Impact of Enhanced Staphylococcus DNA Extraction on Microbial Community Measures in Cystic Fibrosis Sputum

Staphylococcus aureus is a common constituent of the bacterial community inhabiting the airways of persons with cystic fibrosis (CF). Culture-independent studies have shown that this species is often present in relatively high abundance and would therefore be expected to exert a pronounced effect on measures of CF airway bacterial community structure. We investigated the impact of DNA extraction method on pyrosequencing-based measures of Staphylococcus abundance and bacterial community structure in 17 sputum samples from five CF patients. Staphylococcus was detected in fewer samples when DNA was extracted using a standard bacterial lysis method compared to when DNA was extracted using a lysis buffer amended with lysostaphin and lysozyme. The standard lysis method resulted in significantly lower measures of Staphylococcus relative abundance and higher levels of community diversity, richness, and evenness compared to the lysostaphin-lysozyme modified method. Measures of community dynamics in serial sputum samples from the same individual were nevertheless highly concordant between the two DNA extraction methods. These results illustrate the impact of DNA preparation method on measures of Staphylococcus abundance and bacterial community structures in studies of the airways microbiota in CF.


Introduction
Cystic fibrosis (CF), an autosomal recessive genetic disease affecting more than 70,000 people worldwide, is caused by mutations in the CF transmembrane conductance regulator gene. The associated alterations in respiratory tract mucociliary clearance and airway surface fluid composition result in chronic infection and inflammation of the airways, leading to progressive lung disease and respiratory failure [1]. Conventional culturebased microbiology has identified a number of bacterial species that commonly infect the airways of persons with CF, including Pseudomonas aeruginosa, Staphylococcus aureus, Burkholderia cepacia complex, Stenotrophomonas maltophilia, and Achromobacter xylosoxidans [2].
Among these species, S. aureus typically infects patients at an early age, being most frequently isolated from infants and children with CF [2,3]. Recent studies using culture-independent approaches to assess the microbiota of CF airways reiterate that this species is a common constituent of the bacterial communities typically found in CF [4,5,6,7]. As such, it is reasonable to expect that the abundance of Staphylococcus in airway communities would have significant bearing on measures of bacterial community structure and dynamics. However, the isolation of Staphylococcus DNA is hampered by the rigid cell wall that is characteristic of Gram positive bacteria. In preliminary experiments assessing the airways microbiome in CF, we found that the combination of lysostaphin and lysozyme increased the yield of Staphylococcus DNA from sputum samples compared with a standard bacterial cell lysis buffer. We therefore wished to further examine the impact of DNA preparation method on pyrosequencing-based measures of Staphylococcus abundance and overall bacterial community structure in sputum specimens from patients culturepositive for S. aureus.

