Báo cáo khoa học: Treatment of neutral glycosphingolipid lysosomal storage diseases via inhibition of the ABC drug transporter, MDR1 Cyclosporin A can lower serum and liver globotriaosyl ceramide levels in the Fabry mouse model

Treatment of neutral glycosphingolipid lysosomal storage diseases via inhibition of the ABC drug transporter, MDR1

Cyclosporin A can lower serum and liver globotriaosyl ceramide levels in the Fabry mouse model Michael Mattocks1, Maria Bagovich1, Maria De Rosa1,4, Steve Bond2, Beth Binnington1, Vanessa I. Rasaiah2, Jeffrey Medin2,3 and Clifford Lingwood1,4,5

1 Research Institute, The Hospital for Sick Children, Toronto, Canada 2 Ontario Cancer Institute, University Health Network, Toronto, Canada 3 Department of Medical Biophysics, University of Toronto, Canada 4 Department of Laboratory Medicine and Pathology, University of Toronto, Canada 5 Department of Biochemistry, University of Toronto, Canada

Keywords enzyme replacement therapy; Gaucher disease; a-galactosidase; glucosyl ceramide translocase; HUS model

Correspondence C. Lingwood, Research Institute, The Hospital for Sick Children, Toronto, Ontario M5G 1X8, Canada Fax: +416 813 5993 Tel: +416 813 5998 E-mail: cling@sickkids.ca

(Received 20 January 2006, revised 2 March 2006, accepted 10 March 2006)

doi:10.1111/j.1742-4658.2006.05223.x

the ABC transporter, multiple drug resistance We have shown that translocates glucosyl ceramide from protein 1 (MDR1, P-glycoprotein) the cytosolic to the luminal Golgi surface for neutral, but not acidic, gly- cosphingolipid (GSL) synthesis. Here we show that the MDR1 inhibitor, cyclosporin A (CsA) can deplete Gaucher lymphoid cell lines of accumu- lated glucosyl ceramide and Fabry cell lines of globotriaosyl ceramide (Gb3), by preventing de novo synthesis. In the Fabry mouse model, Gb3 is increased in the heart, liver, spleen, brain and kidney. The lack of renal glomerular Gb3 is retained, but the number of verotoxin 1 (VT1)-staining renal tubules, and VT1 tubular targeting in vivo, is markedly increased in Fabry mice. Adult Fabry mice were treated with a-galactosidase (enzyme- replacement therapy, ERT) to eliminate serum Gb3 and lower Gb3 levels in some tissues. Serum Gb3 was monitored using a VT1 ELISA during a post-ERT recovery phase ± biweekly intra peritoneal CsA. After 9 weeks, tissue Gb3 content and localization were determined using VT1 ⁄ TLC over- lay and histochemistry. Serum Gb3 recovered to lower levels after CsA treatment. Gb3 was undetected in wild-type liver, and the levels of Gb3 (but not gangliosides) in Fabry mouse liver were signiﬁcantly depleted by CsA treatment. VT1 liver histochemistry showed Gb3 accumulated in Kupffer cells, endothelial cell subsets within the central and portal vein and within the portal triad. Hepatic venule endothelial and Kupffer cell VT1 staining was considerably reduced by in vivo CsA treatment. We conclude that MDR1 inhibition warrants consideration as a novel adjunct treatment for neutral GSL storage diseases.

Abbreviations BSA, bovine serum albumin; CsA, cyclosporin A; ERT, enzyme replacement therapy; Gb3, globotriaosyl ceramide; GlcCer, glucosyl ceramide; GSL, glycosphingolipid; HUS, hemolytic uremic syndrome; LacCer, lactosyl ceramide; LSD, lysosomal storage disease; MDR1, multiple drug resistance protein 1 (P-glycoprotein); NGS, normal goat serum; VT1, verotoxin 1; TLC, thin layer chromatogram.

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2064

The lysosomal storage diseases (LSD) are genetic deﬁ- ciencies in glycoconjugate catabolism, each due to a lack of a speciﬁc lysosomal sugar hydrolase or its acti- vator protein [1]. The (mainly neurological) symptoms are due to the intracellular accumulation of the enzyme substrate. In the ‘glycosphingolipidoses’, this

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

synthase inhibitors, such as N-butyldeoxynojirimycin, have proven effective in animal storage disease models [19] and in clinical trials for Gaucher disease [20,21]. Such imino sugars, however, also inhibit glucosidase processing of N-linked high mannose oligosaccharides [22] and glycogen breakdown [23].

accumulation results in the formation of lipid inclu- sions and multilamellar structures which prevent nor- mal cell function. Symptoms depend on the enzyme, age of onset and residual enzyme activity [1]. Because only (cid:1) 10% residual enzyme activity may be sufﬁcient to avert clinical symptoms, exogenous enzyme-replace- ment therapy (ERT) has been developed, particularly in the two neutral GSL storage diseases, Gaucher (glucosyl ceramide accumulates) and Fabry (globotria- osyl ceramide, Gb3, accumulates) [2–4]. a-Galactosi- dase administered to Fabry patients is able to reduce serum levels of Gb3 by 50% [5], liver Gb3 and by inference, kidney Gb3 levels [5,6]. In the Fabry mouse model, in which the a-galactosidase is abscent [7], ele- vated serum Gb3 levels (serum Gb3 is undetectable in normal mice) can be deleted by ERT, but tissue Gb3 is more refractory. This may be due, in part, to the direct access of the enzyme to the serum substrate. To digest accumulated Gb3 in tissue, the enzyme must be taken up by cells within the tissue and targeted intracellularly to the lysosome. This is achieved, in vitro at least, via the mannose phosphate receptor pathway and replace- ment a-galactosidase is phosphomannosylated to pro- mote such uptake [8]. Within the Fabry mouse tissues, liver Gb3 is most susceptible to a-galactosidase ther- apy. Although some lowering of spleen and heart Gb3 is seen, renal Gb3 is more resistant [9].

