STED imaging of endogenously tagged ARF GTPases reveals their distinct nanoscale localizations

Abstract

ADP-ribosylation factor (ARF) GTPases are major regulators of cellular membrane homeostasis. High sequence similarity and multiple, possibly redundant functions of the five human ARFs make investigating their function a challenging task. To shed light on the roles of the different Golgi-localized ARF members in membrane trafficking, we generated CRISPR-Cas9 knockins (KIs) of type I (ARF1 and ARF3) and type II ARFs (ARF4 and ARF5) and mapped their nanoscale localization with stimulated emission depletion (STED) super-resolution microscopy. We find ARF1, ARF4, and ARF5 on segregated nanodomains on the cis-Golgi and ER-Golgi intermediate compartments (ERGIC), revealing distinct roles in COPI recruitment on early secretory membranes. Interestingly, ARF4 and ARF5 define Golgi-tethered ERGIC elements decorated by COPI and devoid of ARF1. Differential localization of ARF1 and ARF4 on peripheral ERGICs suggests the presence of functionally different classes of intermediate compartments that could regulate bi-directional transport between the ER and the Golgi. Furthermore, ARF1 and ARF3 localize to segregated nanodomains on the trans-Golgi network (TGN) and are found on TGN-derived post-Golgi tubules, strengthening the idea of distinct roles in post-Golgi sorting. This work provides the first map of the nanoscale organization of human ARF GTPases on cellular membranes and sets the stage to dissect their numerous cellular roles.

Figure 1.

Gene editing with CRISPR-Cas9 highlights the endogenous localization of ARF GTPases. (A–H) ARFs were tagged at their endogenous locus with the self-labeling enzyme Halo (A–D) or 2xALFA tag (E–H). An additional LAP tag linker was added in the case of ARF3 (B). Live cells were stained with the Halo substrate JF646-CA (A, C, and D) and JFX650-CA (B). Fixed cells were immunolabeled with an anti-ALFA primary antibody and ATTO647N-conjugated secondaries. (I) Scatter dot plot with mean and SD represents the quantification of the mean fluorescence intensity at the Golgi area measured in the ARF^EN-2xALFA KI cells as described in Materials and methods. ARF1 n = 22; ARF3 n = 22; ARF4 n = 24; ARF5 n = 20; n = number of total cells from at least three independent experiments. Yellow arrows highlight structures positive for endogenoulsy tagged ARFs. All images were smoothed with a Gaussian filter and background subtracted as described in Materials and methods. Mean fluo. int. Golgi area (A.U.) = mean fluorescence intensities at the Golgi area in arbitrary units. Scale bars are 10 and 2 µm in the cropped images.

Figure S1.

Addition of a LAP tag greatly improves the localization of endogenously tagged ARF3. Endogenously tagged ARF GTPases either with a GS linker (ARF1^EN-Halo and ARF3^EN-Halo) or with a LAP tag linker where GFP was switched with Halo tag (ARF1^EN-LAP-Halo and ARF3^EN-LAP-Halo) labeled with JFX650-CA. While addition of a LAP tag linker improves the membrane and Golgi localization of ARF3, no differences were observed in the localization of ARF1 with or without a LAP tag linker. All images were smoothed with a Gaussian filter as described in Materials and methods. Scale bars are 10 µm.

Figure S2.

