American Jotma) of Epidemiology Copyright ©1992 by The Johns Hopkins University School of Hyg^ne and Public Health
Vol 136, No. 4 Printed in U.S A.
AS rights reserved
Interspecies Transmission and Reassortment of Influenza A Viruses in Pigs and Turkeys in the United States
Stephen M. Wright,1 Yoshihiro Kawaoka,1 Gerald B. Sharp,2 Dennis A. Senne,3 and Robert G. Webster1
Genetic reassortment between influenza A viruses in humans and in animals and birds has been implicated in the appearance of new pandemics of human influenza. To determine whether such reassortment has occurred in the United States, the authors compared the genetic origins of gene segments of 73 swine influenza virus isolates (1976-1990), representing 11 states, and 11 turkey virus isolates (1980-1989), representing eight states. The host origin of gene segments encoding the internal proteins of H1N1 swine and turkey influenza viruses was identified by developing a dot-Wot assay. All gene segments of swine influenza viruses were characteristic of influenza virus genes from that species, indicating that pigs may not be frequent participants in interspecies genetic exchange and reassortment of influenza viruses in the United States. In contrast, 73% of the turkey influenza virus isolates contained genes of swine origin. One turkey isolate was a reassortant having three genes characteristic of avian influenza virus and three of swine origin. These findings document a high degree of genetic exchange and reassortment of influenza A viruses in domestic turkeys in the United States. The molecular biologic techniques used by the authors should aid future epidemiologic studies of influenza pandemics. Am J Epidemiol 1992;136:488-97. epidemiologic methods; genetics; Orthomyxovirus type A, avian; Orthomyxovirus type A, human; Orthomyxovirus type A, porcine; reassortant viruses; swine; turkeys
Influenza A virus is responsible for frequent epidemics and occasional pandemics. Received for publication October 21,1991, and in final form March 11, 1992. Abbreviations' H, hemagglutmin; M, matrix or membrane protein; N, neuraminidase; NP, nucleoprotein; NS, nonstructural; PA, polymerase A; PB1, polymerase B1; PB2, polymerase B2. ' Department of Virology and Molecular Biology, St. Jude Children's Research Hospital, Memphis, TN. 1 Department of Btostatistics and Epidemiology, University of Tennessee, Memphis, TN. 3 National Veterinary Services Laboratories, Ames, LA. Reprint requests to Dr. Robert G. Webster, Department of Virology and Molecular Biology, St. Jude Children's Research Hospital, 332 North Lauderdale, Memphis, TN 38105. The authors thank Dr. Clayton Naeve and the Molecular Resource Center for preparation of oligonucleotide probes and pnmers. They also thank Glenith D. White for preparation of the manuscript This work was supported by US Public Hearth Service research grant AI-29680, National Institutes of Health CORE grant CA-21765, and the American Lebanese Syrian Associated Chantes.
From the years 1957 to 1986, there were 19 epidemics of influenza in the United States, each responsible for 10,000 excess deaths (1). More catastrophic are pandemics, such as "Spanish influenza," which claimed over 25 million lives worldwide in 1918 (2). Influenza A virus has several subtypes defined by changes in two surface proteins, hemagglutinin (H) and neuraminidase (N). H3N2 and H1N1 are the subtypes currently circulating among human populations. The genome of influenza A virus consists of eight segments of RNA. Reassortment is characterized by the exchange of genomic segments between two different influenza viruses that simultaneously infect a cell (3). Such reassortment of the genome has been reported for both avian (4) and human (5, 6) influenza viruses, but there is little information on how often or in which species this event occurs. Genetic reassortment be488
Reassortment of Influenza A Viruses
tween the influenza A viruses infecting humans and other species has been implicated in the generation of pandemic influenza (7). In this study, we have examined the occurrence of interspecies transmission (that is, virus transfer and replication in another species) and reassortment of influenza A viruses. The human influenza virus responsible for the 1957 pandemic contained three gene segments, H (8), N (8), and polymerase Bl (PB1) (9), that are characteristic of avian influenza. Similarly, two viral gene segments (H and PB1) of the 1968 pandemic were closely related to avian influenza. Thus, viral genes from widely different species can be part of the same genome. Two possibilities could account for interspecies transmission or reassortment of influenza A viruses. 1) Reassortment could occur in an avian or human host, although the available evidence indicates that avian influenza viruses are not able to replicate productively in human hosts. 2) An intermediate host could become infected by both avian and human influenza viruses. Swine are promising candidates for this role because they allow productive replication by either avian (10) or human (11) influenza viruses. Moreover, interspecies transmission of influenza virus has been reported to occur between swine and humans in the United States. One swine virus, Swine/Colorado/1/1977 (Sw/CO/1/77) (H3N2), has characteristics of a human virus (12), while the human virus A/Wisconsin/ 3523/1988 (A/WI/3523/88) (H1N1) has swine characteristics (13). It has been difficult to test these ideas because of the lack of techniques capable of accurate, rapid comparison of nucleotide sequences in minute samples of viral RNA. In an effort to assess the likelihood of viral gene reassortment in pigs and turkeys in the United States, we devised a sensitive assay to determine the origin of the gene segments based on the polymerase chain reaction and dot-blot hybridization. The method was successful because influenza virus genes possess conserved regions that remain host-specific (14). Relying on the vast amount of sequence information available for influenza A viruses (Y. Kawaoka and R. G. Webster,
489
unpublished manuscript; 15-22), we were able to design probes that would recognize and bind to genomic regions characteristic of a particular host group, e.g., human, swine, or avian. Reported here are the results of this analysis as applied to influenza A isolates collected over defined periods within geographically distinct areas of the United States.
