Exacerbation induces a microbiota shift in sputa of COPD patients

Little is known about the microbiota shift induced by exacerbation in chronic obstructive pulmonary disease (COPD) patients. The sputa microbiota of COPD patients was evaluated when clinically stable and during acute exacerbations of the disease. Sputa microbiota was analyzed using 16S ribosomal RNA gene pyrosequencing and quantitative polymerase chain reaction-based pathogen detection. Nine COPD patients were enrolled. Pyrosequencing of 16S rRNA genes identified 2,267 unique bacterial operational taxonomic units. Principal microbiota shifts during exacerbation were in either Proteobacteria, Firmicutes or Bacteroidetes. Streptococcus and Moraxella levels were detected during exacerbation in severe (Global Initiative for Chronic Obstructive Lung Disease 3) COPD patients. Most of the clinically-important genera found in the sputum with the pyrosequencing of 16S rRNA gene correlated with specific quantitative polymerase chain reactions for bacteria while respiratory viruses were nearly absent. Sputum microbiotas of exacerbated COPD patients are complex. This pilot study shows a clear shift in the microbiota of patients during exacerbation. The nature of this shift varies from patient to patient in such a way that the treatment should be patient-specific. Further studies are needed to establish the impact of microbial exacerbations on the pulmonary microbiota.


Introduction
Chronic obstructive pulmonary disease (COPD) is characterized by non-fully reversible airflow obstruction. It is currently the 4 th leading cause of death worldwide [1,2], and has been projected by the World Health Organization to become the 3 rd by 2030 [3]. COPD patients often experience episodes of acute symptom worsening such as shortness of breath, wheezing, chronic cough, called acute exacerbations (AE) [4]. In fact, AE are an important cause of mortality and they are associated with the severity of the disease [5,6].
In contrast to what was previously thought, recent microbiota studies have shown that the lungs are not sterile, even in healthy individuals [2,[15][16][17]. The lung microbiota of COPD patients has historically been investigated using culture-based methods with a limited view of biodiversity [11,[18][19][20][21][22]. Moreover, virus cultures are fastidious and have to be specifically adapted for the various virus types. Viral identification is also cumbersome using conventional culture-based methods [23].
Sze et al. characterized the microbiota of eight severe COPD patients using lung tissues and showed that Proteobacteria (44%), Bacteroidetes (28%), Firmicutes (16%), Actinobacteria (6%) and Tenericutes (5%) were the main phyla detected in those tissues. A significant increase of Firmicutes was shown compared to non-COPD patients included in their study (non-smokers, smokers and cystic fibrosis) [15]. Other studies suggested that Firmicutes [2,24,25], Proteobacteria [17] or both [26] are the main phyla found in the lungs of COPD patients. Results from these studies are, however, difficult to compare as the nature of the analyzed samples were different (bronchoalveolar lavage (BAL) or tissues). Importantly, the clinical use of these methods is limited by their invasive nature or their rarity (tissues). Sputum sampling is a rapid, low cost method, and non-invasive since most patients are able to spit spontaneously. However, sometimes saline is required in order to obtain induced sputum. Bronchoscopy with protected specimen brush (PSB) or bronchoalveolar lavage (BAL) require to locally disturb the lung tissue. To our knowledge, only one study compared the microbiota of stable and exacerbated patients using sputum analyses (n = 12) [27]. Huang et al. studied the dynamic (five samples/patient) of the lung microbiota and showed a predominance of Proteobacteria such as Moraxellaceae, Pasteurellaceae, Pseudomonadaceae, and Enterobacteriaceae at exacerbation in the sputum of COPD patients. They also pointed out that inhaled corticosteroids (ICs) may alter the microbiome, however viral analysis were not performed.
In this study, we used 16S rRNA gene pyrosequencing and quantitative PCR-based pathogen detection methods to characterize the sputum microbiota in patients with COPD during stable conditions and, subsequently, during acute COPD exacerbations (AECOPD), testing the hypothesis that there would be a shift sputum microbiotas during the exacerbation period. We also evaluated whether the sputum microbiotas in COPD are influenced by the severity of airflow limitation.

