Sterols and sphingolipids differentially function in trafficking of the Arabidopsis ABCB19 auxin transporter


  • The authors declare that they have no conflicts of interest.

For correspondence (e-mail


The Arabidopsis ATP-binding cassette B19 (ABCB19, P-glycoprotein19) transporter functions coordinately with ABCB1 and PIN1 to motivate long-distance transport of the phytohormone auxin from the shoot to root apex. ABCB19 exhibits a predominantly apolar plasma membrane (PM) localization and stabilizes PIN1 when the two proteins co-occur. Biochemical evidence associates ABCB19 and PIN1 with sterol- and sphingolipid-enriched PM fractions. Mutants deficient in structural sterols and sphingolipids exhibit similarity to abcb19 mutants. Sphingolipid-defective tsc10a mutants and, to a lesser extent, sterol-deficient cvp1 mutants phenocopy abcb19 mutants. Live imaging studies show that sterols function in trafficking of ABCB19 from the trans-Golgi network to the PM. Pharmacological or genetic sphingolipid depletion has an even greater impact on ABCB19 PM targeting and interferes with ABCB19 trafficking from the Golgi. Our results also show that sphingolipids function in trafficking associated with compartments marked by the VTI12 syntaxin, and that ABCB19 mediates PIN1 stability in sphingolipid-containing membranes. The TWD1/FKBP42 co-chaperone immunophilin is required for exit of ABCB19 from the ER, but ABCB19 interactions with sterols, sphingolipids and PIN1 are spatially distinct from FKBP42 activity at the ER. The accessibility of this system to direct live imaging and biochemical analysis makes it ideal for the modeling and analysis of sterol and sphingolipid regulation of ABCB/P-glycoprotein transporters.


Polar streams of the phytohormone auxin regulate organ development, elongation, shoot/root branching and plastic growth responses to light, gravity and touch in plants (Zazímalová et al., 2010; Peer et al., 2011). Polar auxin transport is motivated by chemiosmotic gradients (Raven, 1975). At apoplastic pH (5.5), the principal auxin, indole-3-acetic acid (IAA, pKa = 4.75), enters cells via lipophilic diffusion and anionic uptake mediated by AUXIN RESISTANT 1 and LIKE AUX 1 (AUX1 and LAX) uptake permeases (TC 2.A.18) (Bennett et al., 1996). At cytosolic neutral pH, IAA is anionic and requires facilitators or transporters to exit cells. Full-length PIN-FORMED (PIN) carrier proteins (TC 2.A.69) mediate the tissue-specific directional cellular auxin export required for organogenesis and polar growth, and a subset of ATP binding cassette subfamily B (ABCB) transporters (TC 3.A.1) function coordinately with PINs to mobilize long-distance auxin transport streams (Zazímalová et al. 2010). At the cellular level, ABCBs function in both auxin export and exclusion from the plasma membrane (PM; Yang and Murphy, 2009) to regulate cellular auxin levels (Kubeš et al., 2011). The best-characterized ABCB auxin transporter is ABCB19, which is a primary component of rootward auxin transport along the embryonic axis, interacts directly with PIN1 and is directly regulated by the PHOTOTROPIN 1 photoreceptor in the first step of phototropic bending (Blakeslee et al., 2007; Titapiwatanakun et al., 2009; Christie et al., 2011).

Cholesterol has been shown to activate the mammalian P-glycoprotein ABCB1 transporter (Belli et al. 2009). Live imaging indicates that sterols, saturated phospholipids and sphingolipids coalesce into membrane domains together with a subset of membrane proteins at the trans-Golgi network (TGN) for sorting to the yeast and plant PM (Klemm et al., 2009; Markham et al., 2011). In Arabidopsis, sphingolipids and sterols have been shown to contribute to trafficking of PIN and AUX1 in their respective membrane domains (Carland et al., 2002; Willemsen et al., 2003; Men et al., 2008; Pan et al., 2009; Roudier et al., 2010; Markham et al., 2011). ABCB19 has been shown to stabilize PIN1 at the PM, has been isolated in sterol- or sphingolipid-enriched detergent-resistant membranes and is required for PIN1 retention in those membranes (Blakeslee et al., 2007; Titapiwatanakun et al., 2009). Furthermore, ABCB19 auxin transport activity in heterologous systems is enhanced by structural sterol enrichment, similar to that observed in animal cells (Belli et al., 2009; Titapiwatanakun et al., 2009). However, biochemical evidence associating ABCB19 with detergent-resistant membranes is not sufficient to demonstrate localization in discrete membrane subdomains (Tanner et al., 2011). The work presented herein demonstrates the dependence of ABCB19 trafficking and membrane stability on sterols and sphingolipids, and, in a broader perspective, establishes a model of where and when lipid interactions regulate trafficking of the large number of plant ABC transporters to the PM.


