Proteins of the innate immune system can act as natural inhibitors of influenza virus, limiting growth and spread of the virus in the early stages of infection before the induction of adaptive immune responses. In this study, we identify the long pentraxin PTX3 as a potent innate inhibitor of influenza viruses both in vitro and in vivo. Human and murine PTX3 bound to influenza virus and mediated a range of antiviral activities, including inhibition of hemagglutination, neutralization of virus infectivity and inhibition of viral neuraminidase. Antiviral activity was associated with binding of the viral hemagglutinin glycoprotein to sialylated ligands present on PTX3. Using a mouse model we found PTX3 to be rapidly induced following influenza infection and that PTX3−/− mice were more susceptible than wild-type mice to infection by PTX3-sensitive virus strains. Therapeutic treatment of mice with human PTX3 promoted survival and reduced viral load in the lungs following infection with PTX3-sensitive, but not PTX3-resistant, influenza viruses. Together, these studies describe a novel antiviral role for PTX3 in early host defense against influenza infections both in vitro and in vivo and describe the therapeutic potential of PTX3 in ameliorating disease during influenza infection.
Influenza is an important annual respiratory infection that remains a leading cause of illness and death throughout the world. In the early stages of infection, innate immune mechanisms represent the main line of host defense, acting to limit virus spread in host tissues before the induction of the adaptive response. Innate mechanisms that may contribute to defense against influenza infection include the intrinsic resistance of macrophages, the induction of IFN and proinflammatory cytokines and activation of NK cells, macrophages, and neutrophils in the airways. In addition, serum and airway fluids contain a number of innate proteins capable of recognizing and inhibiting influenza viruses, including members of the collectin and pentraxin families, mucins and salivary scavenger receptor-rich glycoprotein 340 (1).
The antiviral activities of collectins have been well described with strong evidence to suggest that surfactant protein-D (SP-D)4 is of particular importance during influenza infections. Both surfactant protein-A (SP-A) and SP-D mediate a range of activities against influenza virus, including inhibition of hemagglutination, virus neutralization, virus aggregation, and opsonizaton of the virus for interaction with neutrophils (2, 3, 4, 5, 6, 7, 8), with SP-D acting as a more potent inhibitor than SP-A in vitro (4). Levels of SP-D recovered from the airways by lavage increase significantly following infection of mice with influenza virus (3, 9, 10, 11), and SP-D−/− mice exhibit enhanced viral replication and illness (9, 11). SP-A−/− mice also display enhanced susceptibility to influenza infections in some studies (9, 10), but not others (12); however, the effect is less pronounced than that observed in SP-D−/− mice.
Of interest, the mechanisms by which SP-A and SP-D act against influenza virus are different. SP-D functions as a classic β inhibitor, binding in a Ca2+-dependent manner through its lectin domains to oligosaccharides on the viral hemagglutinin (HA) and neuraminidase (NA) glycoproteins (6). As such, the degree or pattern of glycosylation is a major factor in determining the sensitivity of a particular virus strain to inhibition by SP-D (3). In contrast, SP-A inhibits influenza viruses via Ca2+-independent binding of the viral HA to terminal sialic acid expressed on the SP-A molecule, thereby blocking the receptor binding site on the viral HA such that it can no longer access cellular receptors. SP-A is therefore classified as a γ inhibitor, acting in a similar manner to the serum inhibitor α2-macroglobulin (13, 14).
The long chain pentraxin (PTX3) is a 45-kDa protein that assembles to form high m.w. multimers linked by interchain disulfide bonds (15). The C-terminal domain (203 aa) of PTX3 shares homology with the classic short pentraxins C-reactive protein and SAP, whereas the N-terminal domain (178 aa) does not show any significant homology with other known proteins. PTX3 plays a complex nonredundant role in vivo, recognizing a diverse range of pathogens, modulating complement activity by binding C1q and facilitating pathogen recognition by macrophages and dendritic cells (reviewed in Ref. 16). Despite a well documented role in innate host defense against certain bacteria and fungi (17, 18, 19, 20), few studies have addressed the antiviral activities of PTX3. To this end, a recent study described the ability of human PTX3 to bind and inhibit the infectivity of human and murine cytomegaloviruses (21). Furthermore, the short pentraxin SAP has been shown to act as a β inhibitor against influenza viruses, binding in a Ca2+-dependent manner to mannose-rich glycans on the viral HA to inhibit both hemagglutination and viral infectivity (22, 23). Therefore, in this study we have assessed the antiviral activity of PTX3 against a range of influenza viruses, both in vitro and in vivo.
