*

Significant (p < 0.05) difference compared with pleura; no significant differences were found between alveolar segments and junctions.

Fig. 6. Dependence of Young’s modulus (E) on maximum indentation. Data are mean ± SE of measurements performed in alveolar segments of 11 rats.

extended up to 2 lm indentations, which indicates that the measurements obtained with low-amplitude oscillations (75 nm) around an operating indentation of 0.5 lm were not affected by the rigid underlying substrate. The complex shear modulus of lung ECM exhibited scale-free behavior in the frequency range of our measurements (Fig. 7). Both storage and loss moduli displayed the same frequency dependence in all the explored regions of the lung matrix. The storage modulus of alveolar segments at 0.1 Hz was 5.49 ± 3.39 kPa (mean ± SD, n = 11) and increased with frequency as a weak power law (note the log–log scale in Fig. 7), reaching 7.28 ± 4.33 kPa at 11.45 Hz. The coefficients of variation (CoV = SD/mean) of G0 computed from different measurement points taken in the same region were 40%. AFM mechanical measurements are sensitive to the microscale changes in the ECM components located at the surface of the sample. Both AFM and SEM images revealed local heterogeneities of the matrix consistent with the intra-regional variability of G⁄. The loss modulus was an order of magnitude smaller that G0 and exhibited a parallel frequency dependence at low frequencies,

but increased more markedly at frequencies higher than 1 Hz. Consequently, g was 0.12 ± 0.04 (mean ± SD, n = 11) at 0.1 Hz rising to 0.22 ± 0.02 at 11.45 Hz. Both G0 and G00 of alveolar wall junctions were 20% higher than those of alveolar wall segments, but the differences computed at 1.15 Hz were not statistically significant. By contrast, the matrix in pleural regions was markedly stiffer than in alveolar walls. At 1.15 Hz, G0 and G00 of the pleura were threefold higher than those of the alveolar segments (p < 0.05). No significant differences were found in the loss tangent measured at 1.15 Hz in the different regions. The two power law model fitted all the data very well (Fig. 7 and Table 1). At physiological frequencies (0.1–1 Hz), lung scaffold rheology was dominated by a power law with an exponent of 0.05. In agreement with the model, this weak exponent was associated with a marked solid-like behavior (g  0.08). At frequencies higher than 1 Hz, the contribution of the second power law became apparent as a progressive rise in the slope of G00 . The corresponding crossover frequency was 400 Hz in all scaffold regions. Beyond this frequency, matrix behavior would be dominated by the liquid-like rheology of the second power law (g = 2.41). Consistently with the differences observed in G⁄ data, the coefficient A of the first power law was threefold higher in the pleura than in the alveolar walls (p < 0.05), but differences in stiffness between alveolar regions were not significant. No significant differences were found in the coefficient B of the second power law. 4. Discussion

Fig. 7. Complex shear modulus (G⁄) measured at different regions of the decellularized lung matrix. Alveolar wall segments (triangles up), alveolar wall junctions (triangles down) and pleural regions (circles). Close symbols are the real part (G0 , storage modulus) and open symbols are the imaginary part (G00 , loss modulus) of G⁄. Data are mean ± SE of measurements obtained in 11 rats. Solid and dashed lines are fits of the two power law model to G0 and G00 , respectively.

Using AFM, the first direct measurement of local micromechanics of the acellular lung scaffold were obtained. The complex shear modulus was measured over two frequency decades. The storage modulus of the alveolar ECM was 6 kPa, showing small changes between alveolar wall regions. In contrast, the pleural membrane was threefold stiffer than the alveolar walls. At physiological frequencies (0.1–1 Hz), lung matrix mechanics was dominated by a solid-like behavior, with frictional stresses ten times lower than elastic stresses. In the measured frequency range, the storage modulus increased with frequency as a weak power law. The loss modulus showed parallel frequency dependence at physiological frequencies, but increased more markedly at higher frequencies. The lung matrix exhibited scale-free rheology that conformed to the two power law model. Septal segments are lined mainly with type I alveolar epithelial cells that are large (40 lm in diameter) and very thin (1–2 lm in height), providing a semipermeable barrier to allow efficient gas exchange between capillary blood and alveolar gas [20]. Type II alveolar epithelial cells, which secrete lung surfactant and have the capability of differentiating into type I cells, are preferentially located at septal wall junctions [33]. The decellularization process completely eliminated lung cells maintaining alveolar architecture (Fig. 2). AFM and SEM images of alveolar walls revealed a very

