Differential adaptation of microbial pathogens to airways of patients with cystic fibrosis and chronic obstructive pulmonary disease

Abstract

Cystic fibrosis (CF), the most common autosomal recessive disorder in Caucasians, and chronic obstructive pulmonary disease (COPD), a disease of adults, are characterized by chronic lung inflammation, airflow obstruction and extensive tissue remodelling, which have a major impact on patients' morbidity and mortality. Airway inflammation is stimulated in CF by chronic bacterial infections and in COPD by environmental stimuli, particularly from smoking. Pseudomonas aeruginosa is the major bacterial pathogen in CF, while in COPD, Haemophilus influenzae is most frequently observed. Molecular studies indicate that during chronic pulmonary infection, P. aeruginosa clones genotypically and phenotypically adapt to the CF niche, resulting in a highly diverse bacterial community that is difficult to eradicate therapeutically. Pseudomonas aeruginosa clones from COPD patients remain within the airways only for limited time periods, do not adapt and are easily eradicated. However, in a subgroup of severely ill COPD patients, P. aeruginosa clones similar to those in CF persist. In this review, we will discuss the pathophysiology of lung disease in CF and COPD, the complex genotypic and phenotypic adaptation processes of the opportunistic bacterial pathogens and novel treatment options.

Pseudomonas aeruginosa, Haemophilus influenzae, Staphylococcus aureus, Burkholderia ssp., mutator, adaptive radiation

Introduction

Cystic fibrosis (CF) is the most common autosomal recessive disorder in Caucasians, affecting ∼1: 2500 children, with a carrier frequency of 1: 25. In the CF lung, disease develops as a consequence of mutations in the CF transmembrane conductance regulator (CFTR) gene, which encodes a membrane-bound cAMP-regulated chloride channel (Kerem et al., 1989; Riordan et al., 1989; Rommens et al., 1989). The primary structure of CFTR indicates that it belongs to a family of transmembrane proteins called ATP-binding cassette transporters, which form a large family of proteins responsible for the translocation of a variety of compounds across membranes of both prokaryotes and eukaryotes. CFTR is composed of five domains: two membrane-spanning domains, two nucleotide-binding domains (NBDs) and a regulatory domain. Over 1500 mutations and sequence variants have been described to date and reported to the Cystic Fibrosis Genetic Analysis Consortium. Most of these mutations are rare and only four mutations occur at a frequency of >1%. The most frequent mutation, F508del, occurs in the DNA sequence that codes for the NBD1 (Döring & Ratjen, 2005). Abnormal transport of chloride and sodium ions caused by CFTR mutations affects water movement across epithelia, leading to pathophysiological consequences in various organs including the respiratory, gastrointestinal and reproductive tract, the pancreas and the liver. The CF phenotype is rather heterogeneous due to many different mutations in CFTR and the influence of modifier genes (Davies, 2007). Chronic respiratory infections caused by several opportunistic bacterial pathogens are recognized to have the largest impact on morbidity and mortality in CF patients (Høiby & Frederiksen, 2000). Bacterial infections stimulate inflammatory defence mechanisms, leading to tissue destruction and extensive tissue remodelling (Döring & Ratjen, 2007). The resulting emphysema and fibrosis mainly determine the reduced life expectancy in individuals with CF. Once established, these infections are difficult to treat with antibiotics and the pathogens are rarely eradicated (Döring et al., 2004).

Unlike CF, chronic obstructive pulmonary disease (COPD), a chronic, progressive disorder of lung parenchyma and airways, is exclusively a disease of adults, especially of individuals over the age of 60 years. COPD is the fourth leading cause of death globally, accounting for approximately 2.7 million deaths in 2003 (Lopez et al., 2006). COPD is defined by airflow obstruction that is not fully reversible, in contrast to asthma, which is associated with reversible airflow obstruction (Celli & MacNee, 2004). The degree of reduction in forced expiratory volume in 1 s (FEV₁) is used to classify COPD into four stages (mild, moderate, severe and very severe). Although genetic factors influence COPD (Wood & Stockley, 2006; Silverman et al., 2009; Smolonska et al., 2009), these influences are complex and mostly unknown. COPD results from responses to environmental stimuli, particularly from smoking, in genetically susceptible individuals (Celli & MacNee, 2004; Wood & Stockley, 2006). Bacterial pathogens also play an important role in the course and pathogenesis of COPD (Sethi & Murphy, 2008).

Many similarities exist between CF and COPD. In both diseases, exuberant inflammation causes airflow obstruction, tissue destruction and extensive tissue remodelling, leading to emphysema and fibrosis. Thus, inflammation plays an important role in the morbidity and mortality of patients. However, CF and COPD differ with regard to infection. While the role of chronic bacterial infections, particularly caused by Pseudomonas aeruginosa, is well established in CF (Høiby & Frederiksen, 2000), the impact of these infections in patients with COPD is less clear (Murphy et al., 2008). Furthermore, pathogens such as Moraxella catarrhalis and Streptococcus pneumoniae are frequently observed in patients with COPD (Sethi & Murphy, 2001), and yet are seldom present in the airways of CF patients. Because of improved symptomatic treatment strategies and antibiotic therapy, the prognosis of CF individuals has improved considerably and many children now reach adulthood. While antibiotic therapy is beneficial in treating acute exacerbations of COPD (Anthonisen et al., 2002; Ram et al., 2006), this therapy has not been demonstrated to affect the course of the disease. In this review, we will discuss the pathophysiology of lung disease in CF and COPD, bacterial colonization of airways and the complex genotypic and phenotypic adaptation processes of the opportunistic bacterial pathogens in these diseases and novel treatment options.

Pathophysiology of CF lung disease

A variety of innate immune functions have been reported to be dysregulated in CF, and a number of mechanisms have been put forward to explain the fact that practically every CF patient suffers from chronic bacterial lung infections. The inability of mutated CFTR to effectively secrete chloride from respiratory epithelial cells into the airway surface liquid causes excessive water absorption from the airway surface liquid, leading to impaired mucociliary clearance in individuals with CF (Matsui et al., 1998; Boucher et al., 2007) (Fig. 1). This in turn facilitates the colonization of the viscous mucus layer on the respiratory epithelium with bacteria, rather than the colonization of the epithelial surface per se (Ulrich et al., 1998; Worlitzsch et al., 2002). Furthermore, an abnormal accumulation of ceramide in the lungs of CF mice and in epithelial cells from CF patients has been shown to result in an increased death rate of respiratory epithelial cells and DNA deposits on the respiratory epithelium, which facilitates bacterial adherence (Teichgräber et al., 2008). In addition to these mechanisms, which explain the high incidence of chronic bacterial lung infections in CF, other investigators have demonstrated that alveolar macrophages from CF mice exhibit defective killing of internalized P. aeruginosa (Di et al., 2006). After phagocytosis by alveolar macrophages, bacteria enter the phagosome and lysosomes, exhibiting a low pH due to CFTR and vacuolar ATPase, and fuse with the phagosome, forming the phagolysosome. The mature organelle exhibits an acid pH that ensures the final destruction of the ingested bacteria by a variety of proteolytic enzymes. Defective CFTR-dependent acidification of lysosomal pH (Barasch et al., 1991) has been linked to significant survival of P. aeruginosa (Di et al., 2006). The contribution of CFTR to the regulation of pH in intracellular organelles has been discussed controversially (Haggie & Verkman, 2007, 2009) due to experimental difficulties to ensure targeting of identical vesicle populations by pH-sensitive fluorescent dyes. Pseudomonas aeruginosa infection may also be facilitated in CF airways directly by defective CFTR, which, in its functional state, binds the pathogen within lipid rafts and removes it from the epithelial surface via internalization (Pier et al., 2000).

Figure 1

Bacterial killing mechanism of innate immunity in the respiratory tract of CF patients and healthy individuals. (a) In healthy individuals, bacteria entering the mucus layer overlaying the respiratory epithelium are effectively removed from the airways by a functional mucociliary clearance system and killed by CAMPs derived from submucosal glands or epithelial cells, by functional neutrophils (PMN) or macrophages (MΦ) or within epithelial cells after uptake via functional CFTR. (b) In individuals with CF, defective CFTR leads to a highly viscous mucus layer that impairs mucociliary clearance, the migration of CAMPs, neutrophils and macrophages towards the bacterial targets, the uptake of bacteria by CFTR itself, and induces a pH shift to alkaline values in epithelial lysosomes. Because of ceramide accumulation, this results in DNA deposits, which in turn serve as adhesion matrices for bacteria. A similar pH shift in the phagolysosomes of macrophages impairs bacterial killing. Ceramide accumulation also triggers cytokine release, which induces further neutrophil influx (adapted from Döring & Gulbins, 2009).

