a pcr based method for the identification of equine influenza virus from clinical samples
TRANSCRIPT
A PCR based method for the identification of equine
influenza virus from clinical samples
L. Oxburgha,*, AÊ . HagstroÈmb
aDepartment of Veterinary Microbiology, Virology Section, Swedish University of Agricultural Sciences,
Biomedical Centre, Box 585, S-751 23, Uppsala, SwedenbDepartment of Virology, National Veterinary Institute, Biomedical Centre, Box 585, S-751 23, Uppsala, Sweden
Received 22 October 1998; accepted 16 March 1999
Abstract
In this paper we describe the development of a nested RT-PCR assay for the rapid diagnosis and
characterisation of influenza virus directly from clinical specimens. Viral RNA is extracted from
nasal swabs by the guanidine thiocyanate extraction method, and subsequently reverse transcribed.
The complementary DNA is then used as template in a nested PCR reaction. Primers designed for
use in this assay are specific for three templates; (1) the nucleoprotein (NP) gene, (2) the
haemagglutinin gene of the H7N7 equine influenza virus (A1), and (3) the haemagglutinin gene of
the H3N8 equine influenza virus (A2). We show that the assays are specific for the target genes
chosen, and display sensitivity similar to virus isolation. The NP assay detects a variety of different
influenza subtypes, whereas A1 and A2 assays are specific for influenza subtypes H7N7 and H3N8,
respectively. Sequencing of amplicons obtained in the A2 assay yields information on antigenic
regions of the haemagglutinin molecule, and use of this procedure in the routine surveillance of
equine influenza will enable tentative characterisation of circulating viruses despite difficulties in
isolating field strains of the H3N8 subtype. The A1 assay will be useful in ascertaining whether
viruses of the H7N7 subtype still circulate amongst horses, or whether these are extinct. # 1999
Elsevier Science B.V. All rights reserved.
Keywords: Equine in¯uenza; Virus; Horse; Nucleic acids diagnosis-viruses; PCR
Veterinary Microbiology 67 (1999) 161±174
* Corresponding author. Tel.: +46-18-4714030; fax: +46-18-4714572; e-mail: [email protected]
0378-1135/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 1 1 3 5 ( 9 9 ) 0 0 0 4 1 - 3
1. Introduction
A number of outbreaks of severe disease caused by influenza viruses have been
reported during recent years, highlighting the importance of rapid and efficient
identification of viral strains and their origins. Since the circulating flora of influenza
virus in a population has the potential to change through the introduction of viruses or
individual viral genes from other species (Guo et al., 1992; Subbarao et al., 1998; Brown
et al., 1998), most notably birds, methods used in surveillance must have a broad
specificity allowing detection of antigenically and genetically diverse members of the
genus influenza A. Less dramatic changes are seen both in human and equine influenza
viruses. Minor changes in the surface glycoproteins allow the circulation of significantly
antigenically altered viruses and necessitate the exchange of virus strains used in vaccine
production on a regular basis. Additionally, in the case of both the H1N1 and H3N2
influenza viruses of humans, strains with differing receptor-specificity have been shown
to co-circulate (Morishita et al., 1996; Ciappi et al., 1997).
Although large numbers of cases of H3N8 influenza virus infection have been
diagnosed serologically in horses during the past years, attempts to isolate the virus by a
number of laboratories have met with limited success (Leif Oxburgh, unpublished data
and Berndt Klingeborn, personal communication). This finding suggests that currently
circulating strains of the H3N8 subtype do not replicate readily in culture systems in
routine use, and indicates that use of an alternative technique to virus isolation (VI) in the
surveillance of equine influenza virus is necessary.
In contrast to the H3N8 subtype, the H7N7 subtype has not been isolated from the
horse for over 20 years. Serological evidence of the virus has been presented (Mumford
and Wood, 1993), but it is unclear whether these responses represent post-infection titres,
post-vaccination titres or cross-reactivity with another pathogen. A technique for rapid
identification of the virus in clinical samples would facilitate screening in order to
ascertain whether this subtype still is in circulation, or whether it can be removed from
existing vaccines.