T. Luque et al. / Acta Biomaterialia 9 (2013) 6852–6859

dense structure, able to preserve the semipermeable features of the native matrix (Figs. 3 and 4). It should be noted that the compact appearance of the SEM images (Fig. 4) could have been enhanced by the sample dehydration process required by this technique. However, the compact structure of the matrix was confirmed by AFM images obtained under liquid conditions (Fig. 3). Moreover, decellularized lungs could be homogeneously ventilated with no leaks (Fig. 1) demonstrating that alveolar wall integrity was majorly maintained (Fig. 2). Lung parenchymal cells interact with their mechanical environment through micron-size focal adhesions composed of transmembrane integrin receptors that link ECM with cell actin cytoskeleton [10]. Considering the reduced thickness of the alveolar walls (Figs. 3 and 4), pyramidal tips were selected to place the tip precisely in the center of the wall. The contact area (S) between the pyramidal tip and the sample increases with indentation as S = 1.58d2tan2h [2]. For the operating indentation of 0.5 lm applied in this work, the contact area corresponding to the tips is 0.05 lm2 (lateral length 0.2 lm). The tip–matrix contact lateral dimension was therefore more than one order of magnitude lower than the thickness of the wall. This high lateral resolution allowed the local heterogeneity of the matrix to be assessed by taking several measurements within the same region. The 40% intra-regional variability observed could partly reflect different ECM components or fiber distribution. It should be noted that intra-regional heterogeneity was probed on a length scale substantially smaller than the stiffness maps (80  80 lm) obtained by Liu and co-workers [25] using spherical tips in native lung parenchymal strips, which could reflect different lung structures. Moreover, the applied indentation was in the range of local deformations exerted on the microenvironment by active contraction of the cytoskeleton of alveolar cells [15]. Therefore, parenchymal ECM rheology was probed at the length scale at which cell membrane receptors sense their local mechanical environment. The plateau observed in Young’s modulus measured at increasing AFM indentations (Fig. 6) revealed fairly linear mechanics of the matrix when subjected to small local strains. This behavior contrasts with the marked non-linear stress–strain relationship reported in parenchymal strips of native and decellularized lungs subjected to large uniaxial stretching [29,35]. This suggests that lung ECM matrix and parenchymal strips were probed in different strain regimes, or that tissue strip mechanics is substantially determined by the 3-D architecture of the alveolar lattice. The stiffness found with AFM in single acellular alveolar walls of rats is higher than that reported from murine native lung parenchymal strips. Liu and co-workers [25] recently used AFM to measure the stiffness of tissue strips obtained from subpleural regions of degassed mice lungs. They found a median shear modulus of 0.5 kPa, which is one order of magnitude lower than that found in alveolar walls. By subjecting degassed lung parenchymal strips of rat to uniaxial stretch, Cavalcante and co-workers [6] measured a shear modulus of 0.17 kPa. These authors developed a computational model of the alveolar structure and concluded that the stiffness of lung parenchyma is substantially determined by the topology of the alveolar architecture, and estimated that the stiffness of alveolar walls is 10 times greater than that of tissue strips. The present findings show that measurements in lung parenchymal strips underestimate the local mechanical stiffness of ECM sensed by alveolar cells. Moreover, it should be taken into account that fresh lung tissue is populated with living cells at different levels of active contraction, which could also modify the bulk stiffness of the strip. Lung parenchymal matrix exhibits scale-free rheology, which means that there is no characteristic frequency or time scale. At physiological frequencies, the matrix mechanics is dominated by the elastic modulus, which increases slightly with frequency as a power law with an exponent a  0.05. Moreover, the loss tangent