The viscosity of the secretions has several further negative consequences for the pathophysiology of CF lung disease. These secretions are present in submucosal gland ducts of CF patients as a result of mutated CFTR (Engelhardt et al., 1992), and may impair the transport of antimicrobial oligopeptides (Lehrer & Ganz, 2002; Selsted & Quellette, 2003) onto the epithelium (Verkman et al., 2003; Joo et al., 2004) (Fig. 1). They may also negatively affect the migration of neutrophils towards the pathogens within the mucus overlaying the respiratory epithelium (Matsui et al., 2005) (Fig. 1). This may allow bacterial multiplication and profound changes in the bacterial phenotypes that facilitate chronic infection. Furthermore, within the highly viscous mucus, a microaerobic/anaerobic milieu prevails due to oxygen consumption by bacterial pathogens such as P. aeruginosa (Worlitzsch et al., 2002; Hassett et al., 2009) or invading neutrophils (Kolpen et al., 2010). Consequently, the generation of reactive oxygen species (ROS) by neutrophils and other cells is abolished and bacterial killing is impaired (Fig. 2a and b), particularly with regard to pathogens, which are intrinsically resistant to nonoxidative killing, or capable of changing their phenotype towards resistance to nonoxidative killing.

Figure 2

Bacterial sensitivity and resistance to neutrophil-mediated killing in aerobic and anaerobic environments. (a) On the respiratory epithelium of healthy individuals, neutrophils effectively kill the CF-related bacterial pathogens Pseudomonas aeruginosa (red), Staphylococcus aureus (blue) and Burkholderia cepacia (green) by ROS and CAMPs. (b) Neutrophils in the anaerobic milieu in the mucus layer overlaying the respiratory epithelium in individuals with CF are devoid of ROS, leading to intraneutrophil survival of S. aureus (blue), which produces a protective exopolysaccharide coat (red), and B. cepacia (green), which is intrinsically resistant to nonoxidative killing (adapted from Döring & Gulbins, 2009).

Lung inflammation is a key finding in patients with CF (Konstan et al., 1994). While ceramide accumulation has been shown to induce a proinflammatory status in the respiratory tissue of CF mice, which precedes bacterial infection, including an increased production of cytokines and recruitment of macrophages (Teichgräber et al., 2008), inflammation clearly increases considerably in response to persistent bacterial infections. The exaggerated inflammatory response dominated by neutrophils not only facilitates tissue destruction but also induces tissue remodelling (Döring & Gulbins, 2009; Ulrich et al., 2010). An essential role in these pathogenic reactions is related to the imbalance of serine and metalloproteases to their respective inhibitors (Döring, 1999; Ratjen et al., 2002). Particularly, neutrophil elastase has been implicated in the pathophysiology of CF lung disease. The serine protease has been shown to cleave the endogenous serine protease inhibitor α₁-antitrypsin, immunoglobulins, surfactant proteins, complement components and a variety of cell receptors (Döring & Gulbins, 2009). Cleavage of the complement receptor 1 on neutrophils and the activated complement component C3bi on opsonized bacteria creates an ‘opsonin-receptor mismatch’ (Berger et al., 1989; Tosi et al., 1990), which significantly impairs P. aeruginosa killing.

The vicious cycle of persisting bacterial pathogens, encased in biofilms, and decaying neutrophils leads to increasing endobronchial DNA deposition and large mucus plugs, which further obstruct the airways of the patients (Whitchurch et al., 2002; Worlitzsch et al., 2002).

Pathophysiology of COPD

The pathophysiology of COPD is characterized by chronic inflammation of airways as a result of active and passive smoking (Barnes et al., 2000; Celli & MacNee, 2004). The indices of airway inflammation – increased neutrophils, IL-8, TNF-α and neutrophil elastase – are all increased compared with healthy controls (Keatings et al., 1996; Hill et al., 1999a b; Barnes et al., 2003). Chronic colonization of the airways contributes to this airway inflammation (Hill et al., 2000; Parameswaran et al., 2009). However, the degree of airway inflammation does not reach the threshold of patient-perceived symptoms. When airway inflammation crosses this threshold for short episodes, these are called exacerbations (Hill et al., 1999a, b). The course of COPD is punctuated by episodes of acute exacerbations, which are associated with increased airway inflammation compared with baseline levels (Hill et al., 1999a, b; Sethi et al., 2008). Goblet cell hyperplasia and mucus hypersecretion are present in the airways of most COPD patients, especially in those with clinical features of chronic bronchitis (Barnes et al., 2000). Although mucus viscosity is considerably lower in COPD compared with CF, mucociliary clearance of bacterial pathogens may be impaired by the fact that respiratory epithelial cells lose cilia and undergo squamous metaplasia (Hogg & Timens, 2009). Lung inflammation leads to the activation of innate immune cells, particularly macrophages, which have a profound effect on immune cell recruitment into the airways and tissue remodelling (Barnes et al., 2004). In the small airway walls, infiltrating neutrophils and lymphocytes cause airway thickening and remodelling, resulting in reduced airway diameter and airflow (Hogg et al., 2004). Inflammatory cellular infiltration of alveolar walls is associated with the destruction of alveoli and enlargement of air spaces, i.e. emphysema (Cosio et al., 2009). Infiltration especially by CD8+ lymphocytes occurs in pulmonary vessel walls that are thickened due to increased collagen deposition (Cosio et al., 2009). As in CF, chronic lung inflammation is accompanied by an imbalance between proteases and antiproteases in COPD (Stockley et al., 1999). Furthermore, oxidative stress is persistently enhanced (Rahman & MacNee, 1996) as well as a number of other innate immune functions that lead to tissue remodelling (Barnes et al., 2000; Cosio et al., 2009; Hogg & Timens, 2009). Alveolar macrophages from COPD patients have decreased sensitivity to nontypeable Haemophilus influenzae (NTHI) antigens and decreased ability to phagocytose NTHI, compared with normal controls (Berenson et al., 2006a, b). The structural abnormalities in airways, alveoli and pulmonary vasculature vary among COPD patients (Hogg & Timens, 2009).

Bacterial colonization of the CF lung

Chronic bacterial pulmonary infection leading to an irreversible decline in lung function is the main cause of mortality and morbidity in CF patients, with >95% of deaths due to respiratory failure (Dodge et al., 2007). Staphylococcus aureus is the most prevalent opportunistic bacterial pathogen, which chronically infects CF airways in the age group of 0–9 years (55%), while P. aeruginosa is the most prevalent pathogen thereafter (81%) (Høiby & Frederiksen, 2000). At a much lower prevalence, H. influenzae, S. pneumoniae, Escherichia coli, strains of the Burkholderia cepacia complex (Bcc), comprising 10 genomovars, and fast-growing mycobacteria can be present in airway specimens of CF patients (Høiby & Frederiksen, 2000).

In addition, a large number of strict anaerobes have been detected, which may be surprising for an air-filled organ such as the lung. However, the viscous mucus plugs, in which all these bacteria thrive, have been shown to create microaerobic/anaerobic conditions (Worlitzsch et al., 2002), which favour the growth of anaerobes. Anaerobic bacteria have been detected in significant numbers (>10⁵ CFU mL⁻¹) in the lungs of CF patients using classic culturing methods, indicating that a highly diverse bacterial community is present within the CF lung (Brook & Fink, 1983; Jewes & Spencer, 1990; Tunney et al., 2008; Worlitzsch et al., 2009). These include diverse species belonging to the genera Prevotella, Bifidobacterium, Veillonella, Peptostreptococcus and Fusobacterium. Single obligate anaerobic species persisted for up to 11 months in sputum plugs in vivo (Worlitzsch et al., 2009). Whether these bacteria contribute to the pathophysiology of CF lung disease is still an open question. CF patients with and without obligate anaerobes in sputum specimens did not differ in lung function (Worlitzsch et al., 2009).

