Action myoclonus-renal failure syndrome: diagnostic applications of activity-based probes and lipid analysis

Lysosomal integral membrane protein-2 (LIMP2) mediates trafficking of glucocerebrosidase (GBA) to lysosomes. Deficiency of LIMP2 causes action myoclonus-renal failure syndrome (AMRF). LIMP2-deficient fibroblasts virtually lack GBA like the cells of patients with Gaucher disease (GD), a lysosomal storage disorder caused by mutations in the GBA gene. While GD is characterized by the presence of glucosylceramide-laden macrophages, AMRF patients do not show these. We studied the fate of GBA in relation to LIMP2 deficiency by employing recently designed activity-based probes labeling active GBA molecules. We demonstrate that GBA is almost absent in lysosomes of AMRF fibroblasts. However, white blood cells contain considerable amounts of residual enzyme. Consequently, AMRF patients do not acquire lipid-laden macrophages and do not show increased plasma levels of macrophage markers, such as chitotriosidase, in contrast to GD patients. We next investigated the consequences of LIMP2 deficiency with respect to plasma glycosphingolipid levels. Plasma glucosylceramide concentration was normal in the AMRF patients investigated as well as in LIMP2-deficient mice. However, a marked increase in the sphingoid base, glucosylsphingosine, was observed in AMRF patients and LIMP2-deficient mice. Our results suggest that combined measurements of chitotriosidase and glucosylsphingosine can be used for convenient differential laboratory diagnosis of GD and AMRF.

Monocytes were isolated with CD14 MicroBeads (Miltenyi Biotec) according to the manufacturer's protocol. Isolated monocytes were differentiated into macrophages during 7 days in RPMI medium supplemented with 10% human AB serum ( 3 ).

Mice
Mice were housed and plasma was collected according to the local protocol with approval from the review board of the Christian-Albrechts-University in Kiel (Germany). LIMP2-defi cient mice were generated as described in ( 16 ).
Intact fi broblasts were incubated with MDW941 (5 nM) for 72 h, and subsequently the culture medium was collected and the cells were harvested. Detection of fl uorescent GBA in the culture medium was performed upon capture of the enzyme from 1 ml medium using monoclonal antibody 8E4 immobilized to Sepharose beads ( 25 ).
For fl uorescence-activated cell sorting (FACS), fi broblasts and macrophages were incubated with MDW933 (50 nM) for 5 h in the medium ( 8 ). In the case of white blood cells, leukocytes were collected from freshly drawn blood washed with 0.8% (w/v) ammonium chloride solution and lysing the remaining erythrocytes . Leukocytes were next incubated with MDW933 (100 nM) for 30 min in phosphate buffered saline containing 1% (w/v) BSA. FACS analysis was performed with a FACSCalibur (B.D. Bioscience), ex 488 nm, em 530 nm (bandpass fi lter 30 nm) ( 8 ).

Western blotting
SDS-PAGE gels were electroblotted onto a nitrocellulose membrane (Schleicher and Schuell). Membranes were blocked with 5% skimmed milk and 0.05% Tween-20 in Tris-buffered saline (TBS) for 1 h at room temperature and incubated overnight with the primary antibodies at 4°C. Membranes were then washed three times with 0.01% Tween-20 in TBS and incubated with the appropriate IRDye conjugated secondary antibodies for 1 h at room temperature. After washing, detection was performed using the Odyssey® CLx infrared imaging system.

