The enzyme-binding region of human GM2-activator
protein
Michaela Wendeler
1
, Norbert Werth
1
, Timm Maier
2
, Guenter Schwarzmann
1
, Thomas Kolter
1
,
Maike Schoeniger
1
, Daniel Hoffmann
3
, Thorsten Lemm
1
, Wolfram Saenger
2
and Konrad Sandhoff
1
1 Kekule
´
-Institut fu
¨
r Organische Chemie und Biochemie der Universita
¨
t Bonn, Germany
2 Institut fu

Functional deficiencies in this protein result in a fatal neurological storage
disorder, the AB variant of GM2 gangliosidosis. In order to elucidate this
cofactor’s mode of action and identify the surface region of GM2AP
responsible for binding to HexA, we designed several variant forms of this
protein and evaluated the consequences of these mutations for lipid- and
enzyme-binding properties using a variety of biophysical and functional
studies. The point mutants D113K, M117V and E123K showed a drastic-
ally decreased capacity to stimulate HexA-catalysed GM2 degradation.
However, surface plasmon resonance (SPR) spectroscopy showed that the
binding of these variants to immobilized lipid bilayers and their ability to
solubilize lipids from anionic vesicles were the same as for the wild-type
protein. In addition, a fluorescence resonance energy transfer (FRET)-
based assay system showed that these variants had the same capacity as
wild-type GM2AP for intervesicular lipid transfer from donor to acceptor
liposomes. The concentration-dependent effect of these variants on hydro-
lysis of the synthetic substrate 4-methylumbelliferyl-2-acetamido-2-deoxy-6-
sulfo-b-d-glucopyranoside (MUGS) indicated a weakened association with
the enzyme’s a subunit. This identifies the protein region affected by these
mutations, the single short a helix of GM2AP, as the major determinant
for the interaction with the enzyme. These results further confirm that the
function of GM2AP is not restricted to a biological detergent that simply
disrupts the membrane structure or lifts the substrate out of the lipid plane.
In contrast, our data argue in favour of the critical importance of distinct
activator–hexosaminidase interactions for GM2 degradation, and corrobor-
ate the view that the activator ⁄ lipid complex represents the true substrate
for the degrading enzyme.
Abbreviations
BMP, bis(monoacylglycero)phosphate; FRET, fluorescence resonance energy transfer; GM1, ganglioside GM1; GM2, ganglioside GM2;
GM2AP, GM2-activator protein; GSLs, glycosphingolipids; HexA, b-hexosaminidase A; HexB: b-hexosaminidase B; LUV, large unilamellar
vesicle; MUG, 4-methylumbelliferyl-2-acetamido-2-deoxy-b-

encoding the polypeptides involved in GM2 degrada-
tion, the a and b subunit of HexA or the GM2AP,
result in the accumulation of nondegraded glycolipids
within the lysosomal compartment and the develop-
ment of severe neurodegenerative storage diseases
known as GM2-gangliosidoses [4,6,7].
In its mature form present in the lysosomes, the
GM2 activator is a small glycoprotein of 162 amino
acids [8]. It is known to bind avidly to a variety of ani-
onic lipids in vitro [9], to be able to extract several gly-
cosphingolipid monomers from micelles or liposomes
and to transport them as soluble 1 : 1 complexes
between donor and acceptor membranes [10]. The
membrane activity of GM2AP was found to depend
critically on several physiological parameters, most
notably a lateral pressure of the lipid bilayer below
15–25 mNÆm
)1
[11] and the presence of anionic lipids,
in particular bis(monoacylglycero)phosphate (BMP)
[12]. Recently, a novel function of GM2AP in the con-
text of glycolipid antigen presentation via CD1d was
identified [13].
For the function of GM2AP in the lysosomal GM2
hydrolysis, the term ‘liftase’ was coined: it recognizes
the lipid substrate within the membrane plane and lifts
it out of the lipid bilayer, thereby presenting it to the
water-soluble enzyme for degradation [14]. In addition,
it has been suggested that GM2AP modifies the
conformation of the trisaccharide unit of ganglioside

