Human pleural fluid is a potent growth medium for Streptococcus pneumoniae

Empyema is defined by the presence of bacteria and/or pus in pleural effusions. However, the biology of bacteria within human pleural fluid has not been studied. Streptococcus pneumoniae is the most common cause of pediatric and frequent cause of adult empyema. We investigated whether S. pneumoniae can proliferate within human pleural fluid and if growth is affected by the cellular content of the fluid and/or characteristics of pneumococcal surface proteins. Invasive S. pneumoniae isolates (n = 24) and reference strain recovered from human blood or empyema were inoculated (1.5×106CFU/mL) into sterile human malignant pleural fluid samples (n = 11). All S. pneumoniae (n = 25) strains proliferated rapidly, increasing by a median of 3009 (IQR 1063–9846) from baseline at 24hrs in all pleural effusions tested. Proliferation was greater than in commercial pneumococcal culture media and concentrations were maintained for 48hrs without autolysis. A similar magnitude of proliferation was observed in pleural fluid before and after removal of its cellular content, p = 0.728. S. pneumoniae (D39 strain) wild-type, and derivatives (n = 12), each with mutation(s) in a different gene required for full virulence were inoculated into human pleural fluid (n = 8). S. pneumoniae with pneumococcal surface antigen A (ΔpsaA) mutation failed to grow (2207-fold lower than wild-type), p<0.001, however growth was restored with manganese supplementation. Growth of other common respiratory pathogens (n = 14) across pleural fluid samples (n = 7) was variable and inconsistent, with some strains failing to grow. We establish for the first time that pleural fluid is a potent growth medium for S. pneumoniae and proliferation is dependent on the PsaA surface protein and manganese.


Introduction
Pneumonia affects 450 million patients worldwide each year, [1] with Streptococcus pneumoniae the most common bacterial cause. As many as 20-40% of patients with pneumonia develop a simple parapneumonic pleural effusion [2]. Pleural infection can develop when the pleural fluid is secondarily infected, and affects~80,000 people in the UK and USA a year with significant morbidity and mortality [3].
Empyema represents the most severe end of the pleural infection spectrum. It is defined by the presence of bacteria, or pus, in the pleural fluid. S. pneumoniae accounts for the vast majority of pediatric empyema and is among the most frequent causative organisms of empyema in adults [4][5][6]. The incidence of pneumococcal empyema continues to rise globally [7][8][9].
Compared to most other pathogens that cause pneumonia, S. pneumoniae seems to have a particular affinity for causing pleural infection, but the reasons why are not fully understood. During pleural infection, the pleural fluid is exudative and contains a diverse milieu of biologically active molecules. Despite the clinical importance of empyema, no studies have investigated the biological interactions between bacteria and pleural fluids. It is unknown, for example, if bacteria are merely shed from the pleural tissues into the fluid and if the fluid influences bacterial growth. On one hand, exudative fluids are rich in nutrients that can potentially enhance microbial growth; on the other hand, many proteins and enzymes (eg defensins, and lysozymes etc.) in the fluid can serve to defend against bacterial invasion. These questions are difficult to study in empyema fluid as the presence of bacteria can profoundly alter the pleural fluid composition. In clinical settings, patients are often given systemic antibiotics before the fluid is sampled, taking away the opportunity to study the effects of bacterial growth.
Pleural effusion is not unique to pleural infection but is the common end-product of a variety of other non-infective pleural diseases/inflammation that leads to vascular hyper-permeability and plasma leak into the pleural cavity. In this paper, we studied whether the pleural fluid of patients with other non-infective pleural etiologies could support survival and growth of S. pneumoniae strains. We hypothesized that pleural fluid can serve as a growth medium for common empyema bacteria such as S. pneumoniae. We further investigated the role of common pneumococcal surface proteins, in particular nutrient transporters, on pneumococcal growth and survival within human pleural fluid.

Pleural fluid samples
Pleural fluid samples were collected from patients attending Sir Charles Gairdner Hospital (SCGH) that required pleural fluid drainage for clinical indications, with approval (SCGH Human Research Ethics Committee 2012-156).
Pleural fluid samples were collected using aseptic techniques from patients with pleural effusions of non-infective etiologies. All samples were subjected to routine bacterial culture for which the fluid was injected into and transported in blood culture bottles and processed in an accredited hospital laboratory (PathWest, Western Australia). Pleural samples confirmed to be culture-negative were used in experiments. Patients who received antibiotics within 72 hours were excluded. Pleural fluid biochemistry (pH, protein, lactate dehydrogenase and glucose) was measured as previously described [10]. Supernatant was obtained from pleural fluid by centrifugation at 1020×g for 10 minutes.
To standardize counts between experiments, frozen inocula of S. pneumoniae were prepared as previously described [21]. Inocula for D39 strains and non-pneumococcal bacteria were prepared from 18-hour blood agar to a turbidity of 0.5 MacFarland (PathWest) in sodium chloride 0.85% using a Sensititre Nephelometer (Thermo-Scientific; Waltham, USA).

