(Received for publication, December 19, 1996, and in revised form, May 27, 1997)
From the Glycobiology Institute, Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, United Kingdom
The imino sugar N-butyldeoxynojirimycin is an inhibitor of the ceramide-specific glucosyltransferase that catalyzes the first step in glycosphingolipid biosynthesis. It results in extensive glycosphingolipid depletion in cells treated in vitro, without causing toxicity. However, we currently do not know the degree to which glycosphingolipids can be depleted in vivo in a mammalian species. We have therefore administered N-butyldeoxynojirimycin long term to young mice and have found that glycosphingolipid levels are reduced (50-70%) in all tissues examined, without resulting in any overt pathology. When the lymphoid tissues from these mice were examined, they were found to be 50% acellular relative to non-lymphoid tissues. These data implicate a role for glycosphingolipids in the biology of the immune system or indicate an additional as yet unknown activity of N-butyldeoxynojirimycin. Extensive glycosphingolipid depletion resulting from N-butyldeoxynojirimycin administration is therefore well tolerated in adult mice, and this compound may be in an invaluable tool for probing glycosphingolipid functions in vivo. In addition, this drug may be effective in clinical situations where glycosphingolipid depletion would be desirable, such as the in the treatment of the human glycosphingolipidoses.
Glycosphingolipids (GSLs)1 are ubiquitous components of the plasma membranes of eukaryotic cells where they are believed to mediate a number of biological functions (1). Their biosynthesis and catabolism have been extensively studied (2-4), as have the disease states that result from their incomplete catabolism within lysosomes (5). They are exploited as receptors by a number of micro-organisms (6); however, their normal physiological functions remain largely obscure.
GSL biosynthesis involves the co-ordinated action of multiple gene products (2). A key enzyme in this pathway is the ceramide-specific glucosyltransferase, which catalyzes the first step in the GSL biosynthetic pathway, the transfer of glucose to ceramide to form glucosylceramide (GlcCer). The sequential action of glycosyltransferases in the Golgi apparatus converts glucosylceramide into other neutral GSLs and gangliosides (2). The ceramide-specific glucosyltransferase has been cloned recently and is a unique gene product without substantial homology with other known glycosyltransferase genes (7). Mutant B16 melanoma cells lacking this enzyme are not compromised in their growth or morphology, indicating that GSLs are not needed for membrane integrity and do not serve a basic housekeeping function, at least at the level of the single cell (7, 8).
One approach to investigate the role of GSLs in intact organisms is to inactivate the glucosyltransferase either by gene disruption techniques or by using a specific inhibitor. It is not currently known the degree to which GSLs can be depleted in vivo in mammalian species without resulting in pathology (9). The gene knockout approach would determine whether or not total inhibition of GSL synthesis is compatible with embryonic development. There is circumstantial evidence, such as the lack of human disease states that result from defects in GSL biosynthesis, which suggests that GSL expression may play a critical role in early mammalian development (4). The complete lack of GSL biosynthesis during embryogenesis could therefore be potentially lethal. However, in studies of teleost development using Medaka fish in in vitro culture, extensive GSL inhibition resulting from treatment with the ceramide analogue PDMP did not disrupt development (10). The roles of GSLs during vertebrate development and their functions in the biology of the mature organism therefore remain to be elucidated.
The alternative experimental approach which can be taken to explore GSL functions is to use specific inhibitors of enzymes required for the biosynthesis of GSLs. One advantage of this approach is that the consequences of partial GSL depletion can be investigated. An attractive target for enzyme inhibition is the ceramide-specific glucosyltransferase, which catalyzes the first step in the GSL biosynthetic pathway.
We have recently identified two inhibitors of GSL biosynthesis that block the action of the ceramide-specific glucosyltransferase (11, 12). These inhibitors are the imino sugars Nbutyldeoxynojirimycin (NB-DNJ) and N-butyldeoxygalactonojirimycin (NB-DGJ), a glucose and galactose analogue, respectively. In vitro these analogues result in GSL depletion in a wide range of human and murine cell lines (11, 12). However, the extent to which they can reduce GSL levels in vivo is currently unknown, as is the extent to which GSL depletion can be tolerated in an intact adult mammal. The degree to which GSLs depletion is tolerated in vivo is also an important factor when considering GSL biosynthetic inhibitors as potential therapeutic agents for the treatment of the GSL lysosomal storage diseases, where in vivo GSL depletion is the primary clinical objective (9, 11, 12).
