The Journal of Clinical Investigation| January 2004|Volume 113|Number 2
Lysosomal storage disorders (LSDs) arise from func-
tional defects in one or more of the proteins essential
to normal lysosome function. This typically involves
the enzymes that play a critical role in the intracellular
digestion of glycoproteins, glycolipids, glycosamino-
glycans, or other macromolecules (1). GM2 gangliosi-
doses, one of the major LSDs, are caused by an abnor-
mality in the hexosaminidases (Hex’s) (1, 2). Hex A
consists of a heterodimer of a β-subunit (HEXA gene
product) and an α-subunit (HEXB gene product). Hex
B is a homodimer of β-subunits. Mutations in the
HEXA gene cause Tay-Sachs disease, whereas muta-
tions in the HEXBgene cause Sandhoff disease (SD) (1).
Mice with disruptions in the Hexb gene develop an
SD-like illness and therefore have provided a useful
model for investigating the pathophysiology of SD
(3–5). Neurologic dysfunction is the major clinical
manifestation of GM2 gangliosidoses, correlating
closely with the severity of the illness. These neuro-
logical abnormalities have been ascribed in part to
neuronal cell death caused by the accumulation of
both undigested GM2 gangliosides and related
lipids in neuronal lysosomes (6). However, several
recent investigations have suggested that ganglio-
side accumulation in neurons alone cannot com-
pletely explain the nerve cell damage and the short
life span that characterizes SD (Hexb–/–) mice. It was
previously reported that bone marrow transplanta-
tion (BMT) from normal (Hexb+/+) to SD mice sup-
presses neuronal death and improves survival ratios
despite having no effect on either Hex activity or
ganglioside accumulation in the brain (7, 8). Cer-
tain inflammatory processes, including microglial
activation, indicated by cytokine expression were
also observed (8–10). In addition, other recent stud-
ies indicated that antiganglioside antibodies were
present in several neuropathies, including Guillain-
Barré syndrome, amyotrophic lateral sclerosis, and
Fisher syndrome (11–13).
Thus, we hypothesized that substrates such as GM2
and/or GA2 could evoke autoimmune responses
because they could not easily be degraded or cleared
due to specific enzyme deficiencies. As a consequence
of this, autoantibodies would bind to antigens on the
cell surface of neurons and trigger microglial activa-
tion via the Fc receptor (FcR) common γ-chain. Stud-
ies were undertaken to evaluate this hypothesis.
Received for publication July 29, 2003, and accepted in revised form
November 11, 2003.
Address correspondence to: Shoji Yamanaka, Department of
Pathology, Yokohama City University School of Medicine,
3-9 Fukuura, Kanazawa-ku, Yokohama, Kanagawa 236-0004,
Japan. Phone: 81-45-787-2586; Fax: 81-45-786-0191;
Conflict of interest: The authors have declared that no conflict of
Nonstandard abbreviations used: lysosomal storage disorder
(LSD); hexosaminidase (Hex); Sandhoff disease (SD); bone
marrow transplantation (BMT); Fc receptor (FcR); high-
performance, thin-layer chromatography (HPTLC); propidium
iodine (PI); blood-brain barrier (BBB); systemic lupus
erythematosus (SLE); death-associated protein 6 (DAXX).
Possible role of autoantibodies in the pathophysiology
of GM2 gangliosidoses
Akira Yamaguchi,1Kayoko Katsuyama,1Kiyotaka Nagahama,1Toshiyuki Takai,2
Ichiro Aoki,1and Shoji Yamanaka1
1Department of Pathology, Yokohama City University School of Medicine, Yokohama, Japan
2Department of Experimental Immunology and Core Research for Evolution Science and Technology Programme of Japan
Science and Technology Corp., Institute of Development, Aging and Cancer, Tohoku University School, Sendai, Japan
Mice containing a disruption of the Hexb gene have provided a useful model system for the study
of the human lysosomal storage disorder known as Sandhoff disease (SD). Hexb–/–mice rapidly
develop a progressive neurologic disease of ganglioside GM2 and GA2 storage. Our study revealed
that the disease states in this model are associated with the appearance of antiganglioside autoan-
tibodies. Both elevation of serum antiganglioside autoantibodies and IgG deposition to CNS neu-
rons were found in the advanced stages of the disease in Hexb–/–mice; serum transfer from these
mice showed IgG binding to neurons. To determine the role of these autoantibodies, the Fc recep-
tor γ gene (FcRγ) was additionally disrupted in Hexb–/–mice, as it plays a key role in immune com-
plex–mediated autoimmune diseases. Clinical symptoms were improved and life spans were
extended in the Hexb–/–FcRγ–/–mice; the number of apoptotic cells was also decreased. The level
of ganglioside accumulation, however, did not change. IgG deposition was also confirmed in the
brain of an autopsied SD patient. Taken together, these findings suggest that the production of
autoantibodies plays an important role in the pathogenesis of neuropathy in SD and therefore
provides a target for novel therapies.
J. Clin. Invest. 113:200–208 (2004). doi:10.1172/JCI200419639.
The Journal of Clinical Investigation
|January 2004| Volume 113|Number 2
Mice. All mice used in this study were bred and
housed under standard nonsterile conditions. All
animal studies were approved by the Animal Com-
mittee at Yokohama City University. SD mice (Hexb–/–
mice; C57BL/6X129/Sv background) were kindly
provided by R. L. Proia (NIH) and were bred in a
closed colony over 30 generations so that they were
inbred for C57BL/6- and 129/Sv-derived genes (4).
