Download Defining the interaction of perforin with calcium and the

Biochem. J. (2013) 456, 323–335 (Printed in Great Britain)
323
doi:10.1042/BJ20130999
Defining the interaction of perforin with calcium and the phospholipid
membrane
Daouda A. K. TRAORE*†‡1 , Amelia J. BRENNAN§1 , Ruby H. P. LAW*†, Con DOGOVSKI¶**, Matthew A. PERUGINI¶**, Natalya
LUKOYANOVA††, Eleanor W. W. LEUNG‡, Raymond S. NORTON‡, Jamie A. LOPEZ§, Kylie A. BROWNE§, Hideo YAGITA‡‡,
Gordon J. LLOYD*†, Annette CICCONE§, Sandra VERSCHOOR§, Joseph A. TRAPANI§§§, James C. WHISSTOCK*†2 and Ilia
VOSKOBOINIK§,2
*Department of Biochemistry and Molecular Biology, Monash University, Clayton, VIC 3052, Australia, †The ARC Centre of Excellence in Structural and Functional Microbial Genomics,
Monash University, Clayton, VIC 3052, Australia, ‡Medicinal Chemistry, Monash Institute of Pharmaceutical Sciences, Monash University, Parkville, VIC 3052, Australia, §Cancer
Immunology Program, Peter MacCallum Cancer Centre, St Andrews Place, East Melbourne, VIC 3002, Australia, Sir Peter MacCallum Department of Oncology, The University of
Melbourne, Parkville, VIC 3010, Australia, ¶Department of Biochemistry, La Trobe Institute for Molecular Science, La Trobe University, Bundoora, VIC, 3086, Australia, **Department of
Biochemistry and Molecular Biology, Bio21 Molecular Science and Biotechnology Institute, Parkville, VIC 3010, Australia, ††Crystallography, Institute of Structural and Molecular
Biology, Birkbeck College, Malet Street, London WC1E 7HX, U.K., ‡‡Department of Immunology, Juntendo University School of Medicine, Tokyo, 113-8421, Japan, §§Department of
Pathology, The University of Melbourne, Parkville, VIC 3010, Australia, and Department of Genetics, The University of Melbourne, Parkville, VIC 3010, Australia.
Following its secretion from cytotoxic lymphocytes into the
immune synapse, perforin binds to target cell membranes through
its Ca2 + -dependent C2 domain. Membrane-bound perforin then
forms pores that allow passage of pro-apoptopic granzymes into
the target cell. In the present study, structural and biochemical
studies reveal that Ca2 + binding triggers a conformational change
in the C2 domain that permits four key hydrophobic residues
to interact with the plasma membrane. However, in contrast
with previous suggestions, these movements and membrane
binding do not trigger irreversible conformational changes in
the pore-forming MACPF (membrane attack complex/perforinlike) domain, indicating that subsequent monomer–monomer
interactions at the membrane surface are required for perforin
pore formation.
INTRODUCTION
resulting pore typically comprises 20–22 PRF molecules, with
each PRF monomer contributing two β-hairpins to a giant
membrane-spanning β-barrel [9,10].
The PRF C2 domain is unusual in that it has a relatively low
affinity (∼ 200 μM) for Ca2 + in comparison with homologous C2
domains that function inside the cell [11]. This feature probably
represents an important control point, and helps ensure that PRF
does not attack the cell from inside, before its release from
secretory granules into the immune synapse. Although membrane
binding is clearly a prerequisite for the formation of PRF pores,
the exact mechanism and role of Ca2 + -dependent C2 domain
interaction with the membrane has not been characterized. It also
remains unclear whether membrane binding of PRF represents
a trigger for PRF unfolding and insertion into the phospholipid
bilayer of the target cell membrane.
In the present study, we investigated the mechanism of Ca2 + dependent PRF C2 domain membrane binding. We found that
Ca2 + co-ordination by the C2 domain induced conformational
changes that significantly stabilized monomeric PRF. This was
a prerequisite for subsequent hydrophobic interactions with the
plasma membrane and positioning of monomeric PRF such that
it favoured pore formation. Furthermore, we showed that initial
interactions between PRF, Ca2 + and the plasma membrane did
not trigger either oligomerization or irreversible conformational
change within the MACPF domain.
The pore-forming protein, PRF (perforin; PRF1 gene), is stored
and secreted by cytotoxic lymphocytes [CTLs (cytotoxic Tlymphocytes) and natural killer cells] and is essential for the
successful elimination of virus-infected or oncogenic cells [1].
During an immune response, cytotoxic lymphocytes form an
immune synapse with a target cell, and subsequently release PRF
monomers and pro-apoptotic serine proteases (granzymes) into
the synaptic cleft [2]. Subsequent interaction with target cell membranes results in PRF oligomerization into transmembrane pores,
which facilitates the entry of granzymes to initiate apoptosis [3,4].
Carriers of bi-allelic mutations in the PRF1 gene lack cytotoxic
lymphocyte function and develop an aggressive immunoregulatory disorder, familial haemophagocytic lymphohistiocytosis (type
II FHL) [5] or haematological malignancies [6].
PRF consists of an N-terminal MACPF (membrane attack
complex/PRF-like) domain [7–9] and a central EGF (epidermal
growth factor)-containing ‘shelf-like’ structure, beneath which is
positioned a Ca2 + -dependent C2 domain [9]. The C2 domain is
crucial for PRF function, and it governs initial interactions with
the target cell membrane in a Ca2 + -dependent fashion. Following
membrane binding, PRF oligomerizes, and two membranespanning regions (termed TMH1 and TMH2) in the MACPF
domain are released and penetrate the lipid membrane. The
Key words: apoptosis, C2 domain, cytotoxic lymphocyte,
perforin, pore-forming protein.
Abbreviations used: CBR, Ca2 + -binding region; CDC, cholesterol-dependent cytolysin; CTL, cytotoxic T-lymphocyte; EGF, epidermal growth factor;
HRP, horseradish peroxidase; mAb, monoclonal antibody; MACPF, membrane attack complex/perforin-like; PRF, perforin; RBL, rat basophil leukaemia
cell; SRBC, sheep red blood cell; Tev, tobacco etch virus; WT, wild-type.
1
These authors contributed equally to this work.
2
Correspondence may be addressed to either of these joint senior authors (email [email protected] or [email protected]).
The structural co-ordinates reported will appear in the PDB under codes 3W56 and 3W57.
c The Authors Journal compilation c 2013 Biochemical Society
324
D. A. K. Traore and others
EXPERIMENTAL
Cell culture
RBL (rat basophil leukaemia cells RBL-2H3) and Jurkat human
leukaemia target T-cells [12], primary CTLs of Prf1 − / − .OT-1
C57BL/6 mice, EL-4 thymoma target cells (H-2Kb) [13] and
K562 human erythromyeloblastoid leukaemia target cells [14]
were generated and/or maintained as described previously.
analysed for PRF expression by immunoblotting with rat anti(mouse PRF) mAb (monoclonal antibody) P1-8 [19] followed
by secondary HRP (horseradish peroxidase)-linked anti-(rat
Ig) antibody (Dako). Mouse anti-(human actin) mAb (Sigma)
followed by secondary HRP-linked anti-(mouse Ig) (Dako) was
used as a loading control for cell lysates. Signals were amplified by
chemiluminescence and detected on X-ray film (GE Healthcare).
Confocal microscopy
Expression and binding of recombinant PRF
As described previously, mouse WT (wild-type) and mutant
recombinant PRF were expressed and purified, and activity was
assessed using SRBC (sheep red blood cell) (Department of
Veterinary Sciences, The University of Melbourne) lysis and 51 Cr
release assays [12]. The binding and oligomerization efficiency of
PRF was determined as described previously [15], by incubating
PRF and SRBCs in a neutral HE buffer (10 mM Hepes and
150 mM NaCl, pH 7) with 1 mM Ca2 + at either 4 ◦ C or 37 ◦ C for
15 min. To determine the proportion of monomeric PRF that had
oligomerized into SDS-resistant complexes, 8 M urea was added
to the sample buffer [15]. Thermal stability of WT and mutant PRF
was assessed by unfolding temperature analysis using SYPRO®
Orange as described in [3,4].