Unlike all other GSLs, GlcCer is made on the outer leaﬂet of the Golgi bilayer [24] and must be ‘ﬂipped’ into the lumen to access the glycosyltransferases for further carbohydrate elongation. Multiple drug resist- ance protein 1 (MDR1) can function as a glycolipid ﬂippase [25,26]. We showed MDR1 to be responsible for this translocation in the majority of cultured cells [27,28]. The conversion of ceramide to GlcCer and other GSLs has been associated with drug resistance, as a means to avoid ceramide-induced cell death [29,30], although this has been questioned [31]. MDR1- mediated GlcCer translocation into the Golgi could be a component of such resistance. However, we found that MDR1-translocated GlcCer is used only for neut- ral GSL synthesis [28] because inhibition of MDR1 does not affect cellular ganglioside synthesis. This pro- vides a degree of selectivity not available in the other approaches to substrate reduction therapy as a clinical management for Fabry disease. In addition, the long- term clinical experience with drugs that modulate MDR1 in cancer, and, for cyclosporin A (CsA), immu- nosuppression, would provide signiﬁcant advantage, in terms of deﬁned toxicity and dosage. Although MDR1 expression varies within tissues, expression in the kid- ney, and liver [32,33], sites of Gb3 accumulation in the Fabry mouse, make this a feasible approach.

In the Fabry mouse, there is no gross pathology (although a thrombotic deﬁciency has recently been found) [10], but in Fabry disease, the primary pathol- ogy is in the kidney [1], the major site of Gb3 synthesis in man [11], and in the heart, possibly due to the association of Gb3 synthesis with the microvasculature. GSL synthesis in man and mouse are distinct, partic- ularly in the kidney, where Gb3 can be found in the human, but not murine, glomerulus [12–14].

In order to begin to address this potential, we deter- mined the effect of CsA on Gb3 synthesis in the Fabry mouse model. Our studies have further delineated the abnormal Gb3 synthesis in this model and have shown that the inhibition of MDR1 is a viable potential approach to the reduction of Gb3 in both the serum and certain tissues of this model.

Results

MDR1 inhibition in LSD cell lines

(PDMP), or

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2065

Epstein Barr virus (EBV) transformed B-cell lines from Gaucher and Fabry LSD patients were cultured with 4 lm CsA for four days. The GSL fractions were puri- ﬁed and separated by thin layer chromatography (TLC). Figure 1A shows the accumulation of glucosyl ceramide (GlcCer) was prevented in CsA-treated Gau- cher B lymphoblasts. Three cell lines were tested. Glc- Cer accumulated in each, but in one cell line, lactosyl ceramide accumulated also (Fig. 1A, lane 6). In each Despite its clinical success, the extraordinary cost of ERT has limited patient access and promoted the development of alternative strategies. Gene therapy is a candidate strategy for Fabry which may eventually prove the most satisfactory [15]. The third approach has been to develop procedures to restrict the synthesis of Gb3. Two strategies have been developed. Both have focused on inhibitors of glucosyl ceramide synthase. This enzyme is the ﬁrst glycosyl transferase required for the synthesis of most GSLs, including Gb3 (in Fabry disease) and of course, GlcCer (in Gaucher disease). By inhibiting this enzyme, the synthesis of most GSLs (and all gangliosides) is prevented. The glucosyl ceramide synthase substrate mimic, d,l-threo-1-phenyl-2-decan- its oylamino-3-morpholino-1-propanol derivatives with improved selectivity [16], provide one approach [17,18]. Imino sugar-based glucosyl ceramide

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

B

C

D

F

E

A

GlcCer GalCer

GM2

LacCer

GM1

Gb3

Gb4

Gb5

Fig. 1. Effect of cyclosporin A (CsA) on cultured Gaucher and Fabry B-cell line glycosphingolipids (GSLs). The neutral GSL fraction (from 2 · 106 cells per lane) was separated by thin layer chromatogram (TLC) (C ⁄ M ⁄ W 65 : 25 : 4 v ⁄ v ⁄ v). The doublets corresponding to GlcCer and Gb3 are shown by arrows. (A) Gaucher lymphoblastoid cell lines, detected using orcinol spray. Lane 1, GSL standards, GlcCer, GalCer, LacCer, Gb3, Gb4, Gb5 (Forssman) as indicated. Lanes 2, 4, 6, Neutral GSLs of untreated 5072, 5410, 5831 Gaucher cell lines. Lanes 3, 5, 7 Neutral GSLs of CsA-treated 5072, 5410, 5831 cell lines. (B–F) Fabry lymphoblastoid cell line. Cells were grown with 14C-serine. 14C-Radio- labeled GSLs were detected by phosphoimaging. (B) Orcinol detection of total neutral GSL fraction, (C) VT1 overlay of panel B to detect Gb3 only, (D) 14C-metabolic radiolabeled GSL phosphoimage of panel B. Lane 1, GSL standards as in (A, lane 1); lane 2, untreated cells; lane 3, CsA-treated cells. The 14C-radiolabeled species below Gb3 were not characterized. (E) The ganglioside fraction from14C-labeled Fabry cells was separated by TLC (C ⁄ M ⁄ W 60 : 25 : 10 0.2 M CaCl2 v ⁄ v ⁄ v) and detected using orcinol (GM3), or (F) phosphoimaging of the 14C-metabolic labeled species. Lane 1, ganglioside standards GM2 and GM1 as indicated; lane 2, untreated cells; lane 3, CsA-treated cells. The accumulated lymphoid GlcCer in Gaucher cells was eliminated by CsA. The extent to which more complex neutral GSLs were reduced varied between cell lines. CsA treatment of Fabry cells signiﬁcantly reduced the Gb3 and neutral GSL content without effect on the ganglioside proﬁle.

major ganglioside present but additional, more com- plex gangliosides were detected by metabolic labeling. translocation results

Because the Fabry mouse has no a-galactosidase activity and already accumulated Gb3 cannot therefore turnover, we designed a treatment protocol in which the effect of MDR1 inhibition by CsA on accumula- tion of Gb3 via de novo synthesis was assessed.

Tissue Gb3 expression

conﬁrming MDR1 inhibition

case, CsA was found to delete GlcCer and reduce other neutral GSLs present. Inhibition of MDR1- in increased mediated GlcCer access to the cytosolic glucocerebrosidase [34] which is not defective in Gaucher LSD. CsA treatment of a Fabry B-cell line (Fig. 1B–F) also showed signiﬁcant inhibition of accumulated Gb3, monitored by orcinol stain (Fig. 1B) and VT1 ⁄ TLC overlay (Fig. 1C). This indicates residual a-galactosidase activity in this cell line. Metabolic labeling of neutral GSLs (including Gb3) within the Fabry cell line was prevented by CsA (Fig. 1D), reduces de novo Gb3 synthesis. Steady-state levels (Fig. 1E) and metabolically labeled (Fig. 1F) gangliosides in this Fabry cell line were unaffected by CsA. GM3 is the The Gb3 expression proﬁle for various tissues from wild-type and Fabry mice was ﬁrst compared by VT1 TLC overlay (Fig. 2). The Gb3 content was marked increased in the kidney, spleen and liver of Fabry mice. A detectable increase was also observed in the heart.

