Unique Biochemical and Behavioral Alterations in
Drosophila shibirets1Mutants Imply a
Conformational State Affecting Dynamin Subcellular
Distribution and Synaptic Vesicle Cycling
Mai-Lei Chen,1David Green,1Lei Liu,2Yung Carmen Lam,1* Leona Mukai,3
Sujata Rao,4Shobha Ramagiri,4K. S. Krishnan,4Jeff E. Engel,5Jim J.-C. Lin,1
1Department of Biological Sciences, University of Iowa, Iowa City, Iowa 52242
2Interdepartmental Neuroscience Program, University of Iowa, Iowa City, Iowa 52242
3Department of Molecular and Cellular Biology, University of Arizona, Tucson, Arizona 85721
4Department of Biological Sciences, Tata Institute for Fundamental Research, Colaba, Bombay
5Department of Biological Sciences, Western Illinois University, Macomb, Illinois 61455
Received 16 October 2001; accepted 4 June 2002
ABSTRACT Dynamin is a GTPase protein that is
essential for clathrin-mediated endocytosis of synaptic
vesicle membranes. The Drosophila dynamin mutation
shits1changes a single residue (G273D) at the boundary
of the GTPase domain. In cell fractionation of homoge-
nized fly heads without monovalent cations, all dynamin
was in pellet fractions and was minimally susceptible to
Triton-X extraction. Addition of Na?or K?can extract
dynamin to the cytosolic (supernatant) fraction. The
shits1mutation reduced the sensitivity of dynamin to salt
extraction compared with other temperature-sensitive
alleles or wild type. Sensitivity to salt extraction in shits1
was enhanced by GTP and nonhydrolyzable GTP-?S.
The shits1mutation may therefore induce a conforma-
tional change, involving the GTP binding site, that af-
fects dynamin aggregation.
Temperature-sensitive shibire mutations are known to
arrest endocytosis at restrictive temperatures, with con-
comitant accumulation of presynaptic collared pits.
Consistent with an effect upon dynamin aggregation,
intact shits1flies recovered much more slowly from heat-
induced paralysis than did other temperature-sensitive
shibire mutants. Moreover, a genetic mutation that low-
ers GTP abundance (awdmsf15), which reduces the par-
alytic temperature threshold of other temperature-sen-
sitive shibire mutations that lie closer to consensus
GTPase motifs, did not reduce the paralytic threshold of
shits1. Taken together, the results may link the GTPase
domain to conformational shifts that influence aggrega-
tion in vitro and endocytosis in vivo, and provide an
unexpected point of entry to link the biophysical prop-
erties of dynamin to physiological processes at synapses.
© 2002 Wiley Periodicals, Inc. J Neurobiol 53: 319–329, 2002
Keywords: dynamin; shibire; temperature-sensitive mu-
tation; endocytosis; synaptic vesicle; GTPase; Drosoph-
Research was conducted in the laboratories of C.-F. Wu, K. S.
Krishnan, M. Ramaswami (Tucson AZ), and J.-C. Lin.
*Present address: Division of Neuroscience, Baylor College of
Medicine, Houston, TX 77030.
Correspondence to: C.-F. Wu (firstname.lastname@example.org.).
Contract grant sponsor: NIH; contract grant number: HD18577
(C.F.W. and J.L.).
Contract grant sponsor: Department of Science and Technol-
ogy, Government of India (K.S.K.).
Contract grant sponsor: NIH; contract grant number: NS34889
© 2002 Wiley Periodicals, Inc.
Published online in Wiley InterScience (www.interscience.wiley.com).
Dynamin is an essential component of clathrin-medi-
ated endocytosis at synaptic terminals and elsewhere
(reviewed by Urrutia et al., 1997; McNiven, 1998;
Schmid et al., 1998; van der Bliek, 1999). Dynamin is
a GTPase, and GTP binding affects its assembly and
function. Studies indicate conformational interactions
between the N-terminal GTPase domain and other
regions of dynamin (Damke et al., 1995). These in-
teractions are influenced by charge effects such as
phosphorylation and salt concentration (Hinshaw and
Schmid, 1995; Warnock et al., 1996; Carr and Hin-
shaw, 1997). Identified mutations may be used to
study such conformational effects and determine their
functional significance in vivo.
The gene shibire (shi) has been shown to encode
multiple dynamin isoforms in Drosophila melano-
gaster (Chen et al., 1991; van der Bliek and Meyer-
owitz, 1991). Temperature-sensitive mutations of
shibire first demonstrated dynamin’s role in endocy-
tosis (Koenig et al., 1983; Kosaka and Ikeda,
1983a,b). Mutant flies show normal behavior and
viability at permissive temperatures, but higher tem-
peratures (above 26–28°C depending on the mutant
allele) induce paralysis (Grigliatti et al., 1973; Siddiqi
and Benzer, 1976) with ultrastructural and physiolog-
ical evidence of defective vesicle cycling (Ozawa and
Hagiwara, 1976; Koenig et al., 1983, 1998; Kosaka
and Ikeda, 1983a,b; Masur et al., 1990; Kawasaki et
al., 2000). In dissociated neuron cultures, the mutation
shits1causes growth cone collapse and arrests new
outgrowth at 30°C but not at room temperature (Kim
and Wu, 1987). Three temperature-sensitive muta-
tions (shits1, shits2, shits4) all occur within the consen-
sus GTPase domain (van der Bliek and Meyerowitz,
1991; Grant et al., 1998) and show similar pheno-
types, and they have been presumed to affect dynamin
in the same way. However, we serendipitously dis-
covered that under certain biochemical conditions
shits1, but not other mutant alleles, alters the distribu-
tion of dynamin among subcellular fractions in vitro.
shits1also differs from other alleles in vivo in the time
it takes to recover from heat-induced paralysis and in
genetic interactions with a nucleotide diphosphate
(NDP) kinase mutation that reduces GTP synthesis.
