Assessment of Subsurface in-situ Microbial Communities by Biomarkers for Remediation Potential, Monitoring
Effectiveness, and as Rational End-Points
David C. White, Cory Lytle, Sarah J. Macnaughton, John R. Stephen, Aaron Peacock, Carol A. Smith, Ying Dong Gan, Yun-Juan Chang, Yevette M. Piceno
Center for Environmental Biotechnology, University of Tennessee, Knoxville, TN, Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN,
Microbial Insights, Inc., Rockford, TN,
-CEBMicrobial Insights, Inc.
In-situ Microbial Community Assessment
Classical Plate Count < 1.0 to 0.1% of community, takes days, lose community interactions & Physiology
Two Biomarker Methods: DNA: Recover from surface, Amplify with PCR
using rDNA primers , Separate with denaturing gradient gel electrophoresis (DGGE), sequence for identification and phylogenetic relationship. Great specificity
Lipids: Extract, concentrate, structural analysisQuantitative, Insight into: viable biomass, community composition,Nutritional-physiological status, evidence for metabolic activity
Vadose/Capillaryfringe12.2-12.8’
Bridgingcapillary13.5-14.3’
Saturated15-16’
Saturated17-18’
12.4-13.1’Vadose/Capillaryfringe
13.7-14.6’Bridgingcapillary
16-17’Saturated
Mig
ra
tio
n S
tan
da
rd
sM
igra
tio
n S
tan
da
rd
sIncreasing sample depth
Background samples,single bore-hole,triplicate sub-samples
Whey-barrier samples,single bore-hole, triplicatesub-samples.(45 days after amendment)
DGGE3 White et al
Burkholderia and relatives
Bacteroides-Flexibacter-Cytophaga phylum
Arthrobacter and relatives
Background(No treatment)
Whey-Barrier
Key: Prominent bandsderived from:
Neighbor-joining analysis of sequences recoveredfrom DGGE gels and reference sequences
Vadose
Saturated
No box = capillary
LIPID Biomarker Analysis
1. Intact Membranes essential for Earth-based life
2. Membranes contain Phospholipids
3. Phospholipids have a rapid turnover from endogenous phospholipases .
4. Sufficiently complex to provide biomarkers for viable biomass, community composition, nutritional/physiological status
5. Analysis with extraction provides concentration & purification
6. Structure identifiable by Electrospray Ionization Mass Spectrometry at attomoles/uL (near single bacterial cell)7. Surface localization, high concentration ideal for organic
SIMS mapping localization
Cathedral from a Brick Predict impact of Cr contamination (from 50-200,000 ppm) on soil microbial community by artificial neural network (ANN) analysis
PLFA (phospholipid fatty acid) excellent ~x 102-103 ppm Cr with (PLFA).
DNA is “non compressible” ~ perfect code not so influencedBy microniche conditions as cell membranes PLFA is compressible as contains physiological status input Contains “holistic’ information & responds to perturbations Predict it is a Cathedral or a Prison : DNA a perfect brick PLFA a non-linear mixture of bricks and a window
Signature Lipid Biomarker Analysis
Phospholipid Fatty Acid [PLFA] Biomarker Analysis = Single most quantitative, comprehensive insight into in-situ microbial community
Why not Universally utilized?