6857

depends little on frequency, indicating a tight coupling between elastic and frictional stresses. Weak power law viscoelasticity with a constant loss tangent around physiological frequencies has been commonly observed in lung parenchymal strips, collagen gels and single collagen fibers. Lung parenchymal strips of guinea pigs subjected to uniaxial stretching exhibited a storage modulus that increased as a power law with a weak exponent (a = 0.075) over a frequency range spanning two decades around physiological frequencies (0.07–2.4 Hz) [49]. Measurements on collagen gels performed with macroscopic rheometers showed a storage modulus increasing with frequency (0.01–10 Hz) as a power law with an exponent of 0.1 and a loss tangent of 0.1 [22,36]. Viscoelastic behavior was also found in single collagen fibers, as revealed by stress relaxation and stress–strain hysteresis [19,41]. This suggests that the frequency dependence of G⁄ of alveolar and pleural matrix found around physiological frequencies reflects the viscoelastic features of ECM fibers. Alveolar walls are substantially stiffer than ECM matrix produced by cells in culture and by reconstituted ECM gels. A recent study in decellularized cell-derived matrix from human pulmonary fibroblasts reported a stiffness one order of magnitude lower than that found in alveolar walls [44]. Reconstituted collagen gels have been used extensively to mimic ECM structure and mechanics. The stiffness of collagen gels depends on collagen concentration and on a number of details in the preparation protocol. However, the stiffness of reconstituted collagen gels [3,22,23,36,37] is one order of magnitude lower than that found in the acellular lung parenchyma. Moreover, in contrast to the compact structure of alveolar walls (Figs. 3 and 4), collagen gels show a porous mesh consisting of sparsely entangled fibers [36,37]. These differences indicate that elastin and proteoglycans and their molecular interactions are also critical determinants of the structure and stiffness of alveolar ECM [6]. Unlike the weak power law shown by the storage modulus of the parenchymal tissue in the explored frequency band, the loss modulus exhibits a faster rise for frequencies higher than 1 Hz (Fig. 7), which a single power law cannot fit. Two frequency regimes have been reported previously in a large number of cell types. High-frequency cell rheology has been described by adding a second power law with an exponent of 3/4 [8,21] that is characteristic of the entropic dynamics of semiflexible polymers [16–18,27]. An excellent fit was found for all the present data with this two power law model with a crossover frequency of 400 Hz (Table 1). Interestingly, the transition between slow and fast regimes occurs at a frequency similar to that of cells [8]. It should be noted, however, that the present authors could not accurately estimate the exponent of the second power law, since the frequency range of the measurements is far below the crossover frequency. In fact, an excellent fit was also found with a high-frequency power law of exponent 1, which corresponds to Newtonian viscosity (data not shown). Further studies are needed to characterize in more detail the microrheological behavior of the lung matrix at high frequencies. At physiological frequencies, lung matrix rheology is dominated by a weak power law with an exponent in the range of that reported in living cells [8], reconstituted acting networks [14,24] and collagen gels [36,47]. This slow dynamics, characteristic of soft glassy materials [13], may be governed by viscoelastic properties of entangled or crosslinked networks of collagen and elastin fibers and other ECM biopolymers. Small differences were found in alveolar wall stiffness between segments and junctions. This suggests that other additional factors regulate the preferential location of type II cells in alveolar corners. In contrast to the relatively flat surface of septal segments, alveolar corners provide a more curved site and concentrate mechanical stresses transmitted through the lung parenchyma, which could

