Journal of Pharmaceutical and Biomedical Analysis 80 (2013) 107–115

Contents lists available at SciVerse ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis

journal homepage: www.elsevier.com/locate/jpba

Development of a HPLC–UV method for the quantitative determination of four

short-chain fatty acids and lactic acid produced by intestinal bacteria during in
vitro fermentation
S. De Baere a,∗, V. Eeckhaut b, M. Steppe b, C. De Maesschalck b, P. De Backer a, F. Van Immerseel b, S. Croubels a

a
Department of Pharmacology, Toxicology and Biochemistry, Faculty of Veterinary Medicine, Ghent University, Salisburylaan 133, 9820 Merelbeke, Belgium
b
Department of Pathology, Bacteriology and Avian Diseases, Faculty of Veterinary Medicine, Ghent University, Salisburylaan 133, 9820 Merelbeke, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: A rapid and sensitive HPLC–UV method for the quantitative determination of four short-chain fatty acids
Received 25 August 2012 (SCFAs) and lactic acid (LA) produced during in vitro fermentation is presented.
Received in revised form 20 February 2013 Extraction of SCFAs from supernatants of bacterial cultures is aggravated due to their polarity and
Accepted 21 February 2013
volatility. Detection can only be performed at a short, non-selective UV wavelength (210 nm), due to
Available online 28 February 2013
the lack of any significant chromophore. Therefore special attention was paid to the optimization of the
sample preparation procedure and the HPLC–UV conditions. The final extraction procedure consisted of
Keywords:
a liquid–liquid back extraction using diethylether. Prior to HPLC–UV analysis the samples were acidified
Short-chain fatty acids
Lactic acid
(pH < 2) in order to improve retention of the SCFA’s and LA on the Hypersil Gold aQ column.
HPLC–UV Matrix-matched calibration graphs were prepared for all analytes of interest (range 0.5–50 mM) and
In vitro fermentation correlation and goodness-of-fit coefficients were between 0.9951–0.9993 and 3.88–8.27%, respectively.
Limits of detection and quantification ranged from 0.13 to 0.33 mM and 0.5 to 1.0 mM, respectively. The
results for the within-day and between-day precision and accuracy fell within the ranges specified.
The reported validated method has been successfully used for the in vitro screening of supernatants of
bacterial cultures for the presence of butyric acid, aiming to select for butyric acid-producing bacteria.
In addition, the method has been used to determine the production pattern of selected fatty acids by
bacterial species isolated from human feces and chicken caeca.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction several groups, such as members of the genera Bifidobacterium and

In recent years intestinal health has become an important issue
since it is correlated with a reduced risk for the incidence of various
chronic and infectious diseases, both in humans and animals. The
gastrointestinal tract contains numerous bacterial species, of which
Abbreviations: AA, acetic acid; BA, butyric acid; EU, European Union; FA,
formic acid; FDA, Food and Drug Administration; g, goodness-of-fit coefficient; GC,
gas chromatography; HPLC, high-performance liquid chromatography; IS, internal
standard; LA, lactic acid; LLE, liquid–liquid extraction; LOD, limit of detection; LOQ,
limit of quantification; PA, propionic acid; QC, quality control; r, correlation coef-
ficient; Re , extraction recovery; RSD, relative standard deviation; SA, succinic acid;
SCFA, short-chain fatty acid; S/N, signal-to-noise ratio; SPE, solid-phase extraction;
SS, stock solution; UV, ultraviolet detection; WS, working solution.
∗ Corresponding author. Tel.: +32 9 264 73 48; fax: +32 9 264 74 97.
E-mail addresses: Siegrid.debaere@ugent.be (S. De Baere),
venessa.eeckhaut@ugent.be (V. Eeckhaut), marjan.steppe@ugent.be (M. Steppe),
celine.demaesschalk@ugent.be (C. De Maesschalck), patrick.debacker@ugent.be
(P. De Backer), filip.vanimmerseel@ugent.be (F. Van Immerseel),
siska.croubels@ugent.be (S. Croubels).
Lactobacillus, are assigned to have beneficial health effects [1,2].
The activity and composition of the intestinal microbiota is largely
influenced by diet-derived substrates that resist small intestinal
digestion. These substrates are fermented in the gut by bacteria,
resulting in the production of short-chain fatty acids (SCFA) among
which acetic acid, propionic acid and butyric acid are the most
important. In the colonic lumen, the SCFA production is in the
order of acetate > propionate ≥ butyrate (molar ratio ∼ 60:20:20)
and remains fairly constant, although alterations in production and
absorption may occur with dietary changes [1,3–5]. Total SCFA con-
centrations and regional differences in SCFA concentrations are
implicated in colon diseases, such as cancer or inflammatory bowel
disease [1].
In recent years, there is increasing interest in butyrate-
producing bacterial strains in the context of human and veterinary
medicine. Butyrate is known to serve as the direct energy source
for the colonic epithelium, possesses anti-inflammatory properties
and is capable to reinforce the colonic defence barrier [3]. Some
studies have demonstrated preventive effects of butyrate on colon

