A Single Amino Acid Substitution in the Hemagglutinin of H3N2 Subtype Influenza A Viruses Is Associated with Resistance to the Long Pentraxin PTX3 and Enhanced Virulence in Mice

This article is featured in In This Issue, p.1

Respiratory secretions contain a complex mixture of innate immune proteins, including members of the collectin and pentraxin families, which neutralize the infectivity of influenza A viruses (IAV) via mechanisms distinct from that mediated by Ab. These neutralizing inhibitors provide an important barrier against initial infection with IAV and limit virus spread during the first few days of infection. Historically, such proteins have been classified as β- or γ-type inhibitors based on their chemical properties and mechanism of action against IAV (reviewed in Refs. 1, 2).

β-type inhibitors are Ca²⁺-dependent lectins that bind to mannose-rich glycans on the hemagglutinin (HA) and the neuraminidase (NA) glycoproteins of IAV (3) to inhibit hemagglutination, neutralize virus infectivity, aggregate virions, and protect neutrophils from the depressive effects of IAV on respiratory burst responses (reviewed in Refs. 4, 5). It is well established that some members of the collectin superfamily, including human mannose-binding lectin and surfactant protein (SP)-D, act as β-type inhibitors against IAV. In contrast, γ-type inhibitors are soluble sialylated glycoproteins that act as receptor decoys, presenting sialic acids (SIA) that are recognized by the viral HA and resist hydrolysis by NA (6, 7). Of γ-type inhibitors, the antiviral properties of α2-macroglobulin, the major IAV inhibitor in horse serum, have been particularly well characterized (6, 7). More recently, other sialylated glycoproteins, such as the collectin SP-A (8, 9), H-ficolin (10), and the pentraxins, pentraxin 3 (PTX3) (11) and serum amyloid P (SAP) (12), were shown to act as γ-type inhibitors of IAV.

Members of the pentraxin superfamily play diverse roles in physiology, inflammation, and host defense. Pentraxins are a family of evolutionarily conserved proteins characterized by a pentameric structure and by the presence of the pentraxin domain (HxCxS/TWxS, where x represents any amino acid). The short pentraxins C-reactive protein (CRP) and SAP are produced by hepatocytes in response to a range of inflammatory stimuli, in particular to the cytokine IL-6 (reviewed in Ref. 13), and they represent the main acute-phase reactants in humans and mice, respectively. PTX3 protomers are composed of a unique 174-aa N-terminal domain coupled to a C-terminal pentraxin-like domain homologous to SAP and CRP, and protomer subunits assemble into higher-order oligomers stabilized by disulfide bonds (14). PTX3 secretion is induced in response to multiple inflammatory stimuli; however, unlike CRP and SAP, it is produced by a variety of innate immune cells (15, 16). PTX3 is also produced by airway epithelial cells of the lung (17) and can be detected in airway fluids from IAV-infected mice (11).

Studies (11, 12) from our laboratory demonstrated that PTX3 acts as a γ-type inhibitor of particular H3N2 subtype IAV in vitro and in vivo. A single site of N-glycosylation is present in the C-terminal domain (Asn²²⁰) of PTX3, and this is occupied by a complex-type glycan terminating in α(2,3)-linked SIA [SIAα(2,3)] (18). Accordingly, H3N2 strains that recognized SIAα (2,3) were sensitive to the anti-IAV activities of PTX3 (11). The short pentraxin SAP also mediates anti-IAV activity. Early studies reported SAP to be a β-type inhibitor of IAV (19, 20), whereas we recently demonstrated that sialylated glycans expressed by SAP were critical for its anti-IAV activity (12). However, the γ-type inhibitor activity of SAP was Ca²⁺ dependent, which may reflect conformational differences in its secondary structure in the presence or absence of Ca²⁺ (21) that alter the arrangement and/or accessibility of its sialylated glycans to the viral HA. Moreover, SAP expressed SIAα (2,6) and, therefore, inhibited a broader range of human IAV strains than did PTX3.

We previously reported that PTX3 mediated antiviral activity against HKx31 (H3N2), but not PR8 (H1N1), in vitro and that treatment of HKx31-infected, but not PR8-infected, mice with exogenous PTX3 reduced virus growth in the lungs, as well as disease severity (11). However, these strains represent different IAV subtypes and do not provide specific information regarding viral determinants of sensitivity to PTX3. In this study, we selected mutants of H3N2 strains for resistance to PTX3 and identified residues in the viral HA that determine PTX3 sensitivity. PTX3-resistant mutants were more virulent in mice, and this was associated with enhanced virus growth and airway inflammation. Finally, we demonstrate that the activity of either HA or NA can determine the sensitivity of a particular virus strain to inhibition by PTX3.

Human and mouse PTX3 expressed in CHO cells were purified from cell supernatants and biotinylated as described (22). Human SAP purified from human serum (Calbiochem) was biotinylated as described (12).

