Supplementary Information s66

1Supplementary Information

2Sputum samples were collected in the UCSD Adult CF clinic, and microbial DNA was obtained

3following the procedure described (Lim et al 2012). The samples were sequenced using the Ion

4Torrent platform. Sequences were preprocessed according to Ion Torrent quality scores. High

5quality sequence was retained using Prinseq (Schmieder and Edwards 2011a) with the following

6command:

7perl prinseq-lite-0.20.2/prinseq-lite.pl -verbose -trim_left 1 -trim_right 1 -trim_to_len 660 8-derep 1245 -lc_method entropy -lc_threshold 50 -trim_qual_right 15 -trim_qual_left 15 9-trim_qual_type mean -trim_qual_rule lt -trim_qual_window 2 -trim_qual_step 1 -trim_tail_left 105 -trim_tail_right 5 -min_len 60 -min_qual_mean 15 -ns_max_p 1 -rm_header -fastq file.fastq 11-log -out_bad null -out_good 12

13Human sequences were removed from the data using Deconseq (Schmieder and Edwards 2011b)

14with the following command:

15perl deconseq.pl -i 94 -c 90 -f filename -id deconseq_filename -dbs 16hsref,hs1,hs2,hs3,hs4,hsunique -dbs_retain vir,bact -z 3 17 18Regression of Streptococcus species abundance vs 2,3-butanedione concentration was conducted 19in R with the following code and outputs: 20 21 22linear model formula = Medianppt ~ Streptococcus 23 24Residuals: 25 1 2 3 4 5 6 7 26 366.7 -480.2 313.2 -200.5 166.2 363.1 -528.4 27 28Coefficients: 29 Estimate Std. Error t value Pr(>|t|) 30(Intercept) -11.18 271.94 -0.041 0.9688 31Streptococcus 53.93 13.58 3.971 0.0106 * 32--- 33Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 34 35Residual standard error: 434.1 on 5 degrees of freedom

1 36Multiple R-squared: 0.7592, Adjusted R-squared: 0.7111 37F-statistic: 15.77 on 1 and 5 DF, p-value: 0.01063 38Ferritin transcription and iron availability in CF sputum:

39In oxygen depleted CF sputum, transcription of the iron storage protein ferritin provides

40evidence for the presence of an alternative electron acceptor. Abnormal iron and ferritin levels

41are an important part of the pathogenesis of many diseases, not just those specific to iron

42metabolism, including diseases characterized by infection and inflammation (Torti and Torti

432002). Increased iron and ferritin concentrations have been found in CF sputum in several

44studies (Ghio et al 2012, Gifford et al 2012, Reid et al 2007, Stites et al 1998). Iron may be

45liberated from ferritin by microbial proteases (Whitby et al 2006) or neutrophil elastase (Fischer

46et al 2009). In addition to the generation of destructive reactive oxygen species, iron can act as a

47final electron acceptor in anaerobic respiration, and could cycle the redox state of phenazines.

48Transcription of human ferritin is a marker of oxidative stress because it is induced via a defense

49mechanism that is activated by redox changes and other toxic molecules (Mackenzie et al 2008,

50Torti and Torti 2002, Tsuji et al 2000). Because these conditions are likely to exist in the CF lung

51environment, we searched metatranscriptome data from the San Diego CF cohort (Lim et al

522012) for human ferritin transcripts. Compared to control genes that are consistently transcribed

53across tissues, ferritin had one to two orders of magnitude more reads in the CF sputum

54metatranscriptome (Supplemental Figure 3). Thus, the transcription of ferritin, potentially in

55response to the redox state of the CF lung, supports the possibility that microbes in the CF lung

56have access to alternative electron acceptors for anaerobic respiration in the form of Fe3+, and

57phenazines that recycle their redox state through donation of electrons to Fe3+.

2 2 59 60Supplementary Figure Legends

62Supplementary Figure 1. Taxonomy of blast hits to genes involved in 2,3-butanedione

63metabolism (6278 hits) and phenazine synthesis (4109 hits) from 654,000 Roche 454 reads from

6418 samples from 6 CF patients at between 2 and 4 time-points each (Lim et al 2012).

65Supplementary Figure 2. Fraction of metagenomic reads assigned to several Streptoccocus spp.

66in three time-points from patient CF1.

67Supplementary Figure 3. Ferritin transcript levels in CF metatranscriptomes (Lim et al. 2012)

68revealed a higher abundance of ferritin heavy chain (FTH1) and to a lesser extent ferritin light

69chain (FTL) mRNA in comparison to evenly expressed mRNA used as a marker in transcriptome

70studies (PGK1 and PPIA).

71Supplementary Figure 4. Linear model of Streptococcus abundance vs 2,3-butanedione

72concentration (ppt) from the seven CF patients in the cross-sectional study shows that a

73significant amount of variation in 2,3-butanedione production is explained by differences in

74Streptococcus abundance between patients. A) 2,3-butanedione vs Streptococcus abundance plot;

75B) Residuals vs fitted values from the linear model in R; and C) the standardized residuals vs

76theoretical quantiles (Normal Q-Q plot) from the linear model.