Materials and Methods
HKx31 (H3N2) influenza virus is a laboratory-derived high-yielding reassortant of A/PR/8/34 (PR8, H1N1) with A/Aichi/2/68 (Aichi/68; H3N2). Other viruses used included H3N2 subtype viruses, i.e., A/Memphis/1/71 (Mem/71), A/Udorn/307/72 (Udorn/72), A/Port Chalmers/1/73 (PtChal/73), A/Victoria/75 (Vic/75), A/Bangkok/1/79 (Bang/79), A/Beijing/353/89 (Beij/89), A/Beijing/32/92 (Beij/92), A/Guandong/25/93 (Guan/93); and additional H3N2 reassortants A/Texas/1/77 x PR8 (Tex/77-X), A/Philippines/2/82 x PR8 (Phil/82-X); and H1N1 subtype viruses, i.e., PR8, A/NWS/33 (NWS), A/Brazil/11/78 (Brazil/78), A/Victoria/57/83 (Vic/83) and A/Victoria/36/88 (Vic/88).
Additional virus strains tested for sensitivity to PTX3 and SP-D included H3N2 subtype viruses (1997–2005), i.e., A/Sydney/5/97, A/Moscow/10/99, A/Panama/2007/99, A/Philippines/472/2002, A/Kumamoto/102/2002, A/Fujian/411/2002, A/Singapore/37/2004, A/Victoria/523/2004, A/Wellington/1/2004, A/New York/55/2004, A/California/7/2004, A/Brisbane/3/2005, A/Victoria/512/2005, A/Wisconsin/67/2005; additional H1N1 viruses, i.e., A/Beijing/262/95, A/New Caledonia/20/1999, A/Fujian/156/2000, A/England/51/2002, A/Singapore/14/2004, A/Malaysia/1513/2004, A/New Caledonia/9/2004, A/Brisbane/193/2004, A/Shenzhen/141/2005; and type B influenza viruses, i.e., B/Shangdon/7/97, B/Brisbane/32/2002, B/Shanghai/361/2002, B/Jiangsu/10/2003, B/Malaysia/2506/2004, B/Florida/7/2004, B/Ohio/1/2005, B/Victoria/502/2005.
Viruses were obtained from Alan Hampson and Ian Barr, World Health Organization Collaborating Centre for Reference and Research on Influenza, Melbourne, Australia. Influenza viruses were propagated in the allantoic cavity of 10-day-old embryonated hens’ eggs and purified from allantoic fluid as described previously (8).
Reassortant influenza viruses used in this study were generated by eight-plasmid reverse genetics as previously described (24). Viruses were 7:1 reassortants consisting of either the PR8 (H1N1) backbone with the HA or NA gene from Udorn/72 (H3N2; PR8-Udorn/72 HA and PR8-Udorn/72 NA, respectively) or the Udorn/72 backbone with the HA or NA gene from PR8 (Udorn/72-PR8 HA and Udorn/72 PR8 NA, respectively). Eight-plasmid reverse genetics was also used to generate wild-type PR8 and Udorn/72 viruses. The rescued viruses were recovered after 3 days and amplified in the allantoic cavity of 10-day-old embryonated hens’ eggs. The identity of all viruses was confirmed by restriction digestion and full-length sequencing of RT-PCR products for the HA, NP, NA, M, and NS genes.
Proteins, sera, and Abs
Recombinant human (rh)PTX3 was purified from the supernatants of Chinese hamster ovary (CHO) cells stably expressing the protein and biotin-labeled (bPTX3) as described (15). Bio-PTX3 was desialylated under nondenaturing conditions for 4 h at 37°C using 2 mU of Vibrio cholerae type III NA (Sigma-Aldrich) per microgram of protein. Naturally occurring human PTX3 was purified by immunoaffinity from human fibrosarcoma cells 8387 exposed to TNF-α (20 ng/ml) for 24 h (PTX38387). Recombinant and naturally occurring human PTX3 used in this study were glycosylated multimers consisting of 8 covalently bound multimers (15, 25). A cDNA encoding murine PTX3 cDNA was subcloned into the pSG5 expression vector, stably transfected in CHO cells, and recombinant protein was purified from the culture supernatant as described (15). Two mAbs, 2C3 and 6B11 (both IgG1), were obtained by immunization of PTX3−/− mice (18) with purified murine PTX3. C1q from human serum was purchased from Calbiochem. Recombinant human SP-D was expressed by CHO-K1 cells and purified by sequential maltosyl-agarose and gel filtration chromatography; dodecamers were used in all experiments as verified by ultrastructural analysis. The rabbit anti-human SP-D Ab was prepared using purified, C-terminal domains of natural SP-D as Ag (1).