Results
Impact on Staphylococcus prevalence, relative abundance and community diversity, richness and evenness Seventeen sputum samples from five CF patients were selected for study. S. aureus had been cultured from each sample, with recovery graded as rare, few, moderate or numerous by the University of Michigan Health System clinical microbiology laboratory (Table 1). DNA was purified in duplicate from each sputum sample: one preparation used a standard lysis buffer while the other used the same buffer with the addition of lysostaphinlysozyme. All 34 DNA preparations were subjected to bar-coded pyrosequencing of a portion of the 16S rRNA V3-V5 region using protocols developed for the Human Microbiome Project; sequences were analyzed using the software package mothur v1.21 [8] (see Materials and Methods for details).
16S rRNA gene sequences corresponding to Staphylococcus were retrieved from all sputum samples that had been graded as having numerous, moderate or few S. aureus in culture, irrespective of DNA preparation method (Table 1). However, Staphylococcus was detected in none of the four sputum samples graded as having rare growth of S. aureus in culture when DNA was prepared using the standard lysis buffer. In contrast, Staphylococcus was detected in three of these four sputum samples when DNA was prepared using the lysostaphin-lysozyme amended buffer.
In the six sputum samples that were graded as having moderate, few or rare growth of S. aureus in culture, the relative abundance of Staphylococcus was ,1%, regardless of DNA preparation method ( Table 1). In one of the 11 sputum samples graded as having numerous S. aureus in culture (sample 16), the relative abundance of Staphylococcus was ,1% in the sample prepared with standard lysis buffer and 2% in the respective sample prepared with the lysostaphin-lysozyme amended buffer. Together, these seven sputum samples are hereafter referred to as ''Staphylococcus -poor'' samples.
In each of the remaining 10 ''numerous S. aureus in culture'' sputum samples (hereafter referred to as ''Staphylococcus -rich'' samples), the relative abundance of Staphylococcus in the samples prepared with the lysostaphin-lysozyme amended lysis buffer (71%; range 37%-98%) was significantly higher than that in the respective sample prepared with standard lysis buffer (27%; range 4%-75%; p,0.01, paired t-test) ( Table 1 and Figure 1A). For these 10 sputum samples, measures of bacterial community diversity, richness (total number of taxa), and evenness (relative abundance of taxa) were significantly higher in DNA samples prepared with standard lysis buffer compared with DNA samples prepared with lysostaphin-lysozyme amended buffer (Table 1, Figure 1B, 1C and 1D; p,0.05, paired t-test). In contrast, these measures showed little difference between DNA preparation methods in the seven pairs of Staphylococcus -poor samples (p.0.05, paired t-test).
Predictably, the increased relative abundance of Staphylococcus in the samples prepared with lysostaphin-lysozyme resulted in the decreased relative abundance of other taxa in these samples (ie, compared to the same sample prepared with standard lysis buffer). Consequently, some taxa that were found in low abundance (mean relative abundance 0.4%; range 0.2%-2.4%) in samples prepared with standard lysis buffer were undetectable in samples prepared with lysostaphin-lysozyme. For example, an OTU representing Prevotella was detected in all 5 samples from Patient 1 when prepared with the standard lysis buffer, but in only 3 of these 5 samples when prepared with the lysostaphin-lysozyme amended lysis buffer. Similarly, an OTU representing Fusobacterium was Abundance of S. aureus in culture (reported by clinical microbiology laboratory). 2 Relative abundance by pyrosequencing without (2LY) and with (+LY) lysostaphin-lysozyme; no. Staphylococcus sequence reads/no. total sequence reads. 3 Total number of OTUs observed after normalization of sequence reads to 498, the smallest number of sequences obtained among the 34 samples. 4 The rank order of the relative abundance of Staphylococcus. 5 With the exception of sample 16 detected in all 6 samples from Patient 2 when prepared with standard lysis buffer, but in only 4 samples prepared with the amended buffer. In no cases did the increase in relative abundance of Staphylococcus result in failure to detect Pseudomonas in samples prepared with the amended lysis buffer.

Impact on community structures
In addition to the effects on measures of alpha diversity, DNA extraction method also had a marked effect on measures of beta diversity among the Staphylococcus-rich samples. This was evident in the Bray-Curtis distances between communities observed in each member of the 17 pairs of DNA samples. The mean Bray-Curtis distance within pairs of DNA samples from the 10 Staphylococcus -rich sputum samples (0.46) was significantly higher than the mean distance within pairs of DNA samples from the seven Staphylococcus -poor sputum samples (0.13) (p,0.01, twotailed t-test). Bray-Curtis distance-based nonmetric multi-dimensional scaling (NMDS) highlighted these differences (Figure 2A). DNA extraction method had little effect on the Staphylococcus-poor samples, with paired communities residing close to each other on the ordination plot. In contrast, among the Staphylococcus-rich samples, communities in samples extracted with lysostaphinlysozyme were more separated from their respective communities in samples prepared with the standard DNA extraction method.
NMDS indicated that these shifts were driven by Staphylococcus (Figure 2A).

Impact on community dynamics
To investigate the impact of DNA extraction method on bacterial community dynamics, Bray-Curtis distances between five sputum samples obtained from Patient 1 during a 15 month period and six sputum samples obtained from Patient 2 during a 50 month period were assessed using NMDS ( Figure 2B). The two DNA extraction methods resulted in similar community structure dynamics in both patients, as reflected by similar movements on the ordination plot. Procrustes analysis, a statistical shape analysis approach, confirmed that the movement of samples on the ordination plot revealed by the two methods were highly correlated (r = 0.83, p = 0.001). These correlations were also shown by the Mantel test comparing Bray-Curtis distance matrices (Patient 1: Mantel r = 0.77, p = 0.04; Patient 2: Mantel r = 0.91, p,0.01).