Analytical ultracentrifugation of PRF
Sedimentation velocity experiments were conducted in a
Beckman model XL-A analytical ultracentrifuge at a temperature
of 20 ◦ C. A 380 μl sample (0.253 mg/ml) and 400 μl of
reference solution (50 mM Tris, 300 mM NaCl, 10 % glycerol
and 0.02 % azide, pH 7.2) were loaded into a conventional
double sector quartz cell and mounted in a Beckman four-hole
An-60 Ti rotor. Samples were centrifuged at a rotor speed of
40 000 rev./min and the data were collected continuously at a
single wavelength (280 nm), using a step-size of 0.003 cm without
averaging. Solvent density (1.015 g/ml) and viscosity [1.070 cP
(1 poise = 10 − 1 N·s·m − 2 )], as well as estimates of the partial
specific volume (0.719 ml/g) and hydration estimate (0.388 g/g)
at 20 ◦ C were computed using the program SEDNTERP [16].
Sedimentation velocity data at multiple time points were fitted
to a continuous sedimentation coefficient [c(s)] distribution and
a continuous mass [c(M)] distribution model [17,18] using the
program SEDFIT, at a resolution of 200, with M min = 1 kDa,
M max = 150 kDa and at a confidence level (F-ratio) = 0.95.
SEDFIT is available to download from the following website:
http://www.analyticalultracentrifugation.com.
Cytotoxicity assays
The QuikChange® site-directed mutagenesis system was used to
generate point mutations in mouse WT PRF cDNA, which were
cloned into the pIRES-EGFP expression plasmid (Biosciences
Clontech). As described earlier, mouse WT and mutant PRF
plasmids were transiently transfected in RBL cells [11] and
primary CTLs of Prf1 − / − .OT-1 C57BL/6 mice [13] and sorted
for equal mean GFP fluorescence with subsequent assessment of
cytotoxic activity in 51 Cr release assays [11].
Electrophoresis and immunoblotting
Cell lysates, SRBC membranes and purified recombinant PRF
were resolved on SDS/PAGE (9 %) Tris/glycine gels and
c The Authors Journal compilation c 2013 Biochemical Society
RBL cells transiently transfected with mouse WT and mutant PRF
cDNA were fixed in ice-cold methanol and processed and imaged
as described previously [20]. PRF was detected using a primary
mouse anti-PRF mAb (clone 6G7/1F10) [21] followed by Alexa
Fluor® 546-conjugated goat anti-mouse secondary antibody
excited at 543 nm. The cells were imaged with an Olympus
FV1000 Confocal Microscope equipped with a 1 mW 543 nm
green HeNe laser. All images were captured with a PlanApoN
×60 oil immersion objective [NA (numerical aperture) = 1.42].
The images were subsequently processed with the Olympus
Micro FV10-ASW program. Final images are displayed as z-stack
projections.
Protein expression and purification of SmC2P1
A DNA sequence encoding the SmC2P1 (without the signal
peptide) was purchased from Top Gene Technology and cloned
into pAPG110 for storage. The SmC2P1 gene was amplified using
PCR with the following oligonucleotides: 5 -CGAATTCCCATATGCAGCTGCGTCTGTATAATCTGCG-3 and 5 -GCATTATATGTATACCCTGAGCGTGTAAGGATCCAAGCTTCG-3 and
cloned into a modified version of the Escherichia coli expression
vector pCOLD IV (Takara Bio). The pCOLD IV plasmid was
modified by restriction enzyme digest, first using NdeI and
HindIII to excise and allow ligation of the HisTev/maspin
encoding sequence from pHisTev/maspin [22]. Secondly, EcoRI
and HindIII restriction enzymes were used to excise the fulllength maspin gene. Finally, the SmC2P1 gene was ligated via
EcoRI and HindIII restriction sites into the resulting plasmid,
pCOLDIVHisTev/SmC2P1. This plasmid encodes an N-terminal
His6 tag linked to the SmC2P1 cDNA via a Tev (tobacco etch
virus) protease recognition site. Positive clones were confirmed
by DNA sequencing and transformed into Rosetta Gami 2
(DE3)pLysS E. coli cells (EMD Millipore) for expression.
The final amino acid sequence comprises an N-terminal tag
(MRGSHHHHHHENLYFQGQNSHM) followed by residues
21–129 of SmC2P1.
Rosetta Gami 2 (DE3)pLysS E. coli cells (EMD Millipore) were
used as an expression host. Cells were grown in yeast/tryptone
medium [1.6 % (w/v) tryptone, 1 % (w/v) yeast extract and 0.5 %
NaCl] with 100 μg/ml ampicillin at 37 ◦ C with shaking until D600
reached 0.6–0.8. Expression was then induced by the addition of
IPTG (1 mM final concentration) and growth continued overnight
at 16 ◦ C with shaking. Cells were harvested by centrifugation
(8000 g, 20 min, 4 ◦ C), resuspended in buffer A (50 mM Tris/HCl,
pH 8.0, 50 mM NaCl, 10 % imidazole and 0.01 Triton X-100)
containing a tablet of CompleteTM protease inhibitor (Roche) and
lysed by sonication. The lysate was clarified by centrifugation
and the supernatant was subject to purification.
Purification of the His6 -tagged SmC2P1 protein was achieved
with three steps of chromatography columns. The lysate was first
loaded on to a 1 ml HisTrap FF affinity column (GE Healthcare).
The column was washed with five column volumes of buffer A
and five column volumes of buffer B (50 mM Tris/HCl, pH 8.0,
500 mM NaCl and 25 % imidazole). The His6 -tagged SmC2P1
Mechanisms of perforin membrane binding
protein finally eluted with buffer B containing 250 mM imidazole.
Fractions containing the protein were concentrated and loaded on
to a Superdex 200 16/60 gel filtration column pre-equilibrated
with buffer C (50 mM Tris/HCl, pH 8.0, and 150 mM NaCl).
Fractions from the gel filtration steps were pooled, and diluted
1:3 in 50 mM Tris/HCl, pH 8.0, and purified through a 5 ml
HiTrap Q FF column (GE Healthcare). A salt gradient made
of buffer C (50 mM Tris/HCl, pH 8.0, and 50 mM NaCl) and
buffer D (50 mM Tris/HCl, pH 8.0, and 1 M NaCl) was used
to elute the protein. The purity of the protein was assessed by
SDS/PAGE.
Structural analysis of SmC2P1
The His6 -tag purified protein was concentrated to 6 mg/ml
and subject to crystallization trials. Initial screenings of
the crystallization conditions were performed at the Monash
University CrystalMation platform using the following screens:
JCSG + (Qiagen), PEGion HT (Hampton Research) and
Wizard Classic 1 and 2 (Emerald Bio). Crystallization plates
were stored at 293 K and photographs were taken every day.
Crystals appeared after 48 h in two conditions. Optimization was
performed manually using the hanging-drop method. The best
diffracting crystals were obtained in [100 mM Tris/HCl, pH 8.5,
200 mM MgCl2 and 10–20 % (w/v) PEG 8000] for ApoSmC2P1
and [100 mM Tris/HCl, pH 8.5, 200 mM CaCl2 and 16–22 %
(w/v) PEG 3350] for CaSmC2P1.
Before data collection, crystals were soaked in a solution
containing the crystallization mother liquor with 25 % glycerol
for cryo-protection. Crystals were flash cooled in liquid nitrogen
and data collected at 100 K. Screening for the best diffracting
crystal was carried out in-house on a Rigaku MicroMax-007
HF rotating anode mounted with a RAXIS IV + + image plate.
Diffraction data were then collected either in-house or on the
MX1 beamline of the Australian Synchrotron. The data collection
strategy consisted of a full scan (>180 ◦ ) on the crystal with 1 ◦
oscillation range. Exposure time was 60 s in-house and 1 s at the
synchrotron. Data were indexed and integrated with XDS [23] and
intensities scaled with SCALA [24]. Data collection statistics are
summarized in Table 1. The Matthews coefficient and estimation
of solvent content of the crystals was carried out in CCP4
[25].
Crystals of ApoSmC2P1 belong to the space group C2 with
unit cell parameter a = 25.32 Å (1 Å = 0.1 nm), b = 55.22 Å,
c = 89.75 Å, α = 90.00 ◦ , β = 91.17 ◦ and γ = 90.00 ◦ . Analysis
from the Matthews coefficient (V m = 2.52 Å3 Da − 1 ) and the
solvent content (51.30 %), the volume of the unit cell is consistent
with the presence of a single copy of the molecule in the
asymmetric unit. The crystals in presence of Ca2 + belong to the
space group P21 with unit cell parameters a = 43.80 Å, b = 52.75
Å, c = 49.06 Å, α = 90.00 ◦ , β = 107.15 ◦ and γ = 90.00 ◦ . The
Matthews coefficient (V m = 2.18 Å3 Da − 1 ) and the solvent
content suggest (43.53 %) the presence of two monomers in the
asymmetric unit.
