RES E AR C H Open Access
In silico prediction of Gallibacterium anatis
pan-immunogens
Ragnhild J Bager
1
, Egle Kudirkiene
1
, Isabelle da Piedade
1
, Torsten Seemann
2
, Tine K Nielsen
3
, Susanne E Pors
1
,
Andreas H Mattsson
4,5
, John D Boyce
6
, Ben Adler
7
and Anders M Bojesen
1*
Abstract
The Gram-negative bacterium Gallibacterium anatis is a major cause of salpingitis and peritonitis in commercial
egg-layers, leading to reduced egg production and increased mortality. Unfortunately, widespread multidrug
resistance and antigenic diversity makes it difficult to control infections and novel prevention strategies are urgently
needed. In this study, a pan-genomic reverse vaccinology (RV) approach was used to identify potential vaccine
candidates. Firstly, the genomes of 10 selected Gallibacterium strains were analyzed and proteins selected on the
following criteria; predicted surface-exposure or secretion, none or one transmembrane helix (TMH), and presence
in six or more of the 10 genomes. In total, 42 proteins were selected. The genes encoding 27 of these proteins
were successfully cloned in Escherichia coli and the proteins expressed and purified. To reduce the number of
vaccine candidates for in vivo testing, each of the purified recombinant proteins was screened by ELISA for their
ability to elicit a significant serological response with serum from chickens that had been infected with G. anatis.
Additionally, an in silico prediction of the protective potential was carried out based on a protein property
prediction method. Of the 27 proteins, two novel putative immunogens were identified; Gab_1309 and Gab_2312.
Moreover, three previously charact erized virulence factors; GtxA, FlfA and Gab_2156, were identified. Thus, by
combining the pan-genomic RV approach with subsequent in vitro and in silico screening, we have narrowed
down the pan-proteome of G. anatis to five vaccine candidates. Importantly, preliminary immunization trials
indicated an in vivo protective potential of GtxA-N, FlfA and Gab_1309.
Introduction
Poultry meat and eggs are considered very important
and sustainable sources of animal protein worldwide [1],
making efficient strategies to prevent and control the
spread of poultry diseases highly important [2]. Gallibac-
terium anatis is a Gram-negative, non-motile, encapsu-
lated coccobacillus of the Pasteurellaceae family [3,4]
and commonly associated with poultry [5]. Besides con-
stituting a part of the normal microflora of the upper re-
spiratory tract and lower genital tract in chickens [6], it
is also considered a major cause of salpingitis and peri-
tonitis in egg-laying chickens [7-9]. Thus, G. anatis in-
fections lead to a drop in egg production and increased
mortality in commercial layers [10]. Unfortunately, wide-
spread multiple-drug resistance [11] hinders treatment
with traditional antimicrobial agents, while substantial
antigenic diversity [12] among disease-causing field iso-
lates hampers disease prevention by classical vaccination
with inactivated whole cell bacterins. Hence, novel pre-
vention strategies are urgently needed.
The sequencing of the first bacterial genome in 1995
[13] initiated the genomic era and catalyzed a shift from
conventional culture-based approaches to genome-based
vaccinology [14 ]. This gave rise to the Reverse Vaccinol-
ogy (RV) approach [15], in which bioinformatics tools
are used to analyze genome sequences to identify genes
encoding likely protective antigens. The concept of RV
was initially applied to Neisseria meningitidis serogroup
B (MenB) [16], for which conventional vaccine develop-
ment approaches had failed in producing an efficacious
vaccine. Based on the genomic sequence of MenB strain
MC58 [17], five universal vaccine candidates were iden-
tified [18], and the resulting 4CMenB vaccine (Bexsero®)
is now approved in the EU for active immunization of
individuals aged over two months against disease caused
* Correspondence: miki@sund.ku.dk
1
Department of Veterinary Disease Biology, Faculty of Health and Medical
Sciences, University of Copenhage n, 1870 Frederiksberg C, Denmark
Full list of author information is available at the end of the article
VETERINARY RESEARCH
© 2014 Bager et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain
Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article,
unless otherwise stated.