Pleural fluid inoculation
Pleural fluid samples (in 3mL aliquots) were warmed to 37˚C then inoculated with approximately 1.5×10 6 CFU/mL of bacteria and incubated for 24hrs at 37˚C and 5% CO 2 unless specified. Baseline inoculum and final concentrations (CFU/mL) were verified by manual counting log serial dilutions from blood agar. Pleural fluid characteristics and bacteria used in each experiment are presented in Table 3. Pleural fluids before and after centrifugation (n = 11 pairs), to remove cells, were inoculated with S. pneumoniae isolates (n = 25). Six pairs were inoculated within 6 hours from collection and all within 24 hours (range 1 to 18 hours). Growth was compared in parallel to Todd-Hewitt Broth (THB; Thermo-Scientific) containing 17% FCS (Serana Australia, WA), [21], Dulbecco's Modified Eagle Medium (DMEM; Invitrogen, NY) containing 1000mg/L of glucose and sodium chloride 0.85% (n = 3 each) over 48 hours in six pleural fluid samples at 4, 8, 12, 18, 24, 28 and 48 hours. A further 7 pleural fluids were inoculated with non-pneumococcal bacteria (n = 14) and S. pneumoniae clinical isolate (n = 2) controls, at the same starting inoculum of 1.5×10 6 CFU/mL, for comparison purposes. The CFU/mL of bacteria failing to grow at 24 hours was subsequently measured at 4 and 8 hours in pleural fluid (n = 3).
D39 wild and mutant pneumococci (n = 12) were tested in pleural effusions (n = 8). Further experiments with ΔpsaA in parallel to wild type in pleural fluids (n = 16) with and without supplementation of 3μM manganese chloride tetrahydrate (Sigma-Aldrich, Sydney, Australia) were performed. The concentration was determined following concentration dependent supplementation experiments ranging from 0.003 to 30μM of manganese.

Statistical analysis
Statistical analyses were performed using SigmaPlot 12.5 (Systat Software, San Jose, CA). Results are presented as mean (SD) or median (IQR) based on normality of data. The Wilcoxon Signed-Rank test and paired t-test (for non-parametric and parametric data respectively) were used to compare baseline to 24 hour pneumococcal concentrations and Student's t-test to compare growth between pleural fluid and other medium at various time-points over 48 hours. The Mann-Whitney Rank Sum test was used to compare differences between samples containing cells and correspondent supernatant and the Pearson Correlation was used to determine any relationship between spun and unspun samples. Significance was defined as p<0.05.

Optimization and pneumococcal growth conditions
Significant growth was observed when 3mL volumes of pleural fluid were inoculated with 1.5 × 10 6 CFU/mL S. pneumoniae (n = 3), with a median increase of 6875 (IQR 3494-15744) fold at 24 hours compared to baseline. Volumes ranging from 1 to 10mL in both tubes and multi-well plates demonstrated similar growth, however volumes less than 1mL had inconsistent or poor growth (data not shown). Inoculum concentrations in the range of 1.5 x 10 3 to 1.5x 10 6 CFU/mL demonstrated similar bacterial growth at 24 hours; an initial concentration of 10 6 CFU/mL was used across experiments. Samples were incubated for 24 hours following S. pneumoniae proliferated rapidly in all pleural effusions tested All 25 strains of S. pneumoniae proliferated rapidly in all (n = 11) effusion samples tested ( Fig  1) by a median of 3009 fold (IQR 1063-9846) from baseline (from 3.83 × 10 5 CFU/mL to 1.3 × 10 9 CFU/mL), p<0.001 after 24 hours. S. pneumoniae growth was consistent with no  Table 3.
https://doi.org/10.1371/journal.pone.0188833.g001 difference in CFU/mL when data was analyzed as growth per S. pneumoniae isolate, growth in each pleural fluid sample or growth per serotype (p = 0.776). Removal of the cellular content of the pleural fluid by centrifugation did not affect S. pneumoniae growth, with a median fold increase at 24 hours of 2857 (IQR, 1020-9735, p = 0.728 v. growth in uncentrifuged fluid). Growth of S. pneumoniae in pleural fluid with cells correlated closely to growth in pleural fluid without cells present (Fig 2, p<0.001).