Detection of PTX3 or SP-D binding to influenza virus by ELISA
To compare the binding of PTX3 or SP-D to different influenza viruses, wells of a microtiter tray were coated with increasing concentrations of purified influenza virus in 50 μl of TBS (0.05 M Tris-HCl, 0.15 M NaCl (pH 7.2)), blocked for >1 h with 10 mg of BSA per ml, and washed with TBS containing 0.05% Tween 20 (TBST). Wells were incubated for 2 h with 50 μl of biotin-labeled human PTX3 (bPTX3; 0.1 μg/ml) or rhSP-D (1.0 μg/ml) in TBS-T containing 5 mg of BSA per ml and either 20 mM CaCl2 (BSA5-TBST-Ca2+) or 5 mM EDTA (BSA5-TBST-EDTA) and then washed. Binding of bPTX3 was detected by the addition of streptavidin-conjugated HRP (BD Pharmingen). Binding of SP-D was detected by addition of rabbit antiserum against rhSP-D for 2 h, followed by the addition of HRP-conjugated swine anti-rabbit Ig (DakoCytomation). A similar protocol was used to determine binding of bPTX3 (0.25 μg/ml) and asialo-bPTX3 (0.25 μg/ml) to wells coated with increasing concentrations of either HKx31 or C1q.
HA and HA inhibition tests
Tests were performed in round-bottom 96-well microtiter plates at room temperature using 1% v/v chicken erythrocytes. Hemagglutination titers were determined by titration of virus samples in TBS followed by addition of an equivalent volume of chicken erythrocytes. For hemagglutination inhibition tests, dilutions of SP-D or PTX3 were prepared in TBS alone (or in TBS supplemented with 10 mM CaCl2 or 5 mM EDTA) and 4 hemagglutinating units (HAU) of virus was added. Following 30 min of incubation, chicken erythrocytes were added and the ability of PTX3 or SP-D to inhibit virus-induced hemagglutination was assessed. Results are expressed as the minimum concentration of PTX3 or SP-D required to fully inhibit the hemagglutinating activity of 4 HAU of virus and are expressed as μg/ml.
NA and NA inhibition assays
Influenza virus NA activity and its inhibition by PTX3 were measured by an enzyme-linked microplate assay in which Arachis hypogaea (peanut) lectin was used to detect β-d-galactose-N-acetylglucosamine sequences exposed after the removal of sialic acid from fetuin (26). Titrations of NA activity and inhibition of NA activity by PTX3 or SP-D were performed as described (3).
Virus neutralization assay
Neutralization of virus infectivity was measured by fluorescent-focus reduction in monolayers of Madin-Darby canine kidney (MDCK) cells cultured in 96-well plates as described (3). In brief, dilutions of PTX3 were mixed with virus, incubated at 37°C for 30 min and inoculated onto MDCK monolayers. After adsorption of the virus for 45 min at 37°C, the inoculum was removed, fresh medium was added and plates were incubated at 37°C for an additional 6 to 7 h. Cells were fixed and stained with anti-NP mAb A-3, and 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. To test for inhibition by sugars or by mannan, diluted PTX3 was incubated with sugar or mannan for 20 min at room temperature before addition of virus. The appropriate concentration of sugar or mannan was also included in the virus control.
Mice, infections, and PTX3 treatment
C57BL/6, PTX3+/+, and PTX3−/− mice in a mixed C57BL/7 × 129S background were housed in specific pathogen-free conditions in the animal facility at the University of Perugia, Italy. Adult mice (6–10 wk) were used in all experiments. Mice were lightly anesthetized with diethyl ether and infected via the intranasal route with 2 × 104 PFU of influenza virus in 20 μl of saline. Recombinant human PTX3 (1 and 2 mg/kg, via i.p. route) was administered on the day of the infection and daily thereafter for 4 consecutive days. Experiments were performed according to the Italian Approved Animal Welfare Assurance A-3143-01. To determine virus titers in lungs, mice were euthanized and lungs were removed, homogenized, and clarified by centrifugation and supernatants frozen at −70°C. The samples were assayed for infectious virus by measuring plaque formation in MDCK cells in the presence of trypsin. For collection of bronchoalveolar lavage (BAL) fluids, mice were euthanized and lungs were flushed 3 times with a 1-ml volume of PBS through a blunted 23-gauge needle inserted into the trachea.