MATERIALS AND METHODS Viruses
We studied a total of 84 H1N1 influenza A viruses of swine and turkey origin (table 1) which were obtained from the repository at St. Jude Children's Research Hospital, Memphis, Tennessee; all were grown in embryonated chicken eggs. Swine influenza viruses were collected during an epidemiologic survey (19761977) conducted to determine the prevalence of influenza viruses in this species (12). Briefly, nasal swabs from 200 randomly selected pigs were obtained weekly from slaughterhouses in Memphis, Tennessee, and Madison, Wisconsin. A total of 9,400 pigs were sampled, of which 478 were positive for influenza (prevalence, 5.1 percent). Sixty-five H1N1 influenza viruses were selected at random from the 1976-1977 survey for dot-blot analysis. The hosts of the viruses originated from seven states: Arkansas (AR), Georgia (GA), Illinois (IL), Missouri (MO), Mississippi (MS), Tennessee (TN), and Wisconsin (WI) (table 1). Additionally, eight other swine influenza viruses were examined to determine genetic host origins; these viruses represented a 15-year isolation period (1976-1990) from California (CA), Iowa (IA), Kansas (KS), and Minnesota (MN) and were obtained during routine sampling of herds for swine influenza. We also examined 11 H1N1 influenza A viruses from turkeys in the central United States. These samples were collected over a 10-year period (1980-1989) by US Department of Agriculture personnel who monitor flocks for influenza (table 1). The two surface glycoproteins (H and N) on the swine H1N1 influenza viruses studied had previ-
490
Wright et al.
TABLE 1.
H1N1* influenza A viruses of swine (Sw) and turkey (Ty) origin analyzed in a US study Virus
Survey swine virus (1976-1977) Sw/TN/- -/76 (24 samples)
Sw/TN/- -/77 (24 samples)
Sw/WI/--/76 (17 samples)
Isolate no.
State of swine origin
2-6,8-14 15-20,26 21 23, 24, 27 25
Illinois (IL) Tennessee (TN) Georgia (GA) Arkansas (AR) Missouri (MO)
2, 4, 5,15-17, 19, 20, 31, 36-38, 43, 44 9,13 11 12,25-27 29, 30, 35
Missouri (MO)
747, 750, 754, 756, 757, 762, 763, 764, 766, 769, 834, 846, 848, 888, 936, 940, 954
Illinois (IL) Mississippi (MS) Tennessee (TN) Arkansas (AR) Wisconsin (Wl)
Other swine viruses (1976-1990) Sw/Minnesota/27/77 Sw/Wisconsin/11/80 Sw/Kansas/3228/87 Sw/Wisconsin/915/88 Sw/CaJifomia/9007919/90 Sw/lowa/3421/90 Sw/California/9001707/91 Sw/lowa/23239/91 Turkey viruses (1980-1989) Ty/Kansas/4880/80 Ty/lowa/7352/80 Ty/Minnesota/1661/81 Ty/Missouri/1/81 Ty/Colorado/10091/81 Ty/South Dakota/7034/86 Ty/South Dakota/7740/86 Ty/Missourl/21939/87 Ty/North Carolina/1/88 Ty/Ohio/29962/88 Ty/Mlnnesota/12537/89 1
H, hemagglutnln; N, neurairtnidase.