In this study we have administered NB-DNJ to young mice using a range of dosing regimes and monitored GSL biosynthesis. NB-DNJ was used, rather than the galactose analogue NB-DGJ, due to the fact that this compound is available in the multigram quantities required for this study. We have found significant GSL depletion in multiple tissues of mice treated with NB-DNJ without any overt toxicity, indicating that partial inhibition of GSL biosynthesis is well tolerated in vivo. The implications these data have for GSL function and the therapeutic administration of GSL inhibitors for treating GSL lysosomal storage disorders are discussed.
NB-DNJ was a gift from Searle/Monsanto.
Treatment of Mice with NB-DNJAge- and sex-matched (6-week-old female) C57BL/6 mice were fed on a diet of powdered mouse chow (expanded Rat and Mouse Chow 1, ground, SDS Ltd., Witham, Essex, UK) containing NB-DNJ. The diet and compound (both as dry solids) were mixed thoroughly before use, stored at room temperature, and used within 7 days of mixing. Water was available to the mice ad libitum. The mice were housed under standard nonsterile conditions. The mice were maintained on 600 mg/kg/day per mouse for 50 days, and two animals were sacrificed from the control group (untreated, n = 10) and two from the experimental group (diet + NB-DNJ, n = 10). The remaining animals in the experimental group were then maintained on a 1200 mg/kg/day diet for a further 50 days when another two animals from each group were sacrificed. The remaining animals in the treated group were fed on a 1800 mg/kg/day diet for 20 days at which point all remaining animals from each group were sacrificed. In an independent study, animals were maintained on a 2400 mg/kg/day diet for 14 days.
Determination of NB-DNJ Concentrations in Mouse SerumBlood
samples were collected by cardiac puncture, the serum expressed, and
stored at 20 °C prior to assay. Steady state serum levels of
NB-DNJ were determined by isolating compound from 50 µl of
blood using AG-50W-X12 (H+ form) resin (200 µl of packed
beads). Compound and an internal standard (N-hexyl-DNJ or
N-propyl-DNJ) were eluted with a 4 M ammonia
solution. Using radiolabeled compounds, greater than 99% of the
compound was recovered from the resin. Detection of compound was
achieved either by matrix-assisted laser desorption ionization time of
flight mass spectrometry (MALDITOF) or by cation-exchange chromatography. For mass spectrometry (Finnigan MAT), an internal standard was included with normal serum samples to construct a standard
curve with NB-DNJ concentrations ranging from 5 to 40 µg/ml of serum. Cation-exchange chromatography (Dionex BioLC) was
performed by an isocratic elution of a CS-10 column with sodium sulfate
containing 5% (v/v) acetonitrile. Quantitation of NB-DNJ was obtained by peak area estimation in comparison with the internal standard and applying experimentally derived response factors.
Tissue samples were homogenized
in water, lyophilized, 5 mg of dry weight extracted in
chloroform:methanol (2:1, v/v) overnight at 4 °C, dried under
nitrogen, resuspended in 20 µl of chloroform:methanol (2:1, v/v), and
separated by TLC (chloroform:methanol:H20, 65:25:4). The
TLC plate was air-dried, sprayed with -naphthol (1% (w/v), in
methanol) followed by 50% (v/v) sulfuric acid and heat-treated (80 °C for 10 min).
The assay has been described previously (11). Briefly, splenocytes from treated and untreated mice were incubated with fluorescein isothiocyanate-conjugated cholera toxin B chain, for 30 min on ice in the dark. Quantitative flow cytometric analysis has been described elsewhere (11) and was performed using a FACScan Cytometer (Becton Dickinson, Sunnyvale CA). Data on viable cells (propidium iodide dye exclusion) were collected on a 4-decade log10 scale of increasing fluorescence intensity.