HAP1 cells edited at each ARF locus do not present defects in morphology or coats recruitment. ARFs were tagged at their endogenous locus with the self-labeling enzyme Halo in a HAP1 haploid cell line. (A) Live cells were stained with the Halo substrate JFX650-CA. (B) Immunoblot to detect ARF-Halo fusion proteins in HAP1 KI lysate with anti-Halo antibodies and anti-βActin primary antibody as a loading control. (C and D) HAP1 WT and ARF4^EN-Halo KI cells were immunostained with anti-KDELR and secondary antibodies conjugated to Alexa488 (C). Quantification of the normalized mean fluorescence intensity of KDELR at the Golgi area measured in HAP1 WT and ARF4^EN-Halo KI cells as described in Materials and methods. Scatter dot plot shows the mean values (black dots) and SEM for each biological replicate (n = 3). All data points (individual cells) are shown as gray dots (n ≥ 10 cells per replicate). Unpaired, two-tailored t test (ns = non-significant, P > 0.05; D). (E–G) HAP1 WT and ARF^EN-Halo KI cells were immunostained with anti-COPI and anti-clathrin and secondary antibodies conjugated to Alexa488 (E). Quantification of the mean fluorescence intensity of COPI (F) and clathrin (G) at the Golgi area measured in HAP1 WT and ARF^EN-Halo KI cells as described in Materials and methods. Scatter dot plots show the mean values (black dots) and SEM for each biological replicate (n ≥ 3). All data points (individual cells) are shown as gray dots (n ≥ 10 cells per replicate). Ordinary one-way ANOVA versus WT (ns, P > 0.05). (H and I) ARF1^EN-SNAP+ARF4^EN-Halo (H, magenta/green) and ARF4^EN-SNAP+ARF5^EN-Halo (I, magenta/green) double KI HAP1 cells were labeled with JF552-CA and JFX650-BG. Norm. Mean Fluo. Int. Golgi = normalized mean fluorescence intensity at Golgi. All images were background subtracted and smoothed with a Gaussian filter as described in Materials and methods. Scale bars are 10 µm. Source data are available for this figure: SourceData FS2.

Figure 2.

STED reveals the super-resolution localization of known Golgi and vesicular markers. (A–H) HeLa cells were immunostained with antibodies against the Golgi markers indicated in the figure and secondaries labeled with either ATTO647N or AlexaFluor594 to be able to perform dual-color STED experiments. Line profiles in each panel correspond to the dotted boxes in the cropped images. (B) Yellow arrows highlight fenestrations in the Golgi cisterna, while white arrows highlight COPI-positive vesicular structures. (E) White arrow highlights a Golgi-associated ERGIC structure. (I) Graphical summary of the nanoscale localization of Golgi markers. All images were deconvolved, background subtracted, and smoothed with a Gaussian filter as described in Materials and methods. Scale bars are 5 and 1 µm in the cropped images.

Figure S3.

Mapping of Golgi markers with STED microscopy in nocodazole-treated cells. (A–G) HeLa cells were treated with nocodazole (33 µM) for 3 h, fixed and immunostained as indicated in the figure. Secondary antibodies labeled either with ATTO647N or AlexaFluor594 were used to be able to perform dual-color STED experiments. (H and I) Scatter dot plots with mean and SD represent the quantification of the distances from the edge of the cis-labeled (H, GM130) or trans-labeled (I, Golgin97) cisternae to the center of COPI and clathrin vesicles measured in the nocodazole-treated fixed cells as described in Materials and methods. GM130: clathrin n = 35; COPI n = 38 (H). Golgin97: clathrin n = 52; COPI n = 56 (I). n = number of ministacks analyzed from at least three independent experiments. All images were background subtracted and smoothed with a Gaussian filter as described in Materials and methods. Scale bars are 1 µm.

Figure 3.

ARFs segregate on different early secretory membranes. (A–I) ARF1^EN-, ARF4^EN-, and ARF5^EN-2xALFA KI HeLa cells were immunostained with anti-ALFA (magenta) and either anti-ERGIC53 (A–C, green) or anti-GM130 (D–F, green) and secondary antibodies labeled with either ATTO647N or AlexaFluor594 to be able to perform dual-color STED experiments. Line profiles in each panel correspond to the dotted boxes in the cropped images. (G–I) ARF1^EN-, ARF4^EN-, and ARF5^EN-2xALFA KI cells were treated with nocodazole (33 µM) for 3 h and immunostained as described above. (J) Scatter dot plot with mean and SD represents the quantification of the distances from the edge of the ARF-labeled cisternae and cis-Golgi cisternae measured in the nocodazole-treated fixed cells as described in Materials and methods. ARF1 n = 9; ARF3 n = 11; ARF4 n = 15; ARF5 n = 14; Golgin97 n = 17. n = number of ministacks from at least three independent experiments. (K) Graphical summary of the nanoscale localization of ARFs on early secretory membranes. Images were deconvolved, background subtracted (A–F), and smoothed with a Gaussian filter (A–I) as described in Materials and methods. Scale bars are 5 and 1 µm in the cropped images.

