Abstract
Animal Cells and Systems, 2013 Vol. 17, No. 3, 170178, http://dx.doi.org/10.1080/19768354.2013.789452 a a a a,b Yeongjin Kim , Jung Woo Hong , Jung Kim and Jennifer H. Shin * School of Mechanical, Aerospace & Systems Engineering, Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, Room 2214. ME bld. KAIST, 373-1 Guseong-dong, Daejeon 305-701, Republic of Korea; Department of Bio and Brain Engineering, Korea Advanced Institute of Science and Technology, Room 2214. ME bld. KAIST, 373-1 Guseong-dong, Daejeon 305-701, Republic of Korea (Received 12 November 2012; received in revised form 12 March 2013; accepted 21 March 2013) Although cancerous cells and normal cells are known to have different elasticity values, there have been inconsistent reports in terms of the actual and relative values for these two cell types depending on the experimental conditions. This paper investigated the mechanical characterization of normal hepatocytes (THLE-2) and hepatocellular carcinoma cells (HepG2) using atomic force microscopy indentation experiments and the HertzSneddon model, and the results were confirmed by an independent de-adhesion assay. To improve the reliability of the data, we considered the effects of tip geometry and indentation depth on the measured elasticity of the cells. This study demonstrated that THLE-2 cells had a higher elastic modulus compared with the HepG2 cells and that this difference was more significant when a conical tip was used. The inhibitor study indicated that this difference in the mechanical properties of THLE-2 and HepG2 cells was mainly attributed to differential arrangements in the cytoskeletal networks of actin filaments. An independent de-adhesion assay also confirmed that THLE-2 cells had a higher elastic modulus compared with the HepG2 cells, which resulted in a shorter time constant for cellular contractility. Keywords: HertzSneddon model; atomic force microscopy; cellular elasticity; de-adhesion assay; hepatocytes 1. Introduction A wide range of different measurement techniques have been utilized to measure the mechanical properties Cellular mechanical properties and the associated of cells; these include micropipette aspiration, magnetic forces involved in cytoskeletal structures can reflect twisting cytometry, and optical tweezers (Table 1). the specific pathological conditions of a cell. In recent Microscale or nanoscale indentation is another technique years, there have been a considerable number of reports that can depict a mechanical response or property at the on the relationship between the biomechanical proper- single-cell level. Atomic force microscopy (AFM) inden- ties of cells and diseases, such as cancer (Guck et al. tation, for example, has been widely used due to its high 2005; Cross et al. 2007; Suresh 2007a, 2007b), and these resolution and sensitivity as well as its deeper probing changes in the cellular biomechanical properties have range (Shroff et al. 1995; Goldmann & Ezzell 1996; Cross recently emerged as an indication of disease (Plewes et al. 2007). The HertzSneddon (HS) model has been et al. 2000; Cochlin et al. 2002; McKnight et al. 2002; most widely and conveniently used to estimate the elastic Bercoff et al. 2003; Salomon et al. 2008). Therefore, the moduli of cells for the forcedisplacement curves ob- mechanical characterization of cells has the potential tained from AFM measurements (Cross et al. 2007). to be used as a quantitative diagnostic tool. Additional However, the elastic modulus estimated using the HS recent reports have shown that the utilization of the model depends on the tip geometry and indentation physical characterization of cancer versus normal cells depth (Mathur et al. 2001; Costa 2003; Kamgoue´et al. adds another dimension to the analysis of these cells 2007; Unnikrishnan et al. 2007). To improve the diag- for improved diagnosis of cancer (Cross et al. 2007). nostic significance of the measured elasticity, tip geome- Because classical diagnostic methods using cytomor- phology and immunohistochemistry are associated try and indentation depths need to be optimized so that the measured values are reproducible and reliable. There- with difficulties due to morphological uncertainties fore, in this study, we characterized elastic modulus between normal and cancerous cells as well as the extreme biochemical diversity of diseased cells, effective distributions with different indentation depths using use of multiple techniques for diagnostic confirmation two types of tips with different geometries, namely a is desired, given that cancerous cells and normal cells cone and a sphere, to suggest the appropriate tip display common mechanical properties (Cross et al. geometry and indentation depth. Two liver cell types, 2007). normal