Cell Surface Labeling of Splenocytes with IO4/B3H4Splenocytes (3 × 107 mononuclear cells) from two NB-DNJ-treated (2400 mg/kg/day) and two untreated mice were washed twice in PBS, resuspended in 1 ml of PBS containing 1 mM sodium metaperiodate, and left on ice for 5 min. The oxidation was stopped by the addition of 0.2 ml of PBS containing 0.1 M glycerol. The cells were washed twice with PBS, resuspended in 0.5 ml of PBS, and tritiated sodium borohydride added (0.8 mCi; specific activity, 12 Ci/mmol) and incubated at room temperature for 30 min. The cells were washed three times with PBS and radiolabeled GSLs extracted from the cell pellet (2 × 1 ml of chloroform:methanol 2:1 (v/v) overnight at 4 °C and 2 h at room temperature). An equal volume of each extract (equivalent to 0.75 × 107 cells) was taken for analysis by high performance thin layer chromatography using a solvent of chloroform:methanol:0.5% CaCl2 (55:49:9, v/v/v). Radiolabeled GSLs were detected by fluorography and their migration compared with authentic GSL standards. The cell pellets remaining after the lipid extraction were resuspended in 200 µl of distilled water to which SDS was added (1% (w/v) final concentration) and dithiothreitol (0.5 mM, final concentration). The samples were heated to 95 °C for 5 min, spun at 13,000 × g for 5 min, and aliquots taken for radioactivity determination by scintillation counting. Protein content was determined using the BCA method (Pierce, Chester, UK) and 20-µg equivalents of protein analyzed by SDS-PAGE and fluorography before and after treatment with peptide:N-glycosidase F (New England Biolabs (Hitchin, UK) Ltd., 50,000 units/ml final concentration for 4 h at 37 °C).
Immunophenotyping of SplenocytesThe following rat mAbs
were used for this study and have been described previously (13):
187.1, anti-mouse light chain; GK1.5, anti-CD4; anti-Ly-2, CD8
(Becton-Dickinson, Mountain View, CA); and 7.3, anti-Lyt-1. A goat
anti-rat fluorescein isothiocyanate conjugate (Ortho) was used to
detect anti-CD4, anti-
and anti-Lyt-1 binding. The staining
procedure and flow cytometry analysis have been described elsewhere
(13). Data on viable cells (propidium iodide dye exclusion) were
collected on a 4-decade log10 scale of increasing
fluorescence intensity. The percentage of cells staining with the mAbs
was defined by the positioning of cursors on the basis of the unstained
control (autofluorescent background) for directly conjugated mAbs or on
the basis of the secondary only control with the nonconjugated
mAbs.
To monitor the
overall well being of the mice treated with NB-DNJ, the mice
were monitored daily, body weights recorded, and the effects of
NB-DNJ on growth rates determined (Fig.
1). The doses of NB-DNJ given
over the course of the study are indicated with arrows. Over
the treatment period, and with the dosing regime given, the mice on the
NB-DNJ containing diet grew slower than the untreated sex-
and age-matched controls. The final body weights attained at the end of
the study (118 days, 2400 mg/kg/day) were 15% lower in the
NB-DNJ-treated group. When the mice were switched to a diet
rich in glucose (65%, w/w), but lacking complex carbohydrates (AIN 76, SDS), the body weights remained 15% lower than the untreated controls.
The mice had no diarrhea at any stage in the study, in contrast to mice
given comparable or lower concentrations of NB-DNJ by oral
gavage every 8 h.2 The
general appearance, activity levels, and condition of the NB-DNJ-treated mice were comparable with the untreated
controls (data not shown), indicating that the animals tolerated
NB-DNJ in their diet and were not overtly compromised by its
administration.
Serum Levels
Serum levels of NB-DNJ were determined at all the dosing levels and were found to range in a dose-dependent fashion from 18 µM at 600 mg/kg/day to 57 µM at 2400 mg/kg/day (Table I).
|
As a measure of cellular GSL levels,
cell surface ganglioside levels were determined on splenocytes from
treated and untreated mice, using fluoresceinated cholera toxin B chain
as a probe for GM1 ganglioside (11). The cells were assayed
by flow cytometry and binding sites per cell quantitated using
fluorescent microbead standards. The percent reduction in cholera toxin
binding sites per cell was measured at each drug dose (Fig.