FcRγ gene knockout mice (FcRγ–/–mice) (14) with
C57BL/6 were bred with Hexb–/–mice to obtain dou-
bly heterozygous (Hexb+/–FcRγ+/–) mice. To minimize
variations due to differences in genetic background,
all genotypes used in this study were derived from
these cross-breedings, and the sib pairs were used as
a comparison. The hexb and FcRγ genotypes were
determined by PCR using tail DNA (15). Primer
sequences used for FcRγ were as follows: (1) 5′-GCC-
3′. In the FcRγ–/–and WT genotypes, 1.2-kb and 0.24-
kb bands were observed, respectively. After denatura-
tion at 94°C for 1 minute, the PCR reaction was
cycled 30 times at 94°C for 30 seconds and 68°C for
5 minutes. PCR products were separated by elec-
trophoresis through a 2% (w/v) agarose gel and visu-
alized by ethidium bromide staining.
Antibodies. A rabbit IgG antibody against mouse GA2
was obtained from Dia-iatron (Tokyo, Japan). Biotin-con-
jugated anti-N-acetyl GM2 antibody that recognizes
mouse GM2 was obtained from Seikagaku Corp. (Tokyo,
Japan). Alkaline phosphatase–labeled goat antimouse
IgG antibody, which also recognizes human IgG, was
obtained from Southern Biotechnology Associates Inc.
(Birmingham, Alabama, USA). Alexa fluor goat antirab-
bit IgG and fluor donkey antigoat IgG were obtained
from Molecular Probes (Eugene, Oregon, USA). Strepta-
vidin-FITC–conjugated GM2 was purchased from
GIBCO BRL (Tokyo, Japan). Horseradish peroxidase–
conjugated goat antimouse IgG was obtained from ICN
Pharmaceuticals Inc. (Costa Mesa, California, USA).
Human brain tissue. A human brain sample, taken
from a 2-year-old boy with SD who was autopsied at
Nagoya City University Medical School, was kindly pro-
vided by M. Tatematsu (16). This experiment was
approved by the Ethical Committee of Yokohama City
University School of Medicine.
Extraction and analysis of glycolipids. Glycolipids were
extracted from mouse brain as described previously (4),
applied to high-performance, thin-layer chromatogra-
phy (HPTLC) plates (Silicagel 60, Merck, Darmstadt,
Germany), and chromatographed with chloroform/
methanol/0.22% aqueous CaCl2 (60:40:9, v/v/v). The
glycolipids were then visualized by spraying the HPTLC
plates with anthrone–sulfuric acid reagent followed by
heating on a hot plate.
ELISA. Antiganglioside antibodies were determined
by ELISA assays as described previously (17). Briefly,
flat-bottom plates (Greiner Immunoplate, Fricken-
hausen, Germany) were coated with appropriately
diluted soluble antigens (extracted from Hexb–/–mouse
brain and liver) and then blocked with 1% BSA in PBS.
Serum diluted in PBS containing 1% BSA was incubat-
ed on antigen-coupled plates for 2 hours. Unbound
immunoglobulin was washed away with PBS contain-
ing 0.05% Tween 20. Horseradish peroxidase–conju-
gated goat antimouse IgG was added for a further 2
hours and the plates were washed again. A color reac-
tion was obtained by adding orthophenylenediamine-
HCl substrates (Wako, Tokyo, Japan). The data were
determined by OD values at 405 nm with reference to
the OD of normal WT mouse (C57BL/129) serum.
Serum cytokines were detected by the ELISA method
using a Mouse Interleukin-4 ELISA Kit (Endogen Inc.,
Rockford, Illinois, USA) and a Mouse Interferon
Gamma ELISA Kit (Endogen Inc.).
Immunofluorescence study. Organs were removed from
the necropsied animals and processed for frozen or
paraffin sections. The sections were pretreated with
PBS containing 3% BSA (Nissui, Tokyo, Japan) for 15
minutes and incubated overnight with antibody at
4°C. For double staining, the sections were incubated
with propidium iodine (PI) solution for nuclear stain-
ing after the secondary fluorescent antibody reaction.
Stained sections were observed using laser scanning
microscopy (LSM 101, Olympus, Tokyo, Japan).
Immunohistochemistry. For the detection of IgG deposi-
tion, alkaline phosphate–labeled goat antimouse IgG
antibody was used for both mouse and human brain tis-
sue samples using a Histofine kit according to the man-
ufacturer’s instructions (Nichirei Corp., Tokyo, Japan).
Serum transfer experiment with blood-brain barrier dis-
ruption. D-Mannitol was chosen as a hypertonic agent
to open the blood-brain barrier (BBB) osmotically.
The safe limits for its experimental use in animals has
been described elsewhere (18, 19). Mannitol was dis-
solved in PBS at a concentration of 1.6 mol/l and
kept at 37°C. Three hundred microliters of pooled
sera from either WT (Hexb+/+) or Hexb–/–mice that
were between 14 and 15 weeks of age was injected
into the tail vein of 8-week-old SCID mice with 200
µl of 1.6 mol/l D-mannitol. The mice were killed 30
minutes after the injections.
Apoptosis analysis. Apoptotic cells were detected by the
in situ TUNEL method using a DeadEnd Colorimet-
ric TUNEL system (Promega, Madison, Wisconsin,
USA). An apoptotic DNA ladder was detected using a
TACS Apoptotic DNA Laddering Kit and Tissue Sup-
plement Reagents (Trevigen Inc., Gaithersburg, Mary-
land, USA). One microgram of extracted DNA taken
from the brain of a 14-week-old mouse was subjected
to electrophoresis on a 2% (w/v) agarose gel contain-
ing ethidium bromide (0.4 mg/ml).