Culture-independent profiling methodologies (Nocker et al., 2007) such as terminal restriction fragment length polymorphism have considerably advanced our knowledge of the polymicrobial nature of CF lung disease (Rogers et al., 2003, 2004, 2006, 2009; Harris et al., 2007; Bittar et al., 2008; Sibley et al., 2008, 2009; Klepac-Ceraj et al., 2010). For instance, it has been shown by culture-independent methods that the Streptococcus milleri group of viridans streptococci, comprising Streptococcus constellatus, Streptococcus intermedius and Streptococcus anginosus, is present in CF sputum specimens and related to the pathophysiology of CF lung disease (Sibley et al., 2008).

During acute infections, which may resolve after several days, microorganisms reversibly regulate gene expression by responding to environmental signals, and thereby adapt their phenotypes accordingly (Green & Paget, 2004; Gilles-Gonzalez & Gonzalez, 2005). During the course of chronic infections, which may last >10 years, however, mutants can be selected that differ genotypically and phenotypically from the originally infecting strain. Given the large microbiota in the airways of infected CF patients, complex interactions not only occur between microorganisms and the environment but also among members of the community that use primary or secondary metabolites in a multitude of chemical interactions (Duan et al., 2009). A fascinating view is that ‘cheater’ cells that refrain from producing ‘expensive’ metabolites benefit in such communities from those produced by the rest of the population (Keller & Surette, 2006; Duan et al., 2009). The chemical and physical nature of the mucus plugs in CF airways, in which the bacterial pathogens multiply, favours such a microevolution during the course of chronic infection. Interestingly, it has been found that the number of taxa and the phylogenetic diversity and evenness within the microbiota in the airways of CF patients decrease with age, which has been attributed to either repeated antibiotic therapy courses or invasion of P. aeruginosa into the CF airways (Klepac-Ceraj et al., 2010). Microevolution processes have been studied most widely in P. aeruginosa.

Pseudomonas aeruginosa is the most prevalent pathogen

The capacity of P. aeruginosa to cause infections in CF patients and other hosts can be traced back to its considerable adaptability to specific environmental conditions and the production of numerous virulence factors (Lee et al., 2006) (see Box 1). This adaptability is linked to a large number of regulatory genes. For example, in the 6.3 Mb genome of P. aeruginosa PAO1, which contains 5570 predicted ORFs, 521 known or putative regulatory genes have been identified, demonstrating a highly complex gene regulation (Stover et al., 2000). The P. aeruginosa chromosome contains a core genome consisting of the sequences that are common to all strains of the taxon and variable amounts of accessory DNA segments. The accessory genome that can be horizontally transferred among strains represents the flexible gene pool that frequently undergoes acquisition and loss of genetic information and hence plays an important role in the adaptive evolution of bacteria. The flexible gene pool is made up of elements such as bacteriophages, plasmids, insertion elements, transposons, conjugative transposons, integrons and genomic islands. Isolates from CF patients and environment are endowed with a variable repertoire of genomic islands (Klockgether et al., 2007; Wiehlmann et al., 2007). The genes required for pathogenicity in one strain of P. aeruginosa are neither required for nor predictive of virulence in other strains (Lee et al., 2006; Bragonzi et al., 2009), indicating a broad intraclonal diversity of pathogenicity of this organism, yet no difference in virulence. Despite the high overall genome similarity of P. aeruginosa of <0.5% sequence diversity in the core genome, comprising 80% of all genes, differences in phenotypes can be striking even at the intraclonal level of genetically highly related strains (Wehmhöner et al., 2003; Salunkhe et al., 2005a, b).

Table Box 1.

Virulence factors of Pseudomonas aeruginosa and their regulation by quorum sensing

Pseudomonas aeruginosa produces many virulence factors, which can be divided into cell-associated and extracellular virulence factors. Many virulence factors are regulated by the quorum-sensing circuit and other regulatory genes, which have been demonstrated in vitro and in vivo. During the chronic course of lung infections in patients with CF, many mutants have been characterized that are defective in virulence factor expression and quorum sensing (D'Argenio, 2004; Hogardt et al., 2004, 2007; van Delden, 2004; Ramsey & Wozniak, 2005; Lee et al., 2006; Williams & Camara, 2009; Winstanley & Fothergill, 2009). Cell-associated virulence factors Flagella, type IV pili, lipopolysaccharide and three different exopolysaccharides, encoded by the alg, psl and pel clusters, contribute to adhesion of the microorganism to various surfaces. On epithelial cells, flagellar proteins bind to toll-like receptor (TLR)-5 and lipopolysaccharide to TLR-4, leading to the stimulation of innate immune functions and inflammation. The exopolysaccharides have been implicated in biofilm formation. Flagella, lipopolysaccharide and alginate antibodies have been detected in the sera of infected CF patients. Extracellular virulence factors Type I, II and III secretion systems secrete protein toxins and provide contact with eukaryotic cells. Alkaline protease is secreted by the type I secretion system, while the LasB and LasA elastases, protease IV, exotoxin A, phospholipase C and lipase are secreted by the type II secretion system. The type III secretion system injects the exoenzymes S, T, Y and U into eukaryotic cells. Other extracellular virulence factors include the secondary metabolites pyocyanin and other phenazine pigments, siderophores, hydrogen cyanide and rhamnolipids. Most exoenzymes and secondary metabolites have been shown to be cytotoxic in in vitro and in animal models, and antibodies to many of these factors have been detected in the sera of infected CF patients. Quorum-sensing circuit Several virulence factors are regulated in a cell-density-dependent manner by the quorum sensing circuit composed of lasRI, rhlRI and pqs systems. Pseudomonas aeruginosa produces several diffusible signal molecules with autoinducer properties such as N-3-oxo-dodecanoyl homoserine lactone (produced by the LasI enzyme) and N-butanoyl homoserine lactone (produced by theRhlI enzyme), which bind at critical concentrations to the corresponding regulators LasR and RhlR. The lasRI system mainly regulates lasB, lasA, exoA, xcpP and xcpR expression. The rhlRI system regulates the expression of rpoS, rhamnolipid, LasB and LasA elastase, hydrogen cyanide, pyocyanin, lipase and alkaline protease. The Pseudomonas quinolone signal system potentiates the effects of the rhlRI system. Quorum sensing controls not only virulence in P. aeruginosa but also many other genes involved in cellular processes, including chemotaxis and biofilm formation, comprising altogether about 10% of the P. aeruginosa genome. In addition, quorum-sensing molecules also modulate host responses.

Various morphotypes that are not normally present in environmental P. aeruginosa strains have been described in P. aeruginosa strains, isolated from chronically infected CF patients, including strains that are mucoid, smooth, rough, dwarf, colourless or present as small colony variants (SCV) (size: 1–3 mm). The latter may contribute to the progressive pulmonary damage in CF and persistence in CF airways (Häussler et al., 1999, 2003; Lory et al., 2009; Starkey et al., 2009). Additionally, ‘rugose SCVs’ have been detected that have characteristics of enhanced biofilm formation and display increased expression of the pel and psl polysaccharide gene clusters, decreased expression of motility functions and a defect in growth on some amino acid and tricarboxylic acid cycle intermediates as sole carbon sources (Starkey et al., 2009). The psl (polysaccharide synthesis locus) genes are responsible for the production of mannose-rich exopolysaccharides and the pel locus for pellicle formation.