0731-7085/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jpba.2013.02.032
108 S. De Baere et al. / Journal of Pharmaceutical and Biomedical Analysis 80 (2013) 107–115
cancer and adenoma, but the mechanisms of action of butyrate in
relation to colon cancer are not yet clearly defined [1]. Butyrate
has also been used as a possible treatment for bowel inflam-
mation. However, there are many inconsistencies between the
results of various studies [1]. In addition, butyrate has an inhibitory
activity against common occurring pathogens in animals, such as
Salmonella and Clostridium perfringens and could be attributed to
prevent necrotic enteritis in chickens [6,7].
To isolate SCFAs and LA from complex matrices, several sample
preparation procedures have been applied: simple centrifugation
[8,9] in combination with a filtration step [10–12], liquid extrac-
tion [5] or solid-phase extraction [13]. The simple and fast methods
were generally combined with expensive ion exclusion chromatog-
raphy.
The quantitative determination of SCFAs in intestinal samples
or fecal cultures is generally performed by gas chromatography
(GC) in combination with flame-ionization detection [14–16] or
mass spectrometric detection [4,17,18]. GC methods usually rely on
derivatisation of non-volatile SCFAs to improve chromatographic
behavior and to increase molecular mass [14,17]. Because these
derivatisation procedures are complex and time-consuming, some
methods also analyze underivatised SCFAs [4,16,19].
Other authors use high-performance liquid chromatography
(HPLC) for the analysis of SCFAs in various matrices (e.g. beverages,
food products, biological fluids) [2,5,8–10,13–15,20]. Chromatog-
raphy is generally performed using expensive ion exclusion
columns [9–12], or reversed-phase columns [5,8,20] with large
dimensions (length: 25–30 cm, internal diameter: 3.9–7.8 mm, par-
ticle size (dp): 4–10 ␮m), resulting in run-times of more than
30 min [5,8,10,20]. Because organic acids lack any significant chro-
mophore, high molar absorptions occur only at short ultraviolet
(UV) wavelengths (205–210 nm), which can compromise method
selectivity [20]. Therefore, a lot of HPLC methods rely on the
formation of derivatives absorbing at higher UV wavelengths
(>250 nm) [5,8] making these methods more complicated and time-
consuming. Several authors use the less common refractive index
[2,9,11] and electrochemical detectors [10].
The aim of this study was to develop a method for the analy-
sis of 4 SCFAs (FA, AA, PA and BA) and LA in fecal bacterial culture
samples that combines a straightforward and cost-effective sample
preparation procedure with a fast analysis time (15 min). Because
both the extraction of SCFAs and the HPLC–UV detection are aggra-
vated due to their polarity and volatility on the one hand and the
lack of suitable chromophores on the other hand, special attention
was paid to the optimization of the sample preparation proce-
dure and HPLC conditions. The sample preparation procedure had
to combine the maximal removal of matrix interferences with an
acceptable extraction recovery for all analytes of interest. The main
chromatographic challenge was the separation of lactic acid and
acetic acid, since this is a major problem in most HPLC–UV meth-
ods and to keep the run-time below 15 min. The principle of internal
standardization was applied to compensate for analyte loss during
sample preparation, for variations during HPLC–UV analysis and to
calculate the relative retention time, which was used for analyte
identification. The method was subjected to an in-house validation
and thereafter applied for the in vitro screening of fecal microbiota
for their butyric acid-producing abilities [3,21].
2. Materials and methods
2.1. Standards, chemicals and reagents
Standards of formic acid (FA), acetic acid (AA), propionic acid
(PA), butyric acid (BA), valeric acid, iso-caproic acid, succinic acid
(SA), d/l-lactic acid (d/l-LA), 2-ethyl-butyric acid and pimelic acid
were obtained from Sigma–Aldrich (Bornem, Belgium). They were
stored at room temperature until the expiry date.
Stock solutions (SS, 5 M) of FA, AA, PA, BA and LA were prepared
in HPLC water and stored at 2–8 ◦ C for 3 months. For SA, which was
used as internal standard (IS), a stock solution of 0.2 M was prepared
as follows: 1.18 g of SA was transferred in a volumetric flask of
50.0 mL and 25.0 mL of HPLC water were added. After gently mixing,
a 1 M solution of sodium hydroxide (NaOH) was added dropwise,
until complete dissolution of the SA crystals was obtained. HPLC
water was added up to the mark. After mixing, the solution was
stored at 2–8 ◦ C.
Combined working solutions (WS, containing FA, AA, PA, BA and
LA) of 0.5 M and 0.05 M were prepared by diluting the stock solution
appropriately with HPLC water. The working solutions were stored
at 2–8 ◦ C for 3 months. Thereafter, the SS and WS were discarded
and replaced by freshly prepared solutions.
Standard solutions of 0.1 M in water/methanol (50/50, v/v) were
prepared for valeric acid, iso-caproic acid, 2-ethyl-butyric acid and
pimelic acid.
Sodium dihydrogenphosphate (NaH2 PO4 ), phosphoric
acid (H3 PO4 ) and NaOH were obtained from Sigma–Aldrich.
Diethylether and hydrochloric acid (HCl, 37%) were of analytical
grade, whereas water and acetonitrile were of HPLC grade. These
reagents were all purchased from VWR (Leuven, Belgium).
The M2GSC medium pH 6 was prepared according to Barce-
nilla et al. and contained per 100 mL: 30 mL of rumen fluid, 1 g of
casitone, 0.25 of yeast extract, 0.2 g of glucose, 0.2 g of cellobiose,
0.2 g of soluble starch, 0.045 g of K2 HPO4 , 0.045 g of KH2 PO4 , 0.09 g
of (NH4 )2 SO4 , 0.09 g of NaCl, 0.009 g of MgSO4 ·7H2 O, 0.009 g of
CaCl2 , 0.1 mg of resazurin, 0.4 g of NaHCO3 and 0.1 g of cystein
hydrochloride [3,22].
2.2. Experimental and blank samples
Experimental samples were obtained as a part of studies car-
ried out to isolate butyrate producing bacteria from chicken caecal
contents and human feces.
2.2.1. Isolation of butyrate-producing bacteria from chicken
caeca
Results have already been published by Eeckhaut et al. [3,21].
Briefly, a sample of the caecal content per chicken was transferred
into an anaerobic workstation (84% N2 , 8% CO2 and 8% H2 , Ruskinn
Technology, Bridgend, UK) immediately after euthanasia. One gram
of the caecal content was homogenized in 9 mL of anaerobic M2GSC
medium (pH 6), followed by the preparation of a 10-fold dilution
series. From each dilution, 0.3 mL was spread onto agar plates con-
taining M2GSC medium with 1.5% agar. Plates were incubated at
41–42 ◦ C for 48 h. Single colonies were randomly picked from dilu-
tions 10−7 or 10−8 and grown overnight (24 h) in 10 mL of M2GSC
broth or in M2GSC broth supplemented with 8 mM d/l-lactic acid.
After centrifugation of these cultures, 1 mL of the supernatant was
subjected to the sample preparation procedure as described below.
2.2.2. Human study
The aim of this study was to isolate butyrate producing bacteria
colonizing the human colon. The same protocol was used as in the
chicken study but instead of the caecum, fresh fecal samples were
used.
Detailed results will be the subject of a forthcoming article.
2.2.3. Blank samples
For the preparation of calibrator and quality control (QC) sam-
ples, blank M2GSC medium (pH 6) that did not contain rumen fluid
and that was not supplemented with SCFAs, was used.
 S. De Baere et al. / Journal of Pharmaceutical and Biomedical Analysis 80 (2013) 107–115 109