2). At 600 mg/kg/day the splenocytes had
10% fewer cholera toxin binding sites per cell than the untreated aged
matched controls. This increased to an approximate 35% reduction at
doses in the range of 1200-1800 mg/kg/day and to 50% at the highest
dose tested, 2400 mg/kg/day. These data indicate that GSL depletion
occurs in vivo in response to NB-DNJ and is
dose-dependent.
Periodate Labeling
Cell surface glycoconjugates were examined
in more detail by incorporating a radiolabel into cell surface sialic
acid residues. Mild periodate treatment of glycoconjugates results in
the selective oxidation of sialic acid residues. The aldehydes
formed are reduced with NaB3H4 to
stable, tritium-labeled
5-acetamido-3,5-dideoxy-L-arabino-2-heptulosonic acid.
Using the conditions described, this method specifically introduces a
label into cell surface sialic acids, including those carried by
gangliosides (14). The autoradiograph of the high performance TLC
separation of solvent extracted GSLs from splenocytes from two
untreated mice and two treated with NB-DNJ (Fig.
3A) showed a quantitative
decrease (60-70% reduction by densitometry) in labeled species
compared with those from untreated animals. There did not appear to be
a qualitative difference in the gangliosides extracted from labeled
cells from the two groups of mice, nor were there any changes in the
relative abundance of individual gangliosides in individual animals.
These data strongly suggest that the expression of cell surface
gangliosides is decreased in a nonselective manner as a result of
treatment with NB-DNJ. When the protein fraction from these
samples was analyzed by SDS-PAGE and visualized by fluorography, no
discernible differences could be detected in the profile, either before
or after peptide:N-glycosidase F treatment, of sialic acid
containing glycoproteins derived from untreated and
NB-DNJ-treated animals (Fig. 3B). Using the same enzyme incubation conditions, mouse IgM was completely
de-N-glycosylated (results not shown). This indicated that
-glucosidase inhibition was not occurring at the serum levels
achieved; hence, there was no apparent reduction in glycoproteins
carrying sialylated N-glycans.
Chemical Detection of Liver GSLs
To determine the degree of
GSL depletion in a non-lymphoid tissue, liver was
chloroform:methanol-extracted and total lipids analyzed by thin layer
chromatography. A reduction in GSL levels in the mice treated with
NB-DNJ was observed, including GM2 and the
trihexoside species (Fig. 4).
Sphingomyelin levels increased in NB-DNJ-treated mice,
presumably due to the increased levels of ceramide available to go down
this biosynthetic pathway, due to inhibition of GSL biosynthesis.
Removal of the drug from the diet for 2 weeks resulted in relatively
normal liver GSL levels with sphingomyelin being present in quantities
comparable with the untreated controls and the trihexoside species
being detectable. GM2 levels were slower to recover, and
this ganglioside was barely detectable after the 2-week period,
following removal of NB-DNJ from the diet. GSL depletion was
therefore occurring in livers of NB-DNJ-treated mice and was
reversible following drug withdrawal. The fact that GM2
recovery is slow may indicate possible long term effect of the
drug.
Immunophenotyping of Lymphoid Tissues
Spleens and thymuses removed from NB-DNJ mice were observed to be consistently smaller (on a wet weight basis) than those from untreated age matched controls. For example, spleens from 600-1800 mg/kg/day were 30% lighter than the untreated controls (±6%), and at 2400 mg/kg/day they were on average 55% lighter (±6%) than spleens taken from age-matched control mice. This was in contrast to organs such as liver and brain, which were comparable by wet weight, irrespective of NB-DNJ treatment (data not shown).
To investigate the nature of the reduction in lymphoid tissue size in
response to NB-DNJ treatment, the 2400 mg/kg/day animals were investigated further. A cohort of animals were fed on a
NB-DNJ-containing diet and changes in spleen and thymus size
monitored over time (Fig. 5, A
and B). Both tissues reduced in size in response to NB-DNJ treatment, and maximal reduction in organ size
occurred between 7 and 11 days following drug administration. The
lymphoid tissues remained smaller as long as the animals were treated
with NB-DNJ, but returned to normal size after 2 weeks
following withdrawal of the drug from the diet (for example, spleen;
Fig. 6).
