Co‐infection: when whole can be greater than the sum

Abstract

A better understanding of the complex interplay between pathogens and the host immune system during co‐infection of different pathogens could help to develop more efficient therapies.

It has been well known for decades that severe cases of influenza involve secondary infection by pathogenic bacteria causing pneumonia. Yet, it is only recently with the application of modern molecular techniques that scientists have made progress in understanding the underlying interactions between the various pathogens and the immune system. Even though many questions remain unanswered, there is at least a clearer idea of what investigations are required so as to exploit this knowledge for therapeutic options to better treat infections.

The field has expanded beyond studying the exemplary case of influenza and bacterial co‐infection to other viral diseases, such as dengue fever, that can be exacerbated by opportunistic bacteria. It also includes associations between non‐viral diseases, such as malaria and bacteria, where symptoms can be enhanced or prolonged as a result of co‐infection. And then there is the question of co‐infection by two viruses as sometimes happens with dengue and influenza given that there is a seasonal element to both in some regions where the former is prevalent. Finally, it includes plants that can also experience more severe symptoms in the event of co‐infection by both viruses and bacteria.

If there is one common thread linking all these different cases, it is the insight that the aetiology of co‐infection is more complex than had been thought. The earlier assumption was that it is a relatively straightforward case of opportunism, where a bacterium such as Streptococcus aureus that may normally lie dormant exploits the compromised immunity caused by a primary viral infection to become pathogenic itself. That this is over‐simplistic can be seen from the fact that, while co‐infection often enhances and prolongs symptoms, it can in some cases alleviate them.

Influenza and Streptococcus

The history of this research can be traced back to the great Spanish influenza pandemic at the end of World War One. The “Spanish flu” was easily the most severe pandemic on record, and 95% of the mortality was attributed to bacterial co‐infection – at an even higher proportion among healthy adults 1. Several common strains of bacteria were later found in sputum samples from cadavers, but evidence pointed towards Streptococcus pneumoniae as the principle culprit. This was confirmed by a study of the more recent pandemics of 1957 and 1968 where S. aureus was again prevalent 2. The focal point of co‐infection in these cases is the lung as bacteria such as S. aureus reside in the nasal cavities and can migrate there and become pathogenic.

The “Spanish flu” was easily the most severe pandemic on record and 95% of the mortality was attributed to bacterial co‐infection…

In the case of severe influenza involving co‐infection, the worst symptoms and risk of death occur in two waves. The first is associated with a cytokine storm provoked by the influenza virus: the respiratory infection causes damage to the lungs accentuated by early and sometimes excessive infiltration of immune cells. These include mononuclear phagocytes that kill virus‐infected epithelial cells of the alveolar by releasing a lytic molecule called TRAIL 3. Macrophages then play a key role in mediating inflammation by releasing large quantities of pro‐inflammatory cytokines after they encounter necrotic cells and other debris. While necessary to generate the immune response, the release of cytokines can get out of hand and cause severe symptoms and potentially death.

In the absence of secondary bacterial infection, influenza symptoms then normally subside and the patient recovers. But as levels of virus decline, the immune system is dampened down into an anti‐inflammatory state to mitigate the cytokine damage. However, there are still infected and damaged cells that present a target for bacteria already present in the area before they are consumed by phagocytosis.

Role of the immune response

But there was clearly a lot more to it given that co‐infection does not just produce two separate waves of symptoms, but in many cases results in much more severe infection, while in other cases reduces severity. There is clearly some synergy involved and, until recently, little understanding of how the various immune pathways interacted. One significant advance was the insight that the innate immune response is triggered more strongly by dual infection by influenza and bacteria than either on its own, leading to increased production of interferon‐γ‐induced protein‐10 or IP‐10. This could almost be a marker of co‐infection, particularly in the case of severe pneumonia associated with influenza, which was demonstrated in vitro for co‐infection of human monocyte‐derived macrophages with Influenza virus A/H1N1 and S. pneumoniae 4. The same study showed that endogenous miRNA‐200a‐3p, whose expression was synergistically induced following co‐infection, indirectly regulates IP‐10 expression by suppressing cytokine signalling‐6 (SOCS‐6), a well‐known regulator of the JAK‐STAT signalling pathway.

A natural question was whether this elevated expression of IP‐10 was specific to the combination of influenza and S. pneumoniae or occurs in other co‐infections. The same team therefore conducted a follow‐up study involving co‐infections initiated by the respiratory syncytial virus (RSV) 5. “In this work we have shown that bacterial secondary infection in RSV‐infected human macrophages contributes to alter the timing and extent of innate immune response, and notably IP‐10 expression”, commented Olivier Terrier from the Centre International de Recherche en Infectiologie (CIRI) in Lyon, France, and a lead author on both studies. “These results are quite similar to those we published for the model of influenza/S. pneumoniae co‐infection”. This finding in turn raises questions over the detailed interaction between the pathways involved in bacterial and viral immune defences. “Our hypothesis is that both influenza virus and S. pneumoniae induce the same immune response (as reflected by IP‐10), but via different TLR and associated pathways”, Terrier said.