ously been characterized antigenically and were characteristic of swine influenza viruses (12). Similarly, the surface glycoproteins of the turkey influenza isolates were characteristic of avian or swine influenza viruses (unpublished data based on antigenic and sequence analyses).
pital and were 18-20 base pairs in length. Table 2 lists forward and reverse primers and nucleotide locations for each gene that was evaluated. The sequences selected represent conserved regions for all influenza viruses. Oligonucleotide probes
Primers
Primers for the polymerase chain reaction were prepared in the Molecular Resource Center at St. Jude Children's Research Hos-
Oligonucleotides, prepared and purified in the Molecular Resource Center, were made for polymerase (PA, PB1, and PB2), nucleoprotein (NP), matrix or membrane protein
Reassortment of Influenza A Viruses
TABLE 2. Locations of primers used In polymerase chain reactions for analysis of swine and turkey influenza A viruses*
TABLE 3. Host-specific oligonucleotide probes used in a study of swine and turkey influenza A viruses* Gene
Nucleotide location segmentt
Forward primer
Reverse primer
segmentt and probe type
NS M NP PA
1-20 73-92 1,033-1,052 1-19 331-353 949-965
871-890 896-915 1,447_1,467 603-622 1,492-1,511 1,602-1,621
NS Swine Human Avian Control
PB1
PB2
• Forward primers were used lor synthesis of cDNA and amplification during the potymerase chain reaction. Reverse primers were used only for the polymerase chain reaction. These primers represent regions conserved by aD viruses. • t NS, nonstructural; M, matrix or membrane protein; NP, nucteoproteln; PA, polymerase A; PB1, polymerase B1; PB2, polymerase B2.
(M), and nonstructural (NS) gene segments. Four oligonucleotide probes were prepared for each gene: I) swine, 2) human, and 3) avian probes that would bind to their respective sequences, and 4) a control probe that would bind to a region common to all viruses. Control probes were used to verify that viral sequences of the gene being analyzed were indeed present. An example of the probe design is shown in the Appendix table. Table 3 summarizes the nucleotide location of each probe for each gene. Oligonucleotide probes were end-labeled with digoxigenin as described by Farquharson et al. (23). All reagents were obtained from Boehringer Mannheim (Indianapolis, Indiana). The reaction was incubated at 37°C for 5 minutes, followed by ethanol precipitation.
491
Nucleotide location
virus wtth Identical sequence
_. "w-"°-
88-107 266-286 452-470 539-557
Sw/TN/26/77 Udom/72 Mal/NY/6750/78
—* 15 —t
264-288 477-499 579-606 700-719
Sw/TN/26/77 Udom/72 Mal/NY/6750/78
16 16 16
1,350-1,371 Sw/TN/26/77 1,110-1,134 Udom/72 1,326-1,351 Mal/NY/6750/78 1,423-1,442
17 18 18
M Swine Human Avian Control NP Swine Human Avian Control PA Swine Human Avian Control PB1 Swine Human Avian Control PB2 Swine Human Avian Control
519-540 194-212 291-309 587-606
Sw/TN/26/77 NT/60/68 Ty/MN/1661/80
19 20 19
390-411 Sw/TN/26/77 1,029-1,047 Mem/8/88 1,290-1,313 Mal/NY/6750/78 1,492-1,511
9 9 21
1,539-1,561 Sw/TN/26/77 1,408-1,431 Mem/8/88 1,536-1,551 Mal/NY/6750/78 1,567-1,592
22 22 21
* The nucleotide location and the virus having the Identical sequence with the probe are shown. Control probes were based on regions conserved by al influenza viruses. t NS, nonstructural; M, matrix or membrane protein; NP, nucieoprcitein; PA, potymerase A; PB1, polymerase B1; PB2, polymerase B2. t Y. Kawaoka and R. G. Webster, unpublished manuscript.