Measurement of in vivo GBA enzymatic activity
GBA enzymatic activity in intact fi broblasts, leukocytes, and macrophages was measured using 5-(pentafl uorobenzoylamino) fl uorescein di-␤ -D -glucopyranoside (PFB-FDG) ( 26 ) (20 M for with the nucleophile E340 in GBA. Linking a BODIPY moiety to the C6 of cyclophellitol renders an even more potent irreversible ABP of GBA. Different ABPs have been designed by variation of the type of BODIPY (MDW941, Inhibody Red; and MDW933, Inhibody Green). The ABPs spontaneously cross membranes and allow very sensitive labeling of active GBA enzyme molecules in living cells ( 8 ).
The transport of newly formed GBA to lysosomes has been an enigma for a long time. GBA is known not to acquire mannose-6-phosphate moieties (12)(13)(14)(15). The enzyme is not deficient in fi broblasts of mucolipidosis II and III patients suffering from defects in the formation of mannose-6phosphate recognition signals ( 12 ). GBA was found to become membrane bound in the endoplasmic reticulum by interaction with an unknown protein ( 13,14 ). Only a few years ago, the receptor protein was identifi ed as lysosomal integral membrane protein-2 (LIMP2), one of the integral membrane proteins of lysosomes ( 16 ). Next, mutations in the scavenger receptor class B member 2 (SCARB2) gene, encoding LIMP2 protein, were found to cause action myoclonus-renal failure syndrome (AMRF) (MIM*602257), a fatal recessively inherited disorder characterized by glomerulosclerosis, progressive myoclonus epilepsy, ataxia, and accumulation of undefi ned storage material in the brain (17)(18)(19)(20)(21)(22)(23). More recently it has become clear that not all AMRF patients develop renal complications ( 22,23 ). Of interest, AMRF patients do not show the massive occurrence of lipidladen macrophages and similar pathology to GD patients ( 24 ). Consistently, AMRF patients also do not show elevated plasma chitotriosidase. These fi ndings point to cell-type-specifi c consequences of LIMP2 defi ciency. Indeed, LIMP2-defi cient fibroblasts of AMRF patients lack GBA, whereas this appears not to be the case for their white blood cells ( 18 ).
The availability of ABPs allowing sensitive detection of GBA in leukocytes prompted us to study the fate of the enzyme in LIMP2-defi cient cells, obtained from an AMRF patient as well as from LIMP2-defi cient mice. We here report the outcome of these investigations, confi rming cell-type-specifi c reductions in GBA caused by LIMP2 deficiency. Furthermore, we studied the consequences of LIMP2 defi ciency for glucosylceramide and glucosylsphingosine levels, the products of hydrolysis by GBA. We here report the outcome and possible use in the differential diagnosis of AMRF and GD.

Antibodies
The rabbit polyclonal anti-LIMP2 antibody was purchased to Novus Biologicals, Littleton, CO.

AMRF patient materials
Materials from donors were obtained after informed consent. Leukocytes and fi broblasts were obtained from LIMP2-defi cient GBA in patient leukocytes was almost similar when compared with cells from a healthy subject ( Fig. 1C ).
Western blot analysis using anti-LIMP2 antibody showed that LIMP2 protein was expressed at higher levels in fi broblasts compared with leukocytes. LIMP2 was decreased in fi broblasts of AMRF carriers compared with controls ( Fig. 1A, C ). Next, GBA was analyzed in monocyte-derived macrophages. Monocytes were isolated from the blood of an AMRF patient and a healthy subject and differentiated into macrophages. Detection of active GBA in homogenates of the generated macrophages revealed again only a slight reduction of GBA in the patient's cells ( Fig. 1D ).
Next, the amount of active GBA in intact cells was assessed by FACS analysis. Cells were labeled with ABP MDW933, or their GBA activity was determined by incubation with the substrate PFB-FDG ( Fig. 2 ). In fi broblasts of three AMRF patients, active GBA was found to be reduced using ABP labeling ( Fig. 2A ). We also used the cell permeable substrate PFB-FDG for assessing GBA enzymatic activity ( Fig. 2B ). We checked to ascertain whether all detected activity could be ascribed to lysosomal GBA by preincubation with and without CBE, an irreversible inhibitor (data not shown ).
Fibroblasts of an AMRF carrier showed considerable levels of active GBA. Similar experiments showed that the amount of active GBA was relatively high in the blood cells of an AMRF patient ( Fig. 2C ). Signifi cant residual GBA levels (values >30%) were noted in lymphocytes, monocytes, and cultured macrophages of the AMRF patient, both with ABP labeling and PFB-FDG treatment. The same fi nding was made for total leukocytes from LIMP2-defi cient mice ( Fig. 2C ). Measurement of GBA enzymatic activity in cell lysates using the artifi cial substrate 4-methylumbelliferyl-␤ -D -glucoside ( 30 ) gave similar results (data not shown).