Bager et al. Veterinary Research 2014, 45:80
http://www.veterinaryresearch.org/content/45/1/80
by MenB [19]. Since this pioneering MenB project, the
RV approach has been applied to a variety of other import-
ant pathogens [20]. However, the increased availability of
multiple genomes for the same bacterial species has shown
that genomic variability in bacteria is much more extensive
than initially anticipated. Thus, analysis of the genome of a
single strain often fails to address intra-species genetic vari-
ability and limits the effectiveness of genome-wide screens
for vaccine candidates. To overcome this, a pan-genomic
RV model utilizing the global gene repertoire for a species
was proposed by Tettelin et al. [21]. Pan-genomic RV was
first applied to vaccine development in Group B Strepto-
coccus [22], and this study demonstrated the importance of
sequencing multiple strains of a single pathogen for the
identification of vaccine antigens [23]. The application of
in silico and in vitro predictions has not only enabled a
much more rational selection of vaccine candidates, but
has also shown promise at reducing the number of experi-
mental animals needed to verify the effectiveness of vac-
cine candidates.
We report here the use of a pan-genomic RV approach
for identification of novel and conserved immunogens of
G. anatis. By implementing different in silico approaches
and in vitro assays, we screened the Gallibacterium pan-
proteome, resulting in a final sele ction of five proteins
with a high predicted potential as vaccine candidates.
Importantly, preliminary in vitro immunization results
indicate protective potential of at least three of these
candidates including FlfA, which has previously been
tested and confirmed highly protective against homolo-
gous challenge in chickens [24]. Together, these results
provide an important step in the development of a new
and broadly protective vaccine against G. anatis.
Materials and methods
Animal ethics stateme nt
All work on experimental animals was carried out with the
approval of the Danish National Animal Ethics Committee
(Approval no. 2012-15-2934-00339 and 2012-15-2934-
00923).
Gallibacterium strains and growth conditions
The 10 Gallibacterium strains included in the study are
listed in Table 1. The strains were selected based on
their pathogenic potential, prevalence in the field and
genetic diversity, in order to provide as much diversity
as possible within the species. Bacteria were incubated at
37 °C on brain heart infusion (BHI) agar supplemented
with 5% citrated bovine blood in a closed plastic bag or
in BHI broth with aeration.
RV in silico analysis
Sequencing and assembly of G. anatis strains 12656–12
and F149
T
was performed a s described previously [27].
The remainin g genomes were sequenced using the Illu-
mina Genome Analyzer IIx (CD genomics, New York,
USA). Reads were assembled using VelvetOptimiser 2.0
[28]. All 10 genomes were annotated using Prokka v.1.0
[29]. The subcellular localization of the 31 564 anno-
tated proteins from the 10 genomes was predicted using
standalone PSORTb v.3.0 [30]. The presence of N-
terminal signal peptides was predicted using SignalP
Table 1 Gallibacterium strains for included in this study
Strain Biovar Host/tissue Lesions Country
a
Reference
G. anatis bv. anatis
Biovar anatis
F149
T
NA Duck/intestine - DK [3]
Biovar haemolytica
12656-12 4 Chicken/liver + DK [
25]
10672-6 1 Chicken/oviduct + DK [
3]
4895 4 Chicken/NA + MX [
26]
7990 3 Chicken/NA + MX [
26]
Avicor 4 Chicken/heart + MX [
26]
CCM5995 20 Chicken/NA NA Cz [
3]
IPDH 697-78 15 Chicken/NA + G [
3]
G. genomospecies 1
CCM5974 8 Hen/Liver + Cz [
3]
G. genomospecies 2
CCM5976 9 Hen/oviduct + Cz [
3]
a
Cz: Czech Republic, DK: Denmark, G: Germany, MX: Mexic o.
