Structure
Ways & Means
A Fluorescence-Detection Size-ExclusionChromatography-Based Thermostability Assayfor Membrane Protein Precrystallization ScreeningMotoyuki Hattori,1,3 Ryan E. Hibbs,1,3 and Eric Gouaux1,2,*1Vollum Institute2Howard Hughes Medical Institute
Oregon Health and Science University, Portland, OR 97239, USA3These authors contributed equally to this work
*Correspondence: [email protected]
http://dx.doi.org/10.1016/j.str.2012.06.009
SUMMARY
Optimization of membrane protein stability underdifferent solution conditions is essential for obtainingcrystals that diffract to high resolution. Traditionalmethods that evaluate protein stability require largeamounts of material and are, therefore, ill suited formedium- to high-throughput screening of membraneproteins. Here we present a rapid and efficientfluorescence-detection size-exclusion chromatog-raphy-based thermostability assay (FSEC-TS). Inthis method, the target protein is fused to GFP.Heated protein samples, treated with a panel of addi-tives, are then analyzed by FSEC. FSEC-TS allowsone to evaluate the thermostability of nanogram-to-microgram amounts of the target protein under avariety of conditions without purification. We appliedthis method to the Danio rerio P2X4 receptor andCaenorhabditis elegans GluCl to screen ligands,ions, and lipids, including newly designed choles-terol derivatives. In the case of GluCl, the screeningresults were used to obtain crystals of the receptorin the presence of lipids.
INTRODUCTION
Growth of membrane protein crystals for the purpose of deter-
mining the underlying atomic structure traditionally requires
a labor-intensive screening approach to identify conditions
that stabilize the protein of interest. This ‘‘precrystallization’’
screening is usually performed with purified protein, which pla-
ces a requirement for large amounts of purematerial and, neces-
sarily, the ability to purify the protein of interest. Membrane
proteins, particularly those from eukaryotes, are typically ex-
pressed at low levels and are often unstable during and after
purification. The challenge of eukaryotic membrane protein crys-
tallography bears itself out in the Protein Data Bank, where <1%
of the deposited structures are from this elusive category. The
degree to which one can identify, in advance of protein purifica-
tion, conditions and compounds that stabilize a given protein
candidate, the higher the likelihood that protein purification will
Structure 20, 1293
proceed smoothly and that well-diffracting crystals will be
obtained. Atomic structures of membrane proteins are of
tremendous interest from both basic science and drug develop-
ment perspectives; relatively little structural information exists
for this class of proteins, molecular mechanisms of signal trans-
duction across cell membranes are not well understood, and
drug design for targets of many currently prescribed therapeu-
tics (Zambrowicz and Sands, 2003) is rendered relatively blind.
One useful technique for precrystallization screening of mem-
brane proteins is fluorescence-detection size-exclusion chro-
matography (FSEC) (Kawate and Gouaux, 2006). In this method,
the target protein is expressed as a GFP fusion, and the SEC
profile of the resulting fusion protein is monitored by fluores-
cence spectroscopy. This simple method allows for rapid evalu-
ation of protein expression level andmonodispersity of the target
protein using nanograms to micrograms of unpurified material
fromwhole-cell extracts. This method has proved to be a power-
ful tool for screening panels of membrane protein orthologs,
detergents, and conformationally sensitive antibodies, and
thus has led to the recent determination of a number of mem-
brane protein structures from this laboratory, including LeuT,
ApcT, ASIC, P2X, GluA2, and GluCl (Hattori and Gouaux, 2012;
Hibbs and Gouaux, 2011; Jasti et al., 2007; Kawate et al.,
2009; Krishnamurthy and Gouaux, 2012; Shaffer et al., 2009;
Sobolevsky et al., 2009; Yamashita et al., 2005).
There are several methods to evaluate protein stability that
have already been described, but these techniques often require
large amounts of pure protein and, therefore, are unsuitable for
high-throughput screening of eukaryotic membrane proteins.
Fluorescent dye-based thermal stability assays using a thiol-
reactive fluorophore N-[4-(7-diethylamino-4-methyl-3-coumar-
inyl)phenyl]maleimide (CPM) (Alexandrov et al., 2008) have
been developed to evaluate the thermostability of membrane
proteins in a high-throughput manner. Although the method
requires only microgram samples of protein, because of the
high sensitivity of the CPM dye, it requires free cysteine residues
embedded in the protein core. A thermostability assay that relies
upon analytical size-exclusion chromatography was used effec-
tively to screen compounds and conditions for stabilizing
membrane proteins for crystallization (Czyzewski and Wang,
2012; Mancusso et al., 2011), but this latter method requires
a relatively large amount of purified sample because of the low
sensitivity of UV absorbance.
–1299, August 8, 2012 ª2012 Elsevier Ltd All rights reserved 1293
Fluorometer
SECcolumn
Proteinsample
Thermalcycler
Fractioncollector
Fluorescence
Void
Oligomer
Protomer
Figure 1. Flow Chart of FSEC-TS
The protein sample, after heat treatment and centrifugation, is loaded onto
a SEC column, connected to a fluorescence detector to monitor GFP or
tryptophan fluorescence. The panel labeled ‘‘Fluorescence’’ shows a hypo-
thetical elution profile from GFP or tryptophan fluorescence.
