This work was supported by National Institutes of Health grants nos. HL071113, HL087166, HL129907, HL122626, and HL133536.
Contrary to conventional teaching, culture-independent molecular techniques have confirmed that the healthy lung is not sterile, and the lung microbiome is likely established in early life and influences the development of immune responses and pulmonary function and is altered in disease states.
Although the major pathway that microorganisms colonize the uterine cavity is vertical ascension from the vagina, there is emerging evidence suggesting transplacental transfer of microbiota as a route of infection.
The airway microbiome of the newborn lung of term and preterm infants is similar in composition and diversity at birth, but the lung microbiome of infants with bronchopulmonary dysplasia (BPD) differs in composition and is less diverse compared with term and preterm infants without BPD.
The genus Lactobacillus is decreased at birth in infants exposed to chorioamnionitis and in preterm infants in whom BPD develops.
The mycoplasmas Ureaplasma parvum and Ureaplasma urealyticum are commensals in the genital tract, but have been associated with intrauterine infection, preterm birth, and adverse neonatal outcomes, including BPD. Current evidence indicates that these organisms modulate host immune responses.
In 1676, Antonie van Leeuwenhoek discovered bacteria. Subsequently, work by Louis Pasteur and Robert Koch in the 19th century linked specific bacterial species to particular diseases. The development of Koch’s postulates improved identification of the etiology of many infectious diseases and strengthened the approach to microbiologic diagnosis. However, it has become apparent that a reliance on culture prevents the identification of microbial species that are difficult to culture. It has been determined that only 1% of all bacteria can be cultured, and many of the microbial species that normally (or abnormally) inhabit the human body are not identified by culture.
Joshua Lederberg coined the term “microbiome” to signify “the ecological community of commensal, symbiotic, and pathogenic microorganisms that literally share our body space.” The term “microbiota” refers specifically to the community of microorganisms living in a particular environment, while the microbiome refers to the communities of microorganisms and their encoded genes. Microbial diversity measures how much variety exists in a microbial community and is characterized by richness (the number of different bacterial species present) and evenness (the relative abundance of the various species with the microbial community), while the term “dysbiosis” describes a microbial pattern associated with disease states. Culture-independent molecular methods show that the microbiota of humans is far greater than previously recognized. The Human Microbiome Project focused particularly on the normal microbiome of the skin, mouth, nose, digestive tract, and vagina and found that even healthy individuals differ remarkably in the diversity and abundance of the microbiome at the different sites. The relative abundance of members of a microbiome are most often determined by sequencing of the variable regions (V1–3 or V4–6) of the bacterial 16S ribosomal ribonucleic acid (rRNA), but more extensive metagenomic and metatranscriptomic sequencing provide more in-depth data on microbial gene function and possible host interactions.
Because of the relative inaccessibility of the lower airways and the lungs, there has been less research on the airway and pulmonary microbiome compared with the gastrointestinal or oral microbiome. Hilty et al. found that lower airways are not sterile, with approximately 2000 bacterial genomes per square centimeter of surface sampled. They also found that the tracheobronchial tree contained a characteristic microbial flora that differs from the nares and oropharynx and between health and disease. The major colonists in healthy people are anaerobes such as Bacteroidetes (e.g., Prevotella spp.) grown with difficulty in culture, while Proteobacteria (e.g., Haemophilus, Moraxella, Neisseria spp.) are strongly associated with airway disease in chronic obstructive pulmonary disease and asthma. It is possible that microbial immigration from the oral cavity contributes to the lung microbiome during health, although the lungs selectively eliminate Prevotella bacteria derived from the upper airways. In healthy lungs, spatial variation in microbiota within an individual is significantly less than variation across individuals, and bronchoalveolar lavage (BAL) of a single lung segment is probably acceptable for sampling the healthy lung microbiome. A limitation of bronchoscopic sampling of the airways is the possible contamination from the oral or nasal flora. However, direct sampling of lung tissue derived from nonmalignant lung tissue samples from patients with cancer determined that Proteobacteria is the dominant phylum and other common phyla include Firmicutes, Bacteroidetes, and Actinobacteria. Microbiota taxonomic alpha diversity increased with environmental exposures to air particulates, residency in high-density population areas, and smoking pack-years.