6858

T. Luque et al. / Acta Biomaterialia 9 (2013) 6852–6859

result in local variations of the 3-D strain field of the alveoli. Moreover, parenchymal deformation is not affine, and whole lung expansions and compressions could change the angle between two adjacent alveolar wall segments [5,6]. Therefore, cells attached to wall corners may be subjected to bending components of strain that contrast with the more flat deformations of those cells located in septal wall segments. Moreover, the stiffness of alveolar segments is not uniform, with a CoV of 60% (Table 1). Although this CoV could in part reflect the variability of the lung scaffolds or the decellularization process used, septal stiffness heterogeneity together with nonhomogeneous wall thickness and differential tethering and other forces may account for the non-uniform distension of alveolar perimeter segments reported previously [33]. However, pleural matrix was markedly stiffer than alveolar walls and lacked the polygonal structure of internal alveoli. Therefore, local changes in stiffness, thickness, strain and topology of the lung scaffold potentially provide differential mechanical signals sensed by alveolar cells to regulate their spatial distribution, differentiation and function. The mechanical interplay between alveolar cells and ECM depends on their relative mechanical properties. AFM has been used to measure the complex shear modulus of BEAS-2B bronchial and A549 alveolar epithelial type II cells over three frequency decades (0.1–100 Hz) [2]. These cells showed a storage modulus of 0.5 kPa at the lowest frequency and increased with frequency as a power law with an exponent of 0.2. At low frequencies, the loss tangent was 0.3, and the loss modulus increased as a power law with the same exponent as G0 . Moreover, G00 exhibited a progressively steeper rise for frequencies higher than 1 Hz, showing a crossover frequency of 100 Hz. Weak power law rheology is also exhibited by primary airway smooth muscle cells probed with AFM [42] and magnetic twisting cytometry [8]. Lung cells therefore have a solid-like behavior with a weak power law rheology close to that of parenchymal ECM. This similarity suggests a viscoelastic matching between the cell and its microenvironment, which ensures synchronous deformation during breathing [45]. Azeloglu and coworkers [4] measured Young’s modulus with AFM of alveolar epithelial type I and type II cells and of fibroblasts isolated from rat lungs. In the nuclear region, all three cells types showed a similar Young’s modulus of 3 kPa which roughly correspond to G0  1 kPa. Taken together, these data suggest that epithelial and interstitial alveolar cells are 5-fold and 15-fold softer than the alveolar walls and the pleural membrane, respectively. Moreover, the stiffness of the lung parenchymal scaffold is within the range that directs cell differentiation [12], suggesting that lung cells are capable of sensing the mechanical signatures of different lung sites. Two-photon imaging of decellularized rat lungs revealed more fibrilar collagen content in the pleural membrane than in inner alveoli [7]. This is consistent with the greater stiffness found in the pleural membrane, which might provide site-specific clues for the differentiation of mesothelial cells. It should be noted that Engler and co-workers [11] reported expression of myogenic markers in mesenchymal stem cells seeded on polyacrylamide gels with a stiffness (E = 8–17 kPa; G0  3–6 kPa) comparable to that of the alveolar scaffold. In contrast, embryonic stem cells seeded in the alveolar regions of decellularized lungs expressed markers of endothelial and type II alveolar epithelial cells [7]. This suggests that other niche-specific physical cues, apart from substrate stiffness, are involved in guiding stem cell differentiation to the different lung lineages. It is concluded that AFM allows for direct measurement of the local microrheological properties of different regions of acellular lung scaffolds over a wide frequency range. At physiological frequencies, alveolar walls have solid-like behavior, with a storage modulus of 6 kPa and a loss modulus one order of magnitude lower. Alveolar wall segments and junctions have similar stiffness,

whereas pleural membrane is three times stiffer. Stiffness of alveolar walls is five times greater than that reported in alveolar cells. Rheology of the acellular lung parenchyma is very well characterized by a two power law model composed of a weak power law with an exponent 0.05 that accounts for the solid-like regime dominant at physiological frequencies, and a second power law with an exponent of 3/4 that accounts for a liquid-like regime at high frequencies. It is suggested that local differences in mechanical properties and topology of the parenchymal site could constitute relevant mechanical cues to regulate spatial distribution, differentiation and function of lung cells. Acknowledgements The authors wish to thank Rocio Nieto and Miguel A. Rodríguez for their valuable technical help. SEM images were obtained at the Electron Microscopy Unit of the School of Medicine of the University of Barcelona. This research was supported in part by the Spanish Ministry of Economy and Competitiveness (FIS-PI11/00089 and SAF2011-22576). Appendix Appendix:. Figures with essential colour discrimination Certain figures in this article, particularly Figs. 1 and 3 are difficult to interpret in black and white. The full colour images can be found in the on-line version, at http://dx.doi.org/10.1016/j.actbio. 2013.02.044. References [1] Alcaraz J, Buscemi L, Puig-de-Morales M, Colchero J, Baró A, Navajas D. Correction of microrheological measurements of soft samples with atomic force microscopy for the hydrodynamic drag on the cantilever. Langmuir 2002;18:716–21. [2] Alcaraz J, Buscemi L, Grabulosa M, Trepat X, Fabry B, Farré R, et al. Microrheology of human lung epithelial cells measured by atomic force microscopy. Biophys J 2003;84:2071–9. [3] Arevalo RC, Urbach JS, Blair DL. Size-dependent rheology of type-I collagen networks. Biophys J 2010;99:L65–7. [4] Azeloglu EU, Bhattacharya J, Costa KD. Atomic force microscope elastography reveals phenotypic differences in alveolar cell stiffness. J Appl Physiol 2008;105:652–61. [5] Brewer KK, Sakai H, Alencar AM, Majumdar A, Arold SP, Lutchen KR, et al. Lung and alveolar wall elastic and hysteretic behavior in rats: effects of in vivo elastase treatment. J Appl Physiol 2003;95:1926–36. [6] Cavalcante FS, Ito S, Brewer K, Sakai H, Alencar AM, Almeida MP, et al. Mechanical interactions between collagen and proteoglycans: implications for the stability of lung tissue. J Appl Physiol 2005;98:672–9. [7] Cortiella J, Niles J, Cantu A, Brettler A, Pham A, Vargas G, et al. Influence of acellular natural lung matrix on murine embryonic stem cell differentiation and tissue formation. Tissue Eng Part A 2010;16:2565–80. [8] Deng L, Trepat X, Butler JP, Millet E, Morgan KG, Weitz DA, et al. Fast and slow dynamics of the cytoskeleton. Nat Mater 2006;5:636–40. [9] Discher DE, Janmey P, Wang YL. Tissue cells feel and respond to the stiffness of their substrate. Science 2005;310:1139–43. [10] DuFort CC, Paszek MJ, Weaver VM. Balancing forces: architectural control of mechanotransduction. Nat Rev Mol Cell Biol 2011;12:308–19. [11] Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem cell lineage specification. Cell 2006;126:677–89. [12] Engler AJ, Carag-Krieger C, Johnson CP, Raab M, Tang HY, Speicher DW, et al. Embryonic cardiomyocytes beat best on a matrix with heart-like elasticity: scar-like rigidity inhibits beating. J Cell Sci 2008;121:3794–802. [13] Fabry B, Maksym GN, Butler JP, Glogauer M, Navajas D, Fredberg JJ. Scaling the microrheology of living cells. Phys Rev Lett 2011;87:148102. [14] Gardel ML, Shin JH, MacKintosh FC, Mahadevan L, Matsudaira P, Weitz DA. Elastic behavior of cross-linked and bundled actin networks. Science 2004;304:1301–5. [15] Gavara N, Sunyer R, Roca-Cusachs P, Farré R, Rotger M, Navajas D. Thrombininduced contraction in alveolar epithelial cells probed by traction microscopy. J Appl Physiol 2006;101:512–20. [16] Gisler T, Weitz DA. Scaling of the microrheology of semidilute F-actin solutions. Phys Rev Lett 1999;82:1606–9. [17] Gittes F, Schnurr B, Olmsted PD, MacKintosh FC, Schmidt CF. Microscopic viscoelasticity: shear moduli of soft materials determined from thermal fluctuations. Phys Rev Lett 1997;79:3286–9.

