Research ArticleHost-Pathogen Interactions

Inhibition of Autophagy Ameliorates Acute Lung Injury Caused by Avian Influenza A H5N1 Infection

See allHide authors and affiliations

Science Signaling  21 Feb 2012:
Vol. 5, Issue 212, pp. ra16
DOI: 10.1126/scisignal.2001931

Abstract

The threat of a new influenza pandemic has existed since 1997, when the highly pathogenic H5N1 strain of avian influenza A virus infected humans in Hong Kong and spread across Asia, where it continued to infect poultry and people. The human mortality rate of H5N1 infection is about 60%, whereas that of seasonal H1N1 infection is less than 0.1%. The high mortality rate associated with H5N1 infection is predominantly a result of respiratory failure caused by acute lung injury; however, how viral infection contributes to this disease pathology is unclear. Here, we used electron microscopy to show the accumulation of autophagosomes in H5N1-infected lungs from a human cadaver and mice, as well as in infected A549 human epithelial lung cells. We also showed that H5N1, but not seasonal H1N1, induced autophagic cell death in alveolar epithelial cells through a pathway involving the kinase Akt, the tumor suppressor protein TSC2, and the mammalian target of rapamycin. Additionally, we suggest that the hemagglutinin protein of H5N1 may be responsible for stimulating autophagy. When applied prophylactically, reagents that blocked virus-induced autophagic signaling substantially increased the survival rate of mice and substantially ameliorated the acute lung injury and mortality caused by H5N1 infection. We conclude that the autophagic cell death of alveolar epithelial cells likely plays a crucial role in the high mortality rate of H5N1 infection, and we suggest that autophagy-blocking agents might be useful as prophylactics and therapeutics against infection of humans by the H5N1 virus.

Introduction

The spread of the H5N1 strain of avian influenza A virus in recent years poses an increased risk of a global pandemic. To date, more than 500 people worldwide have been infected, and the death rate is ~60% (http://www.who.int/influenza/human_animal_interface/EN_GIP_20120124CumulativeNumberH5N1cases.pdf). Although vaccines are being developed to prevent infection, it is also essential to understand the molecular mechanisms of the disease to develop treatments for infected individuals. Most human cases of H5N1 infection present as severe pneumonia, and the ultimate cause of death is acute respiratory distress syndrome (ARDS), which is the most severe form of acute lung injury (13). The pathogenic determinants of avian influenza A H5N1 infection are direct viral infection of tissues, the host immunological response, or both (47). We previously found that signaling through innate immune pathways downstream of Toll-like receptor 4 (TLR4) causes acute lung injury when avian influenza A H5N1 triggers oxidative stress (8). However, it is not known how direct viral infection of the lung tissue is involved in the pathogenesis of acute lung injury, despite the suggestion that the destruction of the alveolar epithelial cells, which are the major target cell type for viral replication, may contribute to lung pathogenesis (9). Here, we report that autophagic cell death, an alternative programmed cell death pathway, is triggered in lung epithelial cells upon infection by the H5N1 virus, but not by the seasonal H1N1 virus. We provide evidence that the autophagic death of alveolar epithelial cells plays a critical role in bird flu–induced acute lung failure and its associated high mortality.

Results

Lung tissue from an H5N1-infected human patient and mice as well as H5N1-infected human lung adenocarcinoma cells contain autophagosomes

Accumulation of autophagosomes is the hallmark of autophagy (10). We examined electron microscopy (EM) images of human lung infected with H5N1 and found that the tissue contained accumulated autophagosomes, which were not present in EM images of normal human lung tissue (Fig. 1A).

Fig. 1

Autophagy is induced by H5N1 virus in infected human and mouse lungs. (A) EM images of normal human lung and of lung infected by H5N1 virus. The bar graph indicates the percentage of the area of the cell section that contains autophagosomes. (B) EM images of BALB/c mouse lung infected with seasonal H1N1 virus or with H5N1 virus at the indicated times after infection. A more complete time course can be found in fig. S1A. Arrows indicate autophagosomes. The bar graph indicates the percentage of the area in a cell that is occupied by autophagosomes. Data are from three experiments. Bar graph shows the average area ± SEM of the cell that contains autophagosomes from 50 fields.*P < 0.05; **P < 0.01.

We further examined EM images of samples from mouse lungs infected with live H5N1 or live seasonal H1N1 viruses at different time points, and we identified accumulated autophagosomes in the epithelial cells of mouse pulmonary alveoli in tissues infected with equal viral titers of live H5N1 or live seasonal H1N1 virus (Fig. 1B and fig. S1A). When mice were infected with the H5N1 virus, we observed accumulation of autophagosomes 6 hours after infection, which continued to increase in numbers at later time points (up to 96 hours after infection). We observed cell death and lung edema 24 hours after infection, which continued to increase in extent at later time points (up to 96 hours) (Fig. 1B and fig. S1A). When mice were infected with the H1N1 virus, we observed the accumulation of autophagosomes in the lungs 12 hours after infection, which peaked in intensity at 24 to 48 hours after infection, and disappeared at 72 hours; however, we did not observe any pathology consistent with acute lung injury (Fig. 1B and fig. S1A). To further characterize H5N1-induced autophagy, we used A549 cells, a human lung adenocarcinoma cell line, as a model. We infected A549 cells with live H5N1 virus and we observed a substantial increase in the number of autophagosomes compared to that in non-infected A549 cells 3 hours after infection, with cell death occurring 24 hours after infection (fig. S1B). We also infected A549 cells with live seasonal H1N1 virus, and we observed autophagosome accumulation starting at 6 hours after infection, but we did not observe cell death (fig. S1B). The percentage of A549 cell volume occupied by autophagosomes was higher in H5N1-infected cells than in H1N1-infected cells (fig. S1B).