The IAV used in this study were A/PR/8/34 (PR8, H1N1) as well as HKx31 and BJx109, which are high-yielding reassortants of PR8 with A/Aichi/2/68 (H3N2) and A/Beijing/353/89 (H3N2), respectively, and bear the H3N2 surface glycoproteins (23). Other IAV strains used were A/Port Chalmers/1/73 (PC/73; H3N2), A/Udorn/307/72 (H3N2), A/New Caledonia/20/99 (N.Cal/99, H1N1), and A/Auckland/1/09 (Auck/09; A(H1N1)pdm09), as well as γ-irradiated H5N1 strains A/Vietnam/1194/2004, A/Vietnam/1203/2004, A/Myanmar/1/2006, A/Chicken/Indonesia/Wates/2005, A/Chicken/Indonesia/Magelang1631-5/2007, A/Chicken/Laos/Xagthani/2006, A/Duck/NgheAn/NCVD-70/2007, and A/Duck/Vietnam/942B/2004. Viruses were obtained from the World Health Organization Collaborating Centre for Reference and Research on Influenza and from the Commonwealth Scientific and Industrial Research Organization Australian Animal Health Laboratory (Geelong, Australia). A reassortant virus expressing six genes from PR8 in conjunction with the HA and NA of A/Shanghai/2/2013 (Shang/13, H7N9) was kindly provided by The National Institute for Biological Standards and Control (Potters Bay, Hertfordshire, U.K.). IAV were propagated in the allantoic fluid cavity of 10-d embryonated hens’ eggs by standard procedures (3) and stored at −70°C. Viruses were titrated on MDCK cells, as described (24), and purified from allantoic fluid by rate zonal sedimentation in 15–80% w/v sucrose gradients (3).

PTX3-resistant (PTX3^R) mutants of H3N2 subtype IAV were prepared by incubation of virus-containing allantoic fluid with 50 μg recombinant PTX3, in a total volume of 50 μl for 30 min at 37°C, before inoculation into embryonated eggs. The resultant virus was screened for sensitivity to PTX3 by hemagglutination inhibition, and the selection process was repeated two or three times until full resistance to PTX3 had been acquired. The resultant virus was then cloned at high dilution in eggs in the absence of PTX3.

Reverse genetic (RG) viruses used in this study were generated by the eight-plasmid RG technique as described (25). Site-directed mutagenesis was used to introduce nucleic acid mutations to facilitate amino acid substitution Ser¹⁴⁵→Asn in the HA of HKx31. Viruses generated were 6:2 reassortants consisting of six genes from PR8 in conjunction with NA from HKx31 and either the wild-type (WT) HA (RG-HKx31) or the mutant HA (RG-HKx31-PTX3^R).

Hemagglutination tests were performed in round-bottom 96-well microtiter plates at room temperature using 1% (v/v) erythrocytes in TBS (0.05 M Tris-HCl, 0.15 M NaCl [pH 7.2]). In some experiments, turkey erythrocytes were treated with increasing amounts of Vibrio cholerae sialidase (Sigma-Aldrich; 0–60 mU/ml) for 30 min at 37°C, washed, and tested in hemagglutination tests using 4 hemagglutinating units (HAU) of virus (26).

Hemagglutination inhibition (HI) tests were performed by standard procedures in TBS containing 10 mM CaCl₂, and results are expressed as the minimum inhibitory concentration (MIC; in μg/ml) of PTX3 or SAP required to inhibit 4 HAU of virus. To inhibit the enzymatic activity of the viral NA, IAV were incubated with a final concentration of 50 nM zanamavir (4-guanidino-2,3-dehydro-N-acetylneuraminic acid, purchased from GlaxoSmithKline, Melbourne, Australia) for 30 min at room temperature and then added to pentraxin titrations in an HI test.

To determine binding of PTX3 and SAP to WT or PTX3^R HKx31, wells of microtiter plates were coated overnight with 1 μg/ml purified virus, blocked for 2 h at room temperature with TBS containing 10 mg/ml BSA, and washed with TBST. Wells were then incubated with increasing concentrations of biotin-labeled PTX3 or SAP in TBST containing 5 mg/ml BSA and 5 mM CaCl₂. Binding of biotin-labeled pentraxins was detected by addition of streptavidin-conjugated HRP (GE Healthcare).

The ability of PTX3 and SAP to inhibit the activity of the viral NA was measured by ELISA. Biotin-labeled peanut agglutinin (Pierce Biotechnology) was used to detect galactose residues exposed after removal of SIA from fetuin (Sigma-Aldrich). Titrations and inhibition of NA activity by pentraxins were performed as described (11).

Neutralization of virus infectivity by pentraxins was determined by fluorescent-focus reduction in monolayers of MDCK cells, as described (27). Briefly, dilutions of PTX3 or SAP, prepared in TBS containing 10 mM CaCl₂, were incubated at 37°C for 30 min with a standard dilution of virus and then added to MDCK cell monolayer for 1 h at 37°C. Cells were washed, incubated for an additional 6–7 h, and fixed in 80% (v/v) acetone. IAV-infected cells were identified by immunofluorescence microscopy after staining with mAb MP3.1092.IC7, which is specific for the nucleoprotein (NP) of IAV, followed by FITC-conjugated rabbit anti-mouse Ig (Dako). The total number of fluorescent foci in four representative fields was counted and expressed as a percentage of the number of foci in the corresponding area of duplicate control wells infected with virus alone.