The hybridization assay
Viral RNA was isolated from influenzainfected allantoic fluid according to the method of Bean et al. (24) with a single phenolxhloroform extraction. cDNA was prepared and amplified as outlined by Katz et al. (25). Briefly, cDNA was synthesized from viral RNA using reverse transcriptase (Life Sciences, St. Petersburg, Florida) at 42°C for 1 hour. cDNA was amplified for 35 cycles in a DNA thermal cycler (Perkin Elmer Cetus, Norwalk, Connecticut). Negative
controls (distilled water) were included during amplification to ensure that contamination did not occur. Six microliters of the polymerase chain reaction product was applied to a nylon membrane (Zeta Probe; Bio-Rad Laboratories, Richmond, California) through a vacuum manifold apparatus (Minifold; Schlicher & Schuell, Keene, New Hampshire). DNA was fixed to the membrane by
492
Wright et al.
ultraviolet radiation cross-linking (Stratalinker; Stratagene, La Jolla, California) as described by Allefs et al. (26). Hybridization and detection of probe binding were done as described by Casacuberta et al. (27). Briefly, prehybridization and hybridization of filters was conducted in an incubator/rotator at 42°C. Fifty nanograms of the digoxigeninlabeled probe was added and was allowed to hybridize overnight. After washing, anti-digoxigenin alkaline phosphataseconjugated Fab (antigen-binding) fragments (Boehringer Mannheim) were added to filters at room temperature (30 minutes). The bound probe was detected with X-phosphate and nitroblue tetrazolium salt in a colorimetric reaction (Genius Nucleic Acid Detection Kit; Boehringer Mannheim). A brownpurple color indicated probe binding. Four filters and four probes were used for each gene (control, human, swine, and avian). The polymerase chain reaction products of Turkey/Missouri/21939/87 were purified for nucleotide sequencing by Geneclean (Bio 101, La Jolla, California). The nucleotide sequences were determined by the dideoxynucleotide chain termination method (28).
RESULTS Probe specificity
We tested each probe against influenza viruses whose complete sequences were already known, to ensure that each of the probes was binding to its complementary sequence and not to heterologous sequences. Figure 1 displays results using the NS gene segment as an example. Part A shows results of the dot-blot hybridization; part B indicates the number of differences between the probe sequence and the viral sequence. There is no binding when there are five or more base pair differences between the 20mer probe and the virus sequence. When there are three or fewer differences, probe binding is nearly equivalent to binding between completely homologous sequences. A difference of four base pairs results in extremely weak binding, if any. The extent of binding of a 20-mer probe to a region with
four different base pairs depends upon the location of the differences. If differences are clustered at one end of the region, the free end is able to bind weakly, but if differences are scattered throughout the 20-base pair region, no binding occurs. Figure 1 also demonstrates probe discrimination of sequences that are characteristic of a particular host. Sw/CO/1/77, an H3N2 virus isolated from a pig, is characteristically a human virus and is recognized by the human probe but not by swine or avian probes. Although isolated from a human, A/ WI/3523/88, an H1N1 virus, is genetically a swine influenza virus according to nucleotide sequence analysis (13) and is bound only by the swine probe. This test demonstrates several features of the assay. Probes are specific and can differentiate sequences of influenza viruses that are characteristic of different host origins. While a minimum of four base pair differences are required for differentiation under our conditions, the actual location of differences in the sequence and base pair composition may affect probe binding. Interspecies transmission/reassortment in swine
The occurrence of genetic mixing was investigated with swine viruses isolated during the influenza prevalence survey in 1976— 1977. Since reassortment might involve the exchange of a single gene segment, we studied all 390 segments (65 viruses, six genes each). The results were uniformly characteristic of influenza viruses isolated from swine (table 4). Probes specific for influenza sequences from human or avian sources failed to bind to any of the genes. The viruses that were isolated from swine in the epidemiologic survey represented specimens collected over a period of 2 years from seven central/south central states. To expand the time frame of virus isolation, we studied H1N1 swine viruses collected from other geographic locations over a 15-year period (see table 1). Dot-blot analysis of these 48 genes again revealed nucleotide sequences characteristic of swine influenza vi-
Reassortment of Influenza A Viruses
493
Part A Probe for Swine-Specific Conserved Sequence* Probe
Human-Specific Probe
Avian-Specific Probe
1 |^^lj
6
r
t
2 ^H
2
3
'm^^
3
4
^J
5
Hi
7
• ' 1 4
5 6
7
.j£
•
'Probe binding is evidenced by the dark dot
PartB Number of Differences from Probe: Sequence: Control
Swine
Human
Avian
1 - Udom/72 (H3N2)
0
10
0
4
2 = Sw/TN/26/77(H1N1)
0
0
I
*
3 = Mal/NY/6750/78 (H2N2)
0
=5
0
4 = Sw/CO/1/77 (H3N2)
0
W
0
4
5 = Wl/3523/88 (H1N1)
0
1
Vol 136, No. 4 Printed in U.S A.