The structures were determined at 1.60 Å (R = 20.89 %;
Rfree = 23.94 %) and 1.66 Å (R = 14.75 %; Rfree = 17.18 %)
respectively for apoSmC2P1 and CaSmC2P1. The structure of
both proteins were determined using molecular replacement as
deployed in PHASER using the co-ordinates of the PRF C2
domain (residues 413–525) as a search probe. Refinement was
carried out using REFMAC and PHENIX. Structure factors and
final models are available in the PDB under accession codes
3W56 and 3W57. Data collection and refinement statistics are
summarized in Table 1.
Table 1
325
Statistics for X-ray data collection and refinement for SmC2P1
Value in parentheses refer to the highest resolution shell. R merge = hkl i |I i (hkl) − <I (hkl)>
|/ hkl i I i (hkl), where I i (hkl) is the intensity of individual reflections. R pim is the multiplicityweighted R merge . R work = (|F o | − |F c |)/|F c |. R free was calculated using 5 % of randomly selected
reflection, excluded from the refinement.
Parameters
Data collection
Space group
Cell dimensions
a , b , c (Å)
α, β, γ (◦ )
Mosaicity (◦ )
Resolution (Å)
Total number of reflections
Number of unique reflections
Completeness (%)
Redundancy
I /σ I
R merge
R pim
Overall B -factor from Wilson plot (Å2 )
Refinement
Number of molecules/ASU
Resolution limit (Å)
Number of reflections
R work /R free (%)
Number of atoms
Protein
Ions
Water
B -factors (Å2 )
RMSDs
Bond lengths (Å)
Bond angles (◦ )
Ramachandran plot (%)
Favoured region
Outliers
CaSmC2P1
apoSmC2P1
P 21
C2
43.80, 52.75, 49.06
90.00, 107.15, 90.00
0.174
20.01–1.66 (1.75–1.66)
107656 (13062)
24949 (3367)
98.6 (90.9)
4.3 (3.9)
19.8 (5.2)
0.047 (0.267)
0.026 (0.153)
20.46
25.32, 55.22, 89.75
90.00, 91.17, 90.00
0.294
47.03–1.60 (1.69–1.60)
73016 (10548)
16003 (2292)
97.9 (96.5)
4.6 (4.6)
9.0 (2.3)
0.088 (0.399)
0.040 (0.182)
25.142
2
1.66
24932
14.75/17.18
1
1.60
15871
20.89/23.94
1804
8
404
16.99
881
0
152
19.47
0.010
1.08
0.010
1.11
97.71
0.00
99.04
0.00
Statistical analysis
An unpaired Student’s t test was used to compare two groups of
samples, and a one-way ANOVA with Newman–Kuels post-test
analysis was used to compare more than two groups of samples.
RESULTS
PRF is stabilized by Ca2 + binding
In 2010, we determined the structure of monomeric PRF [9]. In
addition to identifying the overall architecture of the molecule,
our data revealed the presence of two Ca2 + atoms bound to the
C2 domain [9]. One cation [Ca2 + (I)] was bound in a canonical
position to conserved residues Asp435 and Asp483 (site I) within
the CBRs (Ca2 + -binding regions) 1 and 2. A second Ca2 + atom
[Ca2 + (II)] was bound in a non-canonical site on the exterior of
CBR3 (including residue Asp490 ). These ions were presumably
scavenged during purification, as no Ca2 + was added to the
crystallization buffers, and thus were likely to be tightly bound.
Through comparison with known C2 domains and owing
to a requirement for high-micromolar concentration of Ca2 +
for PRF membrane binding and pore-forming activity, we
concluded that one or two lower affinity CBRs remained vacant
(different types of C2 domains bind different numbers of Ca2 +
atoms). Unexpectedly, our structural data revealed that Asp429 ,
an invariantly conserved residue that is essential for Ca2 + dependent membrane binding of PRF [11] was located >8 Å
c The Authors Journal compilation c 2013 Biochemical Society
326
Figure 1
D. A. K. Traore and others
Prediction for the calcium-induced structural reorganization of PRF C2 domain
PRF C2 domain backbone (magenta) and the density map (white) are based on [9]. Key residues Asp429 , Trp427 , Tyr430 , Tyr486 and Trp488 are shown as sticks; Tyr486 is absent in the density map. Two
high-affinity Ca2 + ions are represented as spheres (magenta). The model shows that CBR-3 residues Tyr486 and Trp488 ‘point’ towards the lipid membrane. In contrast, CBR-1 residues Trp427 and
Tyr430 are orientated away from the lipid bilayer. We predict that in the presence of elevated Ca2 + , Asp429 of CBR-1 ‘swings’ towards CBR-3; this movement repositions Trp427 and Tyr430 , so they
now face the membrane. Arrows indicate the predicted movement of CBR-1 residues Asp429 , Trp427 and Tyr430 in response to elevated Ca2 + , and the circle (and ellipse) shows the anticipated final
position of the hydrophobic residues Trp427 and Tyr430 with respect to the membrane. (A) Front and (B) side view (90 ◦ rotation) of the C2 domain.
distal from the core CBR and could not possibly co-ordinate
Ca2 + in this position [9]. Accordingly, it was suggested that
engagement of Asp429 in Ca2 + binding could only occur through
major rearrangement in the C2 domain, leading to significant
repositioning of Asp429 (Figure 1). We therefore predicted that
Ca2 + -induced conformational change would position the PRF
C2 domain into a thermodynamically favourable and stable
conformation, as has been demonstrated for other Ca2 + -binding
proteins [26].
To investigate the global effect of Ca2 + on PRF, we first
examined the thermal stability of WT PRF. As expected [4],
at physiological concentrations of Ca2 + (1 mM), purified
PRF acquired a more stable conformation, as detected by a
significant increase in melting temperature (Figure 2A). The
increase in stability was at least partly attributed to the putative
conformational change that occurred when residue Asp429 of
CBR1 co-ordinated Ca2 + , as the melting temperature of the
inactive, but structurally stable, D429A mutant [11] did not
increase in the presence of Ca2 + (Figure 2A). Using analytical
ultracentrifugation, we also discovered that, in contrast with
previous studies [27], PRF maintained its monomeric structure in
solution in the presence of Ca2 + and did not aggregate (Figure 2B).
Taken together, these data suggested that the conformational
change that occurred through Ca2 + co-ordination in the C2
domain of PRF stabilized the monomer, but did not trigger
oligomerization in solution. We next examined the consequence
c The Authors Journal compilation c 2013 Biochemical Society
of Ca2 + -induced conformational changes in the C2 domain on
PRF activity in the presence of a phospholipid bilayer.
Hydrophobic residues at the tip of the PRF C2 domain are required
for efficient interaction with membranes
The CBRs of PRF include two pairs of conserved and exposed
aromatic side chains at the tip of the C2 domain: hydrophobic
residues Trp427 and Tyr430 (in CBR1), and Tyr486 and Trp488
(in CBR3), which we predicted would play a role in plasma
membrane contact and binding [9]. We also noted that the
topology of Trp427 /Tyr430 would be sub-optimal for membrane
binding in apo-PRF, and predicted that these residues would
be repositioned in response to Ca2 + -induced conformational
change of Asp429 to face the membrane (Figure 1).
In order to determine whether the four residues were
involved in membrane binding, we engineered mutants
with several combinations of amino acid substitutions and
tested their cytotoxic function using effector-target cell-based
RBL assays [12]. The mutation of all four residues, to
alanine or serine residues (W427A/Y430A/Y486A/W488A and
W427S/Y430S/Y486S/W488S; Figure 3A) resulted in a complete
loss of function, thus confirming that PRF activity was absolutely
dependent on the presence of exposed aromatic side chains at the
tip of the C2 domain. Even in the most physiological environment
Mechanisms of perforin membrane binding
Figure 2 Ca2 + -induced conformational change of PRF is localized to the
C2 domain
(A) Thermal melting temperature of WT and D429A PRF in the absence and presence (2 mM)
of Ca2 + . Each value represents means +
− S.E.M. for three experiments and statistics were
obtained with an unpaired two-sample for means t test; *P < 0.05; n.s., not significant. (B)
Sedimentation velocity analysis of PRF. Sedimentation velocity experiments were conducted
in a Beckman model XL-A analytical ultracentrifuge as described in the Experimental section.
Continuous mass c (M ) (1/Da) distribution was plotted as a function of molar mass (kDa).
of cytotoxic lymphocytes, the quadruple mutants were unable to
restore the activity of primary PRF-deficient cytotoxic T-cells
from Prf1 − / − mice (Figure 3E). Quadruple substitutions did not
affect protein folding, stability and intracellular localization, as
shown by thermal melt experiments, Western immunoblotting and
immunofluorescence microscopy (Figures 3F and 4A).