The cellular compositions of spleen and thymus after 2 weeks on 2400 mg/kg/day NB-DNJ were examined by flow cytometry (Table II). The spleen contained a higher percentage of T cells relative to the frequencies observed in untreated animals (44% compared with 30%), and the B cell frequencies were reduced from 51% in untreated controls to 45% in NB-DNJ-treated mice. When drug was removed from the diet, within 2 weeks the relative proportions of T and B cells within the spleen returned to normal levels (Table II). The increased proportion of T cells within the spleen was due to an increased frequency of both CD4+ and CD8+ cells, with the greatest increase being in the CD8+ cells (9% positive in the untreated spleen to 16% positive in the NB-DNJ-treated spleen), while CD4+ cells accounted for 20% of cells in the untreated spleen and 31% in NB-DNJ-treated spleen. Histological examination of spleens from untreated and NB-DNJ-treated mice showed no gross morphological differences with normal follicular architecture maintained in spleens from NB-DNJ-treated mice (data not shown).
|
The thymuses from animals treated with NB-DNJ were also
examined. Two-color flow cytometry after 7 days of NB-DNJ
treatment revealed that, in contrast to the untreated thymus, which has a typical composition of approximately 5%
CD4/CD8
, 80%
CD4+/CD8+, 10%
CD4+/CD8
, and 5%
CD4
/CD8+ cells, the NB-DNJ-treated
mice had an increased frequency of single positive cells. For example
the thymuses from NB-DNJ-treated mice contained
approximately 7% CD4
/CD8
, 65%
CD4+/CD8+, 18%
CD4+/CD8
, and 10%
CD4
/CD8+ cells (Fig.
7). There was therefore an increase in
the proportion of single CD4+ or CD8+ cells at
the expense of CD4+/CD8+ double positive
cells.
As GSLs are not essential components of mammalian cells at the single cell level (8), their functions must reside in more complex aspects of mammalian biology. If these functions are to be elucidated, methods and reagents are required that will permit GSL manipulation in vivo as well as in vitro. Although we can inhibit GSL biosynthesis in vitro using a number of pharmacological agents, compounds are required that are well tolerated long term in vivo.
There are currently two main classes of GSL biosynthesis inhibitor that have been characterized to date which specifically target the ceramide-specific glucosyltransferase. The first class of inhibitors described were the ceramide analogues of which the prototypic compound is PDMP (15). By virtue of ceramide mimicry these compounds prevent the glucosylation of ceramide by acting as reversible mixed-type inhibitors (16). However, the highly hydrophobic nature of these compounds makes their long term administration to animals problematic (9). The second group of GSL biosynthesis inhibitors are the N-alkylated imino sugars, such as the glucose analogue NB-DNJ. The nonalkylated parental compound (DNJ) does not inhibit the ceramide-specific glucosyltransferase, and a minimal alkyl chain length is required to render these compounds inhibitory, with butyl and hexyl chains being optimal (11). The alkyl chain may render these compounds short chain ceramide mimics or serve to target these compounds to the membrane environment where the glucosyltransferase resides (9).
The alkylated imino sugars have a major advantage over PDMP and related compounds in that they exhibit minimal cytotoxicity in vitro, even at doses in excess of 2 mM (17). They therefore have the potential to be well tolerated in vivo. In this study we have investigated whether or not we can achieve GSL depletion in vivo in healthy mice treated orally with NB-DNJ and whether GSL depletion is in any way deleterious to the animal.
When mice were treated with various doses of NB-DNJ, serum levels in the range of 18-57 µM were achieved. Similar serum levels (steady state trough level of approximately 20 µM) were achieved in humans during the evaluation of this compound as an antiviral agent when patients were treated with 43 mg/kg/day (18). The pharmacokinetics of NB-DNJ are 2 orders of magnitude poorer in mouse relative to human necessitating high dosing regimes in the mouse to achieve serum levels in the range required to achieve GSL biosynthesis inhibition (5-50 µM) (9, 11, 12).