B

A

Fig. 2. Comparison of the Gb3 content of wild-type and Fabry mouse tissues. GSLs were separated by TLC (C ⁄ M ⁄ W 65 : 25 : 4 v ⁄ v ⁄ v) and visualized using orcinol spray for carbohydrate (A) or VT1 overlay (B) to detect Gb3. Lane 1, GSL standards, from the top: GlcCer, GalCer, LacCer, Gb3, Gb4, Gb5; lanes 2, 4, 6, 8, wild-type; lanes 3, 5, 7, 9, Fabry, lanes 2, 3 heart; lanes 4, 5, spleen; lanes 6, 7 kidney; lanes 8, 9 liver.

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2066

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

ERT and CsA treatment

Effect on serum Gb3

The most notable elevation was seen surprisingly, in the liver in which, under the conditions used, Gb3 was undetected in the wild-type. This indicates Gb3 must undergo rapid turnover in the normal liver. Alternat- ively, liver Gb3 may accumulate via increased serum Gb3 clearance rather than de novo synthesis. Gb3 syn- thase is present in the liver, however [35,36], suggesting de novo synthesis in liver endothelial cell subsets and scavenger accumulation in Kupffer cells (see histology below).

Renal Gb3

Owing to the difﬁculty of drug administration in neo- nates and the availability of well-documented CsA dos- age protocols for adult mice, it was decided that our initial studies on the feasibility of MDR1 inhibition as a potential treatment should be carried out in adult Fabry animals after ERT. a-Galactosidase treatment will eliminate the serum Gb3 levels [9] and the effect of a maintenance dosing of CsA on the recovery of serum Gb3 levels after termination of ERT was determined. ERT is an effective means of eliminating serum Gb3, and Gb3 remained subsequently undetectable in the serum of any animal until 9 weeks post ERT. At this time, the serum Gb3 has recovered for most control mice, whereas the level reached by the CsA-treated mice is reduced by (cid:1) 50% (Fig. 4; P ¼ 0.028). The recovery of serum Gb3 post ERT was found to be, to some extent, variable and some mice within both the control and treated groups did not recover detectable serum Gb3 by the 9-week experiment termination. For responding mice, CsA-treated Fabry mice had serum Gb3 levels of 3.31 ± 1.33 ngÆlL)1 compared with con- trol Fabry serum levels of 8.21 ± 2.27 ngÆlL)1 as determined from a standard curve.

Serum Gb3 levels were monitored in all animals throughout the experimental period. However, at the termination of the experiment a random selection of from control and CsA-treated mice were organs assigned for either GSL extraction or VT1 ⁄ immunohis- tological evaluation.

Effect of CsA on tissue Gb3

Renal Gb3 is the verotoxin receptor responsible for the development of hemolytic uremic syndrome (HUS) in man [14]. HUS is a renal glomerular disease. Gb3 is found in tubules and glomeruli in man [14]. There is no adequate small animal model of VT1-induced HUS because Gb3 is not found in rodent renal glomeruli [13]. Gb3 is present in rodent renal tubules and VT1 induces renal tubular necrosis [37]. We considered that the increased renal Gb3 of the Fabry mouse might extend to the glomerulus to provide a model of the human dis- ease. In the cortex of wild-type kidney, subpopulations of renal tubules were VT1 stained but glomeruli were unreactive, consistent with our previous studies [13]. However, although the VT1 staining of renal tubules is dramatically increased in the Fabry mouse (Fig. 3A), compared with the sporadic VT1 staining seen in the wild-type animal [13], the glomeruli of Fabry mouse kidney remain completely unstained. In Fabry kidney, virtually all cortical tubules were now stained. This indicates that Gb3 is synthesized in all renal tubules in wild-type mice but is rapidly degraded in the majority. VT1 renal tubular targeting in vivo (Fig. 3B) was also signiﬁcantly increased relative to wild-type mice [13], suggesting that Fabry mice should be hypersensitive to VT1. The deparafﬁnization necessary for immunostain- ing precludes identiﬁcation of the tubule type stained. Under the experimental conditions used, no in vivo staining of wild-type kidney tubules was seen (not shown). As with the VT1 cryosection staining, VT1 did not target the renal glomeruli of Fabry mice in vivo.

Serum Gb3

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2067

GSLs were extracted from kidney and liver of control and CsA-treated Fabry mice after 9 weeks recovery post ERT, and from CsA-treated and untreated wild- type mice. The Gb3 content was assessed via VT1 overlay. CsA-dependent differences were seen only in the liver (Fig. 5). The Gb3 content of the liver was increased in Fabry mice and this was reduced in CsA- treated, compared with control mice after recovery from ERT. CsA treatment reduced the liver Gb3 con- tent overall by (cid:1) 50% (P ¼ 0.013). Renal Gb3 content was much greater but signiﬁcant changes after ERT and CsA treatment were not seen (Fig. 5). GM2 gan- glioside is the major ganglioside of mouse liver [39] and the only ganglioside we detected in the Fabry liver GSL extract. GM2 levels were similar in Fabry and wild-type mouse liver. Comparison of Gb3 and GM2 levels (Fig. 5G) clearly show that although liver Gb3 The low level of Gb3 in the Fabry mouse serum and the small volumes available precluded the use of TLC overlay to detect Gb3. A more sensitive VT1-based ELISA assay was used [38]. This assay was linear < 60 ng standard Gb3 and was able to detect > 1 ng Gb3 per lL serum sample. Gb3 in the serum of wild- type mice was below the background of this assay.

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

A

b

a

d

c

B

Fig. 3. Comparison of VT1 staining of wild-type and Fabrys kidney tissue. (A) VT1 staining of cryosections. (a, b) Fabry, (c, d) wild-type kidney cortex. Magniﬁcations: (a) ·16, (b–d) ·40. Glomeruli are marked by arrows. VT1 staining is brown. The section is counter stained with hema- toxylin. (B) In situ staining of renal VT1 bound in vivo. VT1 (50 lg per mouse) injected i.p. and bound within the kidney was immunostained with anti-VT1(without counterstain) in ﬁxed sections after parafﬁn removal. The Fabry cortical section is shown. VT1 in vivo renal tubular targeting is signiﬁcantly increased in the Fabry mouse compared with the 5–10 VT1-labeled tubules which would be seen in an equivalent normal mouse kidney ﬁeld [13]. No VT1 containing glomeruli are seen. Magniﬁcation: ·16.

levels are reduced in the extracts of CsA-treated mice, the level of GM2 is unaffected.