To test the significance of individual amino acids in
CBR1 and CBR3, and identify the minimum number of
aromatic side chains required to restore PRF to WT function,
we generated a series of triple mutants. The presence of
intact Trp427 alone (in Y430A/Y486A/W488A mutant), Tyr430
alone (in W427A/Y486A/W488A mutant) or Trp488 alone (in
W427A/Y430A/Y486A mutant), all permitted ∼ 10 % of WT
PRF activity (estimated as the relative number of killer cells
required to eliminate the same number of targets; Figure 3B). In
comparison, leaving Tyr486 intact (in the W427A/Y430A/W488A
mutant) only rescued ∼ 1 % of activity (Figure 3B), suggesting
that Tyr486 played a less important role in PRF function than the
other three residues. Next, we found that keeping the hydrophobic
residues of either CBR1 or CBR3 intact was equally important for
PRF function in the context of effector cells, as W427A/Y430A
and Y486A/W488A had ∼ 50 % reduced activity compared with
327
WT PRF (Figure 3C). Mutation of an individual residue (i.e.
W427A, Y430A, Y486A or W488A) had no effect on WT PRF
activity (Figure 3D). Therefore maintaining a pair of residues
within CBR1 or CBR3 (i.e. W427A/Y430A and Y486A/W488A)
was critical for maintaining PRF function.
Our functional studies clearly showed that the exposed
aromatic side chains at the tip of the C2 domain are important
for PRF function. To determine whether this is due to
interactions with the plasma membrane, we expressed and
purified W427A/Y430A, Y486A/W488A and the quadruple
mutant W427A/Y430A/Y486A/W488A. First, we tested whether
the mutations influenced Ca2 + -dependent stabilization of PRF,
by investigating changes in the thermal stability of the PRF
mutants. In contrast with the D429A mutant (Figure 2A),
W427A/Y430A/Y486A/W488A, as well as W427A/Y430A and
Y486A/W488A, all gained thermodynamic stability in the
presence of Ca2 + (Figure 4A), suggesting that these mutations
did not influence Ca2 + binding. Next, we investigated the
in vitro cytotoxic activity of each mutant and found that all
three purified PRF mutants had significantly less activity than
WT PRF (Figure 4B), and bound inefficiently to membranes
(Figure 4C). Furthermore, the CBR1 mutant (W427A/Y430A)
had ∼ 4-fold less activity than the CBR3 mutant (Y486A/W488A)
(Figure 4B). In agreement with this, W427A/Y430A bound less
efficiently to the plasma membrane than the Y486A/W488A
mutant, with more unbound recombinant protein detected in the
supernatant (Figure 4C; see lane 2). Of note, the difference in
function between the purified recombinant PRF and the cell-based
assays for W427A/Y430A and Y486A/W488A is not an unusual
observation. In the context of a cellular system, which requires
recognition of a target cell and the formation of an immune
synapse, the delivery of PRF to the target cell is polarized and
concentrated. In addition, PRF may synergize with co-secreted
proteases in cell-based assays. Taken together, this can lead to
underestimation of functional defects in mutant PRF that have
residual activity, as has been observed in the past [11–13]. By
comparison, the recombinant quadruple mutant did not lyse target
cells and was completely devoid of Ca2 + -dependent membrane
binding (Figures 4B and 4C), even though all of the essential
Ca2 + -binding aspartate residues within the C2 domain remained
intact. These results clearly demonstrated that the two pairs of
exposed aromatic side chains at the tip of the C2 domain had
an essential role in PRF membrane binding. Our data strongly
suggested that the conformational change of Asp429 towards
the Ca2 + -binding pocket would position all four hydrophobic
residues in the most favourable topology for membrane binding.
To investigate the effect of Ca2 + on PRF structure directly,
we initially attempted soaking murine PRF crystals in increasing
concentrations (100–250 μM) of Ca2 + . These experiments
invariably resulted in crystal destruction, most likely because
Asp429 forms a key contact in the PRF crystal lattice [9].
Co-crystallization experiments in the presence of appropriate
concentrations of Ca2 + (100–250 μM) have also failed to yield
suitable crystals. Moreover, extensive attempts over many years
to produce the PRF C2 domain alone (using a variety of domain
boundaries) have not yielded folded protein. This latter problem
is unsurprising, since the PRF C2 domain has evolved to form
a modest hydrophobic interface with the EGF/shelf region [9].
To understand how the PRF C2 domain functions, we decided
to investigate whether related proteins could be used as a model
system to study the PRF C2 domain.
Our previous results have shown that the PRF C2 domain is
quite unusual, as it is a type II C2 domain that in terms of amino
acid sequence is most similar to the type I C2 fold variant. Indeed,
the most similar protein structurally characterized to date is
c The Authors Journal compilation c 2013 Biochemical Society
328
Figure 3
D. A. K. Traore and others
Four bulky hydrophobic residues of the C2 domain of PRF contribute to PRF activity
(A–D) Cytotoxic activity of transiently transfected RBL cells expressing WT PRF and hydrophobic-PRF mutants, as determined by 51 Cr release assay using Jurkat T-cells as targets, at the effector/target
(E:T) ratios indicated. All values have been standardized against WT PRF at a 30:1 effector/target ratio (100 %; average maximum lysis was 51.2 +
− 1.8 % S.E.M.), which has been duplicated in each
graph for clarity. Each value represents means +
PRF, and statistics were obtained
− S.E.M. for three independent experiments for each hydrophobic mutant and of ten independent experiments for −WT
with a one-way ANOVA with Newman–Keuls multiple comparison test post-hoc analysis; *P < 0.05. (E) Cytotoxic activity of transiently transfected CTLs of Prf1 / − .OT-1 C57BL/6 mice expressing
WT PRF and the quadruple hydrophobic-PRF mutant, as determined by 51 Cr release assay using SIINFEKL pulsed EL-4 cells as targets, at the effector/target ratios indicated and is representative of
two independent experiments. Each value represents means +
− S.E.M. and statistics were obtained with an unpaired two-sample for means t test; *P < 0.05. (F) Immunoblot for PRF in cell lysates
from transiently transfected RBL cells expressing WT PRF and hydrophobic-PRF mutants. β-Actin was used as a loading control. Molecular masses are indicated in kDa. (G) Immunofluorescence
microscopy of RBL cells expressing WT PRF and hydrophobic-PRF mutants as detected using mouse anti-PRF mAb.
c The Authors Journal compilation c 2013 Biochemical Society
Mechanisms of perforin membrane binding
Figure 4
329
Bulky hydrophobic residues are required for PRF binding to target membranes
(A) Thermal melting temperature of WT and hydrophobic mutant PRF in the absence and presence (2 mM) of Ca2 + . Each value represents means +
− S.E.M. for three experiments and statistics were
obtained with an unpaired two-sample for means t test; *P < 0.05. (B) Activity of purified recombinant WT and hydrophobic mutant PRF as determined by lysis in SRBCs, and 51 Cr release assays
in K562 cells and Jurkat T-cells. Each assay is representative of three independent experiments. Each value represents means +
− S.E.M. (C) Binding of recombinant WT and hydrophobic mutant PRF
to SRBC plasma membranes in the presence and absence of 1 mM Ca2 + at 37 ◦ C. ‘Membrane-bound’ PRF was recovered from the membrane fraction denatured with 8 M urea (lane 1), ‘unbound’
PRF was detected in the supernatant (lane 2) and ‘-Ca2 + ’ shows Ca2 + -independent binding of PRF to membranes (lane 3). Lower panel shows densitometry of percentage of unbound and bound
recombinant PRF and means +
− S.E.M. for three independent experiments. Molecular masses are indicated in kDa.
Munc13-C2B [9,28]. This protein shares 35 % sequence identity,
but importantly contains extensive insertions (including a large
helical region) and deletions in the critical Ca2 + /lipid-binding
region. Thus known structures do not permit us to understand
how many Ca2 + atoms the PRF C2 domain is able to co-ordinate.
Neither do these structures allow us to study the extent of Ca2 + driven conformational change, or whether unique rearrangements
could take place within the PRF C2 domain that, when considered
in the context of the entire molecule, could be predicted to trigger
a conformational change in the MACPF domain.
c The Authors Journal compilation c 2013 Biochemical Society
330
Figure 5
D. A. K. Traore and others
Sequence alignment of SmC2P1 and four PRF-like C2 domains
Bold and boxed amino acids indicate residues essential for Ca2 + binding in PRF. The alignment was performed using ClustalW.
