Journal of Microbiological Methods 55 (2003) 121 – 131

www.elsevier.com/locate/jmicmeth

Identification of species of Brucella using Fourier transform

infrared spectroscopy
M.A. Miguel Gómez a,1, M.A. Bratos Pérez a,*, F.J. Martı́n Gil b,
A. Dueñas Dı́ez a, J.F. Martı́n Rodrı́guez c,2, P. Gutiérrez Rodrı́guez a,
A. Orduña Domingo a, A. Rodrı́guez Torres a
a
Departamento de Microbiologı́a, Hospital Clı́nico Universitario, Facultad de Medicina. Ramón y Cajal 7, 47005 Valladolid, Spain
b
Servicio de Análisis Clı́nicos, Hospital Universitario Rı́o Hortega, Cardenal Torquemada s/n, 47010 Valladolid, Spain
c
Departamento de Estadı́stica, Facultad de Medicina, Ramón y Cajal 7, 47005 Valladolid, Spain
Received 23 October 2002; received in revised form 29 March 2003; accepted 2 April 2003

Abstract

Fourier transform infrared spectroscopy (FTIR) is a technique that has been used over the years in chemical analysis for the
identification of substances and is one that may be applied to the characterisation of microorganisms. The marked tendency of
Brucella towards variation in the smooth rough phase, together with the laboriousness and risk involved in the methods used in
their identification, make their classification difficult. We studied the type strains of the different species and biovars of Brucella
and 11 isolates of human origin of Brucella melitensis, six corresponding to biovar 1, one to biovar 2 and five to biovar 3. The
results of linear discriminant analysis performed using the data provide an above 95% likelihood of correct classification, over
half of which are in fact above 99% for the vast majority of Brucella strains. Only one case of B. melitensis biovar 1 has been
incorrectly classified. The rest of the microorganisms studied (Staphylococcus aureus, Strteptococcus pyogenes, Enterococcus
faecalis, Corynebacterium pseudodiphtheriticum, Clostridium perfringens, Escherichia coli, Acinetobacter calcoaceticus and
Pseudomonas aeruginosa) have been classified correctly in all cases to a likelihood of over 80%. In the graphic representation
of the analysis, a grouping of these can be seen in clusters, which include the different species. One of these comprises B.
melitensis, another Brucella abortus, and another wider one is made up of Brucella suis. The Brucella canis, Brucella ovis and
Brucella neotomae strains appear separate from the previously described groups.
D 2003 Elsevier Science B.V. All rights reserved.

Keywords: Identification; Brucella; FTIR

* Corresponding author. Área de Microbiologı́a, Facultad de Medicina, Ramón y Cajal 7, 47005 Valladolid, Spain. Tel.: +34-983423063;
fax: +34-983423066.
E-mail address: bratos@med.uva.es (M.A. Bratos Pérez).
1
Present address: Servicio de Microbiologı́a, Hospital Universitario de Canarias, Carretera Cuesta Taco s/n, 38320 Santa Cruz de Tenerife,
Spain.
2
Present address: Consejerı́a de Sanidad y Bienestar Social, Servicio de Estadı́stica, Junta de Castilla y León, Avenida de Burgos 5, 47009
Valladolid, Spain.

0167-7012/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0167-7012(03)00120-9
122 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131