Figure S4.

STED microscopy reveals ARF3 clusters segregated from ERGIC and cis-Golgi cisternae. (A and B) ARF3^EN-2xALFA (magenta) KI HeLa cells were fixed and immunostained with anti-ALFA tag and either anti-ERGIC53 (A, green) or anti-GM130 (B, green). Secondary antibodies labeled with either ATTO647N or AlexaFluor594 were used to be able to perform dual-color STED experiments. (C) ARF3^EN-2xALFA KI HeLa cells were treated with nocodazole (33 µM) for 3 h and immunostained with the same primary antibodies described above. Secondary antibodies labeled with either ATTO647N or AlexaFluor594 (GM130)/AlexaFluor568 (ERGIC53) were used. Images were deconvolved, background subtracted (A and B), and smoothed with a Gaussian filter (A–C) as described in Materials and methods. Line profiles in each panel correspond to the dotted boxes in the cropped images. Scale bars are 5 and 1 µm in the cropped images.

Figure S5.

STED microscopy of double-edited ARF1-ARF4 and ARF4-ARF5 cells suggests distinct functions. (A and B) ARF1^EN-2xV5/ARF4^EN-2xALFA (A, green/magenta) and ARF4^EN-2xV5/ARF5^EN-2xALFA (B, green/magenta) double KI HeLa cells were fixed and immunostained with anti-ALFA and anti-V5 primary antibodies and secondary antibodies labeled with either ATTO647N or AlexaFluor594 to be able to perform dual-color STED experiments. (C) ARF1^EN-2xALFA KI HeLa cells were fixed and immunostained with a single anti-ALFA and secondary antibodies labeled with ATTO647N (magenta) and AlexaFluor594 (green). Line profiles in each panel correspond to the dotted boxes in the cropped images. All images were deconvolved, background subtracted, and smoothed with a Gaussian filter as described in Materials and methods. Scale bars are 5 and 1 µm in the cropped images (A and B) or 10 and 2 µm in the cropped images (C).

Figure S6.

Live-cell microscopy of tubular-vesicular ERGIC elements and of control cells. (A) βCOP^EN-Halo (green) KI HeLa cells were transfected with a plasmid encoding for SNAP-ERGIC53 (magenta). Live cells were stained with the Halo substrate JF571-CA together with the SNAP substrate JFX650-BG to perform dual-color live-cell STED. Yellow arrows highlight COPI-positive clusters associated with tubular-vesicular ERGIC elements highlighted by white arrows. (B and C) ARF1^EN-Halo (B, green) and ARF4^EN-Halo (C, green) KI HeLa cells were transfected with a plasmid encoding for SNAP-ERGIC53 (magenta) and labeled with the Halo substrate JF552-CA and the SNAP substrate JFX650-BG. White arrows highlight peripheral ERGICs without any ARF, while yellow arrows highlight structures in which ARFs are co-localizing with peripheral ERGICs. (D) ARF4^EN-Halo (cyan) and ARF1^EN-SNAP (magenta) double KI HeLa cells were transfected with a plasmid encoding for YFP-ERGIC53 (yellow) and labeled with the Halo substrate JF552-CA and the SNAP substrate JFX650-BG. White arrows highlight peripheral ERGICs positive only for ARF4, while yellow arrows highlight peripheral ERGICs positive for both ARF1 and ARF4. (E and F) ARF1^EN-Halo KI HeLa cells were stained with the two Halo substrates JF571-CA (green) and JFX650-CA (magenta) used for-dual color live-cell imaging as a control for chromatic aberrations (E). Line profiles in each panel correspond to the dotted boxes in the cropped images. Scatter dot plot with mean and SD represents the quantification of the distances between the two channels in control ARF1^EN-Halo cells stained with both dyes (negative control) or between ARF1 and ARF4 nanodomains on peripheral ERGICs (ARF1 vs. ARF4). Examples are shown in Fig. 4 E. ARF1 vs. ARF4 n = 71; negative control n = 56 (F). n = number of ERGICs from at least three independent experiments. Images were deconvolved, background subtracted (A and E), and smoothed with a Gaussian filter (A–E) as described in Materials and methods. Scale bars are: 5 and 1 µm in the cropped images (A); 10 and 2 µm in the cropped images (B–D); 10 µm and 500 nm in the cropped images (E).