S. pneumoniae reaches high densities in pleural fluid and avoids autolysis
Growth of S. pneumoniae in pleural fluid was compared to growth in 1) THB with fetal calf serum (17%), a supplemented culture medium commonly used to cultivate pneumococci, 2) DMEM, a glucose rich cell culture medium and 3) 0.85% saline as a control. Rapid proliferation of S. pneumoniae was observed in THB with FCS, DMEM and pleural fluid within 8 hours following inoculation (Fig 3). The pneumococci did not proliferate in saline alone. Maximum CFU/mL were obtained at 8-hours post-inoculation in broth medium with higher or similar concentrations observed in pleural fluid at the same time point. From 8 to 18 hours' post-inoculation, autolysis became evident for bacteria grown in either DMEM or culture medium with S. pneumoniae CFU/mL returning to baseline by 24 to 48 hours. In contrast, concentrations of S. pneumoniae CFU/mL in pleural fluid were maintained at >1000 fold from baseline over 48 hours.

S. pneumoniae strains demonstrated consistent growth whereas growth of other bacterial pathogens was variable
We investigated the growth of other bacterial pathogens (n = 14) across pleural fluid samples (n = 7). Growth measured at 24 hours was variable across other bacterial pathogens (Fig 4) with a median fold change in CFU/mL of 12.0 (IQR 0.2-813.0) compared to 1172 (IQR, 587-1757) for S. pneumoniae controls (n = 2), p<0.001. Furthermore, inconsistent growth between strains of the same bacterial pathogen was evident, including within the viridans streptococci group (n = 3) which are a frequent cause of adult empyema. The bacterial pathogens which failed to grow at 24 hours in pleural fluid did not demonstrate increased growth at earlier time points of 4 or 8 hours (data not shown).

S. pneumoniae proliferation in pleural fluid is dependent on pneumococcal surface adhesin A (PsaA)
To identify factors required for S. pneumoniae growth in pleural fluid, growth experiments were repeated using the D39 wild-type strain and eleven mutant strains (Δply, ΔlytA, ΔpspA, ΔpsaA, ΔluxS, ΔcbpD, ΔpitA, ΔpiuB/piaA, ΔadcAI, ΔadcAII/adcAI, ΔlivH) in eight pleural fluid samples ( Table 2). These mutant strains contain mutations in S. pneumoniae genes required for full virulence either because they encode surface proteins known to be important for host interactions or components of ABC transporters required for nutrient acquisition in vivo [11-17, 19, 20, 22]. All of the mutant strains achieved a similar density of CFU/mL as the wild-type D39 stain with the exception of ΔpsaA. After 24 hours in pleural fluid, the CFU/mL of ΔpsaA had decreased from baseline with a fold change of 0.55 (IQR, 0.08-3.255) whereas the D39 wild strain had increased 1214-fold (503-1746, p<0.001). A decrease in ΔpsaA concentrations  Table 3.
https://doi.org/10.1371/journal.pone.0188833.g003 from baseline was apparent at 4 hours and further decreased at 8 and 24 hours. Importantly, manganese supplementation of pleural fluid by the addition of 3uM of manganese prior to inoculation restored ΔpsaA proliferation in the pleural fluid, increasing the median foldchange in CFU from 0.11 (0.07-501) to 1546 (509-4775) at 24 hours, p <0.001 (Fig 5). 3μM manganese chloride-tetrahydrate was selected following concentration dependent supplementation experiments ranging from 0.003 to 30μM of manganese. The addition of manganese had no impact on growth of the D39 wild type strain. Iron and zinc supplementation of Δpit and ΔadcAII/I mutants respectively had no impact on growth compared to the D39 wild type control (data not shown).

