© 2001 Wiley-Liss, Inc.
Cytometry 44:106 –112 (2001)
Green Fluorescent Protein (GFP)-Dependent Separation
of Bacterial Ghosts From Intact Cells by FACS
W. Haidinger,1* M.P. Szostak,1,2 W. Beisker,3 and W. Lubitz1
1
Institute of Microbiology and Genetics, University of Vienna Biocenter, Vienna, Austria
2
BIRD-C GmbH & Co. KEG, Vienna, Austria
3
National Research Center of Environment and Health, Neuherberg, Germany
Received 13 June 2000; Revision Received 28 February 2001; Accepted 28 February 2001
Background: E. coli and Salmonella ghost preparations,
produced by applying the PhiX174 protein E-mediated
lysis system, contain nonlysed bacteria at a very low per-
centage. To use the ghosts as vaccines, additional methods
have to be identified to remove any viable cell, to end up
in totally inactivated ghost fractions.
Materials and Methods: To increase the purity of ghost
fractions, we established a green fluorescent protein
(GFP)-dependent “in vivo staining” method to be com-
bined with the E-mediated lysis system. Several gfp expres-
sion vectors were constructed, and the corresponding
cellular fluorescence was analyzed. Bacterial fluorescence,
exclusively preserved in nonlysed cells, was utilized to
separate these cells from ghost preparations via flow cy-
tometric sorting.
Results: High-level production of GFP prior to induction
of the lysis system did not affect bacterial growth rates and
caused no inhibitory effects on the subsequent protein
E-mediated lysis of the cells. The population of reproduc-
Empty bacterial cell envelopes (ghosts) of a variety of
Gram-negative bacteria are under investigation as candi-
date vaccines (1,2). Upon expression of the cloned phage
␾X174 lysis gene E, a transmembrane tunnel structure
through the bacterial cell envelope is formed (3,4). Driven
by high osmotic pressure inside the cell, the cytoplasmic
content is expelled into the surrounding media, thus lead-
ing to an empty bacterial cell envelope. Due to the gentle,
nondenaturing production procedure, ghosts share func-
tional and antigenic determinants of the envelope with
their living counterparts. In animal experiments, ghosts
were able to induce both cellular and humoral immune
responses (5–7). Most vector-encoded gene E expression
systems are thermo-inducible by carrying the lethal gene E
under the control of the lambda promoters pL or pR (8).
Unlike Actinobacillus pleuropneumoniae (9) and
Vibrio cholerae (10), where E-mediated lysis completely
accomplishes the inactivation of the pathogens, E. coli or
Salmonella strains give ghost preparations, that still con-
tain small amounts of viable or nonlysed but inactivated
cells. Adequate methods to eliminate all viable bacteria
tive or inactivated but nonlysed cells was highly fluores-
cent at mean intensities 215-fold higher than ghosts,
which exhibited fluorescence at background level. Fluo-
rescent cells could effectively be separated from ghost
preparations via flow cytometric sorting. Cell sorting sub-
sequent to protein E-mediated lysis reduced the number of
viable cells within ghost preparations by a factor of
3 ⫻ 105.
Conclusions: The presented procedure is compatible
with the protein E-mediated lysis system, is highly effec-
tive in separation of nonlysed fluorescent cells, and may
serve as a prototype for ghost-purification in applications
where only a minimum number of viable cells within
ghost preparations can be tolerated. Cytometry 44:
106 –112, 2001. © 2001 Wiley-Liss, Inc.
Key terms: green fluorescent protein; GFPS65T; FACS;
flow cytometry; ghosts; purification; candidate vaccine;
PhiX174 gene E
within these ghost preparations have to be identified. The
classical procedures for inactivation of bacteria, such as
heat, radiation, or chemical treatment, may cause reduc-
tion of immunogenicity through destruction of critical
epitopes. Therefore, these methods were excluded for
further inactivation of the ghost preparation to preserve
the excellent immunogenic features of ghosts, which re-
sult from genetic inactivation by the protein E-mediated
lysis system. Methods to separate nonlysed cells from
ghosts, such as gradient centrifugation, were insufficient
and resulted in ghost fractions of only slightly increased
purity and accompanied by a substantial loss of ghost
material. Chemical indicators such as viability- or DNA-
stains could be used to selectively stain either the popu-
lation of ghosts or the nonlysed cells and to separate them
*Correspondence to: W. Haidinger, Institute of Microbiology and Ge-
netics, University of Vienna, Biocenter, Dr. Bohrgasse 9, A-1030, Vienna,
Austria.