In conclusion, there is a need for a diagnostic procedure which has both the redundancy
required to detect diverse subtypes of influenza virus, and the specificity to discriminate
between these different subtypes. It has previously been shown that the reverse
transcription-polymerase chain reaction procedure (RT-PCR) can be used to detect
influenza virus in nasal swab material from horses (Donofrio et al., 1994). We have
therefore chosen to modify this technique for use in the routine surveillance of equine
influenza virus. As shown by Ilobi and colleagues (Ilobi et al., 1998), the product of the
PCR assay can be sequenced in order to obtain information regarding antigenicity and
receptor-binding affinity of viruses without having to isolate and passage field strains for
subsequent purification of RNA and sequencing. Information of this type is important in
the routine surveillance of equine influenza virus, since significant changes at antigenic
sites on the viral haemagglutinin leading to antigenic drift necessitate the updating of
vaccine strains.
In this paper we report the development of a RT-PCR protocol for use in the laboratory
diagnosis of influenza viruses of equine origin. The procedure consists of three steps; (1)
a general PCR detecting the nucleoprotein gene, capable of detecting both avian and
162 L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174
equine virus strains, (2) a PCR specific for the H3 (A2) equine virus, and (3) a PCR
specific for the H7 (A1) equine virus. This protocol enables laboratory diagnosis of
infection within one day and distinguishes between the two different subtypes of equine
influenza virus. This system can be further modified to discriminate between different
avian and mammalian strains by addition of primer-pairs specific for evolutionarily
divergent genes.
2. Materials and methods
2.1. Abbreviations
Abbreviations used to denote virus strains are shown in Table 1.
2.2. Virus stocks
Stocks of viruses SKA/88, PRA/56, LON/73, MIN/84, SHA/89, H5 and AIC/68 were
obtained from the repository at the National Veterinary Institute. All viruses were
cultured in embryonated hen's eggs, and were used in the experiments without further
purification from allantoic fluid. Haemagglutination titres of the viruses ranged from
1 : 32 to 1 : 256 assayed with 1% chick erythrocytes.
2.3. Clinical samples
Clinical samples consisting of nasal swabs collected from horses were obtained from
the routine diagnostic laboratory at the National Veterinary Institute. Ten samples which
had previously been tested positive by immunofluorescent staining (IF, see below) were
chosen as positive controls (SW1±10), five samples from Iceland where equine influenza
Table 1Influenza viruses used in this study
Virus Strain Abbreviation Subtype Genbank accession number
Haemagglutinin Nucleoprotein
A/equi 2/Skara/88 SKA/88 H3N8 Y14053 ±
A/equi 2/Miami/63 MIA/63 H3N8 M24719 ±
A/equi 2/Solvalla/79 SOL/79 H3N8 Y14054 ±
A/equi 2/SoÈderala/94 SOD/94 H3N8 Y14058 ±
A/equi 2/BollnaÈs/96 BOL/96 H3N8 Y14060 ±
A/equi 1/Prague/56 PRA/56 H7N7 X62552 M63748
A/equi 1/London/73 LON/63 H7N7 X62560 M30750
A/swine/Bavaria/76 BAV/76 H1N1 ± ±
A/human/Aichi/2/68 AIC/68 H3N2 V01085 ±
A/human/Shanghai/16/89 SHA/89 H3N2 AF008668 L07372
A/mink/Sweden/84 MIN/84 H10N4 M21646 M24454
A/chicken/H5N4 H5 H5N4 ± ±
L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174 163
has never been shown to circulate were chosen as negative controls (ISW1±5), and 12
unknown samples from horses displaying symptoms of respiratory infection were chosen
(RSW1±12).
2.4. Analysis of the influenza virus content of the samples
In order to determine the virus content of nasal swabs, the samples were subjected to
VI and IF. In cases where both of these assays were negative, paired acute and
convalescent sera from sampled horses were tested by haemagglutination inhibition (HI)
assay in order to assess whether the animals had been infected with equine influenza
virus.