hepatocytes (THLE-2) and hepatocellular *Corresponding author. Email: j_shin@kaist.ac.kr Current address: Jennifer H. Shin, Graduate School of Medical Science and Engineering, Korea Advanced Institute of Science and Technology, Room 2214. ME bld. KAIST, 373-1 Guseong-dong, Daejeon 305701, Republic of Korea # 2013 Korean Society for Integrative Biology TRANSLATIONAL MEDICINE Animal Cells and Systems 171 Table 1. Experimental techniques and analysis methods for cellular biomechanics. Authors Experimental sample Experimental method Analysis method Shao (2002) Human neutrophils Extended micropipette aspiration Finite element model Jones et al. (1999) Chondrocytes isolated from normal Micropipette aspiration Elastic finite element and osteoarthritic human cartilage model Alexopoulos et al. (2003) Human articular cartilage Micropipette aspiration Elastic model Mathur et al. (2001) Endothelial, cardiac muscle and Atomic force microscopy HertzSneddon skeletal muscle cells model Goldmann and Ezzell Mouse F9 embryonic carcinoma cells Atomic force microscopy HertzSneddon (1996) model Shroff et al. (1995) Cultured rat atrial myocytes Atomic force microscopy Elastic model Cross et al. (2007) Live metastatic cancer cells taken Atomic force microscopy HertzSneddon from the body (pleural) fluids of model patients with suspected lung, breast, and pancreas cancer Kim et al. (2011) Hepatocellular carcinoma cells Atomic force microscopy HertzSneddon model and finite element model Alcaraz et al. (2003) Human alveolar (A549) and Atomic force microscopy Elastic model bronchial (BEAS-2B) epithelial cell lines Guo et al. (2012) Human aortic endothelial cells Atomic force microscopy Hertz model and (HAECs) Sneddon model Chen et al. (2001) HUVEC cells membrane Magnetic twisting cytometry Elastic model Ziemann et al. (1994) G-actin Magnetic twisting cytometry Viscoelastic model Bausch et al. (1999) J774 mouse macrophages Magnetic twisting cytometry Kelvin viscoelastic model He ´non et al. (1999) Human red blood cell membrane Optical tweezer Elastic model (shear modulus estimation) Dao et al. (2003) Human red blood cells Optical tweezer Finite element model carcinoma cells (HepG2), were selected for this study. The HepG2 cells were cultured in a humidified atmosphere elastic moduli of both cells were approximated using the of 5% CO at 378C in Bronchial Epithelial Cell Growth well-known HS linear model. The proper tip geometry Medium (Lonza, CC-3170) and Dulbecco’s Modified and indentation depth were investigated using conical Eagle’s Medium (Lonza, 12-604F), respectively and and spherical AFM tips, and a point-wise HS analysis supplemented with 10% fetal bovine serum (16000-044, was used to improve the diagnostic significance. In GIBCO, USA) and 1% penicillin/streptomycin (15140- addition, the contribution of each cytoskeletal protein 122, GIBCO, USA). The subculture rate was between to the measured mechanical property differences between 1:4 and 1:6.The following three groups of samples were normal and cancerous cells was determined by disrupting tested: untreated control cells, actin-disrupted cells, and each protein network using the appropriate chemical microtubule (MT)-disrupted cells. Actin filaments were inhibitors. These results indicated that disparities in disrupted by treatment with 20 mM cytochalasin D elastic moduli between normal and cancerous cells may (CD; Sigma Chemical, USA), and MTs were perturbed reflect inherent differences in the actin cytoskeletal with 200 mM colchicine (Col; Sigma Chemical, USA). structure and organization. In parallel, an independent Cells were incubated in CD and Col for 30 minutes at de-adhesion assay was performed to confirm that normal 378C. THLE-2 cells have a greater elasticity than do cancerous HepG2 cells, as indicated by the shorter time constant for contraction upon trypsin treatment. 2.2. AFM indentation experiments The Nanowizard II Bioscope (JPK Instruments, Ger- TM many) with the Biocell system (JPK Instruments, 2. Materials and methods Germany), which is a liquid holder for biological 2.1. Sample preparation applications, was used to probe adherent cells in the culture media. The AFM was coupled to an optical THLE-2 (ATCC CRL-2706) and HepG2 (ATCC HB- 8065) were purchased from ATCC. THLE-2 and microscope (Zeiss Axiovert 200, Germany) to acquire 172 Y. Kim et al. optical images and determine surface topography. This (a) combination allows users to position the AFM tip over the region of interest on the cell (Radmacher 1997). To 2R δ z investigate the effects of tip geometries, the following two Conical tip Spherical tip different tip geometries were tested following the calibra- tion of the cantilever’s spring constants using the thermal TM (b) 100 THLE-2 (cone) THLE-2 (sphere) noise method: a conical tip end cantilever (Nanosensor , HepG2 (cone) HepG2 (sphere) Switzerland) with a 0.10 N/m spring constant and a tip angle of 22.58and a 1-mm-sphere-particle-attached can- tilever (Novascan Technologies, Inc., USA) with a 0.12 N/m spring constant. In spectroscopy mode, the interac- tion forces between the tip and the cells were monitored as the deflection of the AFM cantilever and were recorded as a forcedisplacement curve. 