The role of amniotic passage in the egg-adaptation of human influenza virus is revealed by haemagglutinin sequence analyses

Transcription

1 Journal of General Virology (1993), 74, Printed in Great Britain 2047 The role of amniotic passage in the egg-adaptation of human influenza virus is revealed by haemagglutinin sequence analyses James S. Robertson, 1. Carolyn Nicolson, 1 Diane Major,' Edwin W. Robertson 2 and John M. Wood' 1 National Institute for Biological Standards and Control, Blanche Lane, South Mimms, Potters Bar, Hertfordshire EN6 3QG and 2 Medical Centre, Bank Street, Alexandria, Dunbartonshire G83 OLS, U.K. Obtaining an isolate of a human influenza virus in the allantoic cavity of the embryonated hen's egg is more efficient if the clinical sample is initially passaged in the amniotic cavity. To investigate the extent to which the variants present after allantoic propagation are also selected by amniotic passage, clinical virus passaged once in the amnion has been subjected to extensive genetic and antigenic analyses. The data indicate that the natural virus can replicate unrestrictedly within the amnion. However, exposure of amniotic virus to the allantois during the incubation period, which will occur through the hole between the amniotic and allantoic cavities caused by the inoculating needle, allows for the possibility of an egg-adapted variant establishing replication within the allantois and returning to the amnion. These observations illustrate why prior passage in the amnion increases the probability of a variant successfully establishing itself during a subsequent auantoic passage. Introduction The allantoic cavity of the embryonated hen's egg is an efficient and commonly used substrate for the propagation of influenza viruses. However, the initial isolation of human influenza virus in the allantoic cavity is generally considered to be inefficient, especially for influenza B virus, and virus does not replicate efficiently until it has' egg-adapted'. A greater efficiency of isolation can be achieved in eggs by initially passaging the virus in the amniotic cavity prior to cultivation :in the allantoic cavity (Burnet, 1940), and this is the usual route for inoculation to obtain an isolate in the egg (Hoyle, 1968). Based on analysis of virus present in clinical material and of virus propagated exclusively in mammalian tissue culture, there are now considerable data to show that egg-adapted human influenza viruses are variants which differ from non-egg-passaged virus by single amino acid substitutions in the haemagglutinin (HA) in the vicinity of the receptor binding site (Robertson et al, 1985, 1987, 1990, 1991; Katz et al., 1987, 1988, 1990). The eggadapted viruses which have been investigated have all been derived in the allantoic cavity, either directly or after amniotic passage, and it is unknown to what extent selection of variants takes place during amniotic passage. The nucleotide sequence data reported in this paper have been submitted to the EMBL database and assigned the accession number X To address this question we have analysed an influenza B virus passaged solely in the amniotic cavity. Typically, for influenza B virus, egg-adaptation is accompanied by the loss of a specific glycosylation site from the tip of the HA molecule through substitution of either Asn-196 or Thr-198 (Robertson et al, 1985, 1990). By using PCR and M13 cloning we have been able to analyse the sequence of the HA of virus derived in the amniotic cavity of single eggs and have compared it to the sequence present in the original clinical sample and to that of virus passaged further in the allantoic cavity. Methods Amniotic isolation. An influenza B virus, B/NIB/48/90, was isolated in the amniotic cavity of 10-day-old embryonated hens' eggs in the following way and as described by Hoyle (1968). A hole was drilled in the shell at the air-sac end and a small quantity of sterile liquid paraffin applied to the shell membrane covering the chorioallantoic membrane (CAM) to render it transparent. The CAM was pierced with fine forceps and the amniotic sac carefully pinched and gently drawn up clear of the allantoic fluid. Two-hundred microlitres of undiluted clinical material was inoculated into the amniotic sac using a fine gauge needle with confidence that the inoculum had entered the amniotic cavity. The amnion was gently lowered into the allantoic cavity and the egg sealed with Sellotape. After 3 to 4 days incubation at 33 C, the seal was removed and a quantity of allantoic fluid harvested and kept. The amniotic fluid was then collected using a Pasteur pipette. These manipulations were performed with extreme care and satisfactory amniotic inoculations and harvests were obtained for this study. Allantoic passage. One-hundred microlitres of a one tenth dilution of SGM