However, he cautioned that there are still many unanswered questions about the underlying molecular interactions. “The mechanisms underlying the pathogenesis of viral/bacterial co‐infection still remain poorly understood, for several reasons and we are facing two big challenges”, Terrier explained. “First, we need to better understand which the culprits in these mixed infections in the clinics are. When you co‐detect several different pathogens in samples from patients suffering from pneumonia, it is not always easy to be sure which pathogens are really involved in the pathology. In our study published in 2016, in a retrospective analysis of a cohort, we detected several different viral/bacterial associations, with a wide panel of different viral strains and bacterial species. We probably need more data from the clinics to identify the more frequent viral/bacterial associations. Then the other big challenge is to develop a biologically relevant experimental model to study the impact of simple/mixed infections on the host cell”,

Co‐infection of dengue fever

While co‐infection of influenza and pneumonia is now more widely studied, this is less the case with dengue fever, where co‐infection is also associated with severe cases. “Unfortunately, the main conclusion is that to date we are only sure about what we are not sure regarding bacterial co‐infection in dengue”, commented Mattia Trunfio from the University of Turin, Italy. “Although the overall proportion of co‐infections may be as small as 0.2–7%, the absolute number of cases may become significantly worrying considering that dengue proved to be one of the leading emerging pandemic‐prone viruses”.

While co‐infection of influenza and pneumonia are now more widely studied, this is less the case with dengue fever, where co‐infection is also associated with severe cases.

Research to date has been hampered by lack of a common approach to surveillance and analysis of in vivo data, he added. “What we can say is that the clinical course of co‐infections may worsen for dangerous synergism between pathogens and for the delayed diagnosis and treatment”, Trunfio explained. “It is not clear yet how the virus can predispose to bacterial infections, but from mostly in vitro data, we can hypothesize that impairing physical barriers like the epithelial lining, may favour microbial translocation from epithelial surfaces and intestinal lumen into the organism when also antigen‐presenting cells’ functions, macrophages’ activities, Toll‐like receptors expression and the interferon cascade have been altered”.

The next dengue epidemic should present an opportunity to study this obscure feature and address the complications of co‐infection, Trunfio added. “Influenza virus is able to inhibit type I interferons‐mediated responses, to deplete alveolar macrophages, to destroy epithelial lining and to facilitate the expression and the exposure of binding receptors and adherence sites exploited by bacteria to invade the airways”, he said. “Some of these processes may overlap with those caused by dengue virus, but, differently from influenza, most of the data about dengue come from animal or cell culture models, so that it may be premature to state that the physiopathology is similar. Nevertheless, we can also observe some similarities when it comes to synergism and immunopathogenicity: both the viral infections seem to be somehow influenced by bacterial toxins”.

This leads to the question what impact bacterial infections have on viruses during co‐infection, given that the greater focus so far has been the other way round. Yet some people, such as cystic fibrosis patients, suffer from chronic low‐level bacterial infection, which could affect the severity of acute viral infections. In the case of dengue, there is now evidence that some bacteria at least can accentuate disease progression through their interaction with the immune system. “Gram‐negative's Lipopolysaccharide has been proved to prolong and enhance dengue replication”, commented Trunfio.

Another natural question is why two or more virus infections can operate synergistically to accentuate and prolong symptoms. “I was asking myself if there are any data about influenza and dengue co‐infection, since they both seasonally co‐circulate in the tropics, and just last year, a study has been published”, said Trunfio. “Once again, the study has been carried out on an animal model and it is preliminary in nature, but they were able to demonstrate an impressive and lethal synergism between the viruses in impairing host immune responses and revealed the lung as the main targeted organ” 6.

Malaria can elicit severe co‐infection

There is no reason why synergies should not also occur in diseases where the primary or initial pathogen is neither a virus nor a bacterium. Malaria is an obvious candidate given its high prevalence and severity in tropical areas. Indeed, there is evidence that some of the more severe cases involve co‐infection with pathogenic bacteria, even if the mechanisms may be somewhat different. There is also a risk of bacteraemia, especially in children, where pathogenic bacteria invade the bloodstream with associated danger of sepsis.

