Supplemental Information
Fast Swinnex Filtration (FSF): A fast and robust sampling and extraction method suitable for
metabolomics analysis of cultures grown in complex media
Authors:
Douglas McCloskey1, Jose Utrilla 1, Robert K. Naviaux3, Bernhard O. Palsson1,2, and Adam M. Feist1,2
1 Department of Bioengineering, University of California, San Diego, CA 92093, USA.
2 Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark, 2800 Lyngby,
Denmark.
3 Departments of Medicine, Pediatrics, and Pathology, University of California, San Diego School of
Medicine, San Diego, CA 92093, USA.
* Corresponding author. AM Feist, Department of Bioengineering, University of California, San Diego,
9500 Gilman Drive, La Jolla, CA 92093-0412, USA.
Tel.: 1 858 534 9592; Fax: 1 858 822 3120; E-mail: afeist@ucsd.edu
Table of Contents:
Supplemental figure S1
Supplemental figure S2
Supplemental figure S3
Supplemental figure S4
Supplemental figure S5
Supplemental figure S6
Supplemental methods
Supplemental table S1
3
4
5
6
7
8
9
12
Figure S1: Effect of quenching metabolism with liquid nitrogen or the organic extraction solvent (see
methods for details of the quenching protocol) for anaerobic wild-type E. coli grown in 4 g*l-1 glucose
M9 minimal media with YE (Anaerobic YE) and without (Anaerobic M9). The percent increase in energy
charge ratio using the organic extraction solvent compared to liquid nitrogen is shown above the bar for
the organic solvent. The significance of the change is also indicated by the number of ‘*’ next to the
percent change (* indicates a P-value < 0.05, ** indicates a P-value < 0.01, Student’s t-test). The energy
charge ratio was calculated from the average ATP, ADP, and AMP concentrations from triplicate
cultures. The energy charge ratio =
𝐀𝐓𝐏+𝑨𝑫𝑷⁄𝟐
𝑨𝑻𝑷+𝑨𝑫𝑷+𝑨𝑴𝑷
Figure S2: Heat map of concentrations for aerobic, wild-type E. coli cultures sampled and extracted using
80:20 Methanol:Water (MeOH:H2O) fast filtration or direct extraction compared and 40:40:20
Acetonitrile + 0.1% formic acid:Methanol:Water (ACN:MeOH:H2O). Red indicates the highest
concentration; green indicates the lowest concentration. Colors are scaled feature by feature.
Figure S3: PCA of aerobic, wild-type E. coli cultures sampled using FSF with polyvinylidene fluoride
(PVDF) filters and extracted using 40:40:20 Acetonitrile + 0.1% formic acid:Methanol:Water
(ACNwFormic), 100 mM of ammonium formate (AF100mM), 10 mM of ammonium formate (AF10mM),
and 10 mM of triethylammonium acetate (TA10mM).
Figure S4: PLS-DA for aerobic, wild-type E. coli cultures sampled used FSF with polyethersulfone (PES),
mixed cellulose acetate/nitrate (Cellulose), and polyvinylidene flouride (PVDF) filters. All cultures were
extracted using 40:40:20 Acetonitrile + 0.1% formic acid:Methanol:Water (CAN:MeOH:H2O). The score
plot is shown on the left and the loadings plot is shown on the right. The 95% confidence intervals for
the scores are indicated by dotted lines. NADH was increased in the cellulose and PVDF filters by 2.2
and 2.1 fold over the PES filters, respectively; NADPH was increased in the cellulose and PVDF filters by
1.2 and 1.4 fold over the PES filters, respectively. 95% confidence intervals are indicated by the dotted
lines on the scores plots.
Figure S5: PLS-DA between wild-type anaerobic E. coli cultures grow in 4 g*l-1 M9 minimal media
supplemented with 1 g*l-1 of yeast extract (red) and without (blue), and sampled by direct extraction (A
and B), or using the optimized FSF method (C and D). The scores plots are shown on the left (A and C);
the loadings plots are shown on the right (B and D). 95% confidence intervals are indicated by the
dotted lines on the scores plots. The scale for the axis of component 1 between of the scores plot is
different between A and C. The wider axis of plot C indicates a greater discrimination between the YE
and M9 samples as a consequence of less variance of compounds between replicates and greater overall
number of compounds measured in the samples taken by FSF compared to direct extraction.
Figure S6: Heat plot of the mean ion count (n=8) for 98 metabolites in a neat standard mix. The
coefficient of variation (CV%) is given next to the mean ion count. Neat standard mixes were either
dried down in a centrivap and reconstituted in water (CE), extracted using the direct extraction method
(DE), extracted using the FSF method (FSF), or analyzed without any manipulation (ST). The
reconstitution volume for CE, DE, and FSF was the same as the initial volume of the neat standard mix.