Sterol and sphingolipid biosynthetic mutants show phenotypes similar to abcb19

If ABCB19 localization and activity depend on sphingolipids and structural sterols, there should be phenotypic similarities between abcb19 and mutants with altered structural sterols and sphingolipids. The recently characterized tsc10a-2 sphingolipid-deficient mutant was chosen for analyses, as tsc10a-2 retains only 10% of wild-type 3-ketosphinganine reductase activity, with the result that free long chain fatty acid bases accumulate and glucosylceramide (GluCer) sphingolipids are reduced (Chao et al., 2011). Simplified diagrams of Arabidopsis sphingolipid and sterol biosynthetic pathways are shown in Figure S1. The sterol biosynthetic-deficient mutants cyclopropylsterol isomerase 1-1 (cpi1-1), cotyledon vascular pattern 1-3 (cvp1-3), and fackel-J79 (fk-J79) were initially selected for analyses of sterol deficiency (Carland et al., 2002; Willemsen et al., 2003; Men et al., 2008; Pan et al., 2009). The smt1 (orc) and cpi1-1 mutants have been used for analysis of sterol-dependent trafficking of PIN1 and PIN2 (Willemsen et al., 2003; Men et al., 2008); however, cpi1-1 is also defective in brassinosteroid biosynthesis (Men et al., 2008), and the loss of SMT2 in the cvp1-3 null allele also resulted in the knock-down of SMT1 and SMT3 expression (Carland et al., 2002). Fatty acid methyl esters were analyzed in cvp1-3 and cpi1-1 to determine whether the loss of structural sterols results in compensatory increases in lipid chain lengths. Increased long chain fatty acids were observed in cpi1-1 (P < 0.05) and, to a lesser extent, in cvp1-3, compared with the wild type (Figure 1a). Furthermore, cycloeucalenol and 24-methylene cycloartanol levels are increased in cpi1-1, and may partially substitute for stigmasterol and sitosterol in the PM (Men et al., 2008). The levels of fatty acids and sterols did not differ in abcb19 compared with the wild type (P > 0.05; Figure 1a), and sitosterol levels were unchanged in abcb19, but were at or below the detection limit in cpi1-1 and cvp1-3 (Table S1). Precursor accumulation and diversion into cholesterol synthesis apparently occurs in cpi1-1, as a statistically significant increase in cholesterol content was observed in cpi1-1 but not in cvp1-3 (P < 0.05; Figure 1a). Even a minor diversion of sterol biosynthesis to increased cholesterol production is a concern for data analyses, as cholesterol addition has been shown to increase ABCB19 activity in a manner similar to that seen with mammalian ABCBs (Belli et al., 2009; Titapiwatanakun et al., 2009). As CVP1 is expressed in non-epidermal root cells, where ABCB19-GFP is readily visualized, it was concluded that cvp1-3 would be the most informative mutant to use for analyses of the structural sterol regulation of ABCB19.

Figure 1.

Sterol and sphingolipid biosynthetic mutants show phenotypes similar to abcb19. (a) Fatty acid methyl ester and sterol quantitations in wild type, and in sterol mutants cvp1-3 and cpi1-1; *< 0.05, Student's t-test. (b) Five-day-old seedlings show epinastic cotyledons. (c) Adult plants show dwarf stature, loss of apical dominance and reduced leaf size. (d) Hypocotyl length in 3-day-old dark-grown seedlings; mean ± SD; n =25; *< 0.05, Student's t-test. (e) Auxin transport assays. Data are means ± SDs of three independent experiments of 10 seedlings each; *< 0.001, Student's t-test. (f) The abcb19-101 (mdr1-101) and tsc10a mutants showed decreased apical dominance, indicated by increased secondary rosette inflorescences; mean ± SD; n = 10. (g) Quantification of FM4-64 endocytosis in Figure S3, as indicated by the decrease of relative PM signal in epidermal cells in the wild type, cvp1-3 and tsc10a-2; means ± SDs, n = 10 cells from three repeats.

Both cvp1-3 and tsc10a-2 displayed cotyledon epinasty, similar to abcb19 (Figure 1b and Figure S2A,B), and tsc10a-2 exhibited the reduced inflorescence height, loss of apical dominance and reduction in etiolated hypocotyl length (Figure 1c,d and Figure S2B) characteristic of abcb19 (Noh et al., 2001, 2003; Christie et al., 2011). Neither cvp1-3 nor tsc10a-2 displayed the additional phenotypes exhibited by the severely dwarfed abcb1 abcb19 double mutant (Noh et al., 2001) or twisted dwarf 1 (twd1), which lacks a functional copy of the FK506 binding protein 42 (FKBP42) required for ABCB1 and ABCB19 exit from the endoplasmic reticulum (ER; Geisler et al., 2003; Wu et al., 2010). Furthermore, polar auxin transport reductions in cvp1-3 and tsc10a-2 hypocotyls were similar to those in abcb19 (Figure 1e). Growth phenotypes in tsc10a-2 were also similar to those of abcb19, to the extent that the two lines were difficult to distinguish between (Figure 1b,c,f). Although reduced leaf expansion was observed in all three mutants (Figure 1c), cvp1-3 did not exhibit the reductions in inflorescence length and increased number of secondary inflorescences that are characteristic of abcb19. Additional growth phenotypes (leaf yellowing and temperature stress sensitivity) not associated with abcb19 were observed in cvp1-3, and additive effects in root growth were seen in abcb19 cvp1-3 but were not observed in abcb19 tasc10a double mutants (Figure S2). Meaningful whole-plant phenotypic comparisons of abcb19 with cpi1-1 and fk-J79 were impractical because of the severe developmental defects evident in the later mutants. As a control, we grew abcb19 on the sterol biosynthesis inhibitor fenpropimorph (FEN) or the sphingolipid biosynthetic inhibitor fumonisin B1 (FB1). The abcb19 primary root length was partially resistant to 5 μM FEN treatment at 5 days (P < 0.05), the root extension was the same as in the wild type at higher FEN concentrations (P > 0.05) and the abcb19 root extension at 4.5 days was also partially resistant to 250 nM FB1 (P = 0.07) (Figure S2C,D). These results suggest that the loss of structural sterols has a greater general impact on plant growth than the loss of sphingolipids.

