Pandemic Co-infection: A History of Influenza Pandemics
and Secondary Bacterial Pneumonia
Beginning in 1918, as World War I was coming to a close, influenza
pandemic occurred resulting in an estimated 50 million deaths worldwide
[1-4]. In just a few short years, the pandemic had killed well over double
the number of people who had died due to World War I. Termed the
“Spanish Flu”, this pandemic resulted in excessive mortality well beyond
the expected seasonal influenza and targeted young, otherwise healthy
adults with a swiftly deadly disease course [1,5]. Based on preserved lung
tissue sections and autopsy analyses, 95% of these deaths were attributed
to co-infections during the 1918 flu pandemic [5,6]. Since 1918, three
more influenza pandemics have occurred, two with disproportionate rates
of mortality. The H2N2 “Asian Flu” pandemic of 1957-1958 and the H3N2
“Hong Kong” Flu of 1968 [7]. In 1968, the Hong Kong Flu hit the world
in two waves-the first causing excessive mortality in North America, and
the second wave affecting Europe, Asia and Africa between 1968 and 1970
[8,9]. More recently, in 2009, the triple reassortment H1N1 virus, termed
the “Swine Flu”, had killed roughly 285,400 people worldwide by its
completion in 2010 [2,5]. Throughout all these pandemics, co-infections
continued to play key role in lethality, making it crucial to consider these
bacterial co-pathogens when planning for a pandemic [10,11].
In an extensive review of influenza and bacterial co-infections from the
20th century, several more common pathogens were identified including
Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus spp. (in
particular S. aureus), and other Streptococcus spp. [12]. Beyond the threat
of high rates co-infections in pandemics, bacterial-super infections also
contribute to about 65,000 deaths by seasonal influenza virus infections
every year in the United States [2,12], although the rates of bacterial coinfections
were found to be considerably higher during a pandemic than
during the seasonal influenza period-of those bacterial co-infections, 41%
were identified as S. pneumoniae, followed by 25% Staphylococcus spp.,
16% other Streptococcus spp., and about 13% H. influenza [12]. Despite S.
pneumoniae emerging as the predominant strain in 1918, during the 1957
pandemic, the clinical presentation of the disease shifted to a fulminant
pneumonia with severe pulmonary edema and hemorrhage resulting in
rapid death. This was soon attributed to principal co-infection with S.
aureus [13]. By the following pandemic in the late 1960’s, S. pneumoniae
had again emerged as the predominant bacterial co-pathogen.
S. pneumoniae, also termed pneumococcus, is a gram-positive diplococci
that commonly colonize the upper respiratory tract of 20-50% of
healthy children and 8-30% of healthy adults [14]. Although generally
asymptomatic when colonizing the nasopharynx, pneumococcus is
also the most frequently seen bacterial agent in bacterial meningitis,
otitis media, sepsis and all community-acquired pneumonia [14] and
is correlated with an increase in intensive care unit hospitalizations and
death [2]. Pneumococcal disease is difficult to classify because of the
diverse nature of its various strains and serotypes which affect disease
outcomes, co-infection models and transmission [15]. Pneumococci can
express one of over 90 capsule types which greatly alter their pathogenicity,
and makes development of effective vaccines and therapies difficult [15-
17]. Diagnosis is also quite difficult, as many of the bacterial pathogens
seen in co-infection, S. pneumoniae in particular, regularly colonizes
the nasopharynx [1]. As the predominant co-pathogen in influenza coinfection,
this mini review will focus on the proposed contributors to
the pathogenesis of the synergistic co-infection of S. pneumoniae with
influenza, as well as several therapeutic options being considered at this
time.
The Complexity of Co-infection: Why are Influenza Viruses
and Streptococcus pneumoniae Lethally Synergistic?
Pulmonary epithelial barrier damage
It has been shown that mice exposed to influenza have hyper
inflammatory responses with increased bacterial burdens and decreased
pulmonary clearance of S. pneumoniae following co-infection compared
to controls [18]. Although the exact mechanisms behind the lethal
synergism seen with co-infection remain unclear, numerous causative
pathways and pathology have been researched to establish the connection.