Participants
Nine patients were recruited at the COPD clinic of the Institut Universitaire de cardiologie et de pneumologie de Québec (IUCPQ). Patients gave written consent and the study was approved by the IUCPQ's Bureau du comité d'éthique de la recherche (BCÉR) (CÉR 2013-2115). Patients with chronic airflow limitation (forced expiratory volume in 1 s [FEV 1 ]/forced vital capacity [FVC] < 0.7) were recruited based on a history of at least one exacerbation in the previous six months. Patients' inclusion criteria were as follows: a cumulative smoking history ! 10 packs-year and a stable clinical condition with no antibiotics use for at least one month prior to collecting baseline sputum. Patients were then asked to contact the research team as soon as they were experiencing symptoms of clinical deterioration to reassess their sputum during an AECOPD. Patients were judged to have an AECOPD when two out of the three symptoms were seen (increased dyspnea, increased sputum volume and sputum color change (Anthonisen's classification)) [28]. Patients on any antibiotic treatment, with infectious diseases such as HIV, tuberculosis, acute pneumonia and diagnosed with asthma were excluded from the study.

Clinical data
Principal demographic data of each patient such as age, gender, Global Initiative for Chronic Obstructive Lung Disease (GOLD) classification and indices of lung function (FEV 1 , FVC, and FEV 1 /FVC) were collected.

Sampling and sample processing
Sputa were collected from patients either spontaneously or by induction during stable condition and during the exacerbation period, before any use of antibiotics. Induced sputum was produced as follows: a saline solution (0.9%, 3.0%, 4.0% and 5.0%) was nebulised every seven minutes for a maximum of 21 minutes. Patients were asked to spit in a sterile tube.
Sputa were processed within two hours following collection. Briefly, mucus was selected from the specimen, suspended in phosphate buffer saline (1X PBS, pH 7.1; Cellgro, Manassas, Virginia, USA), treated with 0.2% dithiothreitol (DTT, EMD Millipore, Etobicoke, Ontario, Canada), weighed and frozen at -80˚C until DNA and RNA extractions. All sputa samples passed the quality test (< 25 epithelial cells per low-power field).

DNA and RNA extraction
Sputum samples were completely thawed on ice, incubated with one equal volume of 10X DTT (100 mg = 100 μl DTT) at 37˚C for one hour, and vortexed every ten minutes [29]. Samples were then split in two for both DNA and RNA extractions. Prior to RNA extraction, samples were spiked with 1×10 7 MS2 phages as an extraction and reverse transcription (RT) positive control. DNA and RNA were extracted with the MO BIO PowerLyzer DNA Extraction Kit (MOBIO, Carlsbad, California, USA) and AMBION MagMAX Viral RNA/DNA Isolation Kit (Life Technologies, Burlington, Ontario, Canada), respectively. RNA was eluted in 50 μL elution buffer heated at 56˚C and immediately transformed into cDNA by RT. DNA was eluted in 50 μL and stored at -20˚C prior to qPCR. DNA was quantified with a Thermo Scientific NanoDrop 2000 spectrophotometer (Thermo Scientific, Waltham, Massachusetts, USA).

Reverse transcription
RT was performed with the iScript cDNA Kit (Bio-Rad, Mississauga, Ontario, Canada). Bio-Rad's protocol was as follows: 5 minutes at 25˚C, 30 minutes at 42˚C and 5 minutes at 85˚C. Each 40 μL reaction was composed of 30 μL of viral RNA template (limit of detection = <4 ng/μl), 8 μL of 5X iScript reaction mix and 2 μL of iScript reverse transcriptase. cDNA samples were stored at -20˚C prior to qPCR.