To address these questions, and in light of these problems, we
therefore developed an alternative strategy to understand how
the PRF C2 domain binds Ca2 + . We conducted PSI-BLAST
searches in order to identify the C2 domain most similar to
PRF. These searches revealed a unique fish C2 domain-only
protein in the Scophthalmus maximus [29]. This protein, which is
termed SmC2P1, shares 39 % identity with the murine PRF C2
domain [29] (Figure 5) and contains all of the critical conserved
residues implicated previously in Ca2 + binding by PRF (Figure 5).
Crucially, we also noted that the lipid-binding loops were more
similar to PRF than any other known C2 domain (structurally
characterized or uncharacterized), and differed only by four amino
acid insertions (Figure 5). Overall, the high sequence similarity
led us to reason that it represented the most suitable model for
confirming our findings in PRF.
A unique C2 domain-only protein SmC2P1 as a model for PRF C2
domain
We expressed SmC2P1 and crystallized the protein in the presence
and absence of Ca2 + . The 1.60 Å structure of apo-SmC2P1
(Figures 6A–6C) revealed the overall fold is very similar to
that of the PRF C2 domain (0.536 Å deviation over 332 atoms;
Figure 6C). Indeed, Dali searches [30] confirmed that the SmC2P1
structure is more similar to PRF (z-score 17.8) than to any other
structure determined to date. Unsurprisingly, the protein with the
second highest similarity (z-score 16) was Munc13-C2B.
Although the structure of apo-PRF is unknown, we noted that
the overall position and orientation of CBR2 and CBR3 were
extremely similar in the two structures (Figure 6C). Thus the two
Ca2 + atoms already present in the PRF monomer did not seem to
dramatically influence the overall structure of CBR2 and CBR3. In
contrast, CBR1 in apo-SmC2P1, which contains the key residue
Asp35 (equivalent to Asp429 in PRF) could not be resolved in
electron density and was not modelled in the final structure. These
data suggested that this region is mobile in the absence of Ca2 + .
Whereas CBR1 is visible in the PRF crystal structure, we noted
that it makes significant crystal contacts that may influence or stabilize its conformation [9]. Interestingly, similar to PRF, SmC2P1
also underwent a large increase in thermal stability in the presence
of Ca2 + (results not shown) and its membrane-binding activity
was strictly Ca2 + -dependent (as shown by Zhao et al. [29]).
c The Authors Journal compilation c 2013 Biochemical Society
We next determined the 1.66 Å structure of Ca2 + -bound
SmC2P1 (Figures 6A, 6B and 6D–6F). These structures revealed
that three Ca2 + atoms are co-ordinated by the CBRs. Justifying
our strategy, we noted a different pattern of Ca2 + binding in
Munc13-C2B [28]. Most notably, our data revealed that Asp35
(Asp429 in PRF) had swung into the Ca2 + -binding site to coordinate both the site II and site III Ca2 + (Figure 6B). Structural
comparisons between apo- and Ca2 + -bound forms revealed no
other significant Ca2 + -induced conformational changes in the
remainder of the C2 domain.
A SmC2P1 model for PRF C2 domain Ca2 + binding reveals
localized conformational change
To help interpret the hydrophobic-dependent membrane binding
of PRF we used the structure of SmC2P1. These data revealed that
the swing of the Asp35 loop in SmC2P1 had a significant effect on
one of the residues in analogous positions to the PRF hydrophobic
residues discussed above (Figures 6D–6F). Notably, Leu36 (Tyr430 )
on CBR1 had shifted 5 Å towards the membrane-binding plane
(Figures 6D–6F). We also noted a modest shift (approximately
2.4 Å) in the side chain of Phe92 (Trp488 ) on CBR3 away
from its initial position, presumably as a consequence of slight
rearrangements of Asp90 in response to Ca2 + . In contrast, we
saw no movements in the position of Pro33 (Trp427 ) and Thr91
(Tyr486 ). Whereas all three of the latter residues represent relatively
non-conservative amino acid substitutions, we noted that the
backbone position of each residue remained essentially unaltered
in the all of the structures considered (i.e. in apo-SmC2P1, Ca2 + bound SmC2P1 or the known structure of PRF itself). These data
suggested that Ca2 + did not induce significant conformational
changes in any of these residues and only locks the Asp35 (Asp429 )
loop in a stable conformation.
Taken together, our structural and functional data (using three
independent experimental settings) strongly suggested that, in
the absence of Ca2 + , both pairs of essential PRF C2 domain
hydrophobic residues, W427A/Y430A and Y486A/W488A, are
positioned in a sub-optimal conformation with respect to one
another that precludes their interaction with the membrane.
Our study on SmC2P1 suggested that the major Ca2 + -driven
movement in the C2 domain predominantly affected the
position of Leu36 (Tyr430 ). Accordingly, we noted that the PRF
Mechanisms of perforin membrane binding
Figure 6
331
Crystal structures of SmC2P1 and comparison with the C2 domain of PRF
(A) Superimposition of ApoSmC2P1 (green) with CaSmC2P1 (light orange) and the C2 domain of PRF (magenta). For clarity, water molecules completing the co-ordination sphere of the Ca2 + ions
have not been represented. Residues connecting Ser34 to Gly38 (dashed line) were not visible in the electron density map and have therefore not been modelled in the final structure. Ca2 + ions in
CaSmC2P1 are represented as spheres (SmC2P1 coloured grey, PRF coloured magenta). (B) Superimposition of ApoSmC2P1 with CaSmC2P1 showing the re-organization of the residues involved
in Ca2 + ion co-ordination. (C) Superimposition of ApoSmC2P1 with PRF showing the position of the equivalent hydrophobic residues. (D–F) Superimposition of CaSmC2P1 with PRF highlighting
the consequence of Ca2 + binding on the orientation of residues Asp35 (Asp429 ), Pro33 (Trp427 ), Leu36 (Tyr430 ), Thr91 (Trp486 ) and Phe92 (Trp488 ).
W427A/Y430A mutant had the greatest functional defects in the
assays detailed above. With respect to the PRF Y486A/W488A
mutant, we observed in SmC2P1 a small, but noticeable,
movement in Phe92 (Trp488 ); however, we argue this modest
shift is unlikely to be of primary importance in the context
of membrane binding. Instead, we postulated that the global
Ca2 + -driven conformational change in CBR1 of PRF brings this
entire loop into closer proximity with CBR3. Taken together,
these two loops would form a striking groove. The predicted
repositioning of Asp429 in CBR1 of PRF would thus be anticipated
to change the entire environment of CBR3. This would provide
additional structural features to the PRF C2 domain, to promote
an interaction between the aromatic side chains and the plasma
membrane.
Binding to the plasma membrane does not induce irreversible
changes in the PRF MACPF domain
Our functional studies on the C2 domain of PRF, together with
structural studies on the PRF C2 domain-like protein SmC2P1,
defined the localized effect of Ca2 + on the C2 domain to explain
the mechanism of PRF–membrane binding. We noted previously
that patterns of conserved disulfide bonds linking the C2 domain
to the MACPF domain (and in particular the TMH2 region)
may represent a possible pathway through which Ca2 + binding
could directly trigger insertion of PRF into the phospholipid
bilayer [9]. In the present study, other than repositioning of
the Asp35 loop into a stable conformation, we observed no
major structural changes in SmC2P1. When rationalized in the
context of the intact PRF molecule, the predicted movement of
the C2 domain when Asp429 binds Ca2 + would be an unlikely
mechanism for triggering irreversible conformational change in
the PRF MACPF domain. We also noted that the increased thermal
stability of WT PRF (Figure 7A) was entirely reversible through
chelating Ca2 + with EGTA, as measured by the reduction in the
melting temperature to the basal (Ca2 + -free) level. Furthermore,
subsequent addition of excess Ca2 + returned PRF to its more
stable conformation (Figure 7A). These and earlier observations
(Figure 2B) suggested that Ca2 + binding did not cause irreversible
unfurling of TMH1/TMH2 and PRF aggregation. Given these
results, we investigated whether interactions between the PRF
C2 domain and membranes directly promoted irreversible
PRF membrane insertion.