Influenza infection damages the host by causing alveolar epithelial
damage, surfactant disruption and resultant obstruction of small airways
by sloughed cells, mucus and other debris [14,19]. The damage to the
respiratory epithelium leads to exposure of the underlying basement
membrane and progenitor epithelial cells, resulting in an inability of the
respiratory epithelium to repair itself and re-proliferate [20]. As epithelial
damage is worsened, a rise in lethality, likely due to bacteremia, is
appreciated [20,21]. Exposure of the basement membrane and fibrin also
increase bacterial adherence [4]. Pandemic viral infections inflict high
cytotoxicity on the alveolar epithelium, which could possibly contribute
to the increase in proportions of co-infections seen at these times [2,20].
In addition, influenza infection also causes a decrease in mucociliary
clearance and coordination, resulting in failure of removal of bacteria
prior to the adherence to the damaged surfaces in the lung [14].
Receptor exposure and bacterial adherence
The desialylation by influenza viral neuraminidase also participates
in bacterial adherence to epithelial cells. Sialylated mucins act as decoy
receptors for the bacteria [1,3,4,22]. The effects that co-infection has on the
recognition of microbial glycans by lectins enhances this pneumococcal
adhesion, making patients with influenza more susceptible to secondary
pneumonia [22]. Damage of epithelial cells also expose glycanson their
surface, thus enhancing bacterial adherence [22]. A variety of proteins
are altered and displayed on epithelial cells following influenza virus
infections, such as platelet activating factor receptor (PAFr), that promote
bacterial adherence and disease [1,23]. Pneumococci also have a variety
of virulence factors that allow adherence to these newly exposed receptors
on damaged epithelium, laminin and fibrin, including pneumococcal
surface protein A (PsaP) and pneumococcal serine-rich repeat protein
(PsrP) [16]. PsaP is a lipoprotein pneumococcal antigen that aids in
adherence to nasopharyngeal epithelial cells via E-cadherin, while PsrP is
a lung-specific adherin [24].
The innate response: can you have too much of a good thing?
Several studies have highlighted exaggerated immune responses in
contributing to the synergism during bacterial co-infection. Among
innate immune cells, high neutrophil influx has been linked with increased
immunopathology in bacterial super infections following influenza
(Figure 1) [25]. Neutrophils are short lived and terminally differentiated
cells, primarily involved in phagocytic clearance of the bacteria. The
ingested bacteria are destroyed through the generation of potent oxidants
after activation of the NADPH oxidase complex (respiratory burst) or
by lytic enzymes and antimicrobial peptides within the phagolysosome.
After bacterial co-infection, neutrophil numbers become excessive
within hours, but macrophages and dendritic cells do not share the
same disproportionate increase [26]. Myeloperoxidase measurements do
not increase at the same rate as the neutrophil quantity, suggesting that
these rapidly recruited neutrophils will not have the same antibacterial
function that the initial responders did [26]. Functional impairment of
neutrophils is seen through several capacities. Phagocytosis has been
shown to be decreased in both neutrophils and macrophages following
influenza infection [25,26] and several pathways to this reduction have
been evaluated including resistance to phagocytic granule components
[27], and the down regulation of the MARCO receptor due to interferon
production [4,28,29]. Neutrophils and macrophages also have a marked
decrease in reactive oxygen species following co-infection [29]. These cells
can kill pathogens through oxidative burst, which creates toxic reactive
oxygen species through NADPH oxidase complex or myeloperoxidase.
Gram positive bacteria such as S. pneumoniae can have a bacterial
superoxide dismutase that can protect the pathogen from these toxic
species [27].
Figure 1: Neutrophils are key players in co-infection pathogenesis
(A) Influenza damages airway epithelium and exposes receptors priming for bacterial adherence; S. pneumoniae adheres to damaged epithelium
and is able to migrate through pulmonary epithelium. (B) Sentinel cells detect pathogens and damaged cells and recruit neutrophils through a
chemotactic gradient for phagocytosis and bacterial killing; Neutrophils contribute to immunopathology through a variety of mechanisms as illustrated.
(C) Worsened epithelial and endothelial damage due to coinfection results in bacteremia.
Neutrophils can potentially cause worsened inflammatory disease
through the release of neutrophil extracellular traps (NETs). We have
previously shown that excessive neutrophils and NETs contribute to
alveolar-capillary damage after influenza challenge in mice. NETs
formation is dependent on redox enzyme activities [30]. NETs were
first identified as a process of cell death that released DNA, histones and
granular proteins such as elastase and myeloperoxidase to entrap and kill
pathogens [31]. Since the initial identification of NETs, they have also
been shown to be detrimental to the host-particularly through histones
which induce endothelial and epithelial cell damage and worsened disease
[32]. Further, using pneumococcal super infection following influenza,
an extensive accumulation of NETs was recognized, especially in the
damaged areas of the lungs, indicating their potential role in tissue injury.