1. Introduction 2. Material and methods

Traditional methods used for the identification of 2.1. Strains

various microorganisms are based on morphological
characteristics, biochemical reactions, serological When carrying out this work, we used the following
reactions and sensitivity to bacteriophage, etc. How- type strains of the different species and biovars of
ever, all of these methods can often be extremely Brucella: B. abortus biovars 1 (strain 544), 2 (strain
tedious, offering results that may take time and on 86/8/59), 3 (strain Tulya), 4 (strain 292), 5 (strain
occasions not prove totally conclusive. B3196), 6 (strain 870), 7 (strain 63/75) and 9 (strain C
Fourier transform infrared spectroscopy (FTIR) is a 68); B. melitensis biovars 1 (strain 16 M), 2 (strain 63/
technique that has been used over the years in chemical 9) and 3 (strain Ether); B. suis biovars 1 (strain 1030),
analysis for the identification of substances and is one 2 (strain Thomsen), 3 (strain 686), 4 (strain RT63/217)
that may also be applied to the characterisation of and 5 (strain 513); B. canis (strain RM 6/66); B. ovis
whole microorganisms. Infrared spectroscopy is based (strain REO 198); and B. neotomae (strain 5K33). In
on the measurement of the molecular bond vibration addition, isolates of clinical strains of B. melitensis, six
compounds, excited by radiation of a suitable frequen- corresponding to biovar 1, one to biovar 2 and five to
cy, when given the conditions for energy absorption by biovar 3, were also studied. In order to fine-tune and
the molecules. In an infrared spectroscopy spectrum, validate the procedure that is the aim of this study, we
different types of energy vibration appear as bands used the following type strains of other microorgan-
(absorptions), each of which is characterised by two isms supplied by the Spanish Collection of Type
basic parameters: position and intensity. The introduc- Cultures (CECT, Valencia, Spain): Staphylococcus
tion of mathematical functions of Fourier transforma- aureus CECT 86, E. coli CECT 515, Streptococcus
tion and the subsequent ongoing development we are pyogenes CECT 985, Enterococcus faecalis CECT
witnessing of computer systems and management of 481, Corynebacterium pseudodiphtheriticum CECT
statistics have enabled easier processing of the infor- 755, Clostridium perfringens CECT 376, Acineto-
mation obtained from spectra, thus increasing the bacter calcoaceticus CECT 441 and Pseudomonas
interest in and usefulness of spectroscopic techniques. aeruginosa CECT 110.
The early applications of FTIR in microorganisms
date back to the 1950s (Greensteet and Norris, 1957; 2.2. Culture conditions
Goulden and Sharpe, 1958). However, it was not until
1988 that Naumann et al. (1998) reintroduced this All of the strains were cultured in Brain Heart
technique as a useful tool in microbial identification. Infusion Broth (Oxoid, Basingstoke, Hampshire, Eng-
This technique has been successfully applied to the land). Incubation was carried out at 37 jC, in aero-
differentiation of a wide range of both gram-positive biosis with constant stirring. Brucella strains were
as well as gram-negative microorganisms and yeasts incubated for 72 h, and the other microorganisms
(Van der Mei et al.,1993; Lin et al., 1998). for 24 h. B. abortus biovars 1, 2, 3 and 4, and B.
Bacteria belonging to the Brucella genus are clas- ovis were incubated in 10% CO2 atmosphere. B.
sified into six recognised species: Brucella melitensis, abortus biovar 2 and B. ovis were cultivated in Brain
Brucella abortus, Brucella suis, Brucella canis, Bru- Heart Infusion with 5% of horse serum (Oxoid). C.
cella ovis and Brucella neotomae, within some of perfringens was incubated in anaerobiosis.
which different biovars have been recorded. The
marked tendency towards variation in the smooth 2.3. Preparation of the samples
rough phase, together with the laboriousness and risk
involved in the methods used in their identification, Each of the strains was cultured in 250-ml Brain
make their typing difficult. Heart Infusion Broth. After incubation formaldehyde
This work is aimed at the application of FTIR to was added to reach a final concentration of 0.5%, and
the characterisation of Brucella at both the genus and left for 48 h at 37 jC. The suspensions were sub-
species level. sequently centrifuged at 2000  g for 30 min. The
 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131 123