2.3. Sample preparation Table 1

HPLC–UV gradient programmes for the analysis of SCFAs in supernatant from bac-
terial cultures using two different analytical columns.
2.3.1. Spiking of calibration curve and QC samples
Each analysis day, matrix-matched calibrator and QC samples Time (min) MF A (%) MF B (%) Flow rate (mL/min)
were freshly prepared as follows: 950 ␮L of blank medium were Column A: Hypersil Gold aQ (150 mm × 4.6 mm, dp: 5 ␮m)
pipetted into a pyrex extraction tube. Appropriate volumes of the 0.0 100 1.25
combined working solutions (0.5 and 0.05 M) were added to obtain 8.0 100 1.25
8.5 80 20 1.25
calibrators with the following theoretical individual organic acid
13.0 80 20 1.25
concentrations: 0.5 mM (10 ␮L of WS 0.05 M), 1 mM (20 ␮L of WS 13.5 100 1.25
0.05 M), 2.5 mM (50 ␮L of WS 0.05 M), 5 mM (10 ␮L of WS 0.5 M), 18.0 100 1.25
10 mM (20 ␮L of WS 0.5 M), 25 mM (50 ␮L of WS 0.5 M). For the Column B: Hypersil Gold aQ (150 mm × 4.6 mm, dp: 3 ␮m)
highest calibrator sample (individual organic acid concentration: 0.0 100 0.80
50 mM), 10 ␮L of each SS were added to 950 ␮L of blank medium. 3.5 100 0.80
QC samples were spiked at an individual organic acid concen- 4.0 80 20 1.50
10.0 80 20 1.50
tration of 2.5 and 25 mM.
10.5 100 1.50
After spiking, the samples were vortex mixed for 15 s, followed 13.5 100 1.50
by the addition of HPLC water up to a final volume of 1000 ␮L and 14.0 100 0.80
an additional vortex mixing step. 15.0 100 0.80

2.3.2. Study and blank samples
Thousand (1000) microliters of the supernatant from each
bacterial culture and of non-inoculated M2GSC medium (=blank
sample) were pipetted into a pyrex extraction tube.
2.3.3. Liquid–liquid extraction
To each sample of an analytical batch (calibrators, QC, study
samples, except blank sample) were added 50 ␮L of the IS stock
solution (SA, 0.2 M). The samples were vortex mixed and equil-
ibrated at room temperature for 5 min. Thereafter 100 ␮L of
concentrated HCl were added, followed by a vortex mixing step
of 15 s. The samples were extracted for 20 min (by gently rolling)
using 5 mL of diethylether. After centrifugation (5 min, 3500 rpm),
the supernatant was transferred to another pyrex extraction tube
and 500 ␮L of a 1 M solution of NaOH were added. The samples were
extracted again for 20 min, followed by a centrifugation step. The
aqueous phase was transferred to an autosampler vial and 100 ␮L of
concentrated HCl were added. After vortex mixing, a 10-␮L aliquot
was injected onto the HPLC–UV apparatus.
Note: Mobile phase (MF) A: 20 mM NaH2 PO4 adjusted to pH 2.2 using phosphoric
acid; mobile phase B: acetonitrile.
EU and FDA documents [24–26]. Validation experiments were per-
formed on the Hypersil Gold aQ column (150 mm × 4.6 mm i.d.)
with particle sizes of 3 ␮m.
2.5.1. Limit of quantification (LOQ)
The LOQ was defined as the lowest concentration of each SCFA
and LA for which the method was validated with a precision and
accuracy that fell within the recommended ranges (see within-run
accuracy and precision). The LOQ was determined by analyzing at
least 5 spiked medium samples on the same day (tested concentra-
tion levels: 0.5 mM and 1.0 mM).
2.5.2. Limit of detection (LOD)
The LOD was defined as the lowest concentration of each organic
acid that could be recognized by the detector with a signal-to-noise
(S/N) ratio of ≥3.

2.4. HPLC–UV analysis

The HPLC–UV system consisted of a P4000 gradient pump with
vacuum degassing, an AS3000 autosampler, an UV6000 detec-
tor, and a SN4000 module, all from Thermo Separations Products
(Thermo Scientific, Breda, The Netherlands). The autosampler was
set at a temperature of 10 ◦ C.
Chromatographic separation was tested on two Hypersil Gold
aQ columns (150 mm × 4.6 mm i.d.) with particle sizes of 5 ␮m and
3 ␮m (Sercolab, Merksem, Belgium). The HPLC columns were pro-
tected by a guard column of the same type. The columns were
thermostatized at 30 ◦ C. The mobile phase consisted of 20 mM of
NaH2 PO4 in HPLC water (pH adjusted to 2.2 using phosphoric acid)
(A) and acetonitrile (B). A gradient elution was performed as shown
in Table 1. The UV detector was set at a wavelength of 210 nm.
Data processing was performed using the Chromquest version 4.1
software (Thermo Scientific).
2.5. In-house method validation
The proposed quantitative method was validated in-house for
each organic acid, by a set of parameters (i.e. limit of quantification
(LOQ), limit of detection (LOD), linearity, within-run and between-
run precision and accuracy, selectivity and carry-over) that were in
compliance with the recommendations as defined by the European
Community [23] and with reference guidelines defined in other
2.5.3. Linearity
The linearity of the method was evaluated by analyzing cal-
ibration curve samples, which were prepared by spiking blank
medium that did not contain any SCFA or LA at the following
concentration levels: 0.5 mM, 1.0 mM, 2.5 mM, 5.0 mM, 10.0 mM,
25.0 mM and 50.0 mM. The calibration curve samples were
treated in a similar way to the study samples. All calibrator
samples were injected once on the HPLC–UV instrument. The
correlation coefficient (r) was determined and had to be ≥0.99.
In addition, the goodness-of-fit coefficient was determined (g =
 