Genetic and chemical depletion of sterols demonstrate the sterol dependence of root gravitropism and PIN2 endocytosis (Men et al., 2008); therefore, the rates of endocytosis in sterol- and sphingolipid-deficient mutants were analyzed using the styryl dye FM4-64. As expected, apparent FM4-64 internalizations in epidermal cells were decreased in cvp1-3 and tsc10a-2 when compared with wild type (Figure 1g and Figure S3). These results indicate that basic endocytotic mechanisms in cvp1-3 and tsc10a-2 are perturbed in root epidermal cells, where ABCB19 trafficking was to be analyzed.

Structural sterols regulate ABCB19-GFP trafficking between the TGN and the PM

The presence of ABCB19 in detergent-resistant membrane (DRM) fractions was shown by western blot analysis, and treatments with the sterol stripping agent methyl-β-cyclodextrin (MβCD) disrupted the weak ABCB19-GFP signals observed on the PM of epidermal cells in ProABCB19:ABCB19-GFP roots (Titapiwatanakun et al., 2009). The co-localization of ABCB19-GFP with membrane sterols was confirmed by filipin staining and epifluorescence microscopy (Figure 2a). To analyze whether ABCB19-GFP localization is regulated by membrane sterols, ProABCB19:ABCB19-GFP signals in cvp1-3, cpi1-1, fk-J79 and abcb19 were examined using confocal microscopy (Carland et al., 2002; Pan et al., 2009). ABCB19-GFP formed punctate intracellular structures in only 30% of cvp1-3 root cells (Figure 2b,c), was reduced overall in cpi1-1 (Figure 2d) and was variably internalized or reduced in fk-J79 (Figure S4A–C). The variability of ABCB19-GFP localization in sterol-deficient mutants may reflect long-term developmental defects, biosynthetic feedback and/or altered transcriptional regulation of the sterol biosynthetic pathway (Willemsen et al., 2003; Men et al., 2008; Pan et al., 2009; Carland et al., 2002; Figure S1). To circumvent these possibilities, the sterol synthesis inhibitor FEN was used. FEN inhibits the C-14 sterol reductase (FACKEL in Arabidopsis) and sterol isomerases (Figure S1), and could inhibit other sterol biosynthetic enzymes in Arabidopsis. FEN has been shown to inhibit squalene epoxidase in Nectria haematococca var. cucurbitae (Ziogas et al., 1991), and to block demethylation of lanosterol during cholesterol synthesis in 3T3 fibroblasts (Corio-Costet et al., 1988). Treatment of 5-day-old seedlings with 50–100 μm FEN for 2–3 h resulted in highly reproducible ABCB19-GFP puncta (Figure 2e), and long-term treatment (for 2.5 days) showed large aggregations (Figure S6). FEN treatment also caused a shift of ABCB19-HA from PM to Golgi-associated membrane fractions in sucrose density gradient fractionation, as indicated by antisera against the H+-ATPase and H+-PPase AVP2, respectively (Figure S5). This apparent shift must be viewed with some caution, as sterols are first incorporated into the smooth ER membrane and FEN treatment appeared to produce a global change in membrane densities as well as altered distribution of the ER marker BiP. A short-term dependence on steroid signalling could be ruled out, as ABCB19-GFP localization was not altered by treatments with the brassinosteroid (BR) biosynthesis inhibitors triadimefon (TRI) and propiconazole (Asami et al., 2003; Hartwig et al., 2012; Figure 2f,g and Figure S6). Long-term TRI treatment produced little change, whereas long-term PZC treatment produced fuzzy aggregations, similar to long-term FEN treatment (Figure S6). The disruption of PIN1 polar localization has been reported in the structural sterol and BR-deficient mutant smt1 (Willemsen et al., 2003); however, overall PIN1-GFP polarity was not disturbed in cvp1-3 and FEN-treated roots, although some intracellular signals were observed (Figure 2h–j). Also, in contrast to the decreased PIN2 endocytosis previously described for epidermal cells of sterol mutants and after inhibitor treatments (Men et al., 2008; Pan et al., 2009), ABCB19-GFP and PIN1-GFP formed more intracellular bodies after treatment with FEN (Figure 2e,j). It is not clear if these vesicles are derived from endocytosis or exocytosis; however, this effect was not generalized, as PLASMA MEMBRANE INTRINSIC PROTEIN 2A (PIP2A)-GFP localization was unaltered by treatment with FEN (Figure 2k,l).

Figure 2.

Structural sterols regulate ABCB19-GFP trafficking between the trans-Golgi network (TGN) and the plasma membrane (PM). (a) Co-localization of ABCB19-GFP and the sterol stain filipin at the PM. (b) ProABCB19:ABCB19-GFP signal in control roots, (c) in cvp1-3, (d) in cpi1-1, (e) after FEN, (f) tridimefon and (g) propiconazole treatment. (h) ProPIN1:PIN1-GFP signal in control, (i) in cvp1-3, (j) after FEN treatment. (k) Pro35S:PIP2A-GFP signal in control, (l) after FEN treatment. (m) Co-localization of ABCB19-GFP with FM4-64 after MβCD treatment. Five-day-old seedlings were treated with MβCD for 1 h, then stained with 20 μM FM4-64 for 5–10 min, washed twice and imaged immediately. Quantification of co-localization is indicated on the figure as a percentage, mean ± SD, n = 40, eight cells from each root from five roots. (n) Co-localization of ABCB19-GFP with TGN marker SYP61-CFP. (o) Quantification of co-localization of intracellular ABCB19-GFP with markers after FEN and FB1 treatments, mean ± SD, n = 40, eight cells from each root from five roots. Scale bar: 20 μm (A–L); 10 μm (n, o).

