CCDC32 is recruited to clathrin-coated pits.

(A) Representative TIRFM images of ARPE-HPV cells that stably express mRuby-CLCa and eGFP-CCDC32(FL). This dual channel imaging was conducted without siRNA-mediated knockdown. White arrows point to colocalized CLCa and CCDC32 clusters. White ROI is magnified on the right. Scale bar = 5µm. (B) Cohort-averaged fluorescence intensity traces of CCPs (marked with mRuby-CLCa) and CCP-enriched eGFP-CCDC32(FL). 51.3±8.2% of analyzed CCPs showed eGFP-CCDC32(FL) recruitment. Number of tracks analyzed: 23699. (C) Lifetime distributions of all CCPs, CCPs with eGFP-CCDC32 recruitment, and CCPs without eGFP-CCDC32 recruitment.

CCDC32 depletion inhibits Transferrin Receptor (TfnR) uptake and CCP maturation.

(A) Immunoblotting (IB) shows efficient CCDC32 knockdown in ARPE-HPV cells by siRNA treatment. (B) Quantified knockdown efficiency (∼60%) of CCDC32 (n=8). (C) Measurements of the uptake efficiency of TfnR (n=8). % of surface bound = Internalized/Surface bound*100%. Error bars in (B-C) indicate standard deviations. (D-E) Representative single frame images from TIRFM videos (7.5 min/video, 1 frame/s, see videos 1 and 2) and corresponding kymographs from region indicated by yellow lines of ARPE-HPV eGFP-CLCa cells treated with (D) control siRNA or (E) CCDC32 siRNA. Scale bars = 5µm. (F-H) Effect of CCDC32 knockdown on the initiation rates of (F) all CCSs and (G) bona fide CCPs, as well as (H) the % of bona fide CCPs. Each dot represents a video (n=11). Statistical analysis of the data in (F-H)) is the Wilcoxon rank sum test, ***, p ≤ 0.001. (I) Lifetime distribution of bona fide CCPs. Data presented were obtained from a single experiment (n = 11 videos for each condition) that is representative of 3 independent repeats. Number of dynamic tracks analyzed: 125897 for siControl and 105313 for siCCDC32. Shadowed area indicates 95% confidential interval.

CCDC32 depletion inhibits CCP invagination.

(A) Scheme of Epi-TIRF microscopy for measuring the invagination of CCPs using primary/subordinate tracking. Δz(t) denotes the invagination depth of CCPs over time. IE and IT denotes the cohort-averaged fluorescence intensity from Epi and TIRF channels, respectively. kE and kT are the initial growth rate for the Epi and TIRF channel signals, respectively. I0is an additive correction factor. ‘*’ indicates the mass center of clathrin coat. h=115nm is the evanescent depth of TIRF field, see more detail in ref.(Saffarian and Kirchhausen, 2008; Wang et al., 2020). (B) Representative Epi and TIRF microscopy images. Scale bars = 10µm. (C) Epi-TIRF microscopy analysis shows that knockdown CCDC32 strongly inhibited CCP invagination. Top: Cohort-averaged CCP fluorescence intensity traces from Epi and TIRF channels; bottom: calculated Δz(t)/h curves. Data presented were obtained from n = 12 videos for each condition. Number of CCP tracks analyzed to obtain the Δz/h curves: 5998 for siControl and 4418 for siCCDC32. Shadowed area indicates 95% confidential interval.

Flat clathrin lattices accumulate in cells depleted of CCDC32.

(A-B) Representative PREM images of ARPE19 cells treated with (A) control siRNA or (B) CCDC32 siRNA showing flat (blue), dome-shaped (green) and spherical (orange) CCSs. Scale bars = 200nm. (C-E) Quantification of the clathrin-coated structures (CCSs) by (C) shape category, (D) the CCS projection area, and (E) the CCS occupancy on the plasma membrane (PM). Each dot in (E) represents an individual fragment of the PM; the number of cell membrane fragments analyzed is 38 for siControl cells and 35 for siCCDC32 cells from two independent experiments, n = number of CCS. Statistical tests were performed using Mann-Whiney test (D) or unpaired t-test (E). For the Box and whisker plots in (D) and (E), the box extends from the 25th to 75th percentiles, the line in the middle of the box is plotted at the median, and the “+” indicates the mean.

CCDC32 interacts with AP2 through the α appendage domain.