Live H5N1 virus induces autophagy in mouse lungs and human A549 cells

To further confirm the presence of autophagy, we analyzed microtubule-associated protein 1 light chain 3 (LC3). LC3 puncta are an alternative hallmark of autophagy (11). We found that rapamycin and the H5N1 virus efficiently induced the formation of LC3 puncta in A549 cells (fig. S2, A and B). Upon induction of autophagy, cytosolic LC3-I (which is generated by cleavage of newly synthesized LC3 by Atg4B) is further converted to membrane-bound LC3-II (lipidated LC3) by the addition of phosphatidylethanolamine to the C-terminal glycine residue of LC-I (12, 13). We used Western blotting analysis to detect an increase in the abundance of LC3-II protein in the lysates of mouse lungs and A549 cells infected with live H1N1 or H5N1 viruses (Fig. 2, A and B). However, the relative ratio of LC3-II protein abundance to actin protein abundance in the lysates of cells infected with live H5N1 was greater than that in lysates of H1N1-infected cells (Fig. 2, A and B). Thus, we conclude that the H5N1 virus induced autophagy in mouse pulmonary epithelial cells and human lung A549 cells. Previous studies showed that the H1N1 virus induces the accumulation of autophagosomes and LC3-II in A549 cells (14), which we have confirmed with experiments with live seasonal H1N1 virus (fig. S1B and Fig. 2B). A report demonstrated that seasonal H1N1 causes the accumulation of autophagosomes by blocking their fusion with lysosomes (15). To determine whether the accumulation of autophagosomes and LC3-II resulted from increased macroautophagy or from decreased degradation of autophagosomes, we performed experiments in which the lysosomal protease inhibitors E64d and pepstatin A were added to A549 cells infected with live H5N1 virus (16). We observed an increase in LC3-II protein abundance in lysates of A549 cells that were treated with lysosomal protease inhibitors, which suggested that the accumulation of LC3-II upon infection with either live H1N1 virus or live H5N1 virus resulted from increased macroautophagy (Fig. 2C).

Fig. 2

H5N1 virus induces the formation of LC3 puncta in infected mouse lungs and A549 cells. (A) Western blotting analysis of mouse lung infected with seasonal H1N1 virus or H5N1 virus for 48 hours or treated with vehicle for 48 hours. Blots were analyzed with antibodies against the indicated proteins. Bar graph shows the relative abundance of LC3-II protein (normalized to that of β-actin) from three experiments. (B) Western blotting analysis of A549 cells treated for 36 hours with vehicle control or infected for 36 hours with seasonal H1N1 virus or H5N1 virus. Blots were analyzed with antibodies against the indicated proteins. Bar graph shows the relative abundance of LC3-II protein (normalized to that of β-actin) from three experiments. (C) Western blotting analysis of A549 cells infected with H1N1 or H5N1 alone or infected with H1N1 or H5N1 virus after pretreatment for 1 hour with E64d (10 μg/ml) and pepstatin A (10 μg/ml). Blots were analyzed with antibodies against the indicated proteins. Bar graphs show the relative abundance of LC3-II protein (normalized to that of β-actin) from three experiments. *P < 0.05; **P < 0.01.

Live H5N1 virus induces autophagic cell death in mouse lungs and human A549 cells

Although autophagy can cause cell death, it is also thought to be a survival mechanism for cells under certain conditions (17, 18). To determine whether autophagy was directly involved in H1N1- or H5N1-induced cell death, we used the methylthiazol tetrazolium (MTT) assay to measure the viability of A549 cells under different conditions. Whereas A549 viability was substantially decreased upon infection with live H5N1 virus, cell viability was not affected by the live seasonal H1N1 virus at the same viral load (Fig. 3A). Thus, although autophagy was induced by infection with either live seasonal H1N1 or live H5N1 virus, autophagic death of infected A549 cells was caused only by live H5N1 virus.

Fig. 3

Live H5N1 virus induces autophagic cell death in A549 cells. (A) MTT assay of A549 cells treated with either seasonal H1N1 virus or H5N1 virus. Graph shows the percentage cell viability over time. Data are from six experiments. (B) The progeny viral titers were calculated and expressed as TCID50 per milliliter of supernatant of A549 cells infected with H1N1 or H5N1 virus. Data are from three experiments. (C) MTT assay of cell viability in uninfected control A549 cells and A549 cells infected with H5N1 virus in the absence or presence of 3-MA (3 mM), wortmannin (1 μM), LY294002 (10 μM), Z-VAD (50 μM), or necrostatin-1 (50 μM) for 48 hours. Data are from four experiments. (D) Western blotting analysis of control-treated A549 cells and A549 cells infected for 36 hours with H5N1 virus in the absence or presence of 3-MA (3 mM). Blots were analyzed with antibodies against the indicated proteins. Bar graph shows the relative abundance of LC3-II protein (normalized to that of β-actin) from three experiments. (E) MTT assay of the viability of A549 cells that were transfected with control siRNA or Atg5-specific siRNA and then were left uninfected or were infected with H5N1 virus. Data are from four experiments. (F) Western blotting analysis of the effectiveness of knockdown of Atg5 in the A549 cells shown in (E). Data are from three experiments. (G) Western blotting analysis of the effect of Atg5 knockdown on LC3 protein abundance in A549 cells infected with H5N1 virus. A549 cells were transfected with control siRNA or Atg5-specific siRNA and were infected with H5N1 virus for 24 hours. Cell lysates were then analyzed by Western blotting with antibodies against the indicated proteins. Bar graph shows the relative abundance of LC3-II protein (normalized to that of β-actin) from three experiments. (H) The progeny viral titers were calculated and expressed as TCID50 per milliliter of supernatant of A549 cells infected with the H5N1 virus that had previously been transfected with control siRNA or Atg5-specific siRNA. Data are from three experiments.*P < 0.05; **P < 0.01.

To determine whether this difference in cell pathology could be attributed to differential viral replication, we monitored the viral titers of H1N1 and H5N1 during infection of A549 cells. We found that both viruses replicated to a similar extent during the first 48 hours after infection; however, after 72 hours, H5N1 replicated less efficiently than did H1N1 (Fig. 3B), which excluded the possibility that H5N1-induced cell death and pathology were the consequences of differential viral replication. The viability of cells infected with live H5N1 was partially rescued by treatment with the autophagy inhibitors 3-methyladenine (3-MA) (19), wortmannin, and LY294002 (20), which suggested that H5N1-induced A549 cell death might be mediated through autophagy (Fig. 3C). We also treated H5N1-infected A549 cells with other inhibitors of programmed cell death, and found that the apoptosis inhibitor Z-VAD partially rescued H5N1-induced cell death, but not as efficiently as did the autophagy inhibitors (Fig. 3C). The inhibitor of programmed necrosis, necrostatin-1, did not rescue A549 cells from H5N1-induced death (Fig. 3C). Further Western blotting analysis showed that the abundance of LC3-II protein in A549 cell lysates was lower in cells that were treated with the autophagy inhibitor 3-MA compared to that in control cells (Fig. 3D), suggesting that autophagy in the 3-MA–treated cells was decreased.