Fluorescence recovery after photobleaching (FRAP) was performed to assess the effects of FEN treatment on PM and intracellular ABCB19-GFP signals. FRAP analyses revealed that ABCB19-GFP signal recovery at the PM was delayed by FEN treatment (Figure S6A). In contrast, when intracellular ABCB19-GFP signals were analyzed by FRAP following FEN treatment, ABCB19-GFP signals increased within 2 min, and continued to increase for >12 min, whereas control PM ABCB19-GFP signals decreased at the background rate (Video Clips S1 and S2). These results indicate that sterols contribute to the proper targeting of ABCB19-GFP to the PM. Brefeldin A and cycloheximide treatments were not employed to examine endocytotic effects, as ABCB19 is not trafficked via the brefeldin A-sensitive GNOM pathway (Titapiwatanakun et al., 2009). FEN treatment followed by a 5–10 min FM4-64 pulse resulted in the partial co-localization of ABCB19-GFP and FM4-64 signals, which decreased after 30 min, and were largely eliminated after 1 h when FM4-64 was trafficked to the vacuole (Figure S6B). These results indicate that ABCB19-GFP accumulates in the TGN but not in the vacuole with the depletion of sterols. These results were confirmed with the direct stripping of sterols by MβCD, which increased ABCB19-GFP co-localization with FM4-64 at the TGN (Figure 2m).

FEN-induced ABCB19-GFP internalizations were co-localized with established subcellular fluorescent protein markers in the root epidermal cells after FEN treatment. ABCB19-GFP partially co-localized with TGN/early endosome compartments characterized by SYP61-CFP, VTI12-mCherry, RabA2a-YFP or RabA5d-mCherry, but not SEC12-YFP (ER), mannosidase Man49-mCherry (Golgi) or 35S:mCherry (cytosol) (Figures 2n,o and Figure S6C; see 'Experimental Procedures'). ABCB19-GFP also formed intracellular accumulations after FEN treatment in the presence of cycloheximide (CHX), although extended treatment with CHX alone resulted in the formation of additional ABCB19-GFP puncta (Figure S7). This result is presumably caused by the loss of FKBP42, which is required for the folding and exit of ABCB19 from the ER, and is not detected after CHX treatment (Wu et al., 2010). Taken together, these results suggest that structural sterols function in anterograde trafficking of ABCB19-GFP between the TGN and the PM.

Sphingolipids regulate ABCB19-GFP trafficking at the Golgi

ABCB19 has been shown to stabilize PIN1 in sterol- or sphingolipid-enriched detergent-resistant membranes (Titapiwatanakun et al., 2009). Very long chain fatty acid (VLCFA) sphingolipids (approximately 80% of total sphingolipids) were shown to be critical for PIN1 localization (Markham et al., 2011). The localization of ProABCB19:ABCB19-GFP in tsc10a-2 and in abcb19 after treatment with the VLCFA sphingolipid biosynthetic inhibitor fumonisin B1 (FB1) was examined (Markham et al., 2011). ABCB19-GFP formed intracellular puncta in FB1-treated and tsc10a-2 root epidermal cells (Figure 3a–d). Consistent with previous studies (Roudier et al., 2010; Markham et al., 2011), PIN1-GFP formed defined intracellular bodies in FB1-treated and tsc10a-2 roots (Figure 3e–g). Slight effects were seen on AUX1-YFP, but not PIN2-GFP, in tsc10a-2 or after FB1 treatment (Figure S4D–G). As control PIP2A-GFP localization was further examined, PIP2A-GFP, but not ABCB19-GFP, localized to punctate PM structures after salt treatment (Figure S4H–J), as previously shown (Boursiac et al., 2005), but PIP2A-GFP localization was unchanged by treatment with FB1 (Figure S4K).

Figure 3.

Sphingolipids regulate ABCB19-GFP trafficking at the Golgi. (a) ProABCB19:ABCB19-GFP signal in control, (b) after FB1 treatment, (c) in tsc10-2 and (d) in tsc10a-2 treated with FB1. (e) ProPIN1:PIN1-GFP signal in control, (f) after FB1 treatment and (g) in tsc10a-2. (h) Co-localization of ABCB19-GFP with the Golgi marker Man49-mCherry (MAN) and (i) TGN marker VTI12-mCherry. Scale bar: 10 μm.

As tsc10-2 also exhibits reduced GluCer levels (Chao et al., 2011), phenyl-2-hexadecanoylamino-3-morpholino-1-propanol (PPMP), a GluCer synthase inhibitor, was used to analyze whether GluCer reduction would change ABCB19-GFP localization. PPMP treatments resulted in intracellular aggregations of ABCB19-GFP and PIN1-GFP, but not PIP2A-GFP or PIN2-GFP (Figure S4L–O). These results indicate that wild-type distributions of VLCFA and GluCer sphingolipids are required for ABCB19 and PIN1 localization, but not for the localization of all PM proteins.

After FB1 treatment, the majority of intracellular ABCB19-GFP co-localized in the Golgi structures labeled with Man49-mCherry, with a minor overlap of the signals from TGN/early endosomes labeled with VTI12-mCherry, SYP61-CFP, RabA2a-YFP or RabA5d-mCherry (Figures 2n, 3h,i and Figure S6C). ABCB19-GFP co-localization was minimal with SEC12-YFP (ER) or soluble 35S:mCherry (cytosol) (Figure S6C). Long-term FB1 treatment produced increasing numbers of ABCB19-GFP puncta (Figure S6). The addition of CHX did not inhibit the formation of ABCB19-GFP intracellular accumulations after FB1 treatment (Figure S7). These results indicate that VLCFA sphingolipids play a major role in ABCB19-GFP trafficking to the Golgi, TGN and PM; however, it is still not clear whether sphingolipids directly regulate PIN1 localization, as the stability of PIN1 in DRM fractions requires ABCB19 function (Titapiwatanakun et al., 2009).