2.4.1. Virus isolation
Nasal swabs were resuspended in 0.5 ml PBS supplemented with 0.5% penicillin,
0.01% streptomycin and 0.00025% fungizone. 100 ml of this fluid was inoculated into the
allantoic cavity of two embryonated hen's eggs (9±10 days old). After 48 h incubation at
378C, allantoic fluid was harvested, and inoculated undiluted into the allantoic cavities of
embryonated hen's eggs. After three passages, allantoic fluid was assayed for the
presence of virus by haemagglutination assay with 1% chick erythrocytes.
2.4.2. Immunofluorescent staining
Cells in the fluid extracted from nasal swabs were sedimented by low-speed
centrifugation, the supernatant was removed, and the pellet was smeared onto a glass
slide. Fixation with methanol and subsequent immunostaining were performed according
to the protocol described by Harlow and Lane (Harlow and Lane, 1988). Polyclonal rabbit
antiserum raised against strain VIS/90 was used as the primary antibody, and FITC-
labelled anti-rabbit immunoglobin was used as the secondary antibody. Stained slides
were examined using a UV microscope, and the presence of fluorescing cells and cell-
associated material was determined.
2.4.3. Haemagglutination inhibition
In the HI test, sera were assayed for antibodies against the most recently isolated strain
of H3N8 equine influenza virus (BOL/96) using the method described by Klingeborn and
colleagues (Klingeborn et al., 1980). Serum samples were collected on two occasions
from the same animals; (i) simultaneously with the nasal swab, and (ii) 10±14 days after
the initial sampling. A fourfold increase in HI titre between the first and the second
sample was interpreted as indicating that the horse had been infected with equine
influenza virus during or immediately prior to the sampling period.
2.5. RNA purification from nasal swabs and virus stocks
Viral RNA was purified from nasal swab material and allantoic fluid using Promega's
SV Total RNA Purification Kit (Promega Corporation), essentially according to the
manufacturer's instructions. The kit is based on disruption of cellular and viral proteins
164 L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174
by treatment with guanidine thiocyanate, binding of the liberated nucleic acid to a silica
matrix and treatment with DN:ase. Neutralisation of DN:ase activity and several washing
steps ensure that no DN:ase is carried over to the purified RNA preparation. Briefly, 30 ml
of material was suspended in lysis buffer containing guanidine thiocyanate, heated to
708C for 3 min, centrifuged, and the supernatant bound to the silica matrix packed in a
minicolumn. After washing, nucleic acid bound to the column was treated with DN:ase
for 15 min at room temperature, DN:ase activity was neutralised by addition of guanidine
thiocyanate, and the bound RNA was washed twice. Finally, RNA was eluted in 100 ml
DEPC-treated dH2O, and the RNA was immediately frozen in liquid nitrogen in 20 ml
aliquots.
2.6. Primer design
Nucleotide sequences of viral genes shown in Fig. 1 were retrieved from the NCBI
database, and were aligned using the Megalign program of the DNASTAR software package
(DNAstar Inc.). After consensus and non-consensus regions were identified as putative
stretches for primer synthesis, the individual genes sequences were processed using the
OLIGO 4.05 software package (National Biosciences) to identify suitable sequences for use
as oligonucleotides in RT-PCR. The oligonucleotide primers shown in Fig. 1 were
subsequently commercially synthesised (DNA Technology AS, Aarhus, Denmark).