Gb3 tissue histochemistry

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2068

The localization of Gb3 within frozen sections of liver and selected tissues from Fabry and wild-type mice monitored by VT1 binding is shown in Fig. 6. VT1 Gb3 staining was not above background in the wild- type mouse liver (Fig. 6A), but within the Fabry liver VT1 binding detected Gb3 in the stellate Kupffer cells, distributed throughout the section, and in cells lining the portal triad. The levels detected in Fabry mouse liver were reduced in ERT Fabry animals maintained

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

Gb3 positive and many Kupffer cells expressed Gb3 (Figs 6Ba–d,ij). Some vessels within the portal triad were also stained (Fig. 6B d). Extracellular matrix staining in the triad was seen. In the livers of CsA-trea- ted Fabry mice, VT1 staining of Kupffer cells was greatly reduced (Fig. 6Be-h). Gb3 expression in central and portal vein endothelial cells was signiﬁcantly reduced and many vessels negative for Gb3 were observed in CsA-treated mice. Portal triad staining was largely unaffected by CsA treatment (Fig. 6B k,l).

Fig. 4. Effect of CsA treatment on the serum Gb3 levels in Fabry mouse. Serum Gb3 assessment at 9 weeks post ERT. Control Fabry mice (n), CsA-treated Fabry mice ()). Serum Gb3 levels for CsA-treated mice are (cid:1) 50% less than control, P ¼ 0.028.

into the myoﬁbrils, is also evident

In the heart, VT1 staining of a subpopulation of lar- ger blood vessel endothelial cells was seen only in the Fabry mouse (Fig. 7A, compared to Fig. 7B). A patch- work staining which originates from a subset of ﬁbro- cytes between the cardiac muscle ﬁbers and appears to ‘diffuse’ in the Fabry mice (Fig. 7A). The VT1 binding in the Fabry mouse lung (Fig. 7C) was increased in the bronchiolar epithelium. Staining of bronchiolar epithelial cells in the lung was signiﬁcantly elevated compared with the wild-type (Fig. 7D). Although MDR1 is detected in the heart and lung [40], this staining was not consis- tently altered after CsA treatment. There was virtually on CsA during recovery (Fig. 6B) compared with ERT Fabry mice that recovered without CsA. In ERT Fabry recovery control mice, subsets of endothelial cells in the central (Fig. 6B c,d,ij) and portal (Fig. 6B d) vein were

Wild-type liver

Fabry liver

GlcCer

sphingomyelin GSLs C

A

B

LacCer

Gb3 Gb4 Gb5 GM3 GM2 GM1

9

10

1

2

3

4

5

6

7

8 11 12 13 14

1

2

3

4

5

6

7

8

9

10

1

2

3

4 Wild-type kidney

Fabry kidney

Fabry liver GSLs

E

D

Fabry liver sphingomyelin F

G

1

8

2

3

4

5

6

7

9

1

2

3

4

5

6

7

8

9

10

1

2

3

4

5

1

2

3

4

5

Fig. 5. Comparison of Gb3 levels in wild-type and Fabry mouse liver and kidney: relative effect of CsA on Gb3 compared with other sphingo- lipids. (A, B, C, F, G) Liver extracts, (D, E) kidney extracts. (A, C, D) Wild-type, (B, E, F, G) Fabry mice extracts, as indicated. (+) Marks extracts from CsA-treated Fabry mice. Neutral GSLs (A, B, D, E) were separated in C ⁄ M ⁄ W 65 : 25 : 4 v ⁄ v ⁄ v and neutral and acidic GSLs (C, F, G) were separated in C ⁄ M ⁄ 0.8% KCl aq. 60 : 40 : 8 v ⁄ v ⁄ v. Gb3 detection by VT1 ⁄ TLC overlay. (A, B, D, E) Lanes 1–3, 0.5, 1, 2 lg Gb3 standard; lane 4, GM3 ganglioside standard; (B) lanes 5, 7, 9–11 CsA-treated Fabry mice; lanes 6, 8, 12–14 control Fabry mice; (E) lanes 5, 6, 10, CsA-treated Fabry mice; lanes 7–9, control Fabry mice. Liver sphingomyelin and ganglioside detection, (C) (lanes 1,2) and (F) iodine vapour detects liver sphingomyelin (marked * C); (C) (lanes 3, 4) and (G) orcinol spray detects liver GSLs-resorcinol reactive GM2 ganglioside is arrowed. (C) lanes 1 and 3, GSL standards: from the top GlcCer, LacCer, Gb3,Gb4,Gb5 (Forssman), GM3, GM2,GM1; lanes 2 and 4, lipid extract of wild-type liver; (F, G) lane 1, GSL standards; lanes 2 and 4, lipid extracts of control Fabry liver; lanes 3, 5, lipid extracts of CsA- treated Fabry liver. Gb3 is only detected in Fabry, as opposed to wild-type liver (compare A with B, and C with G) and the normal renal Gb3 doublet (D) is markedly enhanced in the Fabry mouse (E). Less Gb3 is detected in the liver of CsA-treated, compared with control Fabry mice (B, G). Although the level of Gb3 is reduced by CsA treatment, the levels of GM2 ganglioside (similar in wild-type, indicated by arrow, and Fabry mouse liver; compare C with G), and sphingomyelin (F) are unaffected. The Gb3 detected in (B) was subject to densitometry and com- pared. The CsA-treated Fabry liver Gb3 values were reduced by 45% (P ¼ 0.013) compared with controls.

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2069

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

A

a

b

B

b

d

a

c

g

e

f

h

j

i

k

l

Fig. 6. Verotoxin staining of frozen liver sections from Fabry mice treated ± CsA. (A) Wild-type (a) compared with Fabry liver (b). VT1 staining in wild-type liver is undetectable. Arrows in (b) indicate some of the VT1-stained (brown) Kupffer cells in the Fabry mouse liver. (B) (a–d, I, j) Untreated, (e-h, k, l) CsA-treated Fabry mice. (Liver sections from three individual mice in each category are shown.) Magniﬁcation: (a–c, e–g) ·40; (d, g, i–l) ·16. * ¼ central veins, p ¼ portal veins. Inserts in (a) and (b) show Kupffer cell staining, and in (d) portal vein endothelial VT1 cell staining. Most VT1 staining is lost after CsA treatment but portal triad staining was retained.