(A) AP2 structure. AD: appendage domain. (B-C) Immunoprecipitation (IP) of eGFP from naive ARPE-HPV cells (control IPs) or ARPE-HPV cells that stably express AP2-α-eGFP (AP2-α-eGFP IPs) using anti-GFP beads. (B) The domain structure of AP2-α-eGFP. (C) Volcano plot after mass spectrometry analysis of the IP samples for protein ID and abundance detection. Green: AP2 subunits; orange: known AP2 interactors. Source data for the volcano plot is available as Supplementary Table 2. (D-F) IP of eGFP from ARPE-HPV cells that stably express eGFP or eGFP-CCDC32(FL) using anti-GFP magnetic beads. (D) The domain structures of eGFP and eGFP-CCDC32(FL). (E) Representative immunoblotting result of n=3 IP samples. (F) Relative AP2 enrichments quantified from immunoblotting results. (G-I) GST pull-down assays. (G) Quantification of immunoblots of the relative enrichment of CCDC32. (H) Coomassie blue stained SDS-page gel of purified GST, GST-AP2-α-AD, and GST-AP2-β-AD. (I) Representative western blot result of n=3 GST pull-down assay from ARPE-HPV eGFP-CCDC32(FL) cell lysate using purified GST, GST-AP2-α-AD, or GST-AP2-β-AD. The amount of bait GST-proteins is shown in (D), and the pulled down eGFP-CCDC32 was detected by immunoblotting (IB) of GFP. Error bars in (F) and (I) are SD (n=3).Two-tailed student’s t-test: ***, p ≤ 0.001.

A central α-helix in CCDC32 mediates CCDC32-AP2 interactions.

(A-C) IP of eGFP from ARPE-HPV cells that stably express eGFP-CCDC32(FL) or eGFP-CCDC32(Δ78-98) using anti-GFP beads. (A) The domain structures of eGFP-CCDC32(FL) and eGFP-CCDC32(Δ78-98). (B) Representative Immunoblotting result of n=3 IP samples. (C) Relative AP2 enrichments that were quantified from (B). Error bars indicate standard deviations. (D) Representative TIRFM images of ARPE-HPV cells that stably express mRuby-CLCa and eGFP-CCDC32(Δ78-98). White ROI is magnified on the right. Scale bar = 5µm. (E) Cohort-averaged fluorescence intensity traces of CCPs and CCP-enriched eGFP-CCDC32(Δ78-98). Number of tracks analyzed: 37892. (F-G) (F) TfnR uptake efficiency and (G) CCP% of ARPE mRb-CLCa cells that stably express eGFP-CCDC32(WT), eGFP, or eGFP-CCDC32(Δ78-98), and with siRNA-mediated knockdown of endogenous CCDC32. Each dot in (G) represents a movie. Statistical analysis of the data in (F) is two-tailed student’s t-test: ns, not significant; ***, p ≤ 0.001. Statistical analysis of the data in (G) is the Wilcoxon rank sum test, **, p ≤ 0.01.

Disease-causing nonsense mutation in CCDC32 loses AP2 interaction capacity and inhibits CME.

(A-C) IP of eGFP from ARPE-HPV cells that stably express eGFP-CCDC32(FL) or eGFP-CCDC32(1-54) using anti-GFP magnetic beads. (A) The domain structures of eGFP-CCDC32(FL) and eGFP-CCDC32(1-54). (B) Representative immunoblotting result of n=3 IP samples. (C) Relative AP2 enrichments quantified from immunoblotting results. (D) Representative TIRFM images of ARPE-HPV cells that stably express mRuby-CLCa and eGFP-CCDC32(1-54). White ROI is magnified on the right. Scale bar = 5µm. (E) Cohort-averaged fluorescence intensity traces of CCPs and CCP-enriched eGFP-CCDC32(1-54). Number of tracks analyzed: 28658. (F-G) (F) TfnR uptake efficiency and (G) CCP% of ARPE-HPV cells that stably express eGFP-CCDC32(FL), eGFP, or eGFP-CCDC32(1-54), and with siRNA-mediated knockdown of endogenous CCDC32. Each dot in (G) represents a movie. Statistical analysis of the data in (F) is two-tailed student’s t-test: ***, p ≤ 0.001. Statistical analysis of the data in (G) is the Wilcoxon rank sum test, **, p ≤ 0.01.

Cartoon illustration of CCDC32-AP2 interactions that regulate CME.

