This article was published in an Elsevier journal. The attached copy is furnished to the author for non-commercial research and education use, including for instruction at the author’s institution, sharing with colleagues and providing to institution administration. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright
Transcript
This article was published in an Elsevier journal. The attached copyis furnished to the author for non-commercial research and
education use, including for instruction at the author’s institution,sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies areencouraged to visit:
Vaccines are the most effective and cost-efficient meansto prevent infectious diseases. The latest trend towards noveland safer vaccines utilises well-characterised antigens, likepurified proteins, peptides, or carbohydrates. These so-calledsubunit vaccines enable focusing of the immune response tothe desired specificity without the risks associated with vac-cines based on whole inactivated or live attenuated pathogens.Unfortunately, such subunit antigens are often poor immuno-gens when administered alone. Therefore, an adjuvant isrequired to potentiate the immune response to the coadmin-istered antigen. Immunopotentiating reconstituted influenza
1 These authors contributed equally to this study.
virosomes (IRIVs) are a safe alternative to alum, which hasbeen used as an adjuvant for decades but is accompanied by ahigh rate of undesirable local and systemic reactions. Essen-tially, virosomes are reconstituted influenza virus envelopeslacking viral genetic material [1–4]. IRIVs are spherical,unilamellar vesicles with a mean diameter of less than200 nm. Their base is a liposome comprised of phosphatidyl-choline (PC), phosphatidylethanolamine (PE) and lipidsderived from the influenza virus. In contrast to liposomes,virosomes contain functional viral envelope glycoproteins:influenza virus hemagglutinin (HA) and neuraminidase (NA)intercalated in the phospholipid bilayer. Thus, reconstitutedinfluenza virosomes retain the receptor-binding and mem-brane fusion activity of influenza virus. Virosomes havebeen demonstrated to be a versatile and efficient carriersystem for a variety of antigens, e.g. proteins, peptides,nucleic acids and carbohydrates [5–12]. IRIVs are one of the
7066 A.R. Kammer et al. / Vaccine 25 (2007) 7065–7074
three adjuvant systems approved by regulatory authorities forhuman use and the only one that has carrier capabilities. Fur-thermore, they have an excellent tolerability, as shown by thetwo products already on the market [13,14].
According to the presumed model of interaction betweenvirosomes and the immune system, the type of immuneresponse induced is dependent on the localisation of theantigen. Antigen on the outer surface of the virosome isrecognised by B cells and stimulates antibody production.In contrast, encapsulated antigens are released into thecytoplasm, thereby directing the antigen to the major his-tocompatibility complex (MHC) class I pathway leading toa stimulation of the CD8+ cytotoxic T lymphocytes (CTL).In addition, antigen associated with virosomes can enter theMHC class II antigen presentation pathway, thus inducingCD4+ helper T cells [2].
Efficient loading can easily be achieved for B and helper Tcell antigens integrated into the virosomal membrane with thewidely used detergent removal procedure [4,15]. In contrast,encapsulation of CTL antigens into virosomes with this pro-cedure is inefficient as the spontaneous encapsulation rate isbelow 1%. Our recently developed procedure using chimericvirosomes (CIRIVs) resulted in good encapsulation rates(≥10%) but the preparation was rather laborious [5]. Thus,there is still a need for improvements of CTL antigen formu-lations with good encapsulation rates inducing strong CTLresponses combined with an efficient adjuvant with minimalside effects. In addition, virosomal vaccines are produced asbuffered aqueous formulations that require storage at sub-ambient temperatures. Especially in the developing world,meeting these storage requirements can be difficult, and vac-cines are often exposed to elevated temperatures that causelosses in vaccine efficacy. Lyophilisation (freeze-drying) is aprocess particularly useful to enhance stability for long-termstorage of biological materials such as proteins and vaccines,and it can make refrigerated storage obsolete for normallytemperature-sensitive materials. However, conventional viro-somes cannot be lyophilised without loss of immunogenicity,particle size homogeneity and the functional activity of theinfluenza HA. Therefore, a new generation of virosomesshould be able to induce strong immune responses, displaya good antigen encapsulation rate and improved storage andstability properties. This should be combined with a simpleand fast preparation method that offers a great flexibility incombining several specific antigens of choice while retainingthe adjuvant properties.
We describe the development and characterisation of sucha novel type of stabilised influenza virosomes (TIRIVs) thatinduce both, strong CTL and antibody responses against spe-cific antigens of choice. TIRIVs can be lyophilised but retainall functional and structural properties of virosomes. Thisfeature allows long-term storage of TIRIVs without interfer-ence with its functions. If desired, antigens can be part ofthe lyophilisate, or the lyophilisate can be reconstituted withthe antigens of choice at an appropriate concentration. Theyalso offer the advantage that antigen can enter the lumen
of the virosome during the reconstitution process. There-fore, TIRIVs offer a flexible antigen delivery vehicle withlong-term storage capability inducing strong cellular and/orhumoral immune responses.
2. Materials and methods
2.1. Viruses
Inactivated and purified influenza A/H1N1 virus strains(A/Singapore/6/86; A/New Caledonia/20/99; A/Beijing/262/95) propagated in the allantoic cavity of embryonatedeggs were obtained from Berna Biotech AG. HA quantifica-tion was performed by single-radial immunodiffusion (SRD)and by estimation on SDS-PAGE after staining with imperialprotein stain (Pierce Biotechnology).
2.2. Peptides and proteins
The following peptides were used: Core 132 (DLMG-YIPLV), OVA 257-264 (SIINFEKL), and AMA49-CPE((1,3-dipalmitoyl-glycero-2-phosphoethanolamino)-N-succ-inyl-GGCYKDEIKKEIERESKRIKLNDNDDEGNKKIIA-PRIFISDDKDSLKCG) (all from Bachem). Ovalbumin(chicken egg albumin, Grade V) from Sigma was used.