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2070

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

Fig. 7. Comparison of Verotoxin staining of other tissues. (A, C, E) Fabry, (B, D, F) wild- type tissue, (insets) CsA-treated Fabry. (A, B) Heart – endothelial staining in Fabry mouse; (C, D) lung – epithelial cell staining increased in Fabry mouse; (E, F) brain micro- vascular endothelial staining in Fabry mouse. Magniﬁcation ·16.

inhibit a-glucosidases as well as glucosyl transferases [22,23].

no VT1 staining of normal brain (Fig. 7F). In Fabry brain, extensive staining of the microvasculature is evi- dent (Fig. 7E). The arachnoid membrane surrounding the brain is extensively stained in Fabry but not nor- mal mouse brain. Although MDR1 is highly expressed in the brain microvasculature [41], ERT is not effective to reduce the level of Gb3 in the brain [9].

Discussion

Inhibitors of glucosyl ceramide The role of MDR1 in GSL synthesis, though estab- lished in vitro, has yet to be understood in vivo. MDR1 knockout mice do not show an overt phenotype, although skin ﬁbroblasts from such mice are, as pre- dicted, defective in neutral GSL synthesis [28]. We showed that an alternative mechanism of Golgi mem- brane GlcCer translocation must exist in HeLa cells [28] because their neutral GSLs are unaffected by CsA. Whereas the liver of MDR1 knockout mice show a GSL complement consistent with the translocase func- tion of MDR1, the GSLs of some other tissues are complicated by the redundancy in this function (stud- ies in progress) and the tissue differences in MDR1 expression. Thus, an effect of MDR1 inhibition on GSL biosynthesis in vivo was by no means assured.

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2071

The differential sensitivity of ganglioside and neutral GSL synthesis to depletion of GlcCer via MDR1 inhi- bition [28] provides an attractive method for the select- ive reduction of neutral GSL synthesis in neutral GSL storage diseases. Other substrate reduction approaches are less selective and hence have greater potential side- effects. synthase prevent the synthesis of both neutral GSLs and gan- gliosides. Although the lack of GSLs can be tolerated in cultured cells [42], the glucosyl ceramide synthase knockout mouse is embryonic lethal [43]. Imino sugars The possibility of using MDR1 inhibition as a new approach to neutral GSL storage diseases is supported by our ﬁnding that CsA completely reverses GSL accu- mulation in Gaucher lymphoblasts, in which there is

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

on prolonged treatment or treatment prior to GSL accumulation. The increased in vivo VT1 renal target- ing in the Fabry mouse suggests increased susceptibil- ity to this toxin compared with wild-type, but the retained lack of glomerular binding indicates that the Fabry mouse will not serve as a model of HUS in man. The increased Gb3 expression in virtually all the renal tubules of the Fabry mouse shows that the lack of Gb3 detection in most tubules of the wild-type mouse [13] is a result of rapid Gb3 turnover, rather than the lack of Gb3 synthesis. A similar effect in man could be important in determining susceptibility to HUS following VTEC infection.

the primary focus of our therefore

an alternative cytosolic mechanism for breakdown [34]. The signiﬁcant effect in Fabry lymphoblasts to reduce Gb3 without effect on ganglioside synthesis supports this approach. Prevention of Gb3 synthesis is the only feasible stratagem in the Fabry mouse and in those Fabry patients with no residual a-galactosidase activ- ity. In such cases, Gb3 already accumulated would not be reversed by MDR1 inhibition, or other mechanisms of substrate-reduction therapy. A protocol using adult Fabry mice was designed to test the efﬁcacy of MDR1 inhibition on de novo Gb3 synthesis, whereby animals were treated by ERT and the effect of CsA on ‘relapse’ of Gb3 accumulation monitored. ERT primarily affects serum and liver Gb3 accumulation [9] and these tissues were study, although the location of Gb3 accumulated in other tis- sues was also investigated.

reductions we have than the modest Our results indicate the feasibility of using inhibition of MDR1 as an approach to the treatment of Fabry disease. Although the efﬁcacy may not, as yet, be as inhibition of MDR1 may prove dramatic as ERT, most beneﬁcial as an adjunct, rather than alternative to ERT. It is clear that the dosage and treatment per- iod in this model needs optimization and the effect of maintenance MDR1 inhibition from birth requires investigation. In addition, more selective inhibitors of MDR1 than CsA are available. CsA is, however, clin- ically used long-term and it might be expected that under such conditions, the effect on Fabry patient tis- sue Gb3 levels might be accumulative and more signiﬁ- cant seen following brief treatment of the Fabry mouse.

Our demonstration that CsA signiﬁcantly reduces Fabry mouse serum and liver Gb3 levels, approaches proof of concept. Total Gb3 extracted from the liver was reduced yet the level of GM2 ganglioside, the only ganglioside we detected in Fabry mouse liver, was not reduced by CsA treatment. This is consistent with our cell culture [28] and Fabry lymphoblast studies in which MDR1 inhibition was found to prevent neutral but not acidic GSL synthesis. Thus in vivo (at least within the liver), as well as in cell culture, GlcCer translocated to the Golgi lumen by MDR1 is a precur- sor for neutral GSL but not ganglioside synthesis. This preferential effect on neutral GSL biosynthesis might be considered as a ‘signature’ for MDR1 involvement.

Other GSL storage diseases in which a similar approach might be beneﬁcial would include Gaucher. In this case, inhibition of GlcCer Golgi translocation should increase exposure to the cytosolic glucosidase (not deﬁcient in Gaucher disease) to effect a reduction in GSL accumulation.

sites, the Gb3

ERT is clinically effective in Fabry patients [4] but neurological symptoms are not addressed and treat- ment with the missing a-galactosidase is extremely costly, such that it is not universally available. The search for alternative or complementary treatment strategies continues [18,19,44]. Our studies suggest a new approach to the inhibition of substrate synthesis. Future work with neonatal Fabry mice is required to establish a ‘proof of principle’ as to the efﬁcacy of an MDR1 inhibition approach.

VT1 staining of Kupffer cells and endothelial cells within the central vein showed signiﬁcantly less Gb3 accumulation after CsA treatment. Because the phago- cytic Kupffer cells are major reticulo-endothelial de- they contain could be gradative serum ⁄ red blood cell-derived and the decrease seen after CsA result from the reduced serum Gb3 levels. Kupffer cells are modiﬁed monocytes that share a common origin with endothelial cells. However, we believe that this is unlikely to be the case because endothelial cell staining within the liver was also reduced and mice in which serum Gb3 was found to remain undetectable after ERT, were nevertheless found to have Gb3 in the hepatic extract and express Kupffer cell Gb3.