0.1 00.511.5 Indenation depth (µm) 2.3. Immunostaining for fluorescence microscopy (c) To analyze the structural changes of both cell types, intracellular structure information was obtained from confocal fluorescence imaging. Cells were fixed in 3.7% formaldehyde for 10 minutes at 378C followed by permeabilization for 15 minutes in 0.2% Triton X- 100. After rinsing with phosphate-buffered saline 5 (PBS), cells were blocked in 3% bovine serum albumin for 60 minutes and probed with an anti-alpha tubulin Cone, Cone, Sphere, Sphere, antibody (1:200; A1112b, Molecular Probes, USA) at p < 0.0001 THLE-2 HepG2 THLE-2 HepG2 48C to visualize the MT network. Cells were then (d) labeled with a secondary antibody conjugated to Alexa Cancer Gray zone Normal Fluor 488 (1:200; A11001, Molecular Probes, USA) for 30 minutes at room temperature, followed by a brief rinse with PBS. The cells were then stained for actin with Alexa Fluor-568 phalloidin (1:50; A12380, Mole- cular Probes, USA) for 30 minutes. Finally, the cell nuclei were labeled with DAPI (300 nM; D1306, Molecular Probes, USA) for 3 minutes, and the cells were then mounted in Vectashield (H-1000, Vector 0 5 10 15 20 25 30 Laboratories, USA). Cells were imaged with a fluores- Elastic modulus (kPa) cence microscope (Axiovert 200M, Zeiss, Germany). Figure 1. (a) Schematic drawings of the HertzSneddon model for conical and spherical tip indentations. The v value 2.4. Hertz Sneddon model represents Poisson’s ratio, and the a, F, and R values The HS model is a linear contact model based on represent the load, radius, and angle of the spherical tip, simplified assumptions of a semi-infinite, homoge- respectively. The dz value represents the indentation depth. neous, and isotropic material. This model has been (b) Elastic modulus variations as a function of indentation depth for THLE-2 and HepG2 cells using conical and the most widely accepted model for elastic modulus spherical tips. (c) Elastic moduli measured with conical and estimations of engineering materials, and its applica- spherical tip indentations for HepG2 and THLE-2 cells. tions have been expanded to include biological tissues THLE-2 cells had a higher elastic modulus than HepG2 and cells (Ottensmeyer 2001; Cross et al. 2007). The cells, and this difference was more significant when a conical elastic modulus equations from this model for conical tip was used. (d) As shown in the histograms of Young’s and spherical tip geometries are defined as follo qffiffiffiffiffiffiffi ws:ffi modulus for a quantitative comparison between HepG2 and 2 3 2 2 E ¼ pFð1 v Þ=2d tan a and E ¼ 3ð1 v ÞF=4 Rd , z z THLE-2 cells, the THLE-2 and HepG2 cells exhibited respectively, where n represents Poisson’s ratio; a, F, R, significantly different trends, featuring a narrow peak for and d represent the angle, load, and radius of the HEP-G2 cells and a broad peak for THLE-2 cells with spherical tip, respectively; and d represents the in- z distinguishable peak values despite a gray zone with over- lapping elasticity values. dentation depth as shown in Figure 1a. The cell was Counts Elastic modulus (KPa) Elastic modulus (kPa) Animal Cells and Systems 173 assumed to be incompressible (Charras et al. 2001). To attributed to a low signal-to-noise ratio due to thermal extract the elastic moduli (E and F) as a function (d ) fluctuations near the plasma membrane (Mathur et al. from the indentation curve, the data were fitted to the 2001). Beyond this point, the elastic modulus stabi- aforementioned HS equations for the given tip geome- lized, and statistically distinctive values for the normal tries. Due to the difficulties in accurately identifying and cancerous cells were obtained. With a spherical tip, the contact point, however, we adopted the method a high modulus region near the plasma membrane suggested by Radmacher (1997), in which they used existed, and the modulus increased beyond 500 nm as two discrete post-contact data points to solve for the indentation depth increased, which may have Young’s modulus rather than fitting the entire post- stemmed from either the nucleus, which has a stiffer contact data set (Radmacher 1997). In this study, we modulus than the rest of the cytoplasm (Dong et al. estimated the elastic modulus of the cells from one data 1991; Maniotis et al. 1997; Caille et al. 