Behavioral testing. Motor function and balance were
measured with a standard rotorod apparatus (Ugo
Basile 7650; BM Apparatus, Tokyo, Japan). The
machine was set to an initial speed of 4 rpm, and the
acceleration was increased by 2 rpm/min. The latency
The Journal of Clinical Investigation|January 2004| Volume 113| Number 2
testing was conducted once a week for 8-week-old
mice until the time when they could no longer walk
due to gait apraxia.
Clinical scores. Neurological activity in mice was
scored using the following criteria: a score was
assigned for the effect when a mouse was lifted by the
tail and also for the ability to roll over when the
mouse was left in a dorsal position. Each score was
graded into the following five categories: 0, normal:
limbs extended and body twisted, able to roll back to
prone from supine position; 1, slight: limbs extended
with slight tremor and twisted body, able to roll back
to prone from supine position; 2, mild: limbs extend-
ed without twisted body, able to roll back to prone
to fall off from the rotating cylinder was also meas-
ured. Mice that fell within 15 seconds were given a sec-
ond trial, and those that did not fall during a 250-sec-
ond period were given a score of 250 seconds. Rotorod
Detection of anti-GM2 and-GA2 antibodies, IL-4 and INF-γby ELISA.
Autoantibodies against GM2 (a) and GA2 (b) in serum were meas-
ured in Hexb–/–(squares) and Hexb+/+(diamonds) mice. Serum IL-4
(c) and INF-γ (d) levels were measured in 14-week-old Hexb–/–and
Hexb+/+mice. n = eight mice per group. *P < 0.01, Student’s t test.
Deposition of IgG in the brain. (a) Immunofluorescent staining with anti-GA2 antibody (green) and nuclear staining with PI (red) is shown
for frozen brain sections. Both cell surface expression and intracellular storage of GA2 in the cerebella of 14-week-old Hexb+/+mice and
Hexb–/–mice are shown (left panel). ML, molecular layer; Pc, Purkinje cells; GL, granular layer. The presence of IgG detected by anti-IgG anti-
body (green) is shown in the same area (right panel). Representative sections with three mice per group are shown. IgG is detected only in
Hexb–/–mice. Original magnification, ×400. (b) Immunohistochemical staining with alkaline phosphatase–labeled anti-IgG antibody shows
IgG deposition in the neurons of the thalamus (red) of 14-week-old Hexb–/–mice with no IgG deposition evident in the thalamus of age-
matched Hexb+/+mice. Representative sections are shown. Methylgreen was used for counterstaining. Original magnification, ×200. (c) Depo-
sition of IgG present in the hippocampal commissure of BBB-disrupted SCID mice following injection of 14-week-old Hexb–/–mice serum.
Prior to injection, the presence of the GA2 (stained green) was confirmed in SCID mice (left). Deposition of IgG was investigated following
BBB disruption. Serum transfer from the Hexb+/+mice did not reveal IgG deposition (middle), whereas the presence of IgG was demonstrated
in the same area of the SCID mice injected with Hexb–/–mice sera (right). Green, presence of Alexa-labeled GA2 (left) or Alexa-labeled IgG
(middle and right); red, nuclear staining with PI. Arrows indicate axons or dendrites that were stained green. Original magnification, ×400.
The Journal of Clinical Investigation
| January 2004| Volume 113| Number 2
from supine position; 3, moderate: flexed and paws
clenched, able to roll back to prone from supine posi-
tion; 4, severe: flexed and paws clenched, unable to
roll back to prone from supine position.
Data analysis. A Student’s t test was employed to eval-
uate the significance of the data. P < 0.05 was consid-
ered statistically significant.
Serum anti-GM2 and GA2 antibodies are elevated in aged
Hexb–/–mice. First, we conducted experiments accord-
ing to our proposed hypothesis that autoantibodies are
associated with the neuronal damage and symptoms of
gangliosidoses. Autoantibodies for the gangliosides
GM2 and GA2 were therefore evaluated by serum
ELISA. Significant elevation of serum antibody levels
was observed in the terminal stages of Hexb–/–diseased
mice that were more than 14 weeks old (Figure 1, a and
b), which was consistent with our hypothesis. Associ-
ated with this autoantibody production, the level of
serum IL-4 (Figure 1c), but not IFN-γ (Figure 1d), was
also elevated, suggesting that certain immune and
inflammatory processes may be occurring in aged
Hexb–/–mice. Thus, the release of autoantigens from
ganglioside-accumulating cells may be involved in the
production of autoantibodies.
IgG deposition is observed on the cell surface of neurons in the
CNS of Hexb–/–mice, but not WT mice. Next we evaluated
the presence of autoantibodies in the CNS. Tissue sec-
tions from various CNS sites were taken from 14-week-
old Hexb–/–mice and analyzed for GA2 accumulation
and IgG deposition (Figure 2a). Immunohistochem-
istry revealed that the Purkinje cells of both WT and
Hexb–/–mice showed specific expression of GA2 on
their cell surfaces. In contrast to Purkinje cells in WT
mice, however, those in the Hexb–/–mice simultane-
ously showed IgG deposition. Similar findings were
observed in various sites of the CNS including the thal-
amus, where extensive apoptosis is known to occur
(Figure 2b), as has been reported previously (8).