AS rights reserved
Interspecies Transmission and Reassortment of Influenza A Viruses in Pigs and Turkeys in the United States
Stephen M. Wright,1 Yoshihiro Kawaoka,1 Gerald B. Sharp,2 Dennis A. Senne,3 and Robert G. Webster1
Genetic reassortment between influenza A viruses in humans and in animals and birds has been implicated in the appearance of new pandemics of human influenza. To determine whether such reassortment has occurred in the United States, the authors compared the genetic origins of gene segments of 73 swine influenza virus isolates (1976-1990), representing 11 states, and 11 turkey virus isolates (1980-1989), representing eight states. The host origin of gene segments encoding the internal proteins of H1N1 swine and turkey influenza viruses was identified by developing a dot-Wot assay. All gene segments of swine influenza viruses were characteristic of influenza virus genes from that species, indicating that pigs may not be frequent participants in interspecies genetic exchange and reassortment of influenza viruses in the United States. In contrast, 73% of the turkey influenza virus isolates contained genes of swine origin. One turkey isolate was a reassortant having three genes characteristic of avian influenza virus and three of swine origin. These findings document a high degree of genetic exchange and reassortment of influenza A viruses in domestic turkeys in the United States. The molecular biologic techniques used by the authors should aid future epidemiologic studies of influenza pandemics. Am J Epidemiol 1992;136:488-97. epidemiologic methods; genetics; Orthomyxovirus type A, avian; Orthomyxovirus type A, human; Orthomyxovirus type A, porcine; reassortant viruses; swine; turkeys
Influenza A virus is responsible for frequent epidemics and occasional pandemics. Received for publication October 21,1991, and in final form March 11, 1992. Abbreviations' H, hemagglutmin; M, matrix or membrane protein; N, neuraminidase; NP, nucleoprotein; NS, nonstructural; PA, polymerase A; PB1, polymerase B1; PB2, polymerase B2. ' Department of Virology and Molecular Biology, St. Jude Children's Research Hospital, Memphis, TN. 1 Department of Btostatistics and Epidemiology, University of Tennessee, Memphis, TN. 3 National Veterinary Services Laboratories, Ames, LA. Reprint requests to Dr. Robert G. Webster, Department of Virology and Molecular Biology, St. Jude Children's Research Hospital, 332 North Lauderdale, Memphis, TN 38105. The authors thank Dr. Clayton Naeve and the Molecular Resource Center for preparation of oligonucleotide probes and pnmers. They also thank Glenith D. White for preparation of the manuscript This work was supported by US Public Hearth Service research grant AI-29680, National Institutes of Health CORE grant CA-21765, and the American Lebanese Syrian Associated Chantes.
From the years 1957 to 1986, there were 19 epidemics of influenza in the United States, each responsible for 10,000 excess deaths (1). More catastrophic are pandemics, such as "Spanish influenza," which claimed over 25 million lives worldwide in 1918 (2). Influenza A virus has several subtypes defined by changes in two surface proteins, hemagglutinin (H) and neuraminidase (N). H3N2 and H1N1 are the subtypes currently circulating among human populations. The genome of influenza A virus consists of eight segments of RNA. Reassortment is characterized by the exchange of genomic segments between two different influenza viruses that simultaneously infect a cell (3). Such reassortment of the genome has been reported for both avian (4) and human (5, 6) influenza viruses, but there is little information on how often or in which species this event occurs. Genetic reassortment be488
Reassortment of Influenza A Viruses
tween the influenza A viruses infecting humans and other species has been implicated in the generation of pandemic influenza (7). In this study, we have examined the occurrence of interspecies transmission (that is, virus transfer and replication in another species) and reassortment of influenza A viruses. The human influenza virus responsible for the 1957 pandemic contained three gene segments, H (8), N (8), and polymerase Bl (PB1) (9), that are characteristic of avian influenza. Similarly, two viral gene segments (H and PB1) of the 1968 pandemic were closely related to avian influenza. Thus, viral genes from widely different species can be part of the same genome. Two possibilities could account for interspecies transmission or reassortment of influenza A viruses. 1) Reassortment could occur in an avian or human host, although the available evidence indicates that avian influenza viruses are not able to replicate productively in human hosts. 2) An intermediate host could become infected by both avian and human influenza viruses. Swine are promising candidates for this role because they allow productive replication by either avian (10) or human (11) influenza viruses. Moreover, interspecies transmission of influenza virus has been reported to occur between swine and humans in the United States. One swine virus, Swine/Colorado/1/1977 (Sw/CO/1/77) (H3N2), has characteristics of a human virus (12), while the human virus A/Wisconsin/ 3523/1988 (A/WI/3523/88) (H1N1) has swine characteristics (13). It has been difficult to test these ideas because of the lack of techniques capable of accurate, rapid comparison of nucleotide sequences in minute samples of viral RNA. In an effort to assess the likelihood of viral gene reassortment in pigs and turkeys in the United States, we devised a sensitive assay to determine the origin of the gene segments based on the polymerase chain reaction and dot-blot hybridization. The method was successful because influenza virus genes possess conserved regions that remain host-specific (14). Relying on the vast amount of sequence information available for influenza A viruses (Y. Kawaoka and R. G. Webster,
489
unpublished manuscript; 15-22), we were able to design probes that would recognize and bind to genomic regions characteristic of a particular host group, e.g., human, swine, or avian. Reported here are the results of this analysis as applied to influenza A isolates collected over defined periods within geographically distinct areas of the United States.