Structure
FSEC-Based Thermostability Assay
In the current study, we present a FSEC-based thermosta-
bility assay (FSEC-TS). In this method, nanogram-to-microgram
quantities of purified or unpurified proteins are incubated over
a range of temperatures and then are applied to a SEC column
in line with a fluorescence detector to monitor GFP or tryptophan
fluorescence. The results from FSEC-TS provide a denaturation
or ‘‘melting’’ temperature (Tm), which is used as a reference point
to test the degree of protein thermostabilization by a panel of
small molecules. We applied this simple and rapid method to
the D. rerio P2X4 receptor and the C. elegans GluCl channel to
screen ligands, divalent cations, and lipids, including new
synthetic cholesterol derivatives. We find that thermostabiliza-
tion of both proteins by different compounds is not qualitatively
altered by the presence of GFP and that, in proof-of-principle
experiments, Tm by FSEC-TS agrees well with Tm determined
by a radioligand binding assay. The results are useful in identi-
fying compounds for stabilizing the proteins during purification
and in crystallization and, in the case of GluCl, have been used
to obtain a distinct, lipid-dependent crystal form.
RESULTS
Validation of FSEC-TSTo validate our assay, we used three proteins: EGFP, D. rerio
P2X4 (zfP2X4) receptor, and C. elegans glutamate-gated
chloride channel a (GluCl). Protein samples were incubated
over a range of temperatures for 10 min using a thermal cycler
(Figure 1), followed by ultracentrifugation to remove precipitated
material. The supernatant was then applied to a SEC column
attached to a fluorescence detector to monitor GFP or trypto-
1294 Structure 20, 1293–1299, August 8, 2012 ª2012 Elsevier Ltd Al
phan fluorescence (Figure 1). In these experiments, we used
detergents that we knew from previous experience would stabi-
lize the native oligomeric state of the receptors.
As shown in Figure 2A, peak profiles from EGFP samples
incubated between 4�C and 65�C were sharp and symmetrical,
indicating that the protein remained monodisperse. In contrast,
a clear thermal shift of the peak profiles was observed from the
samples incubated at temperatures near to and above 70�C(Figure 2A). Aggregation and precipitation after heating resulted
in a significant decrease in the peak height, and a melting curve,
from the fluorescence signal intensity at the respective temper-
atures, was observed (Figure 2B). The calculated Tm is 76�C,which is consistent with the reported melting temperature for
GFP (76�C), the temperature where one half of the intrinsic
fluorescence is lost (Bokman and Ward, 1981; Ward, 1981).
P2X receptors are nonselective cation channels gated by
extracellular ATP and are implicated in diverse physiological
processes, such as synaptic transmission, inflammation, and
taste and pain sensing (Jarvis and Khakh, 2009; North, 2002;
Surprenant and North, 2009). Seven P2X receptor subtypes
assembled into either homomeric or heteromeric trimers (Jarvis
and Khakh, 2009; North, 2002). Recently, our group reported the
crystal structure of zfP2X4 in an apo, closed-channel conforma-
tion (Kawate et al., 2009). Precrystallization screening of P2X
orthologs by FSEC had identified zfP2X4 as a promising target
for the crystallization trials.
The well-behaved, crystallized construct of an N-terminal
EGFP fusion to zebrafish P2X4 (DP2X4-A) (Kawate et al., 2009)
was expressed with an octa-histidine affinity tag and purified
for the assay (see Figure S1A available online). FSEC peak
profiles from the EGFP-tagged DP2X4-A samples incubated
between 4�C and 40�C were sharp and symmetrical (Figure 2C);
these peaks correspond to the functional P2X receptor trimer. A
clear thermal shift of the peak profiles was observed from the
samples incubated around 50�C (Figure 2C). Protein denatur-
ation led to a significant decrease in the trimer peak height and
concomitant increase in the putative monomer peak height. A
clear melting curve from the trimer peak height at the respective
temperatures was observed (Figure 2D). The calculated Tm is
49�C, on the basis of the trimer peak height.
To verify that the result from FSEC-TS is relevant to binding-
competent receptor, we measured the ATP binding activity of
the heat-treated samples (Figure S1B). The experiment was
performed at 100 nM ATP concentration, approximately 5-fold
above the Kd value of 19.6 ± 3.8 nM (Figure S1C). The calculated
Tm is 47.8�C ± 0.3�C, consistent with FSEC-TS (49�C). To assess
the effect of theGFP-tag on Tm, we also obtained amelting curve
from EGFP-free DP2X4-A samples by monitoring tryptophan
fluorescence from the endogenous Trp residues in DP2X4-A
(Figure 2D); the estimated Tm is 50�C, not significantly different
from the Tm for EGFP-tagged DP2X4-A (49�C).GluCl is a pentameric chloride channel gated by glutamate
and ivermectin (Cully et al., 1994) and is a member of the Cys-
loop receptor superfamily (Thompson et al., 2010). This receptor
family also includes the ion channels activated by acetylcholine,
serotonin, g-aminobutyric acid, and glycine. GluCl was identified
as a promising target for crystallization by FSEC screening of
Cys-loop receptor orthologs, which subsequently led to the
determination of the structure of a GluCl-Fab complex with the
l rights reserved
0 10 20 30 40 50 60 70 80
1.0
0.8
0.6
0.4
0.2
0.0
GluClcryst
GluClcryst
+IVM
GluCl-EGFP (wholecell)
GluCl-EGFP (wholecell)+IVM
90
F
Temperature (°C)
No
rma
lize
d F
l sig
na
l4 °C40 °C50 °C55 °C60 °C65 °C70 °C80 °C
Seconds0 100 200 300 400 500 600 700 800 900
0
40
80
120
160
200
240
Flu
ore
scence (
10
3 c
ps)
Oligomer
Void
E
Temperature (°C)
EGFP-zfP2X4AEGFP-zfP2X4A+ATPEGFP-zfP2X4A (wholecell)EGFP-zfP2X4A (wholecell)+ATPnon-EGFP-zfP2X4A
0 10 20 30 40 50 60 70 80 90
No
rma
lize
d F
l sig
na
l
1.0
0.8
0.6
0.4
0.2
0.0
D
Seconds0 500 1000 1500 2000 2500 3000
Flu
ore
sce
nce
(1
03 c
ps)
0
40
80
120
160
200
2404 °C
20 °C35 °C40 °C45 °C50 °C55 °C80 °C
C
VoideGFP
Oligomer
Monomer
Temperature (°C)
eGFP
0 10 20 30 40 50 60 70 80 90
No
rma
lize
d F
l sig
na
l
1.0
0.8
0.6
0.4
0.2
0.0
Seconds0 500 1000 1500 2000 2500 3000
0
50
100
150
200
250
300
350
400
90 °C
20 °C4 °C
65 °C70 °C75 °C80 °C
Flu
ore
sce
nce
(1
03 c
ps)
BA
Figure 2. Thermostability Profiles of zfP2X4 and GluCl
(A, C, and E) Representative FSEC profiles from EGFP (A), EGFP-DP2X4-A (C), and GluClcryst (E), heated at the respective temperatures, detected by GFP
fluorescence (A and C) or by Trp fluorescence (E).