Assay for PTX3 in BAL fluid
The presence of PTX3 in mouse BAL fluid was assayed by ELISA on wells coated with mAb 2C3 (anti-murine PTX3) at 1 μg/ml in 15 mM carbonate buffer (pH 9.6). After washing, wells were blocked with 3% skim milk for 2 h at room temperature, and samples were titrated relative to a standard curve of murine PTX3 (10 ng/ml–9.7 pg/ml). Bound PTX3 was detected with 0.25 μg/ml biotin-labeled mAb 6B11 (anti-murine PTX3) followed by streptavidin-conjugated HRP and substrate.
Binding of PTX3 to purified influenza viruses
SP-D, a collectin present in respiratory fluids, and the short pentraxin SAP bind to influenza viruses in a Ca2+-dependent manner (4, 6, 22, 23) and mediate antiviral activity by binding to mannose residues on the viral HA/NA glycoproteins. Therefore, studies were undertaken to assess the ability of the long pentraxin PTX3 to bind influenza virus by ELISA. Wells coated with increasing concentrations of purified viruses were probed with biotin-PTX3 in the presence or absence of Ca2+ (Fig. 1,A). PTX3 binds to ligands such as the extracellular matrix protein TSG-6 and Omp A from Klebsiella pneumoniae in a Ca2+-dependent manner (20, 27) whereas binding of PTX3 to C1q is Ca2+ independent (15); accordingly, binding to influenza virus strain HKx31 (H3N2) was not inhibited by chelation of Ca2+ with 5 mM EDTA. Furthermore, binding of bPTX3 was not inhibited by inclusion of d-mannose or mannan, but was inhibited by preincubation of HKx31-coated plates with purified PTX3 that was not biotinylated, indicating the specific nature of the interaction between PTX3 and HKx31. In contrast, binding of human SP-D to HKx31 was Ca2+-dependent and inhibited in the presence of either mannose or mannan (Fig. 1 B).
We next tested binding of biotin-PTX3 to a range of purified influenza viruses known to differ in their degree of glycosylation of the HA glycoprotein, namely HKx31 (7 glycosylation sites on HA), Udorn/72 (H3N2; 7 sites), Beij/89-X (H3N2; 9 sites), Phil82-X (H3N2; 10 sites) and PR8 (H1N1; 5 sites); a particular feature of the PR8 strain is the lack of glycosylation sites on the exposed head of the HA molecule (28, 29). PTX3 bound specifically to wells coated with purified HKx31 and Udorn/72, but not to wells coated with equivalent concentrations of Beij/89-X, Phil/82-X or PR8 (Fig. 1,C). SP-D displayed a very different pattern of reactivity, binding with high avidity to heavily glycosylated Phil/82-X and Beij/89-X strains, with intermediate avidity to HKx31 and Udorn/72, and weakly to strain PR8 (Fig. 1 D). Together, these findings indicate that PTX3 binds only to particular strains of influenza virus, and that binding is not determined by the presence of Ca2+, nor by the amount of mannose-containing oligosaccharides on the virus.
Antiviral activity of PTX3 against influenza virus HKx31
The potential of recombinant human PTX3 to mediate antiviral activity against influenza virus strain HKx31 was compared with that of recombinant human SP-D. First, PTX3 was shown to inhibit HKx31-induced hemagglutination of chicken erythrocytes, indicating that binding of PTX3 to virus was sufficient to block access of the viral HA glycoprotein to erythrocyte cell surface receptors (Fig. 2 A). The ability of PTX3 to inhibit hemagglutination was Ca2+ independent and unaffected by the inclusion of mannan. In contrast, hemagglutination inhibition of HKx31 by SP-D was inhibited in the presence of mannan, or by chelation of Ca2+ with EDTA.
