The kinesin KIF16B mediates apical transcytosis of transferrin receptor in AP-1B-deficient epithelia

Abstract

Polarized epithelial cells take up nutrients from the blood through receptors that are endocytosed and recycle back to the basolateral plasma membrane (PM) utilizing the epithelial-specific clathrin adaptor AP-1B. Some native epithelia lack AP-1B and therefore recycle cognate basolateral receptors to the apical PM, where they carry out important functions for the host organ. Here, we report a novel transcytotic pathway employed by AP-1B-deficient epithelia to relocate AP-1B cargo, such as transferrin receptor (TfR), to the apical PM. Lack of AP-1B inhibited basolateral recycling of TfR from common recycling endosomes (CRE), the site of function of AP-1B, and promoted its transfer to apical recycling endosomes (ARE) mediated by the plus-end kinesin KIF16B and non-centrosomal microtubules, and its delivery to the apical membrane mediated by the small GTPase rab11a. Hence, our experiments suggest that the apical recycling pathway of epithelial cells is functionally equivalent to the rab11a-dependent TfR recycling pathway of non-polarized cells. They define a transcytotic pathway important for the physiology of native AP-1B-deficient epithelia and report the first microtubule motor involved in transcytosis.

Figure 1

AP-1B KD MDCK cells display microtubule- and rab11a-dependent transcytosis of basolateral TfR. (A) Model of a WT MDCK cell displaying endosomal compartments and the endosomal itinerary of pIgR, TfR and LDLR postendocytic pathways. (B) A SulfoTag-Tf assay. Tf labelled with biotin and the luminophore SulfoTag (SulfoTag-Tf) provides a signal 10 × higher than ¹²⁵I-Tf, allowing to detect 2 × 10⁻¹⁵ mol per 12 mm filter (the estimated amount of endogenous dog TfR present at the apical membrane of confluent MDCK cells). (C) MT mediate apical transcytosis of TfR. Values of apical transcytosis, basolateral recycling and the intracellular pool of SulfoTag-Tf (60 min) in control WT, nocodazole-treated WT, control AP-1B KD and nocodazole-treated AP-1B KD MDCK cells. (D) Dominant-negative rab11a inhibits apical transcytosis of TfR. A MDCK cell line was generated that stably expresses dominant-negative rab11a fused to monomeric Cherry fluorescent protein (mCh-DN-rab11a) under the control of a tetracycline-repressible promotor. Western blot analysis showed that expression of mCh-DN-rab11a increased four-fold by removing doxycycline for 12 h. (E) RNA levels of μ1B and GAPDH in WT, stable AP-1B KD, transient luciferase KD and transient AP-1B KD MDCK cells. (F) Values of apical transcytosis of SulfoTag-Tf (75 min) in WT and transient AP-1B KD MDCK cells with or without 12-h expression of mCh-DN-rab11a. *P<0.05, **P<0.001. Bars represent mean±standard error. Ap, apical; BL, basolateral; IC, intracellular.

Figure 2

Microtubules mediate transport of basolateral TfR to both ARE and ASE in AP-1B KD MDCK cells. (A) MDCK cells were incubated from the basolateral surface with fluorescent 594-Tf for 15 min and immunostained with anti-rab11a. From left to right, columns show: control WT, nocodazole-treated WT, control AP-1B KD and nocodazole-treated AP-1B KD MDCK cells. (A′) Cells from experiments represented in (a) were quantified for the percentage of pixels of rab11a colocalizing with 594-Tf (top) and the percentage of pixels of 594-Tf colocalizing with rab11a (bottom). Circles correspond to individual cells obtained from different experiments and red lines indicate the median value. (B) MDCK cells were incubated with fluorescent 594-Tf as in (a) and stained for ASE with incubation of 488-WGA from the apical surface for 5 min at 37°C and washed with NADG at 4°C. (B′) Cells from experiments represented in (b) were quantified for the percentage of pixels of 488-WGA colocalizing with 594-Tf (top) and the percentage of pixels of 594-Tf colocalizing with 488-WGA (bottom). **P<0.001. Red line represents the median. Scale bar, 10 μm.

Figure 3

LDLR undergoes apical transcytosis to ARE in AP-1B KD MDCK. MDCK cells were transiently transfected with TfR–GFP (A) LDLR (B) or NBC1 (C) and immunostained for rab11a. (A′, B′, C′) Cells from experiments represented in (a), (b) and (c) were quantified for the percentage of pixels of rab11a colocalizing with the cargo (left) and the percentage of pixels of the cargo colocalizing with rab11a (right). NS, no significance. **P<0.001. Red line represents the median. Scale bar, 10 μm.

Figure 4

Transport of basolateral TfR to ARE requires the plus-end microtubule motor KIF16B. (A, B) WT and AP-1B KD MDCK cells transiently transfected with DN-KIF16B-YFP (a) or DN-KIF5B-CFP (b) were incubated from the basolateral surface with fluorescent 594-Tf for 15 min and immunostained with anti-rab11a. Transfected and non-transfected cells can be identified in the bottom panel by the signal of DN-KIF16B-YFP or DN-KIF5B-CFP. (C) Cells from experiments represented in (a) and (b) were quantified for the percentage of pixels of rab11a colocalizing with 594-Tf (left) and the percentage of pixels of 594-Tf colocalizing with rab11a (right). (D) WT and AP-1B KD MDCK cells nucleofected with either luciferase or KIF16B siRNA were incubated from the basolateral surface with fluorescent 594-Tf for 15 min and immunostained with anti-rab11a. (D′) Cells from experiments represented in (d) were quantified for the percentage of pixels of rab11a colocalizing with 594-Tf (left) and the percentage of pixels of 594-Tf colocalizing with rab11a (right). (D′′) RNA levels of KIF16B and GAPDH in AP-1B KD cells nucleofected with luciferase or KIF16B siRNA. NS, no significance, *P<0.05, ** P<0.001. Red line represents the median. Scale bar, 10 μm.