T
= type strain.
NA = Not available.
Bager et al. Veterinary Research 2014, 45:80 Page 2 of 12
http://www.veterinaryresearch.org/content/45/1/80
v.3.0 [31] and the number of transmembrane helices
(TMHs) was predicted by TMHMM v.2.0 [32]. The pro-
tein conservation among the strains was analyzed using
BLASTp v.2.2.22 [33] with default parameters. The the-
oretical molecular masses and isoele ctric points were
calculated using the pepstats tool in EMBOSS [34]. In
total, 42 proteins were selected (Table 2).
Cloning and small-scale protein expression
Each of the selected genes was amplified from the G.
anatis 12656–12 genome by PCR and cloned into the
Gateway entry vector pENTR™/SD/D-TOPO (Invitrogen).
Primers were designed using Oligo Explorer 1.2 (Gene
Link™, Hawthorne, NY, USA) as described previously [37].
Areas with high predicted hydrophobicity in the N and/or
C terminus were removed, as were predicted signal pep-
tides. In addition, GtxA was cloned as two parts (N- and
C-terminal) due to its size. The E. coli strains and plas-
mids used in this study are listed in Additional file 1.
Genes were cloned and small-scale expressed as described
in Additional file 2 using Gateway cloning and ligation-
independent cloning (LIC) systems. Altogether, 37 expres-
sion clones were constructed for 36 of the 42 selected pro-
teins (two clones were made for GtxA). The primer
sequences used for gene amplification, and the final ex-
pression vector chosen for protein expression from each
gene, are listed in Additional file 3.
Large-scale protein exp ression and purification
All proteins were expressed and purified in large-scale
from E. coli Rosetta 2 (DE3) cells (Novagen, Madison, WI,
USA). Large-scale expression was performed in a custom-
made large-scale expression system (LEX) (Harbinger Bio-
tech, Toronto, Canada) as described previously [38] and in
Additional file 2. Of the 37 expression clones, 27 recom-
binant proteins were successfully purified; the majority
(17) of these proteins had a purity > 90%.
Production of antiserum against G. anatis 12656–12 in
chickens
Two Lohmann brown chickens (21 weeks old) were pur-
chased from a commercial breeder with high biosecurity
standards. The chickens were kept under free indoor
housing conditions and were provided with fresh water
and feed ad libitum. The chickens were swabbed for the
presence of G. anatis by a cloacal swab. After two weeks
of acclimatization the chickens were challenged with 10
5
colony forming unit s (CFU) of G. anatis 12656–12 by
injection into the peritoneal cavity as previously de-
scribed [39] and re-infected 2 weeks after the first infec-
tion. Blood for serum purification was collected from
the brachial vein prior to the first infe ction (pre-immune
antiserum) and one week after the second infection
(hyper-immune antiserum).
Enzyme-Linked Immunosorbent Assay (ELISA)
The putative immunogenicity of each of the purified re-
combinant proteins was assessed by indirect ELISA as
described previously [40], using pooled anti-G. anatis
pre-immune and hyper-immune antiserum. Briefly, Nunc-
Immuno™ MicroWell™ 96-Well Plates (Thermo Scientific,
Waltham, MA, USA) were coated overnight at 4 °C with
0.5 μg recombinant protein (48 wells per protein) diluted
in carbonate-bicarbonate buffer (pH 9.6) (Sigma-Aldrich,
St. Louis, MO, USA). Each well was then washed; this and
all subsequent washing steps consisted of three washes in
350 μL wash buffer (PBS + 0.05% Tween 20). The wells
were blocked for 2 h at room temperature in 200 μL
blocking solution (PBS containing 0.05% Tween 20 and
2% bovine serum albumin (BSA)) and washed. The anti-
body titers were assayed by serial 3-fold dilutions of
chicken serum ranging from 1:200 to 1:48600. All dilu-
tions were prepared in triplicate in dilution buffer (PBS
containing 0.05% Tween 20 and 0.1% BSA), 100 μLwere
added to each well and plates were incubated for 1 h at
37 °C. For each assay, 12 control wells were included,
which contained pure dilution buffer; secondary antibody
was added to 6 of these wells as a measure of background,
and the other 6 wells remained blank as a negative control
for the ELISA. Following incubation, the wells were
washed and 100 μL polyclonal goat anti-chicken IgG (Fc):
HRP (AbD Serotec, Puchheim, Germany), diluted 1:4000
in diluting buffer, were added to each well and the plates
incubated for a further 1 h at 37 °C and then washed. To
detect the binding, 100 μLof3,3′,5,5′-Tetramethylbenzi-
dine (TMB) liquid substrate (Sigma) were added to each
well. The plates were incubated for 2 min and then the
reaction was stopped by addition of 100 μL 1 M HCl.