(%deviation)2
(calculated concentration−nominal concentration)
n−1
with % deviation = ×100
nominal concentration
and n = number of calibrator samples) and had to be ≤10% [26].
2.5.4. Accuracy and precision
Within-day precision and accuracy was evaluated by analyzing
at least 5 blank medium samples, which were spiked with the SCFAs
of interest and LA at a low (2.5 mM) and high (25 mM) concentration
level on the same day. Between-day precision and accuracy was
evaluated using the QC samples which were analyzed with each
batch of samples, run on different days. The acceptance criteria for
accuracy were −20% to +10% of the theoretical concentration. For
the precision, the relative standard deviation (RSD) had to be below
the RSDmax value with RSDmax = 2(1−0.5 log Conc) × 2/3 for within-day
precision and RSDmax = 2(1−0.5 log Conc) for between-day precision.
110 S. De Baere et al. / Journal of Pharmaceutical and Biomedical Analysis 80 (2013) 107–115

Fig. 1. HPLC–UV chromatograms of the analysis of a spiked medium sample (SCFA and LA concentration 25 mM) using Hypersil Gold aQ columns (15 cm × 4.6 mm i.d.) with
particle sizes of (A) 5 ␮m and (B) 3 ␮m.

2.5.5. Selectivity In order to identify an unknown peak in the chromatogram

The selectivity of the method was evaluated with respect to
interferences from matrix-specific compounds. The S/N ratio of
a possible interfering peak in a blank sample had to be below to
the S/N ratio of the analyte(s) in the same elution zone at the LOD
level.
In addition, 10 ␮L of a standard-mix solution (10 mM) contain-
ing all analytes of interest and valeric acid, iso-caproic acid, pimelic
acid and 2-ethyl-butyric acid were injected on the HPLC–UV
instrument, using the chromatographic conditions mentioned
above.
of a real sample, the RTT was calculated (i.e. retention
timepeak /retention timeinternal standard ). The RRT had to be within a
±2.5% interval of the mean RRT of the analyte of interest (FA, LA,
AA, PA or BA) in the spiked samples (i.e. matrix-matched calibrator
and quality control samples) of the same analytical batch [23].
2.5.6. Carry-over
The carry-over was evaluated by analysing a water sample just
after the highest calibrator sample. The eventual analyte concen-
tration in the water sample had to be below the LOD.
 S. De Baere et al. / Journal of Pharmaceutical and Biomedical Analysis 80 (2013) 107–115 111

Fig. 2. HPLC–UV chromatograms of the analysis of a blank M2GSC medium sample (A), a standard-mix solution (analyte concentration: 10 mM) (B), a control sample
containing M2GSC medium supplemented with rumen fluid (C) and a sample containing a chicken microbial isolate (25-3) with a high butyrate-producing capability (D).
Original concentrations of SCFAs were: control sample: FA: <LOQ, LA: <LOQ, AA: 8.28 mM, PA: 2.73 mM and BA: 2.04 mM; sample 25-3: FA: 1.95 mM, LA: 12.99 mM, AA:
5.82 mM, PA: 1.24 mM and BA: 13.97 mM.

3. Results and discussion
3.1. Sample clean-up
Extraction of SCFAs and LA from cultures containing rumen fluid
is aggravated due to their polarity and volatility. In the literature
several sample preparation techniques were reported. Horspool
et al. [13] performed a solid-phase extraction (SPE) using Sep-Pak
C18 columns. Other authors performed a centrifugation step, fol-
lowed by a filtration step using 0.2 ␮m filters to remove bacterial
cells [9–11] or using a microconcentrator with a molecular cutoff of
3000 Da [12]. These sample preparation techniques are simple and
fast, but are generally combined with ion exclusion chromatogra-
phy. In addition, the filtration devices increased the sample analysis
cost.
In our study, a simple sample preparation consisting of a
combination of a 1/5 dilution of the supernatant, followed by cen-
trifugation and filtration using a 0.22 ␮m filtration device, was
tried out during initial experiments. A lot of interfering peaks were
observed in the chromatogram, resulting in a poor resolution and
sensitivity of the analytes of interest. Similar results were obtained
if the sample clean-up consisted only of a protein precipitation
step using a strong acid (trifluoroacetic acid, trichloroacetic acid)
(results not shown). As an alternative, liquid–liquid extractions
in acidic medium using ethylacetate (EtAc), hexane:ethylacetate
(Hex:EtAc, 90/10, v/v), hexane:iso-amylalcohol (Hex:IAA, 90/10,
v/v) and diethylether (ether) as extraction solvents were tried
out. Following extraction, the organic phase was evaporated and
the residue was redissolved in 0.01 N hydrochloric acid prior to
injection onto the HPLC–UV apparatus. Cleaner extracts could be
obtained, but the extraction recoveries for all aliphatic SCFAs were
low (<10% for FA, AA, PA and BA), whereas for d/l-lactic acid the
extraction recovery was about 25%. Further investigations revealed
that the aliphatic SCFAs were lost during evaporation of the organic
phase due to their volatile nature. To circumvent this problem, a
liquid–liquid back extraction was performed using the different
solvents. Following extraction with EtAc or Hex:EtAc (90/10, v/v),
a large interfering peak was observed at the elution zone of LA
and AA. Extraction with ether and Hex:IAA (90/10, v/v) resulted
in chromatograms without interfering peaks. Because extraction
recoveries were higher for all compounds after extraction with
ether (34–94%) compared to Hex:IAA (90/10, v/v) (6.5–66%), the
first solvent was used in the final protocol.
To increase the retention of the acidic analytes on a reversed-
phase HPLC column, the final alkaline aqueous extract was acidified
to a pH < 2, using 100 ␮L of concentrated HCl.
Compared to other methods, our sample preparation procedure
is cost-effective, because no expensive filters or SPE columns are
needed. In addition, the LLE is straightforward, without the need
for evaporation and derivatisation, which makes that the whole
sample preparation procedure can be accomplished in 1 h. Hence,
∼100 samples can be easily processed per analysis day.
3.2. Chromatography
To obtain an optimal separation of the SCFAs, 25–30 cm columns
with internal diameters (i.d.) of 3.9–7.8 mm, packed with ion-
exclusion, ion-exchange or reversed-phase C18 stationary phases
with particle sizes of 4–10 ␮m are generally used [5,8,10,20]. The
first two types of columns are expensive and use dilute sulphuric
acid as the mobile phase, operating at high temperatures (e.g.
60 ◦ C). The aqueous part of the mobile phases used with reversed-
phase C18 columns are water with 5% methanol [8] or phosphate
112 S. De Baere et al. / Journal of Pharmaceutical and Biomedical Analysis 80 (2013) 107–115

