Loss of Vac14, a regulator of the signaling lipid
phosphatidylinositol 3,5-bisphosphate, results
in neurodegeneration in mice
Yanling Zhang*†, Sergey N. Zolov*, Clement Y. Chow‡, Shalom G. Slutsky§, Simon C. Richardson¶, Robert C. Piper¶,
Baoli Yang?, Johnathan J. Nau†, Randal J. Westrick‡, Sean J. Morrison§, Miriam H. Meisler‡, and Lois S. Weisman*,**††
*Life Sciences Institute,‡Department of Human Genetics, **Department of Cellular and Developmental Biology, and§Howard Hughes Medical Institute and
Department of Internal Medicine and Center for Stem Cell Biology, University of Michigan, Ann Arbor, MI 48109; and Departments of†Biochemistry,
¶Physiology, and?Obstetrics and Gynecology, University of Iowa, Iowa City, IA 52242
Edited by Pietro V. De Camilli, Yale University School of Medicine, New Haven, CT, and approved September 17, 2007 (received for review March 12, 2007)
The signaling lipid, phosphatidylinositol 3,5-bisphosphate (PI(3,5)P2),
likely functions in multiple signaling pathways. Here, we report the
characterization of a mouse mutant lacking Vac14, a regulator of
PI(3,5)P2synthesis. The mutant mice exhibit massive neurodegenera-
tion, particularly in the midbrain and in peripheral sensory neurons.
spaces are present in areas where neurons should be present. Similar
vacuoles are found in cultured neurons and fibroblasts. Selective
membrane trafficking pathways, especially endosome-to-TGN retro-
on a mouse with a null mutation in Fig4, presents the unexpected
of neural cells.
Fab1 ? PIKfyve ? PtdInsI(3,5)P2? spongiform ? endosomal traffic
low abundance and the limited number of tools available for their
study, relatively little is known about these lipids.
An interesting property of PI(3,5)P2occurs in yeast, where a
stimulus of hyperosmotic shock induces dramatic and transient
changes in the levels of PI(3,5)P2. The levels of PI(3,5)P2
transiently rise ?20-fold (4). Within 1 minute, the levels rise
5-fold; by 5 minutes, they increase ?20-fold; there is a short
plateau of 10 min, and then PI(3,5)P2levels decrease at a rate
similar to their increase. The rapid decrease in PI(3,5)P2levels
occurs even though the cells remain in hyperosmotic media.
Vacuole volume undergoes transient changes that parallel
PI(3,5)P2levels. That these rapid and transient changes occur
even in the presence of a sustained stimulus strongly suggests
that PI(3,5)P2plays a major role in signaling pathways related to
Several proteins are required for the synthesis and turnover of
PI(3,5)P2. PI(3,5)P2 is synthesized from PI(3)P by the PI(3)P
5-kinase Fab1/PIKfyve/PIP5K3 (5, 6). Fab1 is stimulated by a
regulatory complex that contains Vac14 (7, 8) and Fig4 (4, 9).
Surprisingly, the Vac14/Fig4 complex plays two opposing roles in
the regulation of steady-state levels of PI(3,5)P2. Vac14/Fig4
both activate Fab1 and also function in the breakdown of
PI(3,5)P2 through the lipid phosphatase activity of Fig4
In mammals, generation of PI(3,5)P2is predicted to impact
PI(5)P production. In vitro studies have shown that PI(5)P can be
generated from PI(3,5)P2through the PI(3,5)P23-phosphatase
activity of members of the myotubularin family and related
proteins including MTM1, MTMR1, MTMR2, MTMR3,
MTMR6, and hJUMPY/MTMR14 (12–15). In addition, PIK-
fyve/Fab1 can generate both PI(3,5)P2and PI(5)P in vitro (16).
The source of PI(5)P in vivo has not been established. However,
the generation of PI(5)P from either pathway requires PIKfyve/
he low-abundance signaling lipids, phosphatidylinositol 3,5-
bisphosphate (PI(3,5)P2) and phosphatidylinositol 5-phosphate
Fab1 activity, either to produce the substrate for myotubularin
[PI(3,5)P2] or to produce PI(5)P directly from PI.