The absorbance was read immediately at 450 nm in a
PowerWave XS spectrophotometry (BioTek Instruments,
Winooski, VT, USA).
The antibody titers were calculated for the measured
absorbances at 450 nm [41], using the “Antibody Titers”
online data analysis tool [42]. To compare and rank the
ELISA results, a P/N ratio (P = hyper-immune serum, N =
pre-immune serum) of mean antibody titers was calcu-
lated [43]. All statistical analysis was performed using SAS
version 9.3 (SAS Institute, Cary, NC, USA), and differ-
ences between groups assessed using a one-way t-test.
The recognition of recombinant protein by hyper-immune
serum was deemed significant at P < 0.05, indicating that
the protein was expressed in vivo during G. anatis infec-
tion and elicited a specific immune response.
VacFinder® in silico protein analysis
To further predict the protective potential of each of the
expressed proteins, each of the proteins was analyzed
using the proprietary VacFinder® in silico technology
platform (Evaxion Biotech, LLC, USA). VacFinder® is a
Bager et al. Veterinary Research 2014, 45:80 Page 3 of 12
http://www.veterinaryresearch.org/content/45/1/80
Table 2 The 42 proteins from G. anatis 12656–12 selected for cloning and expression
Protein ID Description Mw (kDa) PSORTb prediction TMHMM
prediction
SignalP
prediction
Acc. no.
Gab_0001 Hypothetical protein 195.9 Outer membrane 0 No ERF78007.1
Gab_0047 Metal-dependent proteases with possible
chaperone activity
36.5 Extracellular 0 No ERF77629.1
Gab_0087 Hypothetical protein 140.9 Extracellular 0 No ERF78219.1
Gab_0091 Outer membrane lipoprotein 15.3 Unknown 0 Yes ERF79624.1
Gab_0151
a
RTX toxins and related
Ca2 + −binding proteins
216.4 Extracellular 0 No FJ917356
Gab_0178 Predicted secreted acid phosphatase 30.5 Unknown 0 Yes ERF78374.1
Gab_0186 Membrane-bound lytic
murein transglycosylase
39.9 Unknown 0 Yes ERF78366.1
Gab_0337 Autotransporter adhesin 142.3 Extracellular 0 No ERF78595.1
Gab_0523 Outer membrane protein and related
peptidoglycan-associated (lipo)proteins
16.4 Outer membrane 0 Yes ERF78926.1
Gab_0572
b
F17-like fimbrial subunit 19.1 Extracellular 0 Yes ERF79277.1
Gab_0574 P pilus assembly protein, porin PapC 94.2 Outer membrane 0 Yes ERF79276.1
Gab_0602 Outer membrane protein 47.8 Outer membrane 0 Yes ERF78505.1
Gab_0652 Organic solvent tolerance protein OstA 90.1 Outer membrane 0 Yes ERF78644.1
Gab_0661 Small protein A (tmRNA-binding) 16 Unknown 0 Yes ERF78651.1
Gab_0925 Hypothetical protein 85.2 Outer membrane
and/or extracellular
0 Yes ERF77284.1