The Atg5-Atg12 complex and Atg6 (also known as Beclin1) are key regulators that are required to activate autophagy (13). We found that the decrease in A549 cell viability after infection with the H5N1 virus was rescued in cells treated with short interfering RNAs (siRNAs) against Atg5 or Atg6, but not a control siRNA (Fig. 3, E and F, and fig. S3, A and B). The abundance of LC3-II protein in cell lysates also substantially decreased upon treatment with siRNA against Atg5 or Atg6 compared to that in cells treated with a control siRNA (Fig. 3G and fig. S3C). Together, these data suggest that the death of A549 cells induced by H5N1 virus was mediated, at least partially, through autophagy. Previous studies indicated that autophagy is beneficial in H1N1 virus replication (14, 15). Thus, we tested whether autophagy was involved in H5N1 replication in our system. We found that titers of live H5N1 virus in infected A549 cells treated with siRNA against Atg5 or Atg6 were similar to those in infected cells treated with a control siRNA (Fig. 3H and fig. S3D), which suggested that autophagy induced by live H5N1 virus did not play an essential role in its replication in A549 cells.

Previous reports indicated that autophagy-related genes, such as Atg5, are regulated by inflammatory cytokines (21, 22). To elucidate whether the rescue of cell viability that we observed earlier was a result of differences in cytokine production during Atg5 knockdown, we performed a series of experiments to measure cytokine production by enzyme-linked immunosorbent assay (ELISA) and reverse transcription–polymerase chain reaction (RT-PCR) assays. We found that the amounts of inflammatory cytokines produced by A549 cells after infection by live H5N1 virus at different times were not affected by knockdown of Atg5 (figs. S4A and S5A).

Autophagy induced by live H5N1 virus in human A549 cells depends on signaling through the Akt-TSC2-mTOR pathway

Autophagy can be induced by inhibiting the mammalian target of rapamycin complex 1 (mTORC1) signaling pathway (23, 24). To probe the mechanism by which H5N1 induced autophagy, we analyzed the mTOR pathway in A549 cells. We found that the extent of mTOR phosphorylation in H5N1-infected cells was reduced compared to that in H1N1-infected cells or control cells (Fig. 4A). We also found that the extent of phosphorylation of the protein S6, a downstream substrate of the mTOR pathway, was decreased in cells infected with live H5N1 but not in cells infected with H1N1 (Fig. 4A). The phosphoinositide 3-kinase (PI3K)–Akt–tuberous sclerosis 2 (TSC2) pathway controls mTOR-regulated autophagic signaling (18, 25, 26). To further examine whether the PI3K-Akt-TSC2 pathway was involved in H5N1-induced autophagy, we transfected A549 cells with an siRNA against TSC2 and found that A549 cell survival was rescued (Fig. 4, B and C). Furthermore, LC3 aggregation was markedly decreased in A549 cells treated with TSC2-specific siRNA compared to that in cells treated with control siRNA (fig. S2C). In addition, the abundance of LC3-II protein was substantially reduced in lysates of A549 cells treated with TSC2-specific siRNA compared to that in cells treated with control siRNA (Fig. 4D), which suggested that TSC2 was required for H5N1-induced autophagic cell death.

Fig. 4

The Akt-TSC2-mTOR pathway is involved in autophagic cell death in A549 cells infected with live H5N1 virus. (A) Western blotting analysis of A549 cells treated for 36 hours with vehicle control or infected for 36 hours with seasonal H1N1 virus or H5N1 virus. Blots were analyzed with antibodies against the indicated proteins. Bar graph shows the relative abundance of phosphorylated mTOR (p-mTOR) protein (normalized to that of total mTOR) from three experiments. (B) MTT assay of the viability of A549 cells transfected with control siRNA or TSC2-specific siRNA. Cells were infected with H5N1 virus for 48 hours before being assessed for viability. Data are from four experiments. (C) Western blotting analysis of the effectiveness of TSC2 knockdown in the A549 cells shown in (B). A549 cells were transfected with control siRNA or TSC2-specific siRNA. After 48 hours, cell lysates were analyzed by Western blotting for the indicated proteins. Bar graph shows the relative abundance of TSC2 protein (normalized to that of β-actin) from three experiments. (D) Western blotting analysis of A549 cells that were transfected with either control siRNA or TSC2-specific siRNA, and then 48 hours later, cells were infected with H5N1 virus for 12 or 24 hours. Blots were analyzed with antibodies against the indicated proteins. Bar graphs show the relative abundance of LC3-II protein (normalized to that of β-actin) from three experiments. (E) Western blotting analysis of the effect of viral infection of A549 cells on the extent of phosphorylation of Akt. A549 cells were infected with seasonal H1N1 or H5N1 for 12 hours. Blots were analyzed with antibodies against the indicated proteins. Bar graphs show average and SEM from three experiments. (F) MTT assay of the viability of A549 cells transfected with vector control plasmid, plasmid encoding constitutively active Akt, or plasmid encoding dominant-negative Akt. After 24 hours, cells were infected with H5N1 for 48 hours. Data are from three experiments. (G) Western blotting analysis of the effect of Akt mutants on autophagy in A549 cells infected with H5N1 virus. A549 cells were transfected with control plasmid, plasmid encoding constitutively active Akt, or plasmid encoding dominant-negative Akt. After 24 hours, cells were infected with H5N1 virus for 24 hours. Blots were analyzed with antibodies against the indicated proteins. Bar graphs show the relative abundance of LC3-II protein (normalized to that of β-actin) from three experiments. (H) Model of the proposed signaling pathway involved in H5N1-induced autophagy. *P < 0.05; **P < 0.01.