A consistent finding in CF airways is the presence of mucoid P. aeruginosa strains that overproduce the exopolysaccharide alginate, a negatively charged polymer of β-1-4-linked d-mannuronate, and its C5 epimer, l-guluronate (Deretic et al., 1995; Boucher et al., 1997). Alginate expression is upregulated by microaerobic environmental conditions that are present in the viscous mucus of CF patients (Worlitzsch et al., 2002). Alginate expression is considered to be an important factor for the persistence of P. aeruginosa in the airways of CF patients as alginate protects the pathogen not only against innate immune functions but also against opsonophagocytosis involving specific antibodies, produced by the adaptive immune system. A key element in alginate regulation is the alternative sigma factor AlgT, also known as AlgU, RpoE or σ²², which induces the expression of algD and increases the expression of regulatory proteins that enhance algD transcription (Ramsey & Wozniak, 2005). The anti-sigma factor MucA inhibits AlgT; thus, inactivation of mucA leads to upregulation of AlgT and conversion to mucoidy. Overproduction of alginate has been linked to biofilm formation in P. aeruginosa– a phenotype that favours chronic persistence of the pathogen in murine and CF lungs (Hoffmann et al., 2005). However, both mucoid and nonmucoid P. aeruginosa clonal strains occur in the airways of CF patients simultaneously (Bragonzi et al., 2009). Furthermore, the finding that a high number of nonmucoid strains from CF patients carry a mucA mutation (Bragonzi et al., 2006) suggests that a mucoid morphotype, intermittently selected in the CF lungs, can be suppressed at a later stage due to other, secondary site mutations (Boles et al., 2004), including mutations in algU (Ciofu et al., 2008). Interestingly, P. aeruginosa rpoS mutants reveal a significantly reduced production of alginate (Suh et al., 1999). Mucoid P. aeruginosa strains from CF patients also contain increased levels of G6PDH activity and expression of the corresponding zwf gene. This allows the bacterium to supply sufficient amounts of precursor for the alginate pathway (Silo-Suh et al., 2005).

Other phenotypic changes of P. aeruginosa strains occurring during the course of chronic CF lung infection include loss of motility (Luzar et al., 1985; Rau et al., 2010), decreased exotoxin A expression (Suh et al., 1999), loss of O antigen in lipopolysaccharide (Hancock et al., 1983) and altered lipid moieties (Ernst et al., 1999; Sabra et al., 2003). The latter change confers enhanced resistance to aminoglycoside antibiotics, cationic antimicrobials such as polymyxin E (colistin) and potentially to cationic peptides produced by the host (defensins). Mutations in nfxB, the repressor of the MexCD-OprJ operon, results in increased resistance to fluoroquinolones (Rau et al., 2010). Furthermore, increased auxotrophy (Taylor et al., 1993; Thomas et al., 2000), defects in type II (Woods et al., 1986) and III secretion (Jain et al., 2004; Lee et al., 2005; Hogardt et al., 2007; Rau et al., 2010), reduced production of proteases and phospholipase C, loss of pyoverdine, pyocins and elastase expression (Hogardt et al., 2007), altered metabolic activities (Silo-Suh et al., 2005) and antibiotic resistance (Thomassen et al., 1979; Oliver et al., 2000; Smith et al., 2006; Hogardt et al., 2007) have been described in P. aeruginosa strains from CF patients. Particularly during acute exacerbations, methionine-dependent auxotrophic mutants of P. aeruginosa have been isolated from CF patients (Taylor et al., 1993). In conclusion, chronic exposure to the CF environment selects for a variety of mutations in P. aeruginosa strains.

The frequency of mutations is increased in hypermutable or mutator strains that reveal defects in genes involved in the DNA repair or error avoidance systems, particularly in the methyl-directed mismatch repair system (Oliver et al., 2002; Ciofu et al., 2005, 2010; Hogardt et al., 2007; Montanari et al., 2007). The percentage of CF patients (60%) carrying at least one hypermutable strain is remarkably high (Montanari et al., 2007). Analysis of the genetic background of the P. aeruginosa mutator phenotypes from CF airways showed that mutS was the most commonly affected gene, followed by mutL (Ciofu et al., 2010). A sequential analysis of CF isolates revealed that mutations in mucA, leading to the mucoid phenotype and lasR, leading to loss of quorum sensing (Box 1), occurred earlier than mutations in the mut genes, showing that hypermutability is not a prerequisite for the acquisition of mucoidy and loss of quorum sensing (Ciofu et al., 2010). Accumulation of mutators in the CF lung is associated especially with multidrug resistance development, thus suggesting that the intensive antibiotic treatment is one of the main selective pressures for the maintenance and amplification of mutators (Oliver et al., 2000; Ciofu et al., 2010). The acquisition of a stable mutator phenotype may confer a selective advantage for bacteria, particularly in stressful and/or fluctuating environments because this may increase the probability of generation of adaptive variants (Taddei et al., 1997).

The genetic analysis of clonal isolates, obtained from the CF lung of single patients over several years, revealed the presence of mutations in many genes (Mahenthiralingam et al., 1994; Anthony et al., 2002; Nguyen & Singh, 2006; Smith et al., 2006; Tümmler, 2006; Hogardt et al., 2007; Hoboth et al., 2009; Mandsberg et al., 2009; Ciofu et al., 2010; Rau et al., 2010). When a P. aeruginosa isolate was compared with its clonal variants, isolated 90 months later from a single CF patient, revealing 68 mutations, the analysis suggested a reduced virulence of the latter strains with regard to their ability to induce acute infections, based on mutations in many virulence genes including type III secretion, quorum sensing and motility (Smith et al., 2006) (Table 1). The progressive microevolution of P. aeruginosa in CF patients, known as adaptive radiation, has been interpreted as an in vivo selection process, resulting in less virulent variants that consequently do less harm to their host than the original colonizing strain (Smith et al., 2006).

Table 1

Gene mutations present in Pseudomonas aeruginosa cystic fibrosis isolates

Name	Annotation	Function	References
mexZ†	PA2020	Transcriptional regulator of multidrug efflux	1
lasR	PA1430	Transcriptional regulator of quorum sensing	1
	PA0313	Probable permease of the ABC transporter	1
mexA	PA0425	RND multidrug efflux membrane fusion protein	1
accC	PA4848	Biotin carboxylase	1
vfr	PA0652	Transcriptional regulator of lasR	1
mexS	PA2491	Probable oxidoreductase	1
exsA	PA1713	Transcriptional regulator of type III secretion	1
	PA0506	Probable acyl-CoA dehydrogenase	1
wspF	PA3703	Probable methylesterase involved in chemosensory signal transduction	1
rpoN	PA4462	RNA polymerase σ-54 factor	1
fleQ	PA1097	Transcriptional regulator of flagella synthesis	1
mexT	PA2492	Transcriptional regulator of multidrug efflux	1
nalD	PA3574	Probable transcriptional regulator	1
ampD	PA4522	β-Lactamase expression regulator	1
	PA0366	Probable aldehyde dehydrogenase	1
cyaB	PA3217	Adenylate cyclase	1
pilB	PA4526	Type 4 fimbrial biogenesis protein	1
	PA1333	Hypothetical protein	1
PA3565	Probable transcriptional regulator	1
PA2312	Probable transcriptional regulator	1
anr	PA1544	Transcriptional regulator of anaerobic metabolism	1
rhlR	PA3477	Transcriptional regulator of quorum sensing	1
phoP	PA1179	Two-component response regulator	1
rhlI	PA3476	Autoinducer synthesis protein	1
	PA4796	Hypothetical protein	1
pqsB	PA0997	Homologous to β-keto-acyl-acyl carrier	1
		protein synthase	1
toxR	PA0707	Transcriptional regulator	1
	PA2435	Probable cation-transporting P-type ATPase	1
PA4420	Hypothetical protein	1
PA2121	Hypothetical protein	1
mutS	PA3620	DNA mismatch repair protein MutS	2–4
mutL	PA4946	DNA mismatch repair protein MutL	3, 4
mutY	PA5147	A/G specific adenine glycosylase	4, 5
uvrD	PA5443	DNA helicase II	3, 4
mucA	PA0763	Anti-sigma factor MucA	4, 6
nfxB	PA4600	Transcriptional regulator	7

*

Numbers relate to the genome of PAO1. References 1, Smith et al. (2006); 2, Hogardt et al. (2007); 3, Montanari et al. (2007); 4, Ciofu et al. (2010); 5, Mandsberg et al. (2009); 6, Anthony et al. (2002); 7, Rau et al. (2010).

†

Genes that are mutated in >5% of isolates.

Although this notion may actually be correct, the definition of virulence also includes the capacity of the pathogen to persist in a given host, causing chronic infection. To test this hypothesis, Bragonzi et al. (2009) isolated P. aeruginosa clones from airways of six CF patients during a period of up to 16.3 years and assessed the virulence of clonal variants in mice by monitoring acute and chronic lung infection. Because the definition of virulence is also dependent on a specific host, they tested the clonal variants in mice with different genetic backgrounds, including CF mice. The data demonstrate that early and intermediate P. aeruginosa CF isolates are more virulent to cause acute infections (e.g. bacteraemia) in mouse lungs than late isolates; however, early, intermediate and late isolates do not differ in their capacity to cause chronic infection (Fig. 3). These findings suggest that clonal expansion of P. aeruginosa strains during microevolution within CF lungs leads to populations with altered, but not reduced virulence in mouse lung models. Thus, it may be questionable to define late isolates as less virulent than early isolates or to describe the pathogenicity of CF lung disease as a process leading to an increasingly benign situation. However, the animal experiments used to determine virulence do not reflect a natural colonization process with initially small numbers of bacteria.