Other PM ABC transporters

The dependence of ABCB19 function on sphingolipids and, to a lesser extent, sterols could be a general characteristic of PM ABC transporters. The auxin exporter ABCB1 and root-specific reversible ABCB4 auxin transporter have also been reported to be associated with sterol and sphingolipid-enriched membranes (Borner et al., 2005; Titapiwatanakun et al., 2009; Kubeš et al., 2011); however, only minor intracellular accumulations of ABCB1-GFP were seen in FB1-treated and tsc10a-2 root epidermal cells (Figure S4P–R). Furthermore, ABCB4 PM localization is particularly sensitive to the introduction of water molecules into the membrane by DMSO (Kubeš et al., 2011). ProABCB4:ABCB4-GFP formed intracellular accumulations and puncta similar to ABCB19 in FB1-treated wild-type and tsc10a-2 roots (Figure S4S–U). ABCG37 is a PM-localized, broad-spectrum transporter from another ABC subclass (Ruzicka et al., 2010). No obvious changes were detected after FB1 or FEN treatment of 35S:ABCG37-GFP roots (Figure S4V–X). These results suggest that sphingolipid interactions regulate the trafficking and stability of ABCB4 and ABCB19 more than the other characterized PM ABC transporters.

ABCB19 is required for sphingolipid-mediated PIN1 trafficking from VTI12 vesicles

Next we analyzed the identity of the discreet PIN1 bodies after FB1 treatment. PIN1-GFP almost completely co-localized with the SNARE VTI12 fused to mCherry in vascular parenchyma cells (where PIN1 is expressed) after FB1 treatment, which resulted in the aggregation of both proteins (Figure 4a,b). In abcb19 mutants, PIN1-GFP formed similar intracellular structures in these cells, and the effect was enhanced by FB1 treatment (Figure 4c,d). In these cells, ABCB19 also partially co-localized with VTI12-mCherry (Figure 4e,f). These data indicate that both ABCB19 and sphingolipids mediate PIN1 function within the same pathway, which may be regulated by the SNARE VTI12 (Sanmartín et al., 2007). These results also indicate that PIN1 internalization after sphingolipid depletion was partially an indirect effect of loss of ABCB19 interaction with PIN1 in vascular cells, where ABCB19 and PIN1 are co-expressed. This result is consistent with previous evidence of PIN1 stabilization in detergent-resistant membranes by ABCB19 (Blakeslee et al., 2007; Titapiwatanakun et al., 2009).

Figure 4.

ABCB19 is required for sphingolipid-mediated PIN1 trafficking from VTI12 compartments. (a) Co-localization of PIN1-GFP with VTI12-mCherry in control (Ctrl) and (b) after FB1 treatment, (c) in pgp19-1 and (d) in pgp19-1 treated with FB1. (e) ProABCB19:ABCB19-GFP signal in control. (f) Co-localization of ABCB19-GFP with VTI12-mCherry after FB1 treatment. (g) ProABCB19:ABCB19-GFP signal control, (h) in twd1-3, (i) after FB1 treatments in twd1-3 and (j) after FEN treatments in twd1-3. (k, l) ABCB19-GFP in twd1-3 after cycloheximide (CHX) treatment. (m) ABCB19-GFP in twd1-3 co-localization with YFP-SEC12 (ER) and (n) NPSN12-mCherry (PM). Scale bars: 10 μm.

FKBP42 function is distinct from sterol and sphingolipid regulation of ABCB19-GFP

The PM localization and activity of ABCB19 requires C-terminal interactions with FKBP42/TWD1 (Geisler et al., 2003; Bailly et al., 2008; Titapiwatanakun et al., 2009). Recent evidence indicates that loss of FKBP42 reduces ABCB19 trafficking from the ER (Wu et al., 2010), and suggests that FKBP42 is a functional ortholog of mammalian FKBP38 that interacts with calmodulin and functions on the ER surface as a co-chaperone for CYSTIC FIBROSIS TRANSMEMBRANE RECEPTOR (CFTR) and P-GLYCOPROTEIN1/ABCB1 (Wang et al., 2006; Walker et al., 2007). Partial retention of ABCB19-GFP at the ER and reduced abundance at the PM in twd1-3 was indicated by co-localization with SEC12-YFP and NPSN12-mCherry (Figure 4g,h,m,n and Figure S8). Reduced PM localization of ABCB19 in twd1-1 (Figure S8A,B) was not the consequence of altered transcription (Figure S8A–C). FKBP42 is required for the folding and maturation of ABCB19, as degradation products and a lower mobility band associated with misfolding comprised most of the signals observed in ABCB19 western blots prepared from twd1-3 (Figure S8D,E). FB1 and FEN treatments of twd1-3 resulted in some additional punctate ABCB19-GFP signals (Figure 4i,j), consistent with post-ER dependence of ABCB19 trafficking on sterols and sphingolipids. CHX treatments reduced ABCB19-GFP signals in twd1-3 without generating additional internalization (Figure 4k,l). These results indicate that ABCB19 exit from the ER requires FKBP42-mediated folding. Structural sterol and sphingolipid regulations appear to take place primarily in the Golgi and post-Golgi compartments.