Supplemental methods:
Expanded description of the FSF sampling and extraction protocol:
The details below follow the workflow given in figure 1 and expand the details presented in the figure
caption for figure 1 in order to facilitate the ability of author researchers to reproduce the method.
1) For aerobic cultures, an accurate volume of culture broth was sampled using a pipette and
transferred to a syringe attached to a Swinnex® filter with the plunger removed. For anaerobic cultures,
an accurate volume of culture broth was collected using a syringe and 18.5 gauge blunt needle. For
anaerobic sealed flasks, an accurate volume of culture broth would be collected using a syringe and 18.5
gauge sharp needle. For anaerobic cultures, the plunger was then extended the volume of the syringe
and then attached to a Swinnex® filter. Using a syringe volume that was a minimum of 2x greater than
the liquid volume it was to contain allowed for a sufficient gas purge of the filter housing to remove
residual culture or filtrate. In practice, we recommend using the largest syringe possible. This step was
found to be the most time-consuming step in the protocol. It should be emphasized, that the step of
accurately sampling the culture broth is shared by both the FSF method and the direct extraction
method.
2) The cells were separated from the culture broth and retained on the Swinnex® filter pad by rapidly
expelling the culture and extra volume gas through the filter housing and into a collection vessel. This
step was found to take only a few seconds for an experienced operator for the sample volumes and
biomass density employed in this study. For culture volumes greater than 5 mL, we recommend using a
larger diameter filter and Swinnex® filter holder.
3) The syringe was quickly removed, and a second syringe loaded with 1 mL of extraction solvent and
labeled biomass pre-cooled to -40°C was quickly attached to the filter housing. The extraction solvent,
labeled biomass, and extra volume gas was rapidly expelled through the filter into another collection
vessel. The large surface to volume ratio on the filter relative to biomass facilitates rapid quenching of
the cell biomass. The extraction solvent and partial cell lysate as well as the filter in the filter housing
was stored in the -80°C for further extraction. The same procedure was repeated for each biological
replicate. The step of removing the sampling syringe, replacing it with the extraction syringe, and
applying the extraction solvent to the filter pad was found to take only a few seconds for an experienced
operator when all materials were prepared ahead of time.
4) The filtrate from step 2 for each replicate was filtered through a fresh Swinnex® filter, and 5)
extracted as in step 3. The twice-filtered media can either be discarded or retained for exo-metabolome
analysis of the culture media. The Swinnex® filter and extraction solvent were placed in the -80°C for
further extraction.
6) The Swinnex® filter from step 3 or 5 was re-extracted with extraction solvent that does not contain
internal standards. The eluent was collected in a 50 mL conical tube.
7) The filter holder was unscrewed over the 50 mL conical so that any residual extraction solvent would
not be lost. The filter disk was removed and placed in the 50 mL conical using tweezers. The inside of
the filter housing that is attached to the syringe was rinsed with a small volume of the extraction solvent
from the 50 mL conical to remove any cells that were detached from the filter disk. The 50 mL conical
with extraction solvent and filter disk were then vortexed for 30 seconds.
8) The extraction solvent and partial cell lysate from step 3 or 5 taken during the sampling procedure
were added to the 50 mL conical and vortexed for an additional 30 seconds. The extraction solvent and
cell lysate were then aliquoted into two eppindorf tubes, and the 50 mL conical and filter disk were
discarded.
9) The cell debris was pelleted by spinning at 16000 RPM at 4°C for 5 minutes. The supernatant was
saved in the -80°C for analysis and the cell debris was discarded. It should be emphasized that all steps
of the protocol were conducted by a single operator without the need for assistance from a second
operator.
Swinnex® filter holder assembly:
Swinnex® filters were assembled by first placing the silicone gasket (i.e., O-ring ) in the sample
inlet housing. The filter was then placed on top of the silicone gasket in the sample inlet housing with
care to ensure that the filter uniformly covered the edges of the silicone gasket. For PES filters, the
orientation of the filter was as recommended by the manufacturer. The sample outlet housing was then
screwed into the sample inlet filter housing. We have found that the order in which the Swinnex®
holder is assembled prevents displacement of the filter from the silicone gasket during sampling.
Swinnex® filter holder cleaning:
Swinnex® filters were cleaned by first allowing the sample inlet and sample outlet housing units
and silicone gasket to soak overnight in 50% methanol. The housings and gasket were then thoroughly
washed with soap. Finally, the housings and gaskets were rinsed with 50% methanol in 0.2uM filtered
water and left to air dry. The Swinnex® filter holders were then assembled as described above and
stored until use.