Figure 3

Virulence of Pseudomonas aeruginosa strains in a murine model of airways infection. C57Bl/6NCrlBR mice were infected with 1–2 × 10⁶ CFU per lung of 32 P. aeruginosa different strains embedded in agar beads. Mortality induced by bacteraemia (red) and survival (grey) were evaluated on challenged mice. Clearance (white) and capacity to establish chronic airways infection (green) after 14 days from challenge were determined on surviving mice. The data show the percentage of mice infected with P. aeruginosa reference strains (PAO1, n=66; PA14, n=14), a pool of environmental strains (n=65) and a pool of clinical strains (n=274), isolated at the onset (n=91), during the intermediate phase of colonization (n=65) or later during chronic infection (n=118) (adapted from Bragonzi et al., 2009).

Pseudomonas aeruginosa proteins are upregulated in the CF lung, including the outer membrane protein OprF and enzymes of the arginine deiminase pathway (Hoboth et al., 2009) (Table 2). These changes indicate an adaptive shift towards constitutive expression of genes required for growth under the nutritional and microaerobic conditions present in the sputum of CF patients. The high viscosity of the CF mucus also leads to a limited diffusion of siderophore molecules and therefore to an increased fitness of siderophore-producing cells (Cornelis et al., 2009; Kummerli et al., 2009).

Table 2

Metabolic adaptation of Pseudomonas aeruginosa strains during chronic lung infection in cystic fibrosis patients based on protein expression

Name	Annotation	Function
rbsB	PA1946	Binding protein RbsB of ABC ribose transporter
aotJ	PA0888	Arginine-ornithine-binding protein
	PA5153	Probable lysine-arginine-ornithine-binding periplasmic protein
PA1260	Probable lysine-arginine-ornithine-binding periplasmic protein
braC	PA1074	Periplasmic branched-chain amino acid transport protein
spuD	PA0300	Polyamine transport protein
oprF	PA1777	Outer membrane porin OprF
oprD	PA0958	Outer membrane porin OprD
fda	PA0555	Fuctose-1,6-bisphosphate aldolase
adk	PA3686	Adenylate kinase
azu	PA4922	Azurin precursor
ccpR	PA4587	Cytochrome c551 peroxidase precursor
leuD	PA3120	3-Isopropylmalate dehydratase small subunit
hisF1	PA5140	Imidazoleglycerol-phosphate synthase, cyclase subunit
arcA	PA5171	Arginine deiminase
arcC	PA5173	Carbamate kinase
	PA2888	Probable biotin-dependent carboxylase
acpP	PA2966	Acyl carrier protein
accB	PA4847	Acetyl-CoA carboxylase
ppa	PA4031	Inorganic pyrophosphatase
trxA	PA5240	Thioredoxin
sucC	PA1588	Succinyl-CoA synthetase (b-chain)
ansB	PA1337	Glutaminase-asparaginase
lpd3	PA4829	Dihydrolipoamid dehydrogenase 3
aceB	PA0482	Malate synthase G (glyoxylate shunt)
atpF	PA5558	ATP synthase B chain
atpH	PA5557	ATP synthase H chain
pdxH	PA1049	Pyridoxamin 5-phosphate oxidase
glnK	PA5288	Nitrogen-regulatory protein P-II 2
atoD	PA1999	Probable CoA-transferase, subunit A
atoB	PA2001	Acetyl-CoA acetyltransferase
	PA0318	Hypothetical protein
PA2575	Hypothetical protein
PA0388	Hypothetical protein
PA3440	Hypothetical protein
PA1677	Hypothetical protein
PA5178	Hypothetical protein
PA3309	Hypothetical protein
PA5339	Hypothetical protein
btuE	PA0838	Glutathione peroxidase
sodB	PA4366	Superoxide dismutase
kynB	PA2081	Kynurenine formamidase
bfrA	PA4235	Bacterioferritin A
katA	PA4236	Katalase
ahpC	PA0139	Alkyl hydroperoxide reductase subunit C
tsaA	PA3529	Probable peroxidase
ibpA	PA3126	Heat-shock protein IbpA
ppiB	PA1793	Peptidyl-prolyl cis-trans isomerase B
groEL	PA4385	GroEL
rpsF	PA4935	30S ribosomal protein S6
	PA4671	Probable ribosomal protein L25
greA	PA4755	Transcription elongation factor GreA
rplL	PA4271	50S ribosomal protein L7/L12
tsf	PA3655	Elongation factor
tufA	PA4265	Elongation factor Tu
guaB	PA3770	Inosine-5′-monophosphate dehydrogenase
pilH	PA0409	Twitching motility protein PilH
fliC	PA1092	Flagellin type B
murF	PA4416	UDP-N-acetylmuramoylalanyl-d-glutamyl-2,6-diaminopimelate-d-alanyl-d-alanyl ligase
ftsZ	PA4407	Cell division protein FtsZ

*

Adapted from Hoboth et al. (2009).

†

Genes whose expression tends to be strongly upregulated during late chronic infection.

Other bacterial pathogens in CF

Staphylococcus aureus also changes its phenotype under microaerobic/anaerobic environmental conditions in vitro and in the CF mucus, forming biofilm-like aggregates (McKenney et al., 1999; Cramton et al., 2001; Ulrich et al., 2007), which differ genotypically and phenotypically from environmental strains and strains from the nasal habitat (Goerke & Wolz, 2004). The exopolysaccharide present in S. aureus biofilms (Cramton et al., 1999), initially named polysaccharide intracellular adhesin and later poly-N-acetyl glucosamine (PNAG), differs structurally from the S. aureus capsule polysaccharides, which are lost during growth in the CF airways due to elevated pCO₂ (∼4%) (Herbert et al., 1997). The PNAG-encased cells confer resistance to nonoxidative killing (Ulrich et al., 2007). Phage mobilization, possibly via transduction, contributes significantly to genome alteration in S. aureus during infection in CF patients (Goerke et al., 2004), and agr mutants are frequently isolated from these specimens (Goerke et al., 2000). Agr (for accessory gene regulator) is a well-characterized global regulatory locus that is expressed from two promoters: P2 and P3 (Novick et al., 1993, 1995). Agr regulates staphylococcal virulence and other accessory gene functions, including the expression of haemolysins. While the P2 RNAII transcript codes for the quorum-sensing elements AgrA, AgrC, AgrB and AgrD, P3 activation, during the late exponential growth phase, leads to the transcription of RNAIII, resulting in the repression of surface proteins such as adhesins or protein A and in the enhanced expression of exoproteins such as haemolysins (Novick et al., 2003). Agr mutants show enhanced biofilm formation in S. aureus, as Agr activation results in the expression of extracellular proteases (Boles & Horswill, 2008) and membrane-active molecules (Kong et al., 2006), such as the δ-toxin, which contribute to the dispersal of biofilms.

Often SCVs of S. aureus, which may persist in respiratory epithelial cells, are detected in CF lung specimens (Sadowska et al., 2002; von Eiff, 2008). SCVs are mostly nonpigmented and nonhaemolytic and their colonies are ∼10 times smaller than those of the parental strain. In general, SCVs are auxotrophic for haemin and menadione, compounds involved in the biosynthesis of electron transport chain components. Most SCVs of S. aureus, isolated from CF patients, are thymidine-auxotrophic due to long-term trimethoprim-sulphomethoxazol treatment (Schneider et al., 2008). Treatment of CF lung infections using the aminoglycoside tobramycin may result in local subinhibitory concentrations that favour the emergence of aminoglycoside-resistant SCVs (Mitchell et al., 2010), involving the global regulator SigB (Biswas et al., 2009). Besides antibiotics, coinfecting P. aeruginosa can also promote the formation of S. aureus SCVs by producing respiration inhibitors such as 4-hydroxy-2-heptylquinoline-N-oxide (Hoffman et al., 2006; Biswas et al., 2009). Molecular characterization of SCVs from CF patients also revealed that they may be defective in the expression of the agr system (Novick et al., 1993). Taken together, the adaptation of S. aureus to the CF lung leads to biofilm formation and SCVs, both of which favour chronic infection states as a result of increased resistance to innate immune defence mechanisms and antibiotic treatment. Without optimized anti-staphylococcal treatment strategies that reduce the bacterial load in the airways (Döring et al., 2004), chronic S. aureus lung infections may be considerably harmful for the CF patient. Indeed, before the introduction of anti-staphylococcal agents, virtually all CF patients were infected with this opportunistic pathogen (Andersen et al., 1938), which was held responsible for the widespread pulmonary changes in CF (di Sant'Agnese & Talamo, 1967).