Haploinsufficiency of human Fab1/PIP5K3 is associated with
Francois–Neetens mouchete ´e corneal fleck dystrophy, a mild
syndrome characterized by vacuoles in the cornea (17). Because
one normal Fab1/PIP5K3 allele is present, affected individuals
likely retain the ability to synthesize PI(3,5)P2and PI(5)P. Thus,
the physiological effect of depletion of these lipids in mammals
was not known. To determine the physiological functions of
PI(3,5)P2in mammals, we characterized a mouse mutant defi-
cient in the Fab1 regulator, Vac14. We chose Vac14 rather than
Fab1 because, in yeast, vac14 but not fab1 mutants maintain a
low level of PI(3,5)P2, and vac14 mutants have a less severe
growth defect. The yeast studies led to the prediction that loss
of Vac14 in mammals might result in a milder defect than loss
The embryonic stem cell clone RRP155 obtained from the Uni-
versity of California Bay Genomics repository (San Francisco, CA)
contains a ?-geo gene-trap vector inserted into intron 1 of Vac14
[supporting information (SI) Fig. 6 a and b]. Chimeric mice were
crossed with C57BL/6J wild-type mice. Offspring with germ-line
transmission of the Vac14-?-geo allele were obtained. Heterozy-
gous mice appeared normal. Just after birth [postnatal day (P)0]
homozygous Vac14?-geo/?-geopups appeared normal externally and
were similar in size to their wild-type and Vac14?/?-geolittermates
(SI Fig. 6c). From midgestation through P0, we observed 39
homozygous mutants among 167 total animals, a value consistent
with the Mendelian prediction of 25%, (P ? 0.78 by the ?2test) (SI
Table 1). All Vac14?-geo/?-geomice died within 1–2 days after birth.
To test whether Vac14 protein was produced in the
Vac14?-geo/?-geomutants, extracts from mouse primary fibroblasts
were tested by Western blot analysis using polyclonal antibodies
raised to human full-length Vac14. The antibody recognizes both
Author contributions: Y.Z. and S.N.Z. contributed equally to this work; Y.Z., S.N.Z., S.G.S.,
C.Y.C., S.C.R., R.C.P., J.J.N., R.J.W., and M.H.M. contributed new reagents/analytic tools;
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Freely available online through the PNAS open access option.
Abbreviations: CI-MPR, cation-independent mannose-6-phosphate receptor; P(n), postna-
tal day (n); PI(3,5)P2, phosphatidylinositol 3,5-bisphosphate; PI(5)P, PI 5-phosphate;
Avenue, Room 6437, University of Michigan, Ann Arbor, MI 48109-2216. E-mail:
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2007 by The National Academy of Sciences of the USA
October 30, 2007 ?
vol. 104 ?
human and mouse Vac14. No Vac14 protein was detected in
the Vac14?-geo/?-geoextracts (SI Fig. 6d). In addition, the levels of
Fab1 are normal in both Vac14?-geo/?-geobrains and cultured
fibroblasts (SI Fig. 6 d and e). Thus, the phenotypes observed in
the Vac14?-geo/?-geomouse are likely because of the absence of
In wild-type mice, Vac14 and Fab1 proteins were expressed in
all tissues examined (SI Fig. 6e). This finding, along with the
observation that these conserved proteins serve housekeeping
functions in yeast, suggested that PI(3,5)P2and/or PI(5)P, are
required in all tissues.
Knockout of Vac14 in yeast is deleterious and results in
abnormally low levels of PI(3,5)P2 (4, 9). To test the role of
Vac14 in the maintenance of steady-state levels of phosphoino-
sitide polyphosphates in mammalian cells, we cultured fibro-
blasts from wild-type and Vac14?-geo/?-geoP0 animals and
measured the levels of these lipids. In yeast, PI(3,5)P2is a very
low-abundance lipid; ?0.08% of the total PI in the cell (4). In
wild-type mouse fibroblasts, the levels of PI(3,5)P2 are even
lower; 0.04 ? 0.003% of the total PI. PI(3,5)P2 levels were
significantly reduced in the Vac14?-geo/?-geomutant, 0.02 ?
0.005% (Fig. 1 a and b). Thus, in mammals, Vac14 has a key role
in the maintenance of steady-state levels of PI(3,5)P2. Consistent
with a defect in the synthesis of PI(3,5)P2, the level of PI(3)P was
elevated in Vac14?-geo/?-geofibroblasts; 0.5 ? 0.2% in the mutant
vs. 0.2 ? 0.02% in wild type. Based on studies in yeast (4, 7, 9),
we postulate that the elevated pool of PI(3)P represents sub-
strate that would normally be consumed by the PI(3)P 5-kinase
Fab1. Because of the absence of Vac14, Fab1 is not properly
activated, and thus its substrate accumulates. The levels of
PI(5)P were also significantly reduced; 0.3 ? 0.07% in wild type
vs. 0.1 ? 0.005% in the mutant (Fig. 1 a and c). The lower level
of PI(5)P may be due either to reduction in Fab1 kinase activity
or to reduction of PI(3,5)P2, the substrate for the 3-phosphatase
activity of myotubularins. Neither PI(3,4)P2nor PI(3,4,5)P3was
detected in the wild-type or mutant fibroblasts. The levels of
these lipids are generally exceedingly low and often cannot be
detected unless cells are stimulated, for example, with PDGF (1,
18). The above analysis of phosphatidylinositol polyphosphates
in primary fibroblasts strongly suggests that, in mammals, Vac14
is required to maintain normal levels of PI(3)P, PI(3,5)P2, and
To further characterize phenotypes associated with loss of
Vac14, sagittal sections of P0 pups were stained with H&E (Fig.
notable were lesions in several regions of the Vac14?-geo/?-geo
brain, including the preoptic area, thalamus, hypothalamus (SI
Table 2). Other regions including the hippocampus and the
cortex were less affected (SI Fig. 7a and SI Table 2). The lesions
appeared to be either neuronal cell bodies filled with a grossly
had formed and degenerated. Because the overall shape and size
of the brain was normal, it appears that Vac14 is not required for
early neural development.