Serum transfer from Hexb–/–mice to BBB-disrupted mice
reveals the presence of antibodies that bind to neurons. To con-
firm the presence of autoantibodies that can bind to
the cell surface of neurons, we disrupted the BBB of
SCID mice by D-mannitol and injected sera from either
Hexb–/–or WT mice. SCID mice were used in this exper-
iment, since they have low background levels of
immunoglobulins. Expression of GA2 was confirmed
in the hippocampal commissure of SCID mice (Figure
2c). Following the disruption of BBB and the transfer
of serum from either WT or Hexb–/–mice, the hip-
pocampal commissure areas of the SCID mice were
analyzed for immunoglobulin binding using anti-IgG
antibodies. IgG-bearing neurons were seen in the SCID
mice that had been injected with Hexb–/–serum, where-
as SCID mice injected with WT serum did not show
IgG binding either on neuronal cell surfaces or in the
cytoplasm (Figure 2c). It is noteworthy that a previous
study reported altered BBB permeability in Hexb–/–
mice (10). Our results indicate that autoantibodies
move to the brain parenchyma via BBB and bind to the
cell surface of neurons in Hexb–/–mice.
Double-knockout (Hexb–/–FcRγ–/–) mice show better clini-
cal outcomes and lower GA2 antibody levels than Hexb–/–
FcRγ+/+mice. It was difficult to determine the relation-
ship between IgG deposition and apoptosis, since
apoptosis was limited to specific areas whereas IgG
deposition was seen diffusely. To determine the role of
autoantibodies more precisely, we generated mice lack-
ing both the FcRγ gene and the Hexb gene by inter-
crossing (14). The double-knockout genotypes were
confirmed by PCR of tail DNA (Figure 3a). As seen in
Figure 3b, the Hexb–/–FcRγ+/+mouse at 14 weeks old was
less active, quite lean, and displayed muscle atrophy
with wide-opening hind limbs to support the body,
whereas the Hexb–/–FcRγ–/–mouse of the same age was
active and mobile. Serum antibodies against GA2 were
also evaluated in the Hexb–/–FcRγ–/–mice. A significant
decrease in serum antibody levels over Hexb–/–FcRγ+/+
mice was noted in 14-week-old Hexb–/–FcRγ–/–mice,
Generation of Hexb–/–FcRγ–/–mice. (a) PCR analysis of mouse tail
DNA. From left lane: Hexb+/+FcRγ+/+, Hexb–/–FcRγ+/+, Hexb–/–FcRγ+/–,
and Hexb–/–FcRγ–/–mice. (b) Clinical appearance of Hexb–/–FcRγ+/+
and Hexb–/–FcRγ–/–mice at 14 weeks. (c) Investigation of the titers
of anti-GA2 antibodies in 14-week-old mice. Anti-GA2 antibody
was at lower levels in Hexb–/–FcRγ–/–mice than in Hexb–/–FcRγ+/+
mice. n = 5–10. *P < 0.05, Student’s t test.
The Journal of Clinical Investigation| January 2004|Volume 113|Number 2
although there was a slight, but statistically significant,
elevation in these levels when compared to WT
(Hexb+/+FcRγ+/+) mice (Figure 3c).
Patterns and levels of ganglioside accumulation in
Hexb–/–FcRγ–/–mice are similar to those seen in Hexb–/–FcRγ+/+
mice. The levels of accumulated GM2 and GA2 at 14
weeks of age were examined in the brains of
Hexb–/–FcRγ–/–mice. Immunohistostaining of the cere-
bral cortex with antibodies against GM2 is shown in Fig-
ure 4a. Hexb–/–FcRγ–/–mice revealed no differences in
GM2 accumulation compared with Hexb–/–FcRγ+/+mice.
We then extracted the gangliosides, and, following analy-
sis by HPTLC, it was revealed that the storage of GM2
and GA2 in the Hexb–/–FcRγ–/–mice did not change
when compared with Hexb–/–FcRγ+/+mice (Figure 4b).
Neurological functions are partially restored and life spans
are extended in Hexb–/–FcRγ–/–mice. Fourteen-week-old
Hexb–/–FcRγ+/+mice showed limb flexion when lifted
by the tail. On the other hand, both WT and
Hexb–/–FcRγ–/–mice of the same age were able to open
their limbs and move actively (Figure 5a). Clinical
scores were judged by (i) the position of the forelimb
and hindlimb observed by tail lifting and (ii) the abil-
ity to roll over. Both female and male Hexb–/–FcRγ–/–
mice always acquired a better score than
Hexb–/–FcRγ+/+mice, demonstrating that they had a
distinctly milder form of the disease with far less
severe neuropathy (Figure 5b).
A rotorod test, which measures balance and mobility,
was performed on Hexb–/–FcRγ–/–mice from 10 to 15
weeks of age (Figure 5c). Both female and male
Hexb–/–FcRγ–/–mice constantly attained better per-
formances than Hexb–/–FcRγ+/+mice during this test.
Analysis of ganglioside storage. (a) Immunohistochemical detection of GM2 in the cerebral cortex of a 14-week-old mouse. Both Hexb–/–FcRγ+/+and
Hexb–/–FcRγ–/–mice, but not WT (Hexb+/+FcRγ+/+), showed GM2 storage in the cerebellum (stained green or yellow). PI was used for nuclear stain-
ing. (b) Extraction and analysis of brain gangliosides. Gangliosides were extracted from the brains of 14-week-old mice and loaded onto HPTLC
plates. Both Hexb–/–FcRγ–/–and Hexb–/–FcRγ+/+mice showed similar levels of GA2 and GM2 storage. M, bovine ganglioside marker.
