Highly Pathogenic H5N1 Influenza Viruses Carry Virulence Determinants beyond the Polybasic Hemagglutinin Cleavage Site

Highly pathogenic avian influenza viruses (HPAIV) originate from avirulent precursors but differ from all other influenza viruses by the presence of a polybasic cleavage site in their hemagglutinins (HA) of subtype H5 or H7. In this study, we investigated the ability of a low-pathogenic avian H5N1 strain to transform into an HPAIV. Using reverse genetics, we replaced the monobasic HA cleavage site of the low-pathogenic strain A/Teal/Germany/Wv632/2005 (H5N1) (TG05) by a polybasic motif from an HPAIV (TG05poly). To elucidate the virulence potential of all viral genes of HPAIV, we generated two reassortants carrying the HA from the HPAIV A/Swan/Germany/R65/06 (H5N1) (R65) plus the remaining genes from TG05 (TG05-HAR65) or in reversed composition the mutated TG05 HA plus the R65 genes (R65-HATG05poly). In vitro, TG05poly and both reassortants were able to replicate without the addition of trypsin, which is characteristic for HPAIV. Moreover, in contrast to avirulent TG05, the variants TG05poly, TG05-HAR65, and R65-HATG05poly are pathogenic in chicken to an increasing degree. Whereas the HA cleavage site mutant TG05poly led to temporary non-lethal disease in all animals, the reassortant TG05-HAR65 caused death in 3 of 10 animals. Furthermore, the reassortant R65-HATG05poly displayed the highest lethality as 8 of 10 chickens died, resembling “natural” HPAIV strains. Taken together, acquisition of a polybasic HA cleavage site is only one necessary step for evolution of low-pathogenic H5N1 strains into HPAIV. However, these low-pathogenic strains may already have cryptic virulence potential. Moreover, besides the polybasic cleavage site, the additional virulence determinants of H5N1 HPAIV are located within the HA itself and in other viral proteins.


Introduction
Highly pathogenic avian influenza viruses (HPAIV) are the causative agents of fowl plague [1,2] which causes devastating economic losses in gallinaceous poultry. In addition, several HPAIV strains are able to infect humans and, therefore, are considered as potential precursors for future influenza pandemics [3]. For infection, the envelope glycoprotein hemagglutinin (HA) precursor HA0 requires proteolytic cleavage by cellular proteases into the two subunits HA1 and HA2. Mammalian and lowpathogenic avian influenza A viruses (LPAIV) carry an HA cleavage site with a monobasic motif susceptible to trypsin-like proteases which confine viral replication to the respiratory or gastrointestinal tract. In contrast, HPAIV possess a polybasic HA cleavage site cleavable by furin [4,5], which is ubiquitous and thus supports systemic viral replication. This polybasic HA cleavage site is the prime virulence determinant of HPAIV [6,7,8] which originate from LPAIV precursors [4,9,10,11,12,13,14,15,16]. Acquisition of a furin recognition motif was shown to result from different events such as recombination of the HA gene with 28S ribosomal RNA [17] or with sequences encoding other viral proteins like the nucleoprotein (NP) gene of an unrelated virus [15] or the HA and matrix protein genes (M) from the same virus [16]. An alternative proposed mechanism is polymerase slippage on template regions with stable secondary structures [13,14]. In mammalian influenza viruses, virulence determinants have been attributed to the HA [18,19,20,21,22], NA [18,23], NS1 [24,25,26,27,28], NP and polymerase proteins [18,29,30,31,32,33,34]. In HPAIV, beside the polybasic HA cleavage site, the caspase cleavage motif in the M2 protein and deletions within the NA stalk region were associated with increased virulence [35,36,37,38,39,40]. Furthermore, introduction of the NS gene from an H5N1 HPAIV into an H7N1 fowl plague strain rendered it virulent for mice [41]. Recently, we demonstrated that the acquisition of a polybasic cleavage site by an LPAIV H3N8 strain is not sufficient for immediate transformation into an HPAIV, and that additional virulence determinants other than the polybasic HA cleavage site are required [42]. However, it remained to be analyzed whether H5 or H7 LPAIV, which are considered HPAIV precursors, have to undergo further evolutionary changes prior to or after acquisition of a polybasic cleavage site. Therefore, we addressed in this study the question whether a polybasic cleavage site engineered into the HA of an H5N1 LPAIV leads to immediate transformation into an HPAIV. To elucidate the virulence potential of all viral genes of H5N1 HPAIV in chicken further, we generated two H5N1 reassortants carrying an HPAIV HA plus the remaining LPAIV genes, or, in reversed composition, the LPAIV HA with engineered polybasic cleavage site plus the HPAIV genes.