MATERIALS AND METHODS Viruses
We studied a total of 84 H1N1 influenza A viruses of swine and turkey origin (table 1) which were obtained from the repository at St. Jude Children's Research Hospital, Memphis, Tennessee; all were grown in embryonated chicken eggs. Swine influenza viruses were collected during an epidemiologic survey (19761977) conducted to determine the prevalence of influenza viruses in this species (12). Briefly, nasal swabs from 200 randomly selected pigs were obtained weekly from slaughterhouses in Memphis, Tennessee, and Madison, Wisconsin. A total of 9,400 pigs were sampled, of which 478 were positive for influenza (prevalence, 5.1 percent). Sixty-five H1N1 influenza viruses were selected at random from the 1976-1977 survey for dot-blot analysis. The hosts of the viruses originated from seven states: Arkansas (AR), Georgia (GA), Illinois (IL), Missouri (MO), Mississippi (MS), Tennessee (TN), and Wisconsin (WI) (table 1). Additionally, eight other swine influenza viruses were examined to determine genetic host origins; these viruses represented a 15-year isolation period (1976-1990) from California (CA), Iowa (IA), Kansas (KS), and Minnesota (MN) and were obtained during routine sampling of herds for swine influenza. We also examined 11 H1N1 influenza A viruses from turkeys in the central United States. These samples were collected over a 10-year period (1980-1989) by US Department of Agriculture personnel who monitor flocks for influenza (table 1). The two surface glycoproteins (H and N) on the swine H1N1 influenza viruses studied had previ-
490
Wright et al.
TABLE 1.
H1N1* influenza A viruses of swine (Sw) and turkey (Ty) origin analyzed in a US study Virus
Survey swine virus (1976-1977) Sw/TN/- -/76 (24 samples)
Sw/TN/- -/77 (24 samples)
Sw/WI/--/76 (17 samples)
Isolate no.
State of swine origin
2-6,8-14 15-20,26 21 23, 24, 27 25
Illinois (IL) Tennessee (TN) Georgia (GA) Arkansas (AR) Missouri (MO)
2, 4, 5,15-17, 19, 20, 31, 36-38, 43, 44 9,13 11 12,25-27 29, 30, 35
Missouri (MO)
747, 750, 754, 756, 757, 762, 763, 764, 766, 769, 834, 846, 848, 888, 936, 940, 954
Illinois (IL) Mississippi (MS) Tennessee (TN) Arkansas (AR) Wisconsin (Wl)
Other swine viruses (1976-1990) Sw/Minnesota/27/77 Sw/Wisconsin/11/80 Sw/Kansas/3228/87 Sw/Wisconsin/915/88 Sw/CaJifomia/9007919/90 Sw/lowa/3421/90 Sw/California/9001707/91 Sw/lowa/23239/91 Turkey viruses (1980-1989) Ty/Kansas/4880/80 Ty/lowa/7352/80 Ty/Minnesota/1661/81 Ty/Missouri/1/81 Ty/Colorado/10091/81 Ty/South Dakota/7034/86 Ty/South Dakota/7740/86 Ty/Missourl/21939/87 Ty/North Carolina/1/88 Ty/Ohio/29962/88 Ty/Mlnnesota/12537/89 1
H, hemagglutnln; N, neurairtnidase.