Amelioration of motor defects in Hexb–/–FcRγ–/–mice. (a) When sus-
pended by their tails, Hexb–/–FcRγ+/+mice flexed their limbs, whereas
Hexb–/–FcRγ–/–mice and Hexb+/+FcRγ+/+mice extended their limbs. (b)
Clinical scores were determined by the degree of neurological sever-
ity and counted as follows: 0 = normal, 1 = slight, 2 = mild, 3 = mod-
erate, 4 = severe. Hexb+/+FcRγ+/+, Hexb–/–FcRγ+/+, and Hexb–/–FcRγ–/–
mice were tested from 10 to 15 weeks. n = 5–10. *P <0.05, Student’s
t test (Hexb–/–FcRγ+/+ mice vs. Hexb–/–FcRγ–/–mice). Squares,
Hexb–/–FcRγ+/+mice; diamonds, Hexb+/+FcRγ+/+mice; triangles,
Hexb–/–FcRγ–/–mice. (c) Rotorod testing was performed by counting
the mean time to fall from the rotating rod. n = 5–10. *P <0.05, Stu-
dent’s t test (Hexb–/–FcRγ+/+mice vs. Hexb–/–FcRγ–/–mice). Squares,
Hexb–/–FcRγ+/+mice; diamonds, Hexb+/+FcRγ+/+mice; triangles,
The Journal of Clinical Investigation
|January 2004|Volume 113|Number 2
The performance of the Hexb–/–FcRγ–/–mice was con-
sistent up to the 12-week point but deteriorated grad-
ually thereafter. The mean survival time of both male
and female Hexb–/–FcRγ+/+mice was 102 days (86–115
days, respectively), whereas that of male and female
Hexb–/–FcRγ–/–mice was 130 days (116–145 days) and
128 days (122–135 days), respectively (Figure 6).
Neurodegeneration of Purkinje cells and apoptosis are less
extensive in Hexb–/–FcRγ–/–mice. Neurodegeneration in
Hexb–/–FcRγ+/+and Hexb–/–FcRγ–/–mice was examined in
the cerebella of 14-week-old animals. As shown in Fig-
ure 7, Purkinje cells of Hexb–/–FcRγ+/+mice of the same
age were either lost or had degenerated. In contrast,
Purkinje cells in the Hexb–/–FcRγ–/–mice were almost as
well preserved as in WT (Hexb+/+FcRγ+/+) mice. We also
assessed the nuclei of the Purkinje cells by PI staining.
In Hexb–/–FcRγ+/+ mice, the nuclei appeared smaller and
degenerated, whereas in Hexb–/–FcRγ–/–mice, the nuclei
appeared morphologically similar to WT. Based upon
these findings, the degeneration of Purkinje cells in
Hexb–/–FcRγ–/–mice was considered to be minimal.
We next evaluated apoptosis in the brains of 14-week-
old mice (Figure 8). Using TUNEL assays, positive cells
suggestive of apoptosis were found in the thalamus of
Hexb–/–FcRγ+/+mice (Figure 8a), whereas in Hexb–/–FcRγ–/–
and WT mice, these TUNEL-positive cells were barely
detectable. The number of TUNEL-positive cells in a
sagittal section was counted (Figure 8b) and was shown
to have decreased significantly in Hexb–/–FcRγ–/–mice
when compared with Hexb–/–FcRγ+/+mice. Apoptotic
changes were also evaluated by a DNA ladder test, in
which extracted DNA from mouse brain was run on an
agarose gel. DNA from Hexb–/–FcRγ+/+mouse brain
appeared as a ladder with a smear pattern, whereas in
Hexb–/–FcRγ–/–mice, a very faint ladder/smear pattern was
visible. From these findings, we conclude that apoptosis
did not occur extensively in the brain of the Hexb–/–FcRγ–/–
mice when compared with Hexb–/–FcRγ+/+mice.
IgG deposition were also observed in the brain of an SD
patient. To determine whether autoimmune reactions
were also present in human SD, we obtained autopsied
tissue samples from the brain of an SD patient and
examined them histologically. Neurons in the cerebel-
lum were found to be distended with evidence of gan-
glioside storage (Figure 9, a and b). IgG deposition was
investigated in the cerebellum of this patient and was
indeed evident in neurons, similar to the findings
observed in Hexb–/–mice (Figure 9, c and d). Thus, the
same immunological mechanisms found in the mouse
may also be involved in human SD.
LSDs comprise a family of more than 40 different dis-
orders that result from deficiencies in various lysoso-
mal enzymes (1). These disorders have much in com-
mon, such as pathological lysosomal swelling due to
lysosomal storage of undegraded substrates and a
mannose-6-phosphate receptor–mediated enzyme
recapture system that has been targeted in the devel-
opment of gene therapy for these diseases (20–23).
Although emerging genetic technologies have uncov-
ered the primary genetic defects of many LSDs, many
uncertainties remain concerning the mechanism(s)
leading to cell injury and disease progression. Extreme
lysosomal swelling may lead to malfunction of cells
that would delay signal transduction, but it is ques-
tionable whether this alone can account for the severe
phenotypes seen in patients. In this study we investi-
gated the pathogenesis of SD mice from the perspec-
tive of autoimmune features.
In the SD mouse model, both GM2 and GA2 accu-
mulate in lysosomes due to defects in Hex’s A and B
(4). These gangliosides are abundant in the neurons
Survival of Hexb–/–FcRγ–/–mice. Male and female Hexb–/–FcRγ–/–mice
(squares, n = 10) and Hexb–/–FcRγ+/+mice (circles, n = 10–15) were
monitored daily. *P < 0.0005, Student’s t test.