Table 1 Table 2
Principal component analysis in each of the windows Discriminant analysis
Window r̄ Principal Variance Strains Likelihood of good Classified as
component explained (%) classification (%) Brucella Other
W1 3.000 – 2.800 F11 97 spp. genera
F12 2 Brucella spp. 96.77 30 1
W2 1.800 – 1.500 F21 55
Other genera 100 0 8
F22 18 Total 97.44 30 9
W3 1.500 – 1.200 F31 88
F32 5 Likelihood of good classification of the strains studied.
W4 1.200 – 900 F41 66
F42 23
W5 900 – 700 F51 98
USA) equipped with an MTC detector. The computer
Percentage of variance explained.
program used was Win FIRST v2.0 Mattson Instru-
ments.
culture sediments were washed three times with dis-
tilled water and centrifuged at 5000  g for 20 min. 2.5. Normalisation of the spectrum
The final sediment was resuspended in 1 ml of
distilled water and lyophilised. From this 0.003 g So as to make all of the spectra comparable, a
was taken and added to 0.3 g of KBr to form a normalisation of the spectra was carried out. This
homogeneous mixture, from which 0.15 g was taken normalisation firstly consisted of an adjustment to a
to form a tablet that was placed in the spectropho- standard base line, the mathematical operation of
tometer to obtain the corresponding FTIR spectrum. which is performed by the apparatus computer system
itself. A value of + 1 is then designated to the
2.4. Obtaining the spectrum maximum transmittance.
The obtained spectrum was subdivided into a series
All the spectra were recorded in the region between of windows or spectral regions (Horbach et al., 1988;
4000 and 400 cm 1 with a Cygmus 100 FT-IR Helm et al., 1991a,b; Van der Mei et al., 1993; Curk et
Spectrometer (Mattson Instruments, Madison, WI, al., 1994; Seltmann et al., 1994).

1
Fig. 1. Graphic representation of the first principal component for two groups of microorganisms in the 1500 – 1200 cm window.
124 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131

Table 3 Table 3 (continued)

Discriminant analysis Strain Microorganism Likelihood of identification as
Strain Microorganism Likelihood of identification as number Brucella Other
number Brucella Other microorganisms
microorganisms
28 Brucella melitensis 0.999819 0.000181
1 Brucella abortus 0.519255 0.480745 biovar 3
biovar 1 29 Brucella melitensis 0.967854 0.032146
2 Brucella abortus 0.999964 0.000036 biovar 3
biovar 2 30 Brucella melitensis 0.538113 0.461887
3 Brucella abortus 0.969441 0.030559 biovar 3
biovar 3 31 Brucella melitensis 0.911385 0.088615
4 Brucella abortus 0.996192 0.003808 biovar 3
biovar 4 32 Staphylococcus 0.000927 0.999073
5 Brucella abortus 0.999361 0.000639 aureus
biovar 5 33 Escherichia coli 0.015084 0.984916
6 Brucella abortus 0.999584 0.000416 34 Enterococcus 0.022196 0.977804
biovar 6 faecalis
7 Brucella abortus 0.998567 0.001433 35 Pseudomonas 0.011563 0.988437
biovar 7 aeruginosa
8 Brucella abortus 0.999981 0.000019 36 Acinetobacter 0.021448 0.978552
biovar 9 calcoaceticus
9 Brucella melitensis 0.976266 0.023734 37 Clostridium 0.476468 0.523532
biovar 1 perfringens
10 Brucella melitensis 0.999153 0.000847 38 Streptococcus 0.008742 0.991258
biovar 2 pyogenes
11 Brucella melitensis 0.994550 0.005450 39 Corynebacterium 0.136866 0.863134
biovar 3 pseudodiphtheriticum
12 Brucella suis 0.999994 0.000006 Likelihood of classification of strains studied.
biovar 1
13 Brucella suis 0.994125 0.005875
biovar 2 W1: between 3000 and 2800 cm 1. Region I is the
14 Brucella suis 0.996042 0.003958
fatty acids where the peaks mark the tension vibrations
biovar 3
15 Brucella suis 0.997754 0.002246 – CH3, >CH2 and QCH of the functional groups present
biovar 4 in the fatty acid components of the bacterial membranes.
16 Brucella suis 0.991336 0.008664 W2: between 1800 and 1500 cm 1 is the region in
biovar 5 which the amide I and amide II bands of proteins and
17 Brucella canis 0.978834 0.021166
peptides are shown.
18 Brucella ovis 0.976772 0.023228
19 Brucella neotomae 0.968888 0.031112 W3: between 1500 and 1200 cm 1. This is the
20 Brucella melitensis 0.975071 0.024929 mixed region as it contains vibrations corresponding
biovar 1 to groups present in fatty acids, proteins, polysacchar-
21 Brucella melitensis 0.865750 0.134250 ides and phosphate derivatives. Some authors (Curk et
biovar 1
al., 1994; Helm et al., 1991b) describe within this
22 Brucella melitensis 0.995880 0.004120
biovar 1 region a sub-region called W31 between 1500 and
23 Brucella melitensis 0.996312 0.003687 1400 cm 1, also known as region II of the fatty acids
biovar 1 where the deformation vibrations are recorded – CH3
24 Brucella melitensis 0.452910 0.547090 and >CH2 of the same functional groups as in W1.
biovar 1
W4: between 1200 and 900 cm 1. Dominated by
25 Brucella melitensis 0.995469 0.004531
biovar 1 the absorption bands known as analogous of the
26 Brucella melitensis 0.908450 0.091550 fingerprints belonging to the carbohydrates present
biovar 2 in the interior of the cell wall.
27 Brucella melitensis 0.892501 0.107499 W5: between 900 and 700 cm 1. Region of the
biovar 3
real fingerprints that shows highly specific spectro-
 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131 125