correct orientation of the tetrasaccharide head-group
with respect to the degrading enzyme’s active site.
The goal of this study was to elucidate the interac-
tion between cofactor and enzyme and to develop a
comprehensive model for the GM2 degradation pro-
ceeding at a phase frontier on intralysosomal mem-
branes. Previously, attempts to identify the lipid- and
enzyme-binding region of GM2AP were hampered
because assay systems for GM2AP, which measure its
stimulatory potential for the degradation of GM2 by
HexA, reflect simultaneously both the interaction of
GM2AP with GM2 and the interaction of the
GM2AP ⁄ GM2 complex with the enzyme. To delineate
in detail the separate effect of the introduced muta-
tions on the lipid- and the enzyme-binding of this
cofactor, we combined a variety of functional and bio-
physical analyses.
Crystallographic analysis of the homodimeric iso-
enzyme HexB [24,25] and subsequent comparative
modelling studies of heterodimeric HexA [25] suggested
that the dimer interface of the enzyme forms the dock-
ing site for the GM2AP ⁄ GM2 complex. Furthermore,
M. Wendeler et al. Enzyme-binding region of GM2-activator protein
FEBS Journal 273 (2006) 982–991 ª 2006 The Authors Journal compilation ª 2006 FEBS 983
based on the crystal structures of the enzyme [24,25]
and of GM2AP [20–22], computational modelling and
theoretical docking studies identified the single short
a helix of GM2AP as the region most likely to be in
direct contact with the enzyme. We therefore expressed
a series of site-directed mutants of GM2AP with amino

of GM2AP and evaluated the biophysical and func-
tional properties of the variant proteins.
Initial homology modelling of HexA based on the
recently published crystal structure of human HexB
[25], and subsequent theoretical docking studies of the
GM2AP structure [20,21] suggested that the a helix of
GM2AP plays a major role in binding to the enzyme.
In this helix, comprising amino acids F111 to P120,
every fourth residue points in the same direction and
the side chains of D113 and M117 were found to be
oriented away from the GM2AP core towards the
putative interface with the enzyme. The first residue
following the GM2AP helix, which likewise points
towards this suggested interaction region, was E123.
The long and charged side chain of glutamate is
known to be particularly well suited for protein–pro-
tein interactions. In contrast, the residues between the
end of the helix and E123 (P120, T121, G122) most
likely serve a structural role only, allowing the loop to
adopt a bent conformation.
We therefore introduced the following mutations in
this region of the protein: D113K, D113A, D113Y,
M117V, E123K, E123A and E123Y. All variant pro-
tein forms were readily produced in the insect cell⁄
baculovirus expression vector system (BEVS) and puri-
fied to homogeneity by cation exchange and sub-
sequent Ni-NTA chromatography. Protein yields of
GM2AP mutants were in the range 4–7 mg purified
protein per litre of expression supernatant, compared
with 7–9 mg of the WT protein.

To assess the membrane activity and lipid-binding
behaviour of the GM2AP variants, their interaction
with immobilized lipid bilayers was measured using
SPR spectroscopy in a Biacore instrument (Fig. 3). All
mutant protein forms showed binding to the same
extent as the WT protein. In the presence of the ani-
onic lipid BMP, which is known to occur in inner
membranes of the acidic compartments and which
Enzyme-binding region of GM2-activator protein M. Wendeler et al.
984 FEBS Journal 273 (2006) 982–991 ª 2006 The Authors Journal compilation ª 2006 FEBS
stimulates membrane activity of GM2AP [12], all
mutants showed a comparable pronounced decrease in
the SPR signal during the subsequent injection of pro-
tein-free buffer. This response has been interpreted as
the solubilization of lipids from the surface of immobi-
Fig. 2. Stimulation of HexA-catalysed ganglioside GM2 degradation
by WT and variant GM2AP. (A) Substrate hydrolysis measured in a
micellar assay system. (B) GM2 degradation measured in a lipo-
somal assay system with LUVs containing 50 mol% PtdCho,
20 mol% Chol, 10 mol% GM2 and 20 mol% BMP. Values repre-
sent means of duplicate measurements; deviations observed were
< 5%.
Fig. 3. Interaction of WT and variant GM2AP with immobilized vesi-
cles measured by SPR spectroscopy. Negatively charged LUVs
containing 20 mol% Chol, 50 mol% PtdCho, 10 mol% GM2 and
20 mol% BMP, were immobilized on a Pioneer L1 sensorchip.
GM2AP (2 l
M) was injected at a flow rate of 20 lLÆmin
)1
in 50 mM