MDCK cells detached from plastic flasks using trypsin-versene were mock infected or infected in suspension with IAV at a multiplicity of infection (MOI) of 10 PFU/cell for 1 h at 37°C in serum-free media. Cells were washed, cultured for an additional 5 h in Teflon-coated pots, and then gently resuspended and used in assays to determine binding of biotin-labeled PTX3 or SAP. Aliquots of 10⁶ cells were incubated on ice with biotinylated PTX3 (2 μg/ml) or SAP (10 μg/ml) in TBS containing 5 mg/ml BSA and supplemented with either 10 mM CaCl₂ or 5 mM EDTA. Cells were washed, and bound pentraxins were detected using streptavidin conjugated to allophycocyanin. Cells were analyzed on a FACSCalibur flow cytometer (BD Biosciences) using 10 μg/ml propidium iodide (PI) to exclude dead cells from the analyses. We used a mAb specific for the IAV H3 HA (mAb 514; a kind gift from Prof. Lorena Brown, Department of Microbiology and Immunology, University of Melbourne) to confirm that a similar proportion of IAV-infected cells (80–90%) expressed the HA glycoprotein at the cell surface in each sample analyzed (data not shown).

RNA was extracted from the cell pellets obtained following centrifugation of virus-infected allantoic fluid using an RNeasy Mini kit (QIAGEN) and converted to cDNA with a QIAGEN Omniscript RT kit using 20 pmol/μl Uni12 primer (5′-AGC AAA AGC AGG-3′). DNA encoding the HA and NA genes was amplified from cDNA, PCR products were separated by agarose gel electrophoresis, and DNA bands of specific size were cut out and extracted (Gel Extraction Kit; MO BIO). Sequencing was performed by Applied Genetics Diagnostics (Department of Pathology, University of Melbourne). H3 numbering (25) was used to align the deduced amino acid sequences.

Subconfluent monolayers of mouse LA-4 airway epithelial cells or MDCK cells in eight-well chamber slides (Lab-Tek; Nunc) were incubated with increasing dilutions of IAV for 1 h at 37°C, washed three times, and incubated in serum-free media, as described (28). At 7–8 h postinfection, cells were fixed in 80% (v/v) acetone and stained with anti-NP mAb MP3.1092.IC7, followed by FITC-conjugated rabbit anti-mouse Ig (Dako). Cells were viewed by fluorescence microscopy, using PI (10 μg/ml) staining to determine the total cell number/field in at least four random fields (minimum total cell count = 100/sample) and anti-NP staining to determine the number of IAV-infected cells/field.

The HA/NA balance of different IAV was determined as described (29). Briefly, duplicate tubes of formaldehyde-fixed LA-4 cells (10⁶ cells) were mixed with 128–256 HAU each IAV for 1 h on ice to allow HA-mediated binding of virus to sialylated receptors on the LA-4 cell surface. One sample was held at 4°C (to inhibit the enzymatic activity of the viral NA), whereas the duplicate tube was transferred to 37°C for 30 min (to allow the enzymatic activity of the viral NA). A third tube received no cells and was held on ice throughout. After incubation, cells were pelleted, and virus remaining in the cell supernatant was quantified by hemagglutination assay.

Multistep growth curves of IAV were generated using MDCK cells cultured in 24-well tissue-culture plates (Thermo Scientific Nunc). Cells were incubated with a low MOI (0.01 PFU/cell) of each IAV for 1 h at 37°C, washed, and cultured in serum-free media containing 2 μg/ml TPCK-treated trypsin (Worthington Biochemical) to allow for multiple cycles of virus replication. At various times postinfection, supernatants were removed, clarified, and assayed for titers of infectious virus by plaque assay on MDCK cell monolayers in the presence of trypsin.

C57BL/6 mice were bred and housed in specific pathogen–free conditions in the Animal Facility at the Department of Microbiology and Immunology, University of Melbourne. Six- to eight-week-old male mice were used in all experiments. Mice were anesthetized and infected with 10³, 10⁴, or 10⁵ PFU IAV in 50 μl PBS via the intranasal route. Control mice were mock infected with 50 μl PBS alone. Mice were weighed daily and assessed for signs of clinical disease. Animals that had lost ≥20% of their original body weight and/or displayed evidence of pneumonia were culled. All research complied with the University of Melbourne’s Animal Experimentation Ethics guidelines and policies. At various times postinfection, mice were culled, and samples including blood, bronchoalveolar lavage fluid (BALF), thymus, and lungs were collected as described (28, 30). Plaque assay was used to determine titers of infectious virus in clarified homogenates prepared from the lungs of IAV-infected mice. Total protein levels in cell-free BALF were determined by Bradford protein dye using a standard curve of BSA. The BD Cytokine Bead Array Mouse Inflammation Kit (Becton Dickinson) was used to determine levels of IFN-γ, TNF-α, IL-6, IL-10, IL-12.p70, and MCP-1 in cell-free BALF, and PTX3 levels in BALF were determined by ELISA, as described (11).

Single-cell suspensions prepared from the BALF and thymus of IAV-infected mice were treated with Tris-NH₄Cl (0.14 M NH₄Cl in 17 mM Tris, adjusted to pH 7.2) to lyse erythrocytes and incubated on ice for 30 min with supernatants from hybridoma 2.4G2 to block FcRs. Cell numbers and viability were assessed via trypan blue exclusion using a hemocytometer. Cells were then stained with appropriate combinations of FITC-, PE-, and allophycocyanin-conjugated mAbs to Gr-1 (RB6-8C5), CD45.2 (104), CD8α (53-6.7), CD4 (GK1.5), B220 (RA3-6B2), TCRβ (H57-597), and NK1.1 (PK136) and analyzed by flow cytometry. PI (10 μg/ml) was added to each sample to exclude dead cells from the analyses, and a minimum of 50,000 PI⁻ cells was collected.