Generation of Recombinant Viruses
As parental strains we used a recent H5N1 LPAIV isolated in Germany in 2005, A/Teal/Germany/Wv632/2005 (H5N1) [43] as well as the first HPAIV H5N1 isolate derived from the outbreak in wild swans on the island of Rügen in February 2006, A/Swan/ Germany/R65/06 (H5N1) [44]. First, plasmid-based reverse genetics systems were established for both strains, resulting in the recombinant viruses TG05 (this study) and R65 [45], respectively. To introduce a polybasic HA cleavage site into TG05 (H5N1), we replaced its monobasic HA cleavage site with a polybasic motif from HPAIV A/Duck/Shanghai/13/01 (H5N1) by site-directed mutagenesis and used this plasmid for generation of the mutant TG05 poly . Considering possible structural constraints within the HA of the parental TG05, we also adapted the amino acids adjacent to the cleavage site to those from the HA of A/Duck/Shanghai/13/01 (H5N1) ( Table 1). Furthermore, we rescued two reassortants carrying the HA from HPAIV R65 plus the remaining seven genes from TG05 (TG05-HA R65 ), or, in reversed composition, the mutated TG05 HA plus the remaining R65 genes (R65-HA TG05poly ). Remarkably, we could not rescue a mutant of TG05 with the polybasic cleavage site region from R65 suggesting structural impairments of the mutated HA.

TG05 poly has in-vitro properties of an HPAIV
To address the question whether the HA cleavage site mutant and the two reassortants are dependent on trypsin for multicycle replication, we performed plaque assays. Whereas the parental TG05 required exogenous trypsin for plaque formation, all the viruses with a polybasic HA cleavage site, TG05 poly , TG05-HA R65 , and R65-HA TG05poly , yielded plaques independent of trypsin ( Fig. 1). Proteolytic HA cleavage was analyzed by immunoblotting. Corresponding to plaque assays, the HA0 precursor of TG05 remained uncleaved in the absence of trypsin in contrast to HA0 of TG05 poly , TG05-HA R65 , and R65-HA TG05poly (Fig. 2), which were cleaved into the HA1 and HA2 fragments. The HA0 cleavage was incomplete in all viruses studied. Multicycle growth kinetics of TG05 showed a strong dependence on exogenous trypsin as the virus failed to replicate efficiently in the absence of trypsin. TG05 poly replicated both in the presence and in the absence of trypsin, although with a delay compared with the other viruses. However, both reassortant viruses TG05-HA R65 and R65-HA TG05poly replicated to high titers irrespective of trypsin (Fig. 3). In summary, these results demonstrate that TG05 poly , TG05-HA R65 , and R65-HA TG05poly undergo multicycle replication in the absence of trypsin in contrast to their parent virus TG05 and, thus, display HPAIV phenotypes in-vitro.

Pathogenicity in chicken
To investigate the virulence of TG05, TG05 poly , TG05-HA R65 , and R65-HA TG05poly in chicken, 10 animals each were infected with 10 5 pfu of the respective virus oculonasally, and observed daily for clinical symptoms for 10 days (Table 2 and Fig. 4). In contrast to the low-pathogenic parental virus TG05 which proved to be completely innocuous as expected, the HA cleavage site mutant TG05 poly and the two reassortants TG05-HA R65 and R65-HA TG05poly were pathogenic in chicken to an increasing degree. Infection with TG05 poly led to temporary non-lethal disease in all animals (predominantly mild symptoms such as ruffled feathers, depression or diarrhea mainly from day 3 to 6). After inoculation with TG05-HA R65 3 of 10 animals died, whereas R65-HA TG05poly displayed the highest lethality as 8 of 10 infected chickens died, thus exhibiting the phenotype of a natural HPAIV (Table 2).

