Supporting Information - PNAS · 3/28/2015 · ions of HNP-1, HNP-2, and HNP-3, respectively,...
Transcript of Supporting Information - PNAS · 3/28/2015 · ions of HNP-1, HNP-2, and HNP-3, respectively,...
Supporting InformationBouslimani et al. 10.1073/pnas.1424409112SI Materials and MethodsMolecular Networking. The molecular network was constructedfrom LC-MS/MS data collected from person 1 and person 2samples (Datasets S2 and S3), as well as from the beauty prod-ucts used by these volunteers (Dataset S4). LC-MS/MS datafrom chemical standards, cosmetic ingredients, bacteria andyeast strains, cultured human basal and differentiated keratino-cytes (Datasets S5–S7), and human skin tissues (Dataset S8)were also included in this network. LC-MS/MS data were con-verted to the .mzXML open file format in Compass data analysissoftware (Bruker Daltonics). This text-based format is used toaccess MS, scan number, and all MS/MS spectra for each MSrun. All .mzXML data were used in the analysis (1). To reducethe redundancy of many spectra from identical molecules,MSCluster was used to group MS/MS spectra that originatedfrom identical compounds together (allowing for a precursor m/ztolerance of 0.05 Da). These clusters were then represented withconsensus spectra, constructed by averaging the peak intensitiesof the constituent spectra. This set of consensus spectra was usedto construct the molecular network.For every possible pair of MS/MS spectra, spectrum–spectrum
similarities were determined using a spectral alignment algo-rithm calculating a modified cosine. Cosine similarity scoresrange from 0 to 1, where 1 indicates perfectly matched MS/MSspectra. A cosine threshold of 0.65 was used to construct thismolecular network, and spectral pairs matched with a cosinebelow this value were discarded. Only MS/MS spectral pairs withat least four matched fragment ions were included in this mo-lecular network, and each MS/MS spectrum was only allowed toconnect to its top 10 scoring matches and only if the match was inthe top 10 scoring matches for both spectra, thus resulting ina maximum of 10 connections per node. The algorithm param-eters included a mass tolerance of 0.05 Da for parent mass and0.05 Da for mass fragments. The molecular network resultingfrom retaining only connections (i.e., edges) between MS/MSspectra as described above was imported in Cytoscape 2.8.1 forvisualization (2). The molecular network was displayed in Cy-toscape as nodes and edges (3), where each node represents aspecific m/z with its corresponding MS/MS consensus spectrum.Network organization was performed using the FM3 layoutplugin (ref. 4 and apps.cytoscape.org/apps/fm3). MS/MS spectracorresponding to the background (Blank for swabs samples,growth media, and solvents) that were not connected to a samplenode were removed. MS/MS data from MassBank (5) and fromMETLIN (6) in .mgf format were compared with the MS/MSconsensus spectra from the male and female samples. Putativeidentifications were imported into the network and displayed inCytoscape as attributes.All of the MS/MS data used will be available as a public dataset
on the MassIVE data repository. Additionally, all identifiedcompounds will be uploaded to the Global Natural Product SocialMolecular Networking (GnPS) knowledge base to facilitate rapidfuture compound identifications for follow-up work.
LTQ-FT-ICR Verification of Heme and HNPs. The verification of theHeme and HNPs by MS/MS and MS/MS/MS was done by LTQ-FT–ion cyclotron (ICR). LTQ analysis was performed on a Fin-nigan LTQ-FT mass spectrometer (Thermo Electron) coupled toan Advion NanoMate 100 (Advion), using a solvent mixture of50:50 methanol/water with 1% formic acid. Two swab samplesfrom the female neck area and female groin area were analyzedto confirm the identification of two predicted molecules using
MS/MS and MS/MS/MS fragmentation: the Heme in the groinarea (Fig. 6B) and HNPs (HNP-1, 2, and 3) in the neck area (Fig.6C). All molecules were detected in the MALDI data. Chemicalstandards of HNP-1 and HNP-3 (AnaSpec, Inc.) were also sub-mitted to tandem MS, and MS/MS (MS2) spectra were com-pared with those of HNP-1 and HNP-3 in the neck swab sample.Full-mass spectra were acquired from both neck swab samplesand standards in IT ion-positive mode, in the mass range m/z200–2,000. IT-MS settings were as follows: 3 microscans andtotal acquisition time of 2 min. MS spectra from the groinsample were acquired in FT-MS mode in the mass range m/z200–2,000 with the following settings: 1 microscan and total ac-quisition time of 5 min. For a direct infusion experiment, with aninternal standard, the mass accuracy will be less than 1 ppm.MS/MS fragmentation was performed on manually selected parentions: m/z 616, [M+H]+ ion of the Heme detected in the groinsample, and m/z 861, 843, and 872 corresponding to [M+4H]4+
ions of HNP-1, HNP-2, and HNP-3, respectively, detected in theneck sample. Parent ions m/z 861 and 872 corresponding to[M+4H]4+ of HNP-1 and HNP-3 standards were also selectedfor MS2. MS2 was performed in IT-positive mode for m/z 861, 843,and 872 in swab samples and standards and in FT-positive modefor m/z 616. MS2 settings were as follows: 5 m/z isolation window(IW) for m/z 616 and IW 5, 3, and 4 m/z for m/z 861, 843, and872, respectively. For standards, the following settings were used:IW 8 m/z for m/z 861 (HNP-1) and IW 4 m/z for m/z 872 (HNP-3);3 microscans in IT and 1 microscan in FT; 2 min of total acquisitiontime; and 35% normalized collision energy. Product ion m/z 557(from MS2 of m/z 616) was manually isolated within a window of6 m/z and submitted to MS/MS fragmentation (MS3). MS3 wasperformed in IT mode with the following settings: 3 microscans andtotal acquisition time of 2 min. Acquired mass spectra were ana-lyzed using Xcalibur software (Thermo Scientific).
