Superresolution Pattern Recognition Reveals the Architectural Map of the Ciliary Transition Zone.

Cell Culture Human

RPE-1 cells (in some cases with stably expressed centrin 2-eGFP) 60 were grown on #1.5 cover glasses coated with poly-L-lysine, in DMEM/F12 supplemented with 10% FBS, 12 mM HEPES, 2.5 mM L-glutamine, 2.4 g/L sodium bicarbonate, and 1% penicillin/streptomycin at 37 °C and 5% CO 2 up to 60% ∼ 90% confluence depending on the experimental conditions, and then incubated in serum-deprived media for 48 hours. For lithium stimulation experiments, cells were treated with 100 mM lithium chloride (L121-100, Fisherbrand) for one hour before fixation.

Antibodies

Ac-tub antibody (anti-acetylated α-tubulin mouse IgG, ab24610, Abcam) was used as a ciliary marker at 1/1000 dilution for epifluorescence imaging and 1/2000 for single-color STED imaging. IFT88 (anti-IFT88 rabbit IgG, 13967-1-AP, Proteintech) was prepared at 1/200 dilution. RPGRIP1L (anti-RPGRIP1L rabbit, HPA039405, Sigma-Aldrich, at 1/300 dilution), TCTN2 (anti-TCTN2 mouse IgG2a, ab119091, Abcam, at 1/500 dilution), MKS1 (anti-BBS13 rabbit IgG, 16206-1-AP, Proteintech, at 1/50 dilution) and TMEM67 antibodies (anti-TEME67 rabbit IgG, 13975-1-AP, Proteintech, at 1/200 dilution) were used for labeling the TZ. For CEP290, two antibodies were used, one for the N-terminal and one for the C-terminal (see comparison in Supplementary Fig. 6 ). For the N-terminal CEP290 antibody, a synthetic peptide consisting of residues 11–24 of mouse Cep290 with a C-terminal cysteine added for conjugation to KLH (IKVDPDDLPRQEEL-C) was used to generate rabbit polyclonal antibodies (Genscript). The resulting antiserum was affinity purified on an agarose support (Thermo Fisher Scientific) according to the manufacturer’s instructions. This N-terminal CEP290 antibody was used at 1/100 dilution. The C-terminal CEP290 (anti-CEP290 rabbit IgG, ab84870, Abcam) against residues 2429–2479 was used at 1/500 dilution. CEP164 (anti-CEP164 rabbit, 45330002, Novus Biologicals) antibody was used at 1/2500 dilution for staining of TFs. Information of isotopes and immunogens of primary antibodies is listed in Supplementary Table 2 . For secondary antibodies, Oregon Green 488 antibodies (goat anti-rabbit or goat anti-mouse, Invitrogen) were used at 1/200 dilution for TCTN2, MKS1, and RPGRIP1L and 1/1000 for TMEM67, CEP290, CEP164, and IFT88. Biotin antibodies (anti-mouse IgG, B6649 or anti-rabbit IgG, B8895, Sigma-Aldrich) were diluted at 1/200 and V500 streptavidin (561419, BD Horizon) was diluted at 1/100 for labeling the TZ/TF/IFT proteins for two-color STED imaging or at 1/500 for ac-tub and IFT88 for single-color STED imaging.

Immunofluorescence

The cells were fixed for 5 min at room temperature with 4% paraformaldehyde (PFA) immediately followed by methanol at −20 °C for 10 min. Following fixation, cells were washed 2–3 times in PBS, permeabilized with 0.1% PBST (PBS + Triton X-100), and then blocked with 1% normal goat serum and 2.5% BSA in 0.1% PBST for 30 min at room temperature. Samples were incubated with primary antibodies in block solution for 1 hour at room temperature, and rinsed at least 3 times in PBST. Cover glasses were then incubated with secondary antibodies or biotin antibodies for 45 min at room temperature, and washed again 3 times in PBST. For biotin antibodies, the cells were stained with streptavidin conjugated to V500 dye (BD Horizon) for 20 ∼ 30 min, followed by at least 5 times in PBST. Finally, the samples were mounted with 86% glycerol in PBS.