Chemostat cultivation conditions:
Growth in an aerobic pH controlled bioreactors was carried out in a 1.3L Bioflo 110 fermentor
with 700 mL of working volume (New Brunswick Scientific, NJ) temperature was controlled at 37 °C by a
heating jacket. pH was maintained at 7.0 by automatic addition of KOH 4N. Agitation speed was set to
800 rpm and increased up to 1200 rpm if necessary to maintain dissolved oxygen above 20% of
saturating value. Glucose supplemented M9 at 4 g/L was added to the reactor at 2 different dilution
rates controlled by a peristaltic pump (0.31 and 0.44 h-1). Chemostat culture was started in a batch
mode by inoculation to a 0.05 OD600; at the end of exponential growth, a pump was used to feed and
remove media at the same rate. Steady-state was achieved in 3-5 residence times, and was verified by
biomass measurements. Metabolomics sampling was carried out during steady-state growth. In order
to achieve a second steady-state at a different dilution rate, a pump was set to a corresponding value
(0.31 followed by 0.44). After 3-5 residence times, a new steady-state was achieved and samples for
metabolomics analysis were taken. Chemostats were operated for a maximum of 4 days to minimize
any adaptive mutations.
Table S1: List of metabolite abbreviations used in this study
met_id
met_name
23dpg
3-Phospho-D-glyceroyl phosphate
35cgmp
3',5'-Cyclic GMP
5oxpro
5-Oxoproline
6pgc
6-Phospho-D-gluconate
acac
acetoacetate
accoa
Acetyl-CoA
acon-C
cis-Aconitate
actp
Acetyl phosphate
adp
ADP
adpglc
ADPglucose
AICAr
5-Aminoimidazole-4-carboxamide riboside
ala-L
L-Alanine
amp
AMP
arg-L
L-Arginine
asn-L
L-Asparagine
asp-L
L-Aspartate
atp
ATP
btn
Biotin
camp
cAMP
cit
Citrate
citr-L
L-Citrulline
cmp
CMP
coa
Coenzyme A
ctp
CTP
dadp
dADP
damp
dAMP
datp
dATP
dcdp
dCDP
dcmp
dCMP
dctp
dCTP
dgmp
dGMP
dhap
Dihydroxyacetone phosphate
dimp
dIMP
ditp
dITP
dtdpglu
dTDPglucose
dtmp
dTMP
dttp
dTTP
dump
dUMP
dutp
dUTP
f1p
fad
fdp
fol
fum
g3p
g6p
gam6p
gdp
glu-L
glx
glyc3p
glyclt
gmp
gsn
gthox
gthrd
gtp
Hexose_Pool_fru_glc-D
his-L
hxan
imp
ins
itp
lac-L
Lcystin
mal-L
met-L
mmal
nad
nadh
nadp
nadph
oaa
orn
oxa
pep
phe-L
phpyr
Pool_2pg_3pg
prpp
D-Fructose 1-phosphate
FAD
D-Fructose 1,6-bisphosphate
Folate
Fumarate
Glyceraldehyde 3-phosphate
D-Glucose 6-phosphate
D-Glucosamine 6-phosphate
GDP
L-Glutamate
Glyoxylate
Glycerol 3-phosphate
Glycolate
GMP
Guanosine
Oxidized glutathione
Reduced glutathione
GTP
Hexose_Pool_fru_glc-D
L-Histidine
Hypoxanthine
IMP
Inosine
ITP
L-Lactate
L-Cystine
L-Malate
L-Methionine
Methylmalonate
Nicotinamide adenine dinucleotide
Nicotinamide adenine dinucleotide - reduced
Nicotinamide adenine dinucleotide phosphate
Nicotinamide adenine dinucleotide phosphate - reduced
Oxaloacetate
L-Ornithine
Oxalate
Phosphoenolpyruvate
L-Phenylalanine
Phenylpyruvate
Pool_2pg_3pg
5-Phospho-alpha-D-ribose 1-diphosphate
pyr
r5p
ribflv
ru5p-D
s7p
ser-L
thr-L
trp-L
tyr-L
udp
ump
ura
urate
uri
utp
xan
Pyruvate
alpha-D-Ribose 5-phosphate
Riboflavin
D-Ribulose 5-phosphate
Sedoheptulose 7-phosphate
L-Serine
L-Threonine
L-Tryptophan
L-Tyrosine
UDP
UMP
Uracil
Urate
Uridine
UTP
Xanthine