In vitro, the shits1mutation modifies the influence
of salt upon dynamin aggregation. Fly head homoge-
nate was fractionated by centrifugation at low speed
followed by high speed to yield two pellets and a
supernatant containing membrane-associated and cy-
toplasmic cell fractions, respectively (Gass et al.,
1995). Dynamin was previously shown to associate
with liposomes in pellet fractions in a charge-sensitive
manner, appearing in the supernatant only when
monovalent cations (Na?or K?) were present (Gass
et al., 1995). We discovered by accident that this salt
requirement is more stringent for shits1than shits2,
shits4, or wild-type dynamins. When tissue grinding
saline contained high Na?(128 mM), less dynamin
shifted to the supernatant in shits1than in the other
genotypes. GTP and nonhydrolyzable GTP-?S mark-
edly enhanced dynamin in the supernatant under these
conditions, but had no effect when monovalent cat-
ions were completely absent. When high K?saline
was used to better approximate intracellular condi-
tions, the distribution of dynamin to supernatant in
shits1increased to resemble other alleles more closely.
In vivo, shits1flies are viable (Grigliatti et al., 1973)
with functional dynamin under normal conditions, but
have substantially longer recovery times than shits2or
shits4after heat-induced paralysis. Furthermore, mu-
tations of awd, an NDP-kinase, substantially reduce
paralytic temperature thresholds in shits2and shits4
(Krishnan et al., 2001) but not in shits1, implying that
shits1dynamin depends less upon GTP abundance for
assembly and membrane association.
These results indicate that an identified residue in
the GTPase domain of dynamin alters both its assem-
bly in vitro and its function in vivo. The shits1muta-
tion replaces neutral glycine with negative aspartate
(van der Bliek and Meyerowitz, 1991) and may in-
crease the flexibility of the linkage between the GTP-
ase domain and regulatory domains closer to the C
terminal (Damke et al., 1984). The interacting effects
of salt concentration and GTP analogs in vitro suggest
altered charge dependence of conformational states.
At the same time, prolonged recovery from paralysis
in vivo suggests that this mutation may trap dynamin
in a metastable state at one point in the endocytic
cycle. The shits1mutation may therefore prove useful
in dissecting the conformational transitions of dy-
namin under physiological conditions and provide a
new entre ´e into studying the endocytic cycle in neu-
MATERIALS AND METHODS
Fly heads were collected, homogenized, and centrifuged as
described previously (Gass et al., 1995). Flies were frozen
by submerging vials in liquid nitrogen and vortexed for 20 s
to separate heads from bodies. Twenty-five heads were
ground in a glass-glass homogenizer (Wheaton, Millville,
NJ) for 50 strokes in 30 ?L of grinding solution. The
homogenizer was then rinsed with an additional 20 ?L of
Chen et al.
grinding solution. The compositions of grinding solutions
(Table 1) were as follows (in mM): Jan and Jan (1976): 128
NaCl, 2 KCl, 4 MgCl2, 0 CaCl2, 35 mM sucrose, 5 HEPES,
pH 7.1; intracellular: 144 KCl, 1 MgCl2, 0.5 CaCl2, 5
EGTA, 10 HEPES, pH 7.1 (Zhao et al., 1995); HCB0
(HEPES Column Buffer with 0 mM NaCl): 2 MgCl2, 1
EGTA, 20 HEPES, pH 7.1 (Muhlberg et al., 1997);
HCB128 and HCB150: same as HCB0 with addition of 128
or 150 NaCl. All grinding solutions also contained protease
inhibitor (1 tablet/50 mL saline; Roche Molecular Bio-
chemicals, Indianapolis, IN, cat. #1697498).
All steps after the liquid nitrogen freezing of flies were
performed at 4°C, except when figure legends designate
“room temperature,” in which case the steps before centrif-
ugation were performed at room temperature. The homog-
enate was centrifuged at 2000?g for 20 min to yield a pellet
(P1) and a supernatant fraction (S1). S1 was transferred to
an airfuge tube and ultracentrifuged (airfuge; Beckman,
Palo Alto, CA) at 130,000?g for 30 min to yield a pellet
(P2) and a cytosolic fraction (S2). P1 and P2 were resus-
pended in 40 ?L of grinding solution. Each fraction was
added to 40 ?L 2X SDS gel sample buffer. Samples were
held at 95–100°C for 3 min and then stored below ?20°C.
For Triton-X experiments this protocol was modified as
follows: 50 fly heads were homogenized in 30 ?L of grind-
ing solution (HCB0 or HCB128) followed by rinsing the
homogenizer with 20 ?L of solution. The homogenate was
incubated for 15 min at 4°C. Then 50 ?L of grinding
solution containing 2% Triton-X was added and the mixture
was incubated for another 15 min before centrifugation.
Ten microliters of P1, P2, and S2 fractions were loaded and
separated by SDS-PAGE (7.5% polyacrylamide) and then
transferred to nitrocellulose (Gass et al., 1995). Antibodies
used were antidynamin-I (BD Transduction Laboratories,
Lexington, KY), anti-?-tubulin (Sigma Chemical Co., St.