qPCR standard curves
Genomic DNA extracted from P. aeruginosa, M. catarrhalis, and H. influenzae were used for standard curves for pathogen-specific qPCRs while S. pneumoniae genomic DNA was used for a standard curve for both total bacteria and S. pneumoniae qPCRs. Genomic DNA of these strains was extracted with the QIAGEN QIAmp DNA mini kit (QIAGEN, Toronto, Ontario, Canada).
Standard curves for viral pathogen quantification by qPCR were obtained from cloning cDNA of RSV

qPCR assays
Total, specific bacterial and viral pathogenic agents were quantified by qPCR. P. aeruginosa, S. pneumoniae, M. catarrhalis, H. influenzae, Influenza A and B, respiratory syncitial virus (RSV) A and B, adenovirus, and human rhinovirus were assessed using published primers and probes (S1 Table, [19,30,31]). All probes and primers were obtained from Integrated DNA Technologies (IDT, Coralville, Iowa, USA). Each qPCR reaction volume was composed of: 7.5 μL of Bio-Rad 2X iQ Supermix (2X reaction buffer with dNTPs, iTaq DNA polymerase, 6 mM MgCl 2 and stabilizers), 1 μM forward primer, 1 μM reverse primer, 0.1 μM probe, 2 μL of DNA or cDNA template, and water (Sigma-Aldrich, St-Louis, Missouri, USA) up to the final volume of 15 μL. The thermal-protocol was as follows: 95˚C for 3 min, followed by 40 cycles of 95˚C for 10 s and 60˚C for 30 s. All samples were analyzed in duplicate and water (Sigma) was used as a negative control. Reactions were manually loaded into Bio-Rad 96-well white plates. Amplifications were performed using a Bio-Rad CFX 96 Real-Time System and data acquired with the on-board CFX Manager Software version 3.0. qPCR efficiencies, detection limits and numbers of total bacteria and pathogenic agents were determined by employing serial 10-fold dilutions of the standard curves. All qPCR efficiencies were between 90-110%, with R 2 values ranging between 0.96-0.99.

pyrosequencing of 16S rRNA genes
Universal primers were used to target the SSU rRNA gene within the V6-8 region of Bacteria [32]. Barcodes (TCB 2009-005) added were from the extended MID set (Roche, Branford, Connecticut, USA). PCR reactions contained: 1X Q5 buffer (New England Biolabs, Ipswich, Massachusetts, USA), 200 μM of each dNTP (Feldan, Quebec, Quebec, Canada), 0.2 μM of each 454 primer (IDT), 1 U of Q5 High-Fidelity DNA polymerase (NEB), and 1 μL of template DNA (10-50 ng). Cycling conditions were as follows: an initial denaturation at 98˚C for 30 s, followed by 30 cycles of denaturation at 98˚C for 10 s, annealing at 55˚C for 30 s, extension at 72˚C for 30 s, and a final extension at 72˚C for 5 min. PCR products were purified using the Axyprep Mag PCR Clean-up Kit (Axygen, Union City, California, USA), and verified with a Bioanalyzer 2100 with DNA 7500 chips (Agilent Technologies, Santa Clara, California, USA). PCR products were then quantified with a NanoDrop 2000 spectrophotometer (Thermo Scientific). The 12 sample-coded amplicons were mixed in equal quantity and 1/8th plate was sequenced on a Roche 454 GS-FLX Titanium platform at the Plate-forme d'Analyses Génomiques de l'Université Laval (Québec, QC, Canada).

Pre-processing and quality control of raw sequences
The datasets generated during the current study are available in the Sequence Read Archives (SRA) repository, identification number: SRP107187. Web link to the datasets: https://trace. ncbi.nlm.nih.gov/Traces/sra/sra.cgi?study=SRP107187. Raw reads were processed using the mothur pipeline (v1.30; http://www.mothur.org/) [33]. Low-quality reads were discarded: (1) containing one or more Ns, (2) short or long reads (<330 and >600 nt), (3) with incorrect forward primer sequence, and (4) homopolymers >8 nt. All nucleotides following the reverse primer were trimmed. Chloroplasts, mitochondria and putative chimeric reads were eliminated within mothur using included reference databases. Quality reads were aligned to the mothur-formatted SILVA reference [34]. The resulting alignments were then manually checked for misalignments and gaps using BioEdit v7.2.5 (Ibis Biosciences, Carlsbad, California, USA). Singletons were removed. Final quality reads, associated with a bar-code, were then randomly re-sampled in order to have the same number of sequences per sample (3,800).