The newborn lung microbiome is even more technologically challenging to sample because available sampling is limited to intubated infants with upper airways sampled by tracheal aspirates of the endotracheal tube and distal airways sampled by tracheal lavage. Despite this significant limitation, some recent studies have evaluated the airway microbiome in preterm infants, specifically in relation to development of bronchopulmonary dysplasia (BPD). Mourani et al. evaluated serial tracheal aspirates (at <72 hours, 7 days, 14 days, and 21 days) from 10 preterm infants who required mechanical ventilation for at least 21 days. Samples were analyzed by quantitative real-time polymerase chain-reaction (PCR) assays for total bacterial load and by pyrosequencing for bacterial identification. Seventy-two organisms were observed in total. Seven organisms represented the dominant organism (>50% of total sequences) in 31 of 32 samples with positive sequences. Staphylococcus and the genital mycoplasmas Ureaplasma parvum and Ureaplasma urealyticum were the most frequently identified dominant organisms, but Pseudomonas , Enterococcus , and Escherichia were also identified. Most infants in this series established either Staphylococcus spp. (Firmicutes) or Ureaplasma spp. (Tenericutes) as the predominant organism by 7 days of age. Lohmann et al. evaluated tracheal aspirates of 25 preterm infants obtained at birth and on days 3, 7, and 28. Bacterial DNA was extracted, and 16S rRNA genes were amplified and sequenced. It was found that Acinetobacter was the predominant genus in the airways of all infants at birth. Infants in whom BPD later developed had reduced bacterial diversity at birth.
Recently we evaluated the airway microbiome of extremely preterm and term infants soon after birth and in preterm infants with established BPD. Tracheal aspirates were collected from a discovery cohort of 23 extremely low-birth-weight (ELBW) infants and 10 full-term (FT) infants (with no respiratory disease) at birth or within 6 hours of birth at the time of intubation, as well as from 18 infants with established BPD in whom samples were obtained at 36 weeks postmenstrual age (PMA) at the time of endotracheal tube change. A validation cohort was used, consisting of tracheal aspirates from extremely preterm infants at a different institution. 16S rRNA sequencing was performed followed by bioinformatics analysis. We were able to detect and characterize bacterial DNA in tracheal aspirates of all ELBW and FT infants soon after birth. The lung microbiome was similar at birth in ELBW and FT infants irrespective of gestational age. Both ELBW and FT infants had a predominance of Firmicutes and Proteobacteria on the first day of life, in addition to Actinobacteria, Bacteroidetes, Tenericutes, Fusobacterium, Cyanobacteria, and Verrucomicrobia ( Fig. 5.1 ). The relative abundance of bacterial phyla and Shannon alpha diversity did not differ between ELBW and FT infants. Compared with FT newborns matched for PMA, the airway microbiome of infants after diagnosis of BPD was characterized by increased phylum Proteobacteria and decreased phyla Firmicutes and Fusobacteria (see Fig. 5.1 ). At the genus level, the most abundant Proteobacteria in BPD patients were Enterobacteriaceae. To confirm the presence of Proteobacteria in the samples from patients with BPD, we also performed specific endotoxin assays. Endotoxin concentrations in the airway were similar between term and preterm infants at birth, but endotoxin levels were increased in infants with established BPD compared with concentrations at birth. Serial samples in five ELBW infants in whom BPD later developed demonstrated a distinct temporal dysbiotic change with decreases in Firmicutes and increases in Proteobacteria over time. It was observed both in the discovery cohort as well as the validation cohort that genus Lactobacillus was less abundant even as early as birth in infants in whom BPD later developed, compared with the infants without BPD ( Fig. 5.2 ). Interestingly, preterm birth was associated with alterations in the vaginal microbial community with decreased relative abundance of Lactobacillus.