scopic spectra, still to be designated to components of mation has been performed in two different ways. In
functional or cellular groups. the first, principal component analysis was applied to
the spectral data of the FTIR data that had been
2.6. Statistical analysis normalised and divided into windows. This was
performed in order to reduce the size of the data.
This consists of the transformation of the infor- Linear discriminant analysis was subsequently per-
mation obtained in the spectra into data that may be formed using the Statistica v4.0 program. The vari-
processed automatically. In this study the transfor- ables included in this analysis were the first two of

Table 4
Discriminant analysis
Strain number Microorganism Likelihood of classification as
Brucella Other Brucella
non melitensis microorganisms melitensis
1 Brucella abortus biovar 1 0.417130 0.372303 0.210567
2 Brucella abortus biovar 2 0.876451 0.000013 0.123536
3 Brucella abortus biovar 3 0.056507 0.037103 0.906390
4 Brucella abortus biovar 4 0.720696 0.002398 0.276906
5 Brucella abortus biovar 5 0.829518 0.000280 0.170201
6 Brucella abortus biovar 6 0.994980 0.000009 0.005011
7 Brucella abortus biovar 7 0.949092 0.000220 0.050688
8 Brucella abortus biovar 9 0.983087 0.000001 0.016911
12 Brucella suis biovar 1 0.993218 0.000000 0.006782
13 Brucella suis biovar 2 0.658587 0.004296 0.337118
14 Brucella suis biovar 3 0.816170 0.001763 0.182068
15 Brucella suis biovar 4 0.562800 0.002045 0.435156
16 Brucella suis biovar 5 0.477352 0.008675 0.513974
17 Brucella canis 0.835758 0.008085 0.156157
18 Brucella ovis 0.999485 0.000059 0.000456
19 Brucella neotomae 0.806365 0.013396 0.180239
32 Staphylococcus aureus 0.000017 0.998868 0.001115
33 Escherichia coli 0.000501 0.984249 0.015250
34 Enterococcus faecalis 0.001593 0.978698 0.019709
35 Pseudomonas aeruginosa 0.009165 0.983700 0.007136
36 Acinetobacter calcoaceticus 0.000104 0.971114 0.028782
37 Clostridium perfringens 0.005211 0.504725 0.490065
38 Streptococcus pyogenes 0.000130 0.989702 0.010168
39 Corynebacterium pseudodiphtheriticum 0.004919 0.866450 0.128631
9 Brucella melitensis biovar 1 0.006101 0.023734 0.970165
20 Brucella melitensis biovar 1 0.006101 0.023734 0.970165
21 Brucella melitensis biovar 1 0.027281 0.146798 0.825922
22 Brucella melitensis biovar 1 0.051477 0.005230 0.943293
23 Brucella melitensis biovar 1 0.554494 0.003361 0.442145
24 Brucella melitensis biovar 1 0.216141 0.526019 0.257840
25 Brucella melitensis biovar 1 0.062180 0.005828 0.931991
10 Brucella melitensis biovar 2 0.186875 0.001165 0.811960
26 Brucella melitensis biovar 2 0.025717 0.100424 0.873858
11 Brucella melitensis biovar 3 0.124247 0.007217 0.868536
27 Brucella melitensis biovar 3 0.054436 0.125236 0.820328
28 Brucella melitensis biovar 3 0.808021 0.000090 0.191888
29 Brucella melitensis biovar 3 0.048349 0.038476 0.913175
30 Brucella melitensis biovar 3 0.058695 0.501420 0.439885
31 Brucella melitensis biovar 3 0.113101 0.107359 0.779541
Likelihood of classification of strains studied.
126 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131

