Discussion
Pseudomonas aeruginosa is one of the most important pathogens in the context of CF (2). Infections with P. aeruginosa can be distinguished into two different groups: (i). intermittent infections in which the infection with P. aeruginosa can be cleared by antipseudomonal treatment, (ii). chronic infections in which P. aeruginosa can be isolated from most of the respiratory samples indicating that the respiratory tract is chronically colonized by this bacterium (26). Infections with P. aeruginosa are correlated with an accelerated decline in lung function and with an increased incidence of respiratory exacerbations (27, 28). Viral infections have been linked to exacerbations in other chronic pulmonary diseases and respiratory virus infections in CF patients seem to be associated with higher viral burden and higher morbidity (29). Investigation of the link between P. aeruginosa and respiratory viruses could therefore identify potential new therapeutic and diagnostic strategies.
We infected human bronchial epithelial cells (BEAS2B and human primary bronchial epithelial cells) with RSV or hRV in the presence of P. aeruginosa conditioned medium. We used conditioned medium to analyze soluble, secreted factors from P. aeruginosa that might interfere with sensitivity to virus infection. Preliminary experiments (data not shown) indicate that also infection with live P. aeruginosa results in inhibition of type III IFN activity and thus recapitulates the effects with conditioned medium. We observed that PAO1 was able to suppress the antiviral response of bronchial epithelial cells toward respiratory viruses (RSV and hRV) which subsequently lead to higher virus titers in secondary infected cells. Due to the stronger effects observed with RSV we subsequently focused on RSV modulation, but findings were similar for hRV (Figures S3–S5). Interestingly, P. aeruginosa strain Boston, a control strain frequently used in Pseudomonas research was not able to modulate the antiviral response. Due to genetic heterogeneity, P. aeruginosa can be divided into three phylogroups (25). Using representative strains of each group (PAO1—group 1; PA14—group 2; PA7—group 3) we observed that only the minor group 3 strain PA7 was not able to modulate the antiviral response, whereas PA14 could, a clone for which global spread has been shown (25, 30). These findings indicate that most of the P. aeruginosa strains found in the environment and causing infections in CF patients are able to modulate the antiviral response since most of P. aeruginosa isolates are of group 1 or group 2. Two earlier studies could also demonstrate that P. aeruginosa is able to modulate the antiviral response of epithelial cells (31, 32). In addition, the first study could demonstrate a difference in virus-induced IFN expression between healthy and CF-derived cells. However, we and the second study did not observe differences in IFN induction. In line with our results, a third study also failed to detect a difference of virus-induced IFN production between healthy and CF-derived epithelial cells (33). Therefore, our observations might also be relevant in diseases where chronic P. aeruginosa infections occur and a role of respiratory viruses in pulmonary exacerbations has been established, e.g., COPD (34). We have not analyzed whether such modulatory activities are also realized in other lung pathogens, but so far the here identified mechanisms are only reported for P. aeruginosa. It has been reported that P. aeruginosa PAK1 infection in a mouse model can induce IFNλ which promotes an inflammatory response and has a negative impact on P. aeruginosa defense in vivo (35). Of note, PAK1 compared to PAO1 has a point mutation in LasR which might affect protease activity (36). Moreover, the PAK strain induced IFNβ in airway epithelial cells, but CF epithelial cells showed a reduced response compared to healthy cells (37). However, so far it had not been reported that CF epithelial cells display increased sensitivity toward viral infection (38) which was the main focus of our study.
The antiviral response can be subdivided into two stages. First, viruses get recognized by nucleic acid receptors that drive the expression of type I and III interferons. Subsequently, secreted IFN I/III activates the canonical transcription factor STAT1/STAT2/IRF9, which in a positive feedback loop again drives further IFN I/III expression plus additional antiviral genes like OAS1/2 or MX1 (39). Interestingly, the initial induction of IFNλ mRNA after virus infection was not altered by PAO1-CM, but nevertheless IFNλ protein and signaling was significantly reduced compared to control or Boston-CM treated cells. Of note and in line with the literature, type I IFN was not induced by RSV in bronchial epithelial cells (Figure S2). Further analysis revealed that P. aeruginosa secretes proteases degrading type III IFN and thereby inhibiting the antiviral response. Of note, we observed that also exogenous type I IFN was degraded (Figure S6) indicating that in a physiological setting type I IFN as produced by immune cells would also get inactivated. Moreover, protease activity of various P. aeruginosa CF isolates correlated significantly with the ability to degrade recombinant IFNλ. Most of the secreted proteases are under the control of the quorum sensing regulator LasR and we could demonstrate that the ability of PAO1, Boston or the longitudinal CF isolates to suppress the antiviral response was associated with functional LasR. It is well-known that LasR is subject to mutations in the course of P. aeruginosa infections in CF patients e.g., it was reported that in a CF cohort 22% of the P. aeruginosa strains have an altered LasR sequence (6, 25, 40, 41). The involvement of LasR is further supported by the fact that LasR deleted CF P. aeruginosa isolates were not able to modulate the antiviral response whereas their parental counterpart did. LasR dependent proteases contributing substantially to the virulence of P. aeruginosa are AprA, LasA, LasB, and PrpL (42, 43). A limitation of the study is that LasR complemented mutants could not be used. However, five independent targeted mutants behave exactly the same way and the loss of the ability to degrade IFNλ correlated with a mutated LasR. Using P. aeruginosa PA14 deleted of either of these proteases showed that AprA is mostly responsible for the modulation of the antiviral response. In line with this, comparison of the protein