Moreover, NETs released during pneumococcal super infection did not
show any bactericidal or fungicidal activities [33,34]. Our recent studies
have shown that NETs generation is dependent on the pneumococcal
capsule thickness and varies with the different serotype infections. The
increase in thickness of the capsule results in enhanced tissue damage and
lung pathology [17]. NETs have been identified in various inflammatory
disease models other than pneumococcal pulmonary co-infection such
as co-infection of otitis media and sepsis [35,36]. Although the complete
pathway for NETs induction has yet to be discovered, S. pneumoniae has
been shown to induce NETs through an enzyme called α-enolasae
[37]. Paradoxically, a pneumococcal endonuclease, EndA, has been
identified as an important virulence factor through its ability to
degrade NETs and diminish their bactericidal response [38]. As with
many other areas of the complex pathogenesis of co-infection, it appears
that NETs too must be balanced between positive effects and those that are
detrimental to the host.
Apoptosis of various cell types also appears to be affected by bacterial
co-infection after influenza. Monocytes express a TNF-related apoptosisinducing
ligand (TRAIL) that can be blocked through CCR2 blockage and
result in decreased bacterial load and protection if administered prior to
co-infection [39]. In vitro, influenza virus has been shown to accelerate
neutrophil apoptosis by enhancing Fas expression and activating caspase,
decreasing neutrophil survival [40]. The significant neutrophil influx
triggered by various viral and bacterial toxins such as PB1-F2 in a coinfection
result in a cytokine storm and can lead to a severely damaging
hyper inflammatory response which can be seen histopathologically as
excessive neutrophilia, sloughing epithelium, hemorrhage, obstructed
airways, pleuritic and large areas of lung consolidation [26].
Toll-like receptors and their contribution to immunopathology
and interferon signaling
Toll-like receptors are an important part of the innate immune
response and recognize conserved patterns in a variety of pathogens.
Upon recognition, these receptors trigger a series of events resulting in
activation of the innate immune response through production of various
pro-inflammatory chemokines, cytokines, interferons and recruitment of
those innate responders such as the neutrophils and macrophages [41]. In
particular, these TLRs can recognize cellular wall components of grampositive
organisms, such as those in S. pneumoniae [42]. Influenza induces
expression of toll-like receptors, such as TLR3 which acts to recognize RNA
and DNA of pathogens after phagocytosis, and this not only sensitizes cells
to secondary infection with pneumococcal pneumonia, but also decreases
bacterial clearance and increases type I interferons, which have been
shown to negatively affect survival in a murine model [43,44]. In addition
to impairment of phagocytosis, production of interferons after recognition
of pathogens by TLRs plays a large role in pathogenesis of co-infection
as well. Type I and II interferons are produced following recognition of
viral nucleic acids by toll-like receptors (TLRs) [1]. The induction of type
I interferon during a primary nonlethal influenza infection was shown
to be sufficient to promote lethality with co-infection of S. pneumoniae
[45]. In addition, mice deficient in type I interferon receptor signaling
has improved survival and bacterial clearance [46]. One mechanism by
which type I interferon release in response to influenza infection results
in worsened bacterial super infection is through the suppression of γδ T
cell production of interleukin-17 (IL-17) [45]. γδ T cells in the lung act
as specialized innate responders and normally produce the majority of
IL-17 in response to a variety of viral and bacterial infections [45,47,48]
which can suppress the effects of bacterial super infection. If type I
interferon signaling is up regulated and IL-17 production suppressed
through decreased γδ T cell function, bacterial colonization in the lungs
is increased causing in deteriorated pathology and disease [45]. With
interferon signaling increase, an impaired production of the neutrophil
attractants CXCL1 and CXCL2 was noted following co-infection. This may
explain some of the impaired neutrophil response to the early phase of coinfection
[46]. Pneumolysin, a cytolytic toxin of S. pneumoniae, induces
substantial inflammation through activation of TLR4 [49]. TLR2 is also
an important mediator of the damage associated with pneumococcal
pneumonia [50]. As discussed, the innate immune response is necessary
early in the disease course, but can result in worsened pathology if the
response remains elevated for too long. Identifying the pathways most
involved in this synergism and filling in the gaps with the pathology of the
disease will not only improve our general knowledge in all co-infections,
but, more importantly help identify therapeutic targets to improve clinical
outcome in those affected.