E-mail: Haiding@gem.univie.ac.at
 E. COLI GHOST VACCINE PURIFICATION
via flow cytometric sorting. However, as all these com-
pounds have to be regarded as potentially harmful and
therefore as nonacceptable by the regulatory authorities
to be used for vaccine preparations, alternative and appli-
cable markers as criteria for flow cytometric sorting are
desired. Here we describe the use of gfp expression vec-
tors for the FACS-assisted purification of ghost preparations.
GFP, a highly fluorescent 27-kDa protein of 238 amino-
acid residues, emits green light (␭max, 509 nm) when
illuminated with either ultraviolet (␭max, 395 nm) or blue
light (␭max, 470 nm) (11,12). As a fluorescent marker
molecule, GFP has been used in a broad range of bacterial
cytometric applications. GFP has been applied as an ex-
pression marker in genetic studies dealing with the iden-
tification and the characterization of potential promoter
sequences of various bacterial strains (13–16). The fea-
tures of these sequences in terms of their potential capa-
bility to initiate the expression of gfp were determined via
flow cytometry. Pathogenic bacterial cells could be inves-
tigated in interaction with their eukaryotic host cells,
using GFP as a bacterial marker for flow cytometric anal-
yses (17). Clones of fluorescent pathogenic bacteria in
association with their host cells could be isolated by flow
cytometric sorting (18). To explore the safety of live
vaccines candidates, GFP has been used to trace the bac-
teria in the gastrointestinal tract of mice for microscopic
and flow cytometric analyses (19). In environmental mon-
itoring experiments, gfp expression in specific bacteria
has been used for the cytometric enumeration of their
total cell numbers (20).
In the present work, the cytoplasmic protein was used
for bacterial in vivo staining before induction of the pro-
tein E-mediated lysis system. As lysing cells expel all their
cytoplasmic content, including previously synthesized
GFP, only nonlysed bacteria should represent fluorescent
particles accessible to fluorescence-activated cell sorting
(FACS), separating intact cells from ghosts. To optimize
fluorescence, several expression vectors encoding gfps65t
(21) were constructed, and the resulting fluorescence
activities were compared. As no living material should be
present in a ghost candidate vaccine preparation, the
model study of GFP-dependent flow cytometric sorting of
bacteria and ghosts was used to evaluate the possibility of
ghost-vaccine purification.
MATERIALS AND METHODS
Plasmids and Bacterial Strains
The plasmids pML1 (22), pAW12 (23), pcI857 (24), pSU1
(25), and pKSII(⫹) (Stratagene, Amsterdam, The Nether-
lands) were described previously. pRH431, a pRSET B deriv-
ative (26) encoding the gfps65t gene, was kindly provided by
Karl Kuchler (Vienna Biocenter, Vienna, Austria). E. coli
KT914 (27), carrying a chromosomal copy of T7 gene 1
under transcriptional control of the lacUV5 promoter, and E.
coli NM522 (28) have also been described.
Growth Conditions
E. coli cells were routinely grown in LB broth (29), con-
taining 2% agar for solid media, supplemented with ampicil-
107
lin (200 ␮g/ml), kanamycin (50 ␮g/ml), or tetracycline (10
␮g/ml) where appropriate. For induction of lacPO, con-
trolled gfp expression IPTG in final concentrations of 0.2 mM
(solid media) or 5 mM (liquid media) was used. Bacteria
were incubated at either 28°C or 37°C for expression of
gfps65t and at 42°C for induction of lysis gene E. Growth and
lysis of bacteria were monitored by measuring the optical
density of liquid cultures at 600 nm (OD600). Bacteria or
ghosts prepared for FACS analysis were collected by centrif-
ugation, washed twice with Tris-buffered-saline (TBS) (30),
and resuspended in TBS or 0.9% NaCl solution.
Materials
Restriction enzymes, other DNA-modifying enzymes,
and buffers were obtained from either New England Bio-
labs (Schwalbach, Germany) or Roche (Vienna, Austria).