Consistent with the involvement of the PI3K-Akt-TSC2 pathway, the amount of phosphorylated Akt (p-Akt) was substantially decreased in A549 cells infected with live H5N1 virus compared to that in cells infected with live seasonal H1N1 virus (Fig. 4E). Ehrhardt et al. showed that the abundance of p-Akt in A549 cells was increased during H5N1 infection (27). Unlike most reported studies (including our own), which showed that untreated A549 cells had basal amounts of p-Akt (2833), the basal amount of p-Akt in the A549 cells used by Ehrhardt et al. was almost undetectable (27). We also examined the effects of dominant-negative and constitutively active Akt variants on cell survival upon infection by live H5N1. We found that the survival of H5N1-infected cells was partially rescued by the presence of constitutively active Akt (Fig. 4F), and the abundance of LC3-II protein in these cells was substantially decreased compared to that in cells transfected with control plasmid (Fig. 4G). Together, our results indicate that H5N1 may induce autophagy by targeting the Akt-TSC2-mTOR signaling pathway (Fig. 4H).

Inhibition of autophagy ameliorates acute lung injury in mice caused by infection with live H5N1 virus

Patients infected with H5N1 virus have been diagnosed with acute lung injury, including a particularly severe form referred to as ARDS (1, 3, 34). Mice infected with H5N1 also develop ARDS-like condition (35). We previously established a mouse model of acute lung injury by infection with H5N1 virus to study the pathogenic mechanisms of this virus (8). Our finding that live H5N1 induced autophagic death in human lung adenocarcinoma cells raised the possibility that autophagy might play a critical role in H5N1-induced acute lung injury in vivo. To test this hypothesis, we administered live H5N1 virus or live seasonal H1N1 virus intratracheally to mice. We observed severe lung inflammation in mice infected with live H5N1 virus, as shown by histopathological analysis and the presence of infiltrating leukocytes, but we did not observe lung inflammation in mice infected with live seasonal H1N1 virus (Fig. 5A). The survival rate of mice infected with live H5N1 was markedly low when compared to the 100% survival rate for mice infected with live seasonal H1N1 virus (fig. S6A). Lung edema, as determined by measurement of the ratio of the weight of wet lung to that of dry lung, was increased to a greater extent by live H5N1 virus than by live seasonal H1N1 virus (fig. S6B). We detected changes in lung elastance in the H5N1-induced model of acute lung injury, whereas lung elastance remained unchanged when mice were infected with live seasonal H1N1 virus (fig. S6C). We found that treatment with the autophagy inhibitor 3-MA reduced the extent of inflammation as well as the increase in LC-II abundance that was caused by infection with live H5N1 virus (Fig. 5A and fig. S6D). We also performed in vivo knockdown of the autophagy regulators Atg5 and Atg6 by intratracheal administration to mice of an Atg5- or Atg6-specific siRNA (Fig. 5B and figs. S6, E and F, and S7A). We then administered live H5N1 virus to the mice 24 hours after the siRNA treatment. Lung inflammation caused by live H5N1 virus (as determined by histopathological analysis) in Atg5- or Atg6-knockdown mice was substantially ameliorated when compared to that in mice treated with control siRNA (Fig. 5B and fig. S7A). We speculate that the inhibition of H5N1-induced autophagic cell death of pneumocytes may reduce the extent of leukocyte cell infiltration into the lungs, thus ameliorating lung inflammation.

Fig. 5

Acute lung injury induced by live H5N1 virus in mice is ameliorated by inhibiting autophagy. (A) H&E staining and analysis of the numbers of infiltrating cells in mouse lung infected with H1N1, H5N1, or H5N1 virus in the presence of 3-MA. Images are representative of three mice. Data in the bar graph are from three experiments. (B) H&E staining and analysis of the numbers of infiltrating cells in mouse lung treated with control or Atg5-specific siRNA that were left uninfected or were infected with H5N1 virus. Images are representative of three mice. Data in bar graph are from three experiments. (C) Viral titers expressed as TCID50 per milliliter of mouse lung infected with virus (n = 3 to 5 mice per time point). (D) Effect of 3-MA on the percentage change in lung elastance in mice infected with H5N1 virus for the indicated times (n = 5 to 7 mice). **P < 0.01.

To test whether differences in the viral replication of live H5N1 and H1N1 viruses in vivo might play a role in the different lung pathologies induced by these viruses, we measured the titers of both viruses in mouse lung tissues and found that the replication of live H5N1 virus was similar to or less than that of live H1N1 virus (Fig. 5C). Therefore, the absence of lung pathology upon infection with H1N1 was not a result of differential viral replication. Furthermore, we found that the decrease in lung elastance caused by H5N1 infection was ameliorated by treatment with the autophagy inhibitor 3-MA, further supporting the idea that autophagy was involved in H5N1-induced acute lung injury in vivo (Fig. 5D).

The survival rates of mice that had been intratracheally infected with live H5N1 virus were markedly increased by prophylactic and therapeutic treatment with 3-MA (Fig. 6, A and B). Mouse survival rates were also substantially increased by prophylactic treatment with wortmannin and LY294002, PI3K inhibitors that inhibit autophagy (fig. S8A). Similarly, lung edema caused by the administration of live H5N1 virus was substantially decreased in mice upon prophylactic and therapeutic treatment with the autophagy inhibitor 3-MA (Fig. 6, C and D), suggesting that inhibition of autophagy had a beneficial effect both before and after infection with live H5N1 virus.

Fig. 6

Mouse survival rate and lung edema induced by live H5N1 virus are improved by inhibiting autophagy. (A) Effect of 3-MA on the survival rates of mice subsequently infected with H5N1 virus (n = 14 mice). (B) Survival rates of mice that were first infected with H5N1 virus and then treated with 3-MA (n = 14 to 17 mice). (C) Wet weight–to–dry weight ratios of lungs harvested from the mice shown in (A) (n = 6 to 10 mice). (D) Wet-to-dry weight ratios of lungs harvested from the mice shown in (B) (n = 5 to 8 mice). (E) Survival rates of mice that were treated with Atg5-specific siRNA or control siRNA before being infected with H5N1 virus (n = 8 to 10 mice). (F) Wet-to-dry weight ratios of lungs harvested from the mice shown in (E) (n = 5 to 7 mice). (G) Measurement of the percentage changes in lung elastance in mice treated with Atg5-specific siRNA or control siRNA before being infected with H5N1 virus (n = 5 to 7 mice). (H) Viral titers expressed as TCID50 per milliliter of lung isolated from the mice shown in (G) (n = 3 to 5 mice per time point). *P < 0.05; **P < 0.01.