As both extremely preterm and term infants had a similar diverse microbiome at birth, it is probable that the airway and lung microbiomes are established before birth, potentially through transplacental passage of bacterial products. This contention is supported by culture-independent studies of the uterine microbiome demonstrating the presence of bacteria in the placenta, fetal membranes, and amniotic fluid of healthy pregnancies that have challenged the notion that the fetus develops in a sterile environment and have increased our understanding of infection-related preterm birth as a polymicrobial diease. Using next-generation sequencing and metagenomic analyses, Aagaard et al. identified a low-abundance microbiota in the placenta of term and preterm placentas including Escherichia coli , Prevotella tannerae , Bacteroides species, and Fusobacterium species. Using 16S rDNA pyrosequencing to identify bacteria in placental membranes from term and preterm deliveries, Doyle et al. found six genera ( Fusobacterium , Streptococcus , Mycoplasma , Aerococcus , Gardnerella , and Ureaplasma ) and one family (Enterobacteriaceae) that were more abundant in preterm membranes or absent in term membranes. There was reduced abundance of genus Lactobacillus and increased abundance of genera Streptococcus , Aerococcus, and Ureaplasma in membranes from preterm infants delivered vaginally.
It is not currently known what proportion of the transferred bacterial DNA from the placenta is from live bacteria and what proportion is from “processed” bacterial products (DNA fragments, cell wall fragments, and so on). Our recent study indicated the presence of both bacterial DNA and bacterial lipopolysaccharide in neonatal airways at the time of birth. Microbiome analysis evaluates bacterial DNA but does not indicate if the DNA is from live bacteria. It has been suggested that evaluation of susceptibility of the bacterial DNA to DNAse I may indicate the proportion of DNA from live bacteria, as live bacterial DNA is DNAse I–resistant, but bacterial DNA from dead bacteria is DNAse I–sensitive; 63% of DNA in porcine bronchoalveolar lavage fluid (BALF) is DNAse I–sensitive, suggesting the majority of airway bacterial DNA is from dead bacteria.
It may be speculated that the establishment of the lung microbiome during fetal life enables the priming of the immune system in the fetus and later recognition of, and response to, bacterial flora encountered after birth. Alterations in the airway microbiome are associated with childhood pulmonary disorders such as asthma. It is also likely that the lung microbiome contributes to normal alveolar development. Yun et al. studied microbiota of sterilely excised lungs from mice of different origin including outbred wild mice caught in the natural environment or kept under non-specific pathogen–free (SPF) conditions as well as inbred mice maintained in non-SPF, SPF, or germ-free (GF) facilities. Metabolically active murine lung microbiota were found in all but GF mice. Bacteria were detectable by fluorescent in situ hybridization on alveolar epithelia in the absence of inflammation. A higher bacterial abundance in non-SPF mice correlated with more and smaller size alveoli (consistent with better alveolarization), which was corroborated by transplanting Lactobacillus spp. lung isolates into GF mice.
There are many potential mechanisms by which the microbiome may modulate lung injury and repair. It is known that manipulation of the gut microbiota may influence lung pathology via the gut-lung axis, but it is not clear if the manipulation of the gut microbiome also simultaneously alters the lung microbiome or if the effects in the lung can be due solely to alterations of the microbiome in the gut. It is possible that bacteria or bacterial products from the gut may be translocated to the systemic circulation and filtered from the pulmonary circulation into the lungs. It is known that the lung microbiome is enriched with gut bacteria in sepsis and the acute respiratory distress syndrome.
In the next section we review the human and experimental evidence that the low-virulence pathogens Ureaplasma parvum and U. urealyticum that are commonly members of the amniotic fluid and placental microbiota contribute to preterm birth and lung injury owing to an augmented dysregulated inflammatory response that contributes to the development of BPD. Recent studies provide new insights into how these organisms evade the host immune response to establish colonization in the intrauterine cavity and fetal/newborn lung and identify these mechanisms as potential therapeutic targets.