3 3 78Supplementary Tables 79 80Supplementary Table 1. BLASTn hits (minimum 60bp and 90% identity and E value of 110- 8110) to genes encoding enzymes for acetoin anabolism (budABC) and catabolism (acoA and acoB) 82in CF metagenomes from (Lim et al 2012). budB budB budA budC acoA acoB Total large small Streptococcus 102 47 33 32 61 60 spp. 335 Other 15 6 0 0 0 1 22 E. coli 44 3 0 0 0 0 47 P. aeruginosa 18 7 0 0 7 7 39 S. maltophila 23 0 0 0 0 0 23 Total 202 63 33 32 68 68 466 83

4 4 84 85 86Supplementary Table 2: Percentage of CF sputum samples containing Streptococcus spp. and 87Pseudomonas spp. in several recent culture-independent 16S rRNA sequence studies of CF mi- 88crobial communities 89 Study Hunter Zhao Delhaes Fodor Goddard Filkins Lim 2012 2012 2012 2012 2012 2012 2012 (Hunter (Zhao (Delhaes (Fodor (Goddard (Filkin (Lim et al et al et al et al et al s et al et al 2012b) 2012) 2012) 2012) 2012) i 2012) 2012) Streptococcu 86 >70 87 95 ~44 79 89 s spp. (%) Pseudomona 97 >97 62 >78 >90 94 56 s spp. (%) # CF 36 6 4 23 15 35 6 Patients Samples 36 126 8 63 43 35 18 90i Lung ex-plant, sputum and mouthwash samples 91

5 5 92 93 94Supplementary Table 3. 2,3-butanedione concentrations from CF patient, healthy control, and 95room samples during 7 time-points (shown in Figure 2). 96 97 2,3-butanedione (ppt) for patient shown in each column Timepoint Day CF1 CF1 CF1 Healthy Room (H5) A 1 4336 4437 1482 261 248 B 36 2850 2537 1861 495 3 C 71 264 333 284 342 596 D 113 5631 4851 4032 388 262 E 142 12544 5335 17270 2287 485 F 162 2235 11802 10867 945 178 G 169 22234 26751 14445 701 241 98

6 6 99 100Supplementary Table 4. 2,3-butanedione concentrations from cross-sectional study of seven CF 101patients, four healthy controls and room samples. 102 patient 2,3-butanedione (ppt) Type Status CF1 2201 CF Suppressive CF1 1370 CF Suppressive CF1 3133 CF Suppressive CF1 2781 CF Suppressive CF2 1008 CF Suppressive CF2 37 CF Suppressive CF2 774 CF Suppressive CF2 562 CF Suppressive CF3 749 CF Suppressive CF3 599 CF Suppressive CF3 1612 CF Suppressive CF4 1124 CF Suppressive CF4 953 CF Suppressive CF4 743 CF Suppressive CF4 87 CF Suppressive CF5 1681 CF Acute CF5 1007 CF Acute CF5 666 CF Acute CF6 81 CF Acute CF6 167 CF Acute CF6 143 CF Acute CF6 273 CF Acute CF7 23 CF Acute CF7 30 CF Acute CF7 34 CF Acute CF7 609 CF Acute H1 43 Control Control H1 347 Control Control H2 132 Control Control H2 357 Control Control H2 1905 Control Control H2 291 Control Control H2 201 Control Control H2 138 Control Control H2 539 Control Control H2 716 Control Control H3 97 Control Control H3 496 Control Control

7 7 H4 3750 Control Control H4 4216 Control Control Room 326 CF1 Room Room 16 CF2 Room Room 10 CF3 Room Room 11 CF5 Room Room 10 CF6 Room Room 30 CF7 Room 103

8 8 104 105Supplementary Table 5. Consequences of 2,3-butanedione (diacetyl) exposure in a wide variety 106of systems. 107 Location Source Concentration Consequence Reference Human lung Popcorn Lung destruction (Egilman and flavoring or (Obliterative Schilling 2012) diacetyl factory Bronchiolitis) Sprague- Purified dose 203–371 ppm Nasal and lung (Hubbs et al Dawley rat diacetyl for 6 hours airway epithelia 2002) lungs (comparable to destruction and factory levels) necrosis C57Bl/6 Purified dose 100 ppm diacetyl 6 Injury to the (Morgan et al mice h/day 5 days/week epithelium of the 2008) for 12 weeks nasal cavity and upper airways Food additive Harvested from Kill gram negatives (Jay 1982) as Lactobacillus preservative Wine, butter, Lactobacillus Flavor Multiple sources popcorn 1900-present (Bartowsky and Henschke 2004) Mosquito Purified dose Firing of CO2 (Turner et al nose detection neurons, 2011) lose ability to use [CO2] to detect prey Microbial Co-cultured B. Nanomolar to Changes in global (Kim et al 2013) culture subtilis with E. micromolar gene expression coli, and purified related to motility dose and antibiotic resistance 108