MS Signal Processing.MALDI-TOF. The collected MALDI-TOF spectra were convertedinto the .mzXML format with the CompassXport routine (BrukerDaltonics) and imported into the MATLAB software (TheMathworks Inc.) by using the mzxmlread routine from theMATLAB Bioinformatics toolbox. First, each spectrum wasnormalized by using the Total Ion Count method by dividingeach spectral intensity by the summary intensity of the spectrum.The spectra without peaks were reduced in intensity usinga heuristic method to compensate for increase in intensity ofsuch spectra after the TIC normalization compared with normalspectra showing a set of peaks. Then, each spectrum was sub-jected to a two-stage baseline correction by applying two timesthe msbackadj routine from the MATLAB Bioinformatics tool-box with the following parameters: WindowSize, 200; Re-gressionMethod, pchip; StepSize, 200; SmoothMethod, none;QuantileValue, 0.05. The mass shift in one of the MALDI platescompared with the other plates was compensated by shifting m/zvalues with data-estimated shifts. The m/z shifts used for thiswere at most 0.08, 0.3, and 0.3 for metabolic, peptidomics, andproteomic spectra, respectively. Finally, all spectra for the samemolecular class (metabolites, peptides, proteins) were inter-polated onto the same grid of m/z values.LC-MS.All LC-MS datasets were converted into the mzML formatby using Compass data analysis software (Bruker Daltonics). ForLC-MS feature extraction, we considered datasets from 36sampling spots (half from male, half from female); the spots weremanually selected to cover all anatomical regions. Then, we
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 1 of 14
applied the OpenMS software to automatically select features foreach of these LC-MS datasets and to store selected features in thefeatureXML format. The FeatureFinderCentroided algorithm wasused; for the parameters (Dataset S10). Each detected featurerepresents a set of m/z RT regions corresponding to isotopes ofa molecule. The m/z and RT widths of these regions are automat-ically detected based on the data. Then, all detected features weremerged into one list. For this analysis, items in the feature list werecompared one by one, and if two features had an intersection areaof more than 10%, only the feature with maximal intensity wasincluded in the final list. Overall, 51,801 LC-MS features wereconsidered in the final list.
Human Skin Cells and Bacterial Strain Cultures. Neonatal humanepidermal keratinocytes (nHEKs) (Life Technologies) and dif-ferentiated nHEKs were cultured in Epilife media (Life Tech-nologies) (Dataset S6), supplemented with epidermal-definedgrowth supplement (Life Technologies) and antibiotic/anti-mycotic (Life Technologies). nHEKs were differentiated by re-moving growth supplements and elevating calcium to 1 mM inthe culture medium for 4 d. We detached 106 basal or differ-entiated cells using trypsin solution (Life Technologies) andwashed them two times with 200 μL PBS (Life Technologies) toremove media traces. Then, 100 μL ethanol (50% solution inwater) was added to cell pellets to mimic metabolite extractionfrom human swab samples. Culture media blanks (controls) werediluted in 50% ethanol in water for analysis.Commensal bacteria (from the following genera: Propionibacte-
rium, Corynebacterium, Staphylococcus, Acinetobacter, Bacillus,Streptococcus, Lactobacillus, Pseudomonas, and Escherichia), allfrequently found on human skin, as well as yeasts (from thefollowing genera: Malassezia and Candida), were cultured inPetri dishes using appropriate media and growing conditions(Dataset S7). Cell biomass collected from each culture was firstwashed in 200 μL water, and then 100 μL of 50% (vol/vol) ethanolin water was added. To capture the background, agar pieces cutfrom blanks containing growth media (controls) were treated fol-lowing the same protocol. We analyzed 50% ethanol extracts fromhuman cells, bacteria, yeast cultures, and blanks using the sameUPLC-MS/MS method described above.For the study of products of triolein hydrolysis, P. acnes was
cultured in Tryptic Soy Agar (TSA) media supplemented or not
with 1% of triolein (Sigma-Aldrich). P. acnes pellets from bothculture conditions as well as agar pieces cut from blanks (TSAand TSA supplemented with 1% of triolein) were first washed in200 μL of water and then extracted in 100 μL of chloroform/methanol (50:50). Extractions were then analyzed using the sameUPLC-MS/MS method.
Preparation of Human Skin Tissues. All experiments that involvehuman skin tissues were performed according to the InstitutionalReview Board approved by the University of California, SanDiego (Project 140144). Normal skin tissues from the back, thescalp, and the neck were obtained from excess healthy tissueremoved during sterile surgical procedures for excision of non-melanoma skin cancers (Dataset S8). One normal skin tissuefrom the back was obtained from the skin bank. The donorsincluded three men 64, 35, and 37 y old, for the scalp, neck, andback skin, respectively. One normal back skin tissue was ob-tained from a man donor 66 y old, via the New York FirefightersSkin Bank (Dataset S8).Sterile biopsies (1 cm) were cut from the dermis to the epi-
dermis and were excised from the center of the excess normal skinsamples immediately after surgery. Epidermis skin tissue (15 cm)was obtained from the skin bank. We used 0.4 g of each skin tissuefor solvent extraction, and 1 mL of solvent was added to 1 g oftissue. Solvent extractions were performed on homogenized tis-sues, using ethanol (50% solution in water), to mimic metaboliteextraction from human swab samples, and using Ethyl acetate/MeOH (1:1). Extractions were performed on ice (4 °C) for 4 h,and then samples were centrifuged and the supernatants wereanalyzed using the same UPLC-QTOF MS/MS method de-scribed above (LTQ-FT-ICR Verification of Heme and HNPs).