2.7. Nested RT-PCR
A nested RT-PCR protocol was designed to optimise sensitivity in order to be able to
amplify small quantities of genetic material present in swabs. A modification of
Promega's Access RT-PCR was used for first-strand synthesis and the first PCR step. The
second (nested) PCR was carried out using a standard protocol. Briefly, a mix containing
3 ml 5x Access reaction buffer, 1 ml reverse primer (to a final concentration of 1 pmol/ml),
1 ml (5 u) Tfl DNA polymerase and 15 ml dH20 was sealed into the bottom of a Hot-Start
tube (Molecular BIO-products) under a layer of wax by melting the wax bead at 808C for
3 min, then rapidly cooling on ice. A mixture containing 7 ml 5x Access reaction buffer,
1 ml dNTP (to a final concentration of 0.2 mM each of dATP, dCTP, dGTP and dTTP),
4 ml MgSO4 (to a final concentration of 1.92 mM), 1 ml forward primer (to a final
concentration of 1 pmol/ml) and 1 ml (5 u) AMV reverse transcriptase was layered on top
of the solidified wax. Two drops of mineral oil were added to prevent evaporation of the
sample during the reverse transcription step. An aliquot of purified RNA was rapidly
thawed and denatured for 15 min at 658C. After snap-cooling on ice, 16 ml of the RNA
solution was added through the mineral oil to the reverse transcription mix. The tube was
subsequently transferred to a thermal cycler (Minicycler, MJ Research) and the following
program was run; 488C for 45 min, 948C for 3 min, then 20 cycles of 948C for 30 s, 528Cfor 1 min and 688C for 2 min. For the NP and A2 reactions, 1 ml of the product of this first
RT-PCR step was used as template in the second PCR. For the A1 reaction, the first RT-
PCR mix was diluted 10 times in dH2O before transfer of a 1 ml aliquot to the second
PCR. This dilution step is necessary in order to minimise appearance of bands caused by
carry-over of primers from the first RT-PCR.
L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174 165
Fig. 1. Primer design for the three nested PCR assays described in this paper. An alignment of influenza virus genes spanning the region that was used for
oligonucleotide synthesis is shown. Sequences corresponding to primers used in the assays are underlined. The DNA sequence is in mRNA sense, with arrows
pointing downstream denoting forward primers, and arrows pointing upstream representing reverse primers. The sequences of reverse primers used in the assay
are the reverse complement of those underlined in the figure. (I) Primer set for the NP RT-PCR, (II) primer set for the A1 RT-PCR, (III) primer set for the A2
RT-PCR; this assay is semi-nested, with the same reverse primer being used in both the first PCR and the nested PCR.
16
6L
.O
xbu
rgh
,A Ê.
Ha
gstro Èm
/Veterin
ary
Micro
bio
logy
67
(1999)
161±174
For the nested PCR reaction three different MgCl2 concentrations were used. For the
NP assay, 5 ml Promega Mg-free reaction buffer, 3 ml MgCl2 (to a final concentration of
1.5 mM), 1 ml dNTP (to a final concentration of 0.05 mM) and 1 ml each primer (to a final
concentration of 1 pmol/ml) were mixed with 1 ml of template. For the A1 assay, 5 ml
Promega Mg-free reaction buffer, 4 ml MgCl2 (to a final concentration of 2 mM), 1 ml
dNTP (to a final concentration of 0.05 mM) and 1 ml each primer (to a final concentration
of 1 pmol/ml) were mixed with 1 ml of template. For the A2 assay, 5 ml Promega Mg-free
reaction buffer, 2 ml MgCl2 (to a final concentration of 1 mM), 1 ml dNTP (to a final
concentration of 0.05 mM) and 1 ml each primer (to a final concentration of 1 pmol/ml)
were mixed with 1 ml of template. The tubes were transferred to a thermal cycler, heated
to 948C and put on hold at this temperature. 5 ml diluted Promega Taq polymerase
containing 2.5 u enzyme was added to each tube at this denaturing temperature. The
thermal cycle was subsequently initiated. Thermal cycle programs used for each of the
assays consisted of 3 min of 948C denaturation followed by 32 cycles of 948C for 1 min,
annealing for 1 min and 728C for 1 min. The cycle was terminated with an elongation
step of 728C for 10 min. Annealing temperatures for the three assays were; 508C for the
NP and A1 assays and 608C for the A2 assay.
2.8. Analysis of the nested PCR product
5 ml of each PCR product was run on a 2% agarose gel in a TAE buffer system and
subsequently stained with ethidium bromide before viewing on a UV lamp. A 100 base
pair ladder (Boehringer Mannheim Scandinavia AB) was used as a molecular weight
marker. The marker contains highlighted bands, with increased intensity at 500, 1000 and
1500 base pairs.