2002; Kim et al. value near the contact point and the other at a 1 mm 2011), or the rigid glass substrate. Because the spherical indentation depth point. tip was bulkier, the substrate effect was more severe for the spherical tip than the conical tip. Moreover, the spherical tip failed to yield a significant difference between the two cell types, contrary to the results 2.5. De-adhesion assay obtained using a conical tip. Based on these results, we To confirm the different elastic moduli of THLE-2 and decided to use a conical tip with an optimal indentation HepG2 cells, a de-adhesion assay, which is a powerful depth of 1 mm, where the effects of thermal fluctuations tool to probe the contractility of adherent cells, was on the cell surface and underlying substrate were performed according to the protocol suggested by Sen minimal and there was a maximal separation between and Kumar (2009). Prior to trypsinization, the media the two cell types. For a comparative study, we plotted was removed, and the cells were briefly washed with the average values at a 1 mm depth for both cell types PBS. Cells were then treated with a warm 0.5% trypsin/ using two types of probes, as shown in Figure 1c. For a EDTA solution (15400, Gibco, USA) for detachment. conical tip, the cellular elastic modulus in THLE2 and Images were acquired with an inverted microscope carcinoma cells exhibited an average of 11.7391.73 (Axiovert 200M, Carl Zeiss) that was equipped with an kPa (n 21) and 3.4890.56 kPa (n 32), respectively incubator chamber for controlled temperature, humid- (Figure 2a), resulting in significantly different averages ity, and CO . Images were taken at 10 second intervals between the two cell types. For the spherical tip until the cells became rounded and eventually detached indentation, the average elastic modulus in the with no further apparent change in the spread area. THLE-2 and HepG2 cells was 0.3890.35 (n 14) The de-adhesion dynamics were quantified by deter- and 0.2690.26 kPa (n 11), respectively, with a P- mining the cell spreading area at different time points value of 0.3694 (Figure 1c). To investigate the feasi- using ImageJ (NIH). The time-dependent normalized bility of these data for diagnostic use, the occurrence area was acquired by dividing the difference between frequencies were counted to generate a quantitative the cell area at time t and the initial spread area [A initial comparison between the two cell types. Although a A(t)] by the difference in area between the first and last gray zone existed with overlapping elasticity values time points (A A ). The normalized area versus initial final between the THLE-2 and HepG2 cells, the THLE-2 time data were then fit to a Boltzmann sigmoidal curve and HepG2 cells exhibited significantly different to yield two characteristic time constants as follows: trends, consisting of a narrow peak for the HepG2 A ¼ 1 ; where t is inversely propor- ðts Þ=s 2 normalized 1þe 1 2 cells and a broad peak for the THLE-2 cells with tional to the cellular elasticity (Sen & Kumar 2009). distinguishable peak values (Figure 1d). The broader distribution may have occurred because THLE-2 cells vary greatly in size. Moreover, the distribution is 3. Results expected to narrow with an increasing number of 3.1. Cellular properties according to the Hertz Sneddon samples. Nonetheless, these results demonstrated the model feasibility of similar data to be used for the purpose of The two tip geometries, namely conical and spherical, increasing the accuracy of cancer diagnosis. Because were compared using the HS model to estimate the the tip geometry and indentation depth dependence on cellular elasticity of THLE-2 and HepG2 cell types. elasticity could lead to unavoidable discrepancies in Figure 1b shows the elastic modulus variations as a terms of the cell properties or incorrect diagnostic function of indentation depth for both THLE-2 and evaluations, especially for spherical tips (Mathur et al. HepG2 cells using the two types of tips. With a conical 2001; Costa 2003; Kamgoue ´ et al. 2007; Unnikrishnan tip, an unusually high modulus was observed at et al. 2007), conical tips are considered to be more indentation depths of less than 0.2 mm, which may be suitable. 