1
Fig. 2. Graphic representation of the first principal component for three groups of microorganisms in the 1500 – 1200 cm window.

each window because they explain the most variance, 3. Results

and the method was validated by the correct classi-
fication of the strains. In the other, the data from the The results of principal component analysis are
second derivatives of spectra were used to carry out shown in Table 1 and Fig. 1. In Table 1 the per-
principal component analysis with the SPSS v7.3 centage of variance explained by each factor can be
program. seen. Fig. 1 is the graphic representation of the first

Fig. 3. Two-dimensional representation of the results of discriminant analysis.

 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131 127

Table 5
Wave numbers of the peaks showing greatest variability in the spectra of the type strain of Brucella biovars
Species Biovar r̄1 r̄2 r̄3 r̄4 r̄5 r̄6 r̄7 r̄8
Brucella abortus 1 769 865 896 944 1527 1552 1689 1743
2 771 866 895 939 1524 1558 1683 1741
3 771 865 889 944 1525 1550 1689 1743
4 769 862 892 948 1525 1550 1689 1743
5 769 869 891 943 1525 1552 1689 1745
6 771 868 896 945 1529 1556 1682 1747
7 767 868 890 949 1527 1550 1689 1745
9 771 864 889 941 1526 1552 1689 1745
Brucella melitensis 1 771 855 880 943 1520 1556 1682 1747
2 771 852 879 941 1519 1556 1682 1747
3 771 854 879 943 1520 1556 1682 1747
Brucella suis 1 769 862 895 949 1520 1556 1682 1747
2 767 860 892 943 1521 1545 1680 1737
3 771 846 891 945 1521 1558 1682 1737
4 771 854 898 945 1520 1550 1682 1743
5 771 852 891 945 1520 1558 1682 1747
Brucella ovis 769 858 883 939 1520 1554 1680 1728
Brucella neotomae 763 856 887 943 1520 1556 1682 1749
Brucella canis 771 866 889 945 1520 1556 1682 1730
Mean 770 860 890 944 1523 1554 1684 1743
S.D. 2.1 6.5 5.6 2.8 3.1 3.5 3.4 5.7

principal component in the 1500– 1200 cm 1 region maximum and minimum values of the group formed
considering two groups: Brucella and other micro- by the rest of the microorganisms fall between the
organisms. It can be seen that although the two maximum and minimum values of the Brucella
groups cannot be totally differentiated (since the group), the differentiation offered by the second

Fig. 4. Graphic representation of principal components of Brucella type strains.