Enzymes were used as specified by the supplier. Plasmid
DNA and DNA fragments were prepared, analyzed, and
manipulated by standard procedures (30). For bacterial
filtration, membrane filters with a pore size of 0.2 ␮m
(Millipore, Vienna, Austria) were used.
Light Microscopy
For epifluorescence microscopy, the standard filter sets
(fluorescein-isothiocyanate observation filter sets) of the
research microscope Olympus AX70 were used.
FACS Analysis
A FACStar Plus flow cytometer (Becton Dickinson, Hei-
delberg, Germany), equipped with an argon-ion laser, was
used to measure forward angle light scatter, right angle
light scatter, and fluorescence of the cells. These param-
eters were acquired as pulse height signals (four decades
in logarithmic scale). The laser was tuned to 488 nm at a
power output of 500 mW, and the fluorescence emission
of GFPS65T was measured using a 530-nm band-pass filter
(BD 530 DF 30, Becton Dickinson). The recorded data files
were transferred by FACSNET software (Becton Dickin-
son) to an IBM-compatible PC system. For data analysis
and graphics, the Data Analysis System (DAS) software
package, version 4.38 (31), was used. For calibration of
the sorting module, standard FITC-labeled beads (2 ␮m;
Becton Dickinson) were used. Flow cytometric sorting
was performed by separating 300 fluorescent particles per
second from ghost preparations. Purified ghost fractions
were collected in 0.9% NaCl solution.
Construction of gfps65t Expression Vectors
Plasmid pRH431 (Fig. 1A), a pRSET B derivative, carries
the gfps65t gene under transcriptional control of the T7
gene 10 promoter. Subcloning gfps65t into the BamHI
site of pRSET B resulted in a gene fusion encoding a
polyhistidine leader peptide at the N-terminus followed by
the T7 gene 10 N-terminus, a tyrosine radioiodination site,
an enterokinase recognition site, and the gfps65t gene.
The gfp-specific ribosome binding site (rbs) has been re-
placed by the rbs of the phage T7 gene 10. Transcription is
driven by the T7 promoter and efficiently terminated by the
T7 gene 10 terminator sequence (32,33). Digestion of
108 HAIDINGER ET AL.

FIG. 1. gfps65t expression vectors. A: The 1.2-kb fragment derived from pRH431 was inserted into several plasmids, thus placing gfps65t under
transcriptional control of different promoter/repressor systems. The fragment encoded T7 expression signals as well as gfps65t gene fusion, consisting of
a polyhistidine leader peptide (H6), the T7 gene 10 amino-terminus (g10), a tyrosine radioiodination site (Y), an enterokinase recognition site (EK), and
the gfps65t gene itself (GFPS65T). gfp expression is controlled by the T7 promoter on pRH431 (A), the lac promoter on pKSGFP (B), the ␭PL promoter
on pGFPPL (C) or the ␭PR/cI1857 system on pGFPPR (D), respectively. term, T7 gene 10 transcriptional termination sequences; rbs, translational initiation
signals of T7 gene 10; colE1, origin of replication; f1, f1 filamentous phage origin of replication; Amp, ampicillin resistance gene; Tet, tetracycline resistance
gene; T7, promoter region of T7 gene 10; lacPO, promoter-operator region of the lac-operon; ␭PL, ␭PR, ␭PRM, leftward, rightward, and rightward
“maintenance” promoters of bacteriophage lambda; cI857, gene, encoding the thermo-sensitive repressor for ␭PR/PL promoters.

pRH431 (3.65 kb) with XbaI and NaeI resulted in a 1.2-kb
DNA-fragment, containing the complete gfps65t gene fusion
placed in between the T7-derived rbs and terminator se-
quences. This DNA-fragment was used for the construction
of all expression vectors described in this work.
Construction of Plasmid pKSGFP
Ligation of the 1.2-kb pRH431-fragment into the XbaI
and NaeI sites of pKSII(⫹) resulted in pKSGFP (3.85 kb),
placing gfps65t under lacPO control. (Fig. 1B).
Construction of Plasmid pGFPPL
Insertion of the 1.2-kb pRH431-fragment into the XbaI
and NruI sites of pSU1 placed the gfps65t gene fusion
under control of the ␭PLpromoter of the resulting plasmid
pGFPPL (5.1 kb; Fig. 1C).