sequence of AprA of PAO1, PA14, and PA7 revealed that PA7/group 3 AprA did not cluster within PA14 or PAO1 (Figure S7). AprA, also known as serralysin or alkaline metalloprotease, is a metalloprotease regulated directly by LasR and has previously been reported to degrade complement, alpha1-proteinase inhibitor, interleukins and interferon gamma (6, 44, 45). It is secreted as an inactive zymogen, which becomes active by the cleavage of a 9-amino acid propeptide either by other proteases (LasA/B) or in an autocatalytic manner. To our knowledge this is the first study showing that AprA is also able to degrade IFNλ thereby modulating the antiviral response of epithelial cells. It is well-known that CF patients produce antibodies against several Pseudomonas antigens including AprA. Moreover, it has been shown that these antibodies are able to block AprA activity (46–49). These antibodies would therefore be able to counteract AprA dependent type III IFN degradation. However, these antibodies need to be present at high titers at the site of infection in the conducting airways. Since high titers are regularly detected only in chronically infected patients neutralizing antibodies are only present when AprA expression is decreased. In addition, anti AprA antibodies are IgG subtypes which get passively secreted in the alveolar space and subsequently transported by the mucocilliary escalator to the airways (50, 51). Considering decreased mucocilliary clearance in CF patients sufficient titers might not be reached in this condition. In line with our results, Bomberger et al. were able to show that CFTR inhibitory factor (CIF), secreted by P. aeruginosa, is able to block presentation of viral antigens on MHC class I of bronchial epithelial cells and recognition by CD8+ cells adding another layer of complexity on how P. aeruginosa is able to modulate the antiviral defense (52).
Chronic infection with P. aeruginosa in CF is subject to a complex adaptation to the CF lung leading to increased biofilm production and a decrease in the expression of various virulence factors, including secreted proteases (26, 53). In line with this, total protease activity of P. aeruginosa isolates derived from CF patients correlated with their potential to degrade IFNλ and, interestingly, the ability to degrade IFNλ was associated with intermittent infection status. At closer analysis, the capacity of IFNλ degradation of P. aeruginosa isolated from chronically infected patients clustered into two groups. Several reasons can be accounted for this observation. First, staging of patients into chronic or intermittent can be challenging for the clinician and is sometimes not entirely correct. Therefore, scientists search for additional biomarkers of chronicity because treatment of the patients is based on this classification (54). Moreover, CF patients could be colonized by several P. aeruginosa strains which do not always display the same phenotype (26, 53). Nevertheless, median IFN degradation activity was significantly decreased in chronic patients, therefore we conclude that chronic patients might have a lower risk of virus infection compared to intermittently infected patients. As discussed before, chronic patients have higher anti-AprA in the serum and this together with decreased protease activity might account for the lower infection rate seen in infected patients.
In order to further investigate a potential link between respiratory viruses and P. aeruginosa infections, we screened respiratory material of CF patients using a multiplex PCR based assay. In line with a similar study, we could detect mainly human rhinovirus and to a much lesser extent RSV, Influenza-A/B, Adenovirus or Parainfluenza Virus (55, 56). Interestingly, CF patients intermittent infected with P. aeruginosa had a higher risk for hRV infection (Odds ratio = 2.374) and displayed higher virus loads in the sputum compared to P. aeruginosa negative or chronically infected patients. However, this observation could just be made if material of the lower respiratory tract (sputum) was analyzed. If also samples from the upper airways (nose, throat) were considered no statistical difference between all groups could be seen. Considering that hRV normally infects the upper respiratory tract, increased detection rates in the lower respiratory airways in P. aeruginosa positive individuals might be an indicator for a higher disease burden of hRV infection, similar to the situation in asthmatic or COPD patients (20).
Pseudomonas aeruginosa infections have been associated with a worse outcome in CF and with an increased exacerbation rate (10, 57). Interestingly pulmonary exacerbation in CF has been associated with respiratory viruses (13, 14, 20, 55), which might indicate an interplay between respiratory viruses and P. aeruginosa infections. This might be more important at early stages of CF when P. aeruginosa infections are not yet chronic. In addition, respiratory virus infections have been associated with the conversion of intermittent infections to chronic infections with P. aeruginosa (22, 23) caused by increased biofilm growth of P. aeruginosa due to increased iron concentrations in the lung (58). Thereby P. aeruginosa would directly benefit from a viral infection. Bacterial superinfections of respiratory viral infections are much better described in the literature than viral superinfections and involvement of interferon λ in this scenario has been suggested as well (59). However, several studies show that gut bacteria are able to foster enteric virus infections (60, 61) even though the underlying mechanisms are different (62, 63). In addition, AprA or another LasR dependent protease might directly modify RSV or hRV virions thereby increasing their infectivity, similar to reoviruses (64).
Taken together we could show that P. aeruginosa is able to suppress the antiviral response of bronchial epithelial cells by the direct degradation of IFNλ. This degradation was dependent on the quorum sensing transcription factor LasR and the protease AprA. In addition, the infection status of CF patients was associated with the potential to degrade IFNλ and with presence of respiratory viruses in sputum. Therefore, we conclude that interfering with the antiviral response might lead to an increased susceptibility of P. aeruginosa infected CF patients for respiratory viruses causing respiratory exacerbations or foster the conversion of intermittent to chronic P. aeruginosa infections.