128 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131

and third quartiles (25 – 75%) is sufficiently signifi- If three groups are made: B. melitensis, Brucella
cant. spp. non melitensis and other microorganisms, the
Discriminant analysis performed using the princi- likelihood falls but good results are still apparent
pal components of each window (Table 2) yields (Table 4). In Fig. 2, which is the graphic representa-
above 95% likelihood of good classification. Table 3 tion of the first principal component of the 1500–
shows the results of all of the strains studied, 1200 cm 1 region, a similar phenomenon to Fig. 1
expressed as the likelihood of classification in one can be observed. In Fig. 3, which is a graphic re-
group or another. The vast majority of the Brucella presentation of discriminant analysis, it can be seen
strains offer a likelihood of correct classification of that with the axis of abscissa, it is possible to separate
over 95% and more than half of all of them are above the Brucella group from that of other microorgan-
99%. Only one case of B. melitensis biovar 1 is isms, and with the axis of the ordinate the B.
classified in the group of other microorganisms. In melitensis group can be separated. Thus, using both
all cases within the group of other microorganisms the axes jointly enables us to separate the three groups:
classification is correct to a likelihood of over 80% on B. melitensis, other Brucella and other microorgan-
average. isms.

Fig. 5. Relationships between the strains of Brucella studied.

 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131 129

Using the normalised spectra of the Brucella first two factors is necessary in order to reach over
strains, the peak wave number of the second deriva- 90% of the explained variance. The 900 –700 cm 1
tive was calculated automatically. In order to facilitate window has traditionally been considered as the
comparison of these numbers, tables were created, fingerprint region, as a result of which, it is not
where those showing greatest variability were selec- surprising that this is the region where the first factor
ted, which turned out to be those situated around the explaining the greatest percentage of variance is to
following wave numbers (m̄ ) (Table 5): 770, 860, 890, be found. The 3000 – 2800 cm 1 region is that
944, 1523, 1684 and 1743 cm 1. which is designated to movements of the fatty acid
Factorial analysis was carried out with the data from component bonds, present in the bacterial mem-
the type strains (Table 5) and principal components branes. The two regions mentioned are considered
were obtained. The wave numbers who contribute in by many authors (Helm et al., 1991a,b; Curk et al.,
the determination of the first principal component are 1994; Holt et al., 1995; Kümerlee et al., 1998) to be
m̄2, m̄3, m̄5 and m̄7; in the second m̄6 and m̄8; and in the amongst the most useful in the identification of
third m̄8. These three principal components account for bacteria, both when taken into account in isolation
36.5%, 16.6% and 14.7%, respectively, of the var- as well as when combinations of the windows are
iance, making an accumulated total of 67.8%. In Fig. studied. Goodacre et al. (1996), in the study carried
4, which is the graphic representation of the factorial out with streptococci and enterococci, found that the
scores of each strain, a grouping of the strains in first two factors of multivariate analysis of the whole
clusters can be seen, including the different species spectrum account for 78% of the variance, but that if
of Brucella. One of these is made up of B. melitensis, the second derivative of the spectrum is used one
another of B. abortus, and another wider one of B. suis. single factor can account for 91%. Timmns et al.
The B. canis, B. ovis and B. neotomae strains appear (1998) found that the first 15 principal components
separate from the previously described groups. When of the whole spectrum of the Candida species
the clinical strains of Brucella were added to the study, account for 97.5% of the variance, and Haag et al.
it was observed that they were included in the group (1996) for actinomycetes need only three principal
comprising the type strains (Fig. 5). From factorial components of the whole spectrum to account for
analysis, it can be deduced that with the three first 90% of the variance.
components 71% of the variance can be accounted for. A strain is classified into a group when the prob-
ability to belonging to this group is higher than the
probability to belonging to other group. The percent-
4. Discussion age of good classification obtained in this study when
two groups are made, Brucella and other microorgan-
The FTIR spectra obtained from complete bacteria isms, is 97.4% overall. The vast majority of the strains
show wide and overlapping absorption bands, derived studied offer likelihood for good classification of
from the multitude of cellular components (proteins, around 99% and only two strains show lower like-
lipids, carbohydrates, amino acids, etc.), that contain lihood. If three groups are considered the likelihood
more information than that offered by all the chemical falls although highly satisfactory results of 82.1% of
components considered individually. In this way, the good classification are still obtained. In Fig. 3, which
spectra of whole bacteria should be considered as is the two-dimensional representation of discriminant
highly specific models that may be unique, even for analysis, it can be seen that with the axis of abscissa
each strain. (root 1) the Brucella group can be separated from that
From our results of multivariate analysis, it may of the other microorganisms. If the second axis (root
be seen that the greatest percentage of variance 2) is taken into account, the group comprising the B.
explained is due to the first principal component of melitensis strains can be separated and, therefore,
the region between 900 and 700 cm 1, with 98%, using the two axes enables the three groups to be
followed by the first factor of the window between differentiated. Percentages of good classification vary
3000 and 2800 cm 1, with 97%. For the remainder in the literature, depending on whether the spectrum
of the spectral regions used, the combination of the as a whole is considered, subdivision in windows or
130 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131