Table 2
Evaluation of the linearity for the analysis of FA, LA, AA, PA and BA in M2GSC culture medium.

FA (mean ± SD) LA (mean ± SD) AA (mean ± SD) PA (mean ± SD) BA (mean ± SD)

r (n = 3) 0.9972 ± 0.0012 0.9989 ± 0.0003 0.9983 ± 0.0014 0.9973 ± 0.0014 0.9951 ± 0.0019
g (%) (n = 3) 6.37 ± 1.31 3.88 ± 0.34 4.55 ± 2.08 6.16 ± 1.24 8.27 ± 1.23

Note: r: correlation coefficient (≥0.99), g: goodness-of-fit coefficient (<10%), n = number of calibration curves, SD = standard deviation.

buffers adjusted to a pH of ∼2 [20,27]. Isocratic [20] or gradient Table 3

Results of the within-run accuracy and precision, LOQ and LOD evaluation for the
elution [8] is performed. Run times varied from ∼30 to 60 min
analysis of FA, LA, AA, PA and BA in M2GSC culture medium.
[5,8,10,20].
In this study, special care has been devoted to the choice of the Theoretical Accuracya (%) Precisiona (%) RSDmax (%)
concentration (mM)
analytical column and of the mobile phase, because it was the aim
to obtain a good separation between LA and AA and to elute all FA
analytes of interest within a reasonable run-time (i.e. ≤20 min). 0.50 −0.3 11.6 (n = 6) 33.5
1.00 −0.2 7.7 (n = 6) 30.2
Since Tormo et al. [27] obtained a good separation between 11
2.50 6.4 5.7 (n = 6) 26.3
organic acids using a mobile phase consisting of 20 mM NaH2 PO4 25.0 1.1 2.5 (n = 6) 18.6
adjusted to pH 2.2 with phosphoric acid (A) and acetonitrile (B),
LOD (mM) 0.13
these mobile phase solvents were chosen for our experiments. Two
Hypersil Gold aQ columns with a length of 15 cm and respective LA
particle sizes of 5 ␮m and 3 ␮m were tested. This type of column 0.50 −2.7 7.3 (n = 6) 33.5
contains a polar endcapped C18 phase which offers superior reten- 1.00 −0.2 2.9 (n = 6) 30.2
tion of polar compounds. Because the resolution between LA and 2.50 2.4 2.2 (n = 6) 26.3
25.0 2.8 1.3 (n = 6) 18.6
AA was not optimal using the 5-␮m column, it was decided to use
a 3-␮m column in the final protocol, resulting in a nearly base- LOD (mM) 0.19
line separation of both compounds. A flow-rate of 0.8 mL/min was
AA
maintained during the first part of the HPLC run (0–4 min) to obtain
0.50 7.7 15.7 (n = 6) 33.5
a good separation of the early eluting SCFAs (FA, AA) and LA. In 1.00 −4.5 9.1 (n = 6) 30.2
order to enhance the elution of BA from the Hypersil Gold aQ col- 2.50 −0.7 9.0 (n = 6) 26.3
umn, both a mobile phase gradient (from 100% A to 80% A) and a 25.0 −0.1 3.4 (n = 6) 18.6
flow-rate gradient (from 0.8 mL/min to 1.5 mL/min) were applied, LOD (mM) 0.23
resulting in a total run-time of 15 min (see Fig. 1B). This is much
shorter than the run-times reported by other authors (Torii et al.: PA
30 min; Sánchez-Machado et al.: 45 min; Czauderna et al.: 59 min; 0.50 −5.3 8.3 (n = 6) 33.5
1.00 −7.3 3.9 (n = 6) 30.2
Kotani et al.: 35 min) [5,8,10,20] and comparable to the run-time
2.50 0.4 6.8 (n = 6) 26.3
of ∼13 min that was reported by Horspool et al. [13]. 25.0 2.3 3.6 (n = 6) 18.6
The change in mobile phase flow-rate and consequently in the
LOD (mM) 0.22
column back pressure resulted in an elevation of the chromato-
graphic baseline signal near the elution zone of PA, as can be seen BA
from Fig. 2. However, the quantification of this compound was not 1.00 −1.5 12.7 (n = 6) 30.2
impaired due to this baseline variation, as can be seen from the 2.50 −3.7 8.1 (n = 6) 26.3
validation results obtained for this compound (Tables 2 and 3). 25.0 6.6 4.1 (n = 6) 18.6

LOD (mM) 0.20

3.3. Internal standard

a
Acceptability ranges: accuracy: −20% to +10%; precision: RSD ≤ RSDmax ; LOD:
S/N = 3.