The Bcc is a group of genetically distinct, but phenotypically similar bacteria that are divided into at least nine species (Coenye et al., 2001). All group members can cause infections in CF patients, albeit with a low prevalence (∼2–3%). Bcc strains are extremely sensitive towards ROS; this is reflected by the minimal mortality of neutropenic mice when infected with this opportunistic pathogen (Chu et al., 2002). However, in sputum plugs in airways of CF patients where ROS are essentially absent due to the anaerobic/microaerobic environment and in patients with the hereditary disorder chronic granulomatous disease (CGD) who suffer from mutations in NADPH oxidase, this opportunistic pathogen is highly virulent due to its intrinsic resistance to nonoxidative killing during phagocytosis by professional phagocytes (Speert et al., 1994; Pollock et al., 1995; Baird et al., 1999; Sahly et al., 2003; Segal et al., 2003; Sousa et al., 2007). Resistance to cationic antimicrobial peptides (CAMPs) has been linked to the presence of 4-amino-4-deoxyarabinose moieties, which are attached to the phosphate residues in the lipid A backbone of Bcc lipopolysaccharides (Cox & Wilkinson, 1991; Vinion-Dubiel & Goldberg, 2003).

While some strains may cause long-term asymptomatic airway colonization, others cause a rapid decline of lung function in CF patients and are highly transmissible, resulting in patient-to-patient spread within and between CF centres (Govan et al., 1993; Mahenthiralingam et al., 2005). For instance, Burkholderia cenocepacia strains of the multilocus enzyme electrophoresis type 12 lineage can cause a fulminate, necrotizing pneumonia, which may occasionally be complicated by rapidly fatal septicaemia termed the cepacia syndrome in CF patients (Corey & Farewell, 1996; Jones et al., 2001).

As with other opportunistic pathogens, Bcc virulence in CF and CGD patients is multifactorial (for a comprehensive review of Bcc virulence factors, see Mahenthiralingam et al., 2005). Besides lipopolysaccharides, which cause significantly more inflammation than does P. aeruginosa lipooligosaccharides (Mahenthiralingam et al., 2005), biofilm formation has been implicated to play a role in the pathophysiology of CF lung disease (Mahenthiralingam et al., 2005). Many Bcc strains can produce biofilms in vitro (Ferreira et al., 2010), associated with the ability to produce N-acyl-homoserine lactones (Conway et al., 2002), which may protect single cells from antibiotic killing (Desai et al., 1998). Also, the ability of Bcc strains to form mixed biofilms with P. aeruginosa has been demonstrated (Riedel et al., 2001). In a CGD mouse model of Bcc lung infection (Sousa et al., 2007), survival and death of the infected animals was correlated to the growth rate of the different Bcc strains. The highest growth rate was found for the epidemic strain J2315, which multiplied from 10³ CFU–10⁸ CFU within 3 days, whereas for example a Burkholderia vietnamensis strain never killed any mouse and reached only ∼10⁴ CFU after 12 days of infection. Thus, genes active in metabolism may be important for Bcc virulence in CF and CGD. Whether Bcc strains adapt genotypically or phenotypically to the CF niche is not known.

As in P. aeruginosa, hypermutable phenotypes have been observed repeatedly in S. aureus (Prunier et al., 2003), H. influenzae (Roman et al., 2004; Watson et al., 2004) and Bcc strains from CF patients (Burns, 2005), suggesting that this mechanism may play a crucial role in the pathogenesis of chronic lung infection.

Bacteria colonization in COPD

In contrast to CF, the role of pathogenic bacteria in the progression of COPD is an area of uncertainty and the subject of active research. Opportunistic pathogenic bacteria including NTHI, M. catarrhalis, S. pneumoniae and P. aeruginosa are detectable in airways of approximately 25–50% of clinically stable COPD patients (Sethi & Murphy, 2008). Pathogenic bacteria in the lower airways induce increased airway inflammation, a hallmark of COPD, even in the absence of symptoms of acute exacerbation (Soler et al., 1999; Hill et al., 2000), supporting the ‘vicious circle’ hypothesis that proposes that persistent bacterial infections cause progressive lung damage in COPD. This notion is further substantiated by the finding that the frequency of positive cultures increases in patients with more severe COPD (Eller et al., 1998). Thus, chronic bacterial and/or viral infections may play a role in the development and progression of COPD (Sethi & Murphy, 2008). However, conclusive evidence for this hypothesis is lacking at present. A key question on which research effort should be focused is whether or not chronic infection of the airways in COPD accelerates the progressive loss of lung function. Understanding the role of chronic infection in the course of the disease would serve as an important guide in directing research in the field. For example, if chronic infection were shown to cause increased progression of the disease, then the priority should be to develop methods such as vaccines or immunomodulators to eradicate chronic colonization.

The causative role of pathogenic microorganisms in acute exacerbations of COPD, however, is well established (Sethi & Murphy, 2008). Opportunistic pathogenic bacteria are responsible for approximately 50% of exacerbations, while virus-induced exacerbations caused by rhinovirus, respiratory syncytial virus, adenovirus, influenza virus and human metapneumovirus (Greenberg et al., 2000; Seemungal et al., 2001) are approximately 30% (Sethi & Murphy, 2008). Coinfections are frequent. Table 3 shows the relative distribution of bacterial pathogens that caused exacerbations in a 10-year prospective study (Murphy & Parameswaran, 2009). Studies of COPD patients using bronchoscopy and protected specimen brush sampling of distal airways, avoiding upper airway contamination, have consistently shown pathogenic bacteria in quantities indicative of infection (Monso et al., 1995; Soler et al., 1999). Moreover, a comparison of exacerbation and stable state patients, using the same technique, showed significantly higher quantities of these bacteria during exacerbation (Fagon et al., 1990; Soler et al., 1998). Longitudinal molecular typing of pathogenic bacteria from sputum cultures showed that acquisition of a new strain was associated with a twofold increase in the incidence of exacerbations (Sethi et al., 2002). Purulent sputum, a marker of exacerbation, is strongly associated with the presence of pathogenic bacteria in sputum (White et al., 2003; Soler et al., 2007).

Table 3

Relative distribution of bacterial pathogens in COPD

Bacterial species	Positive culture	Exacerbation
Haemophilus influenzae	707	84
Moraxella catarrhalis	310	68
Streptococcus pneumoniae	219	11
Pseudomonas aeruginosa	230	15

*

Number of positive cultures for the pathogen, isolated from the sputum of adults with COPD, followed in a 10-year prospective study in Buffalo, NY.

†

Number of exacerbations in the same 10-year prospective study as defined by the acquisition of a new bacterial strain (based on molecular typing of cultured isolates) simultaneous with symptoms of exacerbation (adapted from Murphy & Parameswaran, 2009).

NTHI and M. catarrhalis are prevalent pathogens

Haemophilus influenzae strains may produce antigenically distinct polysaccharide capsules, which distinguishes strain types ‘a’ through ‘f’. In contrast, strains of H. influenzae that colonize and infect the airways in COPD are nonencapsulated as demonstrated by a lack of reactivity with typing sera for each of the six known capsular serotypes. Thus, these strains are referred to as nontypeable, abbreviated NTHI. Haemophilus influenzae strains use several genetic mechanisms that alter either the gene expression or the gene content that allow rapid adaptation and improved fitness to changing environments including phase variation based on slipped-strand mispairing, mediated by short DNA repeats in either the coding regions or the upstream promoter regions of virulence genes (Gilsdorf et al., 2004; Srikhanta et al., 2010). Phase variation results in a large genetic diversity among H. influenzae strains present in different environments, which complicates the interpretation of virulence-related phenotypes. The multifactorial virulence of H. influenzae strains includes factors such as lipooligosaccharides (Schweda et al., 2007), fimbriae encoded by the hif locus, IgA1 protease(s) (Erwin & Smith, 2007) and the ability to form biofilms (Starner et al., 2006).