2.9. Precautions to minimise contamination risks
Sample purification and RT-PCR reaction preparation, second PCR step reaction
preparation and agarose gel analysis were all performed in separate laboratories. All
solutions were aliquoted to volumes suitable for one set of 10 reactions, and each aliquot
was discarded after one use. Negative (dH2O) controls were run with each assay,
comprising every fifth sample in each series. No contamination was detected at any
time.
2.10. Dilution series of virus template to assess the sensitivity of the assays
Titration of egg infectious doses (EID) for SKA/88 and PRA/56 was performed by
inoculation of 10-fold dilutions of these virus stocks into 10-day embryonated hen's
eggs. Allantoic fluid was harvested after 2 days incubation, and the end-point of the
titration was determined by haemagglutination assay with 1% chick erythrocytes. The
lowest concentration of inoculum yielding growth in eggs was designated 1 EID. Titrated
virus was diluted in negative nasal swab fluid so that 30 ml would contain 1 EID, 10 EID
or 100 EID. These samples were subsequently assayed in the same way as the swab
samples.
L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174 167
2.11. Sequence analysis
PCR products generated by each of the assays were analysed by automated sequencing.
SKA/88 PCR products from the NP and A2 assays and the PRA/56 product from the A1
assay were purified using Promega's Wizard PCR preps, and sequenced using the ABI
sequencing kit and the second forward PCR primer. Sequence data was aligned with the
corresponding sequence from the NCBI database to assess homology.
3. Results
3.1. All nested primer sets amplify targets specifically
Oligonucleotide primers were designed to amplify motifs in the influenza virus
genome with differing evolutionary conservation (Fig. 1). The forward primer of each
first PCR step primer pair was used to reverse transcribe viral RNA prior to the first PCR
reaction. The NP primer set amplifies a 241 base pair fragment from each of the viruses
which have been assayed (Fig. 2). These viruses are evolutionarily distantly related and
are isolated from different animal species. The A1 primer set specifically amplifies a 327
base pair fragment from the H7N7 equine influenza viruses. No PCR product can be seen
with any other template. The A2 primer set amplifies a 522 base pair fragment from the
H3N8 equine influenza virus, and does not show any positive result with the other
templates. It can be concluded that the three RT-PCR assays can be used in combination
to identify diverse subtypes of influenza virus, and subsequently to discriminate between
the two subtypes of equine virus.
3.2. The amplified fragments are specific for influenza virus
Sequencing of amplicons from each of the assays reveals a 99±100% homology with
the sequence predicted to be amplified by that primer set (data not shown). SKA/88
Fig. 2. RT-PCR assay of a selection of different viruses to demonstrate the specificities of the three assays.
Assays were performed as described in Materials and methods. Virus used was unpurified allantoic fluid.
168 L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174
amplicons were sequenced in order to verify the NP assay and A2 assay, and the PRA/56
amplicon was sequenced to verify the A1 assay. One single, non-coding substitution
could be seen in the SKA/88 A2 amplicon. From the sequencing data we conclude that
the assays amplify influenza genes specifically and that there is no background generated
by priming of sequences other than those from influenza virus in the template.
3.3. The assay system displays sensitivity similar to VI
Owing to the difficulties in isolation of field strains of the H3N8 subtype (see above
and Table 2), and the lack of clinical material from horses infected with the H7N7
subtype, we chose to compare the detection limit of the RT/PCR assays with VI
indirectly. Egg-adapted strains of both subtypes were used for this purpose. Purification
and amplification of serial dilutions of allantoic fluid containing defined titres of virus
mixed with negative nasal swab material were performed in order to assess sensitivity
(Fig. 3). The NP assay was found to amplify as little as 1 EID of virus, whereas the A1
and A2 assays showed a lower sensitivity, amplifying 10 EID of virus. This demonstrates
the sensitivity of the method, suggesting that it is similar to VI.
3.4. Assay of clinical samples demonstrates the utility of the system for diagnostic use
3.4.1. Assay of positive and negative control samples
Ten nasal swab samples which had previously been tested positive by IF (SW1±10),
and five samples from horses tested negative for antibodies by IF and HI (ISW1±5) were
used as positive and negative controls, respectively. Virus culture in embryonated hen's
eggs was attempted from each of the positive swab samples without result. VI and IF
were performed directly on arrival of the samples at the laboratory. The samples were
Fig. 3. RT-PCR demonstrating the sensitivity of the assay. Egg-grown virus was titrated as described in
Materials and methods. Defined amounts of virus corresponding to egg infectious doses (EID) of 1, 10 and
100 were assayed in the RT-PCR.