The method of internal standardization was applied in order

to compensate for analyte losses during sample preparation and acid, pimelic acid and 2-ethyl-butyric acid. Citric acid showed no
for variations (retention time, sensitivity) during HPLC–UV anal- retention on the analytical column, whereas oxalic acid was not
ysis. Several organic acids were tested for their use as internal baseline separated from LA and AA and pimelic acid eluted just
standard, i.e. citric acid, oxalic acid, succinic acid (SA), iso-caproic after BA. 2-Ethyl-butyric acid and iso-caproic acid eluted at the end

Table 4
Mean results (±SD) of the analysis of FA, LA, AA, PA and BA in supernatant from selected butyrate producing bacteria isolated from chicken caecal content, cultured overnight
in M2GSC medium. To evaluate the production/consumption of SCFAs and LA, the original concentrations obtained after HPLC–UV analysis were corrected by subtracting
the SCFA/LA concentrations in the control sample from the respective SCFA/LA concentration found in the supernatants of caecal culture samples. Samples were analyzed in
triplicate.

Isolate FA (mM) LA (mM) AA (mM) PA (mM) BA (mM)

20-2 3.79 ± 0.17 17.99 ± 0.85 −3.29 ± 0.36 −1.45 ± 0.33 2.60 ± 0.35
37-2 0.98 ± 0.02 18.99 ± 0.19 −4.01 ± 0.10 −1.47 ± 0.57 2.47 ± 0.61
65-2 1.04 ± 0.03 21.58 ± 0.16 −4.10 ± 0.10 −1.93 ± 0.43 2.80 ± 0.15
3983 1.42 ± 0.06 19.98 ± 0.70 −2.64 ± 0.13 −1.55 ± 0.60 3.12 ± 0.30
3989 <LOQ 12.56 ± 0.39 −5.22 ± 0.39 −1.12 ± 0.71 4.54 ± 0.63
13,633 <LOQ 19.14 ± 0.55 −1.14 ± 0.16 −0.78 ± 0.16 −0.29 ± 0.02
SP 3.78 ± 0.14 19.15 ± 0.29 −2.56 ± 0.20 −1.02 ± 0.30 2.74 ± 0.14
25-3 1.65 ± 0.16 12.91 ± 0.39 −2.77 ± 0.29 −1.45 ± 0.14 12.20 ± 0.37
Ca <LOQ <LOQ 8.28 ± 0.54 2.60 ± 1.07 2.15 ± 0.15
a
C = control sample consisting of blank M2GSC broth supplemented with rumen fluid.
 S. De Baere et al. / Journal of Pharmaceutical and Biomedical Analysis 80 (2013) 107–115 113

Fig. 3. Results of (A) the screening of human fecal microbiota for their BA-producing capacities (cut-off level for BA: 5 mM) and of (B) the quantitation of relevant SCFAs (FA,
AA, PA and BA) and LA in selected samples with BA concentration ≥ 5 mM. To evaluate the production/consumption of SCFAs and LA, the original concentrations obtained
after HPLC–UV analysis were corrected by subtracting the SCFA/LA concentrations in the control sample (FA: 0 mM, LA: 0 mM, AA: 7.63 mM, PA: 1.73 mM and BA: 3.05 mM)
from the respective SCFA/LA concentrations found in the supernatants of fecal culture samples.