174 Y. Kim et al. (a) Microtubule Actin Nucleus Microtubule Microtubule Actin Actin (b) Normal Hepatocytes (THLE-2) Liver Heptocellular Carcinoma (HepG2) Nucleus Nucleus Control Cytochalasin D (CD) Colchicine (COL) Control Cytochalasin D (CD) Colchicine (COL) (c) THLE-2 HepG2 THLE-2 HepG2 THLE-2 HepG2 Control CD Col p < 0.001 Figure 2. Chemical perturbation of cytoskeletal structures and elastic moduli measurements. (a) Inherent differences in the cytoskeletal structures between THLE-2 and HepG2 cells. (Scale bar represents 70 mm). THLE-2 cells expressed more prominent actin stress fibers (indicated with arrows), but the structural organization of MTs between cell types was not different. To test whether these differences in actin structures between the two cell types were responsible for the differences in cellular elastic moduli, CD and Col were added to the cells to disrupt each of these cytoskeletal proteins, and the elastic modulus in each of these cases was measured. (b) Fluorescent images of MTs and actin in THLE-2 cells and HepG2 cells treated with the indicated chemicals. Significant depolymerization and aggregation of actin filaments in CD-treated cells (20 mM) and aggregation of MTs in colchicine-treated cells (1 mM) were observed. (Scale bar represents 70 mm). Blown-up images shown in the insets show clear differences between intact and disrupted cytoskeletal filaments. (c) The elastic modulus of HepG2 and THLE-2 cells under control, MT-disruptive (Col), and actin-disruptive (CD) conditions. For all conditions, THLE-2 cells had higher elastic moduli than did cancerous cells. The elastic modulus differences resulting from the actin-disruptive condition were statistically insignificant with a P-value of 0.0141. This result implied that the key contributor for a higher modulus in normal cells is the inherently stronger expression and structural organization of actin filaments in normal cells. 3.2. Cytoskeletal analysis under disruptive conditions the high elasticity of THLE-2 cells. Because cytoskele- tal proteins influence cellular mechanical properties Given that normal THLE-2 cells had a significantly higher elastic modulus than did cancerous HepG2 cells, as the primary force-bearing structure (Stamenovic ´ & we next investigated the key contributor responsible for Coughlin 1999; Wang & Stamenovic ´ 2000; Wu et al. Elastic modulus (kPa) Animal Cells and Systems 175 2000; Svetina et al. 2001), we investigated whether 3.3. Elastic moduli confirmation measured by a de- apparent differences in the actin and MT networks adhesion assay existed between these two cell types. As compared to The AFM results indicated that THLE-2 cells had a HepG2 cells, THLE-2 cells had more prominent actin greater elasticity than HepG2 cells. To confirm this stress fibers, which are responsible for cellular con- difference in elasticity between the two cell types, we tractility, but the structural organization of the MTs performed a de-adhesion assay, as proposed by Sen and in both cell types was not very different qualitatively Kumar (2009). To perform the de-adhesion assay, where both exhibited well-spread filamentous net- THLE-2 and HepG2 cells were cultured on glass work throughout the cytoplasm with higher concen- substrates for 48 hours and were then subjected to trypsin-EDTA treatment to induce detachment. As the tration around the nucleus (Figure 2a). Schematic adherent cells were enzymatically detached, the rate at illustrations shown in the figures qualitatively repre- which the cells rounded was quantified to obtain the sent these differences in two cytoskeletal proteins in time constant for cellular contractility. As shown in THLE-2 and HepG2. To test whether these differ- Figure 3a, the detachment response of both cell types ences in actin structures between the two cell types consisted of the following three well-defined phases: were responsible for the differences in cellular elastic initial lag period, rapid cell contraction, and plateau moduli, we performed an inhibitor study to disrupt phase. These phases were fit to a sigmoidal curve with actin filaments, and we measured the elastic modulus two characteristic time constants, t and t , whose 1 2 in each cell type after treatment with CD, which values represented the mechanical state of the cells. selectively disrupts filamentous actin. For com- Figure 3b shows a schematic representation of the parison, MTs were also disrupted by colchicine treat- detachment process, where t (iii) represents the ment, and the elasticity measurements were performed adhesion strength of the cell relative to the substrate (Figure 2b). (which also depends on cellular contractility) and t A conical tip and an indentation depth of 1 mm (iiiii) characterizes the elasticity (or contractility) of were used for the HS model analysis. As shown in the cell in an inversely proportional manner (Sen & Figure 2c, the average elastic modulus for the THLE- Kumar 2009). As shown in Figure 3c, THLE-2 cells 2 cells was 11.7397.93 kPa (n 21) in control cells showed much shorter t and t values with a higher 1 2 with no chemical treatment, 9.3897.96 kPa (n 19) elasticity compared with HepG2 cells; this result was in MT-disrupted cells, and 4.1892.61 kPa (n 16) in consistent with the AFM-based measurements. Be- actin-disrupted