2.3. Virosome formulation
IRIVs were prepared as described previously [16]. Briefly,32 mg of egg phosphatidylcholine (PC) (Lipoid) and 8 mg of1-oleoyl-3-palmitoyl-rac-glycero-2-phosphoethanolamine(PE) (Bachem) were dissolved in 3 ml of PBS, 100 mMoctaethyleneglycol-mono-(n-dodecyl)ether (OEG-PBS).Two milligrams HA of inactivated influenza A/H1N1virus was centrifuged at 100,000 × g for 1 h at 4 ◦C,and the pellet was dissolved in 1 ml of OEG-PBS. Thedetergent-solubilised phospholipids and viruses were mixedand sonicated for 1 min. This mixture was centrifuged at100,000 × g for 1 h at 18 ◦C. The supernatant was used andvirosomes were formed by detergent removal using 1.25 g ofwet SM2 Bio-Beads (BioRad) for 1 h at room temperaturewith shaking and three times for 30 min with 625 mg of SM2Bio-Beads each. Finally, the virosomes were sterile filtered(0.22 �m).
Virosomes with peptides anchored in the virosomal mem-brane (peptide-IRIVs) were obtained by adding the wantedamount of the peptide-PE conjugate (e.g. AMA49-CPE) tothe dissolved phospholipid mixture following the proceduredescribed above.
Stabilised virosomes (TIRIVs) were prepared by addingsucrose (5% final conc.) and a stabiliser (e.g. 3�-[N-(N′,N′-dimethylaminoethane)-carbamoyl] cholesterol hydrochlo-ride (DC-chol; Sigma), cholesteryl N-(trimethylammonioe-thyl)carbamate chloride (TC-chol), N-[1-(2,3-dioleoylo-xy)propyl]-N,N,N-trimethylammonium chloride (DOTAP)
A.R. Kammer et al. / Vaccine 25 (2007) 7065–7074 7067
(both from Merck Eprova), or dimethyldihexadecylammo-nium bromide (DHAB, Sigma)). The stabiliser was addedto the phospholipids and dissolved in OEG-PBS at a finalconcentration of 1.25 mg/ml if not otherwise indicated. Viro-somes were formed by detergent removal as described above,except that two additional steps with 625 mg of Bio-BeadsSM2 each were performed. The virosomes were sterile fil-tered (0.22 �m) and aliquoted in sterile glass vials. The vialswere frozen at −70 ◦C and then lyophilised with a vacuum of10 Pa at −40 ◦C for 20 h followed by 2 h at 10 ◦C. The closedvials were stored at −20 ◦C or 4 ◦C until use.
To obtain TIRIVs containing the antigen of choice(antigen-TIRIVs), the antigen (e.g. peptides Core 132,OVA 257-264, or AMA49-CPE; chicken egg albumin) wasdissolved in water at the desired concentration. Frozen,lyophilised TIRIVs were removed from the freezer andequilibrated at RT for 2–5 min, before an equal amount ofdissolved antigen was added to the lyophilisate. The vial wasmixed shortly to dissolve the lyophilisate and stored at 4 ◦Cuntil use.
2.4. Size determination
Size determination was performed by dynamic lightscattering using a Zetasizer 1000HS instrument (MalvernInstruments) equipped with a standard 10 mW He–Ne laser(λ = 633 nm) and an avalanche photodiode (APD). Five to20 �l of sample was added to filtered PBS buffer in a finalcuvette volume of 1 ml. The measurements were performedat T = 25 ◦C at the fixed scattering angle of 90◦. The sizedistributions were evaluated by selecting the proper fitting.
2.5. In vitro fusion assay
In vitro fusion measurements were done by fluores-cence resonance energy transfer as described [5]. Thefluorescence-labelled liposomes were composed of 69%(mol/mol) egg phosphatidylcholine, 30% 1,2-dipalmitoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (both from Lipoid),0.75% Bodipy 530/550-DHPE and 0.25% N-Rh-DHPE (bothfrom Molecular Probes). The liposomes were prepared by thehydration method and sized by extrusion through polycar-bonate filters with a pore size of 0.1 �m. Measurements werecarried out with an LS 55 luminescence spectrometer (PerkinElmer Instruments) supplied with a thermostated cuvetteholder and a magnetic stirring device. Typically, 5–20 �lof virosomes (0.1–0.4 nmol phospholipid) and 1 �l of prop-erly diluted labelled liposomes (0.3 nmol phospholipid) weremixed in 2.5 ml poly(methyl methacrylate) micro-cuvettescontaining the assay buffer (5 mM sodium phosphate bufferpH 7.5, 100 mM NaCl) in a final volume of 0.8 ml under con-tinuous stirring. Considering the measurement starting timeas t = 0, fusion was triggered at t = 120 s by addition of HCl1 M to achieve the optimal pH value of ∼5. The resultingincrease in fluorescence was recorded every 5 s at excitationand emission wavelengths of 538 ± 2.5 nm and 558 ± 15 nm,
respectively. At t = 600 s Triton X-100 was added to a finalconcentration of 0.5% (v/v) in order to achieve infinite dilu-tion of the fluorescence molecules corresponding to maximalfluorescence. During the experiment the assay solution waskept under continuous stirring at 42 ◦C (water bath temper-ature). For calibration of the fluorescence scale the initialresidual fluorescence of the liposomes was set to zero andthe fluorescence at infinite probe dilution to 100%.
