Contribution of the Human Parainfluenza Virus Type 3 HN-Receptor Interaction to Pathogenesis In Vivo

Abstract

The envelope of human parainfluenza virus type 3 (HPF3) contains two viral glycoproteins, the hemagglutinin-neuraminidase (HN) protein and the fusion (F) protein. In a previous study, highly fusogenic variant HPF3 viruses were isolated, including two, C-0 and C-22, that exhibit increased avidity for sialic acid receptors due to single amino acid changes in the HN protein and one, C-28, that has decreased neuraminidase activity relative to that of the wild type (wt) and is delayed in the release of virus particles into the supernatant fluid. These variants form very large plaques and destroy a cell monolayer more rapidly than does wt HPF3 in cell culture. These variant viruses allowed us to formulate hypotheses about the roles of HN in pathogenesis. We investigated the behavior of wt HPF3 and the three variant viruses in the cotton rat model. In the cotton rat, there was no delayed clearance of any of the variant viruses compared to that of the wt. The variant plaque morphology was preserved in vivo, and there was no reversion to the wt phenotype in the infected animals. In spite of a slight advantage of wt virus in viral titer, there were no differences in the severities of peribronchiolitis between wt viruses and the variants. However, there were marked differences in severities in alveolitis and interstitial pneumonitis when each of the three variants was compared to the wt, with the variants causing enhanced disease. Thus, despite similar or lower viral titers and similar clearance rates, the variants caused more extensive disease in the lung. The results show that mutations in HN conferring altered fusion properties in cell culture also confer striking differences in the ability of HPF3 to cause extensive disease in the cotton rat lung and that this effect is dissociated from any effect on viral replication.

The Paramyxoviridae family is comprised of several important agents of human pathology, including measles, mumps, respiratory syncytial, and human parainfluenza viruses. Human parainfluenza virus type 3 (HPF3) is the second leading cause of infant and childhood respiratory disease, and no vaccine or antiviral therapy for this agent is currently available.

The envelope of HPF3 contains two viral glycoproteins, the hemagglutinin-neuraminidase (HN) protein and the fusion (F) protein. Infection of cells by HPF3 is initiated by attachment of the virus to the host cell through interaction of the HN glycoprotein with a sialic acid-containing cell surface receptor. Penetration and uncoating of the virus result from F protein-mediated fusion of the viral envelope with the plasma membrane of the cell, which leads to the release of the viral nucleocapsid into the cytoplasm. For fusion to occur, both interaction of the viral HN glycoprotein with its sialic acid receptor and the presence of the viral F glycoprotein are required (5, 6, 8, 11, 12). By virtue of its neuraminidase activity, HN also has a receptor-destroying potential that plays a role in the spread of infection (7).

The hallmark cytopathic effect of acute infection with HPF3 in vitro is extensive cell fusion resulting in syncytium formation, which involves the interaction of F and HN proteins expressed on the surface of an infected cell with the membrane of an adjacent uninfected cell. In a previous study, variant HPF3 viruses that have a greatly increased ability to fuse cells in culture were isolated; this offered a new approach to understanding the mechanism of paramyxovirus-induced cell fusion and the role of the HN protein in this process. The two highly fusogenic variants of HPF3 that were isolated, C-0 and C-22, exhibited increased avidity for sialic acid receptors due to single amino acid changes in the HN protein (12). These studies demonstrated that a key component of HN's function in promoting fusion in cell culture is its avidity of binding to sialic acid-containing receptors.

In continuing to study the role of neuraminidase in the life cycle and pathogenesis of HPF3, we isolated a variant of HPF3 (C-28) that has decreased neuraminidase activity relative to that of the wild type (wt) (7). Analysis of the growth properties of this variant revealed a delay (of 7 h) in the release of virus particles into the supernatant; the addition of exogenous neuraminidase to the culture corrected this delay. These findings implicated the neuraminidase activity of HN in the release of HPF3 virus particles from the surface of the infected cell, thus beginning a new round of infection (7).