Secretion of GBA by LIMP2-defi cient fi broblasts
Misrouting of GBA due to absence of LIMP2 might result in partial secretion of the enzyme to the extracellular space. To study this, cultured fi broblasts were labeled with MDW941 for 72 h and the culture medium was collected. GBA from the medium (1 ml) was immunoprecipitated with the anti-GBA monoclonal antibody (8E4) and visualized by fl uorescence scanning of SDS-PAGE gels ( Fig. 3 ). Indeed, fi broblasts of AMRF patients demonstrated the presence of ABP-labeled GBA in the medium, which was absent in the WT and AMRF carriers. Of note, the molecular mass of GBA seen with SDS-PAGE is determined by its N-linked glycan composition. The enzyme was initially synthesized with four high-mannose type glycans showing a mass of about 62 kDa with SDS-PAGE. In the Golgi apparatus, at least three glycans were modifi ed to sialylated complex type structures rendering masses of 62-66 kDa. After entering the lysosome, the glycans are trimmed by local exoglycosidases resulting in a gradual reduction to 59 kDa ( 31 ). Secreted enzyme largely shows a high molecular mass due to the presence of complex-type glycans, lacking the lysosomal deglycosylation. We presume that the very tiny amounts of lower mass bands, seen in the case of control fi broblasts, 40 M for leukocytes and macrophages) as substrate, exactly as described by Witte et al. ( 8 ). To discriminate the activity of lysosomal GBA from other ␤ -glucosidases, samples were pretreated with and without 1 mM of CBE for 30 min before adding the PFB-FDG. No activity was observed for fi broblasts preincubated with CBE, indicating that activity in the absence of the inhibitor is due to GBA.

Statistical analysis
The results were analyzed using the Student's unpaired t -test. P < 0.05 was considered signifi cant. Data were statistically analyzed using GraphPad Prism 6 software (Graphpad Software, San Diego, CA).