Discussion
S. pneumoniae is a common cause of both pediatric and adult empyema, yet the reasons why this pathogen can cause pleural infections have only been partially investigated. This is the first study to investigate the interactions of bacteria, namely S. pneumoniae, and human pleural fluid ex vivo. We found that pleural fluid provides a rich medium to support the growth of all relevant S. pneumoniae strains tested and that pneumococci appear to be much better adapted to grow in pleural fluid than most other common respiratory pathogens. More importantly, growth of S. pneumoniae in pleural fluid was dependent on PsaA, a manganese transporter making this a potential future therapeutic target. Mutations of other key virulence factors of S. pneumoniae did not affect the growth of pneumococci in human pleural fluid ex vivo. We have previously shown in our mouse model that S. pneumoniae from infected pneumonia tissues invades the pleural cavity within 4 hours of infection via transcytosis across the mesothelial cells [23]. Little information exists on the fate of bacteria after they invade human pleural fluid. This is the first study to investigate these interactions.
To investigate the capacity of pleural fluid to support bacterial growth we used human pleural effusions, most with malignant etiology. Parapneumonic effusions and malignant pleural effusions are both exudative in nature and have indistinguishable biochemical compositions with elevated protein and lactate dehydrogenase and reduced glucose and pH. The effusions used for this study display the same biochemical features of non-infective simple parapneumonic effusions. Our results show that all 25 isolates of S. pneumoniae consistently proliferated across eleven human exudative pleural fluid samples tested with a median increase of approximately 3000-fold from baseline at 24 hours. In contrast, the median growth of other bacterial pathogens in pleural fluid was approximately 10-fold, and interestingly the growth was variable with some bacterial isolates demonstrating similar levels of growth to S. pneumoniae, whilst other isolates consistently failed to grow.
When tested in parallel, the growth of S. pneumoniae was similar to optimal laboratory media at 8 hours demonstrating that pleural fluid represents an excellent growth medium for S. pneumoniae. Furthermore, pneumococci naturally undergo autolysis after reaching stationary phase during growth which was observed in broth at 18 hours, yet bacterial CFU/mL density was maintained for 48 hours after growth in human pleural fluid. This persistent growth is likely to promote significant neutrophil influx and may be a reason why pneumococci are a frequent cause of empyema. This failure of autolysis would ensure S. pneumoniae strains are capable of maintaining high level infection even in a closed environment such as a pleural effusion. The reasons why autolysis did not occur after growth in pleural fluid are not known. The cellular content of the pleural fluid did not influence S. pneumoniae proliferation, indicating it is the pleural fluid itself that provides the medium for S. pneumoniae growth. The ability to grow in pleural fluid did not vary between individual S. pneumoniae isolates or serotypes investigated.
Multiple cell surface proteins that function as adhesins, in complement resistance, autolytic enzymes and nutrient transporters in S. pneumoniae have been identified as essential for virulence and growth of the bacterium. We investigated eleven of these virulence factors for their role during growth in pleural fluid, using mutants that are known to have reduced growth in blood (eg Δply, ΔlytA, ΔpsaA and ΔlivH) [11,20,[24][25][26][27] or in cation depleted media (ΔpiuB/piaA, ΔadcAI/ΔadcAII) [22,28,29]. However, perhaps surprisingly, only one mutant strain demonstrated reduced growth in pleural fluid, the ΔpsaA mutant. PsaA is a lipoprotein required for manganese transport and resistance to oxidative stress [14,30,31]. Mutation of PsaA reduces virulence of S. pneumoniae and growth in laboratory medium that has been depleted of manganese. Providing biochemical complementation of the ΔpsaA mutation by supplementing pleural fluid with 3uM manganese was sufficient to restore growth in pleural fluid, confirming that it is the low concentration of manganese that causes failure of growth of the ΔpsaA strain [31]. This suggests that manganese is rate limiting for growth of S. pneumoniae in pleural fluid yet pneumococci are capable of overcoming this using the PsaA transporter. Potential PsaA inhibitors that have recently been identified open opportunities to control S. pneumoniae proliferation in the pleural space [32]. Interestingly, growth of S. pneumoniae with mutations affecting transporters required for iron or zinc uptake was not affected, suggesting that manganese results are specific for growth within pleural fluid.
The study has a number of potential limitations. We were unable to use pleural fluid of infected etiology in the present study for reasons stated in the introduction. However, malignant pleural effusions have a close resemblance in biochemistry to parapneumonic effusions and were therefore used for the majority of our experiments. Although the cellular content of these effusions may differ, we found that the cellular content of the fluid did not influence the growth/proliferation of S. pneumoniae in vitro. Next, due to the nature of the study, experiments were performed in ex vivo pleural fluid and therefore host defenses, including neutrophils that may modify the bacterial proliferation were not accounted for. However, as a proof of principle study our aim was to investigate the direct effect of pleural fluid on S. pneumoniae proliferation. Furthermore, our data showed no difference in proliferation with or without cellular content of the pleural fluid. Finally, we studied 25 invasive strains of S. pneumoniae that had caused empyema or septicemia. Future studies that include other serotypes may be needed, though the very consistent nature of the results indicates that the majority of S. pneumoniae isolates are likely to grow in pleural fluid.
Pleural fluid is a potent growth medium for S. pneumoniae and may be one reason why pneumococci are a common bacterial cause of empyema. The high density and persistent growth of S. pneumoniae in pleural fluid highlights the importance of draining infected effusions. Pneumococcal growth was independent of the cellular content of pleural fluid but dependent on the PsaA surface protein, making PsaA a potential future therapeutic target. The variable growth of other bacterial pathogens in pleural fluid may account for the lower incidence of otherwise common respiratory pathogens in empyema and warrants further exploration in order to understand pathogenesis of pleural infection.
Supporting information S1