NTHI strains show a highly variable pattern of carriage among COPD patients. Some strains of NTHI, causing exacerbations, can also persist for longer periods of time than is apparent from sputum cultures. Molecular analysis shows that a strain is present in COPD airways during periods when it cannot be recovered by routine culture, indicating that selected strains persist in the airways for months to years. A 7-year prospective study (Murphy et al., 2004) including 104 COPD patients observed 17 episodes of prolonged (>6 months) periods of negative sputum cultures preceded and followed by genotypically identical strains of NTHI. Clearing and reacquisition of the identical strain by the patients was ruled out, because NTHI strains are genetically highly heterogeneous, making the reacquisition of the same strain circulating in the community a highly unlikely event. Furthermore, H. influenzae does not remain viable on inanimate objects in the environment even for minutes. Finally, analysis of sputum samples yielding negative cultures revealed the presence of strain-specific NTHI genes, further establishing persistent colonization in COPD patients. Newly acquired NTHI strains isolated from COPD patients during exacerbations, when compared with persisting strains, induce more airway inflammation, including neutrophil influx and IL-8 secretion from respiratory epithelial cells, and adhere more efficiently to the epithelial cells in vitro (Chin et al., 2005). These data suggest that, as with P. aeruginosa in chronic CF lung infection, adaptive processes may occur that modulate the virulence of this pathogen. However, similar molecular studies as mentioned above for P. aeruginosa (Smith et al., 2006; Bragonzi et al., 2009) have not yet been carried out for sequential NTHI isolates from single COPD patients to substantiate this notion. Analysis of the genome content of strains of NTHI that persist in the airways of COPD patients in comparison with those that are cleared quickly will help elucidate the relative roles of the host and pathogen in the dynamics of infection in COPD airways.

Also, M. catarrhalis is regarded as an important human mucosal pathogen in COPD patients (de Vries et al., 2009; Murphy & Parameswaran, 2009). Virulence factors include a family of ubiquitous surface proteins, including the ubiquitous surface protein A1 (UspA1) and UspA2, which can bind α₁-antichymotrypsin, thereby potentially increasing the protease activity in vivo, which may cause excessive inflammation in COPD patients (Perez Vidakovics & Riesbeck, 2009). In contrast, UspA1 has been shown to bind to the human-specific extracellular immunoglobulin V (IgV)-like amino-terminal domain of carcinoembryonic antigen-related cell adhesion molecule 1, resulting in reduced toll-like receptor 2-initiated transcription factor nuclear factor-κB-dependent inflammation (Slevogt et al., 2008). Which of these mechanisms dominate during M. catarrhalis infection in COPD patients is unclear.

COPD patients who acquire new strains of M. catarrhalis usually clear this species efficiently from their airways in approximately 1 month. In a study including 104 COPD patients who experienced 560 exacerbations, molecular typing identified 120 episodes of acquisition and clearance of M. catarrhalis in 50 patients (Murphy et al., 2005). Reacquisition of the same strain was rare. The rapid establishment of protective antibody responses against the pathogen in these patients suggested that the clearance of M. catarrhalis is a function of a functional immune system (Murphy et al., 2005). The findings that 57 (47.5%) of M. catarrhalis acquisitions were associated with clinical exacerbations (Murphy et al., 2005), and that increased inflammation and tilting of the protease–antiprotease balance in favour of protease activity occurred in airways of COPD patients during exacerbation as well as colonization with M. catarrhalis (Parameswaran et al., 2009), suggest that this pathogen contributes to the long-term decrease of lung function in COPD patients. Calculations revealed that M. catarrhalis likely causes ∼10% of exacerbations of COPD.

Pseudomonas aeruginosa does not commonly show persistence in COPD

In a chronic lung disease such as COPD, which is characterized by chronic inflammation, progressive tissue destruction and increased mucus production, one would expect persisting P. aeruginosa clones undergoing a similar microevolution as described above for the P. aeruginosa strains in CF airways. Indeed, single clones of P. aeruginosa have been identified in COPD patients that persisted over several years and developed an increasing number of mutations during the chronic course of the disease, resulting in reduced motility, decreased protease production and increased antibiotic resistance as also recognized in patients with CF (Martinez-Solano et al., 2008). Furthermore, different morphotypes of P. aeruginosa, including SCVs and mucoid phenotypes, were observed in sputum isolates from a single COPD patient, suggesting that in this disease diversification also plays an important role in bacterial persistence. Additionally, strains that persisted over long periods of time tended to be hypermutable as in CF (Martinez-Solano et al., 2008).

However, the similarity to CF reduces when larger patient groups are investigated longitudinally. In a 10-year study involving 126 COPD patients, only 13 strains persisted for >6 months and only four of these were mucoid (Murphy et al., 2008). These data suggest that most COPD patients had sporadic and intermittent infection of airways with P. aeruginosa, which were repeatedly cleared. Strains from one clone were generally cleared by the patients' immune system and a gap in time existed before the acquisition of a strain from a different clone (Rakhimova et al., 2009). Chronic carriage of a clone was observed in a minority of investigated patients in this study. When a clone chronically persisted in COPD-affected lungs for an extended period of time, the clone often altered the repertoire of its accessory genome. Hence, intraclonal microevolution and the frequent turnover or loss of clones are typical of infections with P. aeruginosa in COPD. This epidemiological signature is divergent from that of the chronic P. aeruginosa infections in patients with CF, the latter being characterized by no or slow turnover of clones (Rakhimova et al., 2009).

Circumstantial evidence suggests that chronic infection of COPD airways by P. aeruginosa occurs in those with more severe COPD, particularly among patients who require mechanical ventilation for severe exacerbations (Miravitlles et al., 1999; Ferrer et al., 2005; Rosell et al., 2005; Nseir et al., 2006), or in other words, patients with more severe reduction in FEV₁, when compared with those with milder disease, are more likely to harbour P. aeruginosa in their lower airways (Eller et al., 1998). Patients with more severe COPD also experience a higher number of acute exacerbations and there is increasing recognition that P aeruginosa is responsible for such exacerbations (Murphy et al., 2008). Whether chronic P aeruginosa infection has a negative impact on lung function and the patients' prognosis requires further study.

The role of P. aeruginosa in the clinical evolution of COPD is not well understood. Unlike in CF, no clear evidence of worsening of lung infection and increased mortality is correlated to chronic P. aeruginosa infection in COPD. Furthermore, unlike in CF, eradication of colonizing pathogenic bacteria, in the absence of exacerbation symptoms, has not been shown to be beneficial in the clinical course of COPD. The questions as to why only some patients develop these chronic infections and why only selected bacterial pathogens establish these infections in COPD remain unanswered at present. Multiple host factors including the severity of COPD, genetic polymorphisms as well as bacterial virulence factors in specific clones may play a role in this context. However, a comparison of P. aeruginosa clones in the airways of patients with COPD with those in the airways of chronically infected patients with CF revealed that despite the disparate geographic origin of the isolates, the dominant P. aeruginosa clones were found with similar frequencies in acutely and chronically infected respiratory tracts irrespective of the aetiology of the disease (Rakhimova et al., 2009) (Fig. 4).

Figure 4

Distribution of Pseudomonas aeruginosa strains isolated from patients with CF and COPD. (a) Venn diagram of the number of P. aeruginosa clonal complexes obtained from the airways of patients with COPD and CF, respectively. Clonal complexes were calculated from the 15-marker genotype of the core genome using the program eburst (Maiden et al., 2006). The intersections represent the clonal complexes that had been identified in both diseases. (b) Clonal complex structure of P. aeruginosa strains collected from various different habitats and geographic origins. The eburst analysis was carried out with the informative SNPs of a microassay of the core genome. Coloured clonal complexes contain airway isolates from patients with COPD (yellow), from patients with CF (blue) and from both patient groups (green). The size of the symbol indicates the number of typed strains with an identical genotype. Clones C and PA14 are the most abundant clones (Wiehlmann et al., 2007). Genotypes only observed in other habitats are depicted in grey. (Figure adapted from Rakhimova et al., 2009.)