The survival of Atg5- or Atg6-knockdown mice infected with live H5N1 virus was also substantially improved when compared to that of infected mice that had been treated with control siRNA (Fig. 6E and fig. S7B), whereas the wet–to–dry lung weight ratio was decreased in Atg5- or Atg6-knockdown mice infected with live H5N1 (Fig. 6F and fig. S7C). The decrease in lung elastance caused by infection with H5N1 was ameliorated in Atg5- or Atg6-knockdown mice compared to that in mice treated with control siRNA (Fig. 6G and fig. S7D), which further supported the involvement of autophagy in H5N1-induced acute lung injury.

To determine whether autophagy affected the replication of H5N1 virus in vivo, we measured viral titers in mouse lungs that were treated with Atg5- or Atg6-specific siRNAs. Virus replication in the lungs of Atg5- and Atg6-knockdown mice was similar in extent to that in mice treated with control siRNA (Fig. 6H and fig. S7E). Therefore, autophagy was unlikely to be involved in regulating the replication of H5N1 virus in mouse lungs. A previous report showed that apoptosis is involved in the infection of alveolar epithelial cells with H5N1 virus (36). To test whether other types of programmed cell death, such as apoptosis or necrosis, played roles in H5N1-induced acute lung injury, we treated mouse models of acute lung injury with the caspase inhibitor Z-VAD or the necrosis inhibitor necrostatin-1. Neither inhibitor substantially improved the changes in survival rate, lung edema, or lung elastance that we observed in H5N1-infected mice (fig. S8, B to G). We therefore conclude that autophagic cell death likely plays a crucial role in H5N1-induced acute lung injury and that inhibition of autophagy could be an effective strategy against H5N1 infection, both prophylactically and therapeutically. As an additional control, we infected mice treated with Atg5 siRNA with live H1N1 virus; however, we could not observe any difference between the survival rates of Atg5-knockdown mice and those of untreated mice (fig. S9, A to C).

Our previous studies showed that innate immune responses are involved in H5N1-induced acute lung injury in mice (8). We further tested whether adaptive immune responses were involved by infecting Rag1-deficient mice with live H5N1 virus. Rag1-deficient mice lack mature B cells and mature T cells. Compared to infected wild-type mice, the Rag1-deficient mice infected with live H5N1 demonstrated no difference in survival rate, lung pathology, or lung edema (fig. S10, A to C). The titers of H5N1 virus in Rag1-deficient mouse lungs were similar to those of infected wild-type mice (fig. S10D). Therefore, we concluded that adaptive immune responses were not involved in acute lung injury in mice infected with live H5N1 virus.

The hemagglutinin protein of H5N1 virus may induce autophagy in A549 cells

A final question was how the infection of mice with H5N1 virus triggered autophagic cell death. To explore this, we generated recombinant hemagglutinin (HA) protein from H5N1 virus (H5) and used it to treat A549 cells. The presence of the H5 protein represents a major difference between the H5N1 and the H1N1 viruses. We found that the formation of autophagosomes and LC3 puncta was increased in cells treated with H5 protein compared to that in control cells (fig. S11, A and B). Treatment with H5-specific siRNA rescued the extent of cell death of A549 cells infected with H5N1 virus, and resulted in a reduction in the amount of LC3-II protein compared to that in cells treated with control siRNA (fig. S11, C and D). Together, our results suggest that the HA protein of the H5N1 virus may be responsible for the induction of autophagic cell death in infected cells.

Discussion

Autophagy induced by influenza A viruses, including the H1N1 strain, is thought to be beneficial to viral replication (14). Thus, it was important to clarify whether H5N1-induced autophagic cell death or acute lung injury in mice occurred as a result of increased viral replication. Our data showed that the inhibition of autophagy in A549 cells and mouse lungs did not affect H5N1 viral titers (Figs. 3H and 6H), which suggested that autophagy induced by H5N1 did not substantially affect virus replication in vitro or in vivo. Thus, we conclude that the H5N1 virus induces autophagic cell death and that the improved survival of H5N1-infected animals upon inhibition of autophagy was not a result of an indirect effect on virus replication.

Our results showed that H5N1 induced the death of alveolar epithelial cells, whereas the H1N1 virus failed to do so at the same multiplicity of infection (MOI) (Fig. 3, A and B). Both H1N1 and H5N1 induced autophagy in A549 cells, whereas induction of autophagic cell death in alveolar epithelial cells was specific to H5N1 (Fig. 3A). The mTOR pathway was inhibited by infection with H5N1, but not H1N1 (Fig. 4, A and E). Gene expression profiling of A549 cells infected with H5N1 or H1N1 virus identified the differential expression of genes whose products are involved in the mTOR pathway (37), which is consistent with our result showing that the mTOR pathway is dysregulated by infection with H5N1 virus, but not with H1N1. These results further support our conclusion that dysregulation of the mTOR pathway plays a critical role in the induction of autophagic cell death by live H5N1 virus.

Here, we demonstrated that the autophagic death of alveolar epithelial cells mediated the molecular pathogenesis of acute lung injury that is induced by live H5N1 virus. A similar mechanism of cell death does not seem to occur in response to infection by the live seasonal H1N1 virus, although previous reports demonstrated that H1N1 can induce autophagy in alveolar epithelial cells (15). We also demonstrated that inhibitors of autophagy markedly improved the survival rates of infected animals when applied prophylactically or therapeutically. In a previous study, we reported that polyamidoamine (PAMAM) nanoparticles induce lung injury in mice through autophagic cell death through a similar dysregulation of the Akt-TSC2-mTOR signaling pathway as occurs during infection with H5N1 virus (38, 39), which suggests that the autophagic death of alveolar epithelial cells may be a common mechanism that mediates the molecular pathogenesis of acute lung injury. Our results provide further mechanistic insight into the molecular pathogenesis of the H5N1 virus and reveal clues for the development of new approaches that could prevent and treat infections during a potential pandemic of avian influenza A H5N1.