9 9 109 110Supplementary Table 6. Consequences of 2,3-butanediol exposure in a wide variety of systems. 111 Location Source Concentration Consequence Reference Bioreactor E. aerogenes Stimulation of (Venkataraman with solid- pyocyanin et al 2011) state production in P. electrodes as aeruginosa, electron anaerobic acceptors (microaerobic) respiration Epithelia cells Vibrio cholera Suppression of (Bari et al inflammatory 2011) signals Rats Purified dose Inhibit neutrophils (Hsieh et al 2007) Arabidopsis Root bacterial Trigger Induced (Han et al community, Systemic 2006) purified dose or Resistance (ISR) B. subtilis and protection from root rot Tobacco plant Pseudomonas Identified 2,3- Enhances aerial (Ryu et al chlororaphis butanediol as the growth, trigger 2004) and purified molecular trigger of plant defense dose ISR, and tried with system through purified R-R-2,3-bd GacS sensor kinase Potato and Dickeya and 40umol/g in Massively (Effantin et al chicory Pectobacteria infection, ND in produced 2,3-bd in 2011) healthy plant tissue soft rot infection, bd metabolism mutants not as destructive 112

10 10 113 114Supplementary Table 7. Ion Torrent sequence quality filtering. Samples are labeled with an 115anonymous patient id (i.e. CF1) followed by a letter indicating the time-point and an abbrevia- 116tion for their disease state (St = Stable, Tr = Treatment and Pt = Post Treatment). 117 Sample Input sequences After Prinseq % length duplicates After Deconseq

1 CF1-L-St 1,780,819 1,494,421 84% 133 57,857 1317296

2 CF2-E-St 1,973,912 1,593,530 81% 150 158,584 925360*

3 CF3-A-St 1,780,474 1,436,594 81% 144 154,059 1,417,311

4 CF4-A-St 5,877,520 4,334,671 74% 139 987,405 4,762,081

5 CF5-A- 1,162,089 964,098 83% 144 59,996 549831 Tr 6 CF6-A-St 1,836,488 1,471,073 80% 136 171,483 1356155

7 CF7-A- 1,428,012 1,593,530 81% 150 580,864 1139292 Pt 8 H2-A 3,956,658 2,992,662 76% 145 476112 2096847

TOTAL 19,795,972 15,880,579 2,646,360 13,564,173

118*half of total available data processed

11 11 119 120Supplementary Table 8. Summary of clinical information, 2,3 butanedione concentration (ppt) 121and abundance of three most prevalent bacteria for CF patients 122 patient CF1 CF2 CF3 CF4 CF5 CF6 CF7 age 40 39 27 52 31 21 44 Rothia 16.6 68.9 79 4.6 73.7 0.0 15.5 Pseudomonas 11.8 1.1 0 77.6 0.0 31 0.0 Streptococcus 39.6 21.5 8 9.2 22.6 0.0 10.6 Acute antibiotic treat- 0 0 0 0 1 1 1 ment Days since acute 62 108 368 315 0 0 5 abx treatment Medianppt 2491 668 749 848 1007 155 32 123

12 12 124 125 126Supplementary Table 9. Pearson correlations between data from Supplementary Table 8, the 127clinical information, 2,3 butanedione concentration (ppt) and abundance of three most prevalent 128bacteria for CF patients 129 pearson correlations with cor() in R age Roth- Pseu- Strep- Acute Days 2,3bd ia do to abx since (ppt) monas coccus abx median age 1 -0.3 0.43 0.28 -0.38 0.21 0.19 Rothia 1.00 -0.63 0.20 -0.19 0.23 0.03 Pseudomonas 1.00 -0.32 -0.23 0.39 -0.02

Streptococcus 1.00 -0.35 -0.24 0.87

Acute antibiotic treatment 1.00 -0.73 -0.52 Days since acute abx treatment 1.00 0.05 Median 2,3bd (ppt) 1.00 130

13 13 131 132Supplementary Table 10. Taxonomic abundance of each sample from cross-sectional study de- 133termined by Metaphlan (results shown in Figure 4). 134 genus 1 Sta- 2 Sta- 3 Sta- 4 Sta- 5 Sta- 6 7 Post- CF aver- Healthy Pooled ble ble ble ble ble Tr. Tr age CF Rothia 16.6 68.9 78.9 4.6 73.7 15.5 36.9 11.3 30.4 Pseu- 11.8 1.1 77.6 30.6 17.3 16.1 5.0 domonas Streptococ- 39.6 21.5 8.3 9.2 22.6 10.6 16.0 40.3 34.0 cus Staphylococcus 63.0 9.0 Bordetella 30.4 4.3 Achromobacter 26.7 3.8 Veillonella 17.8 2.3 1.8 3.1 2.0 Actinomyces 1.3 3.1 3.6 1.1 1.0 3.2 Neisseria 0.0 9.2 Prevotella 5.0 1.3 0.9 3.9 Stenotrophomonas 1.1 0.2 7.7 Gemella 1.2 0.2 4.6 Escherichia 1.1 0.2 4.6 Burkholderia 5.0 0.7 Haemophilus 0.0 3.7 Oribact-eri- 3.6 0.5 um 135