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2072

Heart, lung, brain, kidney and spleen tissue show a clear increase in Gb3 staining in the Fabry, compared with normal mouse but this was not obviously affected by the current CsA protocol. However, because these tissues are less sensitive than liver to ERT [9], the potential beneﬁt of CsA in these tissues might accrue In summary, CsA treatment has been found to reduce the recovery of serum Gb3 levels in Fabry mice following a-galactosidase treatment. In such mice, the expression of Gb3 within the liver is also reduced in comparison with Fabry mice allowed to recover from ERT without MDR1 inhibition. These studies indicate that MDR1 inhibition represents a potential novel adjunct to the current treatment of neutral GSL stor- age diseases.

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

Experimental procedures

EBV-transformed B-lymphoblastoid cell lines from Gaucher type 1 and Fabry disease (kindly supplied by J. Clarke, Hospital for Sick Children, Toronto, Canada) were cul- tured in RPMI +15% fetal bovine serum (FBS) ± 4 lm CsA for 4 days. CsA induces a < 10% reduction in growth rate, compensated for in analyses. Neutral GSLs from equal cell numbers were extracted and separated by TLC as described previously [28]. The ganglioside fraction was pre- pared by anion exchange [28].

CsA treatment of LSD cultured cells

below the detection limit. Gb3 standard was quantitated by sphingosine assay using the method of Naoi et al. [45]. The plates were placed at 37 (cid:1)C overnight to evaporate the sol- vent. All subsequent incubations were performed for 1 h 37 (cid:1)C and washes at room temperature. Wells were blocked with 150 lL of 0.2% bovine serum albumin (BSA) in 50 mm Tris-buffered saline pH 8.0 (BSA-TBS) then washed twice with BSA-TBS. Wells were subsequently incubated with 200 ng per well of VT1 in BSA-TBS, rabbit antiserum against the VT1 B subunit, diluted 1 ⁄ 2000 in BSA-TBS, and ﬁnally goat anti-(rabbit HRP)-conjugate (Bio-Rad Laboratories, Hercules, CA) diluted 1 ⁄ 2000 in BSA-TBS (all 50 lL per well). Verotoxin binding in the wells was visu- alized by incubation with 100 lL per well of 0.5 mgÆmL)1 ABTS in citrate-phosphate buffer, pH 4.0. Absorbance was measured at 405 nm after 30–40 min of colour development at room temperature. Serum Gb3 values were assessed for signiﬁcance using a transformed two-sample Student’s t-test assuming equal variances.

Treatment of Fabry mice

Tissues were homogenized, extracted in 20 vol. chloro- form ⁄ methanol (2 : 1 v ⁄ v) and ﬁltered. The extract was dried under N2 and saponiﬁed overnight in 0.1 n NaOH in MeOH at 37 (cid:1)C [11]. The glycolipid extract was neutral- ized, partitioned against water and was used for VT1 TLC overlay without further puriﬁcation. GSL extracted from 0.5 mg wet weight organ were applied per sample.

Twelve adult mice were treated i.p. with a bolus injection of a-galactosidase (1.5 mgÆkg)1). Six mice were then injec- ted twice a week i.p. with CsA (30 lgÆg)1) and the remain- ing mice served as controls. Similarly six wild-type mice were maintained on CsA and six animals were left untreated. Serum Gb3 levels were monitored for nine weeks post ERT at which time some organs (wild-type and Fabry) were processed for Gb3 extraction, whereas others were processed for VT1 staining of cryosections. Experimentat- ion using the Fabry mouse is necessary to demonstrate the in vivo potential of MDR1 inhibition as an approach to treatment of Fabry disease in man and was carried out under ethical approval. Mice were euthanized under condi- tions of minimized trauma.

Tissue Gb3 extraction

For ganglioside and Gb3 comparison in Fabry liver, the saponiﬁed extract was desalted on a SepPak cartridge after neutralization and total GSL separated by TLC. Sphingo- myelin was detected by iodine, gangliosides by resorcinol and total GSLs by orcinol spray. In this case, lipids equiv- alent to 2 mg liver were applied per sample.

Determination of Gb3 levels in Fabry mouse plasma was performed effectively as described by Zeidner et al. [38].

VT1 TLC overlay of the GSL tissue extracts to detect Gb3 was performed as described [46]. Some TLC overlays were subject to comparative densitometry using the image j 1.34 program. Values were compared using an unpaired Student’s t-test.

Extraction of plasma Gb3 and quantitation by VT1 ELISA

Plasma samples ((cid:1) 20–60 lL) were prepared weekly dur- ing the study and stored at )20 (cid:1)C. End-point plasma vol- umes ranged from 100 to 400 lL. For lipid isolation, plasma samples were extracted overnight with 2 mL of chloroform ⁄ methanol (2 : 1 v ⁄ v) per 100 lL of plasma and then partitioned against 1 ⁄ 5 volume of water. The lower phase was dried under a stream of nitrogen gas and then the residue was dissolved in chloroform. The sample was applied to a silica gel 60 column ((cid:1) 100 mg of silica per 100 lL plasma volume). Neutral lipids were removed by washing with 4 column volumes of chloroform and then neutral glycosphingolipids were eluted with 10 column vol- umes of acetone ⁄ methanol (5 : 1 v ⁄ v). The eluate was dried, dissolved in 10 times the original plasma volume of ethanol and stored at )20 (cid:1)C.

Five-micrometer frozen tissue sections were air-dried over- night at room temperature on the lab bench. When dry, a PAP hydrophobic barrier pen was used to encircle sections. Throughout all incubation steps, slides were kept in a humid chamber at room temperature Sections were blocked with endogenous peroxidase blocker (Universal Block, KPL Inc., Gaithersburg, MD) for 20 min. After extensive rinses with 1· NaCl ⁄ Pi solution, sections were blocked with 1% normal goat serum ⁄ NaCl ⁄ Pi (NGS–NaCl ⁄ Pi) for 20 min. Without washing, sections were then stained with VT1 (200 ngÆmL)1 in NGS–NaCl ⁄ Pi) for 30 min. After ﬁve vigorous rinses with NaCl ⁄ Pi, sections were incubated with rabbit anti-(VT1B

Plasma extracts (50 lL) were added to duplicate ELISA plate wells (Nunc Polysorp DiaMed Mississauga, ON). Dilutions of standard human kidney Gb3 in ethanol were also plated in triplicate. Serum Gb3 levels < 2 ngÆmL)1 were

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2073

VT1 tissue staining

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

Gottesman MM, Brady RO et al. (1997) alpha-Galac- tosidase A deﬁcient mice: a model of Fabry disease. Proc Natl Acad Sci USA 94, 2540–2544.