2 2048 J. S. Robertson and others Table 1. Description of virus samples and their antigenic and genetic characterization Sequence at B/NIB/48/90 Passage Route Fluid harvested HA titre HI with BM15* HA1 residues 196 to 198 Clinical specimen None None Throat wash YDt ND Ash-x- Thr MDCK 1st MDCK Culture fluid 128 < 200 Asn-x -Thr Egg 1.1 1st Amniotic Amniotic 40 ND Asn-x-Thr Allantoic 10 No Asn-x-Thr 2nd Allantoic Allantoic Hetero Egg 1.2 1st Amniotic Amniotic :~ Heterol[ Allantoic > Hetero]l 2nd Allantoic Allantoic Asp-x--Thr Egg 1.3 1st Amniotic Amniotic 320 ND Asn-x-Ala Allantoic Asn-x-Ala 2nd Allantoic Allantoic Asn-x-Ala * MAb BM15 was raised against B/Memphis/6/86 and was a gift from R. G. Webster. t ~D, Not done. :~ Partial agglutination. Heterogeneous sequence consisting maialy of Asn/Asi~x-Ala/Thr. [[ Heterogeneous sequence consisting mainly of Asn/Asp-x Thr. virus derived in the amuion was inoculated into the allantoic cavity using standard procedures, and incubated at 33 C for 3 days. Tissue culture isolation. Virus was isolated on MDCK ceils from clinical material using standard procedures. Serological assays. Haemagglutination and haemagglutination-inhibition (HI) assays were performed in microtitre plates using turkey red blood cells. Sequencing. Sequencing of the HA1 coding region was performed either on PCR-amplified cdna or on M13 clones of the amplified DNA as described previously for influenza B virus (Robertson et al., 1990). Briefly, RNA was extracted from 50 to 100 I J1 of sample, cdna was synthesized using reverse transcriptase and primer B/17/1 (TTTCTAATATCCACAAAATGAAGGC) and the HAl coding region was amplified in a PCR with cloning primers Eco/B/35/1 (CAAAATGAATTCAATAATTGTACTACTCAT) and Bam/B/1092/2 (TCTATATTTGGATCCATTGGCCAGCTT) exactly as described. If insufficient DNA was generated from the first amplification, a sample of cdna was amplified with primers B/17/1 and B/1140/2 (ACCAGCAATAGCTCCGAAGAAACC) which flank the above cloning primers, and a sample of this amplified DNA was subjected to nested PCR with the cloning primers Eco/B/35/1 and Barn/B/1092/2. Amplified DNA was either sequenced directly or cloned into M13mp18/19 using standard techniques. Direct dideoxynucleotide sequencing of PCR DNA was performed using 5' ~2p_ labelled primers specific for the HA1 region. M13 sequencing was performed using the M 13 universal primer or HA-specific primers with [~-3~P]dATP present in reaction mixes. Nucleotide sequences were analysed using the Staden (1982) and the GCG (Devereux et al., 1984) programs. Results Initially, 200 ltl samples of a throat wash containing an influenza B virus (B/NIB/48/90) were inoculated directly into the amniotic cavity of five embryonated hens' eggs as described in Methods, and after incubation for 4 days both the allantoic and amniotic fluids were individually harvested from four of the eggs (one egg was non-viable). As assayed by haemagglutination, three eggs were positive for virus growth and these amniotic samples were then further passaged allantoically. In addition to these egg isolates, virus was derived directly from the clinical specimen on MDCK cells. This provided 11 virus samples for initial analysis: the original clinical material, an MDCK cell-derived virus, three amniotic samples (Am virus), three corresponding samples of allantoic fluid harvested from the eggs infected amniotically (A1 virus), and three allantoic samples from eggs inoculated allantoically with each of the amniotic viruses (AreA1 virus). The haemagglutination titres of the amniotic fluids varied between 40 and 1280 (Table 1) and in each case the corresponding allantoic fluids had a comparable titre. Samples for which there was sufficient virus were analysed by HI assay using a monoclonal antibody (MAb), BM15, which we previously found could discriminate between MDCK cell-derived and egg-adapted variants of influenza B virus. The amniotic virus from egg 1.2 reacted with BM15 in the HI test but showed partial agglutination which indicates that the virus sample was likely to be a mixture. Two of the allantoic fluids obtained from the eggs inoculated amniotically were assayed and both reacted to high titre with BM 15. All three virus samples cultivated specifically in the allantois (AmA1 virus) reacted to high titre with BM15 whereas MDCK-grown virus failed to react. Viral RNA was extracted from each of the 11 samples and the cdna corresponding to the HAl coding region was PCR-amplified. The region of amplified DNA that