Lungs were perfused, inflated, and fixed in 4% formaldehyde, as described (30), and 4-μm sections were prepared and stained with H&E. Sections were assessed for inflammation using a subjective scale from 0 to 5 (0, no inflammation; 1, very mild; 2, mild; 3, moderate; 4, marked; and 5, severe inflammation). Sections were randomized and blinded, and those corresponding to the least and most severe inflammation were assigned scores of 0 and 5, respectively. Peribronchiolar inflammation and alveolitis were then graded in multiple random fields for each section by three independent readers, as described (28, 30).

C57BL/6 mice were lightly anesthetized and infected via the intranasal route with 10⁵ PFU WT or PTX3^R PC/73 in 50 μl PBS. Recombinant human PTX3 (2 mg/kg i.p.) was administered on the day of infection and at days 1 and 3 postinfection. To determine virus titers in the airways, mice were euthanized at day 5 postinfection, and lungs were removed, homogenized, and clarified by centrifugation; supernatants were frozen at −70°C. The samples were assayed for infectious virus by plaque assay on MDCK cells in the presence of trypsin.

For the comparison of two sets of values, a Student t test (two-tailed, two-sample equal variance) was used. One-way ANOVA (parametric), followed by the Tukey post test, was used when comparing three or more sets of values. Two-way ANOVA, followed by the Bonferroni correction, was used when comparing weight change over time in three groups. Survival proportions were analyzed by a two-tailed, log-rank (Mantel–Cox) test. For analysis of histopathological data, a Kruskal–Wallis test (nonparametric) was used, followed by the Dunn post test.

PTX3 and SAP express SIAα(2,3) and SIAα(2,6), respectively, and act as receptor decoys that are recognized by the viral HA, resisting hydrolysis by the viral NA to mediate anti-IAV activity (11, 12). In general terms, mouse-adapted and avian IAV exhibit HA preference for SIAα(2,3), human strains prefer SIAα(2,6), and recent pandemic IAV show dual specificity for SIAα(2,3)/α(2,6). Therefore, we compared a range of IAV strains for sensitivity to PTX3 and SAP in the presence or absence of the NA inhibitor (NAI) zanamavir.

Consistent with previous results (11), an early H3N2 strain [HKx31, expressing the HA/NA from A/Aichi/2/1968 with dual HA receptor preference (31)] was inhibited by PTX3 and SAP in the presence or absence of zanamavir (Table I), whereas a later strain [BJx109, expressing the HA/NA from A/Beijing/353/1989 with preference for SIAα(2,6) (11)] was inhibited only by SAP, and NAI did not enhance activity (Table I). All H1N1 IAV, including A(H1N1)pdm09, were resistant to PTX3 (MIC ≥ 10 μg/ml); however, NAI increased the sensitivity of PR8 and Auck/09 to PTX3, consistent with HA specificity for SIAα(2,3) (32) and SIAα(2,3)/α(2,6) (33), respectively. Inhibition of the N1 NA also increased the sensitivity of the human H1N1 strains N.Cal/99 and Auck/09 (but not mouse-adapted PR8) to SAP, consistent with the preference of human IAV for SIAα(2,6) expressed by SAP. Of note, SIAα(2,3) is the only SIA linkage expressed by natural PTX3 (from TNF-α stimulated human fibrosarcoma cells) or by recombinant PTX3 expressed by CHO cells (12, 18), and our previous studies confirmed that PTX3 from either source displayed a similar spectrum of antiviral activity against H3N2 IAV (12).

Avian H5N1 show HA receptor preference for SIAα(2,3) (34, 35) and are considered a potential pandemic threat. Therefore, we compared the sensitivity of human and avian H5N1 to inhibition by PTX3 and SAP. All avian and human H5N1 strains tested were largely resistant to PTX3 (MIC ≥ 5 μg/ml) and SAP (MIC > 10 μg/ml), and HI activity was not enhanced in the presence of zanamivir (Table I), indicating that the H5 HA cannot bind effectively to SIA expressed by either PTX3 or SAP. The new human H7N9 virus associated with human infections in China during 2013 is a reassortant virus expressing H7 and N9 and other gene segments from avian IAV (36), and it displays dual HA specificity for SIAα(2,3)/α(2,6) (37). Consistent with this, IAV expressing H7N9 from Shang/13 was sensitive to PTX3 and SAP, and activity was enhanced in the presence of NAI (Table I). Together, these data indicate that, in addition to the viral HA, the activity and/or specificity of the viral NA can be an important viral determinant of sensitivity to either PTX3 or SAP.

Because both the viral HA and NA can contribute to PTX3 resistance, we investigated molecular changes associated with the acquisition of PTX3 resistance by H3N2 IAV. HKx31 was incubated with purified human PTX3 and inoculated in embryonated eggs (as described in 2Materials and Methods) to facilitate selection of a PTX3^R mutant. After cloning at high dilution in eggs, WT and PTX3^R HKx31 were compared for sensitivity to PTX3 and SAP using a range of assays, including binding to virus (as determined by ELISA; Fig. 1A), inhibition of IAV hemagglutination activity (Fig. 1B), inhibition of the enzymatic activity of the viral NA (Fig. 1C), and neutralization of viral infectivity (Fig. 1D). In each assay, WT HKx31 was sensitive to both PTX3 and SAP, whereas the PTX3^R mutant was sensitive only to SAP.