Current Prospective Therapeutics and the Efficacy of
Combination Therapies
Antibiotics and combination therapies
Due to the complex nature of co-infection, a wide variety of therapeutic
options and combinations of therapy are being evaluated for efficacy in a
dual infection model of influenza A virus with subsequent pneumococcal
infection. Combination therapies suggest the best results at this time, with
one element of the combination being antibiotic therapy. Several classes
of antibiotics have been evaluated. Although β-lactams were initially
considered a mainstay of treatment for pneumococcal pneumonia, it has
been shown well over the last decade that standalone therapies are no
longer ideal and that combinations with macrolides and fluoroquinolones
are more effective, especially in light of emerging antibiotic resistance [51-
53]. Macrolides such as azithromycin and clarithromycin are bacteriostatic
and work by binding the 50S ribosomal subunit, thereby inhibiting protein
synthesis. In addition to their antimicrobial effects, macrolides also have
an immunomodulatory effect, which poses an additional benefit in
combatting superinfections. Azithromycin in particular has been shown
to improve survival in a mouse model of influenza and pneumococcal
dual infection with almost double the survival rate than ampicillin
(92% versus 56%) as well as improved outcomes over clindamycin [54].
Combination ampicillin and azithromycin for treatment of pneumococcal
pneumonia not only decreases lung inflammation, but also decreases
pulmonary vascular permeability and increases bacterial clearance,
limiting the chances of septicemia [55]. A lower number of inflammatory
cells and proinflammatory cytokines are seen with macrolide treatment
than standalone β-lactams as well as less severe lung histopathology-as
this antibiotic is bacteriostatic, the reduction in an otherwise exacerbated
inflammatory response seen with β-lactam therapy may be due to
lessening in bacterial lysis [50,54]. Another study comparing the effects of
moxifloxacin, a bactericidal drug, with azithromycin in a murine model of
acute bacterial rhinosinusitis supports this as the azithromycin treatment
resulted in rapid bacterial clearance and reduced inflammation compared
with the relatively limited effect of moxifloxacin [56]. Further evaluation
of the potential negative effects of azithromycin in human disease is still
needed, but a 2015 study evaluating cardiotoxicity of azithromycin in
community-acquired pneumonia (CAP) showed that the QT prolongation
suggested to be an adverse effect of therapy was not associated with
treatment, but instead with the disease of pneumonia, regardless of the
therapy administered [57].
Anti-inflammatories
The use of corticosteroids in treatment of bacterial infections is always
a hot topic and one heavily debated. On the one hand, some argue that
the use of an immune inhibitor in combination with an antibiotic to
reduce the bacterial burden can more effectively control the exaggerated
inflammatory response seen in co-infection and that the use of steroids
should improve survival rates. In a murine model, this seems to hold
true-a susceptible murine model for the 2009 H1N1 pandemic showed
that dexamethasone significantly improved survival rate and acute lung
injury [58]. A reduction in the proinflammatory cytokine storm, and
improved clinical outcomes was associated with combination treatment
of dexamethasone and azithromycin in mice [26]. However, what is most
concerning with corticosteroids was highlighted in a retrospective cohort
study from 2011 in which the early use of glucocorticoids was significantly
linked with the development of more severe disease versus patients who
did not receive the drug in pandemic H1N1 [59]. The in vivo benefits in
human disease, particularly in a pandemic setting, are clearly still up for debate.
Toll-like receptor agonists and antagonists are a relatively new area
showing promise as a potential combination therapeutic for pneumococcal
co-infection. Special attention has been given to TLR2, which has been
shown to mediate the extensive tissue damage, lung necrosis and mortality
seen after bactericidal treatment of pneumococcal pneumonia in a murine
co-infection model [50]. This mediation was independent of TLR4 or the
pneumococcal virulence factor, pneumolysin. TLR2 also plays a role in
transmission of disease, likely with a multitude of other factors-when
a TLR2 agonist (Pam3Cys) was administered in a murine model of coinfection,
contact transmission was diminished as well as inflammation
and bacterial shedding [41]. A TLR2 agonist was again seen to reduce the
severity of pneumococcal infection post-influenza in a murine model by
decreasing bacterial loads and pro-inflammatory cytokines, subsequently
leading to decreased vascular permeability and reduced bacteremia [60].