Reduction in the neurodegeneration of Purkin-
je cells in Hexb–/–FcRγ–/– mice and 14-week-old
Hexb–/–FcRγ+/+mice, but not Hexb–/–FcRγ–/–or
Hexb+/+FcRγ+/+, displayed neurodegeneration
and shrinkage of nuclei in Purkinje cells. Arrows
indicate nuclei of Purkinje cells. H&E stain, orig-
inal magnification, ×100. PI stain, original mag-
nification, ×600. ML, molecular layer, Pc, Purk-
inje cells, GL, granular layer.
The Journal of Clinical Investigation|January 2004| Volume 113|Number 2
of the CNS but are also present to some degree in
visceral organs such as the liver, spleen, and thymus.
In our study, we hypothesized that accumulated gan-
gliosides might not be easily degraded and cleared
due to the absence of Hex’s. Thus, such undegraded
substrates could potentially be targeted by the
immune system, resulting in an autoimmune re-
sponse. Similar mechanisms have been suggested in
some cases of systemic lupus erythematosus (SLE),
where low-serum DNase I activity, due to DNase I
mutations, may result in undegraded nuclear DNA
and hence, antinuclear DNA antibody formation
(24). Indeed, DNase I–knockout mice do develop
SLE-like glomerulonephritis with the formation of
anti-chromatin autoimmunity (25). The antigenici-
ty of gangliosides is further indicated by Guillain-
Barré syndrome, amyotrophic lateral sclerosis, and
Fisher syndrome (11–13), and we therefore postulat-
ed that accumulated gangliosides in SD mice might
serve as autoantigens.
An initial series of experiments demonstrated
increased levels of serum anti-GM2 and anti-GA2
antibodies in 14-week-old SD mice that showed
advanced illness. Immunohistochemical studies of
these mice also revealed IgG deposition in neurons of
the CNS. IgG deposition in CNS neurons may be
consistent with a recent report showing increased
BBB permeability not only in SD mice but also in
GM1 gangliosidosis mice (10). Although it is still
uncertain whether antiganglioside antibodies accu-
mulate in neurons, serum from SD mice successfully
transferred IgG deposition to the CNS neurons of
SCID mice that had undergone D-mannitol treat-
ment to disrupt the BBB. These data strongly sug-
gested the presence of specific autoimmunity in SD
mice. We hypothesized that antigen-antibody com-
plexes on cell surfaces would trigger type II hyper-
sensitivity reactions with microglial activation via
FcRγ because microglial activation has been reported
in SD mice (8, 9), and FcRγ-knockout mice are resist-
ant to the induction or spontaneous onset of various
autoimmune diseases (26–28).
To further clarify this hypothesis, we generated and
analyzed Hexb–/–FcRγ–/–double-knockout mice that
lack the FcRγ in SD mice. In the Hexb–/–FcRγ–/–mice,
amelioration of neurological symptoms, life span
increases, reduction in the degeneration of Purkinje
cells, decreased levels of neuronal cell death, and
decreased levels of anti-GA2 antibodies were all
observed. However, as found in BMT experiments, the
accumulation of gangliosides in SD mice was also evi-
Reduction of apoptosis in the brain
of Hexb–/–FcRγ–/–mice. (a) At 14
weeks, Hexb–/–FcRγ–/–mice were ana-
lyzed by in situ TUNEL analysis, and
compared with Hexb–/–FcRγ+/+and
Hexb+/+FcRγ+/+mice. Arrows indicate
apoptotic cells. Original magnifica-
tion, ×100 (thalamus). (b) Numbers of
apoptotic (TUNEL positive) cells were
counted in a sagittal section of the
brain. Data are mean ± SEM (n = 4).
*P < 0.001, Student’s t test. (c) DNA
laddering in the
Hexb+/+FcRγ+/+, Hexb–/–FcRγ–/–, and
Hexb–/–FcRγ+/+mice. One microgram
of brain DNA was subjected to elec-
trophoresis on a 2% agarose gel and
stained with ethidium bromide. The
lane at the left of the panel shows a
1-kb DNA ladder.
IgG deposition in the cerebellar neurons of an SD patient. (a and b)
H&E staining of cerebellar sections showed cells with typical bal-
looning. Arrows indicate enlarged cells with ganglioside storage. (c
and d) Immunohistochemical staining of the cerebellum with alka-
line phosphatase–labeled anti-IgG antibody. Arrows indicate cells
with IgG deposition (stained as red). (a and c) Original magnifica-
tion, ×200; (b and d) Original magnification, ×400. ML, molecular
layer; Pc, Purkinje cells; GL, granular layer.
The Journal of Clinical Investigation
|January 2004| Volume 113|Number 2
dent in Hexb–/–FcRγ–/–mice (7). These data suggest
that this accumulation is one of the factors in the
pathophysiology of LSDs.
Recent substantial studies have suggested that
inflammatory processes are involved in the develop-
ment of the neuropathology of SD mice (8, 10).
Using cDNA microarray analysis, Wada et al. revealed
the upregulation of genes related to an inflammato-
ry process dominated by activated microglia and con-
firmed that there were increased numbers of activat-
ed microglia and increased level of TNF-α mRNA in
both SD mice and an autopsied human SD patient
(8). Gene expression profiling by serial analysis of
gene expression in human GM2 gangliosidoses,
including Tay-Sachs disease, supported the role of
microglial and astroglial activation, and demon-
strated the increased expression of death-associated
protein 6 (DAXX), which encodes a protein that asso-
ciates with the Fas receptor and is involved in apop-
tosis via Fas (9). Moreover, microglial activation
accompanied by the production of many other
cytokines was previously reported in both SD mice
and GM1 gangliosidosis mice (10).