Shannon Analysis. The diversity analysis was performed separatelyfor 16S rRNA amplicon and LC-MS data and separately for thefemale and male individuals. For each sample, we calculated thevalue of the Shannon diversity index for intensities of all features(taxa corresponding to genera for 16S rRNA amplicon and MSfeatures for LC-MS). Then, the values of the Shannon diversityindices for all samples were visualized on the 3D models with thecolormap used for visualization of other data (blue correspondingto the minimal value and red corresponding to the maximal valueof the Shannon index for the individual).
1. Frank AM, et al. (2008) Clustering millions of tandemmass spectra. J Proteome Res 7(1):113–122.
2. Smoot ME, Ono K, Ruscheinski J, Wang PL, Ideker T (2011) Cytoscape 2.8: Newfeatures for data integration and network visualization. Bioinformatics 27(3):431–432.
3. Watrous J, et al. (2012) Mass spectral molecular networking of living microbial colo-nies. Proc Natl Acad Sci USA 109(26):E1743–E1752.
4. Hachul S, Junger M (2004) Drawing large graphs with a potential-field-based multilevelalgorithm. Graph Drawing 3383:285–295.
5. Horai H, et al. (2010) MassBank: A public repository for sharing mass spectral data forlife sciences. J Mass Spectrom 45(7):703–714.
6. Smith CA, et al. (2005) METLIN: A metabolite mass spectral database. Ther Drug Monit27(6):747–751.
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 2 of 14
Fig. S1. Representative topographical maps constructed from (A) UPLC-QTOF data and (B) MALDI data, showing specific and distinct body distributions ofmolecular features in person 1 and person 2. m/z, mass-to-charge ratio; RT, retention time.
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 3 of 14
A
C
E
G
B
D
F
184.0729
311.2562
+MS2(494.32436), 19.9-59.8eV, 5.37min #2107
0
20
40
60
80100
Intens.[%]
100 150 200 250 300 350 400 450 m/z
Human swab Phosphocholine[M+H]+
PO
OOO-
N+
OHO
O
*
125.00178
184.07554
311.25953
+MS2(494.32411), 19.9-59.8eV, 5.31min #2100
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 400 450 m/z
*
98.98523125.00046
184.07482
311.25717 391.22521
+MS2(494.32426), 19.9-59.8eV, 5.34min #2102
0
20
40
60
80100
Intens.[%]
100 150 200 250 300 350 400 450 m/z
Malassezia.
furfur *
Basalkeratinocytes
106.04856
341.303651+
421.27066508.30488
+MS2(526.31366), 20.4-61.1eV, 6.56min #2617
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 400 450 500 m/z
*
341.30188
421.26898
+MS2(526.31360), 20.4-61.1eV, 6.49min #2607
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 400 450 500 m/z
*
106.05098155.00952 242.04311
260.05350
341.30708
421.27268
+MS2(526.31624), 20.4-61.1eV, 6.40min #2566
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 400 450 500 m/z
*
Human swab
Bassalkeratinocytes
Malassezia. furfur
508.3015421.2725
341.3050
260.0545242.0416155.0095
106.0495
88.0395
106.05098155.00952 242.04311
260.05350
341.30708
421.27268
+MS2(526.31624), 20.4-61.1eV, 6.40min #2566
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 400 450 500 m/z
*m/z 341
Malassezia. furfur
METLIN
104.10857 184.06689
339.29313+MS2(522.35317), 20.3-60.9eV, 5.95min #2544
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 400 450 500 m/z
*
98.98379125.00035
184.073721+
339.288921+
+MS2(522.35525), 20.3-60.9eV, 5.92min #2346
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 400 450 500 m/z
PO
OOO-
N+O
O
OH
m/z 184
m/z 339
*
125.00197
184.07537
339.29047
+MS2(522.35544), 20.3-60.9eV, 5.86min #2340
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 400 450 500 m/z
*
Phosphocholine[M+H]+Human swab
Basalkeratinocytes
Candida. albicans
252.267121+
264.26702
282.278601+
+MS2(300.28905), 18.0-54.0eV, 5
0
20
40
60
80
100
Intens.[%]
75 100 125 150 175 200 225 250 275 m/z*
HO
OH
NH2
135.11441
252.26814264.26771
282.27930+MS2(300.29026), 18.0-54.0eV, 5
0
20
40
60
80
100
Intens.[%]
100 125 150 175 200 225 250 275 m/z*
136.06409252.27003
264.27008
282.28034+MS2(300.28933), 18.0-54.0eV, 5
0
20
40
60
80
100
Intens.[%]
75 100 125 150 175 200 225 250 275 m/z*
Human swab
Chemical standard
Differentiated keratinocytes
109.09736238.25130
250.25292
268.262311++MS2(286.27256), 17.6-52.8eV, 5.05min #1962
0
20
40
60
80
100
Intens.[%]
80 100 120 140 160 180 200 220 240 260 280 m/z
OH
OH
NH2
*
238.25220250.25337
268.26392+MS2(286.27434), 17.6-52.8eV, 5.08min #2093
0
20
40
60
80100
Intens.[%]
80 100 120 140 160 180 200 220 240 260 280 m/z*
Human swab
Chemical standard
109.09997135.11499
147.11646
161.13247175.14653
189.16553201.16272
215.18180
243.21041
287.27066
+MS2(369.35121), 18.7-56.1eV, 9.93min #4065
0
5
10
15
2025
Intens.[%]
100 150 200 250 300 350 m/z
91.05811
147.11847161.13408
175.14901 203.17949 229.19582 259.24248273.25745 313.28735
+MS2(369.35161), 18.7-56.1eV, 9.55min #4139
0
5
10
15
20
25
Intens.[%]
100 150 200 250 300 350 m/z
135.11807109.10345 201.16327189.16327 243.21224 287.27440
*
*
HO
H
H H
105.07151
135.11799147.11870
189.16519 219.21239243.21147
259.24329
+MS2(369.35224), 18.7-56.1eV, 9.53min #3932
0
5
10
15
20
25
Intens.[%]
100 150 200 250 300 350 m/z
287.27363273.25900109.10363 203.18250
*161.13508
175.14953
215.18032
Human swab
Chemical standard
Differentiated keratinocytes
O
OH
O NH2
O
O
OHPO
OOH
m/z 184Phosphocholine
m/z 311