Construction of Plasmid pGFPPR
Restriction of pAW12 with XbaI and HpaI, purification
of the 4.8-kb DNA fragment, and ligation with the 1.2-kb
pRH431 fragment led to pGFPPR (6 kb). The ␾X174 gene
E of pAW12, thereby, was replaced by the pRH431 frag-
ment, placing gfps65t under transcriptional control of the
␭PR/cI857 elements (Fig. 1D)
“In Vivo” Staining of Bacterial Cells, Preparation of
Ghosts, and Methods of Analyzing Liquid Fractions
E. coli NM522 was cotransformed with pML1 and pKS-
GFP and grown at 28°C in LB supplemented with ampi-
cillin and kanamycin in the presence of 5 mM IPTG to
induce GFPS65T synthesis. For production of ghosts, the
culture was split and one part was shifted to 42°C to
 E. COLI GHOST VACCINE PURIFICATION
FIG. 2. Flow cytometric analysis of
bacterial fluorescence derived from
different gfps65t expression plas-
mids. The E. coli strain NM522 was
transformed with the plasmids pKS-
GFP (lacPO control), pGFPPR (␭PR/
cI857 control), or pGFPPL (␭PL con-
trol; cotransformed with the plasmid
pcI857), while the E. coli strain
KT914 was transformed with the ex-
pression vector pRH431. The cellular
fluorescence is presented by mean in-
tensities (represented by bars) given
in arbitrary units (Green fluorescence
int. (a.u.) and the standard deviations
of fluorescence intensities (⵫E⵬). The
data sets for each combination of
plasmid/strain are derived from sin-
gle, individual flow cytometric runs
with regard to 10,000 analyzed bacte-
ria.
induce E-lysis, whereas the other part of the culture,
grown at 28°C, was used as control. When no further
decrease of OD600 of the lysis culture could be ob-
served, the cells were harvested by centrifugation (15
min, 10,000g; 4°C), washed twice with 30 ml TBS,
resuspended in 2 ml TBS, and stored at 4°C. Prior to
flow cytometry, the samples were diluted 1:100 in TBS.
FACS-purified ghost fractions were collected in 0.9%
NaCl solution. The number of colony-forming units
(CFU) in nonpurified ghost preparations was deter-
mined by colony counting after growth on agar plates at
28°C for 48 h. The number of total cells in unsorted
fractions was determined using counting chambers.
Small numbers of viable cells in large volumes were
quantified by means of vacuum extraction onto sterile
membrane filters, which were put on nonselective LB-
agar plates and incubated at 28°C. CFU values were
ascertained after a growth period of 48 h.
RESULTS
gfps65t Expression Vectors and Bacterial
Fluorescence
For optimal GFP production in E. coli, three gfps65t
expression vectors (pKSGFP, pGFPPL, and pGFPPR) were
constructed, expressing the gfps65t gene under the con-
trol of the lacPO-promoter, the ␭PL-promoter, and the
␭PR/cI857 system, respectively (Fig. 1). These vectors
were transformed into E. coli NM522. Plasmid pGFPPL
was cotransformed with pcI857, providing constitutive
expression of the cI857 gene in order to ensure the re-
pression of the ␭PLpromoter. The parental plasmid
pRH431, carrying the gfps65t gene under transcriptional
control of the T7 promoter, was transformed into E. coli
KT914, as this strain carries a chromosomal copy of T7
gene 1 encoding the T7 RNA polymerase. Transformants
harboring either pGFPPL or pGFPPR were grown in liquid
109
culture at 37°C to trigger ␭PL/␭PR,-controlled gene expres-
sion. Transformants harboring pRH431 or pKSGFP were
grown at 28°C, and expression of gfps65t derived from
pKSGFP was induced with IPTG. Cells were harvested and
their fluorescence intensity was quantified by flow cyto-
metric analysis (Fig. 2).