Figure 5

KIF16B mediates trafficking of TfR from CRE to ARE. WT MDCK cells transiently transfected with the WT-KIF16B-YFP (full-length KIF16B) (A) or the DN-KIF16B-YFP (motorless KIF16B) (B) were incubated for 25 min from the basolateral surface with 633-Tf (green) followed by a 5 min chase, in which 594-Tf (red) was added to the basolateral medium. The left panel displays the signal of 594-Tf and 633-Tf and the right panel displays the signal of KIF16B-YFP from the same field. Arrows emphasize endosomes containing KIF16B and 633-Tf, but not 594-Tf. (A′, B′) Cells from experiments represented in A and B were quantified for the percentage of pixels of the kinesin colocalizing with 633-Tf or 594-Tf (top) and the percentage of pixels of 633-Tf or 594-Tf colocalizing with the kinesin (bottom) (C, C′, D, D′). The same experiment described in A, A′, B and B′ was carried out in AP-1B KD MDCK cells. (E) WT MDCK cells were treated with nocodazole/cold at 4°C, incubated at 37°C in the absence of the drug for MTs regrowth (see Methods), fixed and immunostained with markers of MTs (α-tubulin), golgi (giantin) and centrioles (γ-tubulin). *P<0.05, **P<0.001. Red line represents the median. Scale bar, 10 μm.

Figure 6

AP-1B-deficient retinal pigment epithelium displays apical transcytosis of TfR. (A) Values of apical transcytosis, basolateral recycling and the intracellular pool of SulfoTag-Tf after 60 min of recycling in WT MDCK and ARPE-19 cells. (B) WT MDCK or ARPE-19 cells were incubated from the basolateral side with fluorescent 594-Tf for 30 min and immunostained with anti-rab11a. (B′) Cells from experiments represented in (b) were quantified for the percentage of pixels of rab11a colocalizing with 594-Tf (left) and the percentage of pixels of 594-Tf colocalizing with rab11a (right). (C) WT MDCK or ARPE-19 cells were incubated with fluorescent 594-Tf as in (b) and stained for ASE with incubation of 488-WGA from the apical surface for 5 min at 37°C and washed with NADG at 4°C. (C′) Cells from experiments represented in C were quantified for the percentage of pixels of 488-WGA colocalizing with 594-Tf (left) and the percentage of pixels of 594-Tf colocalizing with 488-WGA (right). (D) ARPE-19 fully polarized in Transwell filters were electroporated with DN-KIF16B-YFP (Deora et al, 2007), incubated from the basolateral surface with fluorescent 594-Tf for 15 min and immunostained with anti-rab11a. Transfected and non-transfected cells can be identified in the bottom panel by the signal of DN-KIF16B-YFP. (D′) Cells from experiments represented in (d) were quantified for the percentage of pixels of rab11a colocalizing with the 594-Tf (left) and the percentage of pixels of 594-Tf colocalizing with rab11a (right). Bars represent mean±standard error.*P<0.05, **P<0.001. Red line represents the median. Scale bar, 10 μm.

Figure 7

Expression of AP-1B in fibroblastic CHO cells generates a rab11a-independent recycling route. (A) WT and AP-1B KD subconfluent MDCK were incubated with fluorescent (dog) 594-Tf for 30 min and immunostained for rab11a. (A′) Quantification of the percentage of pixels of rab11a colocalizing with 594-Tf (left) and the percentage of pixels of 594-Tf colocalizing with rab11a (right). (B) WT and AP-1B (+) CHO cells were incubated with fluorescent (human) 594-Tf for 30 min and immunostained with anti-rab11a and anti-HA (μ1B-HA). (B′) Quantification of the pixels of rab11a occupied by 594-Tf (left) and the pixels of 594-Tf occupied by rab11a (right). (C) Western blot analysis showing expression of stably transfected μ1B-HA and/or transiently transfected mCh-DN-rab11a in CHO cells. (C′) TfR recycling assay: after 60 min uptake of SulfoTag-Tf, cell were either allowed to recycle for 56 min or lysed immediately afterwards (0-min recycling) (see Methods). Retention was calculated as ‘retention after 56-min recycling/retention after 0-min recycling’; and normalized to ‘control WT CHO’. Bars represent mean±standard error. NS, no significance, *P<0.05, **P<0.001. Red line represents the median. Scale bar, 10 μm.

Figure 8

Model. (A) WT MDCK cells recycle most TfR to the basolateral membrane through BSE and CRE. In contrast, in AP-1B KD MDCK cells a substantial fraction of TfR is trafficked from CRE to ARE by KIF16B and non-centrosomal MTs and delivered to the apical PM in a rab11a-mediated manner. (B) WT CHO cells recycle TfR through perinuclear RE in a rab11a-mediated manner. In contrast, CHO cells stably transfected with AP-1B generate a rab11a-independent route, likely mediated by AP-1B.