In order to test whether PRF binding to the membrane caused
conformational changes that resulted in membrane insertion,
we needed to uncouple PRF binding from the events that led
to pore formation. To investigate this, we exploited a unique
feature of PRF. Unlike other MACPF proteins and CDCs
c The Authors Journal compilation c 2013 Biochemical Society
332
Figure 7
D. A. K. Traore and others
PRF C2 domain binding to target cell membranes does not cause irreversible conformational changes
(A) Thermal melting temperature of WT PRF in the absence of Ca2 + , in the presence of Ca2 + (2 mM Ca2 + ), after chelation of Ca2 + (2 mM Ca2 + plus 2 mM EGTA) and subsequent addition of excess
Ca2 + (4 mM Ca2 + plus 2 mM EGTA). Each value represents means +
− S.E.M. for three experiments and statistics were obtained with a one-way ANOVA with Newman–Keuls multiple comparison test
c The Authors Journal compilation c 2013 Biochemical Society
Mechanisms of perforin membrane binding
(cholesterol-dependent cytolysins), membrane binding of PRF
is dependent on elevated Ca2 + and, critically, it is known
that membrane binding is reversible [15]. If PRF binding
to membranes resulted in substantial MACPF domain
conformational change such as unravelling of the TMH β-hairpins
[9], then material recovered from membranes would be inactive,
as has been demonstrated for other pore-forming toxins such as
bacterial CDCs [31]. Alternatively, if membrane binding did not
promote membrane insertion then we would be able to recover
active PRF from the membrane-bound state.
To test these ideas, we designed our experiments so that the
amount of PRF added to target cells would be insufficient to
induce lysis (or form pores) (Figures 7B–7E). Thus we titrated
increasing amounts of target red blood cells to a fixed amount of
PRF until minimal/no lysis was observed (i.e. pore formation was
impaired through a decreased PRF/cell ratio). Through measuring
cytolysis we determined the number of target SRBCs required for
PRF to achieve a minimal and maximal cell lysis [i.e. 1.5×108
cells/ml (100 % lysis) and 40×108 cells/ml (no lysis) respectively
(Figure 7B)].
We then added 2.8–14 ng of PRF to 40×108 cells/ml
(Figure 7C; the ‘primary bind’). This resulted in 0–20 % cytolysis
depending on PRF concentration (Figure 7F, open squares). The
cells were washed three times using Ca2 + -free buffer in order to
elute membrane-bound PRF (Figure 7C). The supernatants were
combined and added to 27-times fewer fresh SRBCs, 1.5×108
cells/ml (Figure 7C, the ‘secondary bind’). This resulted in a
higher PRF/cell ratio and, as expected, not only allowed PRF to
‘rebind’ (Figure 7G), but also to oligomerize and lyse the targets
(Figure 7F, open circles). As a positive control, we performed the
‘primary bind’ in the absence of Ca2 + , thus preventing PRF from
membrane binding and allowing the maximum cytolysis in the
‘secondary bind’ (Figure 7D, and Figure 7F, closed circles). As
a negative control, both the ‘primary bind’ and the washes were
performed in the presence of Ca2 + that prevented elution of PRF
and, hence there was no lysis in the ‘secondary bind’ (Figure 7E,
and Figure 7F, crosses).
We found that, at higher PRF concentrations, the ‘primary
bind’ resulted in up to 20 % cell lysis, but in these instances
the subsequent lysis in the ‘secondary bind’ experiments was
reduced by the extent of the primary one (Figure 7F, last point,
arrowed). Remarkably, the sum of the primary and secondary lysis
was essentially identical to the control (Figure 7H, open circles),
indicating a complete recovery of fully functional PRF following
its initial membrane binding under conditions that would normally
favour irreversible pore formation (37 ◦ C, pH 7, 1 mM Ca2 + ).
333
Overall, these results suggested that PRF membrane binding
did not directly result in events that would be anticipated to be
irreversible: TMH unwinding, membrane insertion and formation
of the membrane-spanning β-barrel.
DISCUSSION
In the present study, we investigated the mechanism and the role
of Ca2 + binding in PRF activity. We supported our functional
studies by determining the crystal structure of the apo and
Ca2 + -bound forms of SmC2P1. Dali and BLAST searches
reveal that SmC2P1 (which comprises a C2 domain alone), is
more similar to the PRF C2 domain than any other structure
determined to date. By analogy, our structural data suggest that
PRF can co-ordinate three Ca2 + atoms within the CBRs, but
that movement of the Asp429 loop alone represents the major
Ca2 + -driven rearrangement in PRF. Critically, however, when
considered with the results of functional studies, our data reveal
that Ca2 + -induced movement of the Asp429 loop permits relative
re-positioning of four key hydrophobic residues such that they
can properly interact with the lipid membrane.
The C2 domain is one of the most common mammalian
motifs to regulate membrane-associated signalling pathways.
Most C2 domain proteins are activated by intracellular
Ca2 + signals, which drives docking of the proteins to
specific membranes predominantly through electrostatic and/or
hydrophobic interactions [32]. For example, the PKC-α (protein
kinase Cα) C2 domain binds to anionic headgroups of
PS (phosphatidylserine) and PIP2 (phosphatidylinositol 4,5bisphosphate) through electrostatic interactions [33], whereas
docking of the cPLA2 α (cytosolic phospholipase A2 α)
C2 domain is triggered by electrostatic interactions with
subsequent penetration of hydrophobic residues into PC
(phosphatidylcholine)-rich membranes [34]. In comparison with
other C2 domains, however, membrane binding through the C2
domain of PRF is unusual. PRF requires up to 1000-fold higher
concentrations of Ca2 + to restrict its activity to the extracellular
milieu [11], and, as shown in the present study, membrane
binding is regulated by Ca2 + -induced conformational changes
that permit subsequent hydrophobic interactions with the plasma
membrane. Furthermore, despite the significant variation in the
phospholipid composition of plasma membranes between cell
types [35,36], PRF binding and cell lysis have evolved to be
apparently indiscriminate of membrane composition [37]. This
is fundamentally important for host immune defence, and allows
post-hoc analysis; *P < 0.05. (B) We determined the number of SRBCs that were required for minimal and maximal lysis of WT PRF in HE buffer supplemented with 1 mM Ca2 + at 37 ◦ C. We allowed
PRF to first bind to membranes at 4 ◦ C for 10 min (a non-permissive temperature that prevents rapid oligomerization), and the solution was subsequently warmed to 37 ◦ C for 15 min, to permit
oligomerization and cell lysis. We found that a minimum of 1.5×108 cells/ml resulted in 100 % lysis and a maximum of 40×108 cells/ml resulted in no lysis. Each value represents means +
− S.E.M.
for three independent experiments. (C) WT PRF (2.8–14 ng) was added to 40×108 cells/ml of SRBCs in HE buffer containing 1 mM Ca2 + , to a final volume of 1 ml. PRF was bound to membranes at
4 ◦ C for 10 min, with subsequent warming to 37 ◦ C for 15 min (primary bind); cell lysis was assessed by haemoglobin release. The WT PRF bound to SRBCs was then washed off with four washes
of 200 μl of Ca2 + -free HE buffer. Fresh SRBCs (1.5×108 cells/ml) (secondary bind) were subsequently added to the combined supernatant at 4 ◦ C; the mixture was supplemented with 1 mM Ca2 +
and incubated for 10 min to allow PRF to bind to SRBCs. The cell suspension was subsequently warmed to 37 ◦ C (15 min) and lysis was measured. Supernatant containing haemoglobin released
from lysed cells in the primary bind was diluted 27-fold to account for the difference in cell number in the secondary bind (40×108 and 1.5×108 cells/ml respectively). (D) As a negative control, the
same experiment was repeated; however, the primary bind of WT PRF was washed with 200 μl of HE buffer plus 1 mM Ca2 + (four times). Washing in the presence of 1 mM Ca2 + prevented WT PRF
from being removed from SRBCs, and consequently there was no PRF present in the secondary bind. (E) As a positive control, the same experiment was repeated, but the primary bind of WT PRF
was conducted in the absence of Ca2 + thus precluding PRF from specific Ca2 + -dependent membrane binding; the supernatant containing unbound PRF was removed and used in the secondary
bind. (F) An SRBC lysis assay showing that after binding to 40×108 cells/ml in the presence of 1 mM Ca2 + at 37 ◦ C (primary bind), WT PRF remained active when washed off with Ca2 + -free buffer
and then added to 1.5×108 cells/ml in the presence of 1 mM Ca2 + at 37 ◦ C (secondary bind). Haemoglobin-containing supernatant from the primary bind was diluted 27-fold to account for the
difference in cell number between the primary and secondary bind (40×108 and 1.5×108 cells/ml respectively). Each value represents means +
− S.E.M. for at least three independent experiments
and statistics were obtained with an unpaired two-sample for means t test; *P < 0.05. (G) Western immunoblot using 11.2 ng PRF (corresponding to the second highest amount used in Figure 7F).