Detection of active GBA in fi broblasts, leukocytes, and macrophages
To detect active GBA molecules, homogenates of fi broblasts from three AMRF patients (LIMP2 W178X/W178X), two AMRF carriers (LIMP2 W178X/WT), two control subjects, and one type 2 GD patient (GBA L444P/L444P) were labeled with MDW941 (Inhibody Red) and subjected to SDS-PAGE. We demonstrated earlier that MDW941 specifi cally labels lysosomal GBA and no other retaining ␤ -glucosidases in humans (GBA2 and GBA3) ( 8,28 ). Actually, cultured fi broblasts contain no GBA3 and very little GBA2 ( 29 ). GBA, labeled after incubation of fi broblasts with MDW941, was detected by fl uorescence imaging of the slab gel ( Fig. 1A ). Fibroblasts of LIMP2-defi cient AMRF patients showed almost no active GBA. The active GBA was similarly reduced in cells from the type 2 GD patient. In the case of cells from the AMRF carrier, a normal amount of active GBA was detected.
Next, intact fi broblasts were incubated with MDW933 to visualize the active GBA molecules by fl uorescence microscopy ( Fig. 1B ). Multi-spectral imaging was used to differentiate autofl uorescence from true ABP signals. In line with the results in Fig. 1A , AMRF fi broblasts were found to be clearly defi cient in lysosomal GBA compared with the WT.
The fi ndings with leukocytes of an AMRF patient were very different from those of fi broblasts. Labeled active glucosylceramide breakdown as seen in GD patients. We earlier documented the increase of glucosylsphingosine in plasma of symptomatic type 1 GD patients suffering from a primary defect in lysosomal GBA ( 6 ). Because LIMP2 defi ciency also results in a marked reduction of lysosomal GBA, we assumed that the same would occur in AMRF patients.
We therefore determined glucosylceramide and glucosylsphingosine concentrations in the fi broblasts and plasma of AMRF patients as well as in the plasma of LIMP2defi cient mice.
In AMRF fi broblasts, glucosylceramide was not elevated compared with the control range ( Fig. 4A ). Importantly, a marked increase in glucosylsphingosine was noted ( Fig. 4C ). The same, although not to the same extent of elevation, was observed for leukocytes ( Fig. 4D ). We also detected elevated glucosylsphingosine levels in the plasma of three AMRF patients ( Fig. 4E ). No clear concomitant increase in glucosylceramide was observed as compared with the matched control ( Fig. 4B ). To further validate the fi nding of the sphingoid base abnormality, we also examined plasma of LIMP2-defi cient mice. Again, a clear elevation in glucosylsphingosine concentration was observed ( Fig. 4F ).
fi broblasts but not AMRF fi broblasts, stem from lysosomal extrusion/secretion. We next looked for the presence of GBA in the plasma of AMRF patients by measuring its enzymatic activity with the artifi cial substrate 4-methylumbelliferyl-␤ -D -glucoside ( 30 ). We noted an increased activity (10.3 vs. 1.4 nmol of released 4-methylumbelliferone per milliliter of plasma per hour) in the case of fresh specimens from an AMRF patient versus a healthy subject. Importantly, the increased activity was only detected in freshly obtained plasma. Storage of plasma, even frozen, led to inactivation of the enzymatic activity of GBA; in samples frozen for a long period at Ϫ 20°C, less than 2% of the original GBA activity was detected. In line with the data for freshly obtained AMRF patient plasma, LIMP2-defi cient mice showed increased GBA as well (167 nmol/ml of plasma per hour vs. 22 nmol/ml of plasma per hour for WT mice). Given the loss of enzyme activity in plasma, measurement of plasma GBA activity is in practice not a very reliable test to identify AMRF patients.

Lipid abnormalities in relation to LIMP2 defi ciency
The lack of lysosomal GBA as the result of LIMP2 defi ciency might theoretically result in abnormalities in GBA in their lysosomes. We demonstrate here with a variety of techniques that, indeed, white blood cells of AMRF patients have considerable residual GBA, in sharp contrast to fi broblasts. We employed newly designed ABPs that allow fl uorescent labeling of active GBA molecules. We fi rst showed that much less active GBA can be detected in