1. Requires 8 hr extraction with ultrapure solvents [emulsions]. 2. Ultra clean glassware [incinerated 450oC]. 3. Fractionation of Polar Lipids4. Derivatization [transesterification] 5. GC/MS analysis ~ picomole detection ~ 104 cells LOD 6. Arcane Interpretation [Scattered Literature] 7. 3-4 Days and ~ $250
Signature Lipid Biomarker Analysis
NEW Expanded Lipid Analysis
1. Utilizes HPLC not GC [Greatly expanded Molecular Sizes]2. Semi-automated, ~ “Flash” Extraction ~ 1 hr with
fractionation & > recovery from spores 3. Direct analysis of intact lipids [no derivatization]4. Sensitivity ~ Electrospray Ionization [sub femtomolar near
single cell as 100% of analyte ionize not 1%] 5. Specificity ~ Tandem Mass Spectrometry
Neutral loss or gain Select parent ions Analysis of specific product ions
Structural analysis of components in MS/MS [<< Chemical Noise]
Signature Lipid Biomarker Analysis
Lyophilized Soil Fractions, Pipe Biofilm
SFECO2 1. Neutral Lipids
UQ isoprenologues
Derivatize –N-methyl pyridyl Diglycerides Sterols Ergostrerol Cholesterol
ESE Chloroform.methanol
2. Polar Lipids
Transesterify
PLFA
CG/MS
Intact Lipids
HPLC/ESI/MS/MS
Phospholipids PG, PE, PC, Cl, & sn1 sn2 FAAmino Acid PGOrnithine lipidArchea ether lipidsPlamalogens
PHAThansesterify & Derivatize N-methyl pyridyl
3. In-situ Derivatize in SFECO2
2,6 DPA (Spores)
LPS-Amide OH FA
Sequential Extraction & HPLC/ESI/MS analysis ~ 1-2 hrs
Concentration/Recovery
Extraction SFE/ESE
SeparationHPLC/in-line
Fractionation
DetectionHPLC/ESI/MS(CAD)MS
or HPLC/ESI/IT(MS)n
CEBMicrobial Insights, Inc.
Lipid Biomarker Analysis
Sequential High Pressure/Temperature Extraction (~ 1 Hour)
Supercritical CO2 + Methanol enhancer Neutral Lipids, (Sterols, Diglycerides, Ubiquinones)
Lyses Cells Facilitates DNA Recovery (for off-line analysis
2. Polar solvent Extraction Phospholipids CID detect negative ions
Plasmalogens
Archeal Ethers 3). In-situ Derivatize & Extract Supercritical CO2 + Methanol
enhancer 2,6 Dipicolinic acid Bacterial Spores
Amide-Linked Hydroxy Fatty acids [Gram-negative LPS]
Three Fractions for HPLC/ESI/MS/MS Analysis
*Macnaughton, S. J., T. L. Jenkins, M. H. Wimpee, M. R. Cormier, and D. C. White. 1997. Rapid extraction of lipid biomarkers from pure culture and environmental samples using pressurized accelerated hot solvent extraction. J. Microbial Methods 31: 19-27(1997)
Feasibility of “Flash” Extraction
ASE vs B&D solvent extraction*
Bacteria = B&D, no distortionFungal Spores = 2 x B&D Bacterial Spores = 3 x B&D Eukaryotic = 3 x polyenoic FA
[2 cycles 80oC, 1200 psi, 20 min] vs B&D = 8 -14 Hours
CEBMicrobial Insights, Inc.
Q6
Q7Q10
O
O
H3OC
H3OC
CH3
]H
n
197 m/z
Respiratory Ubiquinone (UQ)
Gram-negative Bacteria with Oxygen as terminal acceptor LOQ = 225 femtomole/uL, LOD = 75 femtomole/uL ~ 100 E. coli
Isocratic 95.5/4.5 % methanol/aqueous 1 mM ammonium acetate
CH2O C
O
CH2(CH2)13CH3
CH2OH
CHO C
O
CH2(CH2)13CH3N
CH3
F+
CH3
SO3
N
CH3
O
N
CH3
CH2O C
O
CH2(CH2)13CH3
CHO C
O
CH2(CH2)13CH3
OCH
CH2O C
O
CH2(CH2)13CH3
CH2
CHO C
O
CH2(CH2)13CH3
Pyridinium Derivative of 1, 2 Dipalmitin
C41H73NO5+
Exact Mass: 659.55
Mol. Wt.: 660.02
C6H7NOExact Mass: 109.05
Mol. Wt.: 109.13
neutral loss
C35H67O4+
Exact Mass: 551.50
Mol. Wt.: 551.90
[M+92]+
[M+92-109]+
M = mass of original Diglyceride
LOD ~100 attomoles/ uL
VIABLE NON-VIABLE
O O || ||
H2COC H2COC
| |C O CH C O CH
| |
H2 C O P O CH2CN+ H3
||
|
O
O-
||O
H2 C O H
||O
Polar lipid, ~ PLFA
Neutral lipid, ~DGFA
phospholipase
cell death
Membrane Liability (turnover)
(A) Chromatogram of purified brain and egg yolk derived authentic PG, PE, and PC; (B) Extracted ion chromatogram (EIC) of PG from soil containing 15:0, 16:0, 16:1, 17:0, 17:1, 18:1, 19:1 (see Fig 5); (C) EIC for ions diagnostic of PE from the soil used in B.