Fig. S2. ESI-QTOF MS/MS fragmentation of PCs, PSs, sphingosines, and cholesterol. (A) Lyso-PC (16:1) m/z 494, RT 5.4 min in the human swab sample, basalkeratinocytes, M. furfur ATCC 163486 strain. The loss of the phosphocholine group (m/z 184, [M+H]+) induces the release of fatty acid chain (m/z 311).Identification based on MS/MS fragmentation and METLIN database interrogation. It is important to note that we drew a specific location of the double bond
Legend continued on following page
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 4 of 14
and the double bond is drawn as cis, the location of the double bond, or the cis/trans isomer is not known. If a sufficient sample was available, ozonolysis couldbe used to determine the location of the double bond. What we have here is a match to a family member of Lyso-PC 16:1. (B) PS 18:0/0:0 m/z 526, RT 6.6 min inhuman swab sample, differentiated keratinocytes, andM. furfur strain. Identification based on METLIN database (Fig. S2C). (C) A database match of PS 18:0/0:0m/z 526 from M. furfur, based on METLIN database. ESI-QTOF MS/MS of m/z 526 (RT 6.4) shows that loss of the phosphoserine group releases the fatty acidchain (m/z 341). MS/MS of m/z 526 in M. furfur is identical to PS 18:0/0:0 from the METLIN database. At this level of metabolomic matching, the fatty acid sidechain can also be on the other hydroxyl group. (D) 1-Oleoylglycerophosphocholine (Oleoyl-PC) m/z 522, RT 5.9 min detected in the human swab sample, basalkeratinocytes, and C. albicans. The loss of the phosphocholine group (m/z 184, [M+H]+) induces the release of fatty acid chain (m/z 339). It is important to notethat we drew a specific location of the double bond and the double bond is drawn as cis, the location of the double bond, or the cis/trans isomer is not knownalthough the RTs match. If a sufficient sample was available, ozonolysis could be used to determine the location of the double bond. This is a match toa molecular family member. (E) D-erythro-sphingosine (18:1) m/z 300, RT 5.3 min in the human swab sample, chemical standard, and differentiated kerati-nocytes. It is important to note that although the RTs are the same as the standard, the location of the double bond, stereochemistry, or the cis/trans isomer isnot known. If a sufficient sample was available, ozonolysis could be used to determine the location of the double bond. What we have here is a match toa molecular family member. (F) D-erythro-sphingosine (17:1) m/z 286, RT 5.1 min in the human swab sample and in the chemical standard. It is important tonote that although the RTs are the same as the standard, the location of the double bond, stereochemistry, or the cis/trans isomer is not known. If a sufficientsample was available, ozonolysis could be used to determine the location of the double bond. What we have here is a match to a molecular family member.(G) ESI-QTOF MS/MS fragmentation of m/z 369 [M+H-H20]
+, RT 9.9 min in the human swab sample, cholesterol standard, and differentiated keratinocytes. It isimportant to note that although the RTs are similar to the standard, this did not ID the sterol with 100% certainty. MS also cannot determine the stereochemistry. Whatwe have here is a match to a sterol molecular family member. These are the inherent limitations of ID by MS. An asterisk represents a precursor ion.
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 5 of 14
91.05728109.10273121.09901
157.09717 209.13213 259.10314337.25304
1+
412.283411+
430.293901+
448.30294
+MS2(466.31603), 19.7-59.0eV, 4.20min #1588
0
20
40
60
80
100[%]
100 150 200 250 300 350 400 450 m/z
448.3056
430.2936
412.2836
337.2524
319.2411209.1331
*
Human swab
METLIN
270.28064
282.27951300.28936
+MS2(318.30079), 18.2-54.5eV, 5.17min #2196
02040
6080
100
Intens.[%]
75 100 125 150 175 200 225 250 275 300 m/z
252.26853
253.27132256.26149
264.26900
265.25975
270.28064
271.28315
264.265121+
282.278751+ 300.28917
1+
0
204060
80100
Intens.[%]
75 100 125 150 175 200 225 250 275 300 m/z
+MS2(318.30002), 18.2-54.5eV, 5.12min #2001
252.26865 264.265121+
265.254301+ 270.27931
Human swab
Chemical standard
BA
109.10053126.02171
211.14461227.14358253.19517
261.14331
337.257551+
447.33769
462.265171+
480.280721+
498.285141+
+MS2(516.29900), 20.2-60.7eV, 3.99min #1521
0
20
40
60
80
100Intens.[%]
100 150 200 250 300 350 400 450 500 m/z
498.2883
480.2775
462.2665
337.2520
*
Human swab
METLIN
C D
F
97.10271109.10337121.10261135.11841149.13425163.14932 195.17453
205.19526
247.24295
265.25310
+MS2(283.26398), 17.5-52.5eV, 7.62min #3306
0
20
40
60
80
100
Intens.[%]
80 100 120 140 160 180 200 220 240 260 m/z
177.16448
95.08414
109.10103121.10159
139.11188149.13228
163.14848177.16345191.179381+
209.19042
247.240971+
265.251911+
+MS2(283.26344), 17.5-52.5eV, 7.77min #3137
0
20
40
60
80
100
Intens.[%]
80 100 120 140 160 180 200 220 240 260 m/z
135.11735205.19586 *
O
OH
*
95.08440
107.08359
121.09939135.11669
149.13206163.14710
247.24075
265.25216
+MS2(283.26322), 17.5-52.5eV, 7.74min #3172
0
20
40
60
80
100
Intens.[%]
80 100 120 140 160 180 200 220 240 260 m/z
177.16335
*
EHuman swab
Chemical standard
Pseudomonas. aeruginosa
104.10748
184.07519
313.27362 478.32931
+MS2(496.34151), 17.4-69.6eV, 6.08min #2627
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 400 450 m/z
104.10733124.99941
166.05967
184.07270
478.326771+
+MS2(496.33904), 20.0-59.9eV, 5.86min #2506
0
20
40
60
80
100
ntens.[%]
100 150 200 250 300 350 400 450 m/z
*
*
P
O
OOO-
N+
OH
O
O
m/z 184Phosphocholine
m/z 313
Humanswab
Humanskin tissue
117.07055159.11389 219.21286
367.33397385.34490
1++MS2(385.34573), 18.9-56.6eV, 7.97min #3416
0
200
400
600
800
1000
Intens.