2.6. Antigen and stabiliser quantification
Quantifications were performed by reversed phase-HPLCon an Agilent 1100 Series system (Agilent Technologies)equipped with a sample cooler set at 4 ◦C and a DAD detec-tor set at 210 nm. For Core 132, a SunFire C8 4.6 × 150,5 �m reversed phase column (Waters) was used. 0.1% tri-fluoroacetic acid (TFA) in water/5% acetonitrile (ACN) asmobile phase A and 0.1% TFA in ACN as mobile phase Bwere used. A gradient from 10% B to 40% B over 7 minfollowed by a gradient from 40% B to 60% B over 3 minwith additional 5 min at 60% and a flow of 1 ml/min at 25 ◦Cwas applied. For AMA49-CPE, a ZORBAX Eclipse XDBC8, 4.6 mm × 150 mm, 5 �m reversed phase column (Agi-lent Technologies) was used. 0.1% TFA in water as mobilephase A and 0.1% TFA in methanol as mobile phase B wereused. A gradient from 60% B to 100% B over 15 min withadditional 5 min at 100% and a flow of 1 ml/min at 60 ◦C wasapplied. For OVA 257–264, a 125/4.6 Nucleosil 100-5 C8reversed phase column (Macherey Nagel) was used. Triethy-lammoniumphosphate 10 mmol/l pH 3 in water/1% ACN asmobile phase A and ACN as mobile phase B were used. Agradient from 20% B to 30% B over 10 min and from 30% Bto 100% B over 10 min with additional 10 min at 100% and aflow of 1 ml/min at 60 ◦C was applied. For DC-chol and TC-chol quantification, a SunFire C8 4.6 × 150, 5 �m reversedphase column was used. 0.1% TFA in water as mobile phaseA and 0.1% TFA in methanol as mobile phase B were used.A gradient from 75% B to 100% B over 9 min with addi-tional 5 min at 100% and a flow of 1 ml/min at 50 ◦C wasapplied. Protein quantification was done by estimation onSDS-PAGE after use of the Silver Quest silver staining kit(Invitrogen).
2.7. Mice and immunisations
Immunisation experiments were performed in C57Bl/6,Balb/c (Charles River) and HHD mice transgenic for HLA-A2.1 (A0201) monochain histocompatibility class I moleculeand deficient for both H-2Db and murine �2-microglobulin[17]. Mice were housed in appropriate animal care facilitiesand handled according to international guidelines.
Mice were immunised subcutaneously for cellularresponse or intramuscularly for humoral response with 100 �lof the appropriate formulation. Mice received two injectionsat a 3-week interval and the response was analysed 2 weeksafter the last injection.
7068 A.R. Kammer et al. / Vaccine 25 (2007) 7065–7074
2.8. Intracellular IFNγ staining
Spleen cells (12 × 106) were incubated with 2.5–10 �g/mlspecific peptide or non-relevant peptide (negative control)in complete RPMI medium containing 2 mM l-glutamine,100 U/ml penicillin, 100 �g/ml streptomycin, 5 mM Hepes,5% FCS and 5 × 10−5 M 2-mercaptoethanol at 37 ◦C and5% CO2 in the presence of 5 �g/ml Brefeldin A for 4 h.Cells were stained with FITC-conjugated anti-CD8 antibod-ies, permeabilised, and stained with PE-conjugated anti-IFN�antibodies using the Cytofix/Cytoperm kit following themanufacturer’s instructions (BD Pharmingen). Data wereacquired on a flow cytometer (FACSCalibur; Becton Dickin-son) and analysed with WinMDI2.8 software. Frequency ofIFN�-producing cells was calculated as percentage of IFN�+
and CD8+ cells among total CD8+ cells. The percentage ofpeptide-specific cells was obtained by subtracting the per-centage in samples stimulated with the non-relevant peptidefrom the percentage in samples stimulated with the specificpeptide.
2.9. Cytotoxicity assay
Spleen cells (4 × 106/well) from individual immunisedmice were restimulated for 5 days in 24-well tissue cul-ture plates with 6 × 105 mitomycin C treated (50 �g/mlfor 1 h) stimulator cells in complete RPMI medium con-taining 2 mM l-glutamine, 100 U/ml penicillin, 100 �g/mlstreptomycin, 5 mM Hepes, 10% FCS and 5 × 10−5 M 2-mercaptoethanol at 37 ◦C and 5% CO2. Stimulator cells wereE.G7-OVA (ATCC LGC Promochem). On day 2, 5 U/mlIL-2 was added. Specific cytolytic activity was tested ina standard 51Cr-release assay against 51Cr-labeled targetcells. Positive target cells were E.G7-OVA and negativetarget cells were EL4 (ATCC LGC Promochem). After4 h incubation, 51Cr-release was measured by using a �-counter. Spontaneous and maximum release was determinedfrom wells containing medium alone or after lysis with1 M HCl, respectively. Lysis was calculated by the for-mula: (release in assay − spontaneous release)/(maximumrelease − spontaneous release) × 100. Peptide-specific lysiswas determined as the percentage of lysis of positive targetsminus the percentage of lysis of negative targets. Spon-taneous release was always less than 15% of maximumrelease.
2.10. AMA49-CPE ELISA
Polysorp plates (Nunc) were coated overnight at 4 ◦Cwith 100 �l of a 10 �g/ml solution of the malariapeptide–phosphatidylethanolamine conjugate AMA49-CPEin PBS (pH 7.4). Wells were then blocked with 5% milkpowder in PBS for 2 h at RT, followed by three washeswith PBS containing 0.05% Tween 20. Plates were thenincubated with serial dilutions of the mouse serum inPBS containing 0.05% Tween 20 and 0.5% milk powder
for 2 h at 37 ◦C. After being washed, plates were incu-bated with HRP-conjugated goat anti-mouse Ig antibody(BD Bioscience) for 1 h at 37 ◦C. After being washedagain, OPD-substrate (O-Phenylendiamine tablets, Fluka)was added, and the plates were incubated in the darkat room temperature until the colorimetric reaction hadprogressed sufficiently and reaction was stopped by addi-tion of 100 �l 1 M H2SO4 and optical densities (OD)were read at 492 nm on a Spectra Max Plus (MolecularDevices).