Louis, MO), and anti-CSP (gift from E. Buchner, Wu ¨rz-
burg). Antibody probe staining was developed with the
Enzyme Catalyzed Luminescence (ECL) Western Detection
System (Amersham, Buckinghamshire, UK). All experi-
ments shown were carried out at least three times.
High-temperature paralysis thresholds and recovery times
were examined as described previously (Grant et al., 1998).
Temperature-sensitive mutant shibire fly stocks shits1,
shits2, and shits4were obtained from collections of S. Benzer
(California Institute of Technology, Pasadena, CA) and M.
Ramaswami (U. Arizona, Tucson, AZ) (Siddiqi and Benzer,
1976; Grant et al., 1998). Each allele bears a missense
mutation causing a single amino acid substitution as fol-
lows: shits1, G268D (van der Bliek and Meyerowitz, 1991);
shits2, G141S (van der Bliek and Meyerowitz, 1991); shits4,
P171S (Grant et al., 1998). Homologous residues in mam-
malian dynamin are G273, G146, and P176, respectively
(van der Bliek and Meyerowitz, 1991; Damke et al., 1995).
The shits2and shits4substitutions lie within the N-terminal
GTPase domain, while shits1is at the boundary of this
domain (van der Bliek and Meyerowitz, 1991; Damke et al.,
1995). Lethal shibire alleles shiEM42and shiEM56were also
from M. Ramaswami (Grant et al., 1998). shiST139, an
independently isolated allele (Siddiqi and Benzer, 1976)
with the same amino acid substitution as shits1(data not
shown), was used in some experiments to corroborate shits1
results (data not shown). The abnormal wing discs (awd)
mutation awdmsf15has been described (Krishnan et al.,
2001). Oregon-R (OR) was used as a wild-type control.
Allele-Specific Dynamin Distribution
among Subcellular Fractions
We found distinct biochemical properties of shits1
dynamin in a continuation of previous cell fraction-
Table 1 Grinding Solutions
Jan and Jan*
Grinding solution: compositions (mM), osmolarities (mOsm), and pH. All solutions also included protease-inhibitor (see Materials and
* Jan and Jan (1976).
† Zhao et al. (1995).
‡ Muhlberg et al. (1997).
§Also included 35 mM sucrose.
shits1Alters Properties of Dynamin
ation studies using antidynamin probes (Gass et al.,
1995). Cell fractionation of fly heads yields one pellet
of plasma membrane and other dense components
(P1), a second pellet of vesicular membrane and other
light components (P2), and a supernatant containing
cytosolic and dissolved components (S2) (Gass et al.,
1995). In all genotypes and all conditions, a substan-
tial portion of dynamin remained in the membrane-
associated fractions P1 and P2 (Table 2). However,
the presence of dynamin in the cytosolic S2 fraction
depended upon the combination of experimental con-
ditions and the shibire genotype of the sampled flies.
Homogenization of fly heads in an extracellular sa-
saline, Table 1) led to the discovery that these conditions
cause a marked reduction of immunoreactivity in the S2
(cytosolic) fraction in shits1, compared to other shi mu-
tants and wild-type [Fig. 1(C)]. A similar proportion of
?-tubulin distributed to S2 in all genotypes [Fig. 1(A)],
suggesting that the differential distribution in shits1is
specific to dynamin. Because all three shi mutations are
temperature-sensitive in vivo, in one experiment the liv-
ing flies were held at restrictive temperature (38°C) for
10 min before freezing and grinding at 4°C [Fig. 1(A)];
in another experiment, heads were ground and held for
30 min at room temperature instead of 4°C before cen-
trifugation [Fig. 1(B)]. Neither of these treatments af-
fected the distribution of dynamin in shits1or the other
shits1may have a negative dominant effect in vivo,
as indicated by examinations of temperature-sensitive
paralysis (Kim and Wu, 1990). However, in heterozy-
gotes of shits1with shi?or shits2, dynamin immuno-
Table 2 Dynamin Extraction to S2 Fraction with Different Biochemical Treatments
Jan and Jan’s
Jan and Jan’s
Jan and Jan’s
Jan and Jan’s
Jan and Jan’s
— 2(B), 5(A1)
Presence of S2 immunoreactivity is indicated by a qualitative scale: —, no visible S2 band; ?, S2 lighter than P1; ??, S2 equal to P1;
???, S2 darker than P1. Symbols in parentheses: data not shown in figures.
* This row also includes HCB150 data [Fig. 2(C)] and HCB150 data [Fig. 2(C)].
in shits1. (A) Flies heat-shocked at 38°C for 10 min before
freezing and sample preparation at 4°C. Tubulin immuno-
reactivity was monitored as a protein loading control in
most preparations (shown here only). (B) Samples pro-
cessed at room temperature until centrifugation. (C) Sam-
ples prepared at 4°C without prior heat shock. (D,E) Cyto-
solic dynamin distribution was present in heterozygotes
combining shits1with wild-type, shits2, or lethal shiEM42or
shiEM56alleles. Grinding solution in all experiments was
Jan and Jan saline with 128 mM NaCl (see Table 1). P1 is
the plasma membrane fraction, P2 is the vesicular mem-
brane fraction, and S2 is the cytosolic fraction.
Cytosolic dynamin (S2) was markedly reduced
Chen et al.
reactivity was clearly present in S2 [Fig. 1(D)]. This
indicates that under these conditions at least some
portions of wild-type or shits2dynamin distribute to
supernatant despite the presence of shits1dynamin.