OTU and taxonomic analyses
Final quality reads were clustered into Operational Taxonomic Units (OTUs) at the 97% similarity level, which approximates genus for Bacteria [35,36], using the furthest-neighbor clustering method in mothur, followed by calculations of diversity indices (Shannon and Simpson), rarefaction curves and community similarity values. Bacterial OTUs were taxonomically identified using the SILVA taxonomy outlines and reference sequence set provided with mothur (www.mothur.org/wiki/Silva_reference_files), trimmed to the V6-V8 region.

Patient population
The severity of airflow limitation in the 9 recruited patients was as followed: GOLD 1, n = 1, GOLD 2, n = 3 and GOLD 3, n = 5. Patients' characteristics are presented in Table 1.

Pyrosequencing outcomes
A total of 142,786 raw reads were obtained from 16S rRNA gene pyrosequencing 18 DNA sputum samples from the 9 COPD patients. P6 was excluded from the microbiota profiling analysis because of low sequences numbers (225 sequences), however qPCR analysis were still performed. A total of 68,400 sequences (3,800 sequences per sample) were available for analysis. Overall, after normalization, 2,267 OTUs were identified.

Sputum diversity indices from COPD patients
Alpha-diversity was represented using the Simpson's diversity index (Fig 1). As shown in Fig  1, patient-specific variation in alpha-diversity nicely illustrates a difference from a clinically

Phylum diversity in sputum of COPD patients
At the phylum level, Firmicutes, Proteobacteria and Bacteroidetes were dominant during exacerbations with average proportions of 41%, 28%, and 25% per sample, respectively (data from our datasets, see Materials and methods for datasets info). Actinobacteria (3%), Fusobacteria (2%), Tenericutes, Spirochaetes, and unknowns (<1%) composed the remaining phyla found in the sputum samples. Slight but non statistically-significant phylum shifts were seen in the sputa of COPD patients during the exacerbation when averaged across all samples. However, when using the same approach as alpha-diversity, our paired-patient analysis showed heterogeneous shifts at the phylum level (Fig 2). As an example, the sputum of P4 showed an 82% increase in Firmicutes while there was a 48% decrease in the sputum of P7 during the exacerbation. P7 showed a clear shift from Firmicutes to Proteobacteria (75% increase, Fig 2B).

Microbiota diversity at the genus level in COPD patients
A total of 34 different genera were found in the sputa of COPD patients. Major genus proportional shifts were found during exacerbations compared to the stable states. Dominant and clinically-relevant genera are represented in Fig 3. The paired-patient analysis shows that Streptococcus (27%), Prevotella (23%), Moraxella (16%) and Veillonella (10%) were the dominant genera during the exacerbations (Fig 3A). The exacerbations showed an 88% increase in Streptococcus and a 90% increase in Moraxella in the sputum of P4 and P7, respectively. Clinically-important genera were also screened ( Fig 3B). Streptococcus (27%) and Moraxella (16%), previously described, were the 2 main clinically-important genera found in the sputum of COPD patients during exacerbations. A low proportion of the microbiota detected during the exacerbations was composed of Pseudomonas (1.8%) and Haemophilus (0.7%).