Materials and Methods

Clinical profile of a patient infected with H5N1 virus

Lung samples from a single H5N1-infected human were used. The nasopharyngeal aspirate contained influenza A virus, and PCR analysis detected expression of the gene encoding the H5 protein. The lung sample obtained upon the death of the patient was preserved in liquid nitrogen in preparation for examination by EM. Uninfected human lung samples were obtained from two patients who underwent lung tumor resection; the edges of the resections containing normal lung tissue served as the normal controls.

Viruses, plasmids, and cells

The influenza viruses used in this study were H1N1 [A/new.Coledonia/20/1999(H1N1)] and H5N1 [A/Jilin/9/2004(H5N1)]. Experiments with live viruses were performed in Biosafety Level 3 facilities under governmental and institutional guidelines. The viruses were propagated by inoculation into 10- to 11-day-old SPF (specific pathogen–free) embryonated fowl eggs through the allantoic route. Hemagglutinating allantoic fluid was collected from the eggs, and the virus was stored at −80°C as live virus or after inactivation by treatment with formaldehyde. The EGFP-LC3 plasmid, which encodes a fusion protein of enhanced green fluorescent protein (EGFP) and LC3, was constructed by K. Kirkegaard (Department of Microbiology and Immunology, Stanford University School of Medicine) and was obtained from Addgene. Plasmids encoding dominant-negative and constitutively active Akt variants were provided by P. J. Coffer (Departments of Immunology and Pediatric Immunology, University Medical Center Utrecht, the Netherlands). The human lung adenocarcinoma A549 cell line was purchased from the American Type Culture Collection and was cultured in Ham’s F-12 (Gibco) medium supplemented with 10% fetal bovine serum (FBS), penicillin (100 U/ml), and streptomycin (100 μg/ml) at 37°C and in 5% CO2. Madin-Darby canine kidney (MDCK) cells were purchased from the Peking Union Medical College Cell Culture Center and were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) supplemented with 10% FBS, penicillin (100 U/ml), and streptomycin (100 μg/ml) at 37°C and in 5% CO2.

Antibodies and reagents

Primary antibodies against mTOR, p-mTOR (Ser2481), total Akt, p-Akt (Ser473), total S6, and p-S6 (Ser235/236) were purchased from Cell Signaling Technology. Antibodies against LC3B and β-actin were purchased from Sigma-Aldrich. Antibodies against TSC2 [tuberin (C-20)], Atg5 [APG5 (N-18)], and Atg6 [BECN1 (H-300)] were purchased from Santa Cruz Biotechnology. Horseradish peroxidase (HRP)–conjugated secondary antibodies and Western blotting luminal reagents were purchased from Santa Cruz Biotechnology. 3-MA, wortmannin, LY294002, E64d, pepstatin A, necrostatin-1, and rapamycin were purchased from Sigma-Aldrich. Z-VAD-FMK was purchased from R&D Systems. All siRNAs, with the exception of the siRNA specific for mouse Atg6, were purchased from Ribo Biotechnology. The siRNAs against mouse Atg6 were purchased from Thermo Fisher Scientific.

EM analysis

A549 cells were infected with H1N1 or H5N1 virus at an MOI of 4 for 3, 6, 12, 24, 36, or 48 hours, or cells were treated with recombinant H5 protein (0.5 μM) for 4 hours. The cells were then fixed with 2.5% glutaraldehyde. Ultrathin sections were stained with uranyl acetate and lead citrate and analyzed by transmission EM (TEM). The total areas of the autophagosome and the whole cell were calculated with Adobe Photoshop CS3 Extended software, and the percentage of the cell occupied by the autophagosome was calculated. Four-week-old BALB/c mice were treated intratracheally with vehicle control or virus (106 TCID50) (50% tissue culture infective dose) and were then killed 6, 12, 24, 48, 72, or 96 hours later. The lungs were treated as described for analysis of cells by TEM.

Analysis of LC3-EGFP aggregates

A549 cells were seeded on coverslips in 24-well plates. One day later, cells were transfected with plasmid encoding LC3-EGFP. Forty-eight hours after transfection, cells were incubated with inactivated H5N1 virus at an MOI of 8 or H5 protein (0.5 μM) for 4 hours at 37°C. EGFP-containing (EGFP+) dots in the cell were counted with a Leica laser-scanning spectrum confocal system linked to a microscope (Leica TCS PS2). The images were captured under the 100× oil objective (Plan Apo 1.4) with the confocal acquisition software LCS (Leica). A cell containing 10 or more EGFP+ dots was defined as an LC-3–positive cell. For the TSC2 knockdown assay, A549 cells were first transfected with 50 nM TSC2-specific siRNA or control siRNA. The siRNA sequences for TSC2 were 5′-CGGCUGAUGUUGUUAAAUAdTdT-3′/3′-dTdTUAUUUAACAACAUCAGCCG-5′. After 24 hours, A549 cells were further transfected with plasmid encoding EGFP-LC3. After an additional 36 hours, the effect of treatment with siRNA was determined by Western blotting analysis. In parallel, cells were treated with inactivated H5N1 virus and were analyzed as described earlier.

Western blotting analysis

A549 cells were treated with virus at the indicated MOI or with an equal volume of vehicle. Cells were collected, lysed, and subjected to Western blotting analysis according to standard protocols. Band intensity was analyzed with Quantity One software. Lung tissues were homogenized in ice-cold lysis buffer [50 mM tris-HCl (pH 7.5), 150 mM NaCl, 1.0% Triton X-100, 20 mM EDTA, 1 mM Na3VO4, 1 mM NaF, and protease inhibitors]. Tissue lysates were resolved by SDS–polyacrylamide gel electrophoresis (SDS-PAGE), and proteins were transferred onto a nitrocellulose filter membrane. Membranes were incubated with the appropriate primary antibodies and then with HRP-conjugated secondary antibodies. Binding of secondary antibody was visualized with the Kodak film exposure detection system, and the film was scanned and analyzed.