Cytosolic S2 dynamin was also present when shits1
was heterozygous with either of two lethal alleles,
shiEM42or shiEM56[Fig. 1(E)]. These lethal mutations
are at positions 381 and 401, respectively, outside of
the GTPase domain, and heterozygotes of these al-
leles with shits1show only slightly lower paralysis
temperatures than shits1/shi?(Grant et al., 1998). The
same amino acid substitution that is present in shits1is
also found in shiST139, an independently isolated al-
lele, and shiST139mutant flies are phenotypically in-
distinguishable from shits1(Kim and Wu, 1990). Un-
der various salt conditions, dynamin from shiST139
homogenate distributed identically to shits1(data not
Influence of Cations and GTP Analogs
Because Jan and Jan extracellular saline has high Na?
and low K?, we examined the effects of grinding in
an intracellular saline with low Na?and high K?,
commonly used for electrophysiological patch clamp
recording to minimize cell rundown. This intracellular
saline (144 mM K?) induced the presence (albeit still
reduced) of dynamin in S2 in shits1mutants [Fig.
2(A)]. In contrast, when all monovalent cations were
eliminated by grinding in HCB0, dynamin was lost
from S2 in all shi mutants as well as in wild type [Fig.
2(B)]. Addition of 150 mM Na?to HCB (HCB150)
induced S2 dynamin in all genotypes except shits1
[Fig. 2(C)] (compare with Fig. 1). This indicated that
monovalent cations in the grinding solution facilitate
the dissociation of dynamin, and further suggested
that shits1is more resistant to monovalent salt extrac-
tion when Na?is substituted for K?. Indeed, in pre-
liminary experiments we found that for wild type
(OR) and shits2, dynamin extraction is apparent with
as little as 32 mM Na?(data not shown). In shits1,
dynamin extraction approaches wild-type levels with
300 mM Na?(data not shown). Finally, shits1is more
resistant to salt extraction than wild type at 128 mM
K?(data not shown). These manipulations did not
affect distribution of a synaptic vesicle protein, cys-
teine string protein (CSP), in shits1or other genotypes
Dynamin binds and hydrolyzes GTP during the
endocytic cycle in vivo. An abundance of GTP in the
grinding solution induced a shift of dynamin to S2
(cytosolic fraction) in shits1[Fig. 3(C)]. Surprisingly,
GTP-?S (a nonhydrolyzable GTP analog) was effec-
tive at 10-fold lower concentrations than GTP [Figs.
3(B) and 4(A)]. This indicates that an increase in GTP
hydrolysis is not responsible for altering the distribu-
tion of dynamin under these conditions.
or 150 mM) in extraction of shits1dynamin from membrane
fractions. The ionic composition of grinding salines altered
the distribution of dynamin between plasma membrane
(P1), vesicular membrane (P2), and cytosolic fraction (S2)
in different shi alleles. In contrast, ionic composition did not
affect the distribution of cysteine string protein (CSP) im-
munoreactivity. Grinding solutions were: (A) Intracellular
saline with 144 mM KCl; (B) HEPES column buffer with 0
mM NaCl (HCB0); (C) HEPES column buffer with 150 mM
NaCl (HCB150). All samples were prepared at 4°C.
K?(144 mM) was more effective than Na?(128
from membrane fractions (P1 and P2) to S2 in shits1. Grind-
ing solution was Jan and Jan saline (128 mM NaCl) with:
(A) No GTP analogs; (B) 100 ?M GTP-?S; (C) 1 mM GTP.
All samples were prepared at room temperature. After
grinding, the homogenate was incubated for 30 min before
GTP and GTP-?S shifted a portion of dynamin
shits1Alters Properties of Dynamin
The influence of GTP-?S was only observed in the
presence of monovalent cations [Fig. 4(B)]. Thus,
GTP analogs appeared to enhance the sensitivity to
salt extraction, while not eliminating the requirement
for monovalent cations to solubilize dynamin. These
results suggest that GTP binding promotes a confor-
mational reconfiguration of dynamin that allows ac-
cessibility to monovalent cations, which is essential
for dissociation from the membrane pellets to the
We examined the possibility of membrane association
of the pellet fraction of dynamin. Triton-X treatment
led to the complete solubilization of CSP [Fig. 5; cf.
Fig. 2(B,C)], a protein that is associated with synaptic
vesicle and plasma membranes (Zinsmaier et al.,
1990; Eberle et al., 1998). In contrast, Triton-X did
not mobilize dynamin to the S2 fraction when shits1or
other genotypes were prepared in Na?-free HCB0
[Fig. 5(A2)]. Interestingly, Triton-X did increase the
proportional amount of dynamin in S2 of shits1ground
in high-Na?conditions [Fig. 5(B2)]. However, even
with high Na?, a substantial amount of dynamin
remained in the pellet fractions after Triton-X treat-
ment in all genotypes, which indicates that dynamin
aggregates can be trapped with nonmembranous com-
ponents in the cell, such as cytoskeleton.
These in vitro results, summarized in Table 2,
provide an interesting and unexpected opening for
tide phosphates for enhancing the distribution of dynamin to
S2 in shits1. (A1) GTP-?S. (A2) GTP. Grinding solution in
(A) was HCB128. (B) In HCB0, 200 ?M GTP-?S did not
shift the distribution of dynamin in shits1. After grinding, the
homogenate was incubated for 30 min before centrifugation.
All samples were prepared at 4°C temperature.