3 Amniotic influenza B virus HA sequences 2049 Table 2. Summary of the sequences at HA1 codons 196 and 198 for 300 clones B/NIB/48/90 Clinical specimen Egg 1.1 Egg 1.2 Egg 1.3 * Unpassaged material. Number of clones with sequence Allantoic Amniotic passage passage (lst passage) (2nd passage) Sequence at HA1 residues Amniotic Allantoic Allantoic 196 to 198 fluid fluid fluid Asn-x- Thr 30/30* Ash x-thr 29/30 29/30 His-x-Thr 1/30 -- Asn-x-Asn / 30 Ser-x-Thr 1/ /30 Asn x-ala /30 Asp-x-Thr /30 Asn-x-.Thr 10/30 7/30 2/30 Lys x-thr /30 Asn-x-Asn 3/30 1/30 2/30 Asp-x-Thr 17/30 22/30 25/30 Asn-x--Thr Se~x-Ala -- 1/30 -- Asn-x--Ala 30/30 29/30 30/30 codes for the glycosylation site at HA1 residues 196 to 198 was sequenced directly and the deduced amino acid sequence for residues 196/198 is shown in Table 1. In some instances there was obviou; heterogeneity of the sequence for these codons. Virus in the clinical specimen, the MDCK cell-grown virus, and virus in the amniotic (Am) and corresponding allantoic fluids harvested from egg 1.1 after the amniotic passage each had the glycosylation sequence Asn--x-Thr at 196 to 198. The sequence obtained after allantoic passage of egg 1.1 amniotic fluid was heterogeneous. All samples derived from egg 1.2 had the sequence Asp-x-Thr to varying degrees with obvious heterogeneity in the Am and A1 virus samples. All egg 1.3 samples had the sequence Asn-x-Ala. In order to assess the extent of heterogeneity in the HA1 for each of the virus samples above, the PCRamplified DNA derived from virus present in the clinical sample and the above nine egg-derived samples was cloned into M t3 and a region of approximately 250 bases, which included the codons for residues 196 to 198, was sequenced from 30 clones per virus sample. In this way, 300 clones and approx bases were analysed and the deduced amino acid sequence for residues 196 and 198 for each of the 300 clones is shown in Table 2. As previously observed in the analysis of clinical material (Robertson et al., 1990), all 30 clones derived from the clinical specimen had the potential glycosylation sequence Asn-x-Thr at residues 196 to 198. For egg 1.1, the majority of clones derived from the amniotic (Am) and corresponding allantoic (A1) fluids had the sequence Asn-x-Thr. On passage of this amniotic virus in the atlantoic cavity, the derived clones had various sequences at residues 196 to 198 including Asp-x-Thr and Asn-x-Ala, and none had the sequence Asn--x-Thr. For egg 1.2, the clones from the amniotic and corresponding allantoic fluid had predominantly Asp-x-Thr and Asnx-Thr with more of the former than the latter sequence in both fluids. After allantoic passage of egg 1.2 amniotic fluid, 25 out of 30 clones had the Asp-x-Thr sequence and only two clones had Asn-x-Thr. In all egg 1.2 fluids analysed, a few clones had the sequence Ash-x-Ash and one clone in the (AmA1) allantoic fluid had Lys-x-Thr. For egg 1.3, the clones derived from virus in both the amniotic and the corresponding allantoic fluid, and also in the (AmA1) allantoic fluid derived from the allantoic passage, the sequence Asn-x-Ala predominated, and no clones had the original Asn-x-Thr sequence. Since the nature of the virus population between each of the three amniotic samples was diverse, additional amniotic samples were derived from B/NIB/48/90 clinical material. In this second experiment, 10 eggs were inoculated amniotically with 100 gl of the throat-wash material and after 3 days incubation both the amniotic fluid and a sample of allantoic fluid were harvested, as before. Five amniotic fluids and one corresponding allantoic fluid were positive for virus growth as measured by haemagglutination. The five amniotic samples were then passaged in eggs, allantoically. These five amniotic samples, the one corresponding allantoic sample, and the five allantoic passaged (AmA1) samples were characterized antigenically using MAb BM 15 and the HA 1 region coding for residues 196 to 198 was analysed by direct sequence analysis of PCR-amptified cdna corresponding to the HA1 coding region (Table 3). The HA titre of the five amniotic and the one corresponding allantoic samples varied between 10 and None of the amniotic viruses reacted in HI with MAb BM 15 and direct sequence analysis of the PCRamplified cdna clearly indicated the deduced sequence at HA1 residues 196 to 198 to be Asn-x-Thr for all six samples with no indication of any heterogeneity of the sequence. Each of the five amniotic samples grew successfully upon allantoic passage and four of them now reacted strongly with BM15. Sequence analysis indicated that a variant had been selected in these four samples with obvious heterogeneity within the AmA1 virus derived from egg The AreA1 virus from egg 2.9 retained the sequence Asn-x-Thr with no indication from the sequencing autoradiogram of heterogeneity. With the exception of the codons for residues 196 and 198, the consensus sequence of the entire HA1 coding region of virus present in the clinical sample, the MDCK

