Specificity, promiscuity and localization of ARF protein interactions with NCS-1 and phosphatidylinositol-4 kinase-III beta

Abstract

ADP-ribosylation factor (ARF) proteins are involved in multiple intracellular vesicular transport pathways. Most studies have focused on the functions of ARF1 or ARF6 and little is known about the remaining ARF isoforms. Although the mammalian ARF proteins share a high degree of sequence identity, recent evidence has indicated that they may control distinct trafficking steps within cells. A unanswered issue is the degree of specificity of ARF family members for different interacting proteins. To investigate potential functional differences between the human ARF proteins, we have examined the localization of all human ARF isoforms and their interactions with two ARF1 binding proteins, neuronal calcium sensor-1 (NCS-1) and phosphatidylinositol-4 kinase-IIIbeta (PI4Kbeta). Use of a fluorescent protein fragment complementation method showed direct interactions between ARFs 1, 3, 5 and 6 with NCS-1 but at different intracellular locations in live HeLa cells. Photobleaching experiments indicated that complementation did not detect dynamic changes in protein interactions over short-time scales. A more specific interaction between ARFs 1/3 and PI4Kbeta was observed. Consistent with these latter findings ARF1 but not ARF5 or 6 enhanced the stimulatory effect of PI4Kbeta on regulated exocytosis, suggesting a specific role for class-I ARFs in the regulation of PI4Kbeta.

Figure 1

Sub-cellular localisation of human ARF proteins and NCS-1 in live HeLa cells. Representative images of HeLa cells transfected with Human ARF 1, 3, 4, 5 and 6 proteins tagged with full length EYFP and NCS-1 with full length ECFP. Protein distributions were analysed 24 hrs post transfection by confocal microscopy. Scale bar, 10 μm.

Figure 2

Co-localisation of NCS-1-ECFP with ARFs 1, 3, 4, 5 and 6-EYFP in live HeLa cells. Representative images are shown of HeLa cells co-expressing NCS-1/ARF1 (a), NCS-1/ARF3 (b), NCS-1/ARF4 (c), NCS-1/ARF5 (d) and NCS-1/ARF6 (e). Protein distributions were determined using confocal microscopy 24 hrs post-transfection and extent of co-localisation shown by overlaying EYFP (Green) and ECFP (Red) channels from the same cell. Regions of co-localisation appear yellow in overlay images. Scale bar, 10 μm.

Figure 3

Fluorescent protein fragment reconstitution analysis to examine direct interaction between NCS-1 and ARF1 in live HeLa cells. HeLa cells were transfected with a ECFP-tagged Golgi marker along with both empty split-EYFP cloning vectors (a), an ARF1-YC fragment construct alone (b), an NCS-1 fragment construct alone (c), NCS-1-YN and ARF1-YC in combination (d) and a construct encoding the myristoylation deficient G2A mutant of NCS-1 (e). ECFP (Red) and EYFP (Green) channels from individual cells were combined to show extent of co-localisation of reconstituted EYFP signal with the Golgi marker (yellow regions in overlay images). Scale bar, 10 μm.

Figure 4

Characterisation of the reconstituted NCS-1-YN/ ARF1-YC EYFP fluorescence on the Golgi complex. (a) HeLa cells co-expressing NCS-1-YN/ ARF1-YC were examined and high resolution images of the reconstituted signal in cells co-transfected with Golgi-ECFP are shown. Expanded images of the boxed Golgi complex regions are shown in the inserts. The scale bars represent 5 μm or 1 μm in the insert. (b) HeLa cells were transfected with ARF1-EYFP and NCS-1-ECFP of NCS-1-YN and ARF1-YC EYFP. To determine the dynamics of ARF1-EYFP and the reconstituted NCS-1-YN/ ARF1-YC EYFP fluorescence, cells were imaged and the Golgi complex was photobleached using local high intensity laser light (at the time indicated by the arrow). Fluorescence intensity in the region of interest over the Golgi complex was monitored before and after photobleaching. The data were normalised to the initial fluorescence intensity for each cell and the recovery of fluorescence over the Golgi was monitored over time. The data are shown as mean ± SEM for 13 (ARF1-EYFP) or 14 (NCS-1-YN/ ARF1-YC) cells. The data for ARF1-EYFP was fitted to a double hyperbolic function (blue line) suggesting two components of recovery with half times of around 1 and 10s.

Figure 5

Specificity of human ARF proteins for binding to NCS-1 in live HeLa cells as determined by fluorescent protein fragment reconstitution and determination of protein expression levels by western blotting. An NCS-1-YN fluorescent protein fragment construct was co-transfected into HeLa cells with ARF fragment constructs encoding ARF1 (a), ARF3 (b), ARF4 (c), ARF5 (d) or ARF6 (e). An ECFP-Golgi marker construct was simultaneously co-transfected. ECFP (Red) and EYFP (Green) channels from individual cells were combined to show extent of co-localisation of reconstituted EYFP signal with the Golgi marker (yellow regions in overlay images). Scale bar, 10 μm. Determination of fluorescent fragment protein expression levels by Western blotting (f).

Figure 6

Use of fluorescence fragment complementation to examine interaction of PI4Kβ with ARF1 in live HeLa cells. HeLa cells were transfected with PI4Kβ tagged to full length ECFP (a) and protein localisation determined by confocal microscopy. PI4Kβ-ECFP was co-expressed with ARF1-EYFP (b). The fluorescent fragment variant of PI4Kβ (PI4Kβ-YN) was used to co-transfect HeLa cells along with ARF1-YC (c) and ells were simultaneously co-transfected with an ECFP-tagged Golgi marker. Expanded images of the Golgi complex region of one cell are shown in (d). ECFP (Red) and EYFP (Green) channels from individual cells were combined to show extent of co-localisation (yellow regions in overlay images). Scale bar, 10 μm (a-c) or 5 μm (d).

Figure 7

Use of fluorescent fragment complementation and RNA interference to probe the specificity of human ARF isoform interactions with PI4Kβ in HeLa cells. HeLa cells were co-transfected with PI4Kβ-YN and ARF1-YC (a), ARF3-YC (b), ARF4-YC (c), ARF5-YC (d) or ARF6-YC (e) along with an ECFP tagged Golgi marker. ECFP (Red) and EYFP (Green) channels from individual cells were combined to show extent of co-localisation (yellow regions in overlay images). Scale bar, 10 μm. Determination of protein expression levels of PI4Kβ and ARF isoforms fragment constructs was done by Western blotting (f). To examine the effect of pair-wise knock-down of ARF isoforms, HeLa cells were transfected to express PI4Kβ-ECFP alone as control (g) or alongside RNA interference constructs for ARF1 and 3 or ARF 4 and 5 as indicated.

Figure 8

Analysis of the effects of PI4Kβ and ARF isoforms in PC12 cells. PC12 cells were co-transfected to express growth hormone along with the indicated constructs. Cells co-transfected with YFP-YN were used as controls. Growth hormone released with measured after a 15 min incubation in the absence or presence of 300 μM ATP. The data for release growth hormone were calculated as a percentage of total cellular growth hormone and release elicited by ATP over basal release is shown. Growth hormone release was significantly increased by co-transfection with PI4Kβ and this was enhanced by ARF1 but not ARF5 or ARF6. (n= 20 for all conditions except ARF1-YC where n=6. All data are shown as mean ± SEM and statistical comparisons were carried out using an unpaired t-test.)