Bacterial and viral quantification in sputum of COPD patients
Mean total bacterial charge did not vary between stable (6.5×10 8 ±1x10 9 ) and exacerbated (5.8×10 8 ±9x10 8 ) states (Fig 4A). However P3 and P4 had a two log increase while P1 had a two log decrease in their total bacterial loads ( Fig 4A). Clinically-important pathogens were also detected in high concentrations during exacerbations (Fig 4B). The exacerbations induce a 4 log increase in S. pneumoniae and an 8 log increase in M. catarrhalis in P4 and P7, respectively ( Fig 4B). Our microbiota analysis did not suggest the importance of the genus Haemophilus and the same trend was observed between the qPCR and the microbiota analysis. However, H. influenzae was detected with qPCR in high concentrations in P6 (Fig 4B). Pseudomonas aeruginosa was not detected in any sputum samples with this specific set of primers during qPCR (detection limit = 10 copies/g of sputum), even if our microbiota analysis showed the presence of Pseudomonas (~15%) during exacerbation in the sputum of P5 (Fig 3B). This difference may come from the fact that it might be another pseudomonas species. Out of the 6 targeted respiratory viruses with qPCR, only Influenza B was found in the sputum of P2 in high concentrations during exacerbation, with 3.4×10 7 genome copies per gram of sputum. RSV A and B were detected in P7 and P4, respectively, with low genome copies per gram of sputum (1×10 3 , not shown). Influenza A, HRV and adenovirus were below detection limits (<10-100 genome copies/g of sputum) in all samples.

Comparison of sputum microbiotas in GOLD 2 and GOLD 3 patients during exacerbations
As shown in Fig 5A, diversity indices in GOLD 3 COPD patients were different from those of GOLD 2 patients. Shannon index (a community richness index) was significantly lower and Simpson index (representing evenness) tended to be higher in GOLD 3 patients. These results highlight a shift in microbiotas of GOLD 3 patients compared to GOLD 2. Firmicutes and Proteobacteria were detected in GOLD 3 compared to GOLD 2 COPD patients during exacerbations, while Bacteroidetes seemed to be lowered (data from our datasets, see Materials and methods for datasets info). Two clinically important pathogens were detected in GOLD 3 compared to GOLD 2 patients during exacerbations (Fig 5B). GOLD 3 patients P2 and P4 had increases in S. pneumoniae while P7 and P9 had increases in M. catarrhalis (Fig 5B).

Discussion
In this study, we evaluated the sputa microbiota of patients with COPD during stable conditions and subsequently when they were exacerbated.

Culture-independent detection of respiratory pathogens in COPD sputa
In our study, culture-independent methods were used and showed no difference in mean total bacterial loads in the sputa of COPD patients between stable conditions and during AECOPDs. Our data are, however, within the range of total bacteria in sputum [38]. In contrast, Garcha et al. showed an increased bacterial load in the sputum of exacerbated COPD patients compared to stable conditions [19]. The difference with our study could be related to the small number of samples and to heterogeneity in the behavior of bacterial loads during an AECOPD from one patient to another. For example, P4 had a 2 log increase in total bacterial load during the exacerbation period. Therefore, the finding that the mean total bacterial load did not change during an AECOPD should be interpreted with caution. S. pneumoniae, M. catarrhalis and H. influenzae are known respiratory pathogens linked to COPD exacerbations and they were detected in high concentrations in the sputa of our exacerbated COPD patients (up to 8 log increase compared to a stable state). These results emphasize that these specific respiratory pathogens were likely to have triggered the exacerbation of our COPD patients. P. aeruginosa was not detected in our study, probably because we did not enroll patients with very severe COPD [39]. Among all genera revealed with the pyrosequencing of 16S rRNA gene (E.g. Streptococcus), significant concentration of specific species were detected with qPCR (E.g. S. pneumoniae). These specie specific qPCRs data might not represent the whole population (E.g. streptococci) found in the microbiota data. Nevertheless, our qPCRs data seems to correlate (E.g. P4) with the microbiota data obtained.