Both PTX3 and SAP bind FcRs and can activate leukocyte-mediated phagocytosis (38, 39). Moreover, opsonization of target cells and/or microbes by pentraxins can be enhanced in the presence of complement (reviewed in Ref. 40). Because the viral HA and NA are expressed at the surface of infected cells prior to virus budding and release, IAV-infected cells represent another important target for recognition by pattern recognition receptors of the innate immune system. PTX3 bound to MDCK cells infected with WT, but not PTX3^R, HKx31 in the presence or absence of Ca²⁺, whereas SAP bound to cells infected with either WT or PTX3^R HKx31 in a Ca²⁺-dependent manner (Fig. 1E). An anti-HA mAb confirmed that MDCK cells infected with WT or PTX3^R HKx31 expressed a similar proportion of cell surface HA in all samples analyzed (data not shown).

Proteins expressed on the surface of the IAV virion represent potential targets for recognition by PTX3, and both HA and NA can contribute to PTX3 resistance. Therefore, we sequenced genes encoding the HA and NA of WT and PTX3^R HKx31. The deduced amino acid sequences of NA were identical; however, a single-base substitution (G⁴³⁵→A) gave rise to the amino acid substitution Ser¹⁴⁵→Asn in the HA₁ domain, in the vicinity of the receptor-binding pocket, of the PTX3^R mutant. An identical mutation was detected in HKx31 mutants selected in the presence of purified murine PTX3. These substitutions are distinct from those associated with resistance of H3 IAV to other γ-type inhibitors, such as SAP (Leu²²⁶→ Glu) (12) or equine α2macroglobulin (Leu²²⁶→ Glu) (41), or to β-type inhibitors, such as conglutinin or SP-D (Thr¹⁶⁷→ Asn) (3). Note that full-genome sequencing confirmed no differences in the viral proteins encoded by the six remaining gene segments of WT and PTX3^R HKx31.

H3N2 IAV that circulated in humans prior to 1973 were sensitive to PTX3, but strains after this time had acquired resistance (11). Therefore, we aligned the HA sequences of 13 H3N2 strains (1968–2005) that had been tested for sensitivity to PTX3 (Fig. 2), with a particular focus on amino acid 145 and surrounding residues. Prior to 1975, most H3N2 strains expressed Ser¹⁴⁵, whereas strains after 1973 expressed Asn/Lys¹⁴⁵. Analysis of HA₁ sequences from 108 strains (1968–1975) in public databases indicated that 82.4% expressed Ser¹⁴⁵, but this residue was very rare (2.2%) in the 4208 HA₁ sequences from later (1975–2005) H3N2 strains. Some early strains (8/108) expressed Ile¹⁴⁵, including our laboratory stocks of PC/73. PTX3^R PC/73 mutants selected under experimental conditions in the presence of human PTX3 were characterized by a single amino acid substitution (Ile¹⁴⁵→Arg) in HA (Fig. 2). Thus, using laboratory-derived H3N2 PTX3^R mutants, as well as analysis of HA₁ sequences in public databases, we demonstrate that residue 145 in H3 HA₁ is associated with determining sensitivity or resistance to PTX3.

Changes in the HA sequence, particularly those in or around the receptor-binding pocket, can affect the receptor properties of the viral HA and, therefore, modulate interactions between IAV and target cells. We compared the HA receptor properties of WT and PTX3^R mutants and found that PTX3^R mutants did not differ in their ability to agglutinate chicken erythrocytes enzymatically altered to express either α(2,6)- or α(2,3)-linked Neu5Ac (Supplemental Table IA) or treated with increasing amounts of V. cholerae sialidase to remove terminal SIA (Supplemental Table IB). Erythrocytes from different species show distinct patterns regarding the type of SIA expressed (Neu5Ac and/or Neu5Gc), SIA density, and the proportion of SIAα(2,3) and SIAα(2,6) (42). However, no differences were noted in the ability of WT and PTX3^R viruses to agglutinate erythrocytes from chickens [α(2,3) = α(2,6), Neu5Ac/Neu5Gc], horses [α(2,3), Neu5Gc], turkeys [α(2,6) > α(2,3), Neu5Ac/Neu5Gc], humans [(α(2,3) = α(2,6), Neu5Ac], or rabbits [α(2,6), Neu5AGc > Neu5Ac] (Supplemental Table IC). Thus, we could not detect any differences between WT or PTX3^R viruses in the specificity of the viral HA for SIA.