Relative potencies of GTP analogs and nucleo-
string protein (CSP), with Triton-X extraction. (A1) Grinding in HCB0 eliminated dynamin
distribution to S2. (A2) 1% Triton-X with HCB0 caused all CSP to shift to cytosolic fraction (S2)
but did not significantly affect dynamin. (B1) Grinding in HCB128 led to a major shift of dynamin
to S2 in OR and shits2, but not shits1. (B2) 1% Triton-X with HCB128 again caused all CSP to shift
to S2, and did induce a shift of dynamin in shits1. All samples were prepared at 4°C.
Dynamin does not follow the distribution of the vesicle-associated protein, cysteine
Chen et al.
investigating the biophysical properties of dynamin
and its aggregation. This in itself does not confirm
that these effects have functional significance in
vivo. Although the three fractions, P1, P2, and S2,
correlate to different pools of cellular components
(Gass et al., 1995), our results indicate that dy-
namin may shift between components depending
upon chemical conditions during grinding and cen-
trifugation. Moreover, the distribution of dynamin
differed less between shits1and other genotypes
when using high-K?(144 mM) grinding solution,
an electrophysiological “intracellular” patch pipette
filling solution selected here with the intention of
mimicking intracellular conditions. Finally, the dif-
ference in distribution was seen after processing at
“permissive” temperatures (4°C or room tempera-
ture) that do not induce the shibire mutant pheno-
type in vivo.
Recovery from Paralysis and
Interactions with awd
Behavioral and genetic experiments revealed notable
differences between shits1and the other two temper-
ature-sensitive alleles in vivo that have not been char-
acterized previously. High-temperature conditions ac-
centuated differences between shits1and the other
mutant alleles. After paralysis by brief exposure to
restrictive temperatures (1 min at 38°C), the time to
recovery was nearly 10-fold longer in shits1than in
shits2or shits4[Fig. 6(A)]. Also in shits1but not the
other mutants, recovery time was strongly propor-
tional to the length of heat treatment (data not shown).
In addition, a mutant allele of the NDP kinase gene
awd, which diminishes the availability of GTP in cells
(Krishnan et al., 2001), reduced the temperature
threshold for paralysis by 6°C in shits2and shits4but
only slightly in shits1[Fig. 6(B)]. It is clear that the
mutant flies from paralysis after a heat pulse (1 min, 38°C) took longer in shits1than in shits2or
shits4. (B) The awd allele awdmsf15reduced the temperature threshold substantially for shits2(lower
left) and shits4(lower right), but not for shits1(upper right).
Temperature-sensitive paralysis of shi alleles differs in vivo. (A) Recovery of adult
shits1Alters Properties of Dynamin
shits1mutation does cause physical changes of dy-
namin that have unique functional consequences in
the living fly.
Aggregation and Dissociation of
Dynamin In Vitro
Results of all cell fractionation experiments are sum-
marized qualitatively in Table 2. Sedimentation as-
says have been used in the past to indicate changes in
the aggregation state of dynamin in vitro (Hinshaw
and Schmid, 1995). Similarly, in the cell fractionation
protocol employed here, distribution of dynamin mol-
ecules in the P1 or P2 fragments implies that dynamin
has either aggregated together or become associated
with other molecular components. These pellet frac-
tions (P1 and P2) contain membrane (Gass et al.,
1995), but the majority of dynamin resisted Triton-X
extraction from the pellets to the supernatant (S2)
when Na?was absent, while the vesicle-associated
protein CSP was completely extracted (Fig. 5). This
suggests that dynamin may have complexed with cy-
toskeletal components such as actin (Ochoa et al.,
2000). Some of this Triton-resistant dynamin was
released to the cytosolic supernatant fraction with
high salt (Figs. 2, 4, and 5, and Table 2). These results
indicate that the monovalent cationic strength criti-
cally regulates dynamin aggregation and distribution
during cell fractionation.
Monovalent cations could act by screening nega-
tively charged residues or by altering the conformation
salt extraction of dynamin (Fig. 2), perhaps because K?
carries fewer waters of hydration. Interestingly, clathrin-
mediated (and therefore dynamin-dependent) endocyto-
sis can be interrupted by cytosolic K?depletion (Larkin
et al., 1983), although dynamin-dependent endocytosis
that is not suppressed by K?depletion has also been
shown (Artalejo et al., 1995).
Unlike other temperature-sensitive alleles and wild
type, shits1dynamin resisted salt extraction in grind-
ing solutions with high Na?and low K?. When a
high-K?intracellular patch pipette saline (144 mM
K?) was used for cell fractionation in an attempt to
approach in vivo conditions, dynamin of shits1distrib-
uted more normally than in high-Na?saline [Fig.
2(A)]. However, slight reductions of K?concentra-
tion below 144 mM revealed obvious differences in
dynamin distribution between shits1and other alleles
(data not shown). The critical concentration of K?for
shits1dynamin in vitro thus appears to be close to
physiological K?concentrations that pertain in vivo.
It is therefore not surprising that dynamin in shits1
mutant flies is functional under normal conditions in
vivo, yet differs functionally from other temperature-
sensitive alleles under extreme conditions such as
high temperature or reduced GTP availability (Fig. 6).
The unique reduction of the sensitivity of shits1
dynamin to salt extraction in vitro (Figs. 1, 2, and 5)
suggests that the influence of charge interactions upon
dynamin function that involves interaction among dy-
namin molecules or associations with other partners
may be affected. Together with the slow recovery
from temperature-induced paralysis in vivo [Fig.