We next compared the ability of PTX3 and SP-D to inhibit the activity of the viral NA. Preincubation of HKx31 with PTX3 (Fig. 2,B) or SP-D (Fig. 2,C) inhibited the viral NA in a dose-dependent manner and inhibition by SP-D, but not PTX3, was abolished in the presence of mannan. Preincubation with PTX3 (Fig. 2,D) or SP-D (Fig. 2,E) also inhibited the ability of HKx31 to infect susceptible MDCK cells in a neutralization assay. The neutralizing activity of SP-D, but not PTX3, was inhibited in the presence of mannan. The sensitivity of BJx109 (highly glycosylated), HKx31 (intermediate levels of glycosylation), and PR8 (poorly glycosylated) to inhibition by SP-D or PTX3 was also examined using NA inhibition ELISA (Fig. 2, B and C) and neutralization assay (Fig. 2, D and E). In both assays a clear hierarchy was observed in the sensitivity of virus strains to SP-D (BJx109 > HKx31 > PR8). In contrast, only HKx31 was sensitive to inhibition by PTX3. Together, these studies clearly demonstrate that SP-D and PTX3 mediate antiviral activity against influenza viruses via a distinct mechanism.
Antiviral activity of PTX3 against different strains of influenza virus
We next examined a range of influenza isolates for their sensitivity to PTX3 (and for comparison to SP-D) using an hemagglutination inhibition (HI) assay. In previous studies, we have reported that early and late H3N2 subtype viruses differ in their degree of glycosylation and therefore in their sensitivity to collectins, such that strains isolated after 1977 were particularly sensitive to SP-D (3). Data presented in Fig. 3 show the minimum concentration of PTX3 or SP-D required to inhibit the hemagglutinating activity of a range of H3N2 and H1N1 subtype viruses. Consistent with previous reports, we found later H3N2 isolates (1977–1993) particularly sensitive to HI by human SP-D. In contrast, PTX3 showed a very different pattern of reactivity against H3N2 subtype viruses, with early isolates (1968–1973) sensitive to inhibition by PTX3 but later strains (1977–1993) resistant to the highest concentration tested. Furthermore, whereas H1N1 subtype isolates were either sensitive or resistant to SP-D, all isolates tested were resistant to inhibition by PTX3 (Fig. 3). We tested an additional range of human influenza viruses for sensitivity to PTX3, including recent H3N2 isolates (1997–2005; refer to Materials and Methods), H1N1 isolates (1995–2005; refer to Materials and Methods) and type B influenza virus isolates (1997–2004; refer to Materials and Methods); all strains tested (1995–2005) were found to be resistant to PTX3 (data not shown).
PTX3 binds to the HA glycoprotein of PTX3S strains of influenza virus
To determine which glycoproteins were bound by PTX3, we used eight-plasmid reverse genetics to construct wild-type viruses Udorn/72 (PTX3S) and PR8 (PTX3R), as well as reassortant viruses consisting of 1) PR8 backbone with the HA or NA gene of Udorn/72 (PR8-Udorn72 HA and PR8-Udorn72 NA viruses, respectively); or 2) Udorn/72 backbone with the HA or NA gene of PR8 (Udorn/72-PR8 HA and Udorn72-PR8 NA, respectively). Each virus was tested for sensitivity to PTX3 in HI assay or in a virus neutralization assay. As expected, wild-type Udorn/72 was sensitive to HI (Fig. 4,A) and neutralization (Fig. 4,B) by PTX3 but wild-type PR8 was resistant. This sensitivity to both HI and neutralization by PTX3 was found to correlate with expression of the Udorn/72 HA by the reassortant viruses. Furthermore, cells infected with wild-type Udorn/72 or Udorn/72-PR8 NA bound bPTX3; however, cells infected to similar levels with wild-type PR8 or Udorn/72-PR8 HA did not (Fig. 4 C). Together, these data indicate that PTX3 binds to the HA glycoprotein of sensitive virus strains.
Recognition of sialic acid on PTX3 by influenza virus HA
Collectins such as SP-D, MBL, and conglutinin are classified as β inhibitors of influenza viruses and act via Ca2+-dependent binding to mannose-containing glycans on the viral HA/NA glycoproteins (6, 7, 8). In contrast, α2-macroglobulin and collectin SP-A mediate antiviral activity by providing sialic acid residues that are recognized by the HA glycoprotein of influenza viruses and are classified as γ inhibitors (13, 14, 30). Oligosaccharide analysis of PTX3 has indicated the presence of complex type sugars displaying varying degrees of sialylation attached to the single N-linked glycosylation site at Asn220 in the pentraxin domain of the PTX3 molecule (25). To assess the contribution of oligosaccharide to interactions with influenza viruses we compared the ability of enzymatically desialylated bPTX3 (asialo-bPTX3) to bind to HKx31 by ELISA. Treatment with V. cholerae type IΙΙ sialidase completely inhibited the ability of bPTX3 to bind HKx31 (Fig. 5,A). Consistent with previous reports (25), desialylated bPTX3 bound complement component C1q more effectively than untreated bPTX3 (Fig. 5 B), confirming that the desialylated protein retained binding activity for other known ligands of PTX3. Furthermore, asialo-bPTX3 failed to inhibit hemagglutination or neutralize infectivity of HKx31 (data not shown). Together, these data indicate that like α2-macroglobulin and SP-A, PTX3 provides sialic acid residues which are bound by the HA glycoprotein of certain virus strains, and therefore acts as a γ inhibitor against influenza viruses.