4 2050 J. S. Robertson and others Table 3. Antigenic" and genetic characterization of virus samples (second experiment) Amniotic passage* (lst passage) Allantoic passage (2nd passage) HI titre HI titre B/NIB/48/90 HA with Sequence at HA with Sequence at Egg no. titre BM to 198 fitre BM to < 200 Asn-x-Thr Asn-x-Ala < 200 Asn-x-Thr Asn-x Ala < 200 Asn-x-Thr Ser-x-Thr < 200 Asn-x -Thr 160 < 200 Asn-x Thr < 200 Asn-x Thr Asn x Ala/Pro 2.10 (A1) 10 ND'~ Asn-x-Thr * Five amniotic samples and one corresponding allantoic sample [2.10(A1)] were analysed. t ND, Not done. cell-derived virus and the nine egg-grown viruses derived in the first experiment was identical (not shown). B/NIB/48/90 is antigenically like B/Victoria/2/87 (not shown) and its HA nucleotide sequence is comparable to other B/Victoria-like viruses isolated at the same time (Rota et al., 1992). During sequence analysis of the M13 clones many unique single base substitutions were observed in addition to those substitutions occurring at codons 196 and 198. The majority of these are presumed to derive from a combination of use of reverse transcriptase and Taq polymerase both of which are known to have an error rate compatible with the frequency of single base substitutions observed in this study, and in other (Robertson et al., 1990, 1991) studies. Discussion Previously, we have observed that passage of non-eggadapted influenza B virus in the allantois typically results in the selection of a virus with a substitution in the HA1 at either residue 196 or 198. In this study, after virus was derived from clinical material by a single passage in the amniotic cavity, the nature of the virus varied. For most samples (six out of eight) there was sequence identity at residues 196 and 198 with virus present in the original clinical material. These data indicate that the original virus present in the clinical specimen can replicate successfully within the amniotic cavity without selection. After allantoic passage of the amniotic samples, a variant(s) with a substitution at either Asn-196 or Thr- 198 was selected for all but one of the samples. In this one sample (egg 2.9), the (AreA1) allantoically passaged virus retained the Asn-x-Thr sequence suggesting that no selection had taken place. This has been observed infrequently before (unpublished observations) and is the subject of a separate study. On two occasions, in the first experiment, there was partial or total selection of a variant after a single amniotic passage (eggs 1.2 and t. 3 respectively). From the detailed analysis in the first experiment, the nature of the virus population in the allantoic fluid harvested concurrently with the amniotic fluid was very similar to that present in the amniotic fluid, with respect to residues 196 to 198 (Table 2). This suggests that mixing between the two compartments had occurred. Although amniotic inoculation was performed with extreme care, inoculation into the amnion inevitably results in a hole in the amniotic membrane between the amniotic and the allantoic cavities. During the 4 day incubation, diffusion and movement of the embryo are likely to cause mixing of the fluids between the two compartments and original virus replicating within the amnion may enter the allantoic cavity. If a variant with the appropriate substitution (i.e. loss of the Asn-x-Thr sequence), either present in the inoculum or arising from intermediate growth of the original virus in the amnion, enters the allantoic cavity, it may successfully replicate. If this occurs sufficiently early in infection there is time for considerable amplification of the variant within the allantoic cavity. Such variants will distribute between the allantoic and amniotic cavities where they are likely to replicate. After 4 days it may thus be possible to observe egg-adapted variants within the virus population in both compartments by sequence analysis, even to the extent that they predominate in the population. For the samples derived in the second experiment in which amniotic incubation was for only 3 days, there was no evidence for the presence of egg-adapted variants and virus was detected in only one allantoic sample whose corresponding amniotic fluid had a very high haemagglutination titre. Thus, the increased efficiency of isolating virus in the allantois from a clinical specimen by prior passage in the amnion appears to be due mainly to increased numbers of virions in the amniotic sample being inoculated