The NB-DNJ-treated animals remained healthy, active, and free from any overt signs of pathology for the duration of the experiment, which lasted for 118 days. The only tangible change observed was that the growth rates of NB-DNJ-treated mice were reduced relative to the untreated age-matched controls. One activity of imino sugars such as NB-DNJ is that they are competitive inhibitors for the gastrointestinal disaccharidases sucrase and isomaltase. Inhibition of these enzymes results in osmotic diarrhea and was the major side effect when this compound was tested as a potential anti-human immunodeficiency virus therapeutic in AIDS patients (18). It was therefore surprising that in this study, working at dosing regimes in excess of those known to cause diarrhea in mice given NB-DNJ by oral gavage, we observed no diarrhea in the treated mice. The major difference in the drug administration protocol used in this study was that NB-DNJ was co-administered with a complex carbohydrate containing diet. There would therefore be competition within the gastrointestinal tract between the natural substrate for the disaccharidases and the inhibitor. Also, because small amounts of drug were ingested each time the mice fed, this may also have helped prevent the side effect associated with bolus administration of these compounds, as the concentration of NB-DNJ in the gastrointestinal tract would be considerably lower. Administration of NB-DNJ to patients in small quantities at regular intervals with food rich in complex carbohydrate may be a simple means of minimizing gastrointestinal tract distress and makes the development of an effective prodrug unnecessary (19).
One potential explanation of why the mice have lower body weights relative to the untreated controls is that due to disaccharidase inhibition, glucose absorption is reduced. We investigated whether or not this was a contributory factor by switching the mice to a diet lacking complex carbohydrate and rich in glucose. On this diet the NB-DNJ-treated mice exhibited an identical reduction in body weight as those on the complex carbohydrate diet, suggesting that potential limitation of glucose uptake on the standard diet containing NB-DNJ is not the reason why these animals have lower body weights. NB-DNJ could be acting as an appetite suppressant, causing reduced feeding activity, or it may reflect a change in metabolism within the treated animals.
A profound reduction in cell surface gangliosides was observed in the livers and spleen of NB-DNJ-treated mice, indicating that extensive GSL depletion was occurring. This depletion occurred in all gangliosides, and no selective depletion was observed, which is consistent with the proposed mechanism of action of NB-DNJ as an inhibitor of the first step in GSL biosynthesis. The degree of GSL depletion was much more profound as determined by GSL TLC analysis and cell surface sialic acid labeling, relative to the cholera toxin binding flow cytometric assay. We have consistently observed that T lymphocyte cell lines in tissue culture can be depleted of cholera toxin binding sites if the cells are treated with NB-DNJ (which inhibits GSL biosynthesis and N-glycan processing), but the GSL-specific inhibitor NB-DGJ (12) only results in a partial reduction. If, however, NB-DGJ is co-administered to cells along with DNJ (which only inhibits N-glycan processing), then complete depletion of cholera toxin binding sites is achieved.3 This implicates N-glycan structures on T cells, but not cells of the myeloid lineage, as having the ability to bind cholera toxin. This serves to illustrate the degree to which differential inhibitor experiments using DNJ, NB-DNJ, and NB-DGJ can shed light on the relative contribution to toxin binding of N-glycans versus GSLs. Such studies are applicable to any biological scenario, where it is unclear which class of glycoconjugate is mediating the biological function of interest. When thymocyte proteins were separated by SDS-PAGE, Western-blotted, and probed with cholera toxin, only a small subset of proteins bound the toxin, suggesting that a small group of proteins carry this N-glycan epitope.3 The identity of these proteins and their N-glycans are currently under investigation. The relative affinity of the interaction between cholera toxin B chain for the authentic GM1 ganglioside oligosaccharide and the unknown N-glycan epitope are currently not known. The cell surface sialic acid labeling method and the liver GSL TLC analysis therefore give a more accurate reflection of GSL depletion (50-70%) on cell populations that include T lymphocytes.