Administration of PTX3 to mice reduces morbidity following infection with HKx31
To explore the role of PTX3 in vivo, C57BL/6 mice were infected with either HKx31 (PTX3S) or PR8 (PTX3R) and treated with different doses of human PTX3 daily for 4 days, beginning the day of infection. Mice were monitored for survival over the next 15 days or killed at day 5 postinfection and viral load quantified in the lungs by standard plaque assay. Infection with HKx31 resulted in 40% survival of untreated mice; however, daily treatment with either 1 mg/kg or 2 mg/kg PTX3 promoted 100% survival of HKx31-infected animals (Fig. 6,A). The antiviral effects of PTX3 treatment were reflected in lung viral titers at day 5 postinfection (Fig. 6,B), with viral titers reduced 10- to 100-fold in mice treated with 1 mg/kg or 2 mg/kg, respectively, when compared with those of untreated animals. In contrast, PTX3 treatment had no effect on survival following infection with PR8 virus (Fig. 6,C) and 80% of mice succumbed to disease in the absence or presence of therapeutic PTX3 treatment (2 mg/kg). PTX3 treatment, did however, have some effect in reducing viral load in the lungs of PR8-infected mice treated with 1 mg/ml and 2 mg/kg of PTX3 suggesting an opsonin-independent effect in antiviral resistance (Fig. 6 D). This reduction was not, however, sufficient to protect animals from the severe disease and subsequent lethality of the PR8 virus.
Role of endogenous PTX3 during infection of mice with PTX3S or PTX3R virus strains
We found recombinant murine PTX3 to be active against HKx31 and other “early” H3N2 virus strains, but not against PR8 virus, in vitro. To assess the role of endogenous PTX3 in the murine model of influenza infection we first determined levels of PTX3 present in airway fluids from uninfected and virus-infected mice. After infection with 105 PFU of HKx31, PTX3 levels in BAL were noted to rise during the course of infection (Fig. 7 A). Of interest, infection of mice with an equivalent dose of the PTX3R PR8 virus induced higher levels of PTX3 at day 5 postinfection (p < 0.001 compared with HKx31-infected mice at day 5), likely reflecting the severe disease associated with infection of mice by this virulent virus (31). Inoculation of mice with 105 PFU of HKx31 or PR8 that had been UV inactivated to destroy virus infectivity did not induce levels of PTX3 above those observed in uninfected mice at day 3 or day 5 postinfection (data not shown). The induction of murine PTX3 during influenza infections is consistent with a role in host defense.
To formally address the role of endogenous PTX3 during influenza infection we infected PTX3+/+ or PTX3−/− mice with an equivalent dose of HKx31 or PR8 and assessed viral growth in the lungs 5 days after infection. We found virus titers were consistently higher in the lungs of PTX3−/− infected with HKx31 compared with PTX3+/+ mice (Fig. 7 B) however virus titers were not significantly different in the lungs of PTX3−/− or PTX3+/+ mice infected with PR8. These data indicate an important role for endogenous PTX3 in limiting virus replication following intranasal infection with the PTX3S strain HKx31 but not with the PTX3R PR8 strain.
Innate inhibitors of influenza A viruses, including the collectins SP-A and SP-D and the pentraxin SAP, are constitutively expressed in fluids lining the respiratory tract. Previous reports have described the activity of SAP against influenza viruses in vitro (22, 23), however in vivo studies indicate a limited role in host defense against influenza (32). Herein we report that long pentraxin PTX3 mediates a range of antiviral activities against influenza viruses in vitro but also makes an important contribution to early host defense in vivo. Together, these studies identify PTX3 as a novel inhibitor of influenza viruses and highlight its potential for therapeutic use during influenza infection.