Role of Genital Mycoplasmas in Intrauterine Infection and Neonatal Lung Injury
The Mollicute class consists of at least 200 species, of which humans are the primary hosts for at least 17. Mollicutes are primarily associated with the mucosal surfaces of the urogenital and respiratory tracts. The four urogenital species belong to two different phylogenetic groups within the Mollicute class. Mycoplasma hominis belongs to the Hominis group, while Ureaplasma spp. and M. genitalium belong to the pneumoniae group. Ureaplasma consists of 2 species and 14 serovars. U. parvum contains serovars 1, 3, 6, and 14, and U. urealyticum contains the remaining serovars. The mycoplasma species are the smallest self-replicating, free-living organisms. The M. genitalium genome is the smallest with 580 kilobase pairs (kbp) while U. parvum serovar 3 genome is the second smallest known genome with 751 kbp and the recently sequenced M. hominis genome is 665 kbp. Because of their small genome size, the genital mycoplasma species have limited biosynthetic capacities, requiring a parasitic relationship with a host. These species all lack cell walls and share 247 coding sequences, but they have distinct energy-generating pathways and pathogenic roles in human disease. M. genitalium uses glycolysis, but the Ureaplasma spp. and M. hominis hydrolyze urea and arginine, respectively, to generate adenosine triphosphate. M. genitalium is associated with cervicitis and male urethritis. It has been detected in 9% of BAL fluids of children who had bronchoscopy for chronic respiratory conditions such as asthma. M. hominis is associated with pyelonephritis, bacterial vaginosis, pelvic inflammatory disease, and postpartum endometritis, but it has not been consistently associated with histologic chorioamnionitis or BPD. It is much less commonly isolated as a single organism from amniotic fluid, chorioamnion, or neonatal tracheal or gastric aspirates than are the Ureaplasma spp. The following section focuses on the evidence implicating the Ureaplasma spp. in neonatal lung disease.
Ureaplasma Species: Are There Species- or Serovar-Specific Virulence Factors?
U. parvum is more commonly isolated from clinical vaginal, amniotic fluid, and infant respiratory specimens and is the predominant species in newborn serum and/or cerebrospinal fluid samples detected by PCR. It has been proposed that some serovars have greater association with adverse pregnancy outcomes than others. Abele-Horn et al. reported a higher rate of BPD in infants whose respiratory tracts were colonized with U. urealyticum. In contrast, Katz et al. observed no difference in prevalence of either species detected by PCR between infants with and without BPD. In a recent prospective study of respiratory secretions in infants less than 33 weeks’ gestation, the distribution of Ureaplasma species and serovars was determined by real-time PCR using species- and serovar-specific primers/probes. U. parvum was the predominant species (63%), compared with U. urealyticum (33%). Serovars 3 and 6 alone and in combination accounted for 96% U. parvum isolates. U. urealyticum isolates were commonly a mixture of multiple serovars with serovar 11 alone or combined with other serovars (59%) as the most common serovar. No individual species/serovars or serovar mixtures were associated with moderate to severe BPD. This supports the contention that Ureaplasma virulence is species- and serovar-independent with regard to neonatal lung disease, but this needs to be confirmed. Recent research has shown that clinical isolates often have DNA hybrid genomes, proving that serovar-specific markers have been transferred horizontally. These findings suggest that there could be innumerable serovars or strains based on different combinations of horizontally transferred genes. Thus serotyping for diagnostic purposes or in an attempt to correlate pathogenicity at the serovar level is unlikely to be useful, and host factors are more likely responsible for different outcomes following Ureaplasma infection.
Previously proposed ureaplasmal virulence factors include immunoglobulin A (IgA) protease, urease, phospholipases A and C, and production of hydrogen peroxide. These factors may allow the organism to evade mucosal immune defenses by degrading IgA and injuring mucosal cells through the local generation of ammonia, membrane phospholipid degradation and prostaglandin synthesis, and membrane peroxidation, respectively. Although functionally active IgA protease and phospholipases A and C were found in Ureaplasma spp., the genes that code for these proteins have not been identified in the U. parvum serovar 3 genome. The ureaplasmal enzymes may have unique sequences compared with analogous genes in other species.