The results presented here demonstrate that ABCB19 is trafficked to the PM via the Golgi and TGN via mechanisms that are dependent on sphingolipid and sterol interactions. A convergence point for endocytotic and secretory trafficking (Dettmer et al., 2006), the TGN is the site of secretory vesicle formation and lipid sorting, and also appears to be the site of lipid raft formation (Klemm et al., 2009). In Arabidopsis, structural sterols have been shown to regulate PIN2 endocytosis and polarity (Men et al., 2008), and the results presented here indicate that ABCB19-GFP trafficking between the TGN and PM is also regulated by membrane sterol content. However, PIN2 incorporation into sterol-containing membrane domains occurs as part of endocytotic processes (Men et al., 2008). In contrast, the rapid fluorescence recovery of photobleached ABCB19-GFP after FEN-induced internalization, without a change in PM signals, and the co-localization of ABCB19-GFP with FM4-64 at the TGN after FEN treatment suggest that loss of sterols blocks post-Golgi anterograde ABCB19 trafficking, but not endocytosis. Overall, the results presented here suggest that sterol packing of ABCB19 and association with sterol-enriched membranes occurs at or post TGN prior to the arrival at the PM. This TGN/post-TGN secretory vesicle population may be similar to that characterized by the chimeric sterol-dependent raft protein FusMidGFP in yeast (Proszynski et al., 2004, 2005), and a light membrane fraction from Arabidopsis containing APM1 and Sec14, in which ABCB19 was originally identified (Murphy et al., 2002; Peer et al., 2009; Hosein et al., 2010).

The synthesis of VLCFA occurs via a four-member ER-resident elongase complex, and VLCFAs are then incorporated into sphingolipids by ceramide synthase (Klemm et al., 2009; Markham et al., 2011). The important role of sphingolipids (especially VLCFA sphingolipids and GluCer sphingolipids) in maintaining the membrane domains characterized by ABCB19 is demonstrated by the disruption of ABCB19, PIN1 and VTI12 localization, and by the aggregation of three proteins into the same compartment. Such an effect might be tissue-specific, as VTI12 aggregates with PIN1-GFP in stele cells but not so obviously in epidermal cells (Figure 4a,b and Figure S6C). Similar to previous reports (Markham et al., 2011), another TGN marker RabA2b-YFP formed aggregations in epidermal cells (Figure S6C), but we could not observe co-localization of this marker with PIN1-GFP because of the absence of RabA2b-YFP in the stele. Furthermore, ABCB19-GFP mainly co-localized with the Golgi markers and only partially co-localized with RabA2b-YFP in epidermal cells with reduced VLCFA sphingolipids (Figure 2o). ProPIN2:PIN1-GFP localization was not pursued because ABCB19-GFP is not highly expressed in epidermal cells and ABCB1 was found to interact with PIN2 in epidermal cells (Blakeslee et al., 2007). In silico docking of sitosterol with an ABCB19 structural model threaded on murine ABCB1 (Aller et al., 2009; Bailly et al., 2011) suggests that sitosterol binds to ABCB19 transmembrane domains within the membrane inner (site 1) and outer leaflet (site 2; Figure 5a), both near the predicted IAA binding sites (Bailly et al., 2011). The binding of sphingolipids is probably required for the stability of ABCB19 tertiary structure, and consequently for its protein–protein interactions (such as ABCB19-PIN1 interaction) and trafficking to the PM or the cell plate (Bach et al., 2011).

Figure 5.

Models of interaction of ABCB19 with sterols and VLCFA sphingolipids. (a) In silico docking of sitosterol with an ABCB19 structural model. (b) Working model of ABCB19 trafficking depends on FKBP42, sterols and VLCFA sphingolipids: 1, FKBP42 facilitates ABCB19 folding in the endoplasmic reticulum (ER); 2, ABCB19-sphingolipid interactions start in the Golgi; 3, sterol association with ABCB19 occurs at the trans-Golgi network (TGN).

Plant ABCB proteins mature via the same type of ER luminal scanning and ER surface folding mechanisms mediated by FKBP38/FKBP42 that assure quality control of mammalian PM ABC transporters (Wang et al. 2007; Walker et al., 2007); however, FKBP42 interactions with the ABCB19 C-terminal cytosolic domain also appear to be required to activate or maintain the functionality of ABCB19 at the PM. FKBP42 was originally biochemically identified in PM fractions (Murphy et al., 2002), has been shown to be distributed to the ER, tonoplast and PM (Geisler et al., 2003), and has been shown to activate ABCB1 and ABCB19 transport activity via interactions with the cytosolic C-terminal nucleotide binding/protein interaction loop domain (Bouchard et al., 2006; Bailly et al., 2008). It is still not clear whether ABCB19 maturation and activation by FKBP42 involves the FK506-sensitive peptidyl prolyl-isomerase (PPIase) activity that is the signature of this class of proteins. Although containing tetracopeptide-repeat, PPIase, FK506-binding and calmodulin-binding domains similar to those in mammalian FKBP38, Arabidopsis FKBP42 has no detectable PPIase activity (Kamphausen et al., 2002). More extensive studies are required to determine whether PPIase activity contributes to FKBP42 function in protein maturation. In any case, this maturation process is distinct from the post-Golgi processes affected by structural sterols and sphingolipids.