3. Results
3.1. The concept of stabilised virosomes
To extend the properties of influenza virosomes regardingtheir storage and versatility for the generation of immuneresponses, the standard preparation method was modified(Fig. 1). The aim was to end up with a stable, freeze-driedpreparation of virosomes suitable for long-term storage thatcan be combined with any antigen to generate an immuneresponse of choice (cellular and/or humoral). Inactivatedinfluenza virus of an A/H1N1 strain was collected in apellet and dissolved in a phosphate-buffered saline and non-ionic detergent solution (OEG-PBS). The solubilised viralenvelope was combined with additional phospholipids (eggphosphatidylcholine and synthetic 1-oleoyl-3-palmitoyl-rac-glycero-2-phosphoethanolamine), sucrose and a newlyintroduced stabiliser. Stabilizing molecules were generallycharged molecules like 3�-[N-(N′,N′-dimethylaminoethane)-carbamoyl] cholesterol hydrochloride (DC-chol), cholesterylN-(trimethylammonioethyl)carbamate chloride (TC-chol),N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammoniumchloride (DOTAP) or dimethyldihexadecylammoniumbromide (DHAB). The viral nucleocapsid complex wasremoved by ultracentrifugation, and virosomes were formedby controlled detergent removal from the supernatantby treatment with a hydrophobic resin. The functionallyreconstituted influenza viral membranes with the incor-porated stabiliser were named stabilised virosomes orTIRIVs. Finally, TIRIVs were aliquoted in glass vials andfrozen before lyophilisation (−20 ◦C). The lyophilisationprocess was completed within 2 days and TIRIV werestored until use. The whole process was tested under GMPconditions and the process is fully upscalable to industrialquantities.
3.2. TIRIVs are homogenous and fusogenic
Reconstituted TIRIVs were characterised regarding theiraverage size, homogeneity and fusion activity and com-pared to IRIVs (Table 1). Addition of stabiliser to thevirosomal membrane increased their average diameter in aconcentration-dependent manner. TIRIVs with 1.25 mg/mlof DC-chol or TC-chol had an average diameter of approx.
Author's personal copy
A.R. Kammer et al. / Vaccine 25 (2007) 7065–7074 7069
Fig. 1. Flow chart of the preparation process for lyophilisable virosomes. Lyophilisable virosomes (TIRIVs) containing a stabiliser (e.g. TC-chol, DC-chol,DOTAP) were prepared from inactive influenza viruses by mixing solubilised influenza virus with additional phospholipids and stabiliser. The influenzanucleocapsid complex was removed by ultracentrifugation and virosomes were formed by detergent removal. These virosomes were lyophilised and stored.Homogenous, active and antigen-containing virosomes were formed after reconstitution with an antigen solution of choice. OEG, octaethyleneglycol-mono-(n-dodecyl)ether.
180–190 nm compared to 105–115 nm for IRIVs with no sta-biliser. After lyophilisation and reconstitution, IRIVs withoutstabiliser showed an increased average size and were veryheterogeneous (indicated by a polydispersity index of 1).In contrast, reconstitution of TIRIVs with stabiliser (e.g.DC-chol, TC-chol) resulted in particles with a very similaraverage size compared to the particles before lyophilisation.Additionally, the particles were very homogenous, indicatedby a low polydispersity index of <0.2. In an in vitro fusiontest, TIRIVs were fusogenic before lyophilisation and afterreconstitution to a similar extent (Fig. 2A), whereas no fusionactivity was detectable after reconstitution of lyophilisedIRIVs (Fig. 2B). HA sensitivity to proteases (e.g. bromelain,
thermolysin) was identical for TIRIVs before lyophilisationand after reconstitution, and indistinguishable from freshlyprepared, fusion-active IRIVs (data not shown). This indi-cates the conformational integrity of the HA despite thelyophilisation and reconstitution steps. TIRIV formulationswith a stabiliser concentration below 0.125 mg/ml or above2.5 mg/ml yielded less uniform particles with higher poly-dispersity indices. These results suggest that the stabiliserin TIRIVs protects the HA during the freeze-drying processand directs the reconstitution of virosomes towards particlesof very similar size and homogeneity compared to TIRIVsbefore lyophilisation. The use of other stabilisers (e.g.DOTAP, DHAB, phosphatidylserine, phosphatidylglycerol,
Author's personal copy
7070 A.R. Kammer et al. / Vaccine 25 (2007) 7065–7074
Tabl
e1
Det
erm
inat
ion
ofav
erag
esi
zean
dfu
sion
activ
ityof
IRIV
s(0
.5m
g/m
lHA
final
conc
.)w
ithdi
ffer
ents
tabi
liser
s
Sam
ple
Sam
ple
desc
ript
ion
Add
itive
conc
entr
atio
n(m
g/m
l)A
vera
gedi
amet
erbe
fore
lyop
hilis
atio
nA
vera
gedi
amet
eraf
ter
reco
nstit
utio
nFu
sion
activ
ity
nmP-
inde
xanm
P-in
dexa
Bef
ore
lyop
hilis
atio
nA
fter
reco
nstit
utio
n
1IR
IVs
–10
7.2
0.14
277.
4b1
+−
2IR
IVs
with
DC
-cho
lc0.
125
126
0.03
158.
40.
09+
+3
IRIV
sw
ithD
C-c
hol
0.62
514
9.1
0.06
141.
20.
02+
+4
IRIV
sw
ithD
C-c
hol
1.25
181.
50.