cells. As shown in Figure 2c, the cause cells with a higher elasticity will contract more average elastic modulus for HepG2 cells was 3.489 rapidly upon protease treatment for detachment, we 3.16 kPa (n 26) in control cells with no chemical expected the t values to be lower. Because the de- treatment, 3.7591.83 kPa (n 8) in MT-disrupted adhesion dynamics are closely related to the internal cells, and 1.9791.46 kPa (n 12) in actin-disrupted tension imposed by the tension-bearing actin stress cells. Although normal THLE-2 cells had greater fibers, it is likely that the THLE-2 and HepG2 cells elastic moduli than did the cancerous cells, the exhibit different intracellular cytoskeletal organization. disruption of cytoskeletal proteins, particularly actin As demonstrated by the immunofluorescence images in disruption via CD, resulted in similar elastic moduli Figure 2a, THLE-2 cells showed stronger stress fiber values for both cell types (P-value0.014). This arrays as compared to HepG2 cells. When the stress result implied that upon disruption of actin filaments, fibers were disrupted by CD that interferes with actin the differences in the elastic moduli of the two cell polymerization, cellular elasticity was significantly types were no longer significant, thus indicating that reduced. Because the HepG2 cells are less elastic, these cells should require more time to contract to their the key contributor for a high modulus in normal original state and therefore exhibit a larger time cells must be the inherently stronger expression and constant. Although this assay was unable to provide cell-specific structural organization patterns of actin numerical values of the elastic moduli, it confirmed the filaments. Although MT disruption by colchicine AFM-based elasticity measurements that demon- disrupted MT organization, the consequential effect strated the differences in the cellular mechanical on the elastic modulus was minimal, which was properties between the two cell types. similar to previously published results (Wu et al. 2000).Therefore, we concluded that the elastic mod- ulus difference between HepG2 and THLE-2 cells 4. Discussion arises from their inherently different expression level and structural organization patterns of actin In this study, we reported differences in elasticity, as filaments. obtained by AFM, between HepG2 cells and THLE-2 176 Y. Kim et al. (a) 1.0 0.8 0.6 0.4 0.2 THLE-2 HepG2 0 100 200 300 400 500 600 700 800 900 Time (sec) (b) i ii iii (c) 250 1 1 2 2 THLE-2 HepG2 THLE-2 HepG2 *p<0.0001 Figure 3. De-adhesion dynamics of THLE-2 and HepG2 cells. Cells were sequentially washed with PBS and incubated with warm trypsin and then imaged every 10 seconds until they became rounded and eventually detached. De-adhesion was quantified by plotting the normalized area as a function of time. The normalized area versus time data was fit to a Boltzmann sigmoid equation to determine the t and t time constants. (a) A decrease in cell spreading area at each time point revealed that the normal cells had a relatively 1 2 weaker adhesion strength and higher contractility and therefore required less time for the detachment and roundup compared with the cancerous cells. (b) Schematic illustration of the de-adhesion process, in which the time interval between i and ii (t )representsthe biological adhesion strength, and the time interval between ii and iii (t )reflects the elasticity of the cell. (c) De-adhesion analysis revealed that THLE-2 cells had significantly higher elasticity compared with HepG2 cells. Error bars indicate standard error. De-adhesion time (sec) Normalized Area Animal Cells and Systems 177 cells, and we investigated the key factors contributing (Costa & Yin 1999; Pesen & Hoh 2005). Provided to the differential mechanical properties of the two cell that the experimental conditions were properly se- types. The mechanical characterization based on the lected, these results demonstrate the implications of HS model showed that this model could provide mechanical characterization using both AFM-based diagnostic potential for liver cancer because THLE-2 measurements and de-adhesion assays for use in the and HepG2 cells exhibited significantly different elas- diagnostic confirmation of liver cancer. Prior to any ticity ranges, and there was a significantly higher practical application of such techniques for the purpose average elastic modulus for the THLE-2 cells compared of cancer diagnosis, however, one would need to test with the HepG2 cells. This difference was also con- primary samples from cancer patients. firmed by a de-adhesion assay where THLE-2 cells exhibited much shorter contraction times than did HepG2 cells. Based on these results, we confirmed the Acknowledgments diagnostic potentials