Figure 4.

ARF1 and ARF4 define different populations of ERGICs. (A and B) ARF1^EN-Halo and ARF4^EN-Halo KI HeLa cells were transfected with a plasmid encoding for SNAP-ERGIC53. Live cells were stained with the Halo substrate JF571-CA and the SNAP substrate JFX650-BG and imaged with a STED microscope. (A ⅰ–ⅲ and B ⅰ–ⅲ) White arrows indicate Golgi-associated ERGICs devoid of ARF1 (A i and ii) and positive for ARF4 (B i and ii) and peripheral ERGICs positive for either ARF1 (A iii) or ARF4 (B iii). (C) ARF1^EN-SNAP and ARF4^EN-Halo double KI HeLa cells were labeled with JF571-CA and JFX650-BG and imaged with a STED microscope. (C ⅰ–ⅲ) Overview of Golgi area showing ARF4 Golgi-associated ERGICs devoid of ARF1 (C i, white arrow) and ARF1/ARF4 segregated tubular-vesicular structures (C ii and iii, yellow and white arrows, respectively). (D) Distribution of peripheral ERGIC populations defined by ARF1 and ARF4. The percentage of peripheral ERGICs positive for either ARF1 or ARF4 was calculated in images of either ARF1^EN or ARF4^EN-Halo KI cells transfected with a plasmid encoding for SNAP-ERGIC53. The percentage of ARF1- and ARF4-positive peripheral ERGICs was calculated in images of ARF1^EN-SNAP and ARF4^EN-Halo double KI cells transfected with a plasmid encoding for YFP-ERGIC53. ARF1+ARF4/ERGIC n = 13; ARF1/ERGIC n = 10; ARF4/ERGIC n = 10. n = number of cells from at least three independent experiments. Error bars represent mean and SD. (E) Gallery of peripheral ERGICs showing ARF1/ARF4 segregated nanodomains. (F) Graphical summary of the nanoscale localization of ARF1 and ARF4 on ERGICs. All images were deconvolved, background subtracted, and smoothed as described in Materials and methods. The brightness in the crops was enhanced to highlight the dim peripheral structures. Scale bars are 5 and 1 µm in the cropped images.

Figure 5.

Live-cell STED reveals segregated ARF1 and ARF4 subpopulations of tubular-vesicular structures defined by COPI machinery. (A and B) ARF1^EN/ARF4^EN-Halo (green) and βCOP^EN-SNAP (magenta) double KI HeLa cells were labeled with JF571-CA and JFX650-BG and imaged with a STED microscope. ARF1 and ARF4 tubular-vesicular structures are highlighted by white arrows and COPI clusters by yellow arrows. (C) Double KI cell lines ARF1^EN-, ARF3^EN-, ARF4^EN-, and ARF5^EN-Halo (magenta) and βCOP^EN-SNAP (green) were labeled with JFX650-CA and JF585-BG and subsequently treated with nocodazole (33 µM) for 3 h. (D) Scatter dot plot with mean and SD represents the quantification of the distances from the edge of the ARF-labeled cisternae to the center of COPI vesicles measured in nocodazole-treated live cells as described in Materials and methods. ARF1 n = 62; ARF3 n = 49; ARF4 n = 49; ARF5 n = 48. n = COPI vesicles from at least three independent experiments. All images were deconvolved, background subtracted, and smoothed as described in Materials and methods. Scale bars are 5 and 1 µm in the cropped images.

Figure S7.