100 150 200 250 300 350 m/z
91.05728109.10273121027027.099033 1
157.09717 209.13213 259.10314337.25304
1+
412.28342 1122+288
430.293901+
448..30294
+MS2(466.31603), 19.7-59.0eV, 4.20min #1588
100 150 202 0 252 0 303 0 3350 404 0 454 0 m/mm zz
448448..305305666
430.2936
412.2836
3333337.7.7 222525252444
319.2414444 1209.1331
*3020
Human swab
METLIN
109.10053126.02171
211.144611227464611.14358253.195195 7
261.14331
337.257551+
447.337699
462.26516 71+
480.280721+
4988.2851441+
+MS2(516.29900), 20.2-60.7e1
V, 3.99min #1521
0
20
40
60
80
100Intenene s.[11%]
00
1010100000 1515155000 2020200000 2525255000 3030300000 3535355000 4040400000 4545455000 5050500000 mmm/////mmmmmmmmmmmm zzz
498.2883
480.480480 222777777555
462.2665
337.2520
*++
Human swab
METLIN
C D
104.10748
184.07519
313.27362 478.32931
+MS2(496.34151), 17.4-69.6eV, 6.08min #2627
100 150 200 250 300 350 400 450 m/mm z
100 150 200 250 300 350 400 450 m/mm z
*Humanskin tissue
367.3358
219.2103
159.1153
117.0701*
HO
H
*Humanswab
METLIN
*
*
Intens.
Fig. S3. ESI-QTOF MS/MS ESI-QTOF MS/MS fragmentation of PC 16:0, vitamin D3, Cholic acids, phytosphingosine, and oleic acid. (A) ESI-QTOF MS/MS 2-Pal-mitoyl-sn-glycero-3-phosphocholine (PC 16:0) m/z 496, RT 6 min in the human swab sample and human skin tissue. The loss of the phosphocholine group(m/z 184, [M+H]+) induces the release of the fatty acid chain (m/z 313). Identification is based on MS/MS fragmentation and METLIN database interrogation.
Legend continued on following page
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 6 of 14
(B) Database match to vitamin D3 (7-dehydrocholesterol) based on the METLIN database. ESI-QTOF MS/MS ofm/z 385, RT 8.0 in the swab sample (Top), from theMETLIN database (Down). Here we have a family that has similar fragmentation patterns to vitamin D3. Therefore, we call this the vitamin D3 molecular family.The identification is not 100% certain. This is the limitation of MS-based identification. MS cannot determine stereochemistry. (C) Putative match to glycocholicacid based on METLIN database. MS/MS of m/z 466 [M+H]+, RT 4.2 min in the swab sample (Top), from the METLIN database (Down). This is a match to theglycocholic acid molecular family. (D) Database match to taurocholic acid based on METLIN database. Shown is ESI-QTOF MS/MS of m/z 516 [M+H]+, RT 4.0 minin the swab sample (Top), from the METLIN database (Down). This is a match to the taurocholate molecular family. (E) ESI-QTOF MS/MS fragmentation and RTof phytosphingosine m/z 318, RT 5.2 min detected in the human swab sample (Top) and chemical standard (Down). (F) ESI-QTOF MS/MS fragmentation of cis-oleic acid m/z 283 [M+H]+, RT 7.8 min in the swab sample (Top), Pseudomonas aeruginosa strain (Middle), and chemical standard (Down). We cannot dif-ferentiate between cis/trans nor is the location of the double bond identified. If a sufficient sample was available, ozonolysis could be used to determine thelocation of the double bond. These are the inherent limitations of MS-based ID. An asterisk represents a precursor ion.