A
B
C
PGPE PE
PC
PG
PESeparation on HAISIL reverse phase HL C-18 column, 30 mm x 1mm x 3 μ,95/5 methanol + 0.002% piperidine/water50 μL/min,
post-column modifier 0.02% piperidine in methanol, 10 μL/min.
HPLC/ESI/MS
CEB
• Enhanced Sensitivity• Less Sample
Preparation• Increased Structural
Information• Fragmentation highly
specific i.e. no proton donor/acceptor fragmentation processes occurring
CH2
HC O C
O
R1
CH2OC
O
R2
OP
O
O
OX
ESI (cone voltage) Q-1 CAD Q-3
ESI/MS/MS
Parent product ion MS/MS of synthetic PG Q-1 1ppm PG scan m/z 110-990 (M –H) -
Sn1 16:0, Sn2 18:2
Q-3 product ion scan of m/z 747 scanned m/z 110-990 Note 50X > sensitivity
SIM additional 5x > sensitivity ~ 250X
OHO CH2OH
HO OH
O
CH O
CH2 O
O
O CH2
CH2
CH2 O P
O
O
O-
CH2
CH2
CH2
H2C OH
Archaebacterial Tetraether Lipid
5 ppm
1600 1620 1640 1660 1680 1700 1720 1740m/z0
100
%
1704
1701
1698
16411638
16431695
1664 1680
1706
1707
1713
FW 1640.4
ES+
[M+H]+
[M-2H+Na+K]+
In sim LOQ ~ 50 ppb
Expanded Lipid Analysis Greatly Increase Specificity ~Electrospray Ionization ( Cone voltage between skimmer and inlet ) In-Source Collision-induced dissociation (CID)
Tandem Mass Spectrometry Scan Q-1 CID* Q-3 DifferenceDaughter ion Fix Vary VaryParent ion Vary Fix VaryNeutral loss Vary Vary FixNeutral gain Vary Vary Fix
Select-ion monitoring Fix Fix Fix
*Collision-induced dissociation (CID) is a reaction region between quadrupoles
Lipid Biomarker Analysis
Problem: Rapid Detection of Bacterial Spores & LPS Amide-Linked OH Fatty Acids in Complex Matrices
From the lipid-extracted residue - - - - derivatize (acid methanolysis) & Supercritical Carbon Dioxide + methanol Extract
1. Detect 2,6 dipicolinate with HPLC/ESI/MS/MS 1 hour and 100% not 3 days and ~ 20% viable
2. Detect 3-OH Fatty Acids Amide-linked to KDO in LPS of Gram-negative Bacteria with HPLC/ESI/MS/MS
Enterics & Pathogens 3OH 14:0Pseudomonad's 3OH 10:0 & 3OH 12:0 (Should Dog Drink from Toilet Bowl?)
[M+H]+
[M+Na]+NOCH3
O
H3OC
O
C9H9NO4Exact Mass: 195.05
Mobile phase: MeOH + 1mM ammonium acetateCone: 40V
ES+
ESI Spectrum of 2, 6-Dimethyl Dipicolinate
LOD ~ 103 spores ~ 0.5 femtomoles/ul
Signal Optimization for 2,6 Dimethyl Dipicolinate
0.0E+00
2.0E+07
4.0E+07
6.0E+07
8.0E+07
1.0E+08
1.2E+08
0 20 40 60 80 100 120
Cone Voltage, V
Re
sp
on
se
196 m/z
218 m/z
168 m/z
Microniche Properties from Lipids
1. Aerobic microniche/high redox potential.~ high respiratory benzoquinone/PLFA ratio, high proportions of Actinomycetes, and low levels of i15:0/a15:0 (< 0.1) characteristic of Gram-positive Micrococci type bacteria, Sphinganine from Sphingomonas 2. Anaerobic microniches ~high plasmalogen/PLFA ratios (plasmalogens are characteristic Clostridia), the isoprenoid ether lipids of the methanogenic Archae.