DISCUSSION
The study by Reczek et al. ( 16 ) fi rst revealed that the lysosomal integral membrane protein, LIMP2, mediates traffi cking of newly formed GBA to lysosomes. Soon afterwards Berkovic et al. ( 17 ) demonstrated that LIMP2 deficiency due to mutations in the SCARB2 gene forms the basis for AMRF. Independently, Balreira et al. ( 18 ) described that mutations in the SCARB2 gene cause celltype-specifi c GBA defi ciency. Intriguingly, AMRF patients do not show the lipid-laden macrophages that characteristically occur in GD patients suffering from a primary defect in GBA. In GBA-defi cient GD patients, the macrophages are particularly prone to store glucosylceramide and to transform to so-called Gaucher cells ( 32 ). Gaucher cells produce and secrete unique marker proteins such as chitotriosidase and CCL18 ( 32 ). Characteristically, chitotriosidase is several hundred-fold higher in the plasma of symptomatic GD patients ( 3,33 ). Of interest, AMRF patients do not show elevated levels of chitotriosidase like GD patients ( 18, 24 ) (P. Gaspar and C. Sá Miranda , unpublished observations). This suggests that somehow macrophages in LIMP2-defi cient individuals can still adequately degrade glucosylceramide due to the presence of suffi cient  as substrate for in vivo detection of GBA enzymatic activity. Again, FACS analysis revealed a marked defi ciency in AMRF fi broblasts, but considerable residual enzyme in white blood cell types. We next looked into the fate of GBA in fi broblasts of AMRF patients, observing that these cells abnormally secrete some active GBA to the medium where the enzyme rapidly loses its enzymatic activity. Labile GBA activity is demonstrable in freshly obtained plasma of homogenates of AMRF fi broblasts with an ABP . In white blood cells and cultured macrophages, this abnormality is far less striking. Next we labeled intact cells with an ABP and noted by FACS analysis that AMRF fi broblasts are markedly defi cient in active GBA. Signifi cant residual active GBA was again detected in lymphocytes, monocytes, and cultured macrophages. The same phenomenon was demonstrated in an independent manner by exploiting PFB-FDG Fig. 4. Glucosylceramide and glucosylsphingosine content of cells and plasma in relation to AMRF/LIMP2 defi ciency. Glucosylceramide and glucosylsphingosine were determined as described in Materials and Methods. A: Glucosylceramide (nmol/mg total protein) in AMRF fi broblasts (n = 2). B: Glucosylceramide (nmol/ ml total protein) in human AMRF plasma specimens (n = 2). C: Glucosylsphingosine (pmol/mg total protein) in AMRF fi broblasts (n = 3). D: Glucosylsphingosine (pmol/mg total protein) in AMRF leukocytes (n = 1). E: Glucosylsphingosine (pmol/ml) in human AMRF plasma specimens (n = 3). F: Glucosylsphingosine (pmol/ml) in samples from LIMP2 Ϫ / Ϫ mice (n = 3). AMRF, LIMP2-defi cient patients (LIMP2 W178X/ W178X); GD, GD patients; GD Car, GD carriers; LIMP2 +/+ , WT mice; LIMP2 Ϫ / Ϫ , KO LIMP2 mice. ** P < 0.01, unpaired student's t -test ). AMRF patients. The mechanism by which lysosomes of white blood cells acquire suffi cient GBA is presently not known and a topic of further investigation. It can't be excluded that these cells possess a membrane protein other than LIMP2 that governs intracellular transport of GBA from the endoplasmic reticulum to lysosomes. Alternatively, secreted GBA might be taken up by some endocytotic process and delivered via this secretion-recapture manner to lysosomes of white blood cells.
In lysosomes, assisted by the activator protein saposin C, GBA takes care of degradation of glucosylceramide ( 4 ). Defi ciency of GBA causes formation of glucosylceramide storage tubules, most prominent in macrophages. Of interest, part of the accumulating glucosylceramide during GBA defi ciency is deacylated to the sphingoid base, glucosylsphingosine, which can leave lysosomes and cells ( 6 ). Most likely, acid ceramidase is responsible for glucosylsphingosine formation, because acid ceramidase-defi cient Farber disease fibroblasts have been found to be unable to synthesize glucosylsphingosine upon inhibition of GBA activity ( 34 ).
This explains the marked increase in glucosylsphingosine in the plasma of GD patients. Our investigation of glucosylceramide and glucosylsphingosine concentrations in cultured fi broblasts of AMRF patients revealed that only the sphingoid base, glucosylsphingosine, is markedly increased. Consistent with this, glucosylsphingosine is abnormally high in the plasma of AMRF patients. This fi nding was also made with plasma specimens of LIMP2-defi cient mice. Apparently, in AMRF patients, cells other than macrophages also form glucosylsphingosine during GBA deficiency that is partly released into the circulation.
Our investigation renders a workfl ow for convenient laboratory diagnosis of AMRF prior to sequencing of the SCARB2 gene. If markedly increased plasma glucosylsphingosine is detected in an individual in the absence of elevated chitotriosidase, this is an indication for AMRF . Measurement of GBA activity in white blood cells will not be very informative, but in fi broblasts enzymatic activity needs to be reduced to consider further the diagnosis AMRF. Such a diagnosis can be further substantiated with the sequencing of the SCARB2 gene. Demonstration of functional GBA defi ciency by detection of abnormally high glucosylsphingosine in plasma should be considered as an important step in identifi cation of individuals suffering from a truly functional defi ciency in LIMP2.