(B, D, and F) Representativemelting curves of EGFP (B), EGFP-tagged and EGFP-freeDP2X4-A (D), andGluClcryst andGluCl-EGFP (F). Melting temperatures (Tm)
were determined by fitting the curves to a sigmoidal dose-response equation. The fitted curves are shown as a black line.
See also Figure S1.
Structure
FSEC-Based Thermostability Assay
allosteric agonist ivermectin bound and the channel in an open
conformation (Hibbs and Gouaux, 2011).
We carried out FSEC-TS with GluCl to test the validity of the
assay and to identify compounds for stabilizing the receptor in
a resting, closed channel state. The crystallized construct of
EGFP-free GluCl (GluClcryst) was purified (Figure S1D) and incu-
bated at 4–80�C and applied to a SEC column, with the eluted
material monitored by tryptophan fluorescence (Figure 2E). We
used the TSKgel SuperSW2000 column for GluCl, which requires
only 15 min per sample, to test many different lipid combinations
in a high-throughput manner, as described later. A melting curve
based on the peak height of the pentameric species was
observed (Figure 2F), with a calculated Tm of a remarkably high
Structure 20, 1293
59�C. Determination of Tm based on a radioligand binding exper-
iment using 3H-ivermectin yields a similar melting temperature
of 58�C (Figure S1E).
The FSEC-TS results from EGFP, zfP2X4, and GluCl demon-
strate that a combination of heat treatment and FSEC can be
applied to monitor the temperature-induced unfolding of soluble
and membrane proteins by detecting GFP fluorescence from
GFP-tagged proteins or by measuring tryptophan fluorescence
from endogenous Trp residues.
Evaluation of Ligands for Cocrystallization by FSEC-TSTo test whether FSEC-TS can be used to estimate the effect of
ligands on membrane proteins for crystallization, we evaluated
–1299, August 8, 2012 ª2012 Elsevier Ltd All rights reserved 1295
B
C
D
A
No
rma
lize
d F
l sig
na
l (%
)
)C°
4( lort
no
C
)C°
05( l
ortn
oC
lC
gM
2 lC
aC
2 lC
nZ
2 PT
A
PT
A-P
NT
SD
AP
P
CP
MD
CP
PD
CP
OP
CP
OD
EP
OP
P)4(I
P
SH
C
CP-
HC
edi
sotl
am-
HC
GE
H-H
C
niel
oo
no
M
100
75
50
25
0
Figure 3. Effects of Additives on the Thermostability of zfP2X4
(A) The normalized peak heights from EGFP-fusion DP2X4-A, heat-treated at
50�C for 10 min, in the presence of the respective additives. The peak heights
were normalized to that from the sample at 4�C without any additive.
(B–D) Chemical structures of novel cholesterol derivatives, hexaethyleneglycol
hydroxyethyl cholesterol (B) maltoside hydroxyethyl cholesterol (C), and
phosphocholine hydroxyethyl cholesterol (D), employed in the FSEC-TS
studies are shown. Structures were created using the ChemDraw (Cambridge
Scientific Computing).
See also Figure S2 and Table S1.
Structure
FSEC-Based Thermostability Assay
the thermostabilizing effects of ATP on zfP2X4 and of ivermectin
on GluClcryst. Both ligands bind to their cognate receptors with
high affinity, and, thus, we can use them to examine whether
we can pick up stabilizing effects of tightly bound ligands. In
FSEC-TS, both ATP and ivermectin had dramatic stabilizing
effects, increasing the calculated Tm by 21�C for DP2X4-A and
by 22�C for GluCl (Figures 2D and 2F, respectively). In accor-
dance with these results, crystallization of zfP2X4 with ATP
yields crystals that diffract to 2.8 A resolution (Hattori and
Gouaux, 2012), and crystals of the GluCl-ivermectin complex
diffract to 3.3 A resolution (Hibbs and Gouaux, 2011).
FSEC-TS with Unpurified ProteinOne of the advantages of FSEC using GFP-tagged protein is that
unpurified protein can be analyzed because of the wide spectral
separation between GFP fluorescence and the fluorescence
intrinsic to most proteins. To test whether unpurified protein
can be analyzed using FSEC-TS, EGFP-tagged DP2X4-A and
GluCl obtained from whole-cell lysates were heat-treated in the
presence and absence of ATP and ivermectin, respectively,
and were applied to a SEC column after centrifugation.
1296 Structure 20, 1293–1299, August 8, 2012 ª2012 Elsevier Ltd Al
For EGFP-DP2X4-A, the calculated Tm values in the absence
and presence of ATP are 52�C and 69�C, respectively (Fig-
ure 2D), consistent with the Tm values from the purified EGFP-
fusion DP2X4-A, which are 49�C in the absence of ATP and
70�C in the presence of ATP.