1416
5.4
0.05
++
5IR
IVs
with
TC
-cho
ld1.
2518
70.
217
00.
04+
+6
IRIV
sw
ithD
OTA
Pe1.
2512
60.
1638
3.1b
0.82
++
7IR
IVs
with
DH
AB
f1.
2514
2.2
0.1
258.
8b0.
52+
+8
IRIV
sw
ithPS
g2
165.
20.
2850
9.9b
0.95
+−
9IR
IVs
with
DPP
Gh
1.25
121
0.07
397b
1+
+10
IRIV
sw
ithD
PPA
i1.
2514
30.
0152
8b1
++
aPo
lydi
sper
sity
inde
x(0
,com
plet
eho
mog
enei
ty;1
,no
hom
ogen
eity
).b
Mea
ndi
amet
erof
part
icle
sw
ithdi
ffer
entd
iam
eter
s.c
3�-[
N-(
N′ ,N
′ -Dim
ethy
lam
inoe
than
e)-c
arba
moy
l]ch
oles
tero
lhyd
roch
lori
de.
dC
hole
ster
ylN
-(tr
imet
hyla
mm
onio
ethy
l)ca
rbam
ate
chlo
ride
.e
N-[
1-(2
,3-D
iole
oylo
xy)p
ropy
l]-N
,N,N
-tri
met
hyl]
amm
oniu
msa
lt.f
Dih
exad
ecyl
dim
ethy
lam
mon
ium
brom
ide.
gl-
A-P
hosp
hatid
yl-l
-ser
ine.
h1,
2-D
ipal
mito
yl-s
n-gl
ycer
o-3-
[pho
spho
-rac
-(1-
glyc
erol
)].
i1,
2-D
ipal
mito
yl-s
n-gl
ycer
o-3-
phos
phat
e.
Fig. 2. Hemagglutinin in TIRIVs is in a native and fusion-active confor-mation. Fusion activity of TIRIVs (panel A) and IRIVs (panel B) wasdetermined in FRET measurements at 37 ◦C before lyophilisation and afterreconstitution of the virosomes. Typical results of TIRIVs with 0.5 mg/mlHA and 1.25 mg/ml TC-chol and IRIVs with 0.5 mg/ml HA are shown. Theleft arrowhead in each panel indicates the pH-change to pH 5 and the rightarrowhead indicates the addition of Triton X-100 to 0.5% (v/v). For cal-ibration of the fluorescence scale the initial residual fluorescence was setto zero and the fluorescence at infinite probe dilution to 100% (maximalfluorescence).
and dipalmitoylphosphate) resulted in sub-optimal formula-tions, e.g. average particle size >250 nm and broader particlesize distribution. However, some of these formulations stillinduced antibodies when combined with a specific antigen(data not shown). Lyophilised TIRIVs with DC-chol or TC-Chol showed an excellent stability with regard to particlesize and HA quantification at −20 ◦C, 4 ◦C and even at25 ◦C for the time tested so far (up to 12 months; test ongo-ing). In order to evaluate the immunogenicity of lyophilisedTIRIVs, we performed immunisation studies in mice, inves-tigating both the cellular as well as the humoral immuneresponse.
Author's personal copy
A.R. Kammer et al. / Vaccine 25 (2007) 7065–7074 7071
Fig. 3. CD8+ T cell response in mice immunised with HCV Core 132 peptidein different formulations. HHD mice (six mice/group) were immunised twicesubcutaneously at a 3-week interval with TIRIVs reconstituted with peptidesolution (A), lyophilised IRIVs reconstituted with peptide solution (B), non-lyophilised IRIVs mixed with peptide solution (C), or TIRIVs reconstitutedwithout peptide (D). Spleen cells were isolated 2 weeks after the secondimmunisation and frequency of IFN�-producing CD8+ T cells was analysedas described in the methods. The percentage of peptide-specific cells wasobtained by subtracting the percentage in samples stimulated with a non-relevant peptide from the percentage in samples stimulated with HCV Core132 peptide. Shown are mean values ± standard deviation.
3.3. Antigen-TIRIVs for generating cellular immuneresponses
Lyophilised TIRIVs can be transformed into antigen-containing TIRIVs through addition of the antigen(s) duringthe reconstitution process. For this purpose, the antigen ofchoice was dissolved to the desired concentration, and thelyophilisate was reconstituted in an equal volume corre-sponding to its volume before lyophilisation. Short and gentlemixing resulted in antigen-TIRIVs suitable for immunisationstudies. For the induction of a strong CD8+ T cell response,the presence of the antigen in the lumen of the virosome isadvantageous, as the antigen will be delivered to the cytosol[2,7,10]. For this purpose soluble antigens without membraneanchor were used.
We first analysed the capacity of reconstituted TIRIVswith synthetic peptide epitopes to prime specific cytotoxic Tcells (CTL) in vivo. In a first set of experiments we used theHLA-A2 restricted hepatitis C virus derived peptide HCVCore 132 in HLA-A2.1 transgenic mice. We have previouslyshown that this epitope efficiently induces CTL responseswhen administered in chimeric virosomes [5]. Mice wereimmunised twice with lyophilised TIRIVs reconstituted withCore 132 peptide solution. Induction of CTL was evaluatedby quantifying the peptide-specific IFN�-producing CD8+ Tcells. Five out of six mice mounted a significant response with0.13–1.28% of the CD8+ cells specific for the peptide (Fig. 3,bar A). However, only two out of six mice showed significantnumbers of peptide-specific CD8+ T cells when immunisedwith lyophilised IRIVs without stabiliser that were reconsti-tuted with the same Core 132 peptide solution (Fig. 3, bar B).Mice did not mount a CD8+ T cell response when they wereimmunised with a mixture of IRIVs and Core 132 peptidesolution, indicating that peptide has to be encapsulated intothe virosomes (Fig. 3, bar C).