Overall, these results indicate that ABCB19 association with sphingolipids begins during anterograde trafficking in the Golgi, after completion of folding by FKBP42 in the ER (Figure 5b). Non-association of ABCB19 with sphingolipids results in ABCB19 mislocalization (Figure 5b), and the loss of sterols blocks outbound post-Golgi trafficking of ABCB19 (Figure 5b). The greater hydrophobicity of sphingolipids containing VLCFA compared with sterols suggests a primary role for sphingolipids in establishing the membrane domains in which PIN1 and ABCB19 interact. We propose that the lack of ABCB19 involvement in the trafficking mechanisms that mobilize PINs and the clear dependence of ABCB19 PM localization on sterols and sphingolipids make the Arabidopsis ABCB19 an ideal model system for analyses of lipid regulation of ABCB/P-glycoprotein transporters. This system also provides an ideal platform for analyses of the coordinate regulation of the PIN and ABCB auxin transport systems at the post-translational level.

Experimental Procedures

Plant materials and growth conditions

ProABCB19:ABCB19HA was previously described by Blakeslee et al. (2007), ProABCB19:ABCB19-GFP and ProABCB1:ABCB1-GFP were previously described by Mravec et al. (2008), 35S:FKBP42HA was previously described by Geisler et al. (2003), twd1-1 was previously described by Geisler et al. (2003), twd1-2 and twd1-3 were gifts from Dr Burkhard Schulz, cvp1-3 and fk-J79 were previously described by Pan et al. (2009), cpi1-1 was previously described by Men et al. (2008), tsc10a-2 was previously described by Chao et al. (2011), VHA-a1-GFP was previously described by Dettmer et al. (2006), and Man49-mCherry, Man49-CFP and γ-TIP-mCherry were previously described by Nelson et al. (2007). VTI12-mCherry, RabD1-mCherry and RabA5d-mCherry were previously described by Geldner et al. (2009). Seedlings were grown on 1% phytagar plates, containing quarter-strength MS basal salts, 0.5% sucrose, pH 5.5, at 22°C, with 16 h of daylight at 100 μmol m−2 s−1, except as indicated for specific treatments. Pictures were taken and measurements were performed using imagej (NCBI, Plants on soil were grown in the glasshouse under natural light conditions, and in the winter, the day length was extended to 14 h with high intensity discharge lights (150 μmol m−2 s−1). Growth phenotypes were analyzed in temperature-controlled chambers. See for additional information.

Membrane preparation, two-phase separations and sucrose density gradients fractionation

Five-day-old Arabidopsis seedlings (5–10 g) were homogenized and total membranes were prepared as described previously (Titapiwatanakun et al., 2009). Fresh microsomal membranes were resuspended in 200 μl of buffer (0.33 m sucrose, 3 mm KCl and 5 mm potassium phosphate, pH 7.8, and 20 mg ml−1 complete protease inhibitor cocktail; Sigma-Aldrich,, and added to the two-phase separation system prepared in the same buffer. The final composition of the phase system was 6.2% (w/w) dextran (Mr = 413 000) and 6.2% (w/w) polyethylene glycol (Mr = 3350). After mixing (inverting 30 times), the phases were separated by centrifugation at 1500 g for 5 min. The upper PM phase was repartitioned twice with the lower phase. Likewise, the lower phase (endosomal membrane phase) was repartitioned twice with the upper phase. The final upper and lower phases were diluted with 10 mm BTP-MES, pH 7.8, 1 mm EGTA, 1 mm EDTA, 0.29 m sucrose and 20 mg ml−1 complete protease inhibitor cocktail (Sigma-Aldrich), then centrifuged at 120 000 g for 1 h. Pellets were resuspended in 10 mm BTP-MES, pH 7.8, 1 mm EGTA, 1 mm EDTA, 0.29 m sucrose, 20 mg ml−1 complete protease inhibitor cocktail (Sigma-Aldrich) and 0.2% MβCD (methyl-β-cyclodextrin). For sucrose density gradient separation, 5-day-old ProABCB19:ABCB19-HA seedlings were treated with fenpropimorph (FEN) by spraying 100 μm FEN on seedlings. After 3 h of treatment with FEN, seedlings (approximately 4 g) were homogenized and microsomal membranes were prepared as described previously. (Titapiwatanakun et al., 2009). Microsomal membranes (1 ml) were layered on a 15-ml linear 15–50% sucrose solution. Samples were spun at 100 000 g in an SW 32Ti rotor (Beckman Coulter, for 16 h at 4°C. Then 600-μl samples were taken carefully from top to bottom without disturbing the gradients and 15-μl samples were separated on 8% SDS-PAGE for western blots.

Western blots

Protein concentration was determined with amido black (Yang and Murphy, 2009). Between 10 and 30 μg of protein (as noted) was separated with SDS-PAGE, transferred to pure nitrocellulose membranes and analyzed by western blot. ABCB19 antisera were used at a 1:1000 dilution. H+-ATPase antisera (Agrisera, were used at 1:2000 dilution. The 2E7 monoclonal antibody against the vacuolar H+-ATPase was used at 1:200 dilution. BIP antisera were used at 1:1000 dilution. H+-PPase antisera were used at 1:1000 dilution. HA anti-rabbit antibody and GFP anti-rabbit antibody were from Santa Cruz Biotechnology ( Immunopure® goat anti-rabbit IgG was from Thermo Scientific ( Commercial antibodies were used at concentrations recommended in vendors' manuals.

