【作者】 栗亚；

【导师】 翟文杰；

【作者基本信息】 哈尔滨工业大学 ， 机械设计及理论， 2013， 博士

【摘要】 癌症以其高发病率和致死率已经成为人类面临的重要公共卫生问题，由于癌细胞具有隐蔽性强、无限增殖、侵袭和转移的生物学特性，准确的进行癌症的早期诊断对及时开展治疗最为关键。目前确诊癌症的组织病理学方法需要制备切片，不仅制片过程繁复、成本高、使细胞丧失生理结构，而且诸多因素会影响切片的正确判断，如肿瘤细胞分化程度、细胞的异质性、非特异性染色等，并且这些方法仅限于定性和半定量阶段。如何进行癌症的有效早期诊断已经成为亟待解决的难题，尤其是定量诊断和预报癌变是当前追求的目标。原子力显微镜（AFM）除了能以纳米级别的分辨率在液相中检测活细胞表面的超微形貌，也能通过对细胞施加微弱的作用力获得细胞的力学特性，进而获取细胞内部结构变化和细胞的功能状态信息。尽管AFM已经检测了包括鱼角膜细胞、肿瘤细胞、干细胞等多种细胞的粘弹性，但是其弹性模量值从100Pa-100kPa不等，如此大的差异并非都归因于细胞亚结构的区别，还受到检测条件、方法和模型等方面的影响。因此本文根据准确分析非小细胞肺癌（NSCLC）细胞力学特性的需要，旨在从表面成像参数设定、检测条件、模型选择等角度，提出准确可行的细胞力学特性的评价方法，并据此分析NSCLC细胞力学特性、内部细胞骨架结构及其转移潜能的关系。这也是首次对不同恶性程度的NSCLC细胞力学特性和骨架结构的对比分析。具体研究内容主要包括以下几个方面：提出了优化细胞AFM成像质量的方法，为细胞的力学特性检测奠定基础。利用二次回归正交旋转设计，模拟出了以细胞扫描图像的清晰度为因变量，AFM扫描电压、扫描频率和比例增益为自变量的数学模型，并计算出理论上在扫描电压为0.61V，扫描频率为2.23Hz，比例增益为3.85的条件下，扫描得到的细胞图像最优。以此最佳扫描条件获得了不同恶性程度NSCLC细胞的形貌图，可以较清晰的观察细胞的表面结构，如伪足、分泌颗粒、细胞骨架纤维、细胞间连接甚至核仁等，不同恶性程度的NSCLC细胞的骨架纤维分布存在较大差异。分析多种检测因素对NSCLC细胞弹性模量（E）的影响，提出了肺癌细胞力学特性的准确评价方法，应采用球形针尖以避免锥形针尖对细胞弹性的高估；对于细胞中心和边缘区域的力学检测应分别以半无限和有限厚度的Hertz模型进行准确计算；检测温度应为近细胞生理环境的37℃，细胞形态对细胞弹性无明显影响；应以低于1Hz的加载速率而避免细胞粘性对检测结果的影响。该评价方法为准确检测癌细胞的粘弹性提供了技术保障。提出了以细胞的外部力学特性鉴定肺癌细胞恶性程度的新方法，提供了细胞癌变鉴别的新手段。以三种NSCLC细胞为研究对象，发现恶性程度不同的三种细胞弹性模量关系为：E低恶性>E中恶性>E高恶性。基于AFM的应力衰减实验研究了不同恶性程度的NSCLC的粘性特征，高转移性的大细胞肺癌NCI-H1299与低转移性的A549细胞相比，NCI-H1299细胞的松弛模量ER低26%、Kelvin弹簧常数K1低26%，表观粘性系数μ低23%。证明了肺癌细胞的顺应性和其侵袭性呈正相关，细胞粘弹性的减小使细胞变得更柔软，更有利于转移性癌细胞从原发灶释放、穿过毛细血管的内皮间隙进行内渗和外渗，转移至区域淋巴结甚至远处器官等。利用纳米原位力学检测系统研究了NSCLC细胞的弹性和粘性特征。以Oliver-Pharr方法进行计算，最终得到的细胞的弹性特征：ENCI-H1299<ENCI-H520。观察到了癌细胞在恒力状态下的蠕变现象，将蠕变数据与Voigt模型、Maxwell模型和Kelvin模型进行非线性拟合后，利用Kelvin模型计算两种细胞系的蠕变参数，发现高转移性的大细胞肺癌NCI-H1299的蠕变参数E1、E2、η1和η2明显小于低转移性肺鳞癌细胞NCI-H520(P<0.05)。以细胞骨架图像为研究对象，利用数盒子法计算二值图像的分形维数，发现分形维数可以评价细胞骨架纹理的复杂性特征，可以作为细胞骨架图像定量描述的指标，由此提出了以细胞骨架纤维的分形维数来鉴别癌细胞的辅助方法。由此最终建立了细胞内部骨架结构-细胞外部力学特性-细胞癌变的定量联系。更多还原