T. Luque et al. / Acta Biomaterialia 9 (2013) 6852–6859 [18] Gittes F, MacKintosh FC. Dynamic shear modulus of a semiflexible polymer network. Phys Rev E 1998;58:R1241–4. [19] Grant CA, Phillips MA, Thomson NH. Dynamic mechanical analysis of collagen fibrils at the nanoscale. J Mech Behav Biomed Mater 2012;5:165–70. [20] Harris JB, Chang L-Y, Crapo JD. Rat lung alveolar type I epithelial cell injury and response to hyperoxia. Am J Respir Cell Mol Biol 1991;4:115–25. [21] Hoffman BD, Massiera G, Van Citters KM, Crocker JC. The consensus mechanics of cultured mammalian cells. Proc Natl Acad Sci 2006;103:10259–64. [22] Koch TM, Münster S, Bonakdar N, Butler JP, Fabry B. 3D traction forces in cancer cell invasion. PLoS One 2012;7:e33476. [23] Latinovic O, Hough LA, Daniel Ou-Yang H. Structural and micromechanical characterization of type I collagen gels. J Biomech 2010;43:500–5. [24] Lieleg O, Kayser J, Brambilla G, Cipelletti L, Bausch AR. Slow dynamics and internal stress relaxation in bundled cytoskeletal networks. Nat Mater 2011;10:236–42. [25] Liu F, Mih JD, Shea BS, Kho AT, Sharif AS, Tager AM, et al. Feedback amplification of fibrosis through matrix stiffening and COX-2 suppression. J Cell Biol 2010;190:693–706. [26] Lo CM, Wang HB, Dembo M, Wang YL. Cell movement is guided by the rigidity of the substrate. Biophys J 2000;79:144–52. [27] Morse DC. Viscoelasticity of tightly entangled solutions of semiflexible polymers. Phys Rev E 1998;58:R1237–40. [28] Murrell M, Kamm R, Matsudaira P. Substrate viscosity enhances correlation in epithelial sheet movement. Biophys J 2011;101:297–306. [29] Navajas D, Maksym GN, Bates JHT. Dynamic viscoelastic nonlinearity of lung parenchymal tissue. J Appl Physiol 1995;79:348–56. [30] Orens JB, Garrity Jr ER. General overview of lung transplantation and review of organ allocation. Proc Am Thorac Soc 2009;6:13–9. [31] Ott HC, Clippinger B, Conrad C, Schuetz C, Pomerantseva I, Ikonomou L, et al. Regeneration and orthotopic transplantation of a bioartificial lung. Nat Med 2010;16:927–33. [32] Paszek MJ, Zahir N, Johnson KR, Lakins JN, Rozenberg GI, Gefen A, et al. Tensional homeostasis and the malignant phenotype. Cancer Cell 2005;8:241–54. [33] Perlman CE, Bhattacharya J. Alveolar expansion imaged by optical sectioning microscopy. J Appl Physiol 2007;103:1037–44. [34] Petersen TH, Calle EA, Zhao L, Lee EJ, Gui L, Raredon MB, et al. Tissueengineered lungs for in vivo implantation. Science 2010;329:538–41. [35] Petersen TH, Calle EA, Colehour MB, Niklason LE. Matrix composition and mechanics of decellularized lung scaffolds. Cells Tissues Organs 2012;195:222–31.