Macrophage-activating lipopeptide 2 (MALP-2) is a TLR2/6 agonist
that, when administered prior to pneumococcal co-infection, increases
proinflammatory cytokine and chemokine release and enhances
neutrophil recruitment without creating excessive inflammation, so also
reduces bacterial loads and improves survival [61]. Like TLR2 agonists,
TLR5, or flagellin, agonists also act as immunostimulants. Given in
combination with an antibiotic, flagellin will decrease bacterial load and
boost antibiotic activity by stimulating CXCL1 to recruit neutrophils and
reduce bacteremia [62]. TLR3 also participates in the immunostimulatory
response when stimulated by pneumococcal RNA. TLR3 acts through
TRIF to secrete IL-12. In a co-infection, influenza virus up regulates
TLR3 in dendritic cells, which helps prime the cells for recognition of
pneumococcal disease [43]. In another study, a TLR4 agonist, UT12,
showed promise in improving clinical outcome and disease in a murine
coinfection model after hastening the macrophage recruitment response
[63]. Modulating TLRs is an interesting approach to understanding the
pathogenesis of co-infection and, with further evaluation, may provide
some promising combination therapies to attempt. The timing of therapy
and its clinical relevance should still be carefully considered, as this
therapy is effective when administered after influenza infection, but prior
to secondary infection.
The role of γδ T cells in interferon signaling and IL-17 production
is also being explored as a therapeutic for bacterial super infections.
Since super infected mice inhibit IL-17, resulting in worsened bacterial
replication and disease, the administration of recombinant IL-17 in these
mice has improved bacterial clearance indicating that induction of IL-17
remains a potential novel therapy [45]. In a recent study, recombinant IL-
17F was administered just prior to S. pneumoniae infection in a murine
model and the therapy resulted in decreased bacterial colonization in the
lungs [64]. In general, modulation of IFN-I signaling, IL-17 production
and the function of γδ T cells all remain intriguing areas of study for
treatment of dual infections.
Other potential therapeutics
Multiple other therapies are being evaluated as well. Anti-virals
are a mainstay of treatment and many are looking for alternatives
to oseltamivir. Peramivir is a neuraminidase inhibitor that reduced
mortality in co-infected mice better than oseltamivir by inhibiting viral
replication resulting in improved bacterial clearance and survival [65].
Although oseltamivir has shown effectiveness to both viral and bacterial
neuraminidase, peramivir only seems to inhibit viral neuraminidase
[65,66]. Another neuraminidase inhibiting compound, artocarpin, was
shown to have a bactericidal effect in vitro, reducing pneumococcal
viability by a factor of over 1000, and reduced biofilm formation [66].
Several agents to reduce vascular leakage have also been evaluated with
varying effectiveness including Slit2N, vasculotide, atrial natriuretic
peptide, S1P, activated protein C, and doxycycline [21,67]. Mathieu, et al.
[68] has started evaluating the use of nanoparticles carrying a plant virus
coat protein and ssRNA that trigger a strong innate immune response
in the lung during a co-infection. Vaccinations are also a key area of
research, especially when considering the effect these vaccinations may
have in pandemic preparedness. Pneumococcal capsular polysaccharide
conjugate vaccines have been shown to be very effective (100%) against
otherwise lethal pneumococcal disease, but in co-infection, the results are
not as promising with less than 40% survival with vaccination in a murine
model [69]. The value of the current vaccine is evident already though,
with the vaccine being 84-94% efficacious against the serotypes included
and reducing the severity of disease and risk for hospitalization in those
affected [4]. In the U.S. alone, we have seen a 39% reduction in clinical
pneumonia in children since the vaccine has been introduced [70].
Imagine how effective the current vaccine will be once it’s more available
in developing countries.
Conclusions
Co-infection of S. pneumoniae with influenza promises to be a relevant
disease for many years to come. Despite the many recent advances in our
knowledge base regarding the disease, the complexity of pathogenesis
implies that an effective “shotgun” approach to therapy is doubtful and a
fine-tuned combination of antimicrobial agents with immunomodulators
is likely to be more effective when treating the disease. Because of the
expansive diversity in both influenza viral strains and pneumococcal
disease and their ever-changing patterns of resistance and survival,
therapy effective for one combination may not consistently work for
all. This review touches on a few approaches to consider in therapeutic
design, but continued discovery will be needed to better prepare for the
next pandemic.
Acknowledgements
This work was supported by the National Institute of General Medical
Sciences of the National Institutes of Health under Award Number
P20GM103648. We are thankful to Mr. Benton Rudd for his assistance in
the preparation of the Figure.