For GluCl-EGFP, the apparent Tm values in the absence and
presence of ivermectin are 41�C and 53�C (Figure 2F), respec-
tively. These values are significantly different from the apparent
Tm values from the purified GluClcryst (58�C in the absence of
ivermectin and 81�C in the presence of ivermectin). There are
two likely sources of the difference in Tm between GluCl-EGFP
andGluClcryst. First, unlike in zfP2X4, wherein EGFP is positioned
at the receptor terminus, in GluCl, the EGFP is inserted into an
internal loop, between the third and fourth transmembrane
helices. We suggest that insertion of GFP into this loop destabi-
lizes the protein. Second, GluCl-EGFP contains the full-length,
wild-type sequence of the GluCl gene, whereas GluClcryst is
a truncated construct used for crystallization. Regardless of
the differences in Tm, however, the qualitative effect of iver-
mectin on GluCl thermostability remains unchanged. Taken
together, we conclude that FSEC-TS can be used to analyze
unpurified GFP-tagged protein.
Effect of Divalent Cations on zfP2X4The effects of three different divalent cations (Ca2+, Mg2+, and
Zn2+) on zfP2X4 thermostability were tested by FSEC-TS (Fig-
ure 3A). EGFP-tagged DP2X4-A samples in a series of buffers
containing the respective additives were heat-treated at 50�Cfor 10 min, which is close to the calculated Tm of apo protein.
As in the case for other P2X receptors, ATP-responses at P2X4
receptors can be modulated by divalent cations, such as Zn2+
(Garcia-Guzman et al., 1997; Wildman et al., 1999) and Mg2+
ions (Negulyaev and Markwardt, 2000). The significant divalent
cation-induced stabilization of the trimer peak is observed in
the presence of all tested divalent cation (Figure 3A). Consis-
tently, best crystals of zfP2X4 in an apo, closed state (Kawate
et al., 2009), as well as crystals in the presence of ATP (Hattori
and Gouaux, 2012) were obtained in presence of Mg2+.
Effect of Ligands on zfP2X4The effects of different P2X ligands, 20,30-O-(2,4,6-trinitrophenyl)
adenosine50-triphosphate (TNP-ATP) (antagonist) andpyridoxal-
phosphate-6-azophenyl-20,4’-disulphonic acid (PPADS) (antag-
onist) on the thermostability of GFP-fusion DP2X4-A were tested
by FSEC-TS (Figure 3A). TNP-ATP stabilized the zfP2X4 trimer to
a similarly strong extent as was observed with ATP (Figure 3A).
PPADS, by contrast, destabilized theprotein (Figure 3A). In corre-
lation to the effects of ligands on thermostability, we have
successfully crystallized zfP2X4 in the presence of TNP-ATP
but have not been able to produce crystals with PPADS.
Effect of Lipids on zfP2X4Purification of membrane proteins from their native environment
requires the use of detergents. However, detergent solubilization
of membrane proteins often leads to aggregation, because
membrane proteins tend to be less stable in detergent micelles
than in the lipid bilayer. To remedy this behavior, the addition
of lipids during purification or crystallization can be used to
improve membrane protein behavior for the purposes of
l rights reserved
No
rma
lize
d F
l sig
na
l (%
)
)C°
4(l
ortn
oC
100
75
50
25
0
)C°
75(
lort
no
C
tc
artx
edi
pill
atot
na
eb
yo
S
)1:
1:3(
EP
OP
+G
PO
P+
CP
OP
dipi
Lr
alo
Pni
arB
)1:
2(E
PO
P:C
PO
P
)1:
2(G
PO
P+
CP
OP
)1:
2(S
PO
P+
CP
OP
CP
OP
EP
OP
nile
ym
og
nih
ps
niar
B
)1:
1(C
PO
P+
nile
ym
og
nih
pS
SP
OP
)1:
1(E
PO
P+
nile
ym
og
nih
pS
)1:
1(S
HC
+nil
ey
mo
gni
hp
S
GP
OP
)1:
2(E
PO
P+
SH
C
)1:
1(C
P4
1C
m+
SH
C
)1:
2(e
dim
are
cni
arB
+S
HC
SH
C
edi
sotl
am-
HC
edi
mar
ec
niar
B
)1:
1(E
PO
P+
CP
MD
CP
MD
EP
MD
Figure 4. Effects of Lipids on the Thermostability of GluCl
The normalized peak heights fromGluClcryst heat-treated at 57�C for 10min, in
the presence of the respective lipids. The peak heights were normalized to that
from the sample at 4�C without any additive. See also Figure S3 and Table S2.
Structure
FSEC-Based Thermostability Assay
biophysical characterization, crystallization, and structure deter-
mination (Guan et al., 2006; Hattori et al., 2007; Hibbs and
Gouaux, 2011; Krishnamurthy and Gouaux, 2012; Lemieux
et al., 2003; Sobolevsky et al., 2009; Zhang et al., 2003).
The effects of ten different lipids on the thermostability of
GFP-fusion DP2X4-A were examined by FSEC-TS (Figure 3A)
and also by ATP binding assay (Figure S2). Among the tested
lipids, phosphocholine hydroxyethyl cholesterol (CH-PC), malto-
side hydroxyethyl cholesterol (CH-maltoside), and a hexaethyle-
neglycol hydroxyethyl cholesterol (CH-HEG) are newly designed
cholesterol derivatives (Figures 3B–3D). Hydroxyl groups of
cholesterol are connected to phosphocholine (CH-PC), maltose
(CH-maltoside), or hexaethyleneglycol (CH-HEG) via a hydrox-
yethyl group. Cholesteryl hemisuccinate (CHS), as well as
cholesterol, have been successfully used for purification and
crystallization of GPCRs (Cherezov et al., 2007; Rasmussen
et al., 2011; Rosenbaum et al., 2011; Soriano et al., 2009).