6(A)], this suggests that the mutation may affect the
conformation of dynamin. This is consistent with the
enhancement of salt extraction of shits1dynamin by
GTP analogs in vitro (Figs. 3 and 4). The shits1
mutation lies in the GTP binding domain (van der
Bliek and Meyerowitz, 1991). Binding of GTP or
GTP-?S to the GTPase site could reconfigure shits1
dynamin to make salt extraction more effective. Site-
directed mutagenesis in a recent study (Marks et al.,
2001) indicates the importance of consensus GTP-
binding regions (G1–G4) within the GTPase domain
for GTP hydrolysis or conformational changes impor-
tant to endocytic function. The shits2point mutation
(G141S) lies four residues from the G3 consensus
region, and the shits4mutation (P171S) is midway
between the G3 (131–137) and G4 (200–203) regions
(van der Bliek and Meyerowitz, 1991; Grant et al.,
1998; Marks et al., 2001) [positions of G3 and G4
given in Marks et al. (2001) have been adjusted by ?5
for Drosophila dynamin (van der Bliek and Meyer-
owitz, 1991)]. In contrast, the shits1
(G268D) is 65 residues removed from the G4 consen-
sus region (van der Bliek and Meyerowitz, 1991) and
97 residues from shits4. Our results therefore point to
another functionally important residue or region to be
analyzed further for mediating factors that influence
the endocytic cycle. To distinguish whether the shits1
mutation modifies a function that this region of dy-
namin normally performs, or instead interacts with
another location such as one of the identified GTP-
binding motifs, will require a systematic effort of
site-directed mutagenesis. Such analysis could also
clarify the connection between the in vitro biochem-
ical effects and the in vivo temperature-sensitive phe-
notype of shits1by indicating whether these can be
Functional Implications In Vivo
Work over the past decade has produced a provisional
view of dynamin’s role in the endocytic cycle (Fig. 7)
Chen et al.
with the involvement of GTP binding and hydrolysis,
conformational changes, interactions with other pro-
teins and lipids, and phosphorylation state (Schmid et
al., 1998; Sever et al., 1999; Marks et al., 2001).
These factors are mediated in different regions in the
dynamin molecule, yet they interact functionally. The
shits1mutation lies at the boundary of the GTPase and
pleckstrin homology (PH) domains, and the mutation
has been suggested to affect the conformational inter-
action between this region and the other functional
domains (Damke et al., 1995). These conformational
changes could be influenced by charge effects such as
the phosphorylation state of dynamin, binding to nu-
cleotide triphosphates, or the ionic strength of the
surrounding solution. While the charge sensitivity of
dynamin has previously been noted in vitro (Carr and
Hinshaw, 1997), the shits1mutation points to the
functional importance of charge-dependent conforma-
tional changes in vivo. Dynamin has also been shown
to associate or interact functionally with a wide array
of other molecules that participate in vesicle cycling
(Schmid et al., 1998), and these interactions could be
disrupted by the shits1mutation. It would therefore be
productive to look for genetic interactions of shits1
with mutations affecting such molecular partners.
Mutations of the nucleotide diphosphate kinase
gene awd decrease GTP availability and mimic the shi
phenotype (Krishnan et al., 2001), consistent with the
GTP dependence of dynamin activity in vivo, and in
double-mutant combinations of awd with shits2or
shits4, the threshold temperature for paralysis is re-
duced [Fig. 6(B); Krishnan et al., 2001]. Strikingly,
however, awd did not reduce the threshold tempera-
ture for shits1[Fig. 6(B)]. This suggests that GTP
binding is not a rate-limiting step for dynamin cycling
in shits1to the same degree as in other shibire geno-
possible effects of the shits1mutation. For clarity, binding to GTP or GDP is shown for only one
dynamin at each step, and dynamin phosphorylation and proposed interactions with other com-
pounds are not shown. GTP-bound dynamin tetramers associate with clathrin-coated plasma
membrane (1) and self-assemble into helical collars (2), and might then activate other proteins that
promote budding (Sever et al., 1999). Overlap of the helical ends (3) promotes GTP hydrolysis (4),
which might drive vesicle fission (Schmid et al., 1998; Marks et al., 2001) or signal dynamin to
disassemble as fission progresses (Sever et al., 1999). Dynamin goes to cytosol (5) where it binds
to GTP reconstituted by the AWD nucleotide diphosphate kinase (6). Dashed lines indicate putative
alternative pathways that shits1dynamin might take under abnormal conditions (low GTP or high
temperature). (A) shits1mutant dynamin may not require GTP binding for early stages of endocy-
tosis, as suggested by results from shits1; awdmsf15double mutants [Fig. 6(B)]. (B) shits1dynamin
may become trapped in a metastable aggregated state at restrictive temperatures. (C) This metastable
state may retard the return of shits1dynamin to the normal cycle after a return to permissive
temperatures [Fig. 6(A)].
The hypothetical dynamin cycle in vivo (Schmid et al., 1998; Sever et al., 1999) with
shits1Alters Properties of Dynamin
types (shits2, shits4, or wild type). It has been proposed
that GTP binding may activate dynamin to act as a G
protein (Sever et al., 1999) or that GTP hydrolysis
may drive dynamin to cleave the budding vesicle from
the plasma membrane (Schmid et al., 1998; Marks et
al., 2001). Certainly, both GTP binding and GTP
hydrolysis are associated with dynamin-mediated en-
docytosis (Fig. 7). Our results with shi; awd double
mutants may indicate that shits1dynamin can carry out
early steps in the endocytic cycle without binding to
GTP (Fig. 7, Step A). Alternatively, shits1might in-
crease dynamin’s GTP binding affinity or reduce the
rate of GTP hydrolysis such that GTP-bound dynamin
is available even when GTP abundance is reduced by
At restrictive temperatures, dynamin of tempera-
ture-sensitive shibire mutations may become stuck or
diverted at a stage of the cycle that precedes vesicle
fission (Fig. 7, Step B), leading to endocytic arrest.