In the peripheral nervous system, for example, the trigeminal
ganglion and dorsal root ganglion, the cell bodies from mutant
mice frequently contained vacuoles (Fig. 2b). The degree of
vacuolation ranged from multiple small cytoplasmic vacuoles to
a single, large vacuole that filled the entire cell body. Similar
defects were observed in the brains of Vac14?-geo/?-geoembryonic
day (E)18 embryos (data not shown). The nervous systems of
Percent total PI
a decrease in PI(3,5)P2and PI(5)P levels. (a) PI(3,5)P2and PI(5)P levels decrease,
24 h. Lipids were extracted, deacylated, and analyzed by anion-exchange
HPLC chromatography. To assess whether differences in the lipid levels were
statistically significant, we applied a two-tailed, paired Student’s t test. Sig-
PI(5)P decreased in Vac14?-geo/?-geocells (P ? 0.0070 and 0.0038, respectively).
PI(4)P and PI(4,5)P2levels did not change (P ? 0.36 and 0.87, respectively). For
all measurements, n ? 3. Error bars represent the standard deviations. (b and
c) Regions of HPLC profiles showing PI(3,5)P2(b) and PI(5)P (c) from Vac14?/?
(Left) and Vac14?-geo/?-geo(Right).
The Vac14?-geo/?-geomouse has no detectable Vac14 protein and has
sections of the whole brain of P0 mice. Examples of lesions (holes) in affected
regions, midbrain (Upper) and pons (Lower), are shown on the right with
higher magnification. (Scale bars, 100 ?m.) Insets are also indicated by the
squares in the whole brain. (b) Lesions composed of large intracellular vacu-
oles (arrows) or acellular areas were found in olfactory bulb, trigeminal
Lesions in neural tissue from Vac14?-geo/?-geomutant mice. (a) Sagittal
Zhang et al.
October 30, 2007 ?
vol. 104 ?
no. 44 ?
both Vac14?/?and heterozygous Vac14?/?-geoE18 embryos and
P0 pups appeared normal.
The holes observed in the brains of the Vac14?-geo/?-geomice
suggested that there might be a significant increase in cell death.
Therefore, we surveyed the levels of apoptotic cells in the brains
from three pairs of mutant P0 pups and corresponding wild-type
littermates. A series of 40-?m coronal sections were prepared
from the entire brain. Starting 80 ?m rostral to the crossing of
the corpus callosum every third section was immunostained by
using an antibody for activated caspase-3, a common marker for
apoptotic cells. All sections of Vac14?-geo/?-geomice were com-
parable in size with their wild-type littermates; however, two of
the three mutant brains had enlarged lateral ventricles. Two
confined regions within the forebrains of Vac14?-geo/?-geomice
had significantly increased levels of apoptotic cells (SI Fig. 8). A
region of the septum immediately caudal to the corpus callosum
showed a high number of activated caspase-3-positive cells; few
positive cells were observed in corresponding sections from the
same region of brains from littermate controls (SI Fig. 8a).
Increased apoptosis was also found in the cingulate cortex (SI
Fig. 8b). The highest increase in apoptosis occurred in the
midbrain and hindbrain, particularly in regions where we ob-
served the lesions in H&E-stained sections. Although both
mutant and wild-type brains had cells with activated caspase-3
scattered throughout these regions, the Vac14?-geo/?-geomice had
occurs in the beginning of the midbrain, 400 ?m caudal to the
beginning of the posterior commissure (SI Fig. 9).
Despite the fact that Vac14 and Fab1 are present in all tissues,
the only obvious defects in the Vac14?-geo/?-geoP0 pups were
observed in the nervous system. The spleen, liver, kidney, lung,
heart, and intestine appeared normal in size, cellularity, and
architecture (SI Fig. 7b and data not shown).
Although Vac14?-geo/?-geohippocampal and cortical neurons
appeared normal in vivo, these neurons formed vacuoles in
culture (Fig. 3a) (n ? 50). Vacuoles were not observed in
neurons cultured from wild-type mice. Neurite outgrowth
appeared normal in both the mutant and wild-type neurons.
Although large holes were not observed in other tissues in vivo,
it is likely that other cell types have the potential to form vacuoles.