Organ tropism and tissue lesions in chicken
On day 4 post inoculation, samples from cerebellum, cerebrum, lung, nose, trachea, caecum, duodenum, kidney, and pancreas from three (TG05, TG05 poly and TG05-HA R65 ) and four (R65-HA TG05poly ) additionally infected birds per group were removed and investigated for viral spread in organs and the extent of lesions in affected tissues. No influenza virus antigen or histomorphological lesions were found in organs of TG05 infected birds. In one of the three TG05 poly -inoculated birds, single positive neurons, pneumocytes II, endothelial cells, lymphocytes, and macrophages could be detected in cerebrum, lung, nasal conchae, trachea, caecum, and duodenum. Furthermore, in one TG05-HA R65infected bird, small clusters of infected cells were present in cerebellum and cerebrum suggesting that already the introduction of the R65 HA facilitates neurotropism.
All four R65-HA TG05poly -infected birds exhibited larger and more intensely stained clusters of infected cells within cerebellum, cerebrum, and nasal mucosae; several birds displayed single positive cells in lung, trachea, caecum, duodenum, kidney, and pancreas. After infection with the recombinant HPAIV R65 [45], the picture of infected cells appeared more intense (Table 3, Fig. 5). These results demonstrate a broader organ tropism of R65-HA TG05poly -inoculated birds compared with TG05 poly -inoculated birds demonstrating the relevance of the other viral genes for systemic spread.
On day 10 post inoculation, organs were removed from four surviving birds each of the TG05, TG05 poly and TG05-HA R65infected groups. No antigen was detectable in tissues of these birds, except for a few positive neurons and glial cells in the cerebellum from one TG05-HA R65 -infected bird, and in the cerebrum from two surviving R65-HA TG05poly -infected animals, as well as inflammatory cells in the nasal mucosa from one of those birds, demonstrating the recovery of the animals (Table S1). Taken together, the presence of influenza virus antigen corresponded to degeneration and necrosis of the affected tissues accompanied by lymphocytic and histiocytic infiltration. Extent of organ tropism and severity of lesions corroborate with the course of disease. In HPAIV, viral neurotropism can be facilitated by the HA gene solely and systemic spread is mediated also by the internal protein genes.