Upper panel, PRF bound to 40×108 cells/ml in the presence of 1 mM Ca2 + (primary bind, as per Figure 7C) or in the absence of Ca2 + (positive control, as per Figure 7D). Lower panel, secondary
bind of PRF eluted from the primary bind or remaining in the supernatant (S/N) in the positive control, to 1.5×108 cells/ml in the presence of 1 mM Ca2 + . (H) Comparison between the percentage
lysis of the positive control (no added Ca2 + in the primary bind) and the experiment (1 mM Ca2 + in the primary bind followed by Ca2 + -free washes), where supernatants from the primary and
secondary binds were added together (the supernatant from the primary bind was diluted as described in Figure 7F).
c The Authors Journal compilation c 2013 Biochemical Society
334
D. A. K. Traore and others
PRF to lyse target cells from differentiated organisms, spanning
insects (e.g. Sf9 cells) to humans [20].
Interestingly, mutation of either the CBR1 or CBR3 hydrophobic residues (W427A/Y430A or Y486A/W488A respectively)
impaired binding of PRF to the plasma membrane. Furthermore,
even membrane-bound mutant proteins did not oligomerize and
form pores efficiently in comparison with equal amounts of
bound WT PRF. This suggested that the loss of hydrophobic
moieties on either CBR1 or CBR3 could influence the topology
of bound monomers, thus impairing their capacity to oligomerize
efficiently. Recent studies have revealed that PRF shares structural
homology with bacterial CDCs, with a common MACPF domain
and C2/immunoglobulin membrane-binding domain folds [8].
These data suggested that, similar to the mechanism reported
for the hydrophobic membrane binding loops of bacterial CDCs,
the four bulky hydrophobic residues of PRF anchor the protein in
an orientation that promotes oligomerization [38].
The present study has revealed that the major role of the C2
domain is limited to PRF binding to the target cell membrane. We
hypothesize that subsequent interactions at the membrane surface,
such as PRF monomer–monomer contact, are instead likely to
be responsible for triggering the major conformational changes
that culminate in TMH unwinding, membrane insertion and
formation of the membrane-spanning β-barrel. Of interest, this
mechanism of pore formation has also been proposed for bacterial
CDCs, although comparison of the crystal structures and electron
microscopy structures of CDC and PRF pore forms has revealed
distinct differences [9]. CDCs oligomerize to form a pre-pore,
which subsequently collapses in order to bring the transmembrane
α-helices close enough to insert and span the bilayer [39,40].
It is unknown, however, whether PRF forms a pre-pore, as the
TMH regions of PRF are approximately twice as long as those of
CDCs, and consequently PRF does not appear to collapse during
pore formation [9]. Future studies will therefore be directed at
elucidating the exact mechanism of PRF TMH release, membrane
insertion and pore formation within the immunological synapse.
In conclusion, we have determined the structural and functional
basis for the first and essential step of cytotoxic lymphocytes
effector mechanism: PRF binding to the target cell membrane.
Our data demonstrated that the role of the PRF C2 domain
appears to be limited to anchoring the protein to the target
membrane. We further suggest that Ca2 + likely triggers a single
localized conformational change in the Asp429 loop of PRF. The
consequence of this change is, however, to activate the molecule
with respect to membrane binding. Finally, we showed that Ca2 +
binding alone does not trigger irreversible PRF oligomerization
or indeed insertion into the membrane. Together, our observations
highlight the specialized and unique regulation of PRF pore
formation, which has evolved to provide the immune system with
an exquisitely efficient cytotoxic effector molecule.
AUTHOR CONTRIBUTION
Daouda Traore and Amelia Brennan designed and conducted experiments, and contributed
to the writing of the paper. Ruby Law, Con Dogovski, Matthew Perugini, Natalya
Lukoyanova, Eleanor Leung, Gordon Lloyd, Annette Ciccone and Sandra Verschoor
conducted the experiments. Jamie Lopez, Kylie Browne, Hideo Yagita, Raymond Norton
and Joseph Trapani contributed to the experimental design. James Whisstock and Ilia
Voskoboinik co-ordinated experimental work, and contributed to the experimental design
and writing of the paper.
ACKNOWLEDGEMENTS
We thank the Australian synchrotron (MX1 and MX2) beamline team, the Monash Protein
Crystallization facility and the Monash Protein Production Unit for technical support.
c The Authors Journal compilation c 2013 Biochemical Society
FUNDING
D.A.K.T. is an Australian Research Council (ARC) Super Science Fellow. A.J.B. and J.A.L.
are National Health and Medical Research Council (NHMRC) of Australia Training Fellows.
M.A.P. is an ARC Future Fellow. I.V. is an NHMRC Career Development Fellow. J.C.W.
is an ARC Federation Fellow and an honorary NHMRC Principal Research Fellow. R.S.N.
is an NHMRC Principal Research Fellow. The work was financially supported by Project
Grants from the NHMRC Australia [APP606557 and APP1029295]. We acknowledge
the Wellcome Trust [grant number 079605/2/06/2] for their support for the Electron
Microscopy facilities at Birkbeck College.
REFERENCES
1 Lopez, J. A., Brennan, A. J., Whisstock, J. C., Voskoboinik, I. and Trapani, J. A. (2012)
Protecting a serial killer: pathways for perforin trafficking and self-defence ensure
sequential target cell death. Trends. Immunol. 33, 406–412
2 Trapani, J. A. and Smyth, M. J. (2002) Functional significance of the perforin/granzyme
cell death pathway. Nat. Rev. Immunol. 2, 735–747
3 Froelich, C. J., Orth, K., Turbov, J., Seth, P., Gottlieb, R., Babior, B., Shah, G. M.,
Bleackley, R. C., Dixit, V. M. and Hanna, W. (1996) New paradigm for lymphocyte
granule-mediated cytotoxicity. Target cells bind and internalize granzyme B, but an
endosomolytic agent is necessary for cytosolic delivery and subsequent apoptosis.
J. Biol. Chem. 271, 29073–29079
4 Lopez, J. A., Susanto, O., Jenkins, M. R., Lukoyanova, N., Sutton, V. R., Law, R. H.,
Johnston, A., Bird, C. H., Bird, P. I., Whisstock, J. C. et al. (2013) Perforin forms transient
pores on the target cell plasma membrane to facilitate rapid access of granzymes during
killer cell attack. Blood 121, 2659–2668
5 Stepp, S. E., Dufourcq-Lagelouse, R., Le Deist, F., Bhawan, S., Certain, S., Mathew, P. A.,
Henter, J. I., Bennett, M., Fischer, A., de Saint Basile, G. and Kumar, V. (1999) Perforin
gene defects in familial hemophagocytic lymphohistiocytosis. Science 286, 1957–1959
6 Chia, J., Yeo, K. P., Whisstock, J. C., Dunstone, M. A., Trapani, J. A. and Voskoboinik, I.
(2009) Temperature sensitivity of human perforin mutants unmasks subtotal loss of
cytotoxicity, delayed FHL, and a predisposition to cancer. Proc. Natl. Acad. Sci. U.S.A.
106, 9809–9814
7 Rosado, C. J., Buckle, A. M., Law, R. H., Butcher, R. E., Kan, W. T., Bird, C. H., Ung, K.,
Browne, K. A., Baran, K., Bashtannyk-Puhalovich, T. A. et al. (2007) A common fold
mediates vertebrate defense and bacterial attack. Science 317, 1548–1551
8 Rosado, C. J., Kondos, S., Bull, T. E., Kuiper, M. J., Law, R. H., Buckle, A. M.,
Voskoboinik, I., Bird, P. I., Trapani, J. A., Whisstock, J. C. and Dunstone, M. A. (2008) The
MACPF/CDC family of pore-forming toxins. Cell. Microbiol. 10, 1765–1774
9 Law, R. H., Lukoyanova, N., Voskoboinik, I., Caradoc-Davies, T. T., Baran, K., Dunstone,
M. A., D’Angelo, M. E., Orlova, E. V., Coulibaly, F., Verschoor, S. et al. (2010) The
structural basis for membrane binding and pore formation by lymphocyte perforin. Nature
468, 447–451
10 Reboul, C. F., Mahmood, K., Whisstock, J. C. and Dunstone, M. A. (2012) Predicting
giant transmembrane β-barrel architecture. Bioinformatics 28, 1299–1302
11 Voskoboinik, I., Thia, M. C., Fletcher, J., Ciccone, A., Browne, K., Smyth, M. J. and
Trapani, J. A. (2005) Calcium-dependent plasma membrane binding and cell lysis by
perforin are mediated through its C2 domain: a critical role for aspartate residues 429,
435, 483, and 485 but not 491. J. Biol. Chem. 280, 8426–8434
12 Voskoboinik, I., Thia, M. C., De Bono, A., Browne, K., Cretney, E., Jackson, J. T., Darcy,
P. K., Jane, S. M., Smyth, M. J. and Trapani, J. A. (2004) The functional basis for
hemophagocytic lymphohistiocytosis in a patient with co-inherited missense mutations in
the perforin (PFN1) gene. J. Exp. Med. 200, 811–816
13 Voskoboinik, I., Sutton, V. R., Ciccone, A., House, C. M., Chia, J., Darcy, P. K., Yagita, H.
and Trapani, J. A. (2007) Perforin activity and immune homeostasis: the common A91V
polymorphism in perforin results in both presynaptic and postsynaptic defects in
function. Blood 110, 1184–1190
14 Suck, G., Branch, D. R., Smyth, M. J., Miller, R. G., Vergidis, J., Fahim, S. and Keating, A.
(2005) KHYG-1, a model for the study of enhanced natural killer cell cytotoxicity. Exp.