3. Microeukaryote predation ~ high proportions of phospholipid polyenoic fatty acids in phosphatidylcholine (PC) and cardiolipin (CL). Decrease Viable biomass (total PLFA) 4. Cell lysis ~ high diglyceride/PLFA ratio.
Signature Lipid Biomarker Analysis
Microniche Properties from Lipids
5. Microniches with carbon & terminal electron acceptors with limiting N or Trace growth factors ~ high ( > 0.2) poly β-hydroxyalkonate (PHA)/PLFA ratios
6. Microniches with suboptimal growth conditions (low water activity, nutrients or trace components) ~ high ( > 1) cyclopropane to monoenoic fatty acid ratios in the PG and PE, as well as greater ratios of cardiolipin (CL) to PG ratios.
7. Inadequate bioavailable phosphate ~ high lipid ornithine levels
8. Low pH ~ high lysyl esters of phosphatidyl glycerol (PG) in Gram-positive Micrococci.
9. Toxic exposure ~ high Trans/Cis monoenoic PLFA
Signature Lipid Biomarker Analysis
ANN Analysis of CR impacted Soil Microbial Communities
1. Cannelton Tannery Superfund Site, 75 Acres on the Saint Marie River near Sault St. Marie, Upper Peninsula, MI
2. Contaminated with Cr+3 and other heavy metals between1900-1958 by the Northwestern Leather Co.
3. Cr+3 background ~10-50 mg/Kg to 200,000 mg/Kg.
4. Contained between ~107-109/g dry wt. viable biomass by PLFA; no correlation with [Cr] (P>0.05)
5. PLFA biomass correlated (P<001) with TOM &TOC but not with viable counts (P=0.5)
-CEB
C181W
9C
CI1
70
C181W
7C
C10M
E180
CA
170
CI1
51W
11C
I151A
C161W
5C
CI1
50
C201W
9C
C161W
11C
C10M
E160
CB
R181
CA
150
CI1
60
%TO
MC
160
CC
Y170B
C170
C150
C203W
6C
AC
210
PH
C12M
E160
C161W
7T
C183W
3%
TO
CC
I171W
8C
BR
150B
CB
R170B
C181W
7T
C182A
CB
R170A
BIO
MA
SS
C151 P
CI1
51B
WE
TLA
ND
CB
R150A
CC
Y190
MG
CI1
40
C180
C161W
7C
C230
CB
R160 K
C11
ME
160
C205W
3C
12M
E180
C200
CC
Y170A
CI1
51C
C182W
6C
140
C220
CI1
61
C162
C240
CB
R150C
C204W
6V
IAB
LE
_C
C181W
5C
0.0%
1.0%
2.0%
3.0%
4.0%
5.0%
6.0%
7.0%
Rel
ativ
e se
nsiti
vity
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
110%
Cum
mulative sensitivity
Sensitivity analysis ranks the inputs by importance in predicting [Cr+3]PLFA have a significant larger predictive value than environment parameters (marked with arrows).
PLFA profiles are a can be used as a general purpose biosensorgeneral purpose biosensor
ANN Analysis of CR impacted Soil Microbial Communities
SENSITIVITY (from ANN) 20% of the variables accounted for 50% of the predictive of Cr+3
concentrationOf these 20 %:
18:1w9c (6.6%) Eukaryote (Fungal) correlated with 18:26 (P<0.02)
10Me 16:0 (2.5%) correlated with i17:0 (4.8), 16:1 11c (2.9), i15:0 (3.1) (P<0.001). Thus all are most likely indicative of SRBs or MRBs.