Discussion
Influenza virus virulence is a polygenic trait which had been established by generating reassortants neurovirulent for mice from two apathogenic strains, and reassortants apathogenic in chicken from two virulent strains [46,47]. Furthermore, reassortment studies based on two H5N1 HPAIV strains revealed that the exchange of the HA, NP, M2 or NS1 proteins resulted in increased mortalities and expanded tissue tropism in chicken. Moreover, replacement of the PB2 or PB1 proteins led to decreased replication in tissues and, consequently, a decrease in virulence [48]. These findings can be attributed to involvement of these proteins in virulence or to constraints of gene segment exchanges by reassortment in HPAIV. However, the prime determinant of virulence is the polybasic HA cleavage site [6,7,8] as its conversion into a monobasic motif invariably resulted in loss of virulence [7]. On the other hand, introduction of such a polybasic motif into the HA cleavage site of a low-pathogenic H3N8 strain did not lead to transformation into an HPAIV indicating the existence of additional virulence determinants in the HA and/or the other viral proteins [42].
To shed more light on virulence determinants required for evolution of HPAIV which, in nature, derive from low-pathogenic precursors of subtypes H5 or H7 [4,9,10,11,12,13,14,15,16], we addressed the question whether a low-pathogenic avian strain of the H5N1 subtype would transform into a highly pathogenic strain after introduction of a polybasic HA cleavage site. For that purpose, we replaced the monobasic HA cleavage site from TG05 by a polybasic motif from the HPAIV A/Duck/Shanghai/13/01 (H5N1) resulting in the mutant TG05 poly . Remarkably, attempts to obtain a similar mutant with the HA cleavage site of HPAIV R65 were not successful indicating structural incompatibilities with the TG05 HA.
Although TG05 poly was able to replicate in the absence of trypsin in cell-culture, and, thus, exhibits the in-vitro phenotype of   an HPAIV, this mutant caused only temporary mild disease in chicken and, accordingly, could be detected only in isolated single cells accompanied with minor lesions in affected organs. These observations differ from those obtained by the H3N8 polybasic cleavage site mutants generated from the LPAIV A/Duck/ Ukraine/1/1963 which caused considerably less symptoms in chicken [42] compared with TG05 poly indicating that the lowpathogenic H5N1 strain already carried cryptic virulence determinants. Correspondingly, repeated air sac inoculation of a low-pathogenic H5N3 isolate in 1-day-old-chickens did not lead to high virulence for 4-to-6-week-old chickens, whereas only two further passages in brain yielded an HPAIV [49].
To demonstrate the existence of virulence determinants besides the polybasic cleavage site in the HA or in other genes of H5N1 HPAIV, we generated two reassortants, TG05-HA R65 carrying the HA from the HPAIV R65 plus the remaining seven genes from TG05 and, in reverse gene constellation, R65-HA TG05poly composed of the mutated TG05 HA plus the other R65 genes. Whereas TG05-HA R65 already had a lethality of 30%, R65-HA TG05poly caused a lethality of 80%, displaying the phenotype of a ''natural'' HPAIV. These findings indicate that in H5N1 HPAIV the HA gene alone provides for a certain level of virulence, whereas the complete viral genetic background influences pathogenicity to a major extent.
Compared with the very broad organ tropism of the HPAIV R65 in chicken [50], the mutant TG05 poly and the two reassortants TG05-HA R65 and R65-HA TG05poly displayed generally less viral spread in organs and less extensive lesions in surrounding tissues. However, spread and lesions increased with the two reassortants from which R65-HA TG05poly exhibited the phenotype of an authentic HPAIV. This finding again emphasizes the relevance of HA and the other viral genes for virulence. Moreover, all birds which had viral antigen detectable in inner organs, i.e. those infected by TG05 poly , TG05-HA R65 or R65-HA TG05poly also exhibited a neuronal infection. Thus, the presence of the HPAIV R65 HA alone in a LPAIV background suffices for viral neurotropism which is considered pivotal for the fatal course of fowl plague [51,52,53].
Previous studies on HPAIV strains revealed that (1) removal of the polybasic HA cleavage site results in a drastic decrease in pathogenicity [7]; (2) virulence is a multigenic trait as reassortants derived from HPAI strains can be attenuated [46], two LPAI strains can reassort to a highly virulent virus [47] and reassortants with exchanged HA, NP, M or NS, PB2 or PB1 gene have an altered virulence [48]; (3) NA stalk deletion leads to a certain increase in virulence in an LPAIV [40]; and (4) NS1 can increase virulence of an HPAIV further [41]. Most of these studies started from HPAIV and measured a decline in pathogenicity, which demonstrates that the identified alterations were required for full expression of virulence, or resulted from reassortment events which are prone to incompatibilities. No study so far addressed the question whether acquisition of a polybasic cleavage site is sufficient to transform a LPAIV H5 precursor into a highly pathogenic H5 virus. However, knowledge on this problem is most relevant, since it is required for development of a science-based assessment of the risk of HPAIV emergence from LPAIV field strains. Therefore, we introduced a polybasic HA cleavage site into the low-pathogenic H5N1 strain TG05, a potential HPAI precursor virus, in order to model the early evolution of HPAIV. This approach complements the current knowledge on the relative importance of virulence determinants and HPAIV evolution from the opposite point of view.
Taken together, acquisition of a polybasic HA cleavage site is only one necessary step for evolution of H5N1 HPAIV from lowpathogenic precursor strains, which may already have cryptic virulence potential. Moreover, besides the polybasic HA cleavage site, additional virulence determinants of HPAIV are located within the HA itself and in the other viral proteins. Knowledge of these virulence factors is crucial for an assessment of the risk of transformation of any LPAIV to HPAIV which is highly relevant for the control of notifiable LPAIV H5 or H7 infections in poultry.

