The role of cell body density in ruminant retina mechanics assessed by atomic force and Brillouin microscopy

Abstract

Cells in the central nervous system (CNS) respond to the stiffness of their environment. CNS tissue is mechanically highly heterogeneous, thus providing motile cells with region-specific mechanical signals. While CNS mechanics has been measured with a variety of techniques, reported values of tissue stiffness vary greatly, and the morphological structures underlying spatial changes in tissue stiffness remain poorly understood. We here exploited two complementary techniques, contact-based atomic force microscopy and contact-free Brillouin microscopy, to determine the mechanical properties of ruminant retinae, which are built up by different tissue layers. As in all vertebrate retinae, layers of high cell body densities ('nuclear layers') alternate with layers of low cell body densities ('plexiform layers'). Different tissue layers varied significantly in their mechanical properties, with the photoreceptor layer being the stiffest region of the retina, and the inner plexiform layer belonging to the softest regions. As both techniques yielded similar results, our measurements allowed us to calibrate the Brillouin microscopy measurements and convert the Brillouin shift into a quantitative assessment of elastic tissue stiffness with optical resolution. Similar as in the mouse spinal cord and the developing Xenopus brain, we found a strong correlation between nuclear densities and tissue stiffness. Hence, the cellular composition of retinae appears to strongly contribute to local tissue stiffness, and Brillouin microscopy shows a great potential for the application in vivo to measure the mechanical properties of transparent tissues.

Figure 1. Comparison between retinae of different ruminant species

Hematoxylin and Eosin staining of (a) ovine and (b) bovine retinae. NFL = nerve fiber layer, GCL= retinal ganglion cell layer, IPL = inner plexiform layer, INL = inner nuclear layer, OPL = outer plexiform layer, ONL = outer nuclear layer, IS = photoreceptor inner segments, OS = photoreceptor outer segments. Scale bar: 20 µm.

Figure 2. Comparison between AFM and Brillouin microscopy measurements of ruminant retinae

(a) Cross-section of an ovine retina, stained with MitoTracker Orange. The individual layers of the retina are clearly discernible. NFL = nerve fiber layer, GCL= retinal ganglion cell layer, IPL = inner plexiform layer, INL = inner nuclear layer, OPL = outer plexiform layer, ONL = outer nuclear layer, IS = photoreceptor inner segments, OS = photoreceptor outer segments. Scale bar = 100µm. (b) Elasticity map of that retina assessed by AFM. Each pixel corresponds to an individual measurement; pixels containing no data were removed (cf. supplementary figure 3). K is shown for full indentation; the larger K, the stiffer the tissue. (c) Brillouin image of a bovine retinal cross-section. As in the AFM elasticity map, the apical PRL is the stiffest retinal layer, and other layers can be distinguished based on their mechanical properties. The resolution of Brillouin microscopy was 1 µm in both directions. (d, e) Line profiles of retinal elastic stiffness; data points shown correspond to the median values of each pixel row shown in supplementary figure 3 (d) and figure 2c (e).

Figure 3. Mechanical properties of different tissue layers of ruminant retinae

(a) Apparent elastic moduli K of the different layers of ovine retinae determined by AFM. K was significantly different between tissue layers (P < 0.001, Kruskal-Wallis ANOVA; n = 3). NFL = nerve fiber layer, GCL = retinal ganglion cell layer (n = 98), IPL = inner plexiform layer (n = 113), INL = inner nuclear layer (n = 48), OPL = outer plexiform layer (n = 32), PRL = photoreceptor cell layer (n = 180). Plot including outliers shown in supplementary figure 2a. (b) Brillouin shift of the different layers of bovine retinae. As in the AFM measurements, retinae were mechanically heterogeneous, and the Brillouin shift was significantly different between tissue layers (P < 0.001, Kruskal-Wallis ANOVA). n_NFL/GCL = 362, n_IPL = 420, n_INL = 340, n_OPL = 90, n_PRL = 924. Plot including outliers shown in supplementary figure 2b. Red line = median, blue box = Q1–Q3 percentile. * (P < 0.05); ** (P < 0.01); *** (P < 0.001), Dunn-Sidak Multiple Comparison Test.

Figure 4. Correlation between elastic moduli obtained by AFM and Brillouin microscopy

We found a strong correlation in the log-log linear regression between AFM and Brillouin microscopy data (P-value < 0.01; R² = 0.99), which could be best described by the relationship log(M’) = 0.0678 log(K) + 9.2235. Each dot in the graph corresponds to the median of stiffness values for a defined layer within the retina as measured by AFM and Brillouin microscopy. The error bars indicate the standard error of the median. The dashed line indicates a linear fit on the log-log plot.

Figure 5. Correlation between cell density and stiffness in different retinal layers

(a) Image of an ovine retinal cross-section. Cell nuclei are shown in grey (stained using DAPI). NFL = nerve fiber layer, GCL= retinal ganglion cell layer, IPL = inner plexiform layer, INL = inner nuclear layer, OPL = outer plexiform layer, ONL = outer nuclear layer, IS = photoreceptor inner segments, OS = photoreceptor outer segments. Scale bar: 20 µm (b) Relationship between nuclear density A_norm and K for different retinal layers. Shown are mean values ± SEM of the IPL, INL, and ONL. The best fit was achieved using K_c = 3.19 × A_norm + 181.9; we found a strong linear correlation between K and A_norm (R² = 0.99).