The ureaplasmal MB antigen that contains both serovar-specific and cross-reactive epitopes is the predominant antigen recognized during ureaplasmal infections in humans. It exhibits highly variable size in vitro, clinical isolates in vivo, and in an experimental ovine intraamniotic infection model, suggesting that antigen size variation may be another mechanism through which the organism evades host defenses. Ureaplasmas have multiple other host immune response avoidance mechanisms that facilitate establishing a chronic infection in the amniotic cavity and the neonatal respiratory tract. These include the ability to form biofilms, the presence of multiple nucleases that may degrade neutrophil extracellular traps formed when activated neutrophils release granule proteins and chromatin that kill bacteria, and downregulation of various endogenous antimicrobial peptides by ureaplasmal-mediated chromatin modification alterations, including significantly decreased histone H3K9 acetylation. Because neutrophil extracellular traps are abundant in chorioamniotic membranes with acute chorioamnionitis, Ureaplasma -induced nuclease degradation may explain in part the prolonged subclinical intrauterine infection with these organisms. Further genetic studies of the ureaplasmal genome are likely to identify other virulence factors that may be novel therapeutic targets.
Potential Role of Ureaplasma Species in Preterm Birth and Intrauterine Inflammation
Because Ureaplasma is a commensal in the adult female genital tract, it has been considered of low virulence. However, it has been associated with multiple obstetric complications, including infertility, stillbirth, chorioamnionitis, and preterm delivery. Ureaplasma spp. are the most common organisms isolated from amniotic fluid obtained from women who present with preterm labor (POL) with intact membranes, preterm premature rupture of membranes (pPROM), short cervix associated with microbial invasion of the amniotic cavity, and from infected placentas. The prevalence of infected amniotic fluid with cultivated Ureaplasma as the only microbe ranges from 6% to 9% for pregnancies complicated by POL with intact membranes to 22% for a cohort of women with POL or pPROM. Detection of cultivated Ureaplasma in placental chorion in pregnancies producing VLBW infants ranges from 6% to 10% in homogenized frozen tissue to 28% in fresh tissue and is inversely related to gestational age. Recovery of Ureaplasma from the chorion increased with duration of rupture membranes, suggesting an ascending route of infection. However, Ureaplasma has also been detected in 31% of infected placentas with duration of rupture of membranes less than 1 hour, suggesting the possibility of a preexisting infection. Indeed, Ureaplasma species have been detected in amniotic fluid as early as the time of genetic amniocentesis (16–20 weeks) in up to 13% asymptomatic women. Placentas with the lowest rate of Ureaplasma recovery were from women delivered for preeclampsia or intrauterine growth restriction.
The presence of Ureaplasma as the only identified microbial isolate in the upper genital tract is significantly associated with chorioamnionitis and adverse pregnancy outcomes, including premature delivery, neonatal morbidity, and perinatal death. Placentas colonized with Ureaplasma exhibit a characteristic bistriate inflammatory pattern with maternal-derived neutrophils accumulating in the subchorion and amnion. Experimental models of intrauterine Ureaplasma infection in mice, sheep, and nonhuman primates have been described. Intraamniotic inoculation of U. parvum did not stimulate preterm labor in mice or sheep, but it did stimulate progressive uterine contractions and preterm delivery in rhesus macaques inoculated at 136 days of gestation (80% term), suggesting species differences in the host response or serovar differences in virulence. The rhesus macaque model is the first experimental model to definitely show a causal link between Ureaplasma intrauterine infection and preterm labor.