Examination of fibroblasts cultured from P0 Vac14?-geo/?-geomice
revealed cells with multiple, large vacuoles. Although morpholog-
ically similar to lipid droplets in adipocytes, they are not labeled by
the lipophilic dye Nile red (SI Fig. 10). The vacuoles appear to be
due to the loss of Vac14 function. Wild-type and Vac14?-geo/?-geo
fibroblasts were electroporated with a cDNA encoding human
Vac14 fused to citrine [mCit (19, 20)], a variant of yellow fluores-
cent protein (YFP), or an mCit-only control. Before electropora-
tion, a high percentage (?80%) of Vac14?-geo/?-geofibroblasts were
vacuolated. Subplating confluent mutant fibroblasts with or with-
of cells with enlarged vacuoles (data not shown). Six hours after
transfection, the percentage of Vac14?-geo/?-geocells with vacuoles
was 2-fold lower in those with Vac14-mCit, than in those with mCit
alone (Fig. 3b). This difference increased with time. Twenty-four
were devoid of vacuoles, whereas the number of mCit-transfected
cells with vacuoles rose to ?80%. Note that in the presence of
sion of newly formed vacuoles occurred as well. These results
strongly suggest that vacuolation is due to the loss of Vac14.
Vac14 is an activator of Fab1 in yeast. The lowered levels of
PI(3,5)P2and PI(5)P in mouse fibroblasts indicate that Vac14 is
an activator of Fab1/PIKfyve in mice as well. To directly test
whether the vacuoles observed in Vac14?-geo/?-geomice are due
to diminished Fab1/PIKfyve activity or an as yet unknown
function of Vac14, we overexpressed human Fab1-mCit in
Vac14?-geo/?-geofibroblasts. Notably, the expression of Fab1-
mCit resulted in the disappearance of previously formed vacu-
oles and strongly suppressed the formation of new vacuoles (Fig.
3b). These results are consistent with the postulate that, as in
yeast, the major function of Vac14 in mammals is the regulation
of PI(3,5)P2and its related metabolites.
To determine whether these large vacuoles are predominantly
due to macropinocytosis or whether they arise from internal
membranes, we made time-lapse movies of individual fibro-
blasts. In both mutant and wild-type fibroblasts (SI Movies 1 and
2), vacuoles arose that were due to macropinocytosis. These
vacuoles appeared in regions of the plasma membrane coinci-
dent with ruffling filopodia. The newly formed vacuoles then
rapidly moved inward to regions adjacent to the nucleus and
rapidly disappeared (SI Movies 1 and 2). In the mutant, of 31
cells observed, in 13 examples, massive macropinocytosis oc-
curred; in each case the macropinosomes resolved in a wild-type
time frame (SI Movie 2). In all seven examples, where vacuoles
6 h24 h
Fluorescent cells with vacuoles, %
Vac14?-geo/?-geofibroblasts leads to the suppression of vacuole formation. (a)
E16.5 embryos and cultured for 2 weeks. Images of live cells were taken on a
Nikon TE-2000 inverted microscope with a ?40 phase-contrast objective.
(Scale bars, 20 ?m.) (b) Cells were transfected with the indicated plasmids.
Fluorescent cells were scored for the presence of vacuoles. Each data point is
a mean of four independent experiments. For each experiment ?100 cells
were counted. Error bars represent SD. (c) Representative cells for 6 and 24 h
after transfection are shown. Arrows indicate transfected cells.
www.pnas.org?cgi?doi?10.1073?pnas.0702275104Zhang et al.
arose and persisted, their origin was unclear (SI Movie 3). This
event was not observed in any of the 21 wild-type cells examined.
Vac14 and Fab1 colocalize with each other in yeast (9) and in
mammalian cells (21). Fab1/PIKfyve had been reported to
localize to late endosomes (22), whereas a recent study showed
colocalization with the early endosome (23). In the mammalian
studies, exogenously expressed tagged proteins were used. The
antiserum raised to Vac14 in this study was not suitable for
immunofluorescence. Although there is ambiguity concerning
the precise location of Vac14, based on its association with
PIKfyve/Fab1, it is likely that it localizes to endosomal
To test which organelles are affected by loss of Vac14 and to
labeled by GM130, exhibited a typical perinuclear tubular pattern
in both Vac14?/?and Vac14?-geo/?-geocells. Early endosomes,
labeled by EEA1, were slightly enlarged in Vac14?-geo/?-geofibro-
blasts, suggesting a possible defect in early endosomal organelles.
Note that EEA1 was not detected on the vacuole membranes,
suggesting that the vacuoles are not derived from early endosomes.
In contrast, the limiting membranes of the vacuoles were labeled
with Lamp2, a marker of late endosomes and lysosomes. These
results strongly suggest that vacuoles arise because of swelling of
either or both of these organelles. In addition to vacuoles, Lamp2-
positive dots were observed. These likely represent normal late
endosomes and lysosomes.