localized within the same model membrane and thereby
quenches NBD fluorescence. It has been shown previ-
ously that GM1 and GM2 are transferred almost
equally well between vesicles by GM2AP [10]. These
novel fluorescent gangliosides are easily incorporated
into model membranes, where they exhibit an extremely
low ‘off rate’ and are recognized and degraded by HexA
in the presence of GM2AP, similar to their natural ana-
logue. Upon addition of catalytic amounts of GM2AP
to donor vesicles containing NBD–gangliosides, the
activator-mediated glycolipid transfer from donor to
acceptor liposomes increases the NBD fluorescence,
permitting continuous real-time monitoring of GM2AP
transfer activity. As shown in Fig. 4, the point mutants
examined here showed the same ability as WT GM2AP
to transfer NBD–GM1 from donor to acceptor vesicles.
The initial rate of NBD transfer and the capacity
observed under equilibrium conditions are identical.
It has previously been shown that, in the absence of
ganglioside GM2, the GM2 activator protein inhibits
hydrolysis of the synthetic fluorogenic substrate
MUGS by HexA [5]. This negatively charged substrate
can only be degraded by the a subunit of the enzyme,
and the inhibitory effect of GM2AP has been inter-
preted as being indicative of a physical association
between GM2AP and the enzyme which blocks this
subunit’s active site. As shown in Fig. 5, all variant
forms of GM2AP examined show concentration-
dependent inhibition of MUGS degradation. However,
their inhibitory effect is less pronounced than that of

Values are means of duplicate measurements, deviations observed
were always < 5%.
Fig. 5. Competition experiment. Degradation of the synthetic fluo-
rogenic substrate MUGS by HexA is inhibited in the presence of
GM2AP in a concentration-dependent manner, suggesting that
HexA and GM2AP form a complex which sterically blocks the act-
ive site of the HexA a subunit. In the presence of GM2AP variants,
MUGS degradation is reduced to a lesser extent, indicating that the
physical association of HexA with these mutant forms is weaker.
Enzyme-binding region of GM2-activator protein M. Wendeler et al.
986 FEBS Journal 273 (2006) 982–991 ª 2006 The Authors Journal compilation ª 2006 FEBS
lipid- and enzyme-binding properties, we tested the
membrane activity of GM2AP using SPR spectroscopy
and probed the lipid-transfer properties in a FRET-
based assay system. The ability to bind specifically the
a subunit of HexA could be assessed by studying
the competition of GM2AP and its variants with the
hydrolysis of the synthetic fluorogenic hexosaminidase
substrate MUGS.
Our data clearly indicate that mutations introduced
into the a-helical region of GM2AP drastically
decreased the ability of the activator to stimulate
HexA-catalysed degradation of GM2. However, the
membrane activity of these variants, as well as their
lipid-binding and transfer ability, was identical to that
of the WT protein. All variants could still bind speci-
fically to the a subunit of HexA, but their association
with the enzyme was weaker than for WT GM2AP.
This clearly identifies the region affected by these
mutations, the single a helix of GM2AP, as the major

GM2AP [20,21] and the known disulfide bond connec-
tivity [27] indicate that the disulfide bond formed by
C138 with C112 forms a ‘clamp’ that structurally
restrains the activator’s helical domain. It is therefore
most likely that the mutation C138R distorts the orien-
tation of the helix and compromises the activator’s
ability to interact with the enzyme.
In addition to mapping the activator’s binding inter-
face with the enzyme, our data confirm the critical
importance of distinct protein–protein interactions in
lysosomal GM2 degradation. The activator’s role is
not limited to that of a biological detergent that simply
disrupts the membrane structure. In contrast, the acti-
vator ⁄ glycolipid complex represents the true Michaelis
substrate for the degrading enzyme.
Experimental procedures
Materials
For insect cell culture, serum-free IPL-41 medium was pur-
chased from Invitrogen Life Technologies (Carlsbad, CA).
The baculovirus transfer vector pAcMP3, linearized baculo-
virus DNA (BaculoGold), the transfection kit and the insect
cell lines were from Pharmingen, obtained via BD Bio-
sciences (Erembodegem, Belgium). Phosphatidylcholine (egg
yolk, PtdCho) and cholesterol (Chol) were purchased from
Sigma (Taufkirchen, Germany). BMP was obtained from
Avanti Polar Lipids (Alabaster, AL), Lichroprep RP18 was
Fig. 6. Model for the interface in the ternary complex of GM2AP,
GM2 and HexA resulting from theoretical docking studies based on
the known crystal structures of GM2AP [20–22] and HexB [25]. In
the GM2AP structure b sheets are blue, the single a helix is green,