In order to analyze the factors affecting pathogenesis of HPF3 in vivo and the contribution of HN's functions to pathogenesis, previous studies have now been extended to the animal model. We wanted to determine whether the avidity of virus-receptor interaction and the receptor-destroying neuraminidase activity are major determinants of virulence and/or pathogenesis in the lung. With the use of the cotton rat, we were able to assess whether the cell culture findings translated into pathogenic mechanisms in the lung. The cotton rat is an excellent model for HPF3 lower respiratory infection (18); experimental infection leads to bronchiolitis and interstitial pneumonia, mimicking human disease. Experiments using the cotton rat model for respiratory syncytial virus (RSV) have been used to predict disease outcome in humans, in that RSV pneumonia was prevented by RSV immune globulin (4, 20, 21). Such evidence for experimental relevance to humans is not yet available for HPF3 in the cotton rat because the data regarding therapy for HPF3 obtained in the cotton rat model (19) have not yet been tested in clinical trials. However, the pulmonary disease seen during cotton rat infection with HPF3, unlike that with the mouse model, corresponds to that seen during human infection, making this a useful model for elucidating the basis for HPF3 disease.

Use of receptor-binding and neuraminidase variants of HPF3 in the cotton rat.

The hypothesis tested in the present study was that functions of HN that determine cytopathic effects in cell culture are important determinants of HPF3 pathogenesis in the lung of the cotton rat. Two of the variants used in this study, C-0 and C-22, exhibited higher avidity than wt HPF3 for the sialic acid-containing cellular receptor. The variants are highly fusogenic by several criteria: plaques form more rapidly and are much larger than wt virus plaques, and infection of a monolayer results in much more rapid and widespread cell fusion. Both variants have the wt F gene sequence. Each variant has one point mutation in the HN gene corresponding to a single amino acid change in the HN glycoprotein; in C-0, an A-to-G mutation converts threonine 193 to an alanine, while in variant C-22, a T-to-G mutation converts histidine 552 to a glutamine (12). Growth characteristics of the variant viruses in cell culture do not differ from those of the wt in that the rates of viral replication and the rates of viral protein synthesis are similar in wt- and variant-infected cells. The amount of HN and the ratio of HN to NP protein in wt- and variant-infected cells are similar. Activity levels, substrate specificities, and pH optima of the neuraminidase of C-0 and C-22 were not significantly different from those of the wt.

We also determined whether the neuraminidase activity of HN, which is critical to the outcome of infection in cell culture, is a determinant of the outcome of infection in the lung, by using variant C-28, which has 40% of wt neuraminidase activity (7). C-28, with no alterations in the F protein sequence, has a single point mutation in the HN protein, i.e., a G-to-A mutation at nucleotide 724 that converts aspartic acid 216 to asparagine. C-28 shows an initial growth delay in cell culture due to slow release but then catches up and causes more widespread fusion than does the wt. The increased fusogenicity of C-28 is attributable to its low neuraminidase activity, which leaves more cellular sialic acid receptor available to bind HN.

Viral replication.

Inbred young adult cotton rats (Sigmodon hispidus) of both genders were obtained from the breeding colony at Virion Systems, Inc., housed in large polycarbonate cages, and fed a diet of standard rodent chow and water. The animals were seronegative for adventitious paramyxoviruses, RSV, and other common rodent pathogens.

The rats were infected intranasally with 100 μl containing two different input doses for each virus, 10^5.5 and 10^6.5 PFU. The animals were sacrificed for study at 2, 4, 6, 8, and 10 days after infection to obtain a time course of viral replication and disease progression in the lung. The time points were selected based on the following rationale. At 2 days after infection, C-22, C-0, and C-28 exhibit large plaques in a cell monolayer. Four days represents the peak of wt HPF3 replication in the lung, and 6 days is the time of the peak lung inflammatory response with wt HPF3 (18). The 8- and 10-day time points were chosen in an effort to look for delayed clearance of the variant viruses.

Following sacrifice of the animals by carbon dioxide inhalation, nasal and lung tissues from each animal were bisected (the other half being used for histologic analysis) and homogenized separately as described previously (18), and then they were stored at −70°C until assayed. Virus titers were determined by plaque assay on MA104 African green monkey kidney cells and calculated as PFU per gram of tissue.