We monitored fluid-phase endocytosis to test whether the
vacuoles observed in Vac14?-geo/?-geofibroblasts remain a func-
tional part of the endocytic pathway. Cells were incubated with
Cascade blue dextran for a 16-hour pulse, followed by a 2-hour
chase, to allow the dextran to accumulate in lysosomes. The fluid
phase marker was taken up by Vac14?-geo/?-geofibroblasts; punc-
tate dots were found in both mutant and wild-type cells (Fig. 4b).
However, dextran was not detected in the vacuoles. This finding
strongly suggests that the vacuoles are not part of the normal
endocytic pathway. A similar defect was observed in cells
expressing dominant-negative Fab1/PIKfyve (24).
Receptor-mediated endocytosis is not affected by loss of Vac14.
After starvation and stimulation with EGF, total EGFR levels in
the cell were analyzed at several time points by Western blot
analysis. Three independent pairs of Vac14?/?and Vac14?-geo/?-geo
cells were compared (SI Fig. 11). No significant differences were
observed. This suggests that, in the absence of Vac14, EGFR is still
efficiently transported to lysosomes.
The steady-state localization of the cation-independent
mannose-6-phosphate receptor (CI-MPR) is perturbed in
network (TGN) and endosomes. In most cell types, CI-MPR is
found predominantly in the TGN, and thus is concentrated in a
perinuclear region of the cell. This same localization was observed
in Vac14?/?fibroblasts. In contrast, in Vac14?-geo/?-geofibroblasts
CI-MPR was found in small spots and enlarged vesicles that were
dispersed in the cytoplasm (Fig. 5a). Some of the vesicles were
EEA1 or Lamp2-positive, suggesting that, in Vac14?-geo/?-geofibro-
blasts, there is a defect in retrograde traffic of CI-MPR from late
and early endosomes back to the TGN. Consistent with this
hypothesis, we observed a defect in the trafficking of cathepsin D,
one of the enzymes transported by CI-MPR. Normally, newly
synthesized procathepsin D is transported by CI-MPR from the
TGN to endosomes, where the propeptide is cleaved, giving rise to
intermediate forms. These intermediate forms are transported to
the lysosome and further cleaved into two noncovalently linked
fragments. Western blot analysis of cathepsin D revealed that both
the pro- and intermediate forms accumulated in Vac14?-geo/?-geo
cells (Fig. 5b). The accumulation of procathepsin D is consistent
with a lack of CI-MRP in the TGN. A similar defect was observed
in siRNA-mediated knockdown of Fab1/PIKfyve (23).
Accumulation of the intermediate form of cathepsin D sug-
gests that there might also be a delay in the transition of late
endosomes to lysosomes. This latter defect and the fact that the
swollen Lamp2-positive vacuoles cannot receive fluid-phase
markers are consistent with the hypothesis that loss of Vac14
causes defects in some late endosomal functions.
Phosphatidylinositol polyphosphates regulate diverse functions,
and defects in their metabolism likely contribute to many human
diseases (25). Here, we analyzed a mouse model with undetect-
able Vac14 protein levels and find that loss of Vac14 leads to a
57% decrease in the levels of PI(3,5)P2, a 45% decrease in the
levels of PI(5)P, and a 2.4-fold increase in the levels of PI(3)P.
The mice die perinatally and exhibit profound degeneration in
certain regions of the central and peripheral nervous systems.
Selected regions in the brain are affected, especially the medulla,
the pons, and the midbrain. Increased cell death occurs in these
areas. Affected neurons contain large vacuoles, and vacuoles
also arise in both neurons and fibroblasts cultured from the
The finding that the enlarged vacuoles in cells from the
Vac14?-geo/?-geomice contain markers of the late endosome is
late endosomal markers occur because of overexpression of
dominant-negative Fab1/PIKfvye (26) or depletion of Fab1/
PIKfyve by siRNA (23). They are observed in mice with a null
mutation of the PI(3,5)P2 5-phosphatase Fig4 (33), and in
Caenorhabditis elegans (27) and Drosophilia (28) with mutations
fibroblasts were labeled with GM130 (Golgi marker), EEA1 (early endosome
marker), or Lamp2 (late endosome and lysosome marker). Although the
cis-Golgi appears normal in Vac14?-geo/?-geocells, early endosomes are fre-
quently enlarged. The limiting membrane of the vacuoles in Vac14?-geo/?-geo
cells is labeled with Lamp2. (b) Vacuoles in Vac14?-geo/?-geocells are incapable
of receiving fluid phase marker dextran. Vac14?/?and Vac14?-geo/?-geofibro-
blasts were incubated with dextran Cascade blue for 16 h and chased for 2 h.
(Scale bars, 10 ?m.)
Zhang et al.
October 30, 2007 ?
vol. 104 ?
no. 44 ?
in Fab1/PIKfyve. Moreover, large vacuoles are observed in
fab1? and vac14? yeast mutants (7, 11).