Expression of gfps65t derived from pRH431 resulted in
the highest intensity of fluorescence, with mean signals
800-fold higher than background. PKSGFP- or pGFPPL-
harboring cells were observed to be highly fluorescent at
mean intensities 215-fold and 170-fold higher, respec-
tively, than background. However, the pGFPPL-derived
fluorescence appeared to be less homogeneous compared
to pKSGFP. With the gfps65t gene under transcriptional
control of the lambda pR promoter, very weak fluores-
cence could be detected at mean values 52-fold higher
than background. Moreover, gfps65t continuously ex-
pressed at high levels in E. coli caused no cytotoxic effects
by means of growth stagnation or limitation of viability
(data not shown). However, though the growth rates
were not affected, gfps65t expression apparently influ-
enced the cell shape of E. coli. After synthesis of GFPS65T,
E. coli cells became elongated, and a small percentage of
filamentous cells (⬍1%) could be detected. These mor-
phological changes, however, caused no inhibitory effect
on protein E-mediated lysis. Micrographs taken after lysis
showed two distinct populations: one consisted of com-
pletely lysed cells that were not detectable using epifluo-
rescence microscopy; the other population was com-
posed of cytoplasm-containing fluorescent bacteria (Fig.
3). The lac-promoter-controlled gfp expression system
(pKSGFP) was considered most favorable, as this system is
independent of the presence of a host- or plasmid-en-
coded T7 RNA polymerase and more generally applicable,
as the side effects of gfp expression mentioned above
were restricted to a minor fraction of the bacterial popu-
lation.
110 HAIDINGER ET AL.

FIG. 3. Corresponding micrographs

of E. coli NM522 (pML1, pKSGFP)
after lysis, taken under conditions of
phase contrast or fluorescence. Non-
lysed, cytoplasm-containing cells are
the only particles within ghost prep-
arations that exhibit fluorescence.

Protein E-Mediated Lysis in Combination With
Flow Cytometric Sorting
For production of ghosts, E. coli NM522 (pKSGFP) was
cotransformed with the lysis plasmid pML1 and grown at
28°C in the presence of IPTG for continuous production
of GFP. After reaching mid-log phase, the cultures where
shifted to 42°C to induce gene E expression and harvested
after the lysis process was completed. The ghosts, which
appeared as translucent envelope structures as the result
of loss of the cytoplasmic content, could clearly be dis-
criminated from the occasionally detectable nonlysed
cells containing cytoplasm. Epifluorescence microscopy
revealed that these nonlysed cells were highly fluorescent
in contrary to the population of ghosts, which were not
detectable under these conditions (Fig. 3). By comparing
the number of total cells with the number of colony-
forming units (Table 1), the ratio of viable cells within
ghost preparations was calculated as 1.3%, whereas the
quantity of fluorescent cells within ghost preparations
was determined as 2.5%, indicating the presence of non-
reproductive but nonlysed cells. By flow cytometric anal-
ysis, the nonlysed cells could be well-discriminated from
ghosts, as the mean intensity of fluorescence was 215-fold
higher for the former than for the latter population, which
exhibited fluorescence at background level, and only a
few signals were detected bridging background to bright
fluorescence (Fig. 4).
Four samples containing 107 bacterial particles were
sorted, the ghost fractions were analyzed by microscopy,
and the numbers of total and viable cells were deter-
mined. The purification of ghost preparations was per-
formed at a separation rate of 300 fluorescent particles per
second. As bacterial cell envelopes were nonfluorescent,
sorting gates were established to eliminate all particles
with intensities of fluorescence higher than background,
including nonlysed and partially lysed cells as well as

Table 1
Comparison Between Purity Grades of Ghost Preparations Before and Subsequent to
Flow Cytometric Sorting*

Total cells Fluorescent cells CFU Total cells:CFU

Lysis 1 ⫾ 0.35 ⫻ 107 2.5 ⫾ 0.27 ⫻ 10 5
1.3 ⫾ 0.17 ⫻ 105
8.5 ⫾ 2.8 ⫻ 101
Sort 8.4 ⫾ 2.8 ⫻ 106 3 ⫾ 1 ⫻ 100 3.1 ⫾ 1.2 ⫻ 101 3 ⫾ 1.1 ⫻ 105
*Numbers of total cells, fluorescent cells, numbers of colony-forming units (CFU), and
ratios of total cell number to number of viable bacteria (Total cells:CFU) were determined at
two different time points: immediately after the lysis process stopped (Lysis), and subsequent
to flow cytometric purification (Sort). Numbers are mean average values and standard
deviations derived from data of four ghost preparations. Total cells:CFU values were first
calculated for each ghost preparation separately and subsequently summarized through
statistical evaluation.