of the chromatogram and were therefore not chosen as IS (Fig. 2B).
The best results could be obtained with SA, because this eluted in
the analyte-free zone between the peaks of AA and PA. Because this
acid was not expected to be present in real samples, it was chosen
as IS in the final protocol. Internal standards used by other authors
were iso-caproic acid [8], ethyl-butyric acid [12]. Horspool et al. did
apply the external standard method, because no suitable IS could
be found due to the proximity of the SCFA peaks [13].
3.4. Method validation
For the final procedure, it was decided to use the Hypersil aQ col-
umn (150 mm × 4.6 mm i.d., dp: 3 ␮m), because a good selectivity
could be combined with a reasonable run-time (15 min).
LOQ and LOD. The LOQ and LOD values for FA, LA, AA, PA and
BA are shown in Table 3. As can be seen, the SCFAs of inter-
est and LA can be quantitated in fecal culture samples down to
a level of 0.50 mM (FA, LA, AA and PA) and 1.0 mM (BA), using
the presented method. These levels are low enough to allow
the quantitation of the analytes in the fecal culture samples and
are comparable with other methods reported in the literature
(Horspool et al.: 0.01–0.2 mM/mL caecal liquor) [13]. Some meth-
ods using pre-column derivatisation report LOQ and LOD levels
which are substantially lower (i.e. in the ␮mol/L range) than those
obtained with our method [5,8]. However, these methods are more
complicated and time-consuming due to the additional derivatisa-
tion step. Hence, it can be stated that in the presented method a
good compromise is obtained between simplicity and sensitivity.
Linearity. Results of the evaluation of linearity of FA, LA, AA,
PA and BA in M2GSC culture medium are shown in Table 2. Good
results for linearity (r, g) were obtained for all compounds over the
range tested (0.5–50.0 mM for FA, LA, AA and PA; 1.0–50.0 mM for
BA).
Linearity was also demonstrated by Horspool et al. (0.5–10 mM)
[13] and Torri et al. (4–500 ␮mol/L) [5]. These latter concentra-
tions were substantially lower than in the present method, but the
analytes were converted to their 2-nitrophenylhydrazides prior to
HPLC analysis.
Precision and accuracy. The results of the within-day precision
and accuracy evaluation are summarized in Table 3. The accept-
ability ranges were met for all analytes at the specified levels. The
between-run precision and accuracy was tested by the analysis of
QC samples, spiked at a level of 2.5 and 25 mM and the results fell
within the ranges specified (results not shown).
Selectivity. In Fig. 2A a chromatogram of a blank chicken fecal cul-
ture sample is shown. As can be seen, no endogenous interferences
could be observed at the elution zones of FA, LA, AA and PA. In the
elution zone of BA, small peaks from late-eluting endogenous com-
pounds could be observed. This can explain why a higher LOQ level
for BA (1.0 mM) is obtained, compared to the other SCFAs (0.5 mM).
However, BA can be reliably quantitated as has been shown dur-
ing method validation (Tables 2 and 3). As can be seen in Fig. 2B,
the method was selective because a good separation was obtained
between FA, LA, AA, PA and BA and other organic acids.
Carry-over. In the water sample that was injected after the high-
est calibrator sample, no peaks were observed at the elution zone of
the analytes of interest, indicating that no carry-over was present.
114 S. De Baere et al. / Journal of Pharmaceutical and Biomedical Analysis 80 (2013) 107–115
From the validation results, it can be concluded that the method
can be used for the reliable quantification of the analytes of
interest.
3.5. Analysis of study samples
The applicability of the developed method was shown by the
analysis of the 4 selected SCFAs (FA, AA, PA and BA) in the super-
natant from overnight cultures from bacteria isolated from chicken
caecal content and human feces, respectively. Concentrations of LA
were also determined, because it was the aim to investigate which
of the newly isolated strains was able to produce BA through con-
version of LA as this is thought to be significant in the human colon
and has been shown to be a specific attribute of species within the
clostridial cluster XIVa [28,29].
The results of an in vitro fermentation experiment that was
performed to screen selected human fecal bacteria for their
butyrate-producing capacities (cut-off level for BA: 5 mM) are
shown in Fig. 3A. The concentrations obtained after HPLC–UV
analysis were corrected for the BA concentration (3.05 mM) that
was originally present in the M2GSC medium supplemented with
rumen fluid (=control sample). As can be seen from Fig. 3A, many
isolates consume BA, resulting in a negative BA concentration,
while the butyrate-producing bacteria (cut-off level: 5 mM) can
easily be detected.
At the same time, the other relevant SCFAs (FA, AA and PA)
and LA were quantitated in the selected BA producing samples, as
shown in Fig. 3B.
The results of one experiment with selected chicken caecal
bacteria are shown in Table 4 and Fig. 2D. Each sample was analyzed
in triplicate. The concentration of the SCFAs and LA in each sample
was corrected for the basal concentration of SCFAs and LA in the
M2GSC broth (Fig. 2C). In such a way, the production or consump-
tion of the different SCFAs and LA could be evaluated. In Fig. 2D, a
chromatogram is shown of a bacterial culture containing a strain
isolated from the chicken caecum with high butyrate-producing
capabilities.
The results above show that the method presented in this
manuscript is suitable not only for initial screening purposes with
the aim to select butyrate-producing bacteria, but also for the quan-
titation of the other relevant SCFAs and LA.
4. Conclusions
A straightforward and sensitive HPLC–UV method for the quan-
titative determination of 4 SCFAs (formic acid, acetic acid, propionic
acid and butyric acid) and lactic acid in supernatant from bacterial
cultures was described. The sample preparation procedure, con-
sisting of a liquid–liquid back extraction, in combination with an
acceptable HPLC run-time of 15 min, allowed the extraction and
analysis of many samples a day (up to 100). No derivatisation step
was performed prior to HPLC analysis, which makes the method
simple and cost-effective. The method was in-house validated and
the results for the investigated parameters (linearity, precision and
accuracy, LOQ, LOD, selectivity, extraction recovery) fell within the
specified ranges.
The method was successfully applied for the screening of intesti-
nal bacteria for butyric acid production. In addition, the method
has been used to determine the SCFA and LA production pattern of
selected bacterial species when grown in broth.
The above conclusions indicate the valuability of the presented
method for use in the field of in vitro testing of bacterial SCFA