A B
CD
144.07883
155.05925
+MS2(161.10605), 8.6-25.7eV, 1.37min #400
0
20
40
60
80
100
Intens.[%]
80 90 100 110 120 130 140 150 160m/z*
144.07938+MS2(161.10640), 13.8-41.5eV, 1.02min #244
0
20
40
60
80
100
Intens.[%]
80 90 100 110 120 130 140 150 160m/z*
97.96912
144.080261+
+MS2(161.10642), 13.8-41.5eV, 1.31min #365
0
20
40
60
80
100
Intens.[%]
80 90 100 110 120 130 140 150 160m/z*
Staphylococcus
Human swab
Chemical standard
95.08589 111.11839
121.10092
135.11720149.13258163.14709
177.16474
219.210571+
237.221871+
256.26149
+MS2(255.23208), 16.7-50.0eV, 7.15min #2873
0
20
40
60
80
100
Intens.[%]
100 125 150 175 200 225 250 275 m/z
95.0861107.0860
121.1000135.1157149.1309
163.1467
181.1593
219.2117
237.2219
256.2622
+MS2(255.2312), 16.7-50.0eV, 7.1min #2969
0
20
40
60
80
100
Intens.[%]
100 125 150 175 200 225 250 275 m/z
97.1006109.1010125.0959
135.1166149.1324
163.1482
181.1583195.1743 213.1901
219.2113
237.2216
256.2667
+MS2(255.2329), 16.7-50.0eV, 7.0min #2842
0
20
40
60
80
100
Intens.[%]
100 125 150 175 200 225 250 275 m/z
*
*
*
Acinetobacter junii
Human swab
Chemical standard
119.049141+ 147.04449
1+
175.039971+
192.042131+
207.066291+
+MS2(225.07657), 15.8-47.3eV, 2.75min #973
0
20
40
60
80
100
Intens.[%]
80 100 120 140 160 180 200 220 m/z*
119.04989
123.04460
147.04451
164.04737
175.04006
179.06981
192.04311
207.06644
+MS2(225.07699), 15.8-47.3eV, 2.73min #964
0
20
40
60
80
100
Intens.[%]
80 100 120 140 160 180 200 220 m/z*
Human swab
Chemical standard
95.08679
103.07503
117.09012 135.11619153.12710163.14676173.15526187.16693 201.18429
+MS2(257.24810), 16.7-50.2eV, 7.64min #3075
0
20
40
60
80
100
Intens.[%]
80 100 120 140 160 180 200 220 240 m/z
95.08369103.07498
109.10081117.09026
125.09594135.11641145.12169159.13611173.15220
201.18675
+MS2(257.24784), 16.7-50.2eV, 7.65min #3132
0
20
40
60
80
100
Intens.[%]
80 100 120 140 160 180 200 220 240 m/z
201.1819187.1708173.1523
153.1523135.1129
117.0891
103.0753
95.0843
O
OH
*
*Human swab
METLIN
Pseudomonas. aeruginosa
Fig. S4. ESI-QTOF MS/MS fragmentation of palmitic acid, palmitoleic acid, tryptamine, and sinapic acid. (A) Palmitic acid match based on METLIN database.Shown is the ESI-QTOF MS/MS of m/z 257 [M+H]+, RT 7.6 min in the swab sample (Top), P. aeruginosa strain (Middle), and from the METLIN database (Down).Because there is no double bond, this match is good for a saturated fatty acid. (B) ESI-QTOF MS/MS of m/z 255 [M+H]+, RT 7.1 min in the human swab sample(person 1, head), Acinetobacter junii ATCC 17908, and palmitoleic acid standard. (C) ESI-QTOF MS/MS of tryptamine m/z 161 [M+H]+, RT 1.3 min in the swabsample, Staphylococcus culture, and chemical standard. This is a match to the tryptamine family. (D) ESI-QTOF MS/MS of sinapic acid m/z 225, RT 2.7 min in thehuman swab sample and in chemical standard. This is a match to the sinapic acid family. An asterisk represents a precursor ion.
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 7 of 14
Pers
on 1
Pers
on 2
107.0722
124.9912
187.0281275.2588
355.1540
+MS2(355.2153), 18.6-55.7eV, 5.8min #2323
0
20
40
60
80100
Intens.[%]
100 150 200 250 300 350 m/z
107.07006
187.02407 275.25668
+MS2(355.21449), 18.6-55.7eV, 6.17min #2449
0
20
40
60
80100
Intens.[%]
100 150 200 250 300 350 m/z
107.0706 187.0274 275.2585 355.2089+MS2(355.21467), 11.7-35.1eV, 5.99min #2365
0
20
40
60
80100
Intens.[%]
100 150 200 250 300 350 m/z
*
A B
163.07441
240.23097
343.29394
+MS2(343.29489), 18.4-55.3eV, 4.62min #1786
0
20
40
60
80
100Intens.[%]
100 150 200 250 300 350 m/z
240.2324 +MS2(343.2948), 18.4-55.3eV, 4.6min #1820
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 m/z
183.17550226.21682
240.23252
343.29562
+MS2(343.29686), 11.5-34.6eV, 4.61min #1775
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 m/z
C
135.04454
161.09686
311.16465
+MS2(311.16451), 18.1-54.3eV, 6.00min #2375
0
20
40
60
80
100
Intens.[%]
100 125 150 175 200 225 250 275 300 325 m/z
135.04457
161.09532311.16543
+MS2(311.16531), 18.1-54.3eV, 5.97min #2360
0
20
40
60
80100
Intens.[%]
100 125 150 175 200 225 250 275 300 325 m/z
105.0678118.0761
135.0443
161.0964
255.1020
311.1659
+MS2(311.1664), 18.1-54.3eV, 5.9min #2408
0
20
40
60
80
100
Intens.[%]
100 125 150 175 200 225 250 275 300 325 m/z
163.07473
232.07528250.08527
365.13729+MS2(362.21062), 18.6-55.9eV, 7.32min #2935
0
20
40
60
80100
Intens.[%]
100 150 200 250 300 350 m/z
105.03284 163.07492 204.08125
232.07564
250.08706
365.13517
+MS2(362.21122), 18.6-55.9eV, 7.33min #2941
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 m/z
177.0676
204.0810
232.0773
250.0870
+MS2(362.2123), 18.6-55.9eV, 7.1min #2928
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 m/z
D
Pers
on 2
922.009851+
758.60815642.47986
788.609781+
846.652281+
945.70800829.62495 1003.74953887.66650714.54280
904.694221+
713.540071061.79136
962.736491+
794.56796 1120.83472
1020.777571+
730.567441+
851.606961+
1025.732311+
1083.77371909.64884
1+ 1078.819071+
+MS, 7.11-8.53min #2853-3458
0
20
40
60
Intens.[%]
600 700 800 900 1000 1100 m/z
m0__RA7_01_2332.d: BPC +All MS
14m__BC2_01_2150.d: BPC +All MS0.00.51.01.5
6x10Intens.