Wild-type virus replicated to a titer that was similar to or slightly higher than those of the three variants in the lungs and nose (Fig. 1). The differences were most pronounced in the case of C-0, where peak titers were significantly lower (P ≤ 0.05, t test of summary data) than wt titers in the lungs (days 2 and 4) and nose (day 2), and in the case of C-22, where peak titers were significantly lower than those of the wt in the nose (days 2 and 4). There were no significant differences in the kinetics of replication, with wt and variant viruses being cleared by day 6 in the lungs and day 8 in the nose. Of note, for each of the three variants, the plaque morphology of the variant viruses was preserved in vivo, with no apparent reversion to the wt phenotype. All the plaques from viruses derived from variant-infected animals showed the typical large, round plaques characteristic of each variant (7, 12).

FIG. 1.

Geometric mean virus titers in lung and nasal tissues from cotton rats on days 2, 4, 6, 8, and 10 after infection with C-28, C-22, C-0, and wt HPF3. Each point represents eight (lungs) or four (noses) animals. On days 2 and 4, some titers for C-22 and C-0, as indicated in the text, were significantly lower (P ≤ 0.05) than those for the wt.

Histopathology of lung tissue.

Lungs were inflated intratracheally with 10% neutral buffered formalin in order to maintain the pulmonary architecture. Two uninfected rats were sacrificed at each time point for comparison of tissue pathologies. Following paraffin embedding, 4-μm sections were cut and then stained with hematoxylin and eosin. Three parameters of pulmonary inflammatory changes were scored in each lung section: peribronchiolitis (inflammatory cells, primarily lymphocytes, surrounding a bronchiole), alveolitis (inflammatory cells within alveolar spaces), and interstitial pneumonitis (increased thickness of alveolar walls associated with inflammatory cells). Each of these parameters was scored separately for each histologic section. Prior to scoring, all of the slides were examined to determine the range of pathologies for each of the three parameters. A semiquantitative assessment was made of each type of inflammation, based on the range of lesions seen within this experiment and on extensive prior studies of cotton rat pulmonary pathology. The absence of inflammation was given a score of 0, the theoretical maximum based on the above criteria was given a score of 100, and the score for inflammation falling between the two extremes was estimated while the tissues were viewed under the low-power objective of a microscope. The slides were randomized and scored blindly. Although each of the three types of inflammation was scored with the same scale, the scores are relative and valid only for comparing the same parameter in different sections, not for comparing different parameters.

Histopathologic changes did not correlate with viral titers in any of the three parameters that we measured in the lungs. Peribronchiolitis (Fig. 2) was a prominent finding in all four groups, with slight differences (P < 0.05) on some days between variants and wt but no consistent trends. All three variants caused significant (P ≤ 0.05) alveolitis (Fig. 2) and interstitial pneumonitis (data not shown), whereas no such pathologic changes were produced by the wt virus, despite the fact that the wt achieved higher titers than did the variants. Figure 3 shows representative examples of the histopathology that was seen. Figure 3A shows wt-infected lung tissue with peribronchiolitis but neither alveolitis nor interstitial pneumonitis. Figure 3B shows alveolar tissue from wt-infected lung tissue with no evidence of disease. Figure 3C shows C-0 variant-infected lung tissue with peribronchiolitis that is indistinguishable from that caused by wt. Figure 3D shows alveolar tissue from C-0 variant-infected lung tissue with striking alveolitis (cells within air spaces) and interstitial pneumonitis (thickened alveolar walls).

FIG. 2.

Arithmetic mean pulmonary pathology scores (plus standard errors) for severity of peribronchiolitis and alveolitis. Interstitial pneumonitis showed the same pattern as alveolitis and is not included in this figure. Asterisks indicate values that are significantly higher than values for the wt for the same day postinfection. Each group consisted of eight animals.

FIG. 3.

Representative examples of lung tissue histopathology. (A) wt virus-infected lung tissue showing peribronchiolitis. Magnification, ×64. (B) wt virus-infected lung showing no evidence of alveolar or interstitial pathology. Magnification, ×128. (C) C-0 variant-infected lung showing peribronchiolitis. Magnification, ×64. indistinguishable from that caused by wt. (D) C-0 variant-infected lung showing striking alveolitis (cells within air spaces) and interstitial pneumonitis (thickened alveolar walls).

Histopathology of nasal tissue.