18:17c (4.6%) = Gram negative bacteria
10Me 18:0 (4.3%) (Actinomycetes)
-CEBNABIR
ANN Analysis of CR impacted Soil Microbial Communities
CONCLUSIONS:1. Non-Linear ANN >> predictor than Linear PCA (principal Components Analysis)
2. No Direct Correlation (P>0.05) Cr+3 with Biomass (PLFA), Positive correlation between biomass (PLFA) and TOC,TOM
3. ANN: Sensitivity to Cr+3 Correlates with Microeukaryotes (Fungi)18:19c, and SRB/Metal reducers (i15:0, i 17:0, 16:1w11, and 10Me 16:0)
4. SRB & Metal reducers peaked 10,000 mg/Kg Cr+3
5. PLFA of stress > trans/cis monoenoic, > aliphatic saturated with > Cr+3
-CEBNABIR
Expand the Lipid Biomarker Analysis
1. Increase speed and recovery of extraction “Flash”
2. Include new lipids responsive to physiological status HPLC (not need derivatization)
Respiratory quinone ~ redox & terminal electron acceptorDiglyceride ~ cell lysisArchea ~ methanogensLipid ornithine ~ bioavailable phosphateLysyl-phosphatidyl glycerol ~ low pHPoly beta-hydroxy alkanoate ~ unbalanced growth
3. Increased Sensitivity and Specificity ESI/MS/MS
Signature Lipid Biomarker Analysis
Extract lipids, HPLC/ESI/MS/MS analysis of phospholipids detect specific PLFA as negative ions PLFA 12C Per 13C 16:1 253 269 same as 12C 17:0
16:0 255 271 Unusual 12C 17:0 (269) + 2 13C cy17:0 267 284 12C 18:0 (283) + 13C
18:1 281 299 12C 20:6 , 12C 19:0 with 2 13C 19:1 295 314 12C 21:5 (315), 12C 21:6 (313)
Detection of specific per 13C-labeled bacteria added to soils
13C bacteria added
No 13C bacteria added
0 400ft
N
C4
B5 B7 B9
C8 C10
D9 D11
C16
D17E16 E18
G14
H15 H17 H19 H21I20 I22
J19J21
K22K20
L21M20
J23
N21O22
P23
Q24
N23
O24
P25
U26T27
K28
TANNERY
G18
Q26
RemovedBeach
Grass Pond
Woodland
Swampy/Cattails
Wooded Wetland
Grassy Wetland
Running Water
A
Cannelton Tannery Superfund Site
100,001-300,000
75,001-100,000
50,001-75,000
25,001-50,000
10,001-25,000
7,001-10,000
5,001-7,000
3,001-5,000
2,001-3,000
1,001-2,000
501-1,000
101-500
51-100
1-50
ND
Cr+3 Concentrations Site map
regression
training
testing
Pred
ictiv
e er
ror
cross-validated error
Stop !
classificationvector
Inputprofile hidden layer
Schematic architecture of a three layer
feedforward network used to associate
microbial community typing profiles
(MCT) with classification vectors.
Symbols correspond to neuronal nodes
Generalization is assured
by cross-validation
ANN are universal predictors
Capable oflearning from examples
1 10 100 1000 10000 100000 1E+006
Observed Cr3+ concentration (mg Kg-1)
1
10
100
1000
10000
100000
1E+006
Predicted Cr3+ concentration (mg Kg-1)
training setvalidation setregressionidentity
slope = 1.09
R2 = 0.98
Good Predictive Accuracy at > 100 mg Cr+3 /Kg
Tandem Mass Spectrometers
CEBJPL
Ion trap MSn (Tandem in Time)Smaller, Least Expensive, >Sensitive (full scan)
Quadrupole/TOF> Mass Range, > Resolution
MS/CAD/MS (Tandem in Space)1. True Parent Ion Scan to Derivative Ion Scan2. True Neutral Loss Scan 3. Generate Neutral Gain Scan4. More Quantitative 5. > Sensitivity for SIM6. > Dynamic Range