 E. COLI GHOST VACCINE PURIFICATION
FIG. 4. Flow cytometric analysis of a ghost preparation derived from E.
coli NM522 (pML1, pKSGFP). Bacteria were detected by right-angle light
scatter (Side scatter) as well as green fluorescence. A distinct population
of highly fluorescent bacteria can clearly be discriminated from ghosts,
showing fluorescence at background level. Green fluorescence int. (a.u.),
intensity of green fluorescence per cell given in arbitrary units.
detectable cell aggregates. The sorting process was ac-
companied by a recovery of ghosts, accounting for
roughly 80% in the purified fraction (Table 1). However,
the sorting process did not remove all viable cells, leaving
31 reproductive bacteria within sorted fractions consist-
ing of 8.4 ⫻ 106 particles (mean values). The ratios of total
cells:CFU in unsorted and purified fractions were calcu-
lated by comparing the number of total cells (including
ghosts) to the number of viable bacteria, determined at
two different time points, immediately after lysis and sub-
sequent to flow cytometric sorting. The ratio of one living
cell per 85 ghosts of the unsorted samples (after lysis)
could be reduced to one in 3 ⫻ 105 ghost after purifica-
tion (Table 1). Thus the increase of purity, in terms of
absence of viable cells, could be calculated as 3,500-fold.
In addition to the presented procedure, the method of
flow cytometric sorting of bacteria which exhibit gene
expression detectable by fluorescence.
DISCUSSION
The GFPS65T protein seems to be the appropriate
marker molecule for flow cytometric sorting of cytoplasm-
containing bacteria because of its single excitation peak at
490 nm, which perfectly fits with the wavelength emitted
by the argon laser of FACS machines.
By comparing the results from sort-analyses with the
microscopic examinations of the purified fractions, the
GFP-based sorting procedure can be characterized as
highly efficient, since purification led to ghost fractions
almost devoid of fluorescent particles and with strongly
111
decreased numbers of viable bacteria. Some of the remain-
ing viable cells within sorted samples, however, might be
nonfluorescent and therefore FACS-undetectable, as only
up to two green fluorescent cells could be microscopi-
cally observed in half of the sorted ghost fractions, as
harvested by centrifugation (Table 1). However, previous
studies showed that elevated temperatures can cause GFP-
misfolding (34). In agreement with these studies, we ob-
served a dramatic decline of fluorescence in E. coli NM522
(pKSGFP) at growth temperatures higher than 28°C, lead-
ing to nearly complete loss of fluorescence in overnight
cultures incubated at 37°C. Hence we propose that non-
fluorescent but viable cells, that escaped the E-mediated
lysis process, first emerged after the temperature upshift
for lysis induction to 42°C. Support for this hypothesis
comes from cytometric analysis by the rare detection of
bacteria with decreased cellular fluorescence. This popu-
lation probably consists of bacteria containing various
portions of active and misfolded GFPS65T per cell. Mis-
folding of GFPS65T during synthesis at higher tempera-
tures (37°C) might also explain the weak fluorescence
observed with pGFPPL and pGFPPR, where the gfp ex-
pression systems are controlled by either the lambda pL or
pR promoter combined with the thermosensitive CI857
repressor.
Although flow cytometry was able to purify ghosts from
nonlysed cells in a single purification step by more than
three magnitudes, the presence of nonlysed cells in the
ghost preparation would remain a problem in vaccine
preparations. As most of the remaining reproductive cells
are nonfluorescent, a further purification of the ghost
fraction by FACS could reduce the number of survivors
only insufficiently, and would also be accompanied by a
considerable loss of material and is a time-consuming
process for sorting huge ghost/cell numbers. The method
could have an advantage where small numbers of ghosts
are needed and/or where a minor fraction of living cells
can be tolerated in the ghost pool. Other methods for
removing fluorescent cells, e.g., photoinactivation of fluor-
escent bacteria by destruction of the cells through high-
energy laser pulses (photodynamic “zapping”), also seem
not to be suitable for improving the purity of the ghost
fraction considerably, because of the remaining problem
of nonfluorescent, nonlysed cells. However, as tempera-
ture-dependent misfolding of GFP might be crucial for the
presented purification procedure, it sounds feasible to
reduce the number of nonfluorescent survivors in un-
sorted fractions by using thermostable variants of GFP
(34). Proper folding of the protein at temperatures up to
42°C could prevent a potential decrease in bacterial fluo-
rescence during ghost production. A higher number of
nonlysed cells could therefore become detectable by
FACS, because of the elevated amount of proper folded,
active GFP per cell.