0.00.51.01.5
6x10
6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 Time [min
E
Human swab
Chemical standard
Human swab
Beauty product
Beauty product
Chemical standard
Armpit sample
Beauty product
Beauty product
m/z 788.613, RT 446.7 s
m/z 846.664, RT 456.8 s
m/z 355.215, RT 366 s
Human swab
Beauty product
Chemical standard
Human swab
Beauty product
Chemical standardO
O OS
OH
O
O
m/z 187
m/z 275
O O
O
m/ z 135 m/ z 161
m/z 183
O
OHNN
H
O
m/ z 240
O
O
Nm/ z 204
m/ z 250
m/ z 232
*
*
*
*
*
*
*
*
*
*
Fig. S5. ESI-QTOF MS/MS identification of beauty products and representative maps generated from UPLC-QTOF data. (Upper Left) Representative mapsgenerated from UPLC-QTOF data, showing person 1 and person 2 distribution of m/z 355. The surfactant SLES was mainly localized on person 1’s head. (UpperRight) ESI-QTOF MS/MS fragmentation of beauty products: sodium lauryl ether sulfate, cocoamidopropylbetaine, avobenzone, and octocrylene. (A) ESI-QTOFMS/MS fragmentation of C12-2EO SLES where m/z 355 [M+H]+, RT 6.2 represents C12 alkyl chain SLES and contains 2 ethoxylated units (2EO), without sodiumcounterion (m/z 355), detected in the human swab sample (Top), beauty product (man’s shampoo, Middle), and cosmetic ingredient SLES (Down). The mainfragments correspond to desulfated C12-2EO (m/z 275) and fatty acid chain (m/z 187). It is the most common monomer of the SLES composition, containing 2 or3 ethoxylated units. (B) ESI-QTOF MS/MS fragmentation of m/z 343 [M+H]+, RT 4.6 detected in the human swab sample (Top), beauty product (person 2’sshampoo, Middle), and cosmetic ingredient cocoamidopropylbetaine (C12-CAPB, Down). Dominant fragmentation product (m/z 240) corresponding to thecleavage of the bond between the methylene group and the quaternary nitrogen atom releases the fragment [M+H– (N(CH3)2CH2COO]+. (C) ESI-QTOF MS/MSfragmentation of avobenzone m/z 311 [M+H]+, RT 6.0 min detected in the human swab sample (Top), beauty product (sunscreen lotion, Middle), and cosmeticingredient avobenzone (Down). The alpha-cleavage of the keto form of avobenzone releases the aromatic groups m/z 135 and m/z 161. (D) ESI-QTOF MS/MSfragmentation of octocrylene m/z 362 [M+H]+, RT 7.3 min detected in the human swab sample (Top), beauty product (sunscreen lotion, Middle), and cosmeticingredient octocrylene (Down). Mainly, fragmentation corresponding to the protonated acid occurs at m/z 250 and acylium ion at m/z 232, which theneliminates CO to yield m/z 204. (Lower Left) Representative maps generated from UPLC-QTOF data, highlighting an exclusive person 2 armpit distribution ofm/z 788.613, RT 446.7 s and m/z 846.664, RT 456.8 s. (Lower Right) Polymeric polypropylen glycol (CH2CH2CH2O)n identification. (E, Top) Total ion chro-matogram (RT range, 6–10 min) of beauty product (person 2’s deodorant). (E, Middle) Armpit woman sample, highlighting polymer profile. A difference of 58Da is observed in the average ESI-QTOF mass spectrum (RT range, 7.11–8.53 min), from the beauty product sample, characteristic of the polypropylene glycolsubunit (E, Down). An asterisk represents a precursor ion.
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 8 of 14
Fig. S6. The topographical maps of representative bacteria detected on the skin surface, at each body location, for person 1 and person 2. The color images, asheat maps, show red as the highest percentage of that phylum that was detected and blue as the lowest percentage; all other colors are between. No spotindicates that the specific phylum was not detected on that location. (A) Maps of representative microbial phyla detected on the skin surface, includingActinobacteria, Firmicutes, Proteobacteria, Cyanobacteria, Bacteroidetes (Left), and Staphylococcaceae family (Right). Staphylococcaceae was found in moistareas, such as the foot of both volunteers. (B) Person 1 and person 2 body distribution of genera Lactobacillus and Pseudomonas (Left) and chloroplast classfrom Cyanobacteria phyla (Right). High abundance of Cyanobacteria in person 2’s hands, knees, and lower leg. (C) Person 1’s and person 2’s body distributionof Streptococcus, Haemophilis, and Rothia. They are present on the skin of both individuals, and they were present in larger quantities around the mouth ofthe male.
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 9 of 14
A
135.11798 163.07400
247.24163
265.252121+
339.289891+
+MS2(357.29948), 18.6-55.7eV, 7.23min #2888
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 m/z
105.03336 135.11646
163.07558247.24160
1+
265.252181+
283.26348
339.288511+
+MS2(357.29950), 18.6-55.7eV, 7.38min #2971
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 m/z
- 18 Da (H2O)
- 74 Da- 92 Da (Glycerol)- 110 Da
*
*121.10072
247.24114
265.25046339.28981
+MS2(357.29984), 18.6-55.7eV, 7.34min #3021
0
20
40
60
80
100
Intens.[%]
100 150 200 250 300 350 m/z
*
*
Human swab
Chemical standard
Beauty product
163.07552
365.136261+394.35251
1+438.37949
1+
482.405231+
526.431721+
570.457681+
614.483681+
658.510391+
790.589631+
922.009851+
299.258431+
509.404381+
321.23981 465.377841+
+MS, 6.75min #2698
0
2
4
6
4x10Intens.
