This file contains Tables A to F and Figs. A to K.

Table A, Comparison of data sets from bacterial community profiling. Resemblance matrices were calculated for each dataset and compared using spearman rank correlation (rho). Major discrepancies observed for each pair of data sets are mentioned. For the comparison of SSCP fingerprinting and deep sequencing, only OTUs with a relative abundance of ≥ 5% were considered for this statistical comparison. Similarity was measured by Spearman’s rho and data sets are permuted. Table B, Summary of parameters characterizing the adult cohort of CF patients and the alpha diversity of the bacterial communities. Richness and Shannon diversity index were calculated based on the relative abundances of single OTUs based on NGS sequence reads as detailed in the Materials and Methods. Table C, OTUs observed with deep sequencing exhibiting a minimum relative abundance of 0.5%. Sequence similarity revealed by BLAST is given as well as the closest representative (accession number and name) for each OTU. Comparison with SSCP fingerprinting allowed phylogenetic affiliation of some previously ambiguous OTUs to species level. Streptococcus species identified by SSCP are mentioned with sequence identity scores in percent as well as accession numbers of closest representatives in the end of the table. For Illumina-based data, number of positive samples for each OTU is given as well as minimum and maximum relative abundances in individual samples. Table D, DNA retrieved from 300µl CSF sputum after extraction from boiled vs non-boiled samples (cellular fraction retrieved from the pellet). Table E, Primers used for library preparation in Illumina-based sequencing. Underlined letters denote complementary sequences for the variable region V3 of the 16S rRNA gene. Eleven different barcodes were used, each indicated with bold letters within the primer sequence. In a second PCR, amplicons for libraries were accomplished by adding Illumina-specific indices. Table F, Consensus sequences of for all major OTUs obtained by NGS sequencing of the 16S rRNA gene. Fig. A., Comparison of prevalence of bacteria in the cohort observed by deep sequencing (NGS, white bars) and SSCP fingerprinting (black bars). OTUs were identified to the genus level and listed accordingly on the left. Streptococci were additionally distinguished between alpha-hemolytic streptococci (Streptococcus) and the Streptococcus millerii group. For each diagnostic method, only OTUs with ≥ 5% relative abundance were taken into account for this comparison. Fig. B, Phylogeny of detected Streptococcus species in sputum samples. High sequence variation was observed in bands from SSCP fingerprints. Closest described representatives from databases are given in the taxonomic tree. Associated OTUs defined by deep sequencing are indicated with brackets. Aerobic alpha-hemolytic streptococci are further marked with bold lines. Staphylococcus aureus was included for rooting of the tree. Scale bar represents base substitution per site. Fig. C, Comparisons of the relative abundance of the most dominant OTU in each individual sample assessed with deep sequencing (NGS, white bars) and SSCP fingerprinting (black bars). Sputum samples are labelled by individual numbers and mentioned on the left. The samples are sorted by decreasing similarity of the relative abundances of OTUs between the two molecular methods. Fig. D, Correlation of richness with Shannon diversity index based on the NGS sequence abundance data of all 56 patients (n = 56). Richness and Shannon diversity index are calculated based on the relative abundances of single OTUs based on NGS sequence reads as detailed in the Materials and Methods. The logarithmic regression provided the best fit to the data set. A comparable logarithmic correlation was observed when the data of the longitudinal observations were included (n = 72). Fig. E, Comparison of lung function with Shannon diversity index of all CF patients (n = 55). The Shannon diversity was calculated as in indicated in Fig. D in S1 File. Standard linear and non-linear correlation analyses did not show a clear relationship. Fig. F, Comparison of age with Shannon diversity index of all CF patients (n = 56). The Shannon diversity was calculated as in indicated in Fig. D in S1 File. Standard linear and non-linear correlation analyses did not show a clear relationship. Fig. G, Comparison between age of patients (white bars) and relative abundance of P. aeruginosa (black bars). Samples are sorted by age of the patient at the individual time point of sputum collection. Age of patient is given in years and relative abundance of P. aeruginosa is given in % on y-axis. For each patient, both parameters are indicated. Missing bars for P. aeruginosa indicate its absence in the sample. A median relative abundance of 24.8% was calculated for the bacteria. Median age at time point of sputum collection was 31 years. Fig. H, MDS plot from bacterial community composition observed in sputum samples with superimposed antibiotic treatment. Six categories of antibiotic treatment were defined and are indicated with individual symbols. Antibiotics were given by inhalation (inh.), orally (p.o.), or intravenously (i.v.) within past 14 days before sampling of sputum or by none of these categories (other). MDS plot is based on the respective plots in Fig. 3. Fig. I, Comparison between relative abundance of P. aeruginosa (grey bars) and lung function of the patient (black bars). Samples are sorted by relative abundance of P. aeruginosa. Lung functions of the patients at the individual time point of sputum collection is measured by the predicted FEV1 value. Both, relative abundance of P. aeruginosa and lung function are given in % on y-axis. For each patient, both parameters are indicated. Missing bars for P. aeruginosa indicate its absence in the sample and missing bars for FEV1 indicates no measurement for the patient at time point of sputum collection. A median FEV1 of 35% was calculated for the cohort. Fig. J, Dynamics of community compositions from 13 CF patients from which sputum samples were obtained twice or three times. Each OTU is indicated by a specific colour and further defined in the legend. CF patients are identified by numbers and sputum samples are shown in chronological order, hereby, time intervals to the initial sample are indicated in the figure. Fig. K, 16S rRNA gene based community fingerprint by SSCP for boiled (B1, B2, B3) vs. non-boiled samples (U1, U2, U3). B1 and U1 in the right part of the gel was a repetition of the single strand preparation. (St = Standard using 5 different bacterial species).
(DOC)