loviruses were then generated by cotransfection of Sf9 cells
with the respective transfer vector and linearized BaculoGold
viral DNA according to the manufacturer’s protocol. Pure
viral stocks were obtained using the end-point dilution
method.
For protein expression, Sf9 cells grown in serum-free
IPL-41 to a cell density of 1.5 · 10
6
cellsÆmL
)1
were infec-
ted with a viral stock at a multiplicity of infection of 5. The
medium was harvested 96 h after infection.
Protein purification
Proteins were purified, with modification, following the
method described by Wendeler et al. [28]. Briefly, the super-
natant was first subjected to perfusion chromatography on
a cation exchange resin (Poros HS) using a BIOCADSprint
System HPLC workstation (Applied Biosystems). Fractions
containing recombinant GM2AP were combined and then
further purified by immobilized metal ion affinity chroma-
tography on Ni-NTA-agarose (Qiagen, Hilden, Germany).
The eluted proteins were analysed by electrophoresis on a
12.5% tricine-SDS ⁄ polyacrylamide gel and visualized by
silver staining. The polyclonal antibody raised against WT
GM2AP [29] recognized all GM2AP mutants. Amino acid
exchanges in the recombinant proteins were confirmed by
ESI-TOF-MS.
Micellar GM2AP assay
In a micellar in vitro system, the activity of recombinant

Liposomal assay systems
To mimic more closely the reaction conditions encountered
on intralysosomal vesicles and membrane structures, the
degradation of membrane-bound ganglioside GM2 by hex-
osaminidase and GM2AP was studied in a detergent-free,
liposomal assay system as described previously [12].
Vesicle prepapration
Large unilamellar vesicles (LUVs) of 100 nm were prepared
by the following procedure. Appropriate aliquots of the
lipid solutions PtdCho (50 mm, toluol ⁄ ethanol 2 : 1 v ⁄ v),
BMP (10 mm, chloroform ⁄ methanol 1 : 1 v ⁄ v), Chol
(25.6 mm, chloroform ⁄ methanol 2 : 1 v ⁄ v) and GM2 trit-
ium-labelled in its terminal GalNAc moiety (0.5 mm, tolu-
ol ⁄ ethanol 1 : 1 v ⁄ v) were mixed and dried in a stream of
nitrogen. The lipid mixture was dissolved to a total lipid
concentration of 2 mm in sodium citrate buffer (50 mm,
pH 4.2) and freeze–thawed 10 times in liquid nitrogen to
ensure solute equilibration between trapped and bulk
solutions. Unilamellar vesicles were prepared by passage
through two polycarbonate filters (pore size, 100 nm; Aves-
tin) mounted in tandem in a mini-extruder (Liposo-Fast;
Avestin, Ottawa, Canada) a total of 19 times.
Liposomal assay
The standard incubation mixture using GM2 as substrate
contained the following components in a final volume of
50 lL: BSA (50 lgÆmL
)1
), sodium citrate buffer (pH 4.2,
50 mm), unilamellar liposomes (total lipid concentration:
1mm), GM2AP (2 lm) and HexA (25 mU). Liposomes

amount of hexosaminida se that g enerates 1 lmolÆmin
)1
of
4-methylumbell iferone.
MUGS was used in competition experiments designed to
study the association of GM2AP with the a subunit of
HexA. In a volume of 40 lL, 10 mm citrate buffer, pH 4.2,
2 · 10
)5
m substrate, 6 lg BSA and an appropriate amount
of HexA were incubated at 37 °C for 30 min in the absence
or presence of various amounts of GM2AP (up to 10 lg).
Reactions were stopped by the addition of 5 vol. of a 0.2 m
glycine, 0.2 m Na
2
CO
3
solution and the generated 4-methyl-
umbelliferone determined by measuring the fluorescence at
440 nm after excitation at 365 nm.
UV CD spectroscopy
CD spectroscopy was used to monitor potential distur-
bances of the protein fold upon introduction of the individ-
ual mutations. CD spectra in the range of 250 to 200 nm
were acquired on a JASCO J600 spectropolarimeter with
four accumulations in a 0.1 cm cuvette at 10 °C for solu-
tions of 0.5 mgÆmL
)1
of the mutant proteins in 15 mm
sodium phosphate, 100 mm NaCl, pH 7. Data processing