Discussion
Previous studies of Staphylococcus have enhanced bacterial cell lysis by the addition of various reagents to extraction buffers, including lysostaphin, proteinase K, and/or detergents such as sodium dodecyl sulfate (SDS) [9]. In this study we chose the combination of lysostaphin and lysozyme, which provided considerably greater yield of Staphylococcus DNA than did the use of a standard lysis buffer in preliminary experiments. Although the impacts of this modification on the detection of Staphylococcus and measures of the relative abundance of this and other taxa in CF sputum samples are apparent and predictable, these findings have important implications for studies of the CF airway microbiota.
Zemanick and colleagues [10] recently found that culture was more sensitive in detecting S. aureus in CF respiratory specimens than was a species-specific quantitative PCR (qPCR) assay. DNA was extracted from specimens with a buffer containing 5.7% SDS, but no lysostaphin or lysozyme. For specimens with relatively low quantities of S. aureus (10 3 -10 5 cfu/ml) in culture, the sensitivity of qPCR for detecting this species was only 22%; in specimens with greater quantities of S. aureus (.10 5 cfu/ml) in culture, the sensitivity of qPCR was 87%. We similarly found a lower rate of detection of Staphylococcus using a culture-independent approach compared to culture. With a standard DNA extraction protocol, 13 (76%) of 17 S. aureus culture-positive specimens yielded pyrosequencing reads classified as Staphylococcus. However, with the addition of lysostaphin and lysozyme to the DNA extraction buffer, the sensitivity of pyrosequencing in detecting Staphylococcus increased to 94%.
We further investigated the impact of DNA extraction method on overall measures of bacterial community structure in CF sputum. The addition of lysostaphin-lysozyme resulted in measures of greater Staphylococcus relative abundance and lowered measures of the relative abundance of other taxa, as well as estimates of overall species richness. Considering the prominent role that S. aureus, including methicillin-resistant S. aureus, is believed to play in CF lung disease [2,11] these findings deserve careful consideration when designing studies using culture-independent approaches to investigate the microbiome of the CF airways. This is especially relevant in studies involving children insofar as the prevalence of S. aureus infection in CF is highest among patients aged 6 to 10 years [3]. Mild pulmonary disease and lack of sputum production are common in this age group, necessitating the use of throat swab cultures to assess airway bacterial communities. Relative to culture, the sensitivity of culture-independent methods to detect S. aureus is considerably lower with oropharyngeal swab specimens than with expectorated sputum samples [10].
In summary, DNA preparation method must be taken into account when considering studies describing rates of Staphylococcus detection based on culture-independent methods. Enhancing Staphylococcus cell lysis and DNA extraction would be desirable in studies in which determining the prevalence and/or relative abundance of Staphylococcus is important. In contrast, in studies focusing on the correlation between specific taxa, particularly low abundance species, and disease or when more accurate estimates of total microbial community richness are sought, decreased estimates of the relative abundance of Staphylococcus may be warranted. Importantly, despite the marked effect of DNA extraction method on alpha and beta diversities, especially in Staphylococcus-rich samples, the community dynamics revealed by the two methods are highly correlated.

Ethics Statement
This study was approved by the University of Michigan Institutional Review Board (HUM00003517). As this study was limited to a retrospective analysis of existing unique-identifierencoded (to protect patient anonymity) biologic specimens and clinical data, waiver of subject informed consent was granted.

Sputum samples
Sputum specimens, collected from CF patients during the course of routine care, were obtained from the University of Michigan Health System clinical microbiology laboratory. After removing an aliquot of the sample for bacterial culture, the remainder of the unadulterated sample was recovered, divided into 0.5 ml aliquots, and stored at 280uC until processing for DNA extraction.