even a combination of them. The level to which the Acknowledgements

classification is made, species of the genus or strain,
also needs to be taken into account. The strains of B. abortus, B. melitensis, B. canis, B.
In our study, the most useful absorption bands in ovis and B. suis biovars 2, 3 and 4 as well as the clinical
terms of species differentiation and biovars of Brucella strains of B. melitensis were kindly provided by Dr.
were 860 and 890 cm 1, followed by 1743 cm 1. The Abad from the Regional Brucellosis Laboratory,
first two are found in the most characteristic area of the Valladolid (Spain); and the strains of B. suis biovars 1
spectrum (the real fingerprint area) and the third in an and 5 as well as B. neotomae were supplied by Dr.
area where the spectra are relatively free of interfering Verger, National Agricultural Research Institute, Nou-
peaks. The 1743 cm 1 absorption band was designated zilly (France). This work was supported by a grant of
by Hedrick et al. (1991) to vibration frequency of the Instituto de Salud Carlos III (Red Temática de
carbonyl group in methylic esters of fatty acids, using it Investigación de Brueelosis ref. G03/204) (Spain).
in the differentiation of archaebacteria and eubacteria.
This band has also proved useful in the characterisation
of Bacillus cereus (Lin et al., 1998) and species of References
Listeria (Lefier et al., 1997).
Other bands considered in our study belonging to Curk, M.C., Peladan, F., Hubert, J.C., 1994. Identification bactéri-
the amide I and II regions (1660, 1554 and 1515 enne par spectrométrie infrarouge à transformée de Fourier.
cm 1) correspond to the N – H flexion vibrations, Bull. Soc. Fr. Microbiol. 9, 278 – 283.
Goodacre, R., Timmins, E.M., Rooney, P.J., Rowland, J.J., Kell,
characteristic of H in CO – NH –R mono-substituted
D.J., 1996. Rapid identification of Streptococcus and Enterococ-
amides. cus species using diffuse reflectance-absorbance Fourier trans-
The variability of the absorption band towards 700 form infrared spectroscopy and artificial neural networks. FEMS
cm 1 should be used with caution due to its proximity Microbiol. Lett. 140, 233 – 239.
to the instrumental cutoff resulting from the loss of Goulden, J.D.S., Sharpe, M.E., 1958. The infrared absorption spec-
tra of lactobacilli. J. Gen. Microbiol. 19, 76 – 86.
prism energy.
Greensteet, J.E., Norris, K.P., 1957. The existence of differences
Classifications of bacteria in microbiology have between the infrared absorption spectra of bacteria. Spectro-
traditionally been based on the phenotype expression chim. Acta 9, 177 – 182.
of genetic material. One example of this is the study Haag, H., Gremilich, H.U., Bergman, R., Sanglier, J.J., 1996. Char-
of the oxidative metabolism of Brucella. With the acterization and identification of actinomycetes by FTIR spec-
troscopy. J. Microbiol. Methods 27, 157 – 163.
advent of techniques facilitating the study of DNA,
Hedrick, D.B., Nivens, D.E., Staford, C., White, D.C., 1991. Rapid
the basis for classification has changed, in such a differentiation of archaebacteria from eubacteria by diffuse re-
way that to define a new species a wide ranging flectance Fourier-transform IR spectroscopic analysis of lipid
genetic study is currently necessary not only as preparations. J. Microbiol. Methods 13, 67 – 73.
regards DNA G + C content but also in terms of Helm, D., Labischinski, H., Naumann, D., 1991a. Elaboration of a
procedure for identification bacteria using Fourier-transform IR
the possible links as well as differences with other
spectral libraries: a stepwise correlation approach. J. Microbiol.
microorganisms that phylogenetically might be Methods 14, 127 – 142.
related. Helm, D., Labischinski, H., Schallehn, G., Naumann, D., 1991b.
FTIR is based on the study of absorptions corre- Classification and identification of bacteria by Fourier-transform
sponding to different energy vibrations, by specific infrared spectroscopy. J. Gen. Microbiol. 137, 69 – 79.
Holt, C., Hirst, D., Sutherland, A., MacDonald, F., 1995. Discrim-
atomic groupings present in the chemical com-
ination of species in the genus Listeria by Fourier transform
pounds. In the case of its application to the identi- infrared spectroscopy and canonical variate analysis. Appl. En-
fication of bacteria, it is the result of the different viron. Microbiol. 6, 377 – 378.
absorptions, obtained from differences in the chem- Horbach, I., Naumann, D., Fehrenbach, F.J., 1988. Simultaneous
ical composition of the structural components of the infections with different serogroups of Legionella pneumophila
investigated by routine methods and Fourier transform infrared
bacteria. As a result, classifications thus obtained,
spectroscopy. J. Clin. Microbiol. 26, 1106 – 1110.
especially at below species levels, might not coincide Kümerlee, M., Siegfried, S., Seiler, H., 1998. Rapid and reliable
with those obtained by other methodologies based on identification of food borne yeast by Fourier transform infrared
different systems. spectroscopy. Appl. Environ. Microbiol. 64, 2207 – 2214.
 M.A. Miguel Gómez et al. / Journal of Microbiological Methods 55 (2003) 121–131 131