However, the ester group of CHS can be hydrolysis sensitive,
and cholesterol itself is sparingly soluble. Our cholesterol deriv-
atives are more stable because their head groups are connected
to hydroxyl groups of cholesterol via a hydroxyethyl group. Addi-
tionally, the new head groups of the derivativesmight have stabi-
lizing effects on membrane proteins. Therefore, we tested these
cholesterol derivatives as well as other lipids by FSEC-TS.
Surprisingly, most of the lipids except for the phosphoinosi-
tide, PI(4)P, significantly stabilized the trimer peak of zfP2X4 in
FSEC (Figure 3A), which is largely consistent with thermostabili-
zation observed by the same compounds in a radioligand
binding assay (Figure S2). Among phosphocholine-containing
lipids, the lipids with longer alkyl chains more effectively stabi-
lized the trimer peak (Figure 3A). We also tested one of the phos-
phoglycerol-containing lipids, POPE (Figure 3A). Although the
chain length of POPE is the same as that of POPC, POPC stabi-
lized the trimer peak of zfP2X4 more effectively than POPE did.
Among cholesterol derivatives, CH-HEG was the poorest
additive in terms of both FSEC-TS and the radioligand binding
assay (Figure 3A and Figure S2). On the basis of FSEC-TS,
CHS was slightly better than CH-PC and CH-maltoside, but on
the basis of the binding assay, CH-PC had the strongest stabi-
lizing effect among the tested lipids (Figure 3A and Figure S2).
We also tested monoolein (MO), which has been used for
membrane protein crystallization in the lipidic cubic phase
(Johansson et al., 2009). MO also stabilized the trimer peak of
zfP2X4 by FSEC-TS but was not better than other ‘‘good’’ lipids
such as POPC (Figure 3A).
Among all tested lipids, only PI(4)P totally destabilized the
trimer peak of zfP2X4 by FSEC-TS (Figure 3A), and consistently
in the presence of this lipid the receptor showed almost no ATP
binding activity (Figure S2). Interestingly, phosphoinositides are
known to regulate the activity of P2X4 receptors through direct
interactions (Bernier et al., 2008).
Effect of Lipids on GluClWe previously found that either POPC or DPPC was required for
growth of well-diffracting crystals of the GluCl-Fab-ivermectin
complex (Hibbs and Gouaux, 2011). To screen lipids for
a more stabilizing effect on GluCl, we tested 24 different
lipid combinations using FSEC-TS with pure protein (Figure 4).
GluClcryst samples in a series of buffers containing the respective
Structure 20, 1293
lipids were heat-treated for 10 min at 57�C, which is close to the
calculated Tm of apo protein. Most lipids had some significant
stabilizing effect on GluCl, whereas total soy lipid, brain ceram-
ide, and DMPE did not stabilize GluCl (Figure 4). Among tested
lipids, sphingomyelin and PO-containing lipids consistently had
strong stabilizing effects, and the addition of CHS often further
increased thermostability. The lipids identified in this experiment
have been fed into crystallization trials of GluClcryst with the goal
of obtaining well-diffracting crystals in an ivermectin free condi-
tion, and a new, to our knowledge lipid-dependent crystal form
of GluClcryst has been obtained (Figure S3A). These crystals
are from theC2 space group, currently diffract X-rays to between
3 and 4 A (Figure S3B), and are a promising lead in obtaining the
structure of a closed-channel conformation of a eukaryotic
Cys-loop receptor.
ConclusionsIn this study, we applied FSEC-TS to zfP2X4 and GluCl. This
method carries over the advantages of conventional FSEC: it
requires only nanogram-to-microgram quantities of protein and
allows for screening of different stabilizing conditions in an
efficient manner with both purified and unpurified protein. Using
this method, we screened various conditions, such as ligands
and lipids, to stabilize zfP2X4 and GluCl. Ligand screening
demonstrated that ATP, a biological agonist of P2X, and its
synthetic analog, TNP-ATP, dramatically stabilized zfP2X4,
suggesting that these are suitable ligands for cocrystallization.
Lipid screening showed that our designed cholesterol deriva-
tives and many other lipids significantly stabilized zfP2X4 or
GluCl (Figures 3 and 4). Employing the ligands and lipids identi-
fied in this study has led to a distinct well-diffracting crystal
form of GluCl.
EXPERIMENTAL PROCEDURES
Expression and Purification
The crystallized construct of zfP2X4 (DP2X4-A) was expressed as an
N-terminal EGFP fusion and was purified as described previously (Kawate
–1299, August 8, 2012 ª2012 Elsevier Ltd All rights reserved 1297
Structure
FSEC-Based Thermostability Assay
et al., 2009). The EGFP-tagged and EGFP-free purified proteins were concen-
trated and dialyzed overnight at 4�C against three changes of buffer I (20 mM
HEPES-NaOH [pH 7.0], 80mMNaCl, 20mMKCl, and 2mMcymal-6), and then
were stored at �80�C before use. The crystallized construct of C. elegans
GluCl (GluClcryst)) with a C-terminal 8x-His tag was expressed in Sf9 cells
and purified as described previously (Hibbs and Gouaux, 2011), omitting the
addition of Fab before the SEC purification step.