L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174 169
subsequently stored at ÿ708C for 2±6 months prior to RT/PCR assay. Since there has not
been any circulation of equine H7N7 equine influenza virus in Sweden for over 20 years
(Berndt Klingeborn, personal communication), it was not possible to obtain clinical
material for assessment of the A1 assay. Instead, 100 EID of viruses PRA/56 and LON/73
in allantoic fluid were mixed with negative nasal swab material in order to simulate
clinical material.
As shown in Table 2 and Fig. 4, all samples from infected horses tested positive in the
NP assay, without amplification of template from non-infected horses. As expected, none
Table 2Comparison of results of three different assays for the detection of influenza virus in nasal swab samples
Sample NPa A1b A2c VId IFe HIf
SW1g � ÿ � ÿ � NDh
SW2 � ÿ � ÿ � ND
SW3 � ÿ � ÿ � ND
SW4 � ÿ � ÿ � ND
SW5 � ÿ � ÿ � ND
SW6 � ÿ � ÿ � ND
SW7 � ÿ � ÿ � ND
SW8 � ÿ � ÿ � ND
SW9 � ÿ � ÿ � ND
SW10 � ÿ � ÿ � ND
ISW1i ÿ ÿ ÿ ÿ ÿ ÿISW2 ÿ ÿ ÿ ÿ ÿ ÿISW3 ÿ ÿ ÿ ÿ ÿ ÿISW4 ÿ ÿ ÿ ÿ ÿ ÿISW5 ÿ ÿ ÿ ÿ ÿ ÿRSW1j � ÿ � ÿ ÿ �RSW2 � ÿ � ÿ � ND
RSW3 � ÿ � ÿ � ND
RSW4 � ÿ � ÿ ÿ �RSW5 ÿ ÿ ÿ ÿ ÿ ÿRSW6 � ÿ � ÿ � ND
RSW7 � ÿ � ÿ � ND
RSW8 � ÿ � ÿ � ND
RSW9 � ÿ � ÿ � ND
RSW10 � ÿ � ÿ ÿ �RSW11 ÿ ÿ ÿ ÿ ÿ ÿRSW12 � ÿ � ÿ � ND
a RT/PCR assay for nucleoprotein.b RT/PCR assay for H7 haemagglutinin.c RT/PCR assay for H3 haemagglutinin.d Virus isolation in the allantoic cavity of 9±10-day embryonated hen's eggs.e Immunostaining of virus antigens in respiratory secretions.f Haemagglutination inhibition assay. A four-fold titre increase in paired sera taken with 10±14 days interval isscored as positive.g Positive control samples are designated SW.h Not Done.i Negative control samples are designated ISW.j Unknown samples are designated RSW.
170 L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174
of the swabs tested were positive in the A1 assay. The simulated samples containing
PRA/56 and LON/73, however, showed a positive result, demonstrating the ability of the
assay to identify the H7N7 viruses in the background of RNA present in swab material.
The A2 assay amplified sequences in each of the positive samples, but none of the
negative samples, showed a 100% correlation with IF results. There is a slight variability
in the intensity of bands in each of the three assays, presumably reflecting the abundance
of virus present in the swab. The high molecular weight band seen in some samples
corresponds to the amplicon from the first PCR reaction.