Gab_0999 Rare lipoprotein B 18.2 Unknown 1 Yes ERF79357.1
Gab_1008 Type II secretory pathway, component HofQ 42.1 Outer membrane 0 No ERF79423.1
Gab_1162 P pilus assembly protein, porin PapC 93.8 Outer membrane 0 Yes ERF78846.1
Gab_1164
b
F17-like fimbrial subunit 20.5 Extracellular 0 Yes JX855927
Gab_1192 Surface lipoprotein 28 Cytoplasmic membrane 0 Yes ERF78832.1
Gab_1245 Hemolysin activation/secretion protein 67.8 Outer membrane 1 Yes ERF78979.1
Gab_1283 Long-chain fatty acid transport protein 48.8 Outer membrane 0 Yes ERF77302.1
Gab_1309 Membrane proteins related to metalloendopeptidases 42.7 Unknown 0 Yes ERF77527.1
Gab_1396 Uncharacterized protein conserved in bacteria 146.6 Outer membrane 1 No ERF79175.1
Gab_1397 Outer membrane protein 65.9 Outer membrane 0 Yes ERF79124.1
Gab_1399 Membrane-bound metallopeptidase 47.2 Outer membrane
and/or extracellular
0 Yes ERF79126.1
Gab_1450 Opacity protein and related surface antigens 23.1 Outer membrane 1 Yes ERF79004.1
Gab_1576 Outer membrane receptor for ferrienterochelin
and colicins
23.4 Outer membrane 0 No ERF78509.1
Gab_1631 Cell envelope opacity-associated protein A 43.9 Extracellular 1 No ERF79417.1
Gab_1654 Outer membrane phospholipase A 32.4 Outer membrane 1 Yes ERF79322.1
Gab_1755 Outer membrane protein and related
peptidoglycan-associated (lipo)proteins
27.9 Outer membrane 0 Yes ERF79542.1
Gab_2087 Outer membrane protein 51 Outer membrane 0 Yes ERF78059.1
Gab_2124
c
Outer membrane protein (porin) 41.1 Outer membrane 0 Yes KF160335
Gab_2156
b
F17-like fimbrial subunit 20.7 Extracellular 0 Yes ERF79559.1
Gab_2158 P pilus assembly protein, porin PapC 90.6 Outer membrane 0 Yes ERF79560.1
Gab_2192 Outer membrane protein W 25.7 Outer membrane 0 Yes ERF78317.1
Gab_2224 Outer membrane receptor proteins,
mostly Fe transport
81 Outer membrane 0 Yes ERF78217.1
Bager et al. Veterinary Research 2014, 45:80 Page 4 of 12
http://www.veterinaryresearch.org/content/45/1/80
data-driven machine learning method trained by protein
property pattern recognition on known and protective
B-cell protein antigens (except for exotoxins), aiming at
identifying novel and protective B-cell protein antigens
with a neutralizing opsonizing profile. The machine-
learned prediction is ba sed on specific protein property
features of protein sequences rather than sequence simi-
larity, allowing antigen classification ba sed solely on pro-
tein properties [44]. The output is a list of proteins from
the proteome ranked by their ability to elicit a highly
protective antibody response.
