Infrared spectroscopy as a tool for discrimination between sensitive
and multiresistant K562 cells
Anthoula Gaigneaux, Jean-Marie Ruysschaert and Erik Goormaghtigh
Laboratory of Structure and Function of Biological Membranes, Free University of Brussels, Belgium
Fourier transform infrar ed spectroscopy was p erformed on
human leukemic daunorubicin-sensitive K 562 cells and their
multiresistant counterpart derived by selection. Statistical
analysis, including variable reduction and linear discrimi-
nant analysis was performed on sensitive and multiresistant
cells spectra in order to establish a diagnostic tool for
multiresistant pattern. For each of t he two methods of data
reduction tested [genetic algorith m or principal component
analysis (PCA)] discrimination between the two cell lines was
found to be possible. The best results, obtained with
PCA-reduction, showed an accuracy of 93% on a distinct
test set of spectra. These results d emonstrate the efficiency of
Fourier transform infrar ed spectroscopy for c lassification.
Further analysis o f t he spectral differences indicated that
discrimination between r esistant and sensitive cells was
based o n variations in all cellular contents. Lipid and nucleic
acid decreased, relatively, while the protein content
increased.
Keywords: multiresistance; infrared spectroscopy; multivar-
iate statistics; K562; leukemia.
In recent years, infrared spectros copy has b een a powerful
tool for biodiagnostics [1]. A major advantage of infrared
spectroscopy over more classical t echniques of investigation
is that neither s taining of t he samples nor chemical reagent
additions are necessary. Just a few minutes and a few lLof
a cell suspension are sufficient to obtain a spectrum
representative of all cell constituents.

multiple and s tructurally unrelated drugs [8]. I t m ay be
expressed by cells selected for resistanc e to a single agent.
Many of these m ultiresistant cells differ f rom their sensitive
counterpart by overexpression of a membranous protein of
170 k Da, named P-glycoprotein (P-gp) [9]. Although the
sole presence of P -gp has proven in s ome cell lines to con fer
multidrug resistance phenotype [10], previous studies have
shown that molecular c hanges in lipid an d nucle ic acid
fractions of the cells accompany P -gp overexpression
[11,12].
In this study, w e worked with sensitive (K562/DNS) and
multiresistant (K 562/DNR) human chronic myelogenous
leukemia K562 cells. First, we examined whether infrared
spectroscopy, a ssociated w ith d ata reduction techniques a nd
multivariate statistics, is a ble t o identify multidrug resistant
phenotype in t hese cells with a high accuracy. S econd, we
tried to learn more about biological origin of the spectral
differences that exist b etween the K562-multiresistant cell
line and its sensitive counterpart.
MATERIALS AND METHODS
Cell culture
K562 is a human chronic myelogenous leukemia cell line. In
this study, two different K 562 lines were used. The first cell
line ( cell line A) has been described p reviously [13]. A second
Correspondence to G. Erik, Laboratory of Structure and Function of
Biological Membranes, Free University of Brussels, CP 206/2,
Boulevard du Triomphe, B-1050 Brussels, Belgium.
Fax: + 32 2 650 5382, Tel.: + 32 2 650 5386,
E-mail:
Abbreviations: P-gp, P-glycoprotein; K562/DNS, sensitive K562 cells;

FTIR spectroscopy
An aliquot of cell p ellet was deposited on a g ermanium
crystal (  2–5 · 10
5
cells per smear). The sample was
rapidly e vaporated in N
2
flux to obtain a homogenous film
of entire cells. IR m easurements were recorded be tween
4000 a nd 8 00 cm
)1
by a Bruker E quinox spectrophotometer
(Bruker, Karlsruhe, Germany) containing a liquid
N
2
-refrigerated Mercury Cadmium Telluride detector. Each
spectrum w as obtained by averaging of 256 scans at a
resolution of 4 cm
)1
. T he spectra were baseline c orrected
and nor malized f or equal area b etween 1711 and 1485 cm
)1
.
Spectra were encoded every 1 cm
)1
.
Data analysis
All spectra were treated with i n-house s oftware w orking in a
MATLAB
environment (