STED Microscopy

The home-built continuous-wave STED setup (491-nm for excitation and 592-nm for depletion) was as previously described 44 , with the addition of a 447-nm diode excitation laser (PGL-V-H-447, CNI) for dual-color STED imaging. Three laser beams were merged and focused onto the sample through a 100× oil immersion objective (Olympus UPLSAPO100×−1.4 NA). The fluorescent signals were collected with the same objective and detected by an avalanche photodiode (APD) module (SPCM-AQR-15, PerkinElmer Optoelectronics). For single-color STED imaging of TZ/TF proteins (except TCTN2), ac-tub imaged in the confocal mode (447-nm laser) was used to find cilia only and one TZ/TF protein at a time was later imaged in the STED mode (491-nm laser). Ac-tub was stained with V500 at a low concentration to avoid crosstalk to the channel of TZ/TF proteins labeled with Oregon Green 488 and illuminated at 491 nm for STED imaging. For TCTN2, anti-IFT88 detected with a low concentration of V500 was used as a ciliary marker. For dual-color STED imaging, centrin-eGFP excited at 491 nm was used to identify the ciliary base, while CEP290, RPGRIP1L, MKS1, TMEM67, TCTN2, or CEP164 was labeled with V500 at the 447-nm channel. Since GFP was partially excitable at 447 nm, centrin could be detected in both channels; as a result, two sequential acquisitions of STED images could be readily aligned by correlating the centrin location. To determine the position of IFT88 relative to TCTN2, IFT88 (447-nm laser channel) was first used to locate cilia in the confocal mode and then STED images were acquired for both proteins. For all STED imaging, the excitation power was operated at 0.3–1.5 μW depending on the signal level of TZ/TF/BB/IFT proteins. For CEP290, TMEM67, CEP164, centrin, or IFT88 where the signals were strong, the excitation laser was set close to the lower bound of the working power. For MKS1 where the signals were substantially weaker ( Supplementary Fig. 8 ), the excitation power was generally operated at the upper bound. A medium excitation power close to 1 μW was used for imaging TCTN2 and RPGRIP1L. The power of the depletion laser was set differently for different proteins based on photon counts. The power was generally set to ~100 mW for centrin, 90 ∼ 100 mW for CEP290, TMEM67, CEP164, or IFT88, 70 ∼ 90 mW for TCTN2 and RPGRIP1L, and ~60 mW for MKS1. When signals were weakened after a few acquisition scans, the depletion power could be reduced by ~10 mW to maintain a sufficient signal level. To find an optimal focus, STED images were acquired at different planes every 100 nm around an initial position until the sample was photobleached substantially, usually after 3–5 times of scanning. For all STED scans, a step size of 25 nm and a dwell time of 20 μs were used to gain the maximal signal-to-noise ratio as well as to satisfy the Nyquist criteria for spatial sampling. The registered photon count per 20 μs was generally above 100 for confocal imaging and about 50% reduced during STED acquisition. For image display in figures, we performed deconvolution to enhance the signals and thresholding to enhance contrast relative to the background signals. Transmission Electron Microscopy G1-arrested RPE-1 cells grown on coverslips made of Aclar film (Electron Microscopy Sciences) were fixed in 4% paraformaldehyde and 2.5% glutaraldehyde with 0.1% tannic acid in 0.1 M sodium cacodylate buffer at room temperature for 30 min, postfixed in 1% OsO 4 in sodium cacodylate buffer for 30 min on ice, dehydrated in a graded series of ethanol, infiltrated with EPON812 resin (Electron Microcopy Sciences), and then embedded in the resin. Serial sections (~90 nm thickness) were cut on a microtome (Ultracut UC6; Leica) and stained with 1% uranyl acetate as well as 1% lead citrate. Samples were examined on a JOEL transmission electron microscope.

Data Analysis

The illustrated confocal and STED images were smoothened and contrast-enhanced with generalized Tikhonov regularization 61 . Thresholding was applied to illustrate the FWHM of objects in Fig. 1c–j using ImageJ. For quantitative width measurement of the TZ/TF proteins, the STED images were first cleaned by a mean filter of 1 pixel and background subtracted. For CEP290 and RPGRIP1L, images were fitted with a two-dimensional Gaussian function ( Supplementary Fig. 3M,N ) to identify the principal axes and to find the FWHMs ( Supplementary Fig. 3A,B,G,H ), where c1 and c2 were used to determine the FWHM by multiplying the factor of . The fitting curves agreed well with the data, except that some RPGRIP1L images had a slight intensity dip in the middle. For MKS1, TMEM67, TCTN2, and CEP164, STED images clearly displayed two distinct peaks in the lateral direction, so the lateral FWHMs were determined by measuring the distances of two intensity peaks ( Supplementary Fig. 3C–F,I–L ). To define the reference coordinate in the axial direction, the distal edge of centrin was located by pinpointing the position of the FWHM of the fluorescent intensity along the primary axis of centrin ( Supplementary Fig. 4 ). Note that the visual artifact of the size difference between Figs 1 and 2 (e.g. RPGRIP1L in Figs 1f and 2 b; MKS1 in Figs 1 g and 2c ) was primarily caused by the intensity threshold chosen to show more of centrin; the measured FWHMs of these two figures were very similar and the axial distance was not affected by its visual size. To obtain the 7-color superresolution image, six single-color STED images including centrin, CEP164, CEP290, RPGRIP1L, MKS1, and TCTN2 were first merged using ImageJ (three additive primary colors RGB and three subtractive primary colors CMY). This merged image was then overlapped with the image of TMEM67 (azure color) using ImageJ. To obtain a white-background 7-color superresolution image, a negative process was applied for the whole merged image using ImageJ. To obtain the multicolor superresolution images of IFT88 with TZ/TF/BB proteins, centrin, CEP164, CEP290, RPGRIP1L, TMEM67, and IFT88 were merged using ImageJ with red-hot color assigned for IFT88 images.