However, the time to recovery from paralysis was
nearly 10-fold longer in shits1than in other shitsmu-
tants after a brief (1 min) exposure to high tempera-
ture [38°C; Fig. 6(A)]. Preliminary electron micros-
copy data also indicate that temperature-induced
arrest of endocytosis at the larval neuromuscular junc-
tion is much more prolonged in shits1than in shits2
(data not shown). Moreover, the recovery time for
shits1increases with longer exposure to heat (data not
shown); this effect is much less apparent for shits2or
shits4. This implies that heat treatment causes shits1
dynamin to become trapped in a unique metastable
state that prevents it from functioning normally for
some time after a return to permissive temperatures
(Fig. 7, Step C).
This hypothetical trapped dynamin state in shits1
mutant flies would most likely be associated with
plasma membrane or cytoskeleton. Ultrastructural
analysis shows that dynamin collared pits can form at
restrictive temperatures in shits1(Koenig and Ikeda,
1989), suggesting that shits1mutant dynamin can still
associate with budding vesicles under restrictive con-
ditions. A parallel finding has been reported. When
human dynamin with a point mutation identical to
shits1was overexpressed in HeLa cells (so that the
majority of dynamin was cytosolic at any time), this
dynamin formed cytosolic aggregations when ex-
posed to restrictive temperatures (Damke et al., 1995).
The fact that a single identified base substitution
causes unique effects distinct from other temperature
sensitive alleles both in vitro and in vivo is remarkable
in itself, yet questions can be raised as to whether the
changes observed in vitro are directly related to the
temperature-sensitive phenotype. Whereas the shits1,
shits2, and shits4mutations all affect endocytic cycling
in a temperature-dependent manner, shits1also alters
nontemperature-sensitive properties in vitro, revealed
at the level of salt-dependent extraction and cellular
component association, that could modify membrane
cycling and paralytic behavior in vivo in a subtle
manner, reflected in the lengthening of recovery time
depending on duration of heat exposure and the al-
tered interaction with awdmsf15. It has been difficult to
extrapolate the biochemical and biophysical proper-
ties of dynamin observed in vitro to physiological
conditions in vivo, in part because those conditions
(such as K?concentration) may not be precisely
known, and other proteins that interact with dynamin
in vivo may not have been appropriately reconstructed
in vitro. Moreover, mechanisms of endocytosis are
difficult to observe directly because of the challenges
of fixing and isolating very rapid processes. Consis-
tent with its behavioral and cytological phenotypes in
vivo, the shits1mutation induces a tendency for dy-
namin to aggregate with unique biophysical properties
that can be studied in vitro. These may prove useful
for examining the energetic and kinetic aspects of
dynamin function, and may allow trapping of dy-
namin in what is normally a transition state in the
endocytic cycle. More importantly, these results sug-
gest a way to link biophysical and physiological un-
derstandings of dynamin and its roles in synaptic
L.M. was supported by NIH grant NS34889 to M. Ra-
maswami. We are grateful to E. Buchner for the gift of
anti-CSP antibodies, and to M. Ramaswami for helpful
comments on the manuscript. The initial observation of a
prolonged recovery time for shits1was made by R. Rikhy.
Artalejo CR, Henley JR, McNiven MA, Palfrey HC. 1995.
Rapid endocytosis coupled to exocytosis in adrenal chro-
maffin cells involves Ca2?, GTP, and dynamin but not
clathrin. Proc Natl Acad Sci USA 92:8328–8332.
Carr JF, Hinshaw JE. 1997. Dynamin assembles into spirals
under physiological salt conditions upon the addition of
GDP and ?-phosphate analogues. J Biol Chem 272:28030–
Chen MS, Obar RA, Schroeder CC, Austin TW, Poodry
CA, Wadsworth SC, Vallee RB. 1991. Multiple forms of
dynamin are encoded by shibire, a Drosophila gene in-
volved in endocytosis. Nature (London) 351:583–586.
Damke H, Baba T, van der Bliek AM, Schmid SL. 1995.
Clathrin-independent pinocytosis is induced in cells over-
expressing a temperature-sensitive mutant of dynamin.
J Cell Biol 131:69–80.
Damke H, Baba T, Warnock DE, Schmid SL. 1984. Induc-
Chen et al.
tion of mutant dynamin specifically blocks endocytic
vesicle formation. J Cell Biol 127:915–934.
Eberle KK, Zinsmaier KE, Buchner S, Gruhn M, Jenni M,
Arnold C, Leibold C, Reisch D, Walter N, Hafen E,
Hofbauer A, Pflugfelder GO, Buchner E. 1998. Wide
distribution of the cysteine string proteins in Drosophila
tissues revealed by targeted mutagenesis. Cell Tissue Res
Gass GV, Lin JJ-C, Scaife R, Wu C-F. 1995. Two isoforms
of Drosophila dynamin in wild-type and shibiretsneural
tissue: Different subcellular localization and association
mechanisms. J Neurogenetics 10:169–191.
Grant D, Unadkat S, Katzen A, Krishnan KS, Ramaswami
M. 1998. Probable mechanisms underlying interallelic
complementation and temperature-sensitivity of muta-
tions at the shibire locus of Drosophila melanogaster.
Grigliatti T, Hall L, Rosenbluth R, Suzuki D. 1973. Tem-
perature-sensitive mutations in Drosophila melanogaster.