For GluCl experiments in crude cell extracts, a fusion of GluCl with EGFP
was made with GFP inserted into the full-length, wild-type C. elegans GluCla
receptor between Asn392 and Ser393 in the mature protein sequence. Sf9
cells at a density of 3–4 3 106 cells per ml were infected with baculovirus en-
coding GluCl-EGFP and were harvested 72 hr later by centrifugation; cell
pellets were stored at �80�C.
Radioligand Saturation Binding Assay for zfP2X4
Measurements of total ATP binding was obtained by adding GFP-fusion
DP2X4-A to a final concentration of 30 nM in 250 ml buffer I containing
0–1000 nM radiolabeled ATP where the hot ATP was diluted with cold ATP
in a ratio of 1:4, yielding a final specific activity of 6 Ci/mmol. Reactions
were incubated at 4�C overnight and then were terminated by filtering through
GSWP 02500 nitrocellulose membranes pre-equilibrated with buffer I contain-
ing 100 mM cold ATP. Filters were washed three times with 2 ml of buffer I,
placed in 6 ml of Ultima Gold scintillation cocktail, and counted after 1 hr. Esti-
mates of nonspecific binding were obtained by reactions in the presence of
100 mM cold ATP. The experiment was performed in triplicate, and the data
were fit to a rectangular hyperbola using the GraphPad Prism4 program.
FSEC-TS with Purified Proteins
For EGFP, 500 ng of EGFP in 50 ml of buffer II (20 mM Tris-HCl [pH 8.0]
and 150 mM NaCl) were incubated at 4�–90� for 10 min, using a thermal
cycler and then were centrifuged at 87,000 3 g for 20 min. A fraction of the
supernatant (25 ml) was loaded onto a SEC column (Superose 6 10/300 GL,
GE Healthcare) pre-equilibrated with buffer II and run at a flow rate of
0.5 ml/min.
For zfP2X4, 1 mg of GFP-fusion andGFP-freeDP2X4-A in 50 ml of buffer I was
incubated at 4–80�C and 4–70�C for 10min and was centrifuged at 87,0003 g
for 20 min. Twenty-five microliters of the supernatant were loaded onto a SEC
column (Superose 6 10/300 GL) pre-equilibratedwith buffer III (20mMHEPES-
NaOH [pH 7.0], 80 mMNaCl, 20 mM KCl, and 1 mM n-dodecyl b-D-maltoside,
C12M) and run at a flow rate of 0.5 ml/min.
For GluCl, 2 mg of GFP-free GluClcryst in 100 ml of buffer IV (20 mM Tris-
HCl [pH 7.4], 150 mM NaCl, and 2 mM C12M) were incubated at 4–80�Cfor 10 min, and centrifuged as described above. Twenty-five microliters
of the supernatant were loaded onto a TSKgel SuperSW2000 column
(TOSOH Bioscience), pre-equilibrated with buffer IV, and run at a flow rate of
0.35 ml/min.
The eluent from the SEC columnwas passed through a fluorometer (Kawate
and Gouaux, 2006) (lexc of 480 nm and lem of 510 nm for GFP fluorescence;
lexc of 280 nm and lem of 325 nm for tryptophan fluorescence). The peak
heights from EGFP (Figure 2B) or oligomers (Figure 2D and 2F) were normal-
ized to that from samples incubated at 4�C, and were fit to the Hill equation
using the GraphPad Prism 4 program. Melting temperatures (Tm) were deter-
mined by fitting the curves to a sigmoidal dose-response equation.
Gaussian fitting was performed using PeakFit software (SeaSolve Software,
Inc) as described previously (Kawate and Gouaux, 2006). Initial peak detection
and fitting were performed with the AutoFit peaks I Residuals option. The
obtained Gaussian functions were further refined and fitted to the original
FSEC profile using a least-squares minimization procedure (Figure S1F). The
obtained melting curves based on the peak height are consistent with the
melting curves based on the peak area of EGFP (Figure S1G), zfP2X4 trimer
(Figure S1H) or GluCl pentamer (Figure S1I) estimated by Gaussian fitting.
To test ATP and ivermectin effects, GFP-fusion DP2X4-A in the presence of
1 mM ATP and EGFP-free GluClcryst in the presence of 10 mM ivermectin were
incubated at 4�C for 1 hr and 30 min before heat treatment, respectively.
FSEC-TS with Whole-Cell Lysates
The GFP-fusion of DP2X4-A was expressed in Sf9 insect cells as described
elsewhere (Kawate et al., 2009), and the harvested cells were stored
1298 Structure 20, 1293–1299, August 8, 2012 ª2012 Elsevier Ltd Al
at �80�C before use. Sf9 cells expressing GFP-fusion DP2X4-A from 7.5 ml
culture were resuspended in 1 ml of buffer V (50 mM Tris-HCl [pH 8.0] and
150 mM NaCl) containing 40 mM C12M in the presence and absence of
1 mM ATP, rotated at 4�C for 1 hr, and then centrifuged at 87,000 3 g for
40 min. One hundred microliters of the supernatant was dispensed into thin-
walled PCR tubes and was incubated at the respective temperatures for
10 min, using a thermal cycler. After centrifuge at 87,000 3 g for 20 min,
50 ml of the supernatant was applied to a Superose 6 column, pre-equilibrated
with buffer V containing 1 mM C12M.
Sf9 cell pellets from 30ml culture containing GluCl-EGFP were resuspended
in 3ml of buffer IV containing 40mMC12M, rotated at 4�C for 40min, and then
centrifuged at 87,000 3 g for 40 min. The supernatant was divided into two
tubes, and ivermectin (100 mM final concentration from a 10 mM stock in
DMSO) was added to one of the tubes, and was rotated at 4�C for 10 min.
