SHG-specificity of cellular Rootletin filaments enables naïve imaging with universal conservation

Abstract

Despite growing demand for truly naïve imaging, label-free observation of cilium-related structure remains challenging, and validation of the pertinent molecules is correspondingly difficult. In this study, in retinas and cultured cells, we distinctively visualized Rootletin filaments in rootlets in the second harmonic generation (SHG) channel, integrated in custom coherent nonlinear optical microscopy (CNOM) with a simple, compact, and ultra-broadband supercontinuum light source. This SHG signal was primarily detected on rootlets of connecting cilia in the retinal photoreceptor and was validated by colocalization with anti-Rootletin staining. Transfection of cells with Rootletin fragments revealed that the SHG signal can be ascribed to filaments assembled from the R234 domain, but not to cross-striations assembled from the R123 domain. Consistent with this, Rootletin-depleted cells lacked SHG signal expected as centrosome linker. As a proof of concept, we confirmed that similar fibrous SHG was observed even in unicellular ciliates. These findings have potential for broad applications in clinical diagnosis and biophysical experiments with various organisms.

Figure 1. Truly label-free penta-modal CNOM visualizes hierarchical structures of retinal layers, highlighting novel SHG-indicated filamentous layers.

Spectroscopic imaging was performed with fixed rat retina using our custom microscope. (a) A collection of label-free images of SHG (green) and CARS at 2860 cm⁻¹ (red) and 1570 cm⁻¹ (blue), showing the distributions of harmonophores, lipid, and DNA, respectively. Note the unexpected filamentous layer in the middle of the SHG image. (b) Representative image of retinal layers by H&E staining with the indicated anatomies. OS, IS, and ONL correspond to outer segment, inner segment, and outer nuclear layer, respectively. (c) Representative spectra of SHG for image reconstruction (spot (i) in (a)). (d) Representative retrieved spectra of CARS at the positions of (ii) and (iii) in (a). Right, higher magnification of left at 1800–800 cm⁻¹. Because raw CARS spectra were distorted due to interference between vibrationally resonant and non-resonant coherent signals, we used the maximum entropy method (MEM) to retrieve the raw CARS spectra to Im[χ⁽³⁾] spectra corresponding to the conventional spontaneous Raman scattering spectrum. (e) Schematic of a photoreceptor highlighting the internal filamentous structures connecting cilia and ciliary rootlets. Parentheses indicate marker antibodies. Bar, 20 μm. Color scales indicate intensity (a.u.) or amplitude (a.u.).

Figure 2. Retinal SHG corresponds to photoreceptor ciliary rootlets.

Fixed rat retina was immunostained with the indicated marker antibodies, followed by superimposed observations with DeltaVision as a standard (a) or custom-built multimodal microscope equipped for SHG and additional TPEF modes using 775 nm excitation to detect Alexa Fluor 488 immunofluorescence (b). (a) Ac-tubulin (red) and Rootletin staining (green) are well separated by standard microscopy, marking two distinct filamentous layers: photoreceptor connecting cilia and ciliary rootlet, respectively. See also the schematic in Fig. 1e. (b,c) Superimposition of SHG signal (green) and TPEF-detected single marker staining (red). Note that the SHG signal corresponds to the Rootletin staining, indicating ciliary rootlet, but not to ac-tubulin staining, indicating connecting cilia. DAPI (blue) was used as a nuclear marker. Bars, 10 μm. Color scales indicate intensity (a.u.).

Figure 3. SHG is ascribed to the Rootletin R234 domain, which assembles filaments.

Myc-tagged deletion constructs of human Rootletin were introduced into COS7 cells by transfection. Transfectants were fixed and subjected to multiple microscopic observations. (a) Full-length Myc-human Rootletin assembles intact filaments with periodic cross-striation at the electron-microscopic level, like endogenous Rootletin filaments. (b) Schematic diagram of humanized and Myc-tagged Rootletin deletion mutants: R1, Globular head domain; R2-4, rod domains. See also ref. . (c) Each human deletion mutant assembles a distinct architecture, as revealed by standard immunofluorescence, as previously reported for the murine case. Note that the minimal filament-forming domain is R234. Anti-Myc (green), DAPI (blue). (d) Multimodal microscopy highlights the correlation of SHG with filament assembled from R234 domain. TPEF (green), SHG (red). (e) Quantitation of the SHG in (d), normalized to Myc intensity per cell. (f) Distinct ultrastructures assembled from each fragment: R1, aggregates with high electron density; R123, massive cross-striation; R234, densely packed filaments. (g) Summary of multilateral domain analyses. Bars, 200 nm in (a and f) and 10 μm in (c and d). Error bars represent s.e.m. ****P < 0.001. N = 5 in (e). Color scales indicate intensity (a.u.).

Figure 4. SHG induction shows quantitative agreement with the density of R234 filaments.

(a) Standard immunofluorescence highlighted unsharpened Myc-R234 filaments co-expressed with the FLAG-R123 domain. Anti-Flag (red), Anti-Myc (green), DAPI (blue). (b) SHG induction was weaker under co-expression. TPEF (green), SHG (red). (c) Quantitative reduction of SHG in (b), compared to Myc intensity in each cell. (d) Electron micrograph depicting ultrastructural alteration due to FLAG-R123 co-expression: (left) pure R234 formed packed filaments; (right) FLAG-R123 co-expression resulted in formation of more widely spaced filaments with additional cross-striations, similar to the case of full-length Rootletin (Fig. 3a). (e and f) Quantitation of filaments in (d), as both intensity (e) and number (f) note the correlation to SHG in c. Bars, 10 μm in (a and b), and 100 nm in (d). Error bars represent s.e.m. *P < 0.05 in (c), ****P < 0.001 in (e and f). N = 5, 30, and > 28 in (c,e and f), respectively. Color scales indicate intensity (a.u.).

Figure 5. Cellular SHG is derived from Rootletin filaments.

(a–e) U2OS cells were treated with negative control or Rootletin-specific siRNA, followed by immunofluorescence for multimodal observations. (a) Standard double immunofluorescence confirmed both Rootletin staining proximal to the centrosome (left) and its depletion by siRootletin, accompanied by centrosome splitting (middle and right). Anti-Rootletin (green), SHG (red). (b) Multimodal microscopic observation revealed SHG induction in Rootletin-positive adjacent centrosomes (left) but not in Rootletin-depleted split centrosomes (middle and right). TPEF (green), SHG (red). (c) Immunoblot confirming Rootletin knockdown. (d and e) Quantitative data for (a and b), respectively; note the mutual correlation. (f) Multimodal images of T. thermophila as xy plane; note a lot of fibrous signals in SHG (green). Lipid (red and yellow) as a control. Pericentrin 2 or γ-tubulin was used as a centrosomal marker. DAPI (blue) was used as a nuclear marker. Bars, 2 μm. GAPDH was used as a loading control. Error bars represent s.e.m. ****P < 0.001 and **P < 0.01 in d and e, respectively. N = 10 and 5 in (d and e), respectively. Color scales indicate intensity (a.u.).