180 s. Measurements started with the injection of GM2AP.
FRET-based assay for GM2AP
The fluorescent analogue 2-NBD–GM1 was synthesized as
described previously [26]. Large unilamellar donor vesicles
were prepared with the following composition: PtdCho
(56 mol%), Chol (20 mol%), phosphatidic acid (20 mol%),
2-NBD–GM1 (2 mol%) and rhodamine–PE (2 mol%).
Acceptor vesicles comprised PtdCho (60 mol%), Chol
(20 mol%) and phosphatidic acid (20 mol%). The final
donor vesicle concentration in the assay mixture was
20 nmolÆmL
)1
, the concentration of acceptor vesicles was
100 nmolÆmL
)1
in a total volume of 400 lLin50mm citrate
buffer, pH 4.2. The transfer of NBD–GM1 was started by
the addition of GM2AP to a final concentration of 37.5 nm
(WT or point mutants). Fluorescence measurements were
performed at 28 °C using a quartz cuvette in a Shimadzu
RF5000 instrument (Kyoto, Japan) with an excitation wave-
length of 480 nm and an emission wavelength of 522 nm.
Without the addition of GM2AP, no increase in fluorescence
was observed over a period of 2 h. For each experimental
time point over the interval ranging from 0.5 to 30 min, the
instrument’s shutter was opened for only 3–4 s, resulting in
a total of maximal 52 s illumination of NBD–GM2. Under
these conditions, photobleaching was negligible.
Acknowledgements
We thank Dr Christina Schuette, Luebeck, for provi-

infantile GM2 gangliosidosis: deficiency of a factor
necessary for stimulation of hexosaminidase A-catalyzed
degradation of ganglioside GM2 and glycolipid GA2.
Proc Natl Acad Sci USA 75, 3979–3983.
7 Platt FM & Walkley SU (2004) Lysosomal defects and
storage. In Lysosomal Disorders of the Brain (Platt FM
& Walkley SU, eds), pp. 32–49. Oxford University
Press, New York.
8 Glombitza GJ, Becker E, Kaiser HW & Sandhoff K
(1997) Biosynthesis, processing, and intracellular trans-
port of GM2 activator protein in human epidermal ker-
atinocytes. The lysosomal targeting of the GM2
activator is independent of a mannose-6-phosphate sig-
nal. J Biol Chem 272, 5199–5207.
9 Hama Y, Li YT & Li SC (1997) Interaction of the
GM2 activator protein with glycosphingolipids. J Biol
Chem 272, 2828–2833.
10 Conzelmann E, Burg J, Stephan G & Sandhoff K
(1982) Complexing of glycolipids and their transfer
between membranes by the activator protein for degra-
dation of lysosomal ganglioside GM2. Eur J Biochem
123, 455–464.
11 Giehl A, Lemm T, Bartelsen O, Sandhoff K & Blume A
(1999) Interaction of the GM2-activator protein with
phospholipid-ganglioside bilayer membranes and with
monolayers at the air–water interface. Eur J Biochem
261, 650–658.
12 Werth N, Schuette CG, Wilkening G, Lemm T & Sand-
hof K (2001) Degradation of membrane-bound ganglio-
side GM2 by beta-hexosaminidase A. Stimulation by

rst W & Sandhoff K
(1991) The human GM2 activator protein – a substrate
specific cofactor of b-hexosaminidase A. J Biol Chem
266, 1879–1887.
19 Zarghooni M, Bukovac S, Tropak M, Calhan J &
Mahuran D (2004) An alpha-subunit loop structure is
required for GM2-activator protein binding by beta-
hexosaminidase A. Biochem Biophys Res Commun 324,
1048–1052.
20 Wright CS, Li SC & Rastinejad F (2000) Crystal struc-
ture of human GM2-activator protein with a novel
b-cup topology. J Mol Biol 304, 411–422.
21 Wright CS, Zhao Q & Rastinejad F (2003) Structural
analysis of lipid complexes of GM2-activator protein.
J Mol Biol 331, 951–964.
22 Wright CS, Mi LZ & Rastinejad F (2004) Evidence for
lipid packaging in the crystal structure of the GM2-acti-
vator complex with platelet activating factor. J Mol Biol
342, 585–592.
23 Wendeler M, Hoernschemeyer J, Hoffmann D, Kolter
T, Schwarzmann G & Sandhoff K (2003) Photoaffinity
labelling of human GM2-activator protein – mechanistic
insight into ganglioside GM2 degradation. Eur J Bio-
chem 271, 614–627.
24 Mark BL, Mahuran DJ, Cherney MM, Zhao D, Knapp
S & James MNG (2003) Crystal structure of human
beta-hexosaminidase B: understanding the molecular
basis of Sandhoff and Tay-Sachs disease. J Mol Biol
327, 1093–1109.
25 Maier T, Strater N, Schuette CG, Klingenstein R,