Immunization of layer chickens with recombinant proteins
24 Isa Brown layer chickens (16 weeks old) were pur-
chased from a commercial breeder with high biosecurity
standards. The chickens were swabbed for the presence
of G. anatis by a cloacal swab. The chickens were ran-
domly divided in eight groups of four each and allowed
to acclimatize for one week after arrival. The chickens
were kept under free indoor housing conditions an d pro-
vided with fresh water and feed ad libitum. Each group
was immunized subcutaneously with 100 μg of one of
the five selected recombinant proteins (GtxA-N, FlfA,
Gab_1309, Gab_2156 or Gab_2312) mixed in 0.5 mL of
SEC buffer (50 mM NaP, 150 mM NaCl, 0.5 mM TCEP,
10% glycerol; pH 7.5) and 0.5 mL of Freund’s incomplete
adjuvant (Sigma-Aldrich). As a control (non-immunized),
a group of four chickens was immunized with a placebo
(SEC buffer and Freund’s incomplete adjuvant). Two
weeks after the immunization all chickens were infected
intraperitoneally with 1.5 × 10
6
CFU of G. anatis 12656–
12 as described previously [39]. Forty-eight hours after in-
fection the chickens were euthanized and a post mortem
examination was conducted. To assess the protective ef-
fect of the immunization, the lesions found in peritoneum
of each bird were scored according to three parameters:
(i) inflammatory reaction, (ii) amount of exudate, and (iii)
clarity of the peritoneal surfaces. All parameters were
scored on a scale from 0–3, thus giving a maximum score
of 9. Furthermore, the presence of Gallibacterium was de-
tected by swabbing the peritoneum with a sterile cotton
swab and streaking it onto BHI blood agar. The scorings
of the lesions in the peritoneum were a nalyzed by a Mann
Whitney U test and P < 0.05 were deemed significant.
Multiple sequence alignments
Multiple amino acid sequence alignments of the Gab_1309
and Gab_2312 proteins and their orthologs were prepared
using MAFFT v7.130b [45] and formatted using Jalview
2.8.0b1 [46].
Genbank accession numbers
The genome sequence of G. anatis 12656–12 has re-
cently been made available [47] and was submitted to
Genbank (BioProject ID: 213810, accession number
AVOX00000000). The nucleotide sequence accession
numbers for the genes included in this stud y are listed
in Table 2. Genome sequence reads from the nine Galli-
bacterium strains used for Gab 1309 and Gab_2312 mul-
tiple sequence alignment wer e submitted to the NCBI
Sequence Read Archive (SRA) [48] and can be retrieved
using the study accession number SRP029613.
Results
RV in silico prediction of candidate vaccine antigens
For the identification and selection of putative immunogens,
the genomes of 10 Gallibacterium strains were analyzed
(Table 1). B ased on the central premise that protective
antigens should be accessible to the h ost immune sys-
tem, proteins predicted to be surface-exposed or se-
creted were sele cted from the G. anatis pan-proteome.
Moreover, proteins with m ore than o ne TMH were dis-
carded, ba sed on the premise that they are unlikely to
be transported beyond the inner membrane. In addition,
these proteins have the highest rate of expression failure
during subsequent procedures [16] or are less likely to
be over-express ed in E. coli [49]. Finally, proteins
prese nt in six or more o f the 10 G. anatis genomes were
identified. A protein wa s considered present if a signifi-
cant full length match (E-value < 10
−8
) was obtained.
From a total of 31564 proteins , 162 prote ins were pre-
dicted as extracellular proteins and 482 proteins as
Table 2 The 42 proteins from G. anatis 12656–12 selected for cloning and expression (Continued)
Gab_2274 Outer membrane receptor proteins,
mostly Fe transport
74 Outer membrane 0 Yes ERF77464.1
Gab_2304 Glycerophosphoryl diester phosphodiesterase 41.6 Unknown 0 Yes ERF77421.1
Gab_2312 Autotransporter adhesin 325.7 Outer membrane
and/or extracellular
0 No ERF77433.1
Gab_2347 Outer membrane protein/protective
antigen OMA87
89.3 Outer membrane 0 Yes ERF79042.1
Gab_2348 Outer membrane protein 19.8 Outer membrane 0 Yes ERF79081.1
a
Previously described as GtxA in [35].
b
Previously described in [24 ]. Gab_1164 = FlfA.
c
Previously described as OmpC in [36].
Bager et al. Veterinary Research 2014, 45:80 Page 5 of 12
http://www.veterinaryresearch.org/content/45/1/80