In addition we observed that HA is needed to induce a CTLresponse as HCV Core 132 liposomes with TC-chol were notimmunogenic but we could not show that fusogenic activity ofthe HA is a prerequisite to induce a cellular immune response(data not shown).
Extending on this, out of a selection of four HCV-derivedepitopes including Core 132 mixtures of two peptides –each formulated with TIRIVs – were used for immuni-sation. A significant CD8+ T cell response was inducedagainst all epitopes (mean values of the percentage ofIFN�-producing CD8+ T cells specific for one peptide:0.13–11.22%).
To estimate the dose response of peptide antigen deliveredwith TIRIVs, we have chosen the well-characterised ovalbu-min peptide OVA 257–264. Groups of six C57BL/6 mice eachwere immunised twice with TIRIVs reconstituted with differ-ent amounts of OVA peptide. Fig. 4A and C show percentageof peptide-specific IFN�-producing CD8+ T cells (Fig. 4A) aswell as the antigen-specific lysis at 100/1 E:T ratio (Fig. 4C).Immunisation with the OVA peptide in TIRIVs induced sig-nificant responses, and the lowest amount of peptide thatinduced both IFN�-producing cells as well as lytic activitywas 0.1 �g/dose, corresponding to approximately 0.1 nmolper dose.
We also investigated the capacity of TIRIVs to deliverprotein antigens. Analogous to peptide antigens, mice wereimmunised twice with TIRIVs reconstituted with differentamounts of OVA protein (44.3 kDa). These immunisationsinduced both CD8+ T cells producing IFN� (Fig. 4B) as wellas CD8+ T cells able to lyse E.G7.OVA target cells (EL4 cellstransfected with a plasmid coding for ovalbumin) (Fig. 4D).The lowest dose able to induce detectable responses with bothreadout systems was 10 �g OVA protein/dose correspondingto approximately 0.2 nmol per dose.
For the induction of a strong humoral immune response,the presence of the antigen on the outer surface of the viro-some is deemed optimal. Therefore, the antigen of choiceshould contain a lipid anchor or a transmembrane domain toassure an intercalation in the membrane. In general, there aretwo ways to prepare such antigen-TIRIVs. Addition of theantigen during the formulation process results in lyophilisedantigen-TIRIVs with a strong increase in stability if labileantigens have to be used. Alternatively, TIRIVs are reconsti-tuted with the antigen solution and active and homogenousvirosomes are formed having the antigen integrated in thevirosomal membrane.
AMA49-CPE, a malaria peptide antigen derived from thePlasmodium falciparum apical membrane antigen 1 (AMA-1) attached to a phospholipid anchor was used as modelantigen to test for induction of antibodies in mice [15,18,19].In ongoing tests, lyophilised AMA49-CPE-TIRIVs havebeen stable so far for at least 6 months at 4 ◦C and −20 ◦C
7072 A.R. Kammer et al. / Vaccine 25 (2007) 7065–7074
Fig. 4. CD8+ T cell response in mice immunised with ovalbumin peptide or protein in TIRIVs. C57Bl/6 mice (six mice/group) were immunised twicesubcutaneously at a 3-week interval with TIRIVs containing various doses of ovalbumin peptide (A and C) or protein (B and D). Spleen cells were isolated 2weeks after the second immunisation and the frequency of OVA 257-264 peptide-specific IFN�-producing CD8+ T cells (A and B), and antigen specific lysis(C and D) was analysed as described in methods. The percentage of peptide-specific cells was obtained by subtracting the percentage in samples stimulatedwith a non-relevant peptide from the percentage in samples stimulated with OVA 257–264 peptide. Specific lysis is shown at an E/T-ratio of 100:1. Meanvalues ± standard deviation are depicted.
and for at least 2 months at 25 ◦C, as seen by HPLC. WhenAMA49-CPE was used to reconstitute TIRIVs the amountof integrated peptide was quantified after gelfiltration orimmunoprecipitation by HPLC. For AMA49-CPE more than95% of AMA49-CPE was found associated with the viro-somes (data not shown).
We have previously demonstrated that AMA49-CPE-IRIVs induced parasite growth-inhibitory antibodies in mice[18]. We next investigated the capacity of reconstitutedAMA49-CPE-TIRIVs to induce a humoral immune response.Balb/c mice were immunised intramuscularly with differentvirosomal formulations with the AMA49-CPE peptide. Astrong antibody response against AMA49-CPE was detectedin an ELISA (Fig. 5). We compared TIRIVs contain-ing AMA49-CPE prior to and after lyophilisation withTIRIVs reconstituted with an aqueous solution of AMA49-CPE. The quantification of AMA49-CPE specific antibodiesrevealed no difference between the distinct formulationsand showed that the process of lyophilisation and recon-stitution had no negative effect on the immunogenicity ofthe stabilised virosomes. There was also no detectable dif-ference between AMA49-CPE linked to TIRIVs beforelyophilisation and TIRIVs reconstituted with the anti-gen solution. In addition, sera of mice immunised withAMA49-CPE-TIRIVs or with AMA49-CPE-IRIVs recog-nised AMA-1 in mature schizonts of P. falciparum parasitesin immunofluorescence staining equally well and resultedin the characteristic punctuate staining pattern (data notshown).
Fig. 5. Humoral immune response against AMA49-CPE in mice immu-nised with different virosomal formulations containing 2 �g antigen/dose.Mice were immunised twice intramuscularly at a 3-week interval withnon-lyophilised TIRIVs with the AMA49-CPE linked to the surface (�),lyophilised TIRIVs with the AMA49-CPE linked to the surface andreconstituted with water (�), or TIRIVs lyophilised without antigen andreconstituted with an AMA-49-CPE solution (�). Serum antibodies thatreacted with AMA49-CPE were detected in an ELISA. Shown is the meanvalue ± standard deviation (n = 10).