XIV. A selection of immobile adults. Mol Gen Genet
Hinshaw JE, Schmid SL. 1995. Dynamin self-assembles
into rings suggesting a mechanism for coated vesicle
budding. Nature (London) 374:190–192.
Jan LY, Jan YN. 1976. L-glutamate as an excitatory trans-
mitter at the Drosophila larval neuromuscular junction.
J Physiol (London) 262:215–236.
Kawasaki F, Hazen M, Ordway RW. 2000. Fast synaptic
fatigue in shibire mutants reveals a rapid requirement for
dynamin in synaptic vesicle membrane trafficking. Nat
Kim Y-T, Wu C-F. 1987. Reversible blockage of neurite
development and growth cone formation in neuronal cul-
tures of a temperature-sensitive mutant of Drosophila.
J Neurosci 7:3245–3255.
Kim Y-T, Wu C-F. 1990. Allelic interactions at the shibire
locus of Drosophila: Effects on behavior. J Neurogenet-
Koenig JH, Ikeda K. 1989. Disappearance and reformation
of synaptic vesicle membrane upon transmitter release
observed under reversible blockage of membrane re-
trieval. J Neurosci 9:3844–3860.
Koenig JH, Saito K, Ikeda K. 1983. Reversible control of
synaptic transmission in a single gene mutant of Dro-
sophila melanogaster. J Cell Biol 96:1517–1522.
Koenig JH, Yamaoka K, Ikeda K. 1998. Omega images at
the active zone may be endocytotic rather than exocy-
totic: Implications for the vesicle hypothesis of transmit-
ter release. Proc Natl Acad Sci USA 95:12677–12682.
Kosaka T, Ikeda K. 1983a. Possible temperature-dependent
blockage of synaptic vesicle recycling induced by a single-
gene mutation in Drosophila. J Neurobiol 14:207–225.
Kosaka T, Ikeda K. 1983b. Reversible blockage of mem-
brane retrieval and endocytosis in the garland cell of the
temperature-sensitive mutant of Drosophila melano-
gaster, shibirets1. J Cell Biol 97:499–507.
Krishnan KS, Rikhy R, Rao S, Shivalkar M, Mosko M,
Narayanan R, Etter P, Estes PS, Ramaswami M. 2001.
Nucleoside Diphosphate Kinase, a source of GTP, is
required for dynamin-dependent synaptic vesicle recy-
cling. Neuron 30:197–210.
Larkin JM, Brown MS, Goldstein JL, Anderson RGW.
1983. Depletion of intracellular potassium arrests coated
pit formation and receptor-mediated endocytosis in fibro-
blasts. Cell (Cambridge, MA) 33:273–285.
Marks B, Stowell MHB, Vallis Y, GMills IG, Gibson A,
Hopkins CR, McMahon HT. 2001. GTPase activity of
dynamin and resulting conformation change are essential
for endocytosis. Nature (London) 410:231–235.
Masur SK, Kim Y-T, Wu C-F. 1990. Reversible inhibition
of endocytosis in cultured neurons from the Drosophila
temperature-sensitive mutant shibirets1. J Neurogenetics
McNiven MA. 1998. Dynamin: A molecular motor with
pinchase action. Cell (Cambridge, MA) 94:151–154.
Muhlberg AB, Warnock DE, Schmid SL. 1997. Domain
structure and intramolecular regulation of dynamin
GTPase. EMBO J 16:6676–6683.
Ochoa G-C, Slepnev VI, Neff L, Ringstad N, Takei K,
Daniell L, Kim W, Cao H, McNiven M, Baron R, De
Camilli P. 2000. A functional link between dynamin and
the actin cytoskeleton at podosomes. J Biol Chem 150:
Ozawa S, Hagiwara S. 1976. Synaptic transmission revers-
ibly conditioned by single-gene mutation in Drosophila
melanogaster. Nature (London) 259:489–491.
Schmid SL, McNiven MA, de Camilli P. 1998. Dynamin
and its partners: A progress report. Curr Opin Cell Biol
Sever S, Muhlberg AB, Schmid SL. 1999. Impairment of
dynamin’s GAP domain stimulates receptor-mediated en-
docytosis. Nature 398:481–486.
Siddiqi O, Benzer S. 1976. Neurophysiological defects in
temperature-sensitive paralytic mutants of Drosophila
melanogaster. Proc Natl Acad Sci USA 73:3253–3257.
Urrutia R, Henley JR, Cook T, McNiven MA. 1997. The
dynamins: Redundant or distinct functions for an expand-
ing family of related GTPases? Proc Natl Acad Sci USA
van der Bliek AM. 1999. Functional diversity in the dy-
namin family. Trends Cell Biol 9:96–102.
van der Bliek AM, Meyerowitz EM. 1991. Dynamin-like
protein encoded by the Drosophila shibire gene associated
with vesicular traffic. Nature (London) 351:411–414.
Warnock DE, Hinshaw JE, Schmid SL. 1996. Dynamin
self-assembly stimulates its GTPase activity. J Biol Chem
Zhao ML, Sable EO, Iverson LE, Wu C-F. 1995. Functional
expression of Shaker K?channels in cultured Drosophila
“giant” neurons derived from Sh cDNA transformants:
Distinct properties, distribution, and turnover. J Neurosci
Zinsmaier KE, Hofbauer A, Heimbeck G, Pflugfelder GO,
Buchner S, Buchner E. 1990. A cysteine-string protein is
expressed in retina and brain of Drosophila. J Neuroge-
shits1Alters Properties of Dynamin