One hundred-microliter aliquots of the supernatant were dispensed into thin-
walled PCR tubes and were incubated at the respective temperatures for
10 min using a thermal cycler. After centrifugation at 87,000 3 g for 20 min,
20 ml of the supernatant was applied to a TSKgel SuperSW2000 column,
pre-equilibrated with buffer IV.
The eluent from the SEC column was passed through a fluorometer (Kawate
and Gouaux, 2006) (excitation, 480 nm; emission, 510 nm for GFP fluores-
cence). Data were normalized and fit to the Hill equation, as described above.
FSEC-TS Screening of Additives
To determine the effect of lipids on the stability of zfP2X4, 5 mg of GFP-fusion
DP2X4-A in 250 ml of buffer I were rotated at 4�C for 1 hr after addition of 2.5 ml
of a 1003 lipid suspension (20% DMSO and 80% buffer I). The mixture was
next incubated at 50�C for 10min and then centrifuged. Twenty-fivemicroliters
of the sample was used for FSEC as described above. The rest of the sample
was stored at 4�C for use in binding assays; 1003 additive solutions were
added to GFP-fusion DP2X4-A at the respective final concentration shown
in Table S1.
For GluCl, 850 ng of GluClcryst in 10 ml of buffer IV were rotated at 4�C for
40 min after the addition of 90 ml of lipid stock solution in buffer IV containing
40 mM C12M. The mixture was next incubated at 59�C for 10 min and then
centrifuged. A 25 ml aliquot of the sample was used for FSEC as described
above. Lipids were added to GluClcryst at the respective final concentrations
as shown in Table S2.
Binding Assay of Heat-Treated DP2X4-A
These experiments were performed as described above except that the
binding was initiated by adding heat-treated GFP-fusion DP2X4-A to a
final concentration of 30 nM in 250 ml of buffer I containing 100 nM ATP
(1:4 3H:1H; 6 Ci/mmol final specific activity). The entire experiment was
performed in triplicate.
Binding Assay of Heat-Treated GluClcrystBinding of 3H-ivermectin to GluClcryst after heat treatment was performed
using a scintillation proximity assay (SPA). A solution of pure GluClcryst was
made in buffer IV at a subunit concentration of 200 nM, and 70 ml were aliquot-
ted into thin-walled PCR tubes and incubated for 10 min at temperatures
ranging from 4�C to 80�C. A suspension was made containing 2 mg/ml
SPA beads (YSi copper his tag, GE Healthcare) and 100 nM ivermectin
(1:9 3H:1H; 5 Ci/mmol final specific activity; 3H-ivermectin purchased from
American Radiolabeled Chemicals and 1H-ivermectin purchased from Sigma)
in buffer IV. The SPA suspension was mixed 1:1 (50 ml to 50 ml) with the protein
solution by pipetting in a 96-well microtiter plate, and the cpm data were
collected after 18 hr of incubation at room temperature.
SUPPLEMENTAL INFORMATION
Supplemental Information includes three figures and two tables and can be
found with this article online at http://dx.doi.org/10.1016/j.str.2012.06.009.
ACKNOWLEDGMENTS
We thank A. Penmatsa for providing the lipid screening kit for GluCl, J. Michel
for technical help, and L. Vaskalis for assistance with figures. This work was
l rights reserved
Structure
FSEC-Based Thermostability Assay
supported by the Uehara Memorial Foundation (M.H.), the American Asth
Foundation (E.G.), and the National Institutes of Health (R.H. and E.G.). E.G.
is an investigator with the Howard Hughes Medical Institute.
Received: April 6, 2012
Revised: June 7, 2012
Accepted: June 13, 2012
Published: August 7, 2012
REFERENCES
Alexandrov, A.I., Mileni, M., Chien, E.Y., Hanson, M.A., and Stevens, R.C.
(2008). Microscale fluorescent thermal stability assay for membrane proteins.
Structure 16, 351–359.
Bernier, L.P., Ase, A.R., Chevallier, S., Blais, D., Zhao, Q., Boue-Grabot, E.,
Logothetis, D., and Seguela, P. (2008). Phosphoinositides regulate
P2X4 ATP-gated channels through direct interactions. J. Neurosci. 28,
12938–12945.
Bokman, S.H., and Ward, W.W. (1981). Renaturation of Aequorea green-
fluorescent protein. Biochem. Biophys. Res. Commun. 101, 1372–1380.
Cherezov, V., Rosenbaum, D.M., Hanson, M.A., Rasmussen, S.G., Thian, F.S.,
Kobilka, T.S., Choi, H.J., Kuhn, P., Weis, W.I., Kobilka, B.K., and Stevens, R.C.
(2007). High-resolution crystal structure of an engineered human beta2-
adrenergic G protein-coupled receptor. Science 318, 1258–1265.
Cully, D.F., Vassilatis, D.K., Liu, K.K., Paress, P.S., Van der Ploeg, L.H.,
Schaeffer, J.M., and Arena, J.P. (1994). Cloning of an avermectin-sensitive
glutamate-gated chloride channel from Caenorhabditis elegans. Nature 371,
707–711.
Czyzewski, B.K., and Wang, D.N. (2012). Identification and characterization of
a bacterial hydrosulphide ion channel. Nature 483, 494–497.
Garcia-Guzman, M., Soto, F., Gomez-Hernandez, J.M., Lund, P.E., and
Stuhmer, W. (1997). Characterization of recombinant human P2X4 receptor
reveals pharmacological differences to the rat homologue. Mol. Pharmacol.
51, 109–118.
Guan, L., Smirnova, I.N., Verner, G., Nagamori, S., and Kaback, H.R. (2006).
Manipulating phospholipids for crystallization of a membrane transport
protein. Proc. Natl. Acad. Sci. USA 103, 1723–1726.