4. Discussion
Influenza virosomes represent an established vaccinedelivery system registered for human use with an excellentsafety profile. IRIVs have already been combined with dif-
Author's personal copy
A.R. Kammer et al. / Vaccine 25 (2007) 7065–7074 7073
ferent types of antigens such as proteins, peptides, nucleicacids or carbohydrates. In order to induce a significant CTLresponse, antigens have to be encapsulated into virosomes.Using conventional preparation methods this process is eitherinefficient or complex. Lyophilisation of virosomes followedby reconstitution with an antigen solution might be an alterna-tive to facilitate antigen encapsulation and to induce a strongcellular immune response. However, conventional IRIVs can-not be lyophilised without loss of the functional activity ofthe influenza hemagglutinin and immunogenicity.
In this study we report the development of stabilised viro-somes (TIRIVs) that can be reconstituted after lyophilisationwith an aqueous solution of antigen(s) to obtain vaccinesstrongly stimulating CD8+ CTLs. We found that immunisa-tion of mice with soluble peptide epitopes formulated withTIRIVs resulted in efficient induction of CD8+ T cells com-parable to the response seen with the previously describedchimeric virosomes [5]. However, the GMP manufacturingprocess of TIRIVs is easier and the preparation of the viroso-mal antigen formulation is faster. Moreover, a combinationof different antigens in the same vaccine is simple to preparewith the concept of TIRIVs. During TIRIV reconstitution afraction of the antigen-containing solution enters the lumenof the virosome. The amount of encapsulated antigen dependson the antigen type, the volume and the concentration ofthe solution used for reconstitution; up to 15% of the anti-gen was located inside the particles (data not shown). Theimmunogenicity of TIRIVs might be dependent on peptideentering the virosome, since a simple mixture of peptide solu-tion with previously described IRIVs was not immunogenicat all. However, immunogenicity of TIRIVs reconstitutedwith peptide antigen was comparable to IRIVs with encap-sulated peptide (CIRIVs) [5]. This finding is in accordancewith the proposed model of interaction between virosomesand the immune system where antigens encapsulated in viro-somes would preferably enter the MHC class I presentationpathway [2]. Immunogenicity of lyophilised and reconsti-tuted virosomes was strongly impaired if the virosomes wereprepared without stabiliser. This might be due to a loss offunctional properties, bigger size and/or very heterologoussize distribution.
Induction of CTL responses by peptide antigen loadedTIRIVs were found to be very efficient. Mice immu-nised with a minimal dose of 0.1 �g (approx. 0.1 nmol)of an ovalbumin-derived peptide in TIRIVs mounted aCTL response. Furthermore, proteins, too, can be efficientlydelivered by TIRIVs. Immunisation of mice with TIRIVsreconstituted with ovalbumin protein resulted in efficientinduction of CTLs at a minimal dose of 10 �g protein (approx.0.2 nmol).
TIRIVs were also potent inducers of humoral responsesagainst peptide antigens linked to their surface. TheAMA49-CPE peptide antigen derived from the Plasmod-ium falciparum apical membrane antigen 1 combined withTIRIVs were capable of mounting a strong antibody responseagainst the AMA49-CPE peptide, similar to previously
described AMA49-CPE-IRIVs that induced parasite growth-inhibitory antibodies in mice [18]. This response was notinfluenced by the lyophilisation and reconstitution processwhen the peptide was already added during the formulationprocess prior to lyophilisation. In addition, an identical anti-body response was observed when TIRIVs without peptidewere reconstituted with the AMA49-CPE peptide solu-tion and used to immunise mice. During the reconstitutionprocess, the peptide-PE conjugate was almost completelyintegrated into the virosomal membrane via its phospholipidmoiety.
The ability to lyophilise a virosomal vaccine has a greatimpact on the stability of the vaccine and reduces the need forthe cold chain requirements associated with liquid vaccines,which are costly and impractical for developing countries.This stabilizing effect was achieved by adding specific stabi-lizing molecules to virosomes like DC-chol or TC-chol, twopositively charged cholesterol derivatives. Although othermolecules were found to stabilise IRIVs at least partially, bestresults were obtained with lipid-based molecules containingsecondary or tertiary amines. These stabilisers were addedduring the formulation process and integrated into the viro-somal membrane during the reconstitution of the TIRIVs;addition of stabilisers to IRIVs after formulation did notresult in a stabilizing effect. The integration of stabilisermolecules resulted in virosomes with superior propertiesregarding lyophilisation and storage while conserving theirstructural and functional properties. These properties maybe a result of the protection of the influenza HA in an activestate by the stabilizing molecule during the freeze-drying pro-cess. While conventional IRIVs displayed a changed proteasesensitivity profile after lyophilisation, TIRIVs after lyophili-sation and reconstitution did show a profile comparable tountreated IRIVs. The stabilizing molecules also seemed tobe involved in retaining a very homogenous population ofparticles after reconstitution, presumably best achieved withcholesterol-based structures.