COPI-positive clusters are observed near all ARFs. (A–D) Fixed ARF1^EN-, ARF3^EN-, ARF4^EN-, and ARF5^EN-2xALFA KI HeLa cells were immunostained with anti-ALFA (magenta) and anti-β′COP (CM1, green) and secondary antibodies labeled with either ATTO647N or AlexaFluor594 to be able to perform dual-color STED experiments. Line profiles in each panel correspond to the dotted boxes in the cropped images. All images were deconvolved, background subtracted, and smoothed with a Gaussian filter as described in Materials and methods. Scale bars are 5 and 1 µm in the cropped images.

Figure 6.

ARF1 and ARF3 are the sole TGN-localized ARF members. (A–E) ARF1^EN-, ARF3^EN-, ARF4^EN-, and ARF5^EN-2xALFA KI HeLa cells were immunostained with anti-ALFA tag (magenta) and anti-Golgin97 (green) and secondary antibodies labeled with either ATTO647N or AlexaFluor594 to be able to perform dual-color STED experiments. (E) ARF1^EN-, ARF3^EN-, ARF4^EN-, and ARF5^EN-2xALFA KI HeLa cells were treated with nocodazole (33 µM) for 3 h and immunostained as described above. (F) Scatter dot plot with mean and SD represents the quantification of the distances from the edge of the ARF-labeled cisternae and Golgin97-labeled TGN cisternae measured in nocodazole-treated fixed cells as described in Materials and methods. ARF1 n = 10; ARF3 n = 12; ARF4 n = 10; ARF5 n = 12; GM130 n = 17. n = number of ministacks from at least three independent experiments. Images were deconvolved, background subtracted (A–D), and smoothed with a Gaussian filter (A–E) as described in Materials and methods. Scale bars are 5 and 1 µm in the cropped images.

Figure 7.

Type I ARFs define tubular-vesicular clathrin-positive structures on the TGN and the cell periphery. (A–D) ARF1^EN-, ARF3^EN-, ARF4^EN-, and ARF5^EN-Halo (magenta) and SNAP-CLCa^EN (green) double KI HeLa cells were labeled with JF571-CA together with JFX650-BG (A and C) or JFX650-CA and JF585-BG (B and D) and imaged with a STED microscope. ARFs tubular-vesicular structures are highlighted by white arrows and clathrin clusters by yellow arrows. (E) Double KI cell lines were labeled with JFX650-CA and JF585-BG and then treated with nocodazole (33 µM) for 3 h. (F) Scatter dot plot with mean and SD represents the quantification of the distances from the edge of the ARF-labeled cisternae to the center of clathrin vesicles measured in nocodazole-treated live cells as described in Materials and methods. ARF1 n = 33; ARF3 n = 30; ARF4 n = 19; ARF5 n = 19. n = clathrin vesicles from at least three independent experiments. All images were deconvolved, background subtracted, and smoothed with a Gaussian filter as described in Materials and methods. Scale bars are 5 and 1 µm in the cropped images.

Figure 8.

ARF1 and ARF3 define segregated nanodomains on the TGN and localize to TGN-derived tubular-vesicular trafficking intermediates. (A) ARF1^EN-Halo and ARF3^EN-LAP-SNAP double KI HeLa cells were labeled with JF571-CA and JFX650-BG. (A ⅰ–ⅳ) Crops highlight segregation of gene-edited ARF1 and ARF3 on the Golgi (A i and ii) and their localization on TGN-derived (A iii) and peripheral (A iv) tubular-vesicular trafficking intermediates. ARF3 nanodomains are highlighted by white arrows in the crops or a black arrow in the line profile graph (A i and ii). ARF1- and ARF3-positive tubular-vesicular trafficking intermediates are highlighted by white arrows (A iii and iv). (B) Graphical summary of the nanoscale localization of ARF1 and ARF3. Images were deconvolved, background subtracted, and smoothed with a Gaussian filter as described in Materials and methods. The brightness in the crops was enhanced to highlight the dim distal structures. Scale bars are 5 and 1 µm in the cropped images.

Figure 9.

Model proposing distinct functions of different ARF paralogs in ER-Golgi transport and TGN export.