8 Chiba Y, Sakuraba H, Kotani M, Kase R, Kobayashi

K, Takeuchi M, Ogasawara S, Maruyama Y, Nakajima T, Takaoka Y et al. (2002) Production in yeast of alpha-galactosidase A, a lysosomal enzyme applicable to enzyme replacement therapy for Fabry disease. Glyco- biology 12, 821–828.

9 Ioannou YA, Zeidner KM, Gordon RE & Desnick RJ

6869) (1 : 1000 in NGS–NaCl ⁄ Pi) for 30 min, washed and then incubated with HRP-conjugated goat anti-(rabbit IgG) (1 : 500 in NGS–NaCl ⁄ Pi) for 30 min. Following washing, sections were developed using DAB substrate for 5 min. To stop the DAB reaction, sections were dipped in distilled water for 4 min. Hematoxylin counterstain was applied for 30 s; excess staining was removed by immersing sections in distilled water for 4 min and ‘blued’ by immersing in tap water for 4 min. Sections were then dehydrated for 2 min. in each of 70%, 95% and 100% ethanol, cleared in xylene for 5 min and mounted in Permount.

For in vivo VT1 distribution, 50 lg VT1 was injected i.p. Mice were killed after one hour and organs removed, ﬁxed, sectioned and deparafﬁnized sections stained with anti-VT1 as described previously [13] without counterstain.

Acknowledgements

(2001) Fabry disease: preclinical studies demonstrate the effectiveness of alpha-galactosidase A replacement in enzyme-deﬁcient mice. Am J Hum Genet 68, 14–25. 10 Eitzman DT, Bodary PF, Shen Y, Khairallah CG, Wild SR, Abe A, Shaffer-Hartman J & Shayman JA (2003) Fabry disease in mice is associated with age-dependent susceptibility to vascular thrombosis. J Am Soc Nephrol 14, 298–302.

11 Boyd B & Lingwood CA (1989) Verotoxin receptor gly- colipid in human renal tissue. Nephron 51, 207–210. 12 Lingwood CA (1994) Verotoxin-binding in human renal

sections. Nephron 66, 21–28.

This work was supported by CIHR grant #MT13073 (CAL) and NIH grant #HL70569 (JAM). We thank Dr J. Phillips (Dept Pediatric Laboratory Medicine HSC), for help in liver histology analysis.

References

13 Rutjes N, Binnington B, Smith C, Maloney M & Ling- wood C (2002) Differential tissue targeting and patho- genesis of Verotoxins 1 and 2 in the mouse animal model. Kid Intl 62, 832–845.

1 Schuette CG, Doering T, Kolter T & Sandhoff K

14 Chark D, Nutikka A, Trusevych N, Kuzmina J & Ling- wood C (2004) Differential carbohydrate epitope recog- nition of globotriaosyl ceramide by verotoxins and monoclonal antibody: Role in human renal glomerular binding. Eur J Biochem 271, 1–13.

(1999) The glycosphingolipidoses – from disease to basic principles of metabolism. Biol Chem 380, 759–766. 2 Brady RO (2003) Gaucher and Fabry diseases: from understanding pathophysiology to rational therapies. Acta Paediatr Suppl 92, 19–24.

15 Yoshimitsu M, Sato T, Tao K, Walia JS, Rasaiah VI,

3 Heukamp LC, Schroder DW, Plassmann D, Homann J

Sleep GT, Murray GJ, Poeppl AG, Underwood J, West L et al. (2004) Bioluminescent imaging of a marking transgene and correction of Fabry mice by neonatal injection of recombinant lentiviral vectors. Proc Natl Acad Sci USA 101, 16909–16914.

& Buttner R (2003) Marked clinical and histologic improvement in a patient with type-1 Gaucher’s disease following long-term glucocerebroside substitution. A case report and review of current diagnosis and manage- ment. Pathol Res Pract 199, 159–163.

16 Lee L, Abe A & Shayman JA (1999) Improved inhibi- tors of glucosylceramide synthase. J Biol Chem 274, 14662–14669.

4 Wilcox WR, Banikazemi M, Guffon N, Waldek S, Lee P, Linthorst GE, Desnick RJ & Germain DP (2004) Long-term safety and efﬁcacy of enzyme repla- cement therapy for Fabry disease. Am J Hum Genet 75, 65–74.

5 Schiffmann R, Kopp JB, Austin HA 3rd, Sabnis S,

17 Abe A, Arend LJ, Lee L, Lingwood CA, Brady RO & Shayman JA (2000) Glycosphingolipid depletion in Fabry disease lymphoblasts with potent inhibitors of glucosylceramide synthase. Kid Intl 57, 446–454.

Moore DF, Weibel T, Balow JE & Brady RO (2001) Enzyme replacement therapy in Fabry disease: a rando- mized controlled trial. JAMA 285, 2743–2749.

18 Abe A, Wild SR, Lee WL & Shayman JA (2001) Agents for the treatment of glycosphingolipid storage disorders. Curr Drug Metab 2, 331–338.

19 Platt FM, Jeyakumar M, Andersson U, Heare T, Dwek RA & Butters TD (2003) Substrate reduction therapy in mouse models of the glycosphingolipidoses. Phil Trans R Soc Lond B Biol Sci 358, 947–954.

20 Zimran A & Elstein D (2003) Gaucher disease and the

6 Schiffmann R, Murray GJ, Treco D, Daniel P, Sellos- Moura M, Myers M, Quirk JM, Zirzow GC, Borowski M, Loveday K et al. (2000) Infusion of a-galactosidase A reduces tissue globotriaosylceramide storage in patients with Fabry disease. Proc Natl Acad Sci USA 97, 365–370.

7 Ohshima T, Murray GJ, Swaim WD, Longenecker G,

clinical experience with substrate reduction therapy. Phil Trans R Soc Lond B Biol Sci 358, 961–966.

Quirk JM, Cardarelli CO, Sugimoto Y, Pastan I,

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2074

M. Mattocks et al.

MDR1 inhibition and GSL storage disease

cytosolic broad-speciﬁcity beta-glucosidase activity. Clin Chim Acta 216, 11–21.