In reversion of our task, the method of flow cytometric
sorting of bacteria which exhibit gene expression detect-
able by fluorescence might be very useful to separate
pools of cells with different intensities of fluorescence in
studies of mutation analyses of target genes or expression
112 HAIDINGER ET AL.
systems. As the viability of the bacteria was not impaired
by the sorting process, sorting of bacteria from commu-
nities might be feasible with features of expressing pro-
teins or components providing signals detectable for the
FACS machine.
LITERATURE CITED
1. Lubitz W, Witte A, Eko FO, Kamal M, Jechlinger W, Brand E, Marchart
J, Haidinger W, Huter V, Felnerova D, Stralis-Alves N, Lechleitner S,
Melzer H, Szostak MP, Resch S, Mader H, Kuen B, Mayr B, Mayrhofer
P, Geretschlager R, Haslberger A, Hensel A. Extended recombinant
bacterial ghost system. J Biotechnol 1999;73:261–273.
2. Eko FO, Witte A, Huter V, Kuen B, Furst-Ladani S, Haslberger A,
Katinger A, Hensel A, Szostak MP, Resch S, Mader H, Raza P, Brand E,
Marchart J, Jechlinger W, Haidinger W, Lubitz W. New strategies for
combination vaccines based on the extended recombinant bacterial
ghost system. Vaccine 1999;17:1643–1649.
3. Witte A, Wanner G, Lubitz W. Dynamics of PhiX174 protein E-medi-
ated lysis of Escherichia coli. Arch Microbiol 1992;157:381–388.
4. Schön P, Schrot G, Wanner G, Lubitz W, Witte A. Two-stage model for
integration of lysis protein E of bacteriophage PhiX174 into the cell
envelope of Escherichia coli. FEMS Microbiol Rev 1995;17:207–212.
5. Hensel A, Huter V, Katinger A, Raza P, Strnistschie C, Roesler U,
Brand E, Lubitz W. Intramuscular immunization with genetically in-
activated (ghosts) Actinobacillus pleuropneumoniae serotype 9 pro-
tects pigs against homologous aerosol challenge and prevents carrier
state. Vaccine 2000;18:2945–2955.
6. Haslberger AG, Kohl G, Felnerova D, Mayr UB, Furst-Ladani S, Lubitz
W. Activation, stimulation and uptake of bacterial ghosts in antigen
presenting cells. J Biotechnol 2000;83:57– 66.
7. Eko FO, Mayr UB, Attridge SR, Lubitz W. Characterization and immu-
nogenicity of Vibrio cholerae ghosts expressing toxin-coregulated
pili. J. Biotechnol 2000;83:115–123.
8. Jechlinger W, Szostak MP, Witte A, Lubitz W. Altered temperature
induction sensitivity of the lambda pR/cI857 system for controlled
gene E expression in Escherichia coli. FEMS Microbiol Lett 1999;173:
347–352.
9. Hensel A, van Leengoed LAG, Szostak M, Windt H, Weissenböck H,
Stockhofe-Zurwieden N, Katinger A, Stadler M, Ganter M, Bunka S,
Papst R, Lubitz W. Induction of protective immunity by aerosol
application of candidate vaccines in a dose-controlled pig aerosol
infection model. J Biotechnol 1996;44:171–181.
10. Eko FO, Hensel A, Bunka S, Lubitz W. Immunogenicity of Vibrio
cholerae ghosts following intraperitoneal immunization of mice. Vac-
cine 1994;12:1330 –1334.
11. Morin JG, Hastings JW. Energy transfer in a bioluminescent system.
J Cell Physiol 1971;77:313–318.
12. Ward WW, Cody CW, Hart RC, Cormier MJ. Spectrophotometric
identity of the energy transfer chromophores in Renilla and Ae-
quorea green fluorescent proteins. Photochem Photobiol 1980;31:
611– 615.
13. Valdivia RH, Falkow S. Bacterial genetics by flow cytometry: rapid
isolation of Salmonella typhimurium acid-inducible promoters by
differential fluorescence induction. Mol Microbiol 1996;22:367–378.
14. Dehio M, Knorre A, Lanz C, Dehio C. Construction of versatile
high-level expression vectors for Bartonella henselae and the use of
green fluorescent protein as a new expression marker. Gene 1998;
215:223–229.