of AFM-based measurements This research was supported by grants from the Fundamental and de-adhesion assays for liver cancer diagnoses. Research Project (Korean Institute of Machinery and Mate- Because the mechanical properties of cells are closely rials) and Basic Science Research Program (National Re- associated with the internal tension induced by actin search Foundation of Korea funded by the Ministry of Education, Science and Technology; 2010*22871). The cytoskeletal forces, we confirmed that THLE-2 cells authors are grateful to Mina Kim from the Soft Biomecha- have more prominent actin stress fibers than do HepG2 nics and Biomaterials Laboratory of KAIST for cell pre- cells. Upon disruption of these stress fibers by inhibit- paration and for her enthusiastic help. The authors are also ing actin polymerization with CD treatment, the grateful to Dr. Junhee Lee and Dr. Wandu Kim from the difference between the elastic moduli of the two cell Nanotechnology Research Team of the Korea Institute of types became much less significant. Based on these Machinery and Materials for their technical assistance. observations, we concluded that the difference in the cellular elasticity of THLE-2 and HepG2 cells could be References attributed mainly to differential arrangements of actin Alcaraz J, Buscemi L, Grabulosa M, Trepat X, Fabry B, stress fibers. Farre ´ R. 2003. Microrheology of human lung epithelial These results demonstrate that the HS model has cells measured by atomic force microscopy. Biophys J. limitations, including property variation, which depend 84:20712079. on the indentation depth and tip geometry. To improve Alexopoulos LG, Haider MA, Vail TP, Guilak F. 2003. diagnostic significance, the mechanical AFM measure- Alterations in the mechanical properties of the human ments require the appropriate tip geometry and in- chondrocyte pericellular matrix with osteoarthrit. J dentation depth. Because the P-value resulting from a Biomech Eng. 125:323333. Bausch AR, Mo ¨ ller W, Sackmann E. 1999. Measurement of comparison between HepG2 and THLE-2 cells using a local viscoelasticity and forces in living cells by magnetic conical tip exhibited more significant results than those tweezers. Biophys J. l76:573579. obtained with aspherical tip with an indentation depth Bercoff J, Chaffai S, Tanter M, Sandrin L, Catheline S, Fink of less than 1500 nm, the conical tip indentation was M, Gennisson JL, Meunier M. 2003. In vivo breast tumor selected as more appropriate for accurate liver cancer detection using transient elastography. Ultrasound Med Biol. 29:13871396. diagnoses. The variation in the modulus for indentation Caille N, Thoumine O, Tardy Y, Meister JJ. 2002. Contribu- depths of less than 200 nm may have been due to tion of the nucleus to the mechanical properties of thermal fluctuation at the cell surface (Mathur et al. endothelial cells. J Biomech. 35:177187. 2001). As the probe indents deeper into the cell, the Charras GT, Lehenkari PP, Horton M. 2001. Atomic force apparent elastic modulus gradually increases because microscopy can be used to mechanically stimulate the measurement is affected by the underlying substrate osteoblasts and evaluate cellular strain distributions. Ultramicroscopy. 86:8595. or nucleus as the deformation by the AFM tip becomes Chen J, Fabry B, Schiffrin EL, Wang N. 2001. Twisting larger (Radmacher 1997; Costa & Yin 1999). In the integrin receptors increases endothelin-1 gene expression case of a spherical tip, the apparent difference in the in endothelial cells. Am J Physiol Cell Physiol. elastic moduli between THLE-2 and HepG2 cells 280:C1475C1484. resulted in a steeper incline at a deeper probing range. Cochlin DL, Ganatra RH, Griffiths DF. 2002. Elastography However, substrate contribution is of less importance in the detection of prostatic cancer. Clin Radiol. 57:1014 in the case of a conical tip indentation, and this Costa KD. 2003. Single-cell elastography: probing for disease contribution has been shown to be negligible when with the atomic force microscope. Dis Markers. 19:139 the strain is less than 25% (Mahaffy et al. 2000). Therefore, indentation depths for AFM experiments Costa KD, Yin FC. 1999. Analysis of indentation: implica- should be selected after taking into account the tions for measuring mechanical properties with atomic nucleus, substrate, and thermal fluctuation effects force microscopy. J Biomech Eng. 121:462471. 178 Y. Kim et al. Cross SE, Jin YS, Rao J, Gimzewski JK. 2007. Nanomecha- [dissertation]. Boston (MA): Massachusetts Institute of nical analysis of cells from cancer patients. Nat Nano. Technology. 2:780783. Pesen D, Hoh JH. 2005. Micromechanical architecture of the Dao M, Lim CT, Suresh S. 2003. Mechanics of the human endothelial cell cortex. Biophys J. 88:670679. red blood cell deformed by optical tweezers. J Mech Phys Plewes DB, Bishop J, Samani A, Sciarretta J. 2000. Visualiza- Solids. 