Viral characterization during exacerbation in COPD sputum
The importance of viruses in COPD exacerbation has been previously underestimated, but several studies have recently highlighted their contributions [40][41][42][43]. Overall, key pulmonary viruses were mostly absent in the sputum of our COPD patients even if Influenza B was detected in one exacerbated patient. In a future study, random viral sequencing will be used instead of focusing on specific viruses. On average, patients waited six days following the onset of symptoms before visiting the clinic. This could have reduced the likelihood of identifying viruses in our specimens. The Influenza B positive patient was seen at the COPD clinic within three days of symptom onset, the recommended delay, suggesting that a shorter period between onset of symptoms and sputum collection might increase the odds of identifying viruses during AECOPD.

Sputum microbiota during AECOPD
COPD is a heterogeneous disease and this is further highlighted by the sputa microbiota at baseline and during AECOPD (Fig 2). As a result, it was not possible to draw a unique conclusion regarding the behavior of the sputa microbiotas during AECOPD. Nevertheless, using a paired-patient analyses, the identity of which microorganism may be responsible for that exacerbation in our COPD cohort could be established. However, we are aware that we cannot rule out the possibility that these shifts may be the results of normal temporal dynamics.
Three out of nine patients (P2, 4 and 5) had increases in Gram-positive bacteria (Firmicutes, which includes Streptococcus, Staphylococcus and Veillonella) during exacerbations, while three others (P7, 8 and 9) had their sputum enriched in Gram-negative bacteria (Proteobacteria, which includes Moraxella, Haemophilus and Pseudomonas). These population changes have also been shown in other studies for these same two phyla [2,15,27]. Two patients out of nine (P1 and 3) showed a different microbiota shift involving increases in Bacteroidetes (which includes Prevotella). At the genus level, P1 and P3 had an increase in Prevotella relative abundances during exacerbations. Prevotella, a Gram-negative anaerobic bacteria, has never before been linked to exacerbated episodes, even though it has already been linked to chronic bronchitis [44]. Perhaps, anaerobic bacteria may be underestimated in the onset of COPD exacerbations. Overall, microbiota shifts were seen in all COPD patients during exacerbation and interestingly our specific pulmonary pathogens detected with qPCRs correlate nicely with the microbiota findings.

Clear microbiota shifts in the sputa of GOLD 3 COPD patients
Studying the sputum microbiotas and confirming the results with qPCR, microorganisms that were likely responsible for the AECOPDs were identified. In fact, the pathogens detected in high concentrations in the sputa during AECOPDs were either bacterial (S. pneumoniae (P2, P4) or M. catarrhalis (P7,P9)) or viral (Influenza B (P6, GOLD 1)). Interestingly, four of these patients are classified as GOLD 3 (P2, 3, 4, 7 and 9). Shannon diversity index decreased in the sputa of GOLD 3 patients during exacerbation compared to GOLD 2 patients (P1, 5 and 8; Fig  5A). Perhaps additional studies are required to clearly evaluate the impact of disease severity on the microbiota. Severe COPD state could predispose patients to a more pathogenic microbiota instead of the actual microbiota inducing disease severity.

Study limitations
One limitation of this pilot study is the small number of evaluated patients, a critique that also applies to other studies on the same topic [45]. However, this limitation was somewhat mitigated by a strong study design involving pair-wise comparisons, enabling us to study potential shifts in the sputa microbiotas of each individual patient. As previously mentioned, COPD patients waited, on average, six days before visiting the COPD clinic. Shortening this delay may result in better viral characterization during exacerbations.

Conclusions
Little is known about the comparative microbiotas of stable and exacerbated COPD patients using a non-invasive method. Our study shows that the sputum microbiotas of COPD patients is complex due to the heterogeneous nature of COPD and the small number of evaluated patients in our study, it is not possible to make general treatment recommendations for AECOPD. Nevertheless, our data contributes to a better understanding of which microorganisms could be induced during exacerbations. As stated before, we cannot rule out the possibility of random fluctuations over time. Our study, however, clearly shows dramatic microbiota shifts occurring during AECOPD, mostly as it relates to Firmicutes or Proteobacteria. However, the nature of these shifts varies from patients to patients in such a way that the treatment should be patient-specific. Additional studies are required to further investigate the impact of microbial diversity in COPD patients.
Supporting information S1