Next, we compared the ability of WT and PTX3^R viruses to bind, infect, and replicate productively in mammalian cells in vitro. The opposing activities of the viral HA (binding to cell surface SIA) and NA (cleaving cell surface SIA) must be carefully balanced to allow for efficient IAV entry and exit from host cells; mutations that affect the specificity and/or avidity of either HA or NA can disrupt this balance (reviewed in Ref. 43). HA/NA balance was assessed by incubating IAV with fixed mouse airway epithelial (LA-4) cells at 4°C (when HA binds but the viral NA is inhibited) or at 37°C (when HA binds to SIA and the viral NA is active) before determining levels of virus in cell-free supernatants (Supplemental Fig. 1A). Compared with controls (i.e., virus not incubated with any cells), incubation of virus with cells at 4°C reduced levels of virus in the supernatant; however, essentially all virus was recovered at 37°C, and no differences were noted between WT and PTX3^R viruses. Moreover, WT and PTX3^R mutants showed no difference in their ability to infect the LA-4 mouse airway epithelial cell line, as assessed by immunofluorescence at 8 h postinfection (Supplemental Fig. 1B). Finally, MDCK cells were infected at a low MOI (0.01 PFU/cell) in the presence of trypsin to allow for multiple cycles of virus replication in cell culture. At various times, cell-free supernatants were removed, and titers of infectious virus released from IAV-infected cells were determined. In these experiments, no significant differences were detected between WT and PTX3^R viruses at any time point tested (Supplemental Fig. 1C). Together, these data indicate that resistance to PTX3 did not alter the intrinsic ability of IAV to infect and replicate in mammalian cells.

We next compared the virulence of WT and PTX3^R mutants of HKx31 and PC/73 in mice. Initial studies examined weight loss (Fig. 3A, 3C) and lethality (Fig. 3B, 3D) of mice following intranasal inoculation with increasing doses of WT or PTX3^R viruses. Overall, PTX3^R mutants displayed enhanced virulence in mice compared with their respective WT viruses. Mice infected with 10⁴ PFU of PTX3^R HKx31 lost more weight (day 7, p < 0.01; day 8, p < 0.001, two-way ANOVA) and showed reduced survival (p < 0.01, two-tailed, log-rank Mantel–Cox test) compared with animals infected with an equivalent dose of WT HKx31. PTX3^R PC/73 was also more virulent than was WT PC/73, and PTX3^R-infected mice showed enhanced weight loss postinfection with 10³ PFU (day 7–8, p < 0.001; day 9–10, p < 0.05, two-way ANOVA), 10⁴ PFU (day 5–7, p < 0.001; day 8, p < 0.01, two-way ANOVA), and 10⁵ PFU (day 6, p < 0.05; day 7–9, p < 0.001; day 10, p < 0.01, two-way ANOVA). Moreover, mice infected with 10⁵ PFU of PTX3^R PC/73 showed reduced survival (p < 0.001, two-tailed, log-rank [Mantel–Cox] test) compared with mice infected with WT PC/73.

To examine viral replication, mice were infected with 10³ PFU of WT or PTX3^R HKx31 or with 10⁴ PFU of WT or PTX3^R PC/73, and viral titers were determined at day 3 (Fig. 4A) and day 7 (Fig. 4B) postinfection. At day 3, there were no differences between either pair of WT and PTX3^R viruses with regard to virus replication in the lungs. At day 7, both PTX3^R mutants grew to higher titers in the lungs compared with their respective WT control viruses (Fig 4B, p < 0.001, Student t test).

We characterized the cellular infiltrate and the soluble mediators in BALF recovered from mice 7 d postinfection with 10⁴ PFU of either WT or PTX3^R PC/73. Numbers of total BALF cells recovered from WT or PTX3^R PC/73–infected mice were similar, with no significant differences in the numbers of neutrophils or CD8⁺, CD4⁺, or B220⁺ lymphocytes, although a modest increase in NK cells was observed (Supplemental Fig. 2A).

Because dysregulated cytokine responses are associated with IAV-induced disease, we next determined the levels of proinflammatory chemokines and cytokines in the airways of IAV-infected mice. Levels of IFN-γ, TNF-α, IL-10, and IL-12.p70 were similar in cell-free BALF from mice infected with PTX3-sensitive (PTX3^S) or PTX3^R viruses (data not shown); however, MCP-1 and IL-6 were significantly higher in BALF from PTX3^R-infected mice (Fig. 5A). It is well established that elevated MCP-1 and IL-6 are associated with lung injury and disease in IAV-infected mice (28, 30). Consistent with this, the increased wet/dry ratio of the lungs (Fig. 5B) and enhanced protein levels in cell-free BALF (Fig. 5C) from mice infected with PTX3^R PC/73 were indicative of increased vascular leakage and lung injury. Lung sections from WT- and PTX3^R-infected mice were also examined for histopathological changes at day 7 postinfection (Fig. 5D). H&E-stained lung sections were blinded, randomized, and scored by three independent readers for peribronchiolar inflammation and alveolitis. Analysis of histological sections indicated that peribronchiolar inflammation and alveolitis were more severe in mice infected with PTX3^R PC/73 (p < 0.05, Kruskal–Wallis test, Fig. 5E).

In humans and mice, thymic atrophy and leukopenia are systemic markers that have been associated with severe influenza infections (28, 30, 44). Infection of mice with PTX3^R PC/73 was associated with reduced cellularity of the thymus, characterized by a marked reduction in the numbers of double-positive, but not single-positive or double-negative, thymocytes (Supplemental Fig. 2B). Total blood leukocyte numbers were also reduced in mice infected with PTX3^R PC/73, and this was associated with decreased numbers of neutrophils, B lymphocytes, and CD4⁺ and CD8⁺ T cells (Supplemental Fig. 2C). Together, these data demonstrate that infection of mice with PTX3^R PC/73 results in more severe airway and systemic disease.