Antimicrobial treatment for CF patients

For chronic P. aeruginosa lung infection in CF patients, antibiotic therapy has been a cornerstone, allowing a largely improved life expectancy of the affected patients due to stabilization of lung function; by contrast, lung function rapidly decreases in untreated CF patients. Commonly applied antibiotics include the aminoglycoside tobramycin, the β-lactam antibiotic ceftazidime, the fluoroquinolone ciprofloxacine and the polymyxin E colistin (Döring et al., 2000, 2004). However, this therapy regimen, even when applied as recommended (Döring et al., 2000, 2004), i.e., aggressively every 3 months with high doses of inhaled antibiotics, does not eradicate the chronically persisting P. aeruginosa from the airways of CF patients. The highly viscous mucus plugs and the biofilm-like macrocolonies of P. aeruginosa within these plugs, which impede the penetration of antibiotics, are assumed to be responsible for this negative outcome (Xu et al., 1998; Wimpenny et al., 2000; Worlitzsch et al., 2002; Döring et al., 2004). As mentioned above, the stressful environment in the CF lung also favours the emergence of hypermutable (or mutator) strains of P. aeruginosa, which are deficient in the DNA mismatch repair system. Mutator strains produce antibiotic-resistant mutants at elevated rates (Oliver et al., 2000; Hogardt et al., 2007). Combination therapy using antibiotics having different modes of action against P. aeruginosa could be superior to monotherapy in this situation. Recently, the combination of the fluoroquinolone ciprofloxacin and the amphipathic oligopeptide polymyxin E (colistin) has yielded promising results in killing P. aeruginosa biofilms in vitro, based on the concept of combining antibiotics that have different targets in the biofilm cells. While biofilm cells exhibiting low metabolic activity were killed by colistin, ciprofloxacin was found to specifically kill the subpopulation of metabolically active biofilm cells (Pamp et al., 2008). Whether this strategy is successful in CF patients still needs to be tested.

To avoid the development of resistance and in an attempt to eradicate P. aeruginosa early, CF patients were treated with antibiotics immediately after the first detection of the pathogen. In this case, reported eradication rates were above 95% (reviewed in Döring et al., 2000, 2004).

In addition to antibiotic therapy, other treatment strategies may also be successfully used in CF patients to fight bacterial lung infection. Based on the abnormal accumulation of ceramide in CF airways, leading to DNA deposits on the respiratory epithelium, the sphingomyelinase blocker amitriptyline is currently under investigation in clinical trials (Riethmüller et al., 2009). Furthermore, recombinant human DNAse may be helpful in this context, as evidenced in experimental mouse models of P. aeruginosa lung infection (Teichgräber et al., 2008) and in CF patients with regard to S. aureus (Frederiksen et al., 2006).

An important question is whether antibiotic therapy directed against P. aeruginosa is also effective against the large number of strict anaerobes present in the mucus plugs within the airways of CF patients. Two studies suggest that this is not the case (Tunney et al., 2008; Worlitzsch et al., 2009). If strict anaerobes actually did contribute to lung disease in CF patients, additional antibiotics, specifically active against these pathogens, should be administered to CF patients.

In the case of Bcc infections, there is no doubt that these microorganisms cause lung disease in CF patients. Because of the intrinsic resistance of Bcc strains to many antibiotics, combination therapy with two to three different antibiotics according to minimal inhibition concentration assessment is recommended (Döring et al., 2004).

Antimicrobial treatment for COPD patients

Antibiotic therapy regimens in COPD have focused on the treatment of exacerbations. However, only approximately 50% of acute exacerbations in COPD patients are caused by bacterial pathogens, a situation that complicates the interpretation of results in such studies. Nevertheless, the data available strongly suggest that antibiotic therapy for acute exacerbations in COPD patients reduces complications and hastens recovery. In a placebo-controlled study, including 173 COPD patients who suffered from 362 acute exacerbations during the study period, the success rate was 68.1% in the antibiotic group vs. 55% in the placebo group (Anthonisen et al., 1987). Most of the success was seen in patients with more severe exacerbations, indicating that patients with more severe exacerbations benefit more from antibiotic therapy. A systematic review analysed 11 placebo-controlled trials (917 patients) of antibiotics in exacerbations of COPD. Combining the data from three trials that used objective definitions of COPD to qualify for inclusion showed a reduction in mortality with antibiotic use and reduction in the rates of treatment failure (Ram et al., 2006). Also, other meta-analyses of placebo-controlled trials revealed a substantial benefit for the antibiotic treatment of acute exacerbations in COPD patients, emphasizing the pathogenic role of bacteria in the course of this disease (Quon et al., 2008; Roede et al., 2009).

Another goal of antimicrobial therapy is to slow or to prevent the worsening of pulmonary function. The use of antibiotics in stable COPD patients to prevent progressive worsening of lung function in the absence of exacerbations has yielded unclear results due to the lack of large studies with stratified COPD patients. Nevertheless, a recent meta-analysis of data from trials conducted in 1950s and 1960s showed that ‘prophylactic antibiotics’ produced a relative risk reduction of 9% for the occurrence of exacerbations (Black et al., 2003). However, even this modest gain should be interpreted with caution because the trials varied markedly in methodology and duration and many of the antibiotics used in these trials would not be expected to be effective with the current susceptibility patterns of bacteria. In a randomized trial, prophylactic erythromycin reduced the frequency of exacerbations in a group of COPD patients and increased the mean interval between exacerbations (Seemungal et al., 2008). Whether the benefits are due to antimicrobial, anti-inflammatory or both effects of erythromycin is not clear. In a recent pilot study involving 13 COPD patients, colonized with multi-drug-resistant P. aeruginosa, inhalation of tobramycin (300 mg twice daily for 2 weeks) reduced the sputum density of P. aeruginosa and the exacerbation rate (Dal Negro et al., 2008). This small uncontrolled study is promising, but needs to be replicated in larger controlled trials that examine the impact of such a treatment on lung function and clinical outcomes.

The success of early treatment with antibiotics in reducing the incidence of airway infection in CF raises the question of whether a similar approach might be effective in adults with COPD, particularly with regard to P. aeruginosa infection. The majority of adults with COPD who acquire P. aeruginosa clear the organism without antibiotic therapy (Murphy et al., 2008). However, patients with more advanced COPD, persistent infection with P. aeruginosa and severe exacerbations are more likely to benefit from antibiotic therapy. Therefore, COPD patients hospitalized with exacerbations, particularly with severe exacerbations revealing purulent sputum masses, and patients with severe underlying COPD having exacerbations should receive antibiotics, according to a decision-making algorithm (Sethi & Murphy, 2008).

It is not known whether the subset of COPD patients who have persistently positive sputum cultures for P. aeruginosa experience adverse consequences of chronic infection, analogous to patients with CF. This is an important question that needs to be addressed in carefully designed clinical trials. The results will have important implications in understanding the pathogenesis of P. aeruginosa infection in COPD and in the clinical management of these patients.

Concluding remarks

The pathophysiology of CF and COPD lung disease is complex. While chronic pulmonary inflammation triggered by opportunistic pathogens is the cause of the more or less rapidly decreasing lung function in CF, the role of bacterial infection in the pathogenesis of COPD is less well understood. Chronic P. aeruginosa infection plays a critical role in the course of CF, whereas only a small minority of adults with COPD experience chronic P. aeruginosa infection. In COPD, the predominant pathogens are H. influenzae and M. catarrhalis. While in CF airways single P. aeruginosa clones, which adapt genotypically and phenotypically to the CF niche, persist in the majority of patients due to a heavily impaired innate immune system, a frequent turnover and loss of clones are typical of infections with P. aeruginosa in COPD. However, once the extensive tissue remodelling and tissue destruction has reached a certain threshold (which has not yet been determined), opportunistic pathogens such as P. aeruginosa may persist and develop epidemiological signatures similar to those in CF. To prevent this point of no return is a challenge for microbiologists and clinicians.

Acknowledgements

The authors want to thank Peter Michael Weber, Children's Clinic, University of Tübingen, Germany, for excellent graphic art.

Editor: Dieter Haas