It was found when the cell surface glycoproteins were studied from
spleen cells that had been radiolabeled specifically in sialic acid
residues that there was no evidence of a reduction in complex sialic
acid-containing glycans. This suggests that -glucosidase inhibition
is not occurring in splenocytes at the serum levels of
NB-DNJ achieved in this study. This is paradoxical, considering the Ki values for NB-DNJ
against the ceramide-specific glucosyltransferase and
-glucosidase I
are 7.4 and 0.22 µM, respectively. However, the topology
of the two enzyme targets of NB-DNJ are very different, with
the glucosyltransferase being located with its catalytic domain on the
cytosolic face of an early Golgi compartment (7, 20-23), whereas
-glucosidase I is a lumenal endoplasmic reticulum-resident enzyme
(24). NB-DNJ therefore not only has only to cross the plasma
membrane to inhibit the glucosyltransferase, but has to also cross the
endoplasmic reticulum membrane to access and inhibit
-glucosidase I. This difference in topology of the two enzymes may result in serum
levels of NB-DNJ in the micromolar range being inhibitory
against the glucosyltransferase but not
-glucosidase. Presumably,
lumenal endoplasmic reticulum concentrations of NB-DNJ
required to inhibit
-glucosidases were not achieved using this
dosing regime. This is in agreement with in vitro studies in
which 50 µM NB-DNJ exhibits minimal inhibitory
activity against
-glucosidase I (25). Therefore, despite the dual
inhibitory activity of NB-DNJ against the GSL biosynthetic
pathway and N-glycan processing pathway, in practice these
two activities are readily dissociated by working at serum levels in
the range where GSL biosynthesis inhibition occurs, but not inhibition
of
-glucosidases I and II.
The brain is a very rich source of GSLs, but the half-lives of both cells and GSLs within brain are long, and minimal impact is observed when adult mice are studied over the time course of this study (data not shown). However, when a mouse model of Tay-Sachs disease was studied, accumulation of GM2 was prevented in the brain of NB-DNJ-treated animals, demonstrating that the compound crosses the blood-brain barrier and inhibits GSL biosynthesis (26).
The lymphoid tissues from mice treated for 7 days or longer with NB-DNJ were consistently small compared with their untreated age-matched counterparts. This was not reflected in non-lymphoid tissues such as brain and liver, where organ sizes were the same irrespective of treatment (data not shown). On average the glands were approximately 50% smaller in the 2400 mg/kg/day group after 7 days of treatment. The basis for this reduction in size was unknown and unanticipated, as the mice were showing no evidence of being immune compromised, as they were disease-free and healthy despite being housed in a nonsterile environment. Both spleen and thymus exhibited normal lymphoid tissue architecture at the level of hematoxylin/eosin staining, when examined histologically (data not shown). In spleen, in addition to having only approximately 50% of the cellular content of untreated control spleen, there was an increase in the percentage of cells staining with T cell-specific markers, and there was a small reduction in the percentage of B lymphocytes. The thymus was also 50% acellular and had an even more dramatic change in phenotype with the percentage of single positive cells (CD4+ or CD8+) doubled relative to normal thymus, with the percentage of double positive cells (CD4+/CD8+) reduced. We currently do not understand the underlying mechanism responsible for these changes. If these changes are the result of perturbation of glycoconjugate biosynthesis, they most probably result from reduced GSL biosynthesis, as the N-glycan-processing pathway is not inhibited at the serum levels achieved (Fig. 4). However, we cannot rule out a glycoconjugate-independent mechanism due to a direct or indirect response to NB-DNJ treatment. The kinetics indicate that the changes in lymphoid tissue cellularity begin within 24 h of treatment and are maximal by days 7-11. The effect persists as long as the drug is maintained in the diet. However, within 2 weeks of withdrawal of NB-DNJ from the diet, the organs regained their normal cellular complement and were indistinguishable in size and composition from untreated controls.
The natural history of lymphoid tissues is highly complex, with cell numbers being maintained within normal limits by homeostatic mechanisms, which are poorly understood. Within the spleen, perturbation in hematopoiesis within the bone marrow could reduce the available pool of circulating cells, which would reduce the size of the lymphoid organs. Alternatively, disruption of normal cellular trafficking could also be manifested as reduced cell numbers within lymphoid tissues. Cell proliferation within the glands could also reduce cell number. However, as much of the proliferation in the spleen occurs in germinal centers as part of an antigen-specific acquired immune response, it would be envisaged that this could have an effect on immune competence. Although the NB-DNJ-treated mice have not as yet been experimentally challenged with T cell-dependent and T cell-independent antigens, the fact that they are not succumbing to environmental infections is suggestive that they are not immune-deficient.
