(F-213) Developing a Rigorous Method for AFM Force Mapping of Optic Nerve Tissue

Introduction:
Atomic force microscopy (AFM) is widely used for measuring biomechanical properties of biological materials. While there is extensive literature on AFM for studying cell mechanics in culture, measurements of whole tissue cryosections, especially for soft tissues, are less well-characterized. Soft tissue AFM biomechanical measurements can be variable due to the highly heterogenous nature of biological tissue samples, and a well-validated data analysis pipeline is necessary to draw reliable conclusions from Young’s modulus estimates. Force mapping, also referred to as force-volume measurements, can provide spatial information about the tissue stiffness across a sample. The objective of this study is to validate an AFM force mapping and data analysis pipeline that characterizes the mechanical properties of cryosectioned mouse optic nerve head tissue in a repeatable and rigorous manner.

Materials & Methods:
Eyes were enucleated from 10–11-month-old male and female mice (n=5), flash frozen, and longitudinally cryosectioned (16 µm thickness). AFM measurements were performed in the glial lamina region, within approximately 100 µm of the posterior sclera. An MFD-3D AFM (Asylum Research, Santa Barbara, CA) with a 10 µm diameter spherical probe attached to a silicon nitride cantilever (0.01 N/m) was used to acquire force maps (4x4 grid; 40x40 µm area). The Hertz model was used to fit force-indentation curves and determine the effective Young’s modulus at each location. Each force map was repeated, and the Young’s modulus was averaged between the two measurements. Measurements where indentation depth into the tissue exceeded 2 µm were removed due to substrate effects. Outliers were defined and excluded based on agreement between repeat measurements (Cook’s outlier removal), as well as force-displacement curve fitting quality. Only eyes where more than 30 force measurements remained after outlier removal were included in analysis to ensure validity of statistical tests.

Results:
Young’s modulus values followed a lognormal distribution, with a geometric mean and multiplicative standard deviation (SD) across animals of 1.51 ⋇ 4.26 kPa. The log-transformation and use of geometric mean and SD better represents the data compared to the arithmetic mean and SD of 4.27 ± 8.06 kPa, which would imply that a large portion of Young’s modulus values were negative. Approximately 21% of measurements were defined as outliers and excluded due to poor Hertz model fitting in one of two measurements or Cook’s distance outlier criteria, which resulted in a 0.19 decrease in geometric SD compared to analysis without repeating force maps.

Conclusions:
The glial lamina region of the optic nerve head consists mostly of astrocytes and retinal ganglion cell axons, with some blood vessels and extracellular matrix. Due to the softness and heterogenous composition of the tissue, obtaining a repeatable Young’s modulus value with AFM was challenging. Here, the use of repeated measurements, outlier exclusion, and lognormal data transformation increased confidence in Young’s modulus estimations.

Discussion:
Although we tested our methodology using mouse optic nerve head tissue, the proposed approach should enable more consistent and repeatable AFM stiffness measurements of other soft tissues. This work also builds on existing AFM literature by measuring stiffness in tissue cryosections rather than cultured cells, capturing mechanical properties in situ, and allowing direct measurements on animal models of disease where other phenotypic information about the tissue is known. In the future, we will apply this pipeline to compare optic nerve head stiffness between wild type and transgenic animal models of optic neuropathy, as well as to measure other tissues.