3.4.2. Assay of unknown samples
Twelve nasal swab samples sent in to the National Veterinary Institute for diagnosis
were assayed using the RT/PCR technique, in addition to IF and VI. In cases where
neither IF or VI were positive, HI assay was performed on paired serum samples to
ascertain whether the sampled horses had been infected with H3N8 influenza virus. The
results shown in Table 2 demonstrate a complete correlation of RT/PCR with IF in cases
where the IF result was positive. In total, four samples were negative by IF. Viral nucleic
acid could be detected in two of these using RT/PCR. These two samples were positive by
Fig. 4. RT-PCR of clinical samples. Ten positive samples (SW) were assayed as described in Materials and
methods. Cultured SKA/88 (H3N8 sybtype) virus was included as a control. A1 influenza virus strains
were not available as clinical samples and 100 EID virus in allantoic fluid diluted in a negative swab
sample were used instead. Negative samples (ISW) consisted of a series of nasal swabs collected from
horses in Iceland where equine influenza does not circulate. These horses were serologically tested in order
to confirm that they were free from infection. The results of the assay shown in the figure are tabulated
together with results of virus isolation, immunoflurescent staining and haemagglutination inhibition in
Table 2.
L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174 171
HI, indicating that infection had taken place in both horses, and that concentrations of
viral antigen in the samples were below the detection limit for IF. The other two IF-
negative samples were negative by all techniques tested, indicating that symptoms
displayed by these horses were not caused by equine influenza virus infection. VI was not
successful for any sample.
4. Discussion
The data presented in this paper demonstrate that the PCR can be a valuable tool in the
surveillance of equine influenza virus. Several aspects of our method make it suitable for
this purpose. First, the method is versatile, allowing amplification either of evolutionarily
conserved regions, giving an assay recognising many different subtypes of virus, or non-
conserved regions, giving a discriminating assay, identifying individual virus subtypes.
Since there are many influenza virus gene sequences in the genome database at this time,
it should be possible to tailor the method to discriminate between most influenza virus
subtypes. Secondly, the method is rapid. Identification of the infecting virus subtypes can
be performed within 1 day of receipt of clinical material, and more detailed information
can be obtained by sequencing of the amplicon (see below). Thirdly, the method does not
rely on the presence of viable virus in the swab sample, allowing diagnosis and
identification to be made on material which has been inactivated during transport, stored
for a long time and which contains virus which will not grow in the culture system used
for isolation at the laboratory. The fourth, which may be the most important aspect is that
the RT-PCR generates an amplicon which can be used for nucleotide sequencing of genes
coding for antigenic determinants of circulating influenza viruses.
Since equine influenza viruses undergo antigenic drift,(Burrows and Denyer, 1982;
Berg et al., 1990; Oxburgh et al., 1993, 1998; Daly et al., 1996), it is important to monitor
the variation in antigenicity of circulating viruses. The best method to this end is
serotyping using the HI assay with mono- and polyclonal antibodies. A major drawback
of this method, however, is that it necessitates isolation and culture of virus. As
demonstrated in this paper, VI is often unsuccessful. A tentative evaluation of the
antigenicity of a new strain could be made from sequencing of the haemagglutinin gene
segment amplified in this RT/PCR assay. Sufficient sequence data has been generated to
enable comparisons of antigenic regions of strains to be made, and major antigenic
changes could be identified.
The protocol described in this paper is, however, labour-intensive and costly, and
cannot be used to replace standard routine methods such as immunoassay of swab
material and serological testing using the HI test. It should, however, be incorporated as a
complement to these methods to facilitate the identification of viruses circulating in new
outbreaks, and as a rapid method for typing virus from outbreaks with unusual
characteristics. The finding by Donofrio and colleagues (Donofrio et al., 1994) that viral
RNA is present in the upper airways during a limited time, from day 3 to approximately
day 8 post infection suggests that pooled samples of nasal swabs from symptomatic
horses in the same stable could be assayed in order to make a certain laboratory dia-
gnosis of an outbreak. Any detailed studies of circulating virus genotypes such as
172 L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174
nucleotide sequencing would, however, have to be performed on samples from individual
horses.
This technique will be extremely useful in elucidating the epidemiology of the H7N7
viruses. The lack of isolations of this virus during the past years raises the question of
whether it could be extinct, and a detailed survey of samples using RT-PCR should give a
picture of whether this is actually a fact.
The versatility of this assay system enables it to be used in the identification of a broad
spectrum of influenza viruses. Addition of primer sets specific for other subtypes of the
virus will enable characterisation of an unknown influenza virus to be made from a
clinical sample within a day, and will give important information regarding the subtypes
of virus. This information will help in monitoring the epidemiology of influenza virus in
various animal species.