Cell viability assays

A549 cells were treated with virus at an MOI of 4 or with an equal volume of vehicle for 48 hours. Cell viability was then determined by MTS [(3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium)] assay (Promega). In the drug rescue assays, 3-MA (final concentration: 3 mM), wortmannin (1 μM), LY294002 (10 μM), Z-VAD (50 μM), or necrostatin-1 (50 μM) was added 1 hour before viral administration. In the siRNA knockdown group, A549 cells were treated as described earlier. The sequence of Atg5-specific siRNA was as follows: 5′-GUGAGAUAUGGUUUGAAUAdTdT-3′/3′-dTdTCACUCUAUACCAAACUUAU-5′. The sequence of Atg6-specific siRNA was as follows: 5′-CUCAGGAGAGGAGCCAUUUdTdT-3′/3′-dTdTGAGUCCUCUCCUCGGUAAA-5′. After 48 hours, the cells were treated with virus, and cell viability was then determined as described earlier. In the Akt rescue group, A549 cells were first transfected with plasmids encoding wild-type, dominant-negative, or constitutively active Akt. After 48 hours, cells were treated with virus. Cell viability was then determined as described earlier.

Measurement of viral titers

A549 cells grown in 12-well plates were treated with control siRNA or Atg5-specific siRNA. H5N1 virus was added to the cell monolayer at an MOI of 4. Cells were then incubated for 3, 6, 12, 24, 36, or 48 hours, and the supernatants of infected cultures were harvested and serially diluted in serum-free medium. Tenfold dilutions were used to inoculate MDCK cells in a 96-well plate, and infected cells were maintained in culture for 72 hours. Virus titers were calculated with the Reed-Muench method and were expressed as TCID50 per milliliter of supernatant. In Atg5 knockdown experiments, mice were treated with siRNA, and then the mice were treated intratracheally with either vehicle control or virus (106 TCID50). Mice were killed at 6, 12, 24, 48, 72, or 96 hours after infection, and lungs were collected and homogenized in serum-free DMEM. Measurement of TCID50 was performed as described earlier.

Real-time quantitative PCR analysis

A549 cells grown in six-well plates were treated with control siRNA or Atg5-specific siRNA. Cells were infected with live H5N1 virus at the indicated time points. Total RNA was isolated with Trizol reagent (Invitrogen). Complementary DNA (cDNA) was synthesized from 1.5 μg of total RNA with the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). PCR amplification assays were performed with the FastStart Universal SYBR Green Master mix with Rox (Roche) on an ABI 7500 Real-Time PCR System (Applied Biosystems). Samples were normalized on the basis of the expression of the gene encoding human glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as a reference. The specific primers used were as follows: interferon-α (IFN-α) forward: 5′-GGCACAAATGAGCAGAAT-3′; IFN-α reverse: 5′-CCAAGCAGCAGATGAATC-3′; IFN-β forward: 5′-TCAAGGACAGGATGAACTT-3′; IFN-β reverse: 5′-GACATTAGCCAGGAGGTT-3′; interleukin-1β (IL-1β) forward: 5′-TCTCTTCAGCCAATCTTCA-3′; IL-1β reverse: 5′-CCACTGTAATAAGCCATCATT-3′; IL-1α forward: 5′-GGAGATGCCTGAGATACC-3′; IL-1α reverse: 5′-TTCTTAGTGCCGTGAGTT-3′; tumor necrosis factor–α (TNF-α) forward: 5′-CCTCTTCTCCTTCCTGATC-3′; TNF-α reverse: 5′-TTGCTACAACATGGGCTA-3′; NOD-like receptor family, pyrin domain containing 3 (NLRP3) forward: 5′-CCGATGATGAGCATTCTG-3′; NLRP3 reverse: 5′-CCTGTCTTGGTAGAGTGT-3′; activating signal cointegrator-2 (ASC-2) forward: 5′-GATGCTCTGTACGGGAAG-3′; ASC-2 reverse: 5′-GCTGGTGTGAAACTGAAG-3′.

Measurement of cytokine release

A549 cells grown in six-well plates were treated with control siRNA or Atg5-specific siRNA. Cells were assessed at the indicated time points, when cell supernatants were collected. The amounts of cytokines in the culture supernatants were measured with the Bio-Plex suspension array system (Bio-Rad). Amounts of IFN-β secreted by cells were determined by ELISA (R&D Systems) according to the manufacturer’s instructions.

Animal handling

Animal experiments were conducted in the animal facility at the Institute of Basic Medical Sciences, Peking Union Medical College (PUMC) and the Institute of Military Veterinary Medicine, Academy of Military Medical Sciences in accordance with governmental and institutional guidelines. BALB/c mice were purchased from the Institute of Laboratory Animal Science, PUMC. B6.129S7-Rag1tm1Mom/J mice and wild-type mice were gifts from the Model Animal Research Center of Nanjing University. The mice were caged in a pathogen-free facility in groups of five or fewer mice and were fed laboratory autoclavable rodent diet ad libitum.

Preparation of lung tissue for histopathological examination

Four-week-old BALB/c mice were treated intratracheally with either vehicle control or virus (106 TCID50). After 5 days, mice were killed, and lungs were fixed in formalin and embedded in paraffin. Ultrathin sections were stained with hematoxylin and eosin (H&E). The number of inflammatory cells was counted per 1000× field. In the 3-MA rescue group, 3-MA (15 mg/kg) was injected intraperitoneally 2 hours or 30 min before the administration of H5N1 virus. In the Atg5 siRNA rescue group, the siRNAs indicated earlier were used to knock down Atg5 expression in mouse lungs.

Analysis of lung elastance

BALB/c mice were intratracheally treated with either vehicle control or virus (10 μg/g) after receiving anesthesia. Elastance was measured by Buxco pulmonary function testing (PFT) every 30 min during spontaneous breathing periods for a total time of 4 hours. In the rescue group, 3-MA (15 mg/kg) was injected intraperitoneally 30 min before administration of the virus. In the Z-VAD-FMK group, a single, intravenous injection of Z-VAD-FMK [0.25 mg in 10% dimethyl sulfoxide (DMSO)] was made 15 min before administration of the virus, which was followed by three intravenous injections of Z-VAD-FMK (0.1 mg each) per hour. Control mice were injected with the same volume of 10% DMSO in sterile phosphate-buffered saline (PBS). In the necrostatin-1 group, necrostatin-1 (10 mg/kg) was injected intraperitoneally 30 min before administration of inactivated H5N1 virus. In the Atg5 siRNA rescue group, the siRNAs indicated earlier were used to knock down Atg5 expression in mouse lungs.