The effect in the thymus could reflect reduced recruitment to the gland of bone marrow-derived precursors, changes in T cell selection, increased frequency of apoptosis, or slower egress of single positive cells from the organ. Further detailed studies will be required to establish the basis for the changes observed in response to NB-DNJ treatment and also to determine whether or not these changes result from GSL depletion or an unrelated activity of this compound.
The reduction in spleen volume as a result of NB-DNJ has practical implications in light of the potential application of this drug for the treatment of Gaucher disease (9, 11, 12). Splenic enlargement is a hallmark of this disease and is used as a disease parameter that is monitored when evaluating clinical improvement (27). If NB-DNJ acts in a similar way in humans the spleen may be reduced in size, irrespective of an impact on lysosomal storage. Other clinical parameters must therefore be used as markers of therapeutic outcome, as spleen size may be an unreliable marker due to this activity of NB-DNJ on lymphoid tissues.
The demonstration of extensive GSL depletion in the lymphoid organs and liver of a normal mammal treated with NB-DNJ now paves the way for using NB-DNJ as an in vivo tool for manipulating GSL levels in vivo to probe GSL functions. Furthermore, the fact that mice with up to 70% GSL depletion are not compromised by the depletion raises interesting questions about how much GSL depletion can be achieved without causing pathology. It may well be the case that a critical threshold level exists above which the GSL-mediated biology can proceed unimpeded, but below which dysfunction occurs. There are several potential ways cells could compensate for GSL depletion, for instance relying on sphingomyelin to mediate adhesion (28) or by compensation through other glycoconjugates. There have been two recent reports implicating compensatory mechanisms in which GSL species can potentially function for a GSL species that is absent from genetically engineered knockout mice. For instance, in mice deficient in the galactosyltransferase that synthesizes GalCer, the mouse has high levels of GlcCer in its myelin as this biosynthetic pathway remains intact. It is postulated that both monohexoside species may be capable of supporting myelination in the developing mice, although the stability of the myelin rich in GlcCer (a lipid not normally found in myelin) rather than the usual GalCer has impaired stability (29). The second example comes from similar studies in which the consequences of making a mouse null for GM2/GD2 synthase has been investigated, which renders these mice devoid of complex gangliosides (30). There is no major phenotype in these mice, and it is postulated that the simpler gangliosides GM3 and GD3 could substitute functionally for the more complex gangliosides that are absent in these mice. Both studies emphasize the potential biological redundancy intrinsic to this group of highly related GSL species. Also, in view of the relatively high degree of lateral mobility exhibited by GSLs, even an extensively depleted cell surface could remodel its GSLs into small GSL-enriched microdomains, which could have sufficient localized GSL density to mediate their functions. This may be particularly relevant to this study, as all GSL species have been partially depleted, but 30-50% of the GSLs are still synthesized due to incomplete inhibition of the glucosyltransferase. Reorganization of these residual GSLs on the plasma membrane may permit the maintenance of biological functions. Further studies are required to see if higher serum levels of NB-DNJ can inhibit GSL biosynthesis to the point where pathology does result.
It has been demonstrated that GSL accumulation is not intrinsically toxic to cells unless the accumulation excedes a critical threshold level and leads to pathology. This has been elegantly demonstrated in studies of mouse models of the GSL storage diseases, Tay-Sachs and Sandhoff's disease (31, 32). Taken together with the data in this report this would suggest that GSLs can be depleted or accumulated by mammalian cells within certain limits, without resulting in cellular dysfunction.
In conclusion, GSL levels can be extensively depleted long term in vivo without overt toxicity. GSL biosynthesis inhibitors such as NB-DNJ can now be evaluated in animal models of GSL storage diseases and also used as in vivo tools for probing GSL functions in mature and developing mice.
We thank David Smith, Julia MacAvoy, and Hamish McMath for excellent technical assistance; Searle/Monsanto for NB-DNJ; and Colin Beesley for photography.