51:22592280. tion and quantification of breast cancer biomechanical Dong C, Skalak R, Sung KL. 1991. Cytoplasmic rheology of properties with magnetic resonance elastography. Phys passive neutrophils. Biorheology. 28:557567. Med Biol. 45:15911610. Goldmann WH, Ezzell RM. 1996. Viscoelasticity in wild- Radmacher M. 1997. Measuring the elastic properties of type and vinculin-deficient (5.51) mouse F9 embryonic biological samples with the AFM. IEEE Eng Med Biol carcinoma cells examined by atomic force microscopy Mag. 16:4757. and rheology. Exp Cell Res. 226:234237. Salomon G, Kollerman J, Thederan I, Chun FKH, Budaus L, ¨ ¨ Guck J, Schinkinger S, Lincoln B, Wottawah F, Ebert S, Schlomm T. 2008. Evaluation of prostate cancer detec- Romeyke M. 2005. Optical deformability as an inherent tion with ultrasound real-time elastography: a compar- cell marker for testing malignant transformation and ison with step section pathological analysis after radical metastatic competence. Biophys J. 88:36893698. prostatectomy. Eur Urol. 54:13541362. Guo Q, Xia Y, Sandig M, Yang J. 2012. Characterization of Sen S, Kumar S. 2009. CellMatrix de-adhesion dynamics cell elasticity correlated with cell morphology by atomic reflect contractile mechanics. Cell Mol Bioeng. 2:218 force microscope. J Biomech. 45:304309. He ´non S, Lenormand G, Richert A, Gallet F. 1999. A new Shao JY. 2002. Finite element analysis of imposing femto- determination of the shear modulus of the human newton forces with micropipette aspiration. Ann Biomed erythrocyte membrane using optical tweezers. Biophys J. Eng. 30:546554. 76:11451151. Shroff SG, Saner DR, Lal R. 1995. Dynamic micromecha- Jones WR, Ting-Beall HP, Lee GM, Kelley SS, Hochmuth nical properties of cultured rat atrial myocytes measured RM, Guilak F. 1999. Alterations in the Young’s modulus by atomic force microscopy. Am J Physiol. 38:C286 and volumetric properties of chondrocytes isolated from C292. normal and osteoarthritic human cartilage. J Biomech. Stamenovic ´ D, Coughlin MF. 1999. The role of prestress and 32:27. architecture of the cytoskeleton and deformability of Kamgoue ´ A, Ohayon J, Tracqui P. 2007. Estimation of cell cytoskeletal filaments in mechanics of adherent cells: a Young’s modulus of adherent cells probed by optical and quantitative analysis. J Theor Biol. 201:6374. magnetic tweezers: influence of cell thickness and bead Suresh S. 2007a. Biomechanics and biophysics of cancer cells. immersion. J Biomech Eng. 129:523530. Acta Mater. 55:39894014. Kim Y, Kim M, Shin JH, Kim J. 2011. Characterization of Suresh S. 2007b. Nanomedicine: elastic clues in cancer cellular elastic modulus using structure based double detection. Nat Nanotechnol. 2:748749. layer model. Med Biol Eng Comput. 49:453462. Svetina S, Bozic B, Derganc J, Zeks B. 2001. Mechanical and Mahaffy R, Shih C, MacKintosh FC, Ka ¨ s J. 2000. Scanning functional aspects of membrane skeletons. Cell Mol Biol probe-based frequency-dependent microrheology of poly- Lett. 6:677690. mer gels and biological cells. Phys Rev Lett. 85:880883. Unnikrishnan GU, Unnikrishnan VU, Reddy JN. 2007. Maniotis AJ, Chen CS, Ingber DE. 1997. Demonstration of Constitutive material modeling of cell: a micromechanics mechanical connections between integrins cytoskeletal approach. J Biomech Eng. 129:315323. filaments, and nucleoplasm that stabilize nuclear struc- Wang N, Stamenovic ´ D. 2000. Contribution of intermediate ture. Proc Natl Acad Sci USA. 94:849854. filaments to cell stiffness, stiffening, and growth. Am J Mathur AB, Collinsworth AM, Reichert WM, Kraus WE, Physiol Cell Physiol. 279:C188C194. Truskey G. 2001. Endothelial, cardiac muscle and skeletal Wu ZZ, Zhang G, Long M, Wang HB, Song GB, Cai SX. muscle exhibit different viscous and elastic properties as 2000. Comparison of the viscoelastic properties of determined by atomic force microscopy. J Biomech. normal hepatocytes and hepatocellular carcinoma cells 34:15451553. under cytoskeletal perturbation. Biorheology. 37:279 McKnight AL, Kugel JL, Rossman PJ, Manduca A, Hart- mann LC, Ehman RL. 2002. MR elastography of breast Ziemann F, Ra ¨ dler J, Sackmann E. 1994. Local measure- cancer: preliminary results. Am J Roentgenol. 178:1411 ments of viscoelastic moduli of entangled actin networks Ottensmeyer MP. 2001. Minimally invasive instrument for in using an oscillating magnetic bead micro-rheometer. vivo measurement of solid organ mechanical impedance Biophys J. 66:22102216.
Journal
Animal Cells and Systems
– Taylor & Francis
Published: Jun 1, 2013
Keywords: Hertz–Sneddon model; atomic force microscopy; cellular elasticity; de-adhesion assay; hepatocytes