For examination of nasal tissues, each rat skull was placed in 10% formalin and then decalcified prior to sectioning. Multiple coronal sections, at 1-mm intervals, were prepared for each animal. Three parameters were scored separately for each specimen: epithelial cell damage, epithelial infiltration (primarily neutrophils), and exudate (humoral or cellular) into the nasal airspace.

None of the four viruses caused significant pathologic changes in any of the three parameters that we examined (data not shown), a finding consistent with earlier observations on wt HPF3 infection in cotton rats (18).

Implications of enhanced disease caused by receptor-binding and neuraminidase variants of HPF3.

In this study, we investigated the behavior of wt HPF3 and three variant viruses in the cotton rat model, i.e., the high-affinity variant viruses C-22 and C-0 (12) and the low-neuraminidase variant C-28 (7). Since these highly fusogenic variants form very large plaques and destroy a cell monolayer more rapidly than does wt HPF3 in cell culture, we wanted to determine whether this would translate into increased virulence in an animal. The results showed no delay in clearance of any of the variant viruses compared to that of the wt. The variant viruses, while being cleared at the same time as for the wt, took longer to reach their peak titers. The variant plaque morphology was preserved in vivo, with the viruses that emerged from the infected animals showing the typical large plaques characteristic of these variants, and there was no apparent reversion to the wt phenotype in the infected animals. In spite of the slight advantage of the wt virus in viral titer, there were no differences in the severity of peribronchiolitis between the wt and the variants. However, there were marked differences in the severities of alveolitis and interstitial pneumonitis when each of the three variants was compared to the wt. Thus, despite similar or lower viral titers and similar clearance rates, the variants caused more extensive disease in the lung. There was no significant nasal pathology in any of the infected animals, a finding consistent with the previous studies of HPF3 disease in the cotton rat (18).

The C-22 and C-0 variants cause increased fusion in cell culture due to an increased avidity of HN for its sialic acid-containing receptor (12). We found that while C-22 and C-0 have no advantage in terms of replication in the lung, and in fact replicate to a slightly lower titer than does the wt virus, they cause enhanced pathology. The altered HN, with increased avidity for its receptor, is thus one determinant of pathogenesis in the lung.

C-28 has only approximately 40% of the neuraminidase activity of wt HPF3, causing an initial growth delay due to slow release from the surface of the infected cell. The C-28 variant's release eventually catches up to that of the wt, as accumulated virions allow for adequate neuraminidase activity. Another characteristic of this variant is that it causes more widespread fusion in cell culture than does the wt, since lower neuraminidase activity leaves more receptors available on adjacent cells to interact with HN. It was of great interest, therefore, to determine whether this virus causes less disease due to the initial growth lag or whether its higher fusogenicity, seen in cell culture, results in more severe disease in the lung.

It has been proposed that for influenza virus, the neuraminidase activity may be important for removing respiratory tract mucin sialic acids, allowing the virus to reach its target cells (1). While research using a neuraminidase-deficient influenza virus mutant shows that neuraminidase function is not absolutely required for replication in the respiratory tract of mice (10), it is possible that such a mechanism contributes to pathogenesis. There is a precedent for decreased virulence as a result of decreased neuraminidase activity for influenza viruses in animals (16, 26). It was therefore possible that C-28 might be less virulent than wt HPF3. In addition to the above-mentioned possibility, the delay in release of C-28 virus progeny from the host cell surface might prevent the virus from spreading to new cells in the lung, thus preventing severe disease. However, our findings here demonstrate that C-28, with an HN mutation causing deficiency of neuraminidase, is associated with enhanced pathology.

The possibility that the balance between receptor-binding and neuraminidase activities might be critical to the life cycle of HPF3 has been suggested by several lines of evidence. It was previously shown (13) that the level of neuraminidase determines whether the outcome of an HPF3 infection in culture will be acute infection with cell fusion or persistent infection without cell fusion. Other studies have shown that under the selective pressure of exogenous neuraminidase present in an infected cell culture, which serves to remove a portion of the available sialic acid receptors, two distinct types of viral variants emerge: (i) variants with decreased neuraminidase (7), and (ii) variants with increased receptor binding avidity (12, 14). The fact that both types of variants emerged under the same selective pressure of receptor scarcity suggests that a change in either function could compensate for receptor scarcity.