That these large vacuoles have late endosomal markers and
that PI(3,5)P2is required for the retrograde transport from late
endosomes to the TGN (ref. 23 and this study) and also for
sorting a subset of cargoes into the lumen of multivesicular
bodies (8, 29–31), strongly suggests that the large vacuoles arise
from late endosomes and occur because of defects in membrane
traffic from the late endosome.
The enlarged vacuoles are likely because of lowered levels of
PI(3,5)P2rather than elevation of PI(3)P. siRNA knockdown of
Vps34, the PI(3)-kinase that produces the PI(3)P substrate used by
in large vacuoles that contain late endosomal markers (32).
It is also unlikely that lowered levels of PI(5)P cause the
enlarged vacuoles. Mutations in myotubularins, which are pre-
dicted to lower PI(5)P levels, have not been found to produce
The spontaneous mouse mutant, pale tremor, has a loss of
function mutation in the Fig4 gene (33). We recently showed
of PI(3,5)P2to PI3P, Fig4 has a second opposing function and is
required for the production of PI(3,5)P2. In yeast, Vac14 and
Fig4 act in the same pathway, form a complex with each other,
and each knockout strain, vac14? and Fig4?, has a partial loss
of PI(3,5)P2(4, 7–9, 11). Our recent studies suggest that Fig4
serves the same dual role in mammals. Cells cultured from the
Fig4 mutant mouse have significantly reduced PI(3,5)P2(33).
Importantly, the Fig4 and Vac14 mouse mutants have similar
defects in levels of PI(3,5)P2, and acquire profound vacuolation
in the same types of neurons. The similar pattern of neurode-
generation strongly supports the hypothesis that the neurode-
generation observed in both mice is a direct consequence of
defects in the metabolism of PI(3,5)P2and/or PI(5)P. Moreover,
human mutations of Fig4 result in the peripheral nerve disorder
Charcot–Marie–Tooth syndrome (CMT4J) (1). This finding
strongly suggests that mutations in Vac14 and Fab1/PIKfyve
could result in Charcot–Marie–Tooth synddrome as well as
other neurological disorders.
Mutations in myotubularin-related genes, which are predicted
to have reduced levels of PI(5)P, also cause neuropathic diseases
in mice and humans (34–36), but these defects are distinct from
the defects observed in the Vac14 or Fig4 mouse mutants. The
neurological defects are not as severe, vacuolation does not
occur, and the myotubularin mutations are not lethal.
Mutations in proteins that regulate other phosphoinositides
have been shown to lead to neurological defects (37–40). None of
these genes is known to affect levels of PI(3,5)P2, and the pheno-
types are distinct from those observed in the Vac14 or Fig4 mice.
Although Vac14 has a wide tissue distribution, neurons are
particularly sensitive to reduction of PI(3,5)P2. This may be due to
the presence of extraordinarily long processes in the affected
unidentified neuronal-specific endosomally derived organelle. In-
deed, during subcellular fractionation of rat brain, Vac14 was
enriched in microsomal membranes from synapses (41).
The finding that loss of PI(3,5)P2leads to neurodegeneration
may be directly linked to recent discoveries that some forms of
Alzheimer’s disease correlate with defects in SORL1 (42) and
Vps35 (43), which function in the same pathway. Vps35 is part
of the retromer complex and in humans is required to form
tubular carriers from the endosomes that travel by retrograde
traffic to the trans-Golgi network (reviewed in ref. 44). Notably,
fibroblasts from the Vac14?-geo/?-geomice show a defect in this
pathway. Thus, loss of PI(3,5)P2in neurons may be particularly
deleterious because of subsequent defects in membrane traffic
from endosomes to the TGN.
The discovery that PI(3,5)P2is critical to neuronal health may
lead to a molecular understanding of some forms of neurode-
generation and will hopefully lead to new therapeutic ap-
proaches to treat these serious diseases.
Materials and Methods
HPLC. For separation of glycerophosphoinositides, two elution
gradients were used at 1 ml/min flow rate. (pump A: H2O; pump
B: 1M (NH4)2HPO4, pH3.8). Gradient 1: 0% B 5 min; 0–2% B
15 min; 2% B 80 min; 2–12% B 20 min; 12% B 20 min; 12–80%
GroPIns(3,4)P2from GroPIns(3,5)P2): 0% B 5 min; 0–2% B 15
min; 2% B 80 min; 2–10% B 20 min; 10% B 65 min; 10–80% B
40 min; 80% B 20 min; 80–0% B 5 min. Positions of
GroPIns(3,4,5)P3 determined by
received as gifts from Lucia Rameh (Biomedical Research
Institute, Boston, MA). Positions of GroPIns(4)P and
GroPIns(4,5)P2 were confirmed with yeast glycerophosphoi-
nositide extracts (4). Position of GroPIns(5)P was based on its
migration relative to GroPIns(4)P (13, 45). The elution position
of GroPIns(5)P is close to that of GroPIns(4)P; the shoulder of
GroPIns(4)P overlapped GroPIns(5)P. For GroPIns(5)P, we
32P-labeled standards were
blasts. (a) CI-MPR is trapped in a mixed population of endosomes. Vac14?/?