100 200 300 400 500 600 700 800 900 m/z
281.24868
619.49092
933.70383
299.25882
735.37949354.31801
+MS, 6.69min #2720
0
20
40
60
80
100
Intens.[%]
100 200 300 400 500 600 700 800 900 m/z
321.24054
281.24744
C Human swab
Propionibacterium acnes culture + Triolein
313.2731
257.2457
239.2361
221.2257123.1160109.1008
95.0851
109.10104123.11600 149.05987
221.22554
239.237421+
257.24760
313.273791+
+MS2(331.28456), 18.3-54.9eV, 7.21min #2898
0
20
40
60
80
100
Intens.[%]
75 100 125 150 175 200 225 250 275 300 325 m/z
- 74 Da*
- 92 Da Glycerol
- 110 Da
- 18 DaH2O
B
Human swab
METLIN
Fig. S7. ESI-QTOF MS/MS fragmentation of (A and B) monoacylglycerol compounds and (C) ESI-QTOF MS spectrum from the human swab sample and theP. acnes + triolein sample. (A) ESI-QTOF MS/MS fragmentation of monoolein m/z 357, RT 7.4 min in a swab sample and beauty product (person 2’s shampoo)and chemical standard. Shown are three main fragments ions: m/z 283, m/z 265 released after the loss of the glycerol group (92 Da), and m/z 247 that cor-responds to the fatty acid chain. MS/MS fragmentation of m/z 357 matches one of the monoolein in the METLIN database. MS/MS cannot differentiate theposition of the double bond or the cis versus trans isomer. If a sufficient sample was available, ozonolysis could be used to determine the location of the doublebond. Further, the position of the fatty acid on the glycerol backbone is not known with 100% certainty. This is an inherent limitation with ID by MS. (B) ESI-QTOF MS/MS fragmentation of 1-Monopalmitin in human swabs and from the METLIN database. One cannot differentiate the location of the fatty acid. (C)ESI-QTOF-MS spectrum from the human swab sample and P. acnes + triolein sample. Both spectra exhibit [M+H]+ ion of m/z 299 as well as [M+H-H20]
+ (m/z281) and [M+Na]+ (m/z 321) (red arrows) showing the presence of the oxidized oleic acid in both the swab sample and P. acnes culture containing triolein. Anasterisk represents a precursor ion.
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 10 of 14
0
10
20
30
40
50
60
70
80
90
100
Rel
ativ
e A
bund
ance
861,27275
840 845 850 855 860 865 870 875 880m/z
0
10
20
30
40
50
60
70
80
90
100
Rel
ativ
e A
bund
ance
872,36366
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Rel
ativ
e A
bund
ance
861,27274
843,36365
872,27274851,27274
867,18183852,27274
878,36365
844,37831
0
10
20
30
40
50
60
70
80
90
100 843,62802
872,38432
861,38805
873,13500
862,13871860,37525
0
10
20
30
40
50
60
70
80
90
100 861,38922
862,13986
862,89210
840 845 850 855 860 865 870 875 880m/z
0
10
20
30
40
50
60
70
80
90
100 872,38446
873,13597
873,88534
0
10
20
30
40
50
60
70
80
90
100
Rel
ativ
e A
bund
ance
861,27275
840 845 850 855 860 865 870 875 880m/z
0
10
20
30
40
50
60
70
80
90
100
Rel
ativ
eA
bund
ance
872,36366
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Rel
ativ
eA
bund
ance
861,27274
843,36365
872,27274851,27274
867,18183852,2728 74
878,36365
844,37831
00
1000
2000
3000
400
5000
6000
7000
8000
9000
10000 843,62802
872,38432
861,38805
873,13500
862,13871860,37525
0
10
20
30
40
50
60
70
80
90
100 861,38922
862,13986
862,89210
840 845 850 855 860 865 870 875 880m/z
0
10
20
30
40
50
60
70
80
90
100 872,38446
873,13597
873,88534
B
A
Fig. S8. Identification of HNP in human swab samples. (A) IT-MS (Left) and FT-MS (Right) spectra collected from the neck swab sample, standard HNP-1 m/z861 [M+4H]4+, and standard HNP-3 m/z 872 [M+4H]4+. (B, Left) ESI-MS/MS fragmentation of m/z 861 [M+4H]4+ in the swab sample and standard HNP-1 and(B, Right) ESI-MS/MS fragmentation of m/z 872 [M+4H]4+ in the swab sample and standard HNP-3.
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 11 of 14
Fig. S9. Correlation between molecular and bacterial body distribution in person 2. (A) Groin molecular and bacterial communities in person 2. Colocalizationanalysis reveals more than 1,000 molecules (UPLC-QTOF data) and six bacteria genera (Anaerococcus, Peptoniphilus, Negativicoccus, Peptostreptococcus,Prevotella, and Sutterella) in the groin area of person 2. (B) Representative map constructed fromMALDI-TOF data, showing person 2’s body distribution ofm/z616 corresponding to the Heme. High abundance of Heme in the groin area of person 2. (C) Heme group identification in person 2 swabs using ESI MS/MS. MS/MS fragmentation of Heme m/z 616 [M+H]+ (Upper) and MS/MS fragmentation of product ion m/z 557 [M+H]+ (Lower). (D) Representative map generatedfrom UPLC-QTOF data that shows the localization of Lyso-PC 16:1 (m/z 494.327, RT 327.6 s) in the groin area of person 2’s body.
Movie S1. 3D_Person1_annotation.
Movie S1
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 12 of 14
Movie S2. 3D_Person2_annotation.
Movie S2
Dataset S1. 16S rRNA sequencing results
Dataset S1
Dataset S2. Person 1 sample list with body locations
Dataset S2
Dataset S3. Person 2 sample list with body locations
Dataset S3
Dataset S4. Person 1 and person 2 beauty product samples
Dataset S4
Dataset S5. List of chemical standards used for molecule identification
Dataset S5
Dataset S6. List of cultured human skin cells
Dataset S6
Dataset S7. List of cultured human bacteria and yeast strains
Dataset S7
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 13 of 14
Dataset S8. List of human skin tissues
Dataset S8
Dataset S9. MS/MS fragmentation parameters
Dataset S9
Dataset S10. LC-MS feature detection. (To open this file in TOPPAS/OpenMS, rename the file extension to .toppas.)
Dataset S10
Bouslimani et al. www.pnas.org/cgi/content/short/1424409112 14 of 14