In this study we describe marked differences in the mechanisms by which long pentraxin PTX3 acts against influenza viruses when compared with the collectin SP-D, or to the short pentraxin SAP. PTX3 acts as a potent γ inhibitor, providing sialylated ligands that mimic the structure of the cellular receptors used by influenza viruses thereby blocking the receptor-binding site of HA. In contrast, SAP (and SP-D) act as classic β inhibitors, binding to mannose-rich glycans on the viral HA to sterically block access of the viral HA to cell surface receptors (22, 23). Despite in vitro activity against influenza viruses, SAP−/− mice did not display enhanced susceptibility to influenza infection (32). The sequence and regulation of SAP have diverged markedly from man to mouse such that murine SAP binds influenza virus poorly when compared with human SAP; even so, no protection was afforded to SAP−/− mice transgenic for human SAP following influenza infection (32). PTX3 is highly conserved in evolution with ortholog genes described in mammals as well as in birds and the most ancient vertebrate Takifugu rubripes (33). Consistent with this notion, we found murine PTX3 to bind effectively to purified HKx31 and to mediate a range of antiviral activities in vitro at equivalent doses to that of human PTX3 (data not shown). Furthermore, mice with a targeted deletion in the PTX3 gene were markedly more susceptible to HKx31 infection (Fig. 7,B), and treatment of C57BL/6 mice with human PTX3 effectively reduced the viral load in the lungs of infected animals (Fig. 6 B). Together, these data indicate that endogenous murine PTX3 is an important antiviral host defense protein and can play a role in limiting influenza infection in vivo.
Naturally occurring γ inhibitors of influenza virus have been known to exist in serum of different species (including human) and in fluid secretions such as lung and pleural fluids for many years (reviewed in Ref. 34, 35). The high molecular-weight glycoprotein α2-macroglobulin is the major inhibitor in horse, guinea pig and pig sera and equine α2-macroglobulin has been particularly well characterized (14, 36, 37). Resistance to equine α2-macroglobulin is associated with a single amino acid alteration at position 226 in the receptor binding pocket of HA and is associated with a change in receptor specificity from sialic acid (SA) linked to galactose by α2,6 linkages (SAα2,6Gal) to SAα2,3Gal (38, 39). We found wild-type HKx31 and a horse serum-resistant mutant of this virus to be equally sensitive to inhibition by murine or human PTX3 (data not shown), indicating definitive differences in mode of action between equine α2-macroglobulin and PTX3. The molecular mechanisms underlying the ability of human PTX3 to bind to and inactivate influenza viruses will be the subject of future studies.
Innate inhibitors such as PTX3 could influence the selection of influenza virus receptor variants in natural hosts; however, direct experimental support for this hypothesis is lacking. The H3N2 subtype has circulated widely in the human population since its introduction from an avian source in 1968. Previous studies have demonstrated that early H3N2 subtype isolates recognize both SAα2,3Gal and SAα2,6Gal linkages whereas isolates from 1975 to 1994 recognize only SAα2,6Gal (40). Thus, it appears that evolution of this subtype has selected for receptor variants focused on recognition of SAα2,6Gal receptors. Our findings that early H3N2 isolates (1968–1973) were sensitive to PTX3 while later isolates (1975–2005) were resistant correlate with the presence of SAα2,3Gal residues on PTX3, such that later H3N2 isolates are unable to recognize this linkage. Indeed, biochemical analysis of human PTX3 confirmed the presence of SAα2,3 on human PTX3 (Antonio Inforzato, unpublished observations). Neutralizing γ inhibitors such as PTX3 (and perhaps SP-A) bearing SAα2,3Gal may provide additional selective pressure to drive evolution of human viruses toward a decreased affinity for SAα2,3Gal. Such evolution would not be expected compromise the fitness of virus because SAα2,6Gal is the predominant receptor type present on epithelial cells lining the human respiratory tract.
PTX3 may interfere with influenza virus infection in vivo in a number of ways. Neutralization of virus infectivity may result from direct blocking of virus attachment or entry into host cells, whereas inhibition of viral NA by PTX3 may inhibit the release of newly formed virus particles from the surface of infected cells in a manner analogous to that of NA-specific Abs (41). Incubation of influenza virus with other sialylated inhibitors, such as SP-A, gp340, and mucin, induces viral aggregation (2) which could reduce levels of infectious virus and promote clearance by mucociliary and phagocytic mechanisms. We have also found that PTX3 binds to influenza virus-infected epithelial cells (Fig. 4 C), suggesting the potential for destruction before the release of newly synthesized virions. To this end, PTX3 has been shown to act directly as an opsonin, increasing the phagocytic activity of macrophages for particular yeasts (17, 18). Given that PTX3 binding to apoptotic cells enhanced deposition of complement components C1q and C3 (42), opsonization of virus-infected cells with complement components may further promote further uptake and destruction of virus-infected cells by professional APCs.