Lefier, D., Hirst, D., Holt, C., Williams, A.G., 1997. Effect of Seltmann, G., Voigt, W., Beer, W., 1994. Application of physico-
sampling procedure and strain variation in Listeria monocyto- chemical typing methods for the epidemiological analysis of
genes on the discrimination of species in the genus Listeria by Salmonella enteritidis strains of phagetype 25/17. Epidemiol.
Fourier transform infrared spectroscopy and canonical variates Infect. 113, 411 – 424.
analysis. FEMS Microbiol. Lett. 147, 45 – 50. Timmns, E.M., Howel, S.A., Alsberg, B.K., Noble, W.C., Goodacre,
Lin, S.F., Schraft, H., Griffiths, M.W., 1998. Identification of Ba- R., 1998. Rapid differentiation of closely related Candida spe-
cillus cereus by Fourier transform infrared spectroscopy (FTIR). cies strains by pyrolysis-mass spectrometry and Fourier trans-
J. Food Prot. 61, 921 – 923. form-infrared spectroscopy. J. Clin. Microbiol. 36, 367 – 374.
Naumann, D., Fijala, V., Labischinski, H., Giesbrecht, P., 1988. The Van der Mei, H.C., Nauman, D., Busscher, H.J., 1993. Grouping
rapid differentiation and identification of pathogenic bacteria oral streptococcal species using Fourier-transform infrared Spec-
using Fourier Transform Infrared spectroscopic and multivariate troscopy in comparison with classical microbiological identifi-
statistical analysis. J. Mol. Struct. 174, 165 – 170. cation. Arch. Oral Biol. 38, 1013 – 1019.