In the presence of pPROM, cultivated Ureaplasma as the sole microbe was associated with increased leukocytes and proinflammatory cytokines (interleukin [IL]-6, IL-1β, and tumor necrosis factor α [TNF-α]) in amniotic fluid and increased cord blood IL-6 concentrations, indicating a robust inflammatory response to this infection. However, amniotic fluid IL-8 levels were higher and the amniocentesis-to-delivery interval was shorter when preterm labor was caused by a combination of Mycoplasma/Ureaplasma and other bacteria than Mycoplasma/Ureaplasma alone. Ureaplasmal bacterial load determined by quantitative PCR in amniotic fluid of women who delivered preterm was associated with histologic chorioamnionitis, preterm labor, PROM, and BPD. The bacterial load correlated with amniotic fluid IL-8 concentrations. Although the majority of women in whom subclinical Ureaplasma amniotic cavity infection is detected midtrimester deliver at term, those with elevated amniotic fluid IL-6 levels have increased risk for an adverse pregnancy outcome, including fetal loss and preterm delivery. In the rhesus monkey U. parvum intrauterine infection model, uterine activity was preceded by a rise in amniotic fluid leukocytes, inflammatory cytokines, prostaglandins (PGs) PGE 2 and PGF 2α , and matrix metalloproteinase-9, demonstrating that Ureaplasma alone stimulates the mediators of preterm labor. Recently Lal and coworkers demonstrated that Ureaplasma spp. stimulate neutrophil matrix metalloproteinase-9 release and express the serine protease prolyl endopeptidase that together induce collagen fragmentation, resulting in release of the tripeptide PGP (proline-glycine-proline), a neutrophil chemoattractant. These findings implicate Ureaplasma spp. in the causal pathway of preterm rupture of membranes and neutrophil influx causing chorioamnionitis.
In vitro studies have provided additional evidence supporting the contention that Ureaplasma spp. stimulate inflammation in the intrauterine compartment. Plasma from placental whole blood (source of maternal circulating leukocytes) that had been preincubated with U. parvum serovar 3 clinical isolate stimulated IL-1β and PGE 2 secretion by chorioamnion explants. High inoculum (10 6 color changing units [CCU]/mL), but not low inoculum (10 2 –10 4 CCU/mL), heat-killed U. urealyticum serotype 8 stimulated TNF-α, IL-10, and PGE 2 production by choriodecidual explants in vitro. In contrast, heat-killed U. parvum serovar 1 laboratory reference strain failed to stimulate a significant increase in cytokine and PGE 2 response in fetal membrane explants derived from term placentas. The apparent low virulence of Ureaplasma in these in vitro studies may be due, in part, to the use of a laboratory reference strain rather than more virulent clinical isolates, or killed rather than live organisms. Alternatively, a decreased capacity to stimulate an inflammatory response in the intrauterine compartment may allow Ureaplasma infections to persist for long periods of time.
Ureaplasma spp. and Neonatal Lung Injury
Ureaplasma respiratory tract colonization in the newborn has been associated with higher incidence of pneumonia and BPD. The rate of Ureaplasma respiratory tract colonization in infants with birth weights less than 1500 g ranges from 20% to 45%, depending on study entry criteria and frequency of sampling and detection methods. In a recent cohort of infants at 33 weeks’ gestation, Ureaplasma spp. were detected by combined culture/PCR during the first week of life in tracheal aspirates or nasopharyngeal specimens in 35% of infants. Ureaplasma -colonized infants are more likely to be born extremely preterm (<28 weeks’ gestation) and to be delivered by spontaneous vaginal delivery following preterm labor or preterm premature rupture of membranes. Typically they experience less respiratory distress in the first week of life with clinical deterioration in the second week, requiring increased oxygen and ventilatory support. Ureaplasma respiratory tract colonization is associated with a peripheral blood leukocytosis and early radiographic emphysematous changes of BPD. These findings may be explained, in part, by an in utero onset of the inflammatory response and lung injury. Indeed, neonatal Ureaplasma respiratory colonization was associated with BPD in infants exposed to antenatal histological chorioamnionitis. Clinical, radiographic, and laboratory characteristics of neonatal Ureaplasma respiratory tract colonization are summarized in Box 5.1 .