5 Amniotic influenza B virus HA sequences 2051 allantoically. The frequency of a variant within non-eggadapted virus which is capable of replication in the allantois is appoximately 10 -~ (Schild et al., 1983). Since clinical material is likely to contain < 10 ~ p.f.u./ml, the probability of the appropriate variant being present within a clinical specimen is slight. Amplification of this material within the amnion (to 10 ~ to 108 p.f.u./ml) increases the likelihood that the allantoic inoculmn will contain a variant capable of replication within the allantois. In addition, there is the possibility that by leakage of virus replicating within the amnion into the allantois, an egg-adapted variant has the opportunity of being selectively amplified and co-harvested with the amniotic virus. No egg-variants were observed in six of the amniotic samples, but it is possible that some enrichment of variants to levels > 1 in 105 had occurred and this would also contribute to an increased efficiency of isolation within the allantois. Other possibilities exist for the appearance of variants during amniotic passage, e.g. selection within other tissues of the embryo, and for the appearance of virus within the allantoic compartment, e.g. by swallowing and passage through the gut of the embryo. However, since selection of variants by direct allantoic passage has been observed previously, selection within the allantois of virus leaking from the amnion remains the most probable explanation. These observations and conclusions are comparable to and similarly explain the original observations on egg adaptation of human influenza A (H1N1) virus made 50 years ago by Burnet & Bull (1943). These were that clinical material passaged in the amnion had agglutinating characteristics described as 'O' (original) but which changed to' D' (derived) upon repeated passage in the amnion or by further passage in the allantois. Katz & Webster (1992) recently demonstrated that influenza A virus isolated on primary chick kidney cells was identical to that isolated on MDCK cells whereas subsequent passage in the allantois selected a variant. It would appear that it is solely the cells of the allantois which present a restriction to the growth of original human influenza viruses. References BURNET, F. M. (1940). Influenza virus infections of the chick embryo lung. British Journal of Experimental Pathology 21, BURNET, F.M. & BULL, D.R. (1943). Changes in influenza virus associated with adaptation to passage in chick embryos. Australian Journal of Experimental Biology and Medical Science 21, DEWREUX, J., HAEBERLI, P. & SMrrnxEs, O. (1984). A comprehensive set of sequence analysis programs for the VAX. Nucleic Acids Research 12, HOYLE, L. (1968). The It~uenza Viruses, pp Wien: Springer- Verlag. KATZ, J.M. & WEBSTER, R.G. (1988). Antigenic and structural characterization of multiple subpopulations of H3N2 influenza virus from an individual. Virology 165, KATZ, J. M. & WEBSTER, R. G. (1992). Amino acid sequence identity between the HA1 of influenza A (H3N2) viruses grown in mammalian and primary chick kidney cells. Journal of General Virology 73, KATZ, J.M., N~VE, C.W. & WEBSTER, R.G. (1987). Host cellmediated variation in H3N2 influenza viruses. Virology 156, KATZ, J. M., WANG, M. & WEBSTER, R. G. (1990). Direct sequencing of the HA gene of influenza (H3N2) virus in original clinical samples reveals sequence identity with mammalian cell-grown virus. Journal of Virology 64, ROBERTSON, J.S., NAEVE, C.W., WEBSTER, R.G., BOOTMAN, J.S., NEWMAN, R. & SCHILD, G.C. (1985). Alterations in the hemagglutinin associated with adaptation of influenza B virus to growth in eggs. Virology 143, ROBERTSON, J.S., BOOTMAN, J.S., NEWMAN, R., OXFORD, J.S., DANn~LS, R. S., WEBSTER, R. G. & SCHmD, G. C. (1987). Structural changes in the hemagglutinin which accompany egg adaptation of an influenza A (H1N1) virus. Virology 160, ROBERTSON, J. S., BOOTMAN, J. S., NICOLSON, C., MAJOR, D., ROBERT- SON, E. W. & WOOD, J. M. (1990). The hemagglutinin of influenza B virus present in clinical material is a single species identical to that of mammalian cell-grown virus. Virology 179, ROBERTSON, J. S., NICOLSON, C., BOOTMAN, J. S., MAJOR, D., ROBERT- SON, E.W. & WOOD, J.M. (1991). Sequence analysis of the haemagglutinin (HA) of influenza A (H1N1) viruses present in clinical material and comparison with the HA of laboratory-derived virus. Journal of General Virology 72, ROTA, P.A., HEMPHILL, M.L., WHISTLER, T., REGNERY, H.L. & KENDAL, A. P. (1992). Antigenic and genetic characterization of the haemagglutinins of recent cocirculating strains of influenza B virus. Journal of General Virology 73, SCmLD, G. C., OXFORD, J. S., DE JoNa, J. C. & WEBSTER, R. W. (1983). Evidence for host ceil selection of influenza virus antigenic variants, Nature, London 303, STADEN, R. (1982). Automation of the computer handling of gel reading data produced by the shotgun method of DNA sequencing. Nucleic Acids Research 10, (Received 26 February 1993; Accepted 24 May 1993)