There is no reason why synergies should not also occur in diseases where the primary or initial pathogen is neither a virus nor a bacterium

There are at least two possible explanations as to why children in particular are at risk of bacteraemia associated with malaria. One is that malaria damages the immature spleen in some children reducing its efficacy as a blood filter and therefore encouraging spread of bacteria. But the bacteria still have to enter the blood stream and an alternative or perhaps complementary hypothesis is that they leak in through the gut owing to increased intestinal permeability. Such increased permeability can be caused by elevated production of mast cells—leucocytes whose production is stimulated by infection – combined with higher histamine production in the plasma. As a result, co‐infection with malaria and non‐typhoidal Salmonella serotypes (NTS) can lead to life‐threatening bacteraemia, while NTS diarrhoea on its own normally clears up without severe symptoms in healthy individuals.

There is additional evidence that malaria encourages development of pathogenic bacteria in the gut before entering the bloodstream, according to Patricia Gomez from the Barcelona Institute for Global Health in Spain. She referred to a 2015 paper, which demonstrated in a mouse model how infection with the malaria parasite Plasmodium yoelii altered the gut microbiome so as to reduce resistance to colonization by NTS 7. “Malaria infection seems to cause intestinal dysbiosis, allowing bacteria that are either harmless or under control to grow in the diminishing presence of good bacteria that normally protect us from their invasion”, Gomez explained, adding that this still has to be shown in humans. “It is totally plausible though, because gut microbiota can change within hours, for instance in response to antibiotics, hence this mechanism of rendering our gut mucosa open to the invasion of local, normally innocuous, bacteria is totally possible during falciparum malaria”.

Co‐infection in plants

While co‐infection research has largely been studied in humans and animal models, it is likely that it occurs across eukaryotic life including plants. Indeed, it has been documented in rice and a few other crops where the phenomenon has implications for agriculture 8. One broad conclusion was that the impact of co‐infection can be negative as well as positive in that it can dampen down the symptoms of one or another pathogen or accentuate them. This study focused on interactions between two of the major rice pathogens in Africa: the Rice yellow mottle virus (RYMV) and the bacterium Xanthomonas oryzae pathovar oryzicola (Xoc). The authors found that while the virus had a positive effect on both bacterial multiplication and symptoms, the bacterium in turn reduced viral multiplication.

It appears that RNA silencing mechanisms are again involved, which could explain why exposure to one pathogen can trigger some resistance against another. The study is being followed up by further attempts to understand the molecular mechanisms involved through sRNA and NGS sequencing, commented Charlotte Tollenaere from the Institute of Research for Development, Marseille, France, and one of the authors. “The direct benefit in agriculture is a deeper understanding of the factors driving infection in agro‐ecosystems”, she said. “In addition, our work on co‐infection has various applications for the deployment of genetic resistance, nowadays the most eco‐friendly approach to control crop diseases. Firstly, pertinent deployment of genetic resistance requires detailed knowledge of spatio‐temporal pattern of considered diseases. Second, studying the mechanisms of co‐infection, such as RNA silencing, is likely to give new insight on molecular interactions between plants and pathogens and consequently important knowledge to develop new resistance. Finally, the study of the impact of co‐infection on pathogen evolution will shed new light on the drivers of pathogen evolution conducive to break plant resistance by incorporating a neglected factor, the other diseases circulating in the agro‐ecosystem”.

Potential interventions

The potential for co‐infection to reduce as well as accentuate symptoms has also been observed in animals and potentially humans, noted Lauren Bakaletz, Director of the Centre for Microbial Pathogenesis at The Research Institute at Nationwide Children's Hospital, Colombus, Ohio, USA. “I can easily envision a scenario where one infectious agent induces, for example, a robust immune response to that particular microbe, however this could then also be instrumental in clearing a second pathogen as a collateral benefit”, she said. ”[…] Jeff Weiser and his laboratory have provided an example of this in a mouse model which when colonized with just S. pneumoniae, a PMN response that does not clear this microorganism occurs. However, if that mouse is co‐colonized with both S. pneumoniae and nontypeable Haemophilus influenzae (both of which are considered human pathogens), the PMN response is amplified and results in clearance of S. pneumonia” 9.

The main implication in the case of influenza is to vaccinate vulnerable people against the seasonal varieties and continue research into more effective anti‐viral therapies.

When it comes to therapeutic implications, it is too early for direct applications, even if there might be potential for using anti‐inflammatory drugs. The main implication in the case of influenza is to vaccinate vulnerable people against the seasonal varieties and continue research into more effective anti‐viral therapies. There can also be a case for prescribing antibiotics to people at high risk of respiratory disease: broad‐spectrum antibiotics are routinely prescribed to malaria patients to give some protection against co‐infection. Yet, developing new therapeutic measures to treat or prevent co‐infection would require more basic research to disentangle the complex relationships between hosts and multiple pathogens and thereby identify molecular markers and targets for intervention.

EMBO Reports (2018) 19: e46601