Generation of recombinant viruses
Viral genes from TG05 were cloned into the plasmid pHWSccdB as described in [45]. Site-directed mutagenesis of the HA cleavage site region was performed by the Quikchange TM protocol (primer sequences available upon request). All recombinant viruses were rescued essentially as described [32] with the addition of plasmids expressing the polymerase proteins and the nucleoprotein genes from A/PR/8/34 (H1N1) (a kind gift from Peter Palese) except for R65-HA TG05poly . The recombinant R65 [45] differs from the native isolate [44]   Plaque assay and growth curves Plaque assays were performed on MDCK cells as described previously [54] with 400 ml inoculum either in the presence of 2 mg/ml N-tosyl-L-phenylalanine chloromethyl ketone (TPCK)treated trypsin (Sigma, Taufkirchen, Germany), or in the absence of trypsin. For determination of growth curves, DF-1 cells were inoculated at a multiplicity of infection of 10 23 in the presence of 1 mg/ml TPCK-treated trypsin or in the absence of any exogenous protease. From two independent experiments, supernatants were harvested at 0, 8, 24, 48, and 72 h post inoculation and resulting infectious virus was titrated by plaque assay on MDCK cells in the presence of trypsin.

Western blots
DF-1 cells were infected at a multiplicity of infection of 0.1 in the presence of either 1.0 mg/ml TPCK-treated trypsin or no exogenous protease in ISCOVES DMEM and 0.2% bovine serum albumin (BSA) (MP Medicals, Heidelberg) until a cytopathic effect appeared (48 h for TG05, TG05-HA R65 , and R65-HA TG05poly ; 7 d for TG05 poly ). Cells were lysed with 46 Laemmli buffer [55] containing 4% sodium dodecyl sulfate (SDS) and were inactivated by heating at 95uC for 5 min. Lysates were separated on a 10%-SDS polyacrylamid gel and electrotransferred to a nitrocellulose membrane. For detection of HA, a polyclonal rabbit antibody to the HA protein from A/Chicken/Vietnam/P41/2005 (H5N1), expressed by a vaccinia virus vector [56], (1:20,000; incubated 1 h at room temperature) and as secondary antibody a rabbit-specific goat immunoglobulin G fragment conjugated with horseradish peroxidase (1:10,000; 1 h at room temperature, Biovision, USA) were used. Antibody binding was visualized by chemiluminescence (Supersignal West Pico Chemiluminescence Kit from Pierce, Bonn).

Animal experiments
Ten 2-week-old White Leghorn specific-pathogen-free chickens per group were infected oculonasally with 10 5 PFU per animal. Each bird was observed daily for 10 days for clinical signs and classified as healthy (0), ill (1) (exhibiting one of the following: respiratory symptoms, depression, diarrhea, cyanosis, edema, or central nervous symptoms), severely ill (2) (severe or more than one of the previously mentioned symptoms), or dead (3) as described previously [57]. When birds are too sick to eat or drink, they are killed humanely and scored as dead to the next observation day [57].

Histopathology and immunohistochemistry
Samples from cerebellum, cerebrum, lung, nose, trachea, caecum, duodenum, kidney, and pancreas of additionally infected chicken were taken on day 4 and 10 post infection, formalin-fixed and processed for paraffin-wax-embedding according to standardized procedures. Immunohistochemical detection of influenza A virus nucleoprotein (NP) and hematoxylin eosin staining was performed as described [58].
For R65 [45], samples from cerebellum and lung were taken on day 2 from two 3-week old chickens infected oculonasally with 10 6 TCID 50 /animal.