6859

[36] Piechocka IK, van Oosten AS, Breuls RG, Koenderink GH. Rheology of heterotypic collagen networks. Biomacromolecules 2011;12:2797–805. [37] Raub CB, Suresh V, Krasieva T, Lyubovitsky J, Mih JD, Putnam AJ, et al. Noninvasive assessment of collagen gel microstructure and mechanics using multiphoton microscopy. Biophys J 2007;92:2212–22. [38] Rico F, Roca-Cusachs P, Gavara N, Farré M, Rotger M, Navajas D. Probing mechanical properties of living cells by atomic force microscopy with blunted pyramidal cantilever tips. Phys Rev E 2005;72:021914. [39] Roca-Cusachs P, Almendros I, Sunyer R, Gavara N, Farré R, Navajas D. Rheology of passive and adhesion-activated neutrophils probed by atomic force microscopy. Biophys J 2006;91:3508–18. [40] Saha K, Keung AJ, Irwin EF, Li Y, Little L, Schaffer DV, et al. Substrate modulus directs neural stem cell behavior. Biophys J 2008;95:4426–38. [41] Shen ZL, Kahn H, Ballarini R, Eppell SJ. Viscoelastic properties of isolated collagen fibrils. Biophys J 2011;100:3008–15. [42] Smith BA, Tolloczho B, Martin JG, Grütter P. Probing the viscoelastic behavior of cultured airway smooth muscle cells with atomic force microscopy: stiffening induced by contractile agonist. Biophys J 2005;88:2994–3007. [43] Song JJ, Kim SS, Liu Z, Madsen JC, Mathisen DJ, Vacanti JP, et al. Enhanced in vivo function of bioartificial lungs in rats. Ann Thorac Surg 2011;92:998–1006. [44] Soucy PA, Werbin J, Heinz W, Hoh JH, Romer LH. Microelastic properties of lung cell-derived extracellular matrix. Acta Biomater 2011;7:96–105. [45] Suki B, Stamenovic D, Hubmayr R. Lung parenchymal mechanics. Compr Physiol 2011;1:1317–51. [46] Wallis JM, Borg ZD, Daly AB, Deng B, Ballif BA, Allen GB, et al. Comparative assessment of detergent-based protocols for mouse lung de-cellularization and re-cellularization. Tissue Eng Part C Methods 2012;18:420–32. [47] Yang YL, Leone LM, Kaufman LJ. Elastic moduli of collagen gels can be predicted from two-dimensional confocal microscopy. Biophys J 2009;97:2051–60. [48] Yeung T, Georges PC, Flanagan LA, Marg B, Ortiz M, Funaki M, et al. Effects of substrate stiffness on cell morphology, cytoskeletal structure, and adhesion. Cell Motil Cytoskeleton 2005;60:24–34. [49] Yuan H, Ingenito EP, Suki B. Dynamic properties of lung parenchyma: mechanical contributions of fiber network and interstitial cells. J Appl Physiol 1997;83:1420–31. [50] Yusen RD, Shearon TH, Qian Y, Kotloff R, Barr ML, Sweet S, et al. Lung transplantation in the United States, 1999–2008. Am J Transplant 2010;10:1047–68.