15. Kohler S, Ouahrani-Bettache S, Layssac M, Teyssier J, Liautard JP.
Constitutive and inducible expression of green fluorescent protein in
Brucella suis. Infect Immun 1999;67:6695– 6697.
16. Kohler R, Bubert A, Goebel W, Steinert M, Hacker J, Bubert B.
Expression and use of the green fluorescent protein as a reporter
system in Legionella pneumophila. Mol Gen Genet 2000;262:1060 –
1069.
17. Kremer L, Baulard A, Estaquier J, Poulain-Godefroy O, Locht C. Green
fluorescent protein as a new expression marker in mycobacteria. Mol
Microbiol 1995;17:913—922.
18. Valdivia RH, Hromockyj AE, Monack D, Ramakrishnan L, Falkow S.
Applications for green fluorescent protein (GFP) in the host-pathogen
interactions. Gene 1996;173:47—52.
19. Geoffroy MC, Guyard C, Quatannens B, Pavan S, Lange M, Mercenier
A. Use of green fluorescent protein to tag lactic acid bacterium strains
under development as live vaccine vectors. Appl Environ Microbiol
2000;66:383—391.
20. Unge A, Tombolini R, Molbak L, Jansson JK. Simultaneous monitoring
of cell number and metabolic activity of specific bacterial populations
with a dual gfp-luxAB marker system. Appl Environ Microbiol 1999;
65:813– 821.
21. Heim R, Cubitt AB, Tsien RY. Improved green fluorescence. Nature
1995;373:663– 664.
22. Szostak MP, Hensel A, Eko FO, Klein R, Auer T, Mader H, Haslberger
A, Bunka S, Wanner G, Lubitz W. Bacterial ghosts: non-living candi-
date vaccines. J Biotechnol 1996;44:161–170.
23. Witte A, Lubitz W. Biochemical characterization of PhiX174 protein
E-mediated lysis of Escherichia coli. Eur J Biochem 1989;180:393–
398.
24. Remaut E, Tsao H, Fiers W. Improved plasmid vectors with a ther-
moinducible expression and temperature-regulated runaway replica-
tion. Gene 1981;22:103–113.
25. Schüller A, Harkness RE, Rüther U, Lubitz W. Deletion of C-terminal
amino acid codons of PhiX174 gene E: effect on its lysis inducing
properties. Nucleic Acids Res 1985;13:4143– 4153.
26. Kroll DJ, Abdel-Malek Abdel-Hafiz H, Marcell T, Simpson S, Chen C,
Gutierrez-Hartmann A, Lustbader JW, Hoeffler JP. A multifunctional
procaryotic protein expression system: overproduction, affinity puri-
fication, and selective detection. DNA Cell Biol 1993;12:441– 453.
27. Tedin K, Altanerova U, Bläsi U. Construction of a transducible cas-
sette encoding T7 RNA polymerase gene 1 in Escherichia coli. J
Microbiol Methods 1995;21:173–180.
28. Gough J, Murray N. Sequence diversity among related genes for
recognition of specific targets in DNA molecules. J Mol Biol 1983;
166:1–19.
29. Miller JH. Experiments in molecular genetics. Cold Spring Harbor,
NY: Cold Spring Harbor Laboratory Press; 1972.
30. Sambrook J, Fritsch EF, Maniatis T. Molecular cloning: a laboratory
manual, 2nd ed. Cold Spring Harbor, NY: Cold Spring Harbor Labo-
ratory Press; 1989.
31. Beisker W. A new combined integral-light and slit-scan data analysis
system (DAS) for flow cytometry. Comput Methods Programs Biomed
1994;42:15–26.
32. Studier FW, Moffatt BA. Use of bacteriophage T7 polymerase to direct
selective high level expression of cloned genes. J Mol Biol 1986;189:
113–130.
33. Rosenberg AH, Lade BN, Chui D, Lin S, Dunn, JJ, Studier FW. Vectors
for selective expression of cloned DNAs by T7 RNA polymerase.
Gene 1987;56:125–135.
34. Siemering KR, Golbik R, Sever R, Haseloff J. Mutations that suppress
the thermo-sensitivity of green fluorescent protein. Curr Biol 1996;
6:1653–1663.