Based on our findings and the reports of other lab-
oratories, we speculate that antibody-ganglioside
complexes are important for the induction of
microglial activation that is mediated via FcRγ in SD
mice. Data showing that BMT ameliorates neurolog-
ic manifestations in SD mice can be explained by
decreased anti-GM2 and anti-GA2 antibody forma-
tion, which we have observed (unpublished data).
Activated microglia would produce cytokines includ-
ing TNF-α, which is known to induce apoptosis
through DAXX (29, 30). The neurodegeneration that
is observed in Purkinje cells of SD mice may therefore
be the result of proinflammatory cytokines from acti-
vated microglia, although the precise mechanisms
underlying this remain to be determined. Another
possible mechanism by which activated microglia may
cause SD is through phagocytosis via the FcRγ chain.
Antibody-ganglioside complexes may become op-
sonized and could potentially be either phagocytosed
or injured by microglia. Apoptotic neurons bearing
these antibody-ganglioside immune complexes would
therefore be phagocytosed and internalized by mi-
croglia. As a result of these mechanisms, neuronal
injury, degeneration, and cell death are induced. As a
result, microglial activation would accelerate antigen
presentation and inflammatory responses that aug-
ment increasing levels of autoantibody production. A
pathological study is currently ongoing in our labo-
ratory to precisely evaluate microglial activation in
Hexb–/–FcRγ–/–mice. Preliminary results indicate that
there is reduced microglial activation, which supports
Importantly, the autoantibody mechanisms evident
in mouse models appear to exist also in human SD.
Although we have so far only demonstrated the pres-
ence of IgG deposition in a CNS specimen from a sin-
gle SD patient, it is highly possible that similar mech-
anisms exist in other cases of LSDs, including gan-
gliosidoses. The manifestation of these may well
depend on both the antigenicity and the levels of
antigens. Immunological disturbances of the sphin-
golipidoses have been reported in some cases of
Gaucher disease in which patients have significantly
higher concentrations of polyclonal immunoglobu-
lin (31). Gaucher disease mice generated by Mizuka-
mi et al. also showed systemic inflammatory changes
(32). It would be very interesting in the future to
investigate the autoantibody production associated
with other LSDs.
Our results clearly demonstrate the involvement of
the FcRγ gene and the possible role of an autoim-
mune response in the development of SD. Although
much still remains to be elucidated, our present
investigation has given us novel insights and fur-
thered our understanding of the pathogenesis of SD.
The contribution of autoimmunity to the pathogen-
esis of LSDs will be an important factor when devel-
oping effective new treatments to be used in combi-
nation with other therapies such as BMT (7, 33, 34),
enzyme replacement therapy (35), and substrate dep-
rivation (33, 36).
We thank Richard L. Proia for providing the SD mice
and Masae Tatematsu for the generous donation of
the human SD brain autopsy tissue. We also thank
Kyoko Suzuki, Eizo Iseki, Haruto Hojo, Yoji
Nagashima, Yoshihiro Kiuchi, Dennis M. Klinman,
Rieko Ijiri, Michiko Ehara, and Yoshitsugu Kojima
for technical assistance and helpful discussions. This
project was carried out with financial support from
a grant-in-aid for scientific research from the Japan-
ese Society for the Promotion of Science and a grant
in support of Promotion of Research from Yoko-
hama City University.
1.Neufeld, E.F. 1991. Lysosomal storage diseases. Annu. Rev. Biochem.
2.Gravel, R.A., et al. 2001. The GM2 gangliosidoses. In The metabolic and
molecular basis of inherited disease. C.R. Scriver et al., editors. McGraw-Hill.
New York, New York, USA. 3827–3876.
3.Yamanaka, S., Johnson, O.N., Norflus, F., Boles, D.J., and Proia, R.L.
1994. Structure and expression of the mouse beta-hexosaminidase genes,
Hexa and Hexb. Genomics. 21:588–596.
4.Sango, K., et al. 1995. Mouse models of Tay-Sachs and Sandhoff diseases
differ in neurologic phenotype and ganglioside metabolism. Nat. Genet.
5.Phaneuf, D., et al. 1996. Dramatically different phenotypes in mouse
models of human Tay-Sachs and Sandhoff diseases. Hum. Mol. Genet.
6.Huang, J.Q., et al. 1997. Apoptotic cell death in mouse models of GM2
gangliosidosis and observations on human Tay-Sachs and Sandhoff dis-
eases. Hum. Mol. Genet. 6:1879–1885.
7.Norflus, F., et al. 1998. Bone marrow transplantation prolongs life span
and ameliorates neurologic manifestations in Sandhoff disease mice.
J. Clin. Invest. 101:1881–1888.
8.Wada, R., Tifft, C.J., and Proia, R.L. 2000. Microglial activation precedes
acute neurodegeneration in Sandhoff disease and is suppressed by bone
marrow transplantation. Proc. Natl. Acad. Sci. U. S. A. 97:10954–10959.
9.Myerowitz, R., et al. 2002. Molecular pathophysiology in Tay-Sachs and
Sandhoff diseases as revealed by gene expression profiling. Hum. Mol.
208 Download full-text
The Journal of Clinical Investigation|January 2004|Volume 113|Number 2
10.Jeyakumar, M., et al. 2003. Central nervous system inflammation is a
hallmark of pathogenesis in mouse models of GM1 and GM2 gan-
gliosidosis. Brain. 126:974–987.