Hematol. 33, 1160–1171
15 Baran, K., Dunstone, M., Chia, J., Ciccone, A., Browne, K. A., Clarke, C. J., Lukoyanova,
N., Saibil, H., Whisstock, J. C., Voskoboinik, I. and Trapani, J. A. (2009) The molecular
basis for perforin oligomerization and transmembrane pore assembly. Immunity 30,
684–695
16 Laue, T. M., Shah, B. D., Ridgeway, T. M. and Pelletier, S. L. (1992) Computer-aided
interpretation of analytical sedimentation data for proteins. In Analytical
Ultracentrifugation in Biochemistry and Polymer Science (Harding, S. E. and Horton, J.
C., eds), pp. 90–125, Royal Society of Chemistry, Cambridge
Mechanisms of perforin membrane binding
17 Schuck, P. (2000) Size-distribution analysis of macromolecules by sedimentation velocity
ultracentrifugation and lamm equation modeling. Biophys. J. 78, 1606–1619
18 Schuck, P., Perugini, M. A., Gonzales, N. R., Howlett, G. J. and Schubert, D. (2002)
Size-distribution analysis of proteins by analytical ultracentrifugation: strategies and
application to model systems. Biophys. J. 82, 1096–1111
19 Kawasaki, A., Shinkai, Y., Kuwana, Y., Furuya, A., Iigo, Y., Hanai, N., Itoh, S., Yagita, H.
and Okumura, K. (1990) Perforin, a pore-forming protein detectable by monoclonal
antibodies, is a functional marker for killer cells. Int. Immunol. 2, 677–684
20 Brennan, A. J., Chia, J., Browne, K. A., Ciccone, A., Ellis, S., Lopez, J. A., Susanto, O.,
Verschoor, S., Yagita, H., Whisstock, J. C. et al. (2011) Protection from endogenous
perforin: glycans and the C terminus regulate exocytic trafficking in cytotoxic
lymphocytes. Immunity 34, 879–892
21 Konjar, S., Sutton, V. R., Hoves, S., Repnik, U., Yagita, H., Reinheckel, T., Peters, C., Turk,
V., Turk, B., Trapani, J. A. and Kopitar-Jerala, N. (2010) Human and mouse perforin are
processed in part through cleavage by the lysosomal cysteine proteinase cathepsin L.
Immunology 131, 257–267
22 Law, R. H., Irving, J. A., Buckle, A. M., Ruzyla, K., Buzza, M., Bashtannyk-Puhalovich, T.
A., Beddoe, T. C., Nguyen, K., Worrall, D. M., Bottomley, S. P. et al. (2005) The high
resolution crystal structure of the human tumor suppressor maspin reveals a novel
conformational switch in the G-helix. J. Biol. Chem. 280, 22356–22364
23 Kabsch, W. (2010) Xds. Acta Crystallogr., Sect. D: Biol. Crystallogr. 66, 125–132
24 Evans, P. (2006) Scaling and assessment of data quality. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 62, 72–82
25 Collaborative Computational Project, Number 4 (1994) The CCP4 suite: programs for
protein crystallography. Acta Crystallogr., Sect. D: Biol. Crystallogr. 50, 760–763
26 Shin, O. H., Xu, J., Rizo, J. and Sudhof, T. C. (2009) Differential but convergent functions
of Ca2 + binding to synaptotagmin-1 C2 domains mediate neurotransmitter release. Proc.
Natl. Acad. Sci. U.S.A. 106, 16469–16474
27 Young, J. D., Podack, E. R. and Cohn, Z. A. (1986) Properties of a purified pore-forming
protein (perforin 1) isolated from H-2-restricted cytotoxic T cell granules. J. Exp. Med.
164, 144–155
28 Shin, O. H., Lu, J., Rhee, J. S., Tomchick, D. R., Pang, Z. P., Wojcik, S. M.,
Camacho-Perez, M., Brose, N., Machius, M., Rizo, J. et al. (2010) Munc13 C2B domain
is an activity-dependent Ca2 + regulator of synaptic exocytosis. Nat. Struct. Mol. Biol. 17,
280–288
335
29 Zhao, L., Hu, Y. H., Sun, L. and Sun, J. S. (2010) SmC2P1, a C2 domain protein from
Scophthalmus maximus that binds bacterial pathogen-infected lymphocytes and reduces
bacterial survival. Fish Shellfish Immunol. 29, 786–792
30 Holm, L. and Rosenstrom, P. (2010) Dali server: conservation mapping in 3D. Nucleic
Acids Res. 38, W545–W549
31 Schuerch, D. W., Wilson-Kubalek, E. M. and Tweten, R. K. (2005) Molecular basis of
listeriolysin O pH dependence. Proc. Natl. Acad. Sci. U.S.A. 102, 12537–12542
32 Corbin, J. A., Evans, J. H., Landgraf, K. E. and Falke, J. J. (2007) Mechanism of specific
membrane targeting by C2 domains: localized pools of target lipids enhance Ca2 +
affinity. Biochemistry 46, 4322–4336
33 Evans, J. H., Murray, D., Leslie, C. C. and Falke, J. J. (2006) Specific translocation of
protein kinase Cα to the plasma membrane requires both Ca2 + and PIP2 recognition by
its C2 domain. Mol. Biol. Cell 17, 56–66
34 Malmberg, N. J., Van Buskirk, D. R. and Falke, J. J. (2003) Membrane-docking loops of
the cPLA2 C2 domain: detailed structural analysis of the protein-membrane interface via
site-directed spin-labeling. Biochemistry 42, 13227–13240
35 Maccarrone, M., Nieuwenhuizen, W. E., Dullens, H. F., Catani, M. V., Melino, G., Veldink,
G. A., Vliegenthart, J. F. and Finazzo Agro, A. (1996) Membrane modifications in human
erythroleukemia K562 cells during induction of programmed cell death by transforming
growth factor β1 or cisplatin. Eur. J. Biochem. 241, 297–302
36 Nouri-Sorkhabi, M. H., Agar, N. S., Sullivan, D. R., Gallagher, C. and Kuchel, P. W. (1996)
Phospholipid composition of erythrocyte membranes and plasma of mammalian blood
including Australian marsupials; quantitative 31 P NMR analysis using detergent. Comp.
Biochem. Physiol., Part B: Biochem. Mol. Biol. 113, 221–227
37 Antia, R., Schlegel, R. A. and Williamson, P. (1992) Binding of perforin to
membranes is sensitive to lipid spacing and not headgroup. Immunol. Lett. 32,
153–157
38 Ramachandran, R., Heuck, A. P., Tweten, R. K. and Johnson, A. E. (2002) Structural
insights into the membrane-anchoring mechanism of a cholesterol-dependent cytolysin.
Nat. Struct. Biol. 9, 823–827
39 Czajkowsky, D. M., Hotze, E. M., Shao, Z. and Tweten, R. K. (2004) Vertical collapse of a
cytolysin prepore moves its transmembrane β-hairpins to the membrane. EMBO J. 23,
3206–3215
40 Tilley, S. J., Orlova, E. V., Gilbert, R. J., Andrew, P. W. and Saibil, H. R. (2005) Structural
basis of pore formation by the bacterial toxin pneumolysin. Cell 121, 247–256
Received 30 July 2013/23 August 2013; accepted 27 September 2013
Published as BJ Immediate Publication 27 September 2013, doi:10.1042/BJ20130999
c The Authors Journal compilation c 2013 Biochemical Society