Interactions between residues 78-98, corresponding to a central α-helix of CCDC32, and AP2 are essential for recruitment of CCDC32 to CCPs. Depleting CCDC32 (ΔCCDC32) inhibits CCP stabilization and invagination, resulting in enhanced CCS abortion and accumulated flat clathrin assembly intermediates. As a compensation effect, CCS initiation is enhanced and CCP maturation is faster, resulting in only slightly reduced cargo internalization. EAPs: endocytic accessory proteins.

(A) Western blotting indicates similar expression levels of exogenous eGFP-CCDC32(FL), eGFP, eGFP-CCDC32(1-54) and eGFP-CCDC32(Δ78-98) in ARPE-HPV mRuby-CLCa cells. Note that the anti-CCDC32 antibody does not detect the eGFP-CCDC32(Δ78-98) as well as full-length and is unable to detect eGFP-CCDC32(1-54); (B) siCCDC32 knocked down endogenous CCDC32 but not exogenously introduced eGFP-CCDC32(FL) from ARPE-HPV mRuby-CLCa cells.

(A) Representative TIRFM images of ARPE-HPV cells that stably express mRuby-CLCa and EGFP. White ROI is magnified on the right. Scale bar = 5µm. (B) Cohort-averaged fluorescence intensity traces of CCPs and CCP-enriched EGFP. Number of tracks analyzed: 46833.

(A-B) siRNA-mediated knockdown of CCDC32 does not affect AP2 expression level. (A) Representative immunoblotting result of AP2 subunits from n=3 biological repeats. (B) Quantification of results from (A). Error bars indicate standard deviations. Statistical analysis is student’s t-test: ns, not significant. (C) All the four subunits of intact AP2 efficiently co-IP with eGFP-CCDC32(FL).

Fraction of CCPs in lifetime cohorts. Data was obtained from ARPE-HPV eGFP-CLCa cells that were treated with siCCDC32. The obtained data was processed with cmeAnalysis.

CCDC32 knockdown shifts the shapes of clathrin-coated structures.

(A-B) Representative PREM images of ARPE-HPV cells treated with control siRNA (A) or CCDC32 siRNA (B). Scale bars = 200 nm. (C-D) Quantification of the CCS shape category (C) and CCS occupancy on PM (D). Each dot represents an individual fragment of the plasma membrane; the number of plasma membrane fragments analyzed is 38 for siControl and 35 for siCCDC32 from two independent experiments. Statistical tests were performed using either unpaired t-test or Mann-Whiney test depending on normality of distributed values.

The sequence and structure of CCDC32.

(A) Amino acid sequence of CCDC32. Blue font: canonical AP2 binding motifs; red font: predicted coiled-coil domain (aa78-98). (B) AlphaFold 3.0 modelling of full-length CCDC32 (Jumper et al., 2021; Varadi et al., 2021). Residues with pLDDT > 0.8 are labelled in red. (C) co-IP of eGFP-CCDC32 did not pulldown endogenous CCDC32. Red-cycled area was overexposed on the right. (D-E) AlphaFold 3.0 modelling of CCDC32 interaction with AP2-α (gene: AP2A2). For all panels, CCDC32 is colored by pLDDT and AP2-α is grey. (D) CCDC32-AP2A2 complex. CCDC32 residues with pLDDT > 0.7 are labelled in red. (E-F) Close up view of CCDC32-AP2A2 complex near CCDC32 residues (E) 12-28, 37-43 and (F) 66-91. Hydrophobic CCDC32 and AP2-α residues at the interface are labelled in red and black, respectively. Two canonical α-AD binding motifs in CCDC32 (17DLW19 and 39FSDSF43 can be docked on the AP2 appendage domain with high confidence; aa78-98 docks with high confidence to an alpha helix in the α-subunit encoded by aa418-438. Alphafold 3.0 modelling was performed using Alphafold Server (Abramson et al., 2024) and structure images were generated using ChimeraX (Meng et al., 2023).

A collection of representative TIRFM images from the main text showing cells that stably express the same amount of eGFP-CCDC32(FL), eGFP, eGFP-CCDC32(Δ78-98), and eGFP-CCDC32(1-54).

Diffuse PM fluorescence is greatest for eGFP-CCDC32(FL), diminished, but detectable for eGFP-CCDC32(Δ78-98), but absent for eGFP-CCDC32(1-54) and control eGFP. Scale bar = 5µm. Blue dashed line indicates the background PM fluorescence of eGFP.

ID and abundances of key proteins detected in IP samples by Mass Spectroscopy analysis.

#1-#3: 3 independent repeats of control IPs; #4-#6: 3 independent repeats of AP2-α-eGFP IPs.

List of primers.