fermentation.
Acknowledgement
The authors wish to thank Ms. J. Lambrecht for her laboratory
assistance.
References
[1] J.M.W. Wong, R. De Souza, C.W.C. Kendall, A. Emam, D.J.A. Jenkins, Colonic
health: fermentation and short chain fatty acids, J. Clin. Gastroenterol. 40 (2006)
235–243.
[2] P. Moura, F. Simões, F. Gírio, M.C. Loureiro-Dias, M.P. Esteves, PCR monitoring of
Lactobacillus and Bifidobacterium dynamics in fermentations by piglet intestinal
microbiota, J. Basic Microb. 47 (2007) 148–157.
[3] V. Eeckhaut, F. Van Immerseel, S. Croubels, S. De Baere, F. Haesebrouck, R.
Ducatelle, P. Louis, P. Vandamme, Butyrate production in phylogenetically
diverse Firmicutes isolated from the chicken caecum, Microb. Biotechnol. 4
(2011) 503–512.
[4] F. Bianchi, M. Dall’Asta, D. Del Rio, A. Mangia, M. Musci, F. Scazzina, Develop-
ment of a headspace solid-phase microextraction gas chromatography–mass
spectrometric method for the determination of short-chain fatty acids from
intestinal fermentation, Food Chem. 129 (2011) 200–205.
[5] T. Torii, K. Kanemitsu, T. Wada, S. Itoh, K. Kinugawa, A. Hagiwara, Measure-
ment of short-chain fatty acids in human faeces using high-performance liquid
chromatography: specimen stability, Ann. Clin. Biochem. 47 (2010) 447–452.
[6] F. Van Immerseel, F. Boyen, I. Gantois, L. Timbermont, L. Bohez, F. Pasmans, F.
Haesebrouck, R. Ducatelle, Supplementation of coated butyric acid in the feed
reduces colonization and shedding of Salmonella in poultry, Poult. Sci. 84 (2005)
1851–1856.
[7] L. Timbermont, A. Lanckriet, J. Dewulf, N. Nollet, K. Schwarzer, F. Haesebrouck,
R. Ducatelle, F. Van Immerseel, Control of Clostridium perfringens-induced
necrotic enteritis in broilers by target-released butyric acid, fatty acids and
essential oils, Avian Pathol. 39 (2010) 117–121.
[8] M. Czauderna, J. Kowalczyk, Lactic acid can be easily and precisely determined
by reversed-phase high performance liquid chromatography with pre-column
derivatization, J. Anim. Feed Sci. 17 (2008) 268–279.
[9] K. Manderson, M. Pinart, K.M. Tuohy, W.E. Grace, A.T. Hotchkiss, W. Wid-
mer, M.P. Yadhav, G.R. Gibson, R.A. Rastall, In vitro determination of prebiotic
properties of oligosaccharides derived from an orange juice manufacturing
by-product stream, Appl. Environ. Microb. 71 (2005) 8383–8389.
[10] A. Kotani, Y. Miyaguchi, M. Kohama, T. Ohtsuka, T. Shiratori, F. Kusu, Determi-
nation of short-chain fatty acids in rat and human feces by high-performance
liquid chromatography with electrochemical detection, Anal. Sci. 25 (2009)
1007–1011.
[11] E. Gomez, K.M. Tuohy, G.R. Gibson, A. Klinder, A. Costabile, In vitro evaluation of
the fermentation properties and potential prebiotic activity of Agave fructans,
J. Appl. Microbiol. 108 (2010) 2114–2121.
[12] J. Fernandes, A.V. Rao, Th.M.S. Wolever, Different substrates and methane
producing status affect short-chain fatty acid profiles produced by in vitro
fermentation of human faeces, J. Nutr. 130 (2000) 1932–1936.
[13] L.J.I. Horspool, Q.A. McKellar, Determination of short-chain fatty acids in equine
caecal liquor by ion exchange high performance liquid chromatography after
solid phase extraction, Biomed. Chromatogr. 5 (1991) 202–206.
[14] A.J. Richardson, A.G. Calder, C.S. Stewart, A. Smith, Simultaneous determina-
tion of volatile and non-volatile acidic fermentation products of anaerobes by
capillary gas-chromatography, Lett. Appl. Microbiol. 9 (1989) 5–8.
[15] A. Pompei, L. Cordisco, S. Raimondi, A. Amaretti, U.M. Pagnoni, D. Matteuzzi, M.
Rossi, In vitro comparison of the prebiotic effects of two inulin-type fructans,
Anaerobe 14 (2008) 280–286.
[16] B.U. Metzler-Zebeli, R.T. Zijlstra, R. Mosenthin, M.G. Gänzle, Dietary calcium
phosphate content and oat ␤-glucan influence gastrointestinal microbiota,
butyrate-producing bacteria and butyrate fermentation in weaned pigs, Micro-
biol. Ecol. 75 (2011) 402–413.
[17] G.A. Mills, V. Walker, H. Mughal, Headspace solid-phase microextraction with
1-pyrenyldiazomethane in-fibre derivatisation for analysis of faecal short-
chain fatty acids, J. Chromatogr. B 730 (1999) 113–122.
[18] V. De Preter, G. Van Staeyen, D. Esser, P. Rutgeers, K. Verbeke, Development of
a screening method to determine the pattern of fermentation metabolites in
faecal samples using on-line purge-and-trap gas chromatographic–mass spec-
trometric analysis, J. Chromatogr. A 1216 (2009) 1476–1483.
[19] M.L. Stewart, D.A. Timm, J.L. Slavin, Fructooligosaccharides exhibit more rapid
fermentation than long-chain inulin in an in vitro fermentation system, Nutr.
Res. 28 (2008) 329–334.
[20] D.I. Sánchez-Machado, J. López-Cervantes, O. Martínez-Cruz, Quantification of
organic acids in fermented shrimp waste by HPLC, Food Technol. Biotechnol.
46 (2008) 456–460.
[21] V. Eeckhaut, F. Van Immerseel, E. Teirlynck, F. Pasmans, V. Fievez, C. Snauwaert,
F. Haesebrouck, R. Ducatelle, P. Louis, P. Vandamme, Butyricicoccus pullicaeco-
rum gen. nov., sp. Nov., an anaerobic, butyrate-producing bacterium isolated
from the caecal content of a broiler chicken, Int. J. Syst. Evol. Microb. 58 (2008)
2799–2802.
[22] A. Barcenilla, S.E. Pryde, J.C. Martin, S.H. Duncan, C.S. Stewart, C. Henderson,
H.J. Flint, Phylogenetic relationships of butyrate-producing bacteria from the
human gut, Appl. Environ. Microbiol. 66 (2000) 1654–1661.
 S. De Baere et al. / Journal of Pharmaceutical and Biomedical Analysis 80 (2013) 107–115
[23] Commission Decision 2002/657/EC implementing Council Directive 96/23/EC
concerning the performances of analytical methods and the interpretation of
results, Official Journal of the European Communities, No L221, 17/08/2002,
Decision of 12 August 2002, European Commission, Directorate General for
Public Health and Consumers Protection.
[24] R.J. Heitzman (Ed.), Veterinary Drug Residues, Report Eur. 14126-EN, Commis-
sion of the EC, Brussels-Luxembourg, 1994.
[25] VICH GL 49 (MRK) Final, Guidance for industry: validation of analytical methods
used in residue depletion studies, Rockville, MD, USA, February 2011.
[26] J. Knecht, G. Stork, Percentage and logarithmic procedures for calculation of
calibration curves, Fresenius Z. Anal. Chem. 270 (1974) 97–99.
115
[27] M. Tormo, J.M. Izco, Alternative reversed-phase high-performance liquid chro-
matography method to analyse organic acids in dairy products, J. Chromatogr.
A 1033 (2004) 305–310.
[28] C. Bourriaud, R.J. Robins, L. Martin, F. Kozlowski, E. Tenailleau, C. Cherbut, C.
Michel, Lactate is mainly fermented to butyrate by human intestinal microflo-
ras but inter-individual variation is evident, J. Appl. Microbiol. 99 (2005) 201–
212.
[29] S.H. Duncan, P. Louis, H.J. Flint, Lactate-utilizing bacteria, isolated from human
feces that produce butyrate as a major fermentation product, Appl. Environ.
Microbiol. 70 (2004) 5810–5817.