Acknowledgements
This work was made possible by a grant from the Swedish Horse Racing Board (ATG).
References
Berg, M., Desselberger, U., Abusugra, I.A., Klingeborn, B., Linne, T., 1990. Genetic drift of equine 2 in-
fluenza A virus (H3N8), 1963±1988: analysis by oligonucleotide mapping. Vet. Microbiol. 22(2/3), 225±
236.
Brown, I.H., Harris, P.A., McCauley, J.W., Alexander, D.J., 1998. Multiple genetic reassortment of avian and
human influenza A viruses in European pigs, resulting in the emergence of an H1N2 virus of novel
genotype. J. Gen. Virol. 79(12), 2947±2955.
Burrows, R., Denyer, M., 1982. Antigenic properties of some equine influenza viruses. Arch. Virol. 73(1),
15±24.
Ciappi, S., Azzi, A., Stein, C.A., De Santis, R., Oxford, J.S., 1997. Isolation of influenza A (H3N2) virus with
`̀ O''!`̀ D'' phase variation. Res. Virol. 148(6), 427±431.
Daly, J.M., Lai, A.C., Binns, M.M., Chambers, T.M., Barrandeguy, M., Mumford, J.A., 1996. Antigenic and
genetic evolution of equine H3N8 influenza A viruses. J. Gen. Virol. 77(4), 661±671.
Donofrio, J.C., Coonrod, J.D., Chambers, T.M., 1994. Diagnosis of equine influenza by the polymerase chain
reaction. J. Vet. Diagn. Invest. 6(1), 39±43.
Guo, Y., Wang, M., Kawaoka, Y., Gorman, O., Ito, T., Saito, T., Webster, R.G., 1992. Characterization of a new
avian-like influenza A virus from horses in China. Virology 188(1), 245±255.
Harlow, E., Lane, D., 1988. Antibodies: a Laboratory Manual, Cold Spring Harbor Laboratory Press.
Ilobi, C.P., Nicolson, C., Taylor, J., Mumford, J.A., Wood, J.M., Robertson, J.S., 1998. Direct sequencing of the
HA gene of clinical equine H3N8 influenza virus and comparison with laboratory derived viruses. Arch.
Virol. 143, 891±901.
Klingeborn, B., Rockborn, G., Dinter, Z., 1980. Significant antigenic drift within the influenza equi 2 subtype in
Sweden. Vet. Rec. 106(16), 363±364.
Morishita, T., Nobusawa, E., Nakajima, K., Nakajima, S., 1996. Studies on the molecular basis for loss of the
ability of recent influenza A (H1N1) virus strains to agglutinate chicken erythrocytes. J. Gen. Virol. 77(10),
2499±2506.
Mumford, J., Wood, J., 1993. In: Proc. WHO/OIE meeting: consultation on newly emerging strains of equine
influenza, 18±19 May 1992, Animal Health Trust, Newmarket, Suffolk, UK, Vaccine 11(11), 1172±
1175.
L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174 173
Oxburgh, L., Akerblom, L., Fridberger, T., Klingeborn, B., Linne, T., 1998. Identification of two antigenically
and genetically distinct lineages of H3N8 equine influenza virus in Sweden. Epidemiol. Infect. 120(1),
61±70.
Oxburgh, L., Berg, M., Klingeborn, B., Emmoth, E., Linne, T., 1993. Equine influenza virus from the 1991
Swedish epizootic shows major genetic and antigenic divergence from the prototype virus. Virus Res.
28(3), 263±272.
Subbarao, K., Klimov, A., Katz, J., Regnery, H., Lim, W., Hall, H., Perdue, M., Swayne, D., Bender, C., Huang,
J., Hemphill, M., Rowe, T., Shaw, M., Xu, X., Fukuda, K., Cox, N., 1998. Characterization of an avian
influenza A (H5N1) virus isolated from a child with a fatal respiratory illness (see comments). Science
279(5349), 393±396.
174 L. Oxburgh, AÊ . HagstroÈm / Veterinary Microbiology 67 (1999) 161±174