Hattori, M., and Gouaux, E. (2012). Molecular mechanism of ATP binding and
ion channel activation in P2X receptors. Nature 485, 207–212.
Hattori, M., Tanaka, Y., Fukai, S., Ishitani, R., and Nureki, O. (2007). Crystal
structure of the MgtE Mg2+ transporter. Nature 448, 1072–1075.
Hibbs, R.E., and Gouaux, E. (2011). Principles of activation and permeation
in an anion-selective Cys-loop receptor. Nature 474, 54–60.
Jarvis, M.F., and Khakh, B.S. (2009). ATP-gated P2X cation-channels.
Neuropharmacology 56, 208–215.
Jasti, J., Furukawa, H., Gonzales, E.B., and Gouaux, E. (2007). Structure of
acid-sensing ion channel 1 at 1.9 A resolution and low pH. Nature 449,
316–323.
Johansson, L.C., Wohri, A.B., Katona, G., Engstrom, S., and Neutze, R. (2009).
Membrane protein crystallization from lipidic phases. Curr. Opin. Struct. Biol.
19, 372–378.
Kawate, T., and Gouaux, E. (2006). Fluorescence-detection size-exclusion
chromatography for precrystallization screening of integral membrane
proteins. Structure 14, 673–681.
Kawate, T., Michel, J.C., Birdsong, W.T., and Gouaux, E. (2009). Crystal struc-
ture of the ATP-gated P2X(4) ion channel in the closed state. Nature 460,
592–598.
Structure 20, 1293
Krishnamurthy, H., and Gouaux, E. (2012). X-ray structures of LeuT in
substrate-free outward-open and apo inward-open states. Nature 481,
469–474.
Lemieux, M.J., Song, J., Kim, M.J., Huang, Y., Villa, A., Auer, M., Li, X.D., and
Wang, D.N. (2003). Three-dimensional crystallization of the Escherichia coli
glycerol-3-phosphate transporter: a member of the major facilitator super-
family. Protein Sci. 12, 2748–2756.
Mancusso, R., Karpowich, N.K., Czyzewski, B.K., and Wang, D.N. (2011).
Simple screening method for improving membrane protein thermostability.
Methods 55, 324–329.
Negulyaev, Y.A., and Markwardt, F. (2000). Block by extracellular Mg2+ of
single human purinergic P2X4 receptor channels expressed in human embry-
onic kidney cells. Neurosci. Lett. 279, 165–168.
North, R.A. (2002). Molecular physiology of P2X receptors. Physiol. Rev. 82,
1013–1067.
Rasmussen, S.G., Choi, H.J., Fung, J.J., Pardon, E., Casarosa, P., Chae, P.S.,
Devree, B.T., Rosenbaum, D.M., Thian, F.S., Kobilka, T.S., et al. (2011).
Structure of a nanobody-stabilized active state of the b(2) adrenoceptor.
Nature 469, 175–180.
Rosenbaum, D.M., Zhang, C., Lyons, J.A., Holl, R., Aragao, D., Arlow, D.H.,
Rasmussen, S.G., Choi, H.J., Devree, B.T., Sunahara, R.K., et al. (2011).
Structure and function of an irreversible agonist-b(2) adrenoceptor complex.
Nature 469, 236–240.
Shaffer, P.L., Goehring, A., Shankaranarayanan, A., and Gouaux, E. (2009).
Structure and mechanism of a Na+-independent amino acid transporter.
Science 325, 1010–1014.
Sobolevsky, A.I., Rosconi, M.P., and Gouaux, E. (2009). X-ray structure,
symmetry and mechanism of an AMPA-subtype glutamate receptor. Nature
462, 745–756.
Soriano, A., Ventura, R., Molero, A., Hoen, R., Casado, V., Cortes, A., Fanelli,
F., Albericio, F., Lluıs, C., Franco, R., and Royo, M. (2009). Adenosine A2A
receptor-antagonist/dopamine D2 receptor-agonist bivalent ligands as phar-
macological tools to detect A2A-D2 receptor heteromers. J. Med. Chem. 52,
5590–5602.
Surprenant, A., and North, R.A. (2009). Signaling at purinergic P2X receptors.
Annu. Rev. Physiol. 71, 333–359.
Thompson, A.J., Lester, H.A., and Lummis, S.C. (2010). The structural basis of
function in Cys-loop receptors. Q. Rev. Biophys. 43, 449–499.
Ward, W.W. (1981). Properties of the coelenterate green-fluorescent protein.
In Bioluminescence and Chemiluminescence: Basic Chemistry and
Analytical Applications, M. DeLuca and W.D. McElroy, eds. (New York:
Academic Press), pp. 235–242.
Wildman, S.S., King, B.F., and Burnstock, G. (1999). Modulation of ATP-
responses at recombinant rP2X4 receptors by extracellular pH and zinc. Br.
J. Pharmacol. 126, 762–768.
Yamashita, A., Singh, S.K., Kawate, T., Jin, Y., and Gouaux, E. (2005). Crystal
structure of a bacterial homologue of Na+/Cl—dependent neurotransmitter
transporters. Nature 437, 215–223.
Zambrowicz, B.P., and Sands, A.T. (2003). Knockouts model the 100 best-
selling drugs—will they model the next 100? Nat. Rev. Drug Discov. 2, 38–51.
Zhang, H., Kurisu, G., Smith, J.L., and Cramer, W.A. (2003). A defined protein-
detergent-lipid complex for crystallization of integral membrane proteins: the
cytochrome b6f complex of oxygenic photosynthesis. Proc. Natl. Acad. Sci.
USA 100, 5160–5163.
–1299, August 8, 2012 ª2012 Elsevier Ltd All rights reserved 1299