Quantitative real-time PCR

Total RNA was isolated from rosette leaves using the RNeasy Mini Kit according to the vendor's manual (Qiagen, Total RNA (3 μg) was used for first-strand cDNA synthesis with the BioScript RNase H Minus reverse transcription kit according to the vendor's manual (Bioline, The specific intron-spanning primer pairs for ABCB19 and FKBP42 were: B19 QrtF, 5′-CTATAGCCGAGAGAATCTCAGTA-3′, B19 QrtR, 5′-CAGAGATAGTTGCTGAGCAAAGT-3′; TWD1 QrtF, 5′-CAGTATGAAATGGCCATAGCATACA-3′, TWD1 QrtR, 5′-CTTCTGTCAACACAATGTTGCAGT-3′. Real-time PCR was performed using Icycler (Bio-Rad Laboratories, and Evagreen (Biotium,

Cytological studies

Protein localizations in Arabidopsis seedlings were visualized using an LSM 710 laser spectral scanning confocal microscope (Zeiss,, with a 40× water immersion lens with zoom of 2–4 and pixel dwell of 6.3 μs. The master gain was always set to less than 823, with a digital gain of 1. All reagents were from Sigma-Aldrich, unless otherwise noted. Inhibitor treatments were administered to 5-day-old seedlings on soaked filter paper in a solution of quarter-strength MS medium, pH 5.5, as indicated: N-1-naphthylphthalamic acid (NPA), 10 μm 5 h; fumonisin B1 (FB1), 3 μm, 16–24 h; propiconazole, 5–10 μm, 2–24 h; fenpropimorph (FEN), 50–100 μm, 1–3 h; triadimefon (TRI), 2–10 μm, 2–24 h; MβCD 50 μm, 1–4 h; cylcoheximide (CHX) and FEN, 50 μm CHX + 100 μm FEN, 2–3 h; FM4-64, 5 μm (Invitrogen, Molecular Probes, for 15–90 min, washed once; for FEN + FM4-64, treated with FEN for 2 h then treated with FM4-64, 20 μm, 5–10 min, washed twice. For GFP acquisition: 488 nm (20%) excitation and 493–598 nm emission using Zeiss LSM 710 (Zeiss). For FM4-64, RFP and mCherry: 594 nm (15%) excitation and 599–647 nm emission. For YFP: 514 nm (15%) excitation and 519–621 nm emission. For CFP: 405 nm (5%) excitation and 454–581 nm emission. For co-localization of CFP and GFP, tracks 1 and 2, 405 and 488 nm excitation; for GFP and RFP, or mCherry or FM4-64, 488 and 594 nm excitation, and 493–551 and 598–735 nm emission, respectively; for GFP and YFP, λ mode, 10-nm steps, 488 nm (15%) excitation, 494–601 nm emission, 786 master gain, digital gain 1 and pixel dwell of 6.3 μs, images were then linear unmixed with corresponding reference spectra: GFP (ABCB19-GFP) and YFP (YFP-SEC12 or YFP-RabA2a). Filipin staining was performed as described previously (Boutté et al., 2011), with some modifications. Briefly, seedlings were incubated at room temperature (∼25 °C) in fenpropimorph (100 μm) for 2 h and filipin (30 μm) for 10 min prior to image collection. Images were collected using a Plan-Neofluar 40× (1.3 numerical aperture) objective (Zeiss). Wide-field microscopy was conducted using a Zeiss Axio Observer.Z1 and images were captured with an AxioCam MRc 5 CCD camera (Zeiss), controlled by axiovision (Zeiss). The following filter sets were used to distinguish between fluorophores: green fluorescent protein (GFP) (479.5–494.5 nm excitation; 510–560 nm emission) and filipin (320–360 nm excitation; 460–500 nm emission). Illumination was provided by an X-Cite series 120Q, using 12% light intensity. Exposure times ranged from 300 to 500 ms. For FRAP analysis, seedlings were treated with ±100 μm FEN for 2 h and immediately placed on slides with 100 μm FEN for imaging. For GFP capture an argon laser was used to provide excitation at 488 nm, with emission at 493–598 nm. Attenuation of the laser was set at 10% for acquisition and 100% for photobleaching. The pinhole was set to 90 μm and the time of pixel dwell was 3.15 μs. The master gain was set at 810 and the digital offset was 6. Quantification of co-localization of ABCB19-GFP with other markers was analyzed using zen 2009 (Zeiss). Briefly, representative images from five seedlings were selected for analysis. Intracellular regions from eight cells were selected in each image. The intracellular regions did not include the signal at the PM. For FM4-64 experiments, the PM or cytosolic region was also selected, and the signal in a 2-μm section from a z-stack was analyzed and the ratio was calculated. The number of pixels of green channel (region 1), red (or purple) channel (region 2) and overlapping region 3 were obtained by zen 2009 (Zeiss). The percentage of co-localization was calculated by the equation: co-localization percentage = 100 × [region 3/(region 1 + region 2 + region 3)]. All images were processed with zen 2009 (Zeiss) and photoshop (Adobe, Briefly, the total B19-GFP intracellular signals were counted and then those that co-localized with the markers were counted. Then the percentage of signal that co-localized with the markers was calculated. As the level of co-localization in the control samples was <1%, we did not use a correction factor, as this was within the standard deviation.

Auxin transport assays

Auxin transport assays were as described previously (Blakeslee et al., 2007).

FAMES and cholesterol quantitations

The FAMES and cholesterol quantitations were as described previously (Titapiwatanakun et al., 2009).


We thank Dr David Salt for supplying the tsc10a-2 mutant before publication. We also thank Dr Boosaree Titapiwatanakun for input on the initial conceptualization of this work. We also thank Dr Joshua J. Blakeslee for auditing the initial data sets generated for this project. The work was supported by the Department of Energy, Basic Energy Sciences, grant no. DE-FG02-06ER15804 to ASM. We declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.