Examination of a temperature-sensitive Newcastle disease virus (NDV) variant and two sequential revertant viruses revealed that alterations in neuraminidase can compensate for alterations in binding (24, 25). The original NDV variant, with an amino acid substitution at position 129, was deficient in binding erythrocytes; a second mutation, at position 175, reduced neuraminidase activity but restored binding; the third sequential mutation, at position 193, partially restored neuraminidase activity. This sequential evolution suggests that the balance between the two activities is the determinant of selective advantage and survival. While the recently determined crystal structure of NDV (2) suggests that the catalytic and binding functions of HN reside in a single site, other data suggest that it is possible for individual mutations to affect one function without affecting the other (17).

The C-28 infection of cotton rats in this study has illuminated the question of whether HPF3 neuraminidase has a role, direct or indirect, in HPF3 pathogenesis. The results suggest that C-28 HN's deficiency in neuraminidase does not cause marked defects in the replication ability of the virus but does cause more intense disease in the lung. It will be of great interest to test a recently characterized variant that is completely neuraminidase deficient; while C-28 has 40% of wt neuraminidase activity, this variant is completely neuraminidase deficient yet is able to bind and enter cells efficiently (17).

In future studies, we plan to use strategies developed in vitro for interrupting HPF3 infection to block HPF3 in the lung. We have recently identified receptor analog molecules that are effective in blocking virus-cell interaction in cell culture (3, 9), and we hope to test these analogs for antiviral capacity in the cotton rat model. 4-Guanidino-Neu5Ac2en (4-GU-DANA; zanamivir) is a sialic acid transition state analogue designed for the influenza virus neuraminidase catalytic site that possesses antiviral activity at nanomolar concentrations in vitro. It has been shown (3) that 4-GU-DANA inhibits HN-mediated binding of HPF3 to host cell receptors as well as HN's neuraminidase activity. 4-GU-DANA reduced the area of plaques formed by the neuraminidase-deficient variant C-28a (3, 17), confirming that its interference with cell-cell fusion is unrelated to inhibition of neuraminidase activity. Thus, for HPF3, 4-GU-DANA and its analogs have an affinity not only to the neuraminidase active site of HN but also to sites important for receptor binding and cell fusion. We have recently generated a 4-GU-DANA-resistant HPF3 virus variant (ZM1) by serial passage in the presence of 4-GU-DANA (15). ZM1 exhibited a markedly fusogenic plaque morphology and harbored two HN gene mutations, one shared with the fusogenic variant C-0 and a second mutation conferring the 4-GU-DANA-resistant property. The sensitivity to 4-GU-DANA of the neuraminidase activity and the receptor binding potential of ZM1 were greatly reduced relative to those of the wt and C-0. ZM1 also retained infectivity at 15-fold-higher concentrations of 4-GU-DANA than did the wt and C-0. We are presently testing the effect of 4-GU-DANA and analog compounds on HPF3 replication and pathogenesis of lung disease in the cotton rat, by use of wt and fusogenic variants, and are investigating the behavior of this 4-GU-DANA-escape variant ZM1 in the cotton rat.

An interesting finding of this study is that enhanced disease caused by the variant viruses does not correlate with an increased presence of infectious virus. This is reminiscent of data obtained for cotton rats by vaccination with a formalin-inactivated RSV vaccine followed by reinfection with wt RSV (22). Upon reinfection with the virus, viral replication was strikingly reduced in the vaccinated rats; however, lung pathology was enhanced. The vaccine-enhanced disease resulted in alveolitis and interstitial pneumonitis, similar to the lesions seen in the present study, and developed in the face of greatly reduced viral replication. The enhancement of disease by vaccine was abrogated by the use of the adjuvant 3-deacylated monophosphoryl lipid A, which altered the pathologic response to reinfection without affecting the protective effect in terms of viral titer (23). RSV vaccine disease enhancement is thought to be an immunopathologic process, and these studies highlighted the fact that reduction in viral titer does not necessarily benefit the infected host. In the present study, the alveolitis and interstitial pneumonitis developed after infection with the variant viruses, in the face of viral titers that are similar to, or lower than, those of the wt virus. The results thus show that mutations in HN conferring altered fusion properties in cell culture also confer striking differences in the ability of HPF3 to cause extensive disease in the cotton rat lung and that this effect is dissociated from any effect on viral replication.