11.Yuki, N. 1999. Pathogenesis of Guillain-Barre and Miller Fisher syn-
dromes subsequent to Campylobacter jejuni enteritis. Jpn. J. Infect. Dis.
12.Niebroj-Dobosz, I., Jamrozik, Z., Janik, P., Hausmanowa-Petrusewicz,
I., and Kwiecinski, H. 1999. Anti-neural antibodies in serum and cere-
brospinal fluid of amyotrophic lateral sclerosis (ALS) patients. Acta
Neurol. Scand. 100:238–243.
13.O’Hanlon, G.M., et al. 2000. Peripheral neuropathy associated with anti-
GM2 ganglioside antibodies: clinical and immunopathological studies.
14.Takai, T., Li, M., Sylvestre, D., Clynes, R., and Ravetch, J.V. 1994. FcR
gamma chain deletion results in pleiotrophic effector cell defects. Cell.
15.Sango, K., Yamanaka, S., Ajiki, K., Tokashiki, A., and Watanabe, K. 2002.
Lysosomal storage results in impaired survival but normal neurite out-
growth in dorsal root ganglion neurons from a mouse model of Sand-
hoff disease. Neuro. Pathol. Appl. Neurobiol. 28:1–17.
16.Tatematsu, M., et al. 1981. Sandhoff disease. Acta Pathol. Jpn.31:503–512.
17.Taki, T., Nishiwaki, S., Ishii, K., and Handa, S. 1990. A simple and spe-
cific assay of glycosyltransferase and glycosidase activities by an enzyme-
linked immunosorbent assay method, and its application to assay of
galactosyltransferase activity in sera from patients with cancer. J. Biochem.
18.Bhattacharjee, A.K., Nagashima, T., Kondoh, T., and Tamaki, N. 2001.
The effects of the Na(+)/Ca(++) exchange blocker on osmotic blood-
brain barrier disruption. Brain Res. 11:157–162.
19.Rapoport, S.I., Fredericks, W.R., Ohno, K., and Pettigrew, K.D. 1980.
Quantitative aspects of reversible osmotic opening of the blood-brain
barrier.Am. J. Physiol. 238:R421–R431.
20.Yamaguchi, A., et al. 2003. Plasmid-based gene transfer ameliorates vis-
ceral storage in a mouse model of Sandhoff disease. J. Mol. Med.
21.Neufeld, E.F., and Fratantoni, J.C. 1970. Inborn errors of mucopolysac-
charide metabolism. Science. 169:141–146.
22.Yew, N.S., and Cheng, S.H. 2001. Gene therapy for lysosomal storage dis-
orders. Curr. Opin. Mol. Ther. 3:399–406.
23.Bou-Gharios, G., Abraham, D., and Olsen, I. 1993. Lysosomal storage
diseases: mechanisms of enzyme replacement therapy. Histochem. J.
24.Walport, M.J. 2000. Lupus, DNase and defective disposal of cellular
debris. Nat. Genet. 25:135–136.
25.Napirei, M., et al. 2000. Features of systemic lupus erythematosus in
Dnase1-deficient mice. Nat. Genet. 25:177–181.
26.Takai, T. 2002. Roles of Fc receptors in autoimmunity. Nat. Rev. Immunol.
27.Ravetch, J.V., and Bolland, S. 2001. IgG Fc receptors. Annu. Rev. Immunol.
28.Abdul-Majid, K.B., et al. 2002. Fc receptors are critical for autoimmune
inflammatory damage to the central nervous system in experimental
autoimmune encephalomyelitis. Scand. J. Immunol. 55:70–81.
29.Shinoda, S., et al. 2003. Formation of a tumour necrosis factor receptor
1 molecular scaffolding complex and activation of apoptosis signal-reg-
ulating kinase 1 during seizure-induced neuronal death. Eur. J. Neurosci.
30.Yang, X., Khosravi-Far, R., Chang, H.Y., and Baltimore, D. 1997. Daxx, a
novel Fas-binding protein that activates JNK and apoptosis. Cell.
31.Marti, G.E., et al. 1988. Polyclonal B-cell lymphocytosis and hypergamma-
globulinemia in patients with Gaucher disease. Am. J. Hematol.29:189–194.
32.Mizukami, H., et al. 2002. Systemic inflammation in glucocerebrosidase-
deficient mice with minimal glucosylceramide storage. J. Clin. Invest.
33.Jeyakumar, M., et al. 2001. Enhanced survival in Sandhoff disease mice
receiving a combination of substrate deprivation therapy and bone mar-
row transplantation. Blood. 97:327–329.
34.Krivit, W., Peters, C., and Shapiro, E.G. 1999. Bone marrow transplanta-
tion as effective treatment of central nervous system disease in globoid
cell leukodystrophy, metachromatic leukodystrophy, adrenoleukodys-
trophy, mannosidosis, fucosidosis, aspartylglucosaminuria, Hurler,
Maroteaux-Lamy, and Sly syndromes, and Gaucher disease type III. Curr.
Opin. Neurol. 12:167–176.
35.Barton, N.W., et al. 1991. Replacement therapy for inherited enzyme
deficiency-macrophage-targeted glucocerebrosidase for Gaucher’s dis-
ease. N. Engl. J. Med. 23:1464–1470.
36.Tifft, C.J., and Proia, R.L. 2000. Stemming the tide: glycosphingolipid
synthesis inhibitors as therapy for storage diseases. Glycobiology.