Next, we examined the levels of PTX3 in BALF from mice infected with WT or PTX3^R PC/73 at days 3 and 7 postinfection (Fig. 6A). At either time, PTX3 levels from WT- or PTX3^R-infected mice were significantly higher than levels in BALF from naive animals. PTX3 levels in BALF from mice infected with WT or PTX3^R PC/73 were not different at day 3 postinfection, but PTX3^R-infected mice had significantly higher PTX3 levels at day 7 postinfection.

Cell-free BALF collected from mice 7 d after IAV infection was incubated on ice with 10% v/v chicken erythrocytes to remove residual virus and then assessed for its ability to neutralize IAV. BALF from naive or WT PC/73–infected mice did not neutralize WT or PTX3^R PC/73 to significant levels (Fig. 6B). In contrast, BALF from PTX3^R-infected mice neutralized WT PC/73 more potently than did naive BALF (p < 0.01, one-way ANOVA). Although there was a tendency for BALF from PTX3^R-infected mice to neutralize PTX3^R PC/73, this was not significant compared with naive BALF and may reflect sensitivity to other innate proteins, such as SP-D, which are known to increase in BALF following IAV infection (Fig. 6B) (45, 46).

Because airway epithelial cells secrete PTX3 following exposure to a range of stimuli, including LPS and TNF-α (15, 17), we tested whether IAV might also induce PTX3 from human (A549) airway epithelial cells. Although TNF-α induced PTX3 secretion, PTX3^S (HKx31) or PTX3^R (HKx31-PTX3^R) viruses did not (Supplemental Fig. 3), nor did virus inhibit TNF-α–induced PTX3 from epithelial cells. Together, these data indicate that IAV infection of epithelial cells does not induce or inhibit PTX3 production. Instead, the enhanced PTX3 levels in BALF from IAV-infected mice are likely a consequence of the surrounding cytokine milieu.

Therapeutic treatment with recombinant human PTX3 was shown to ameliorate disease severity in a number of murine models, including murine CMV (47), mouse hepatitis virus (48), and Pseudomonas aeruginosa (49). We demonstrated previously that PTX3 treatment reduced virus growth in mice infected with HKx31 (H3N2, PTX3^S) but not PR8 (H1N1, PTX3^R) (11). Given the marked differences between these viruses, we refined our studies to determine the effect of therapeutic PTX3 treatment of IAV replication in the airways of mice infected with WT or PTX3^R PC/73, which only differ by a single amino acid in HA₁. In these studies, mice were challenged with 10⁵ PFU and analyzed at day 5 postinfection. Compared with mock-treated animals infected with IAV, PTX3 treatment significantly reduced viral titers in lungs from mice infected with WT (p < 0.01, one-way ANOVA) but not PTX3^R PC/73 (Fig. 7A).

To confirm that a single amino acid substitution in the viral HA (Ser¹⁴⁵→Asn) was the critical determinant of sensitivity to PTX3, we used RG to generate 6:2 reassortant viruses expressing the HA and NA of HKx31 and all internal components derived from PR8. RG viruses expressing WT HKx31 HA (RG-HKx31) or HKx31 HA containing the S145N substitution (RG-HKx31-PTX3^R) were rescued, amplified in eggs, and characterized in vitro and in vivo. Compared with RG-HKx31, RG-HKx31 PTX3^R was resistant to both hemagglutination inhibition (Fig. 8A) and neutralization (Fig. 8B) by human PTX3 in vitro. Furthermore, RG-HKx31 PTX3^R grew to higher titers than did RG-HKx31 in mouse lung 7 d postinfection with 10³ PFU of each virus (Fig 8C). Full genome sequencing confirmed that site 145, and not other unidentified mutations, accounts for increased virulence in the mouse model and resistance to PTX3.

The current study focused on understanding the viral determinants that contribute to the sensitivity of different IAV to PTX3. First, we demonstrated that the viral HA or NA can be the critical determinant of PTX3 sensitivity. HA-mediated recognition of SIAα(2,3) is an essential feature, but it cannot be used as a predictor of PTX3 sensitivity. For H1N1 IAV, the susceptibility of SIA expressed by PTX3 to the viral NA was also critical in determining sensitivity. Moreover, H5N1 viruses display clear HA preference for SIAα(2,3) (34, 35) yet were resistant to PTX3, even in the presence of NAI. Some avian viruses discriminate between the core structures of oligosaccharides (50), and HA-glycan conformation analysis suggests that the size and shape of SIA-bearing glycans, rather than specific linkage, are important for H5-mediated recognition (51). Therefore, it may be that the type of SIA presented by PTX3 is not recognized by the H5 HA, suggesting further complexity in the interplay between the fine specificity of the viral HA and the particular SIAα(2,3)-rich glycans expressed by PTX3.

H3N2 viruses associated with the 1968 Hong Kong pandemic acquired their HA gene from an avian source (52) and exhibited dual specificity for SIAα(2,3)/(2,6), (31), which has been reported for H1N1 (53), H2N2 (54), and A(H1N1)pdm09 (33) viruses associated with pandemics in 1918, 1957, and 2009, respectively. In fact, some early pandemic viruses were reported to retain a preference for SIAα(2,3) over SIAα(2,6) (54, 55). Early H3N2 IAV were sensitive to PTX3, and A(H1N1)pdm09 viruses were sensitive in the presence of NAI (Table I). Recent H7N9 viruses associated with human disease also displayed HA specificity for both SIAα(2,3) and SIAα(2,6) (37) and were inhibited by PTX3 (Table I). Together, these data suggest that PTX3 may represent a barrier limiting the introduction of new subtypes into humans and/or limiting initial spread and severity of pandemic IAV.