In conclusion, our novel type of influenza virosomes,TIRIVs, shows all the positive aspects previously describedfor virosomes with respect to fusion activity and immuno-genicity. But in contrast, TIRIVs can be lyophilised underGMP compliant conditions and at an industrial scale withoutchanging the functional and structural properties after recon-stitution. This innovative feature is a significant improvementof an already excellent vaccine carrier and accepted adju-vant for several reasons. Most importantly, TIRIVs inducestrong CTL and antibody responses against specific antigens,e.g. peptides, proteins or carbohydrates. These antigens canbe efficiently delivered with TIRIVs even in combinations.Depending on the desired type of immune response (cellularand/or humoral) antigens with or without membrane anchorcan be formulated alone or in combinations. At the same timedesired antigens can already be part of the lyophilisate, orthe lyophilisate can be reconstituted with the antigens at thedesired concentrations. Thus, TIRIVs offer a broad flexibil-ity for vaccine formulation. Finally, lyophilisation strongly
7074 A.R. Kammer et al. / Vaccine 25 (2007) 7065–7074
improves the stability of vaccines and renders them less sen-sitive to elevated storage temperatures. This is of particularimportance for vaccines for the developing world, where stor-age at subambient temperatures often cannot be consistentlymaintained. This extends the use of virosomes as adjuvantand carrier system to very labile antigen structures.
Therefore, TIRIVs provide a superior antigen deliveryvehicle with approved adjuvant effect, long-term storagecapability and broad versatility, enabling their use forthe induction of strong humoral and/or cellular immuneresponses.
Acknowledgments
We would like to thank Dr. F. Lemonnier and Dr. G.Inchauspe for the HHD transgenic mice and EL-4S3-RobHHD cell line, S. Rosenfellner, D. Oberholzer, Ch. Krueger,P. Coro and B. Neuhaus for excellent technical assistanceand Dr. Ch. Moser for inspiring discussions and critical read-ing of the manuscript. This work was partially supported bythe European Commission, Brussels, Belgium (QLRT-2002-01329).
References
[1] Gluck R, Metcalfe IC. Novel approaches in the development ofimmunopotentiating reconstituted influenza virosomes as efficient anti-gen carrier systems. Vaccine 2003;21(7–8):611–5.
[2] Huckriede A, Bungener L, Stegmann T, Daemen T, Medema J, PalacheAM, et al. The virosome concept for influenza vaccines. Vaccine2005;23(Suppl 1):S26–38.
[4] Zurbriggen R. Immunostimulating reconstituted influenza virosomes.Vaccine 2003;21(9–10):921–4.
[5] Amacker M, Engler O, Kammer AR, Vadrucci S, Oberholzer D, CernyA, et al. Peptide-loaded chimeric influenza virosomes for efficient invivo induction of cytotoxic T cells. Int Immunol 2005;17(6):695–704.
[6] Cusi MG, Terrosi C, Savellini GG, Di Genova G, Zurbriggen R,Correale P. Efficient delivery of DNA to dendritic cells mediated byinfluenza virosomes. Vaccine 2004;22(5–6):735–9.
[7] Felnerova D, Viret JF, Gluck R, Moser C. Liposomes and virosomesas delivery systems for antigens, nucleic acids and drugs. Curr OpinBiotechnol 2004;15(6):518–29.
[8] Liu X, Siegrist S, Amacker M, Zurbriggen R, Pluschke G, SeebergerPH. Enhancement of the immunogenicity of synthetic carbohydratesby conjugation to virosomes: a leishmaniasis vaccine candidate. ACSChem Biol 2006;1(3):161–4.
[9] Schumacher R, Amacker M, Neuhaus D, Rosenthal R, GroeperC, Heberer M, et al. Efficient induction of tumoricidal cytotoxicT lymphocytes by HLA-A0201 restricted, melanoma associated,L(27)Melan-A/MART-1(26-35) peptide encapsulated into virosomesin vitro. Vaccine 2005;23(48–49):5572–82.
[10] Westerfeld N, Zurbriggen R. Peptides delivered by immunostimulat-ing reconstituted influenza virosomes. J Pept Sci 2005;11(11):707–12.
[11] Zurbriggen R, Amacker M, Kammer AR, Westerfeld N, Borghgraef P,Van Leuven F, et al. Virosome-based active immunization targets sol-uble amyloid species rather than plaques in a transgenic mouse modelof Alzheimer’s disease. J Mol Neurosci 2005;27(2):157–66.
[12] Waelti ER, Gluck R. Delivery to cancer cells of antisense l-mycoligonucleotides incorporated in fusogenic, cationic-lipid-reconstitutedinfluenza-virus envelopes (cationic virosomes). Int J Cancer 1998;77(5):728–33.
[13] Gluck R. Influenza immunization. Biologicals 1997;25(2):221–5.[14] Loutan L, Bovier P, Althaus B, Gluck R. Inactivated virosome hepatitis
et al. Exploiting conformationally constrained peptidomimetics andan efficient human-compatible delivery system in synthetic vaccinedesign. Chembiochem 2001;2(11):838–43.
[16] Zurbriggen R, Novak-Hofer I, Seelig A, Gluck R. IRIV-adjuvantedhepatitis A vaccine: in vivo absorption and biophysical characterization.Prog Lipid Res 2000;39(1):3–18.
[17] Pascolo S, Bervas N, Ure JM, Smith AG, Lemonnier FA, Per-arnau B. HLA-A2.1-restricted education and cytolytic activity ofCD8(+) T lymphocytes from beta2 microglobulin (beta2m) HLA-A2.1monochain transgenic H-2Db beta2m double knockout mice. J ExpMed 1997;185(12):2043–51.
[18] Mueller MS, Renard A, Boato F, Vogel D, Naegeli M, Zurbriggen R,et al. Induction of parasite growth-inhibitory antibodies by a viroso-mal formulation of a peptidomimetic of loop I from domain III ofPlasmodium falciparum apical membrane antigen 1. Infect Immun2003;71(8):4749–58.
[19] Pfeiffer B, Peduzzi E, Moehle K, Zurbriggen R, Gluck R, PluschkeG, et al. A virosome-mimotope approach to synthetic vaccine designand optimization: synthesis, conformation, and immune recognitionof a potential malaria-vaccine candidate. Angew Chem Int Ed Engl2003;42(21):2368–71.