【Abstract】 Cancer has become an important public health issue in humanity with its highincidence and fatality rate. Cancer could undergo a hidden unlimited proliferation,invasion and metastasis, therefore accurate early diagnosis of cancer is critical forcarrying out treatment in a timely manner. Histopathology, a definitive diagnosismethod of cancer, needs preparation of tissue slices. Not only these producing processare complicated and high costing which make cells loss physiological structure, andmany other factors will affect the correct judgment of the slice, such as the degree oftumor cell differentiation, cell heterogeneity, non-specific staining etc.. These methodsare restricted to the qualitative and semi-quantitative stage. Effective early diagnosis ofcancer has become a problem demanding prompt solution. The current pursuit of goal isquantitative diagnosis and prognosis of cancer, especially.Atomic force microscope (AFM) can detect the ultrastructural surface morphologyof living cells with nanometer level resolution in the liquid phase, it also can obtain themechanical characteristics of cell through appling a weak force on cell, and thus toobtain the cell internal structural changes. AFM has evaluated viscoelasticity of manydifferent kinds of cells including fish corneal cells, tumor cells, stem cells etc. whoseelastic modulus ranged from100Pa to100kPa. Such large difference is not onlyattributed to the difference between the cellular sub-structures, but also can be affectedby the impact of the detection conditions, methods and models, etc.. Based on the needof accurate analysis of non-small cell lung cancer (NSCLC) cell mechanical properties,this article aimed at the establishment of accurate and viable cancer cell mechanicalproperties evaluation method from surface imaging parameters setting, testingconditions, model selection perspective. Analysis of the relationship among themechanical properties, internal cytoskeleton structure and cellular metastatic potentialof NSCLC cells were performed. This is the first comparative analysis of themechanical properties and cytoskeletal fibers of NSCLC cells with different malignancydegree. This research mainly includes the following aspects:A novel method was proposed to optimize the quality of AFM imaging of livingcells and this layed a foundation for the detection of the mechanical properties of livingcells.A quadratic regression orthogonal design was made to simulate a mathematicalmodel for cell surface scanning. The clarity of cell image surface was defined as thedependent variable, while the AFM scanning setpoint, scanning rate and proportionalgain were defined as the independent variable in this mathematical model. Theoretically,the optimal cell image can be obtained at the scan setpoint of1.31V, the integral gain of1.93, and the proportional gain of3.96. The topography images of NSCLC cells with different malignancy degree were taken by this optimal scanning conditions. Thesurface structure of the cell can be observed clearly, such as cell pseudopod, protrusion,secretory granules, cytoskeletal fiber, cellular connection, nucleolus etc.. Greatdifference exists among the skeleton fibers of NSCLC cells with different degree ofmalignancy.Impact of a variety of detection factors on NSCLC cell elastic modulus (E) wereanalyzed in AFM experiments. Thus a new evaluation method was established for theaccurate identification of mechanical properties of lung cancer cells. Spherical tip isessential to avoid elasticity overestimation which might be caused by pyramidal tip.Mechanical detection at cell center and edge can be accurately evaluated by thesemi-infinite and finite thickness Hertz model, separately. Experiment environmentshould be similar to the physiological environment of37°C. No significant influenceof cell morphology has on cell elasticity. Loading rate should be lower than1Hz toavoid the cellular viscous effect. This evaluation method provided technical support forthe accurate detection of cancerous cell viscoelasticity.A new method was proposed for the identification of lung cancer cells withdifferent malignancy degree by external mechanical characteristics. A new means wasprovided for identification of cancerous cell. Three NSCLC cell lines were investigatedin this study. The elastic modulus relationship of the three NSCLC cells yielded Elowmalignant> Emedium malignant> Ehighly malignant. The viscous characteristics of NSCLC cellswith different malignant degree were established based on the AFM stress relaxationtesting. Compared with the low metastatic A549cell, high-metastatic NCI-H1299cellhad a26%lower relaxation modulus ER,26%lower Kelvin spring constant K1,23%lower apparent viscosity coefficient μ. These data proved that compliance andinvasiveness of lung cancer cell was positive correlation. The reduced viscoelasticitymakes cell softer and more conducive to the release of metastatic cancer cells from theprimary tumor, through the capillary endothelial gap endosmosis and extravasation andmetastasis to regional lymph nodes or distant organs.Elasticity and viscosity characteristics of lung cancer cells were investigated bynanoindentation testing system. The Oliver-Pharr method was used to calculate theelastic modulus of cells and yielded ENCI-H1299<ENCI-H520. Creep phenomenon of cancercells at constant force was observed. Nonlinear fitting between the creep data and Voigtmodel, Maxwell model and Kelvin model was carried out. Kelvin model, which hadperfect fitting with creep data, was used to calculate the creep parameters of both celllines. It was proved that the creep parameters E1, E2, η1and η2of the high-metastaticNCI-H1299cells were significantly lower than the counterparts of NCI-H520cells (P<0.05). The cellular cytoskeleton images were transformed into binary images and fractal box counting was used to calculate the fractal dimension of the binary images. Itwas proved that the fractal dimension can evaluate the complexity of the cytoskeletontexture and can be used as an indicator of quantitative description of the cytoskeletonimage. Thus an ancillary method was presented to identify cancer cells by fractaldimension counting of cytoskeleton. Eventually a quantitative relationship wasestablished, which focused on the cytoskeleton structure, cell mechanical properties andcarcinogenesis.更多还原