Early H3N2 viruses were sensitive to PTX3 but rapidly developed resistance as they circulated in humans. To understand the molecular basis underlying resistance to PTX3, H3N2 mutant viruses were selected under experimental conditions in the presence of purified PTX3. Amino acid substitutions at residue 145 of HA₁, which forms part of a loop lying adjacent to the receptor-binding site (56), were associated with resistance to PTX3, and similar substitutions were implicated in the natural evolution of human H3N2 strains from PTX3^S to PTX3^R (Fig. 2). It was suggested that changes at residue 145 can alter the orientation of SIA within the receptor-binding site (57); however, we did not detect any differences in receptor specificity between WT and PTX3^R viruses or in their ability to bind, infect, and replicate in MDCK or LA-4 cells in vitro. Together, these data indicate that resistance to PTX3 did not come at a major cost to virus fitness, nor was it associated with the acquisition of compensatory mutations in either HA or NA.

The receptor preference of H3N2 viruses evolved from recognition of SIAα(2,3) and SIAα(2,6) by early strains to recognition of only SIAα(2,6) by strains after 1975 (31). Selective pressures on HA receptor specificity are likely to include the nature of SIA receptors expressed on target cells, as well as soluble extracellular proteins that can neutralize virus infectivity. This concept is not a new one, and the contributions of these two pressures to the selection of virus variants with optimal receptor-binding phenotypes was demonstrated in vitro using inhibitors from a range of animal sera (57–59). In humans, it is conceivable that widespread expression of SIAα(2,6) receptors in the airways (60), in concert with selective immune pressure from the SIAα(2,3)-rich neutralizing inhibitor PTX3, was a factor underlying the evolution of H3N2 receptor specificity in humans. However, because substitutions at residue 145 alone did not result in detectable differences in the HA specificity of PTX3^R mutants, it is likely that additional changes in the HA of circulating H3N2 strains also contributed to their evolution toward a preference for SIAα(2,6).

SP-D, a major constituent of human airway fluids (45, 61), is a potent inhibitor of highly glycosylated seasonal IAV; however, pandemic viruses are poorly glycosylated and, therefore, are largely resistant to this lectin (27, 45, 62). PTX3 acts as a SIAα(2,3)-rich “receptor decoy” that is induced following IAV infection, consistent with a role in preventing infection and limiting spread of some pandemic IAV (or avian-origin IAV) in humans. Such a role could be particularly important because lectin-mediated innate defenses against these viruses are limited. In contrast, SAP expresses SIAα(2,6) (12, 63) and is likely to be effective against seasonal IAV that have evolved receptor preference for this linkage and cannot evade SIAα(2,6) recognition. Other sialylated inhibitors, such as gp340 (64), H-ficolin (10), and SP-A (65), express both SIAα(2,3) and SIAα(2,6) and, therefore, may contribute to immunity against potential pandemic and/or seasonal strains. Moreover, NAI potentiate the anti-IAV activity of PTX3 and SAP (Table I), as well as other γ-inhibitors, including mucins, SP-A, and H-ficolin in vitro (10, 66). Thus, NAI treatment blocks the ability of the viral NA to inactivate sialylated inhibitors in mucus, saliva, and other airway secretions, thereby enhancing their anti-IAV activity and contributing to virus clearance and recovery.

PTX3 plays a complex multifactorial role in infection and inflammation. Elevated levels of PTX3 in serum and/or BALF were associated with poor outcomes in patients with acute lung injury or acute respiratory distress syndrome (reviewed in Ref. 67). During IAV infection, high levels of endogenous PTX3 in airway fluids from PTX3^R PC/73–infected mice were associated with pulmonary inflammation and lung injury (Fig. 6A). However, early treatment with recombinant PTX3 ameliorated disease severity following infection with PTX3^S virus (Fig. 7), suggesting that the timing and regulation of PTX3 production are important in determining its role during IAV infection. During murine CMV infection, therapeutic PTX3 treatment induced dendritic cell maturation and enhanced NK cell and T cell activation (47), consistent with an important protective role mediated via its immunostimulatory properties. Therapeutic PTX3 treatment reduced virus replication in the lungs of mice infected with PTX3^S virus, but not PTX3^R virus, suggesting that the direct antiviral effects of PTX3 were most important in limiting IAV in this model. Although our results demonstrate a clear increase in the virulence of PTX3^R IAV in mice, the mechanisms underlying this phenomenon have yet to be fully elucidated. For example, subtle changes in HA preference could enhance the avidity of the PTX3^R HA for sialylated receptors in the murine airways, resulting in enhanced virus infection and virulence. In future studies, the use of PTX3-null mice will allow for a clearer understanding of the role that endogenous PTX3 plays in controlling IAV infection in vivo.

We thank St Jude Children’s Hospital (Memphis, TN) for providing the pHW2000 plasmid for RG.

The authors have no financial conflicts of interest.