ously been characterized antigenically and were characteristic of swine influenza viruses (12). Similarly, the surface glycoproteins of the turkey influenza isolates were characteristic of avian or swine influenza viruses (unpublished data based on antigenic and sequence analyses).
pital and were 18-20 base pairs in length. Table 2 lists forward and reverse primers and nucleotide locations for each gene that was evaluated. The sequences selected represent conserved regions for all influenza viruses. Oligonucleotide probes
Primers
Primers for the polymerase chain reaction were prepared in the Molecular Resource Center at St. Jude Children's Research Hos-
Oligonucleotides, prepared and purified in the Molecular Resource Center, were made for polymerase (PA, PB1, and PB2), nucleoprotein (NP), matrix or membrane protein
Reassortment of Influenza A Viruses
TABLE 2. Locations of primers used In polymerase chain reactions for analysis of swine and turkey influenza A viruses*
TABLE 3. Host-specific oligonucleotide probes used in a study of swine and turkey influenza A viruses* Gene
Nucleotide location segmentt
Forward primer
Reverse primer
segmentt and probe type
NS M NP PA
1-20 73-92 1,033-1,052 1-19 331-353 949-965
871-890 896-915 1,447_1,467 603-622 1,492-1,511 1,602-1,621
NS Swine Human Avian Control
PB1
PB2
• Forward primers were used lor synthesis of cDNA and amplification during the potymerase chain reaction. Reverse primers were used only for the polymerase chain reaction. These primers represent regions conserved by aD viruses. • t NS, nonstructural; M, matrix or membrane protein; NP, nucteoproteln; PA, polymerase A; PB1, polymerase B1; PB2, polymerase B2.
(M), and nonstructural (NS) gene segments. Four oligonucleotide probes were prepared for each gene: I) swine, 2) human, and 3) avian probes that would bind to their respective sequences, and 4) a control probe that would bind to a region common to all viruses. Control probes were used to verify that viral sequences of the gene being analyzed were indeed present. An example of the probe design is shown in the Appendix table. Table 3 summarizes the nucleotide location of each probe for each gene. Oligonucleotide probes were end-labeled with digoxigenin as described by Farquharson et al. (23). All reagents were obtained from Boehringer Mannheim (Indianapolis, Indiana). The reaction was incubated at 37°C for 5 minutes, followed by ethanol precipitation.
491
Nucleotide location
virus wtth Identical sequence
_. "w-"°-
88-107 266-286 452-470 539-557
Sw/TN/26/77 Udom/72 Mal/NY/6750/78
—* 15 —t
264-288 477-499 579-606 700-719
Sw/TN/26/77 Udom/72 Mal/NY/6750/78
16 16 16
1,350-1,371 Sw/TN/26/77 1,110-1,134 Udom/72 1,326-1,351 Mal/NY/6750/78 1,423-1,442
17 18 18
M Swine Human Avian Control NP Swine Human Avian Control PA Swine Human Avian Control PB1 Swine Human Avian Control PB2 Swine Human Avian Control
519-540 194-212 291-309 587-606
Sw/TN/26/77 NT/60/68 Ty/MN/1661/80
19 20 19
390-411 Sw/TN/26/77 1,029-1,047 Mem/8/88 1,290-1,313 Mal/NY/6750/78 1,492-1,511
9 9 21
1,539-1,561 Sw/TN/26/77 1,408-1,431 Mem/8/88 1,536-1,551 Mal/NY/6750/78 1,567-1,592
22 22 21
* The nucleotide location and the virus having the Identical sequence with the probe are shown. Control probes were based on regions conserved by al influenza viruses. t NS, nonstructural; M, matrix or membrane protein; NP, nucieoprcitein; PA, potymerase A; PB1, polymerase B1; PB2, polymerase B2. t Y. Kawaoka and R. G. Webster, unpublished manuscript.