DNA extraction
Samples from frozen stock were thawed on ice and incubated with an equal volume of 10% Sputolysin (EMD Chemicals, Gibbstown, NJ) at 37uC for 30 min with pulse vortexing at 5 min intervals. Samples that were not of uniform consistency after 30 min were mechanically homogenized for 10 sec with a tissue homogenizer (Omni International, Kenneshaw, GA). Samples were then mixed with 0.9 volume of MagNA Pure Bacterial Lysis Buffer (Roche Applied Science, Indianapolis, IN) either with or without the addition of lysostaphin (final concentration 0.14 mg/ml, Sigma-Aldrich Corp., St. Louis, MO) and lysozyme (final concentration 2.9 mg/ml, Sigma-Aldrich Corp.). Lysostaphin-lysozyme containing preparations were incubated another 30 min at 37uC. All samples were then transferred to microfuge tubes containing garnet beads (MO BIO Laboratories, Inc., Carlsbad, CA) and agitated in a Mini-Beadbeater-8 (Biospec Products Inc., Bartlesville, OK) for 1 min on maximum setting. After digesting with Proteinase K (final concentration 1.4 mg/ml) for 10 min at 65uC, samples were again agitated in the Beadbeater and incubated at 95uC for an additional 10 min. DNA was purified using a MagNA Pure Compact System (Roche) according to the manufacturer's DNA Bacteria v3.1 protocol. As a negative control, the same procedure was used with sterile water replacing sputum in each extraction; no PCR products were detected in any experiment, indicating lack of contamination of any of the reagents used. Barcode and adaptor sequences are found at http://www.hmpdacc. org/doc/HMP_MDG_454_16S_Protocol_V4_2_ 102109.pdf. PCR amplification was performed on 2 ml of DNA template in a total volume of 25 ml containing 16 AccuPrime Buffer II (Invitrogen Corp., Carlsbad, CA), 320 mM of each primer, and 0.03 U/ml AccuPrime High Fidelity Taq polymerase. Reactions were heated at 95uC for 2 min followed by 30 cycles of 95uC for 20 sec, 50uC for 30 sec, and 72uC for 5 min. The concentration of amplicons in each reaction was determined in triplicate using the PicoGreen fluorescent assay (Invitrogen Corp.), and amplicons were pooled in groups of 96 in equal proportion before being sequenced via 454 FLX-titanium according to manufacturer's specifications. A water sample was included in each pyrosequencing run as a negative control.

Sequence processing and statistical analysis
Raw sequences were analyzed using mothur v1.21 [8] to remove sequences containing homopolymers greater than 8 bp, mismatches in the barcode or primer, one or more ambiguous bases, or an average quality score below 35 over a moving window of 50 bp. Remaining sequences that were at least 200 bp but less than 590 bp in length were further curated to remove chimeric sequences using UCHIME [12] and to reduce sequencing noise by a preclustering methodology [13] before being assigned to operational taxonomic units (OTUs) using an average neighbor algorithm with a 0.03 dissimilarity cutoff. The consensus taxonomy of each OTU was identified at the genus level using the Bayesian method [14]. The total number of reads for each community was normalized to 498, the smallest number of reads among the samples included in the study, to control for differences in sequencing depth before alpha and beta diversity measures were calculated. Community diversity was measured using non-parametric Shannon indices [15]. The number of observed OTUs was used as a measure of community richness. Community evenness was measured with Shannon indices-based measure of evenness. Beta diversity was measured using Bray-Curtis dissimilarity coefficients.
Procrustes analysis was used to compare the correlation of NMDS configurations by scaling and rotating one to the other. The statistical significance of the correlation between the two configurations was calculated by a permutation procedure (protest function) [16]. The Mantel test, used to test the correlation between Bray-Curtis based community measures, was performed with 10000 permutations measuring the Spearman correlation. Both the Procrustes analysis and the Mantel test were performed using the R package vegan (http://CRAN.R-project.org/package = vegan). Paired t-test was used to compare the relative abundance of Staphylococcus, and community diversity, richness, and evenness between sets of samples. Two-tailed t-test was performed to compare the average Bray-Curtis distances between the Staphylococcus-rich and Staphylococcus-poor samples. A p,0.05 was considered statistically significant.