Our in vivo data describe a role for both human and murine PTX3 in limiting the growth of PTX3S virus strains in the respiratory tract of influenza virus-infected mice. In addition, we report that therapeutic treatment of C57BL/6 mice with human PTX3 also reduces the growth of the resistant PR8 strain (Fig. 6,D). Although therapeutic treatment of uninfected B6 mice with PTX3 does not enhance serum cytokine levels (data not shown), others have reported PTX3 to facilitate production of cytokines in an inflammatory or infectious environment (43, 44). Future studies will determine the effect of therapeutic PTX3 treatment on the production of proinflammatory cytokines in the airways during infection with PTX3S/PTX3R virus strains; cytokines such as TNF-α have been reported to modulate influenza virus-induced inflammation and to mediate direct antiviral activity (45, 46). The PTX3R PR8 virus did, however, grow to similar titers in mouse lung in PTX3+/+ and PTX3−/− mice (Fig. 7,B) and C57BL/6 mice were equally susceptible to PR8-induced disease despite PTX3 treatment (Fig. 6 C), arguing that the direct antiviral role of PTX3 may be the most important in protection against influenza infection in vivo.
Treatment of C57BL/6 mice with human PTX3 clearly demonstrate its therapeutic potential in ameliorating disease and reducing morbidity associated with influenza infection, however as recent H1N1, H3N2 and type B influenza viruses are resistant to PTX3 it is likely that the role of PTX3 during current human influenza infections is limited. This does not, however, rule out the potential for endogenous or therapeutic PTX3 to modulate disease should a new subtype emerge within the human population in the future. As our data using H3N2 virus strains demonstrated, when a new subtype emerges it may be sensitive to inhibition by PTX3 and the acquisition of resistance may be a feature associated with its continued survival in the human population. Given the devastating morbidity and mortality that may be associated with the appearance of a new subtype of human influenza, the role of PTX3 as an innate antiviral protein may be particularly important in future influenza outbreaks.
Respiratory secretions contain a complex mixture of innate proteins, including PTX3 and members of the collectin family. Early and late strains of H3N2 subtype viruses differ in their degree of glycosylation and therefore in their sensitivity to collectins, with heavily glycosylated strains isolated after 1977 being particularly sensitive to SP-D (Ref. 3 and Fig. 3). PTX3 and SP-D may play distinct but cooperative roles in innate host defense against H3N2 subtype viruses; PTX3 sensitivity was a feature of early virus strains (1968–1975) whereas SP-D sensitivity was a feature of later (1977–1993) and currently circulating strains. Studies using SP-D−/− mice (9, 11) have demonstrated the importance of SP-D in limiting disease in mice infected with late H3N2 subtype viruses that we have shown to be resistant to PTX3. Together, innate inhibitors may play a cooperative role in limiting initial virus replication and spread but the predominant inhibitor acting to limit virus growth may be different for particular virus strains.
We thank Dr. Marica Sassano from Tecnogen Società Per Azioni for the supply of PTX3 for the in vivo studies, Ian Barr and Robert Shaw from World Health Organization Collaborating Centre for Reference and Research on Influenza (Melbourne, Australia) for coordinating testing of additional type A H1N1 and H3N2 subtype viruses and type B influenza viruses, and Dr. Robert Webster, St. Jude Children’s Research Hospital (Memphis, TN) for provision of the plasmid vector used to create the reverse engineered viruses for this study.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Project Grant 400226 and a Programme Grant from The National Health and Medical Research Council of Australia, and by Research Grant RIGP/05/CR90 from Tecnogen Società Per Azioni Italy (to L.R.). E.C.C. was supported by Grants HL44015 and HL29594 from the National Institutes of Health. A.M. was supported by Cariplo Foundation (Project Next Generation Optical Network for Broadband European Leadership), Telethon, Project Fluinnate, and Mugen from the European Commission. P.C.R. is a National Health and Medical Research Council R. D. Wright Research Fellow.
Abbreviations used in this paper: SP-D, surfactant protein D; SP-A, surfactant protein A; HA, hemagglutinin; NA, neuraminidase; CHO, Chinese hamster ovary; HAU, hemagglutinating unit; MDCK, Madin-Darby canine kidney; BAL, bronchoalveolar lavage; HI, hemagglutination inhibition; SA, sialic acid.