Measurement of wet/dry ratios of mouse lungs

Four-week-old BALB/c mice were intratracheally treated with either vehicle control or virus (106 TCID50) after receiving anesthesia. In the 3-MA prevention group, either 3-MA (15 mg/kg) or an equal volume of vehicle was injected intraperitoneally 2 hours and 30 min before administration of the H5N1 virus. In the 3-MA therapeutic group, either 3-MA (15 mg/kg) or an equal volume of PBS was injected intraperitoneally 1, 3, and 8 hours after the administration of H5N1 virus. In the Z-VAD-FMK group, a single intravenous injection of Z-VAD-FMK (6 mg/kg in 10% DMSO) was given 6 hours after virus administration, which was followed by injections of Z-VAD-FMK (6 mg/kg) every 24 hours. Control mice were injected with the same volume of 10% DMSO in sterile PBS. In the necrostatin-1 group, necrostatin-1 (10 mg/kg) was injected intraperitoneally 6 hours after the administration of the H5N1 virus, which was followed by one injection of necrostatin-1 (10 mg/kg each) every 24 hours. In the Atg5 siRNA rescue group, the siRNAs indicated earlier were used to knock down Atg5 expression in mouse lungs. Mouse lungs were assessed for their wet weight 5 days after the administration of the virus and were dried in a 65°C oven for 24 hours before their dry weight was measured.

Analysis of mouse survival rates

Four-week-old BALB/c mice were treated as described earlier for measurement of wet/dry ratios of lungs. The rates of survival of each group (at least 10 mice per group) were recorded consecutively for 8 days. Experiments were repeated at least three times and were analyzed by Kaplan-Meier survival analysis.

Knockdown of Atg5 and Atg6 in mouse lungs

Four-week-old BALB/c mice were intratracheally treated with either control siRNA or an Atg5-specific siRNA duplex (100 μg in 100 μl) in diethyl pyrocarbonate (DEPC)–treated normal saline. The siRNA sequence for Atg5 is 5′-ACCGGAAACUCAUGGAAUAdTdT-3′/3′-dTdTUGGCCUUUGAGUACCUUAU-5′. All animals were monitored throughout a 24-hour recovery period. Knockdown efficiency was determined by Western blotting analysis. Histopathological examination and determination of mouse lung elastance, lung wet/dry ratio, and survival rate were then determined as described earlier. Knockdown of Atg6 by siRNA was performed as described earlier. The siRNA sequence for Atg6 is 5′-GUACCGACUUGUUCCCUAUdTdT-3′/3′-dTdTCAUGGCUGAACAAGGGAUA-5′.

Statistical analysis

All data are presented as means ± SEM. Measurements at single time points were analyzed by analysis of variance (ANOVA), and, if they demonstrated significance, they were further analyzed by a two-tailed t test. Time courses were analyzed by a repeated measurements (mixed model) ANOVA with Bonferroni post–t tests. Survival data were subjected to Kaplan-Meier survival analysis. All statistical tests were performed with GraphPad Prism 5.0 (GraphPad Software). P < 0.05 indicates statistical significance.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/5/212/ra16/DC1

Fig. S1. Autophagy is induced by H5N1 infection of mouse lungs and A549 cells.

Fig. S2. LC3 puncta occur in H5N1-infected A549 cells and decrease in abundance upon treatment with TSC2-specific siRNA.

Fig. S3. Autophagic death of A549 cells induced by H5N1 infection is substantially rescued by Atg6-specific siRNA.

Fig. S4. The production of proinflammatory cytokines by H5N1 infection is not altered by Atg5-specific siRNA.

Fig. S5. The increased immune response induced by H5N1 infection is not altered by Atg5-specific siRNA.

Fig. S6. Mouse model of acute lung injury induced by infection with live H5N1 virus.

Fig. S7. Acute lung injury in mice induced by live H5N1 virus was ameliorated by Atg6-specific siRNA.

Fig. S8. Acute lung injury in mice induced by live H5N1 virus was not ameliorated by the selective inhibitors of apoptosis or necrosis.

Fig. S9. Mice infected by H1N1 virus and treated with Atg5-specific siRNA do not develop acute lung injury.

Fig. S10. Rag1 knockout mice develop acute lung injury induced by H5N1 virus.

Fig. S11. The HA protein of H5N1 virus may induce autophagy in human A549 cells.

References and Notes

Acknowledgments: We thank Z. Han, W. Xi, H. Wang, R. Lu, X. Chen, Z. Pan, R. Sheng, and X. Liu for technical support; H. Zhang and J. Ma for experiments performed in mouse embryonic fibroblast cells with formulated chicken inactivated virus vaccines; and J. Penninger, L. Li, and H. Pickersgill for helpful discussions and for editing the manuscript. Funding: This work was supported by the National Natural Science Foundation of China (30625013, 30623009, and 81000764), the Ministry of Science and Technology of China (2009CB522105), and the Ministry of Health (2009ZX10004-308). D.L. acknowledges support by the Science and Technology Commission of Shanghai Municipality (07pj14096). Author contributions: Y. Sun, C.L., X.J., Z.Z., H.W., S.R., F.G., W.N., Y.Z., Y.Y., J.T., and C.Z. performed the experiments; Y. Shu and R.G. provided H5N1-infected human lung tissue; H. Liu, P.Y., K.L., and S.W. constructed the H5-expressing cell line and expressed and purified the H5 protein; H. Lu, X.L., and L.T. prepared live virus; J.S. took EM photos of human tissue; X.G. provided Rag1 knockout mice; X.T., Y.Q., and K.-F.X. provided normal human lung tissue; D.L. provided helpful ideas; C.J. and N.J. designed the experiments and analyzed the data; and C.J. wrote the manuscript. Competing interests: C.J., S.R., H. Liu, F.G., H.W., Y. Sun, and C.L. have filed for a patent in China for the use of autophagy inhibitors in the prevention or treatment of acute lung injury. The file number: PCT/CN2010/072331.
View Abstract

Stay Connected to Science Signaling

Navigate This Article