and Vac14?-geo/?-geofibroblasts were fixed as above and double labeled with
CI-MPR (green) and EEA1 or Lamp2 (red). Images were taken on a Zeiss
confocal microscope. Peripheral regions of cells marked by squares are shown
in an amplified view. (b) Pro- and intermediate cathepsin D accumulate in
Vac14?-geo/?-geofibroblasts. Steady-state levels of cathepsin D in whole-cell
lysates from Vac14?/?and Vac14?-geo/?-geofibroblasts were analyzed by West-
Endosome-to-TGN trafficking is defective in Vac14?-geo/?-geofibro-
www.pnas.org?cgi?doi?10.1073?pnas.0702275104Zhang et al.
quantified the right half of the GroPIns(5)P peak and estimated
seen, we integrated the values from the same elution position as
the latter half of GroPIns(5)P peak and multiplied by a factor of 2.
All experiments were performed in compliance with the guide-
lines of the University Committee on Use and Care of Animals of
the University of Michigan. Animals were housed and cared for in
accordance with National Institutes of Health guidelines.
Generation of Mice, Antibodies, Inositol Labeling, Western Blot,
Primary Fibroblast and Neuron Cultures, Histology, Immunofluores-
cence, Immunohistochemistry, Dextran Uptake, Nile Red Staining,
Electroporation, EGFR Degradation, and Live-Cell Imaging. See SI
Materials and Methods for detailed experimental procedures for
generation of the mice, antibodies, inositol labeling, Western blot,
primary fibroblast and neuron cultures, histology, immunofluores-
cence, immunohistochemistry, dextran uptake, Nile red staining,
electroporation, EGFR degradation, and live-cell imaging.
We thank Drs. James Dowling, David Ginsburg, Jason Gestwicki, and
John Stokes for helpful discussions; Drs. Kristen Verhey and Jennetta
for help with the EGFR degradation assay. The mAb for LAMP-2 was
developed by Dr. Thomas August and provided by the Developmental
Studies Hybridoma Bank (developed under the auspices of the National
Institute of Child Health and Human Development and maintained by
the Department of Biological Sciences, University of Iowa). This work
was supported by National Institutes of Health Grants R01 GM50403
(to L.S.W.) and R01 GM24872 (to M.H.M.). Generation of the
Vac14?-geo/?-geomouse was supported in part by Pilot Grant P50
C.Y.C. was supported by Genetics Predoctoral Training Grant NIH-T32
GM07544. S.C.R. was supported by Engineering and Physical Sciences
Research Council (EP/C013220/1) and the American Heart Association
(0120475Z and 0325605Z).
1. Whiteford CC, Brearley CA, Ulug ET (1997) Biochem J 323(Pt 3):597–601.
2. Dove SK, Cooke FT, Douglas MR, Sayers LG, Parker PJ, Michell RH (1997)
3. Rameh LE, Tolias KF, Duckworth BC, Cantley LC (1997) Nature 390:192–196.
4. Duex JE, Tang F, Weisman LS (2006) J Cell Biol 172:693–704.
5. Gary JD, Wurmser AE, Bonangelino CJ, Weisman LS, Emr SD (1998) J Cell
6. McEwen RK, Dove SK, Cooke FT, Painter GF, Holmes AB, Shisheva A, Ohya
Y, Parker PJ, Michell RH (1999) J Biol Chem 274:33905–33912.
7. Bonangelino CJ, Nau JJ, Duex JE, Brinkman M, Wurmser AE, Gary JD, Emr
SD, Weisman LS (2002) J Cell Biol 156:1015–1028.
8. Dove SK, McEwen RK, Mayes A, Hughes DC, Beggs JD, Michell RH (2002)
Curr Biol 12:885–893.
9. Duex JE, Nau JJ, Kauffman EJ, Weisman LS (2006) Eukaryot Cell 5:723–731.
10. Rudge SA, Anderson DM, Emr SD (2004) Mol Biol Cell 15:24–36.
11. Gary JD, Sato TK, Stefan CJ, Bonangelino CJ, Weisman LS, Emr SD (2002)
Mol Biol Cell 13:1238–1251.
12. Walker DM, Urbe S, Dove SK, Tenza D, Raposo G, Clague MJ (2001) Curr
13. Schaletzky J, Dove SK, Short B, Lorenzo O, Clague MJ, Barr FA (2003) Curr
14. Tronchere H, Laporte J, Pendaries C, Chaussade C, Liaubet L, Pirola L,
Mandel JL, Payrastre B (2004) J Biol Chem 279:7304–7312.