21 Futerman AH, Sussman JL, Horowitz M, Silman I & Zimran A (2004) New directions in the treatment of Gaucher disease. Trends Pharmacol Sci 25, 147–151. 22 Tian G, Wilcockson D, Perry VH, Rudd PM, Dwek

RA, Platt FM & Platt N (2004) Inhibition of alpha-glu- cosidases I and II increases the cell surface expression of functional class A macrophage scavenger receptor (SR-A) by extending its half-life. J Biol Chem 279, 39303–39309.

23 Andersson U, Reinkensmeier G, Butters TD, Dwek RA & Platt FM (2004) Inhibition of glycogen breakdown by imino sugars in vitro and in vivo. Biochem Pharmacol 67, 697–705.

35 Steffensen R, Carlier K, Wiels J, Levery SB, Stroud M, Cederen B, Nilsson SB, Bennett EP, Jersild C & Clau- sen H (2000) Cloning and expression of the histo-blood group Pk UDP-galactose: Galbeta 1–4Glcbeta 1-Cer alpha 1, 4-galactosyltransferase. Molecular genetic basis of the p phenotype. J Biol Chem 275, 16723–16729. 36 Kojima Y, Fukumoto S, Furukawa KTO, Wiels J, Yok- oyama K, Suzuki Y, Urano T & Ohta M (2000) Mole- cular cloning of globotriaosylceramide ⁄ CD77 synthase, a glycosyltransfease that initiates the synthesis of globo series glycosphingolipids. J Biol Chem 275, 15152– 15156.

24 Lannert H, Gorgas K, Meißner I, Wieland FT & Jeckel D (1998) Functional organization of the Golgi appara- tus in glycosphingolipid biosynthesis. J Biol Chem 273, 2939–2946.

37 Wadolkowski EA, Sung LM, Burris JA, Samuel JE & O’Brien AD (1990) Acute renal tubular necrosis and death of mice orally infected with Escherichia coli strains that produce Shiga-like toxin type II. Infect Immun 58, 3959–3965.

25 van Helvoort A, Smith A, Sprong H, Fritzsche I, Schin- kel A, Borst P & van Meer G (1996) MDR1 P-Glyco- protein is a lipid translocase of broad speciﬁcity, while MDR3 P-glycoprotein speciﬁcally translocates phospha- tidyl choline. Cell 87, 507–517.

26 Eckford PD & Sharom FJ (2005) The reconstituted

38 Zeidner KM, Desnick RJ & Ioannou YA (1999) Quan- titative determination of globotriaosylceramide by immunodetection of glycolipid-bound recombinant vero- toxin B subunit. Anal Biochem 267, 104–113.

P-glycoprotein multidrug transporter is a ﬂippase for glucosylceramide and other simple glycosphingolipids. Biochem J 389, 517–526.

39 Suzuki M, Nakamura K, Hashimoto Y, Suzuki A & Yamakawa T (1986) Mouse liver gangliosides. Carbo- hydr Res 151, 213–223.

27 Lala P, Ito S & Lingwood CA (2000) Transfection of MDCK cells with the MDR1 gene results in a major increase in globotriaosyl ceramide and cell sensitivity to verocytotoxin: role of P-gp in glycolipid biosynthesis. J Biol Chem 275, 6246–6251.

40 Croop JM, Raymond M, Haber D, Devault A, Arceci RJ, Gros P & Housman DE (1989) The three mouse multidrug resistance (mdr) genes are expressed in a tis- sue-speciﬁc manner in normal mouse tissues. Mol Cell Biol 9, 1346–1350.

28 De Rosa MF, Sillence D, Ackerley C & Lingwood C

(2004) Role of multiple drug resistance protein 1 in neu- tral but not acidic glycosphingolipid biosynthesis. J Biol Chem 279, 7867–7876.

41 Schinkel AH, Wagenaar E, Mol CA & van Deemter L (1996) P-Glycoprotein in the blood–brain barrier of mice inﬂuences the brain penetration and pharmacologi- cal activity of many drugs. J Clin Invest 97, 2517–2524. 42 Ichikawa S, Sakiyama H, Suzuki G, Hidari KI & Hira- bayashi Y (1996) Expression cloning of a cDNA for human ceramide glucosyltransferase that catalyzes the ﬁrst glycosylation step of glycosphingolipid synthesis. Proc Natl Acad Sci USA 93, 4638–4643.

29 Lavie Y, Cao H, Bursten SL, Giuliano AE & Cabot MC (1996) Accumulation of glucosylceramides in multidrug- resistant cancer cells. J Biol Chem 271, 19530–19536. 30 Liu YY, Han TY, Yu JY, Bitterman A, Le A, Giuli- ano AE & Cabot MC (2004) Oligonucleotides block- ing glucosylceramide synthase expression selectively reverse drug resistance in cancer cells. J Lipid Res 45, 933–940.

43 Yamashita T, Wada R & Proia RL (2002) Early devel- opmental expression of the gene encoding glucosyl- ceramide synthase, the enzyme controlling the ﬁrst committed step of glycosphingolipid synthesis. Biochim Biophys Acta 1573, 236–240.

44 Siatskas C & Medin JA (2001) Gene therapy for Fabry

31 Norris-Cervetto E, Callaghan R, Platt FM, Dwek RA & Butters TD (2004) Inhibition of glucosylceramide synthase does not reverse drug resistance in cancer cells. J Biol Chem 279, 40412–40418.

disease. J Inherit Metab Dis 24 (Suppl. 2), 25–41.

32 Chin KV & Liu B (1994) Regulation of the multidrug

resistance (MDR1) gene expression. In Vivo 8, 835–841.

33 Ernest S, Rajaraman S, Megyesi J & Bello-Reuss EN

(1997) Expression of MDR1 (multidrug resistance) gene and its protein in normal human kidney. Nephron 77, 284–289.

34 Forsyth G, Romero K, Alverson J, VanderJagt D & Glew R (1993) Variable expression of leukocyte

45 Naoi M, Lee YC & Roseman S (1974) Rapid and sensi- tive determination of sphingosine bases and sphingo- lipids with ﬂuorescamine. Anal Biochem 58, 571–577. 46 Nutikka A, Binnington-Boyd B & Lingwood C (2003) Methods for the identiﬁcation of host receptors for Shiga toxin. In Methods in Molecular Medicine (Philpot D & Ebel F, eds), pp. 197–208. Humana Press, Totowa, NY.

FEBS Journal 273 (2006) 2064–2075 ª 2006 The Authors Journal compilation ª 2006 FEBS

2075