The hybridization assay
Viral RNA was isolated from influenzainfected allantoic fluid according to the method of Bean et al. (24) with a single phenolxhloroform extraction. cDNA was prepared and amplified as outlined by Katz et al. (25). Briefly, cDNA was synthesized from viral RNA using reverse transcriptase (Life Sciences, St. Petersburg, Florida) at 42°C for 1 hour. cDNA was amplified for 35 cycles in a DNA thermal cycler (Perkin Elmer Cetus, Norwalk, Connecticut). Negative
controls (distilled water) were included during amplification to ensure that contamination did not occur. Six microliters of the polymerase chain reaction product was applied to a nylon membrane (Zeta Probe; Bio-Rad Laboratories, Richmond, California) through a vacuum manifold apparatus (Minifold; Schlicher & Schuell, Keene, New Hampshire). DNA was fixed to the membrane by
492
Wright et al.
ultraviolet radiation cross-linking (Stratalinker; Stratagene, La Jolla, California) as described by Allefs et al. (26). Hybridization and detection of probe binding were done as described by Casacuberta et al. (27). Briefly, prehybridization and hybridization of filters was conducted in an incubator/rotator at 42°C. Fifty nanograms of the digoxigeninlabeled probe was added and was allowed to hybridize overnight. After washing, anti-digoxigenin alkaline phosphataseconjugated Fab (antigen-binding) fragments (Boehringer Mannheim) were added to filters at room temperature (30 minutes). The bound probe was detected with X-phosphate and nitroblue tetrazolium salt in a colorimetric reaction (Genius Nucleic Acid Detection Kit; Boehringer Mannheim). A brownpurple color indicated probe binding. Four filters and four probes were used for each gene (control, human, swine, and avian). The polymerase chain reaction products of Turkey/Missouri/21939/87 were purified for nucleotide sequencing by Geneclean (Bio 101, La Jolla, California). The nucleotide sequences were determined by the dideoxynucleotide chain termination method (28).
RESULTS Probe specificity
We tested each probe against influenza viruses whose complete sequences were already known, to ensure that each of the probes was binding to its complementary sequence and not to heterologous sequences. Figure 1 displays results using the NS gene segment as an example. Part A shows results of the dot-blot hybridization; part B indicates the number of differences between the probe sequence and the viral sequence. There is no binding when there are five or more base pair differences between the 20mer probe and the virus sequence. When there are three or fewer differences, probe binding is nearly equivalent to binding between completely homologous sequences. A difference of four base pairs results in extremely weak binding, if any. The extent of binding of a 20-mer probe to a region with
four different base pairs depends upon the location of the differences. If differences are clustered at one end of the region, the free end is able to bind weakly, but if differences are scattered throughout the 20-base pair region, no binding occurs. Figure 1 also demonstrates probe discrimination of sequences that are characteristic of a particular host. Sw/CO/1/77, an H3N2 virus isolated from a pig, is characteristically a human virus and is recognized by the human probe but not by swine or avian probes. Although isolated from a human, A/ WI/3523/88, an H1N1 virus, is genetically a swine influenza virus according to nucleotide sequence analysis (13) and is bound only by the swine probe. This test demonstrates several features of the assay. Probes are specific and can differentiate sequences of influenza viruses that are characteristic of different host origins. While a minimum of four base pair differences are required for differentiation under our conditions, the actual location of differences in the sequence and base pair composition may affect probe binding. Interspecies transmission/reassortment in swine
The occurrence of genetic mixing was investigated with swine viruses isolated during the influenza prevalence survey in 1976— 1977. Since reassortment might involve the exchange of a single gene segment, we studied all 390 segments (65 viruses, six genes each). The results were uniformly characteristic of influenza viruses isolated from swine (table 4). Probes specific for influenza sequences from human or avian sources failed to bind to any of the genes. The viruses that were isolated from swine in the epidemiologic survey represented specimens collected over a period of 2 years from seven central/south central states. To expand the time frame of virus isolation, we studied H1N1 swine viruses collected from other geographic locations over a 15-year period (see table 1). Dot-blot analysis of these 48 genes again revealed nucleotide sequences characteristic of swine influenza vi-
Reassortment of Influenza A Viruses
493
Part A Probe for Swine-Specific Conserved Sequence* Probe
Human-Specific Probe
Avian-Specific Probe
1 |^^lj
6
r
t
2 ^H
2
3
'm^^
3
4
^J
5
Hi
7
• ' 1 4
5 6
7
.j£
•
'Probe binding is evidenced by the dark dot
PartB Number of Differences from Probe: Sequence: Control
Swine
Human
Avian
1 - Udom/72 (H3N2)
0
10
0
4
2 = Sw/TN/26/77(H1N1)
0
0
I
*
3 = Mal/NY/6750/78 (H2N2)
0
=5
0
4 = Sw/CO/1/77 (H3N2)
0
W
0
4
5 = Wl/3523/88 (H1N1)
0
1