15. Tosch V, Rohde HM, Tronchere H, Zanoteli E, Monroy N, Kretz C, Dondaine
N, Payrastre B, Mandel JL, Laporte J (2006) Hum Mol Genet 15:3098–3106.
16. Sbrissa D, Ikonomov OC, Deeb R, Shisheva A (2002) J Biol Chem 277:47276–
17. Li S, Tiab L, Jiao X, Munier FL, Zografos L, Frueh BE, Sergeev Y, Smith J,
Rubin B, Meallet MA, et al. (2005) Am J Hum Genet 77:54–63.
18. Hawkins PT, Jackson TR, Stephens LR (1992) Nature 358:157–159.
19. Griesbeck O, Baird GS, Campbell RE, Zacharias DA, Tsien RY (2001) J Biol
20. Zhang J, Campbell RE, Ting AY, Tsien RY (2002) Nat Rev Mol Cell Biol
21. Sbrissa D, Ikonomov OC, Strakova J, Dondapati R, Mlak K, Deeb R, Silver R,
Shisheva A (2004) Mol Cell Biol 24:10437–10447.
22. Ikonomov OC, Sbrissa D, Shisheva A (2001) J Biol Chem 276:26141–26147.
23. Rutherford AC, Traer C, Wassmer T, Pattni K, Bujny MV, Carlton JG,
Stenmark H, Cullen PJ (2006) J Cell Sci 119:3944–3957.
24. Ikonomov OC, Sbrissa D, Foti M, Carpentier J-L, Shisheva A (2003) Mol Biol
25. Pendaries C, Tronchere H, Plantavid M, Payrastre B (2003) FEBS Lett
26. Ikonomov OC, Sbrissa D, Mlak K, Kanzaki M, Pessin J, Shisheva A (2002)
J Biol Chem 277:9206–9211.
27. Nicot AS, Fares H, Payrastre B, Chisholm AD, Labouesse M, Laporte J (2006)
Mol Biol Cell 17:3062–3074.
28. Rusten TE, Rodahl LM, Pattni K, Englund C, Samakovlis C, Dove S, Brech A,
Stenmark H (2006) Mol Biol Cell 17:3989–4001.
29. Odorizzi G, Babst M, Emr SD (1998) Cell 95:847–858.
30. Reggiori F, Pelham HRB (2002) Nat Cell Biol 4:117–123.
31. Shaw JD, Hama H, Sohrabi F, DeWald DB, Wendland B (2003) Traffic
32. Johnson EE, Overmeyer JH, Gunning WT, Maltese WA (2006) J Cell Sci
33. Chow CY, Zhang Y, Dowling JJ, Jin N, Adamska M, Shiga K, Szigeti K, Shy
ME, Li J, Zhang X, et al. (2007) Nature 448:68–72.
34. Bolino A, Muglia M, Conforti FL, LeGuern E, Salih MA, Georgiou DM,
Christodoulou K, Hausmanowa-Petrusewicz I, Mandich P, Schenone A, et al.
(2000) Nat Genet 25:17–19.
U, Mruk DD, Feltri ML, et al. (2004) J Cell Biol 167:711–721.
36. Bonneick S, Boentert M, Berger P, Atanasoski S, Mantei N, Wessig C, Toyka
KV, Young P, Suter U (2005) Hum Mol Genet 14:3685–3695.
37. Cremona O, Di Paolo G, Wenk MR, Luthi A, Kim WT, Takei K, Daniell L,
Nemoto Y, Shears SB, Flavell RA, et al. (1999) Cell 99:179–188.
38. Di Paolo G, Moskowitz HS, Gipson K, Wenk MR, Voronov S, Obayashi M,
Flavell R, Fitzsimonds RM, Ryan TA, De Camilli P (2004) Nature 431:415–422.
39. Lowe M (2005) Traffic 6:711–719.
40. Nystuen A, Legare ME, Shultz LD, Frankel WN (2001) Neuron 32:203–212.
41. Lemaire JF, McPherson PS (2006) FEBS Lett 580:6948–6954.
42. Rogaeva E, Meng Y, Lee JH, Gu Y, Kawarai T, Zou F, Katayama T, Baldwin
CT, Cheng R, Hasegawa H, et al. (2007) Nat Genet 39:168–177.
43. Small SA, Kent K, Pierce A, Leung C, Kang MS, Okada H, Honig L, Vonsattel
JP, Kim TW (2005) Ann Neurol 58:909–919.
44. Bonifacino JS, Rojas R (2006) Nat Rev Mol Cell Biol 7:568–579.
45. Tolias KF, Rameh LE, Ishihara H, Shibasaki Y, Chen J, Prestwich GD, Cantley
LC, Carpenter CL (1998) J Biol Chem 273:18040–18046.
Zhang et al.
October 30, 2007 ?
vol. 104 ?
no. 44 ?
100 µ µm
20 µ µm
EGF 0103060 90120010306090 120 (min)