double c rossing-over, and data were divided in nin e subsets
to cross-validate the models.
As the solutions proposed by this method are not
deterministic, running the algorithm several times allows a
more precise solution to be obtained. Only the wavenum-
bers selected in more than 80% of all m odels built wer e kept
in the final model.
Linear discriminant analysis (LDA).Thisstatistical
multivariate method is supervised. It searches for the
variables containing the greatest interclass variance and
the smallest intraclass variance, and constructs a linear
combination of the variables to discriminate between the
classes. The rule i s constructed with training set of samples,
and further tested with the test set. We performed LDA in
standard method, i.e. including all the variables in the
model.
RESULTS
Spectral information contained in a cell IR spectrum
Figure 1 (line A) shows a representative spectrum of K 562/
DNS cells, whic h can be divided in three regions. The
absorption between 300 0 and 2800 cm
)1
is dominated by
the stretching vibration of CH
2
and CH
3
groups mainly
contained in fatty acids of the cell. The band at 2963 cm
)1

C¼O b onds arises at 1650 cm
)1
(amide I). The deformation
of protein amide N–H bond appears at 1 540 cm
)1
(amide
II) [15]. The 1450 and 1400 cm
)1
bands arise from the side
chain o f p roteins [ 15], but the C –H bend ing vibration of
fatty acids at 1467 and 1450 cm
)1
[3] and the carboxylate
vibration o f fatty acids at 1400 cm
)1
[17] are s uperimposed.
Absorptions between 1300 and 900 cm
)1
arise mainly
from phosphate associated with nucleic acids, i.e. DNA
and RNA. The a bsorption bands a t 1245 a nd 1087 cm
)1
are c haracteristic o f asymmetric and symmetric pho spho-
diester vibration of nucleic acids [15]. In glycogen-poor cells
such as lymphocytes, B enedetti et al. assigned the shoulders
present at 1117 and 1020 cm
)1
to RNA and DNA,
respectively [1 8].
Classification by LDA

PCA was performe d on t he training set. At this stage, only
two or three principal c omponents were sufficient to obtain
a partial separation between the two cell lines; Fig. 2 shows
the s pectra reduced with PCA proje cted on vector 2 and
vector 4. Each one of these two vectors (Fig. 3) h as
features at characteristic wavenumbers of nucleic acids,
lipids, and protein. I t is i nteresting to note that, in the
second vector, a negative influence of 1625 cm
)1
(attribut-
ed to a b eta sheet secondary structure of proteins) is
associated with a positive value of 1 667 cm
)1
(a helix
secondary structure). This may reflect a m odification i n t he
global secondary structure c omposition in the c ells.
Reduced training set spectra was used for model building
in LDA. The results obtained show 100% of co rrect
classification f or the t raining set. F or the test sets ( Table 2),
the global accuracy was 93%.
Table 1 . Results of LDA f or spectra of t he test s e t w he n spectra were
reduced b y genetic algorithm. Overall a ccuracy on the training set w as
100% and overall accuracy on the t est set is 73%. Actual assi gnments
in columns, LDA p redicted assignments i n rows.
K562/DNS K562/DNR Accuracy
K562/DNS 7 0 100%
A line
K562/DNS 6 0 100%
B line
K562/DNR 6 5 45%

Figure 4 reports the m ean spectra of K562/DNS (curve a),
K562/DNR (curve b), a nd the spectrum obtained b y
difference between these two cells lines (sensitive cells
spectrum minus its r esistant counterpart; curve c). Regions
where the spectra of the two cell lines are significantly
different were determined by Student t-test (shaded areas).
In the CH region (3000–2800 cm
)1
), spectra display
significant differences, indicating that the resistant cells have
a p rotein/lipid ratio higher t han s en sitive cells (Table 3). The
decrease of intensities at 1740 cm
)1
(C¼O bonds of lipids) is
also consistent with a decrease of lipid content in multire-
sistant cells. In addition , t he ratio of CH
3
/CH
2
,calculatedas
the ratio of absorbance at 2871 cm
)1
over the a bsorbance a t
2853 c m
)1
, is also significantly higher in K562/DNR cells
than in K562/DNS cells (Table 3). T hese res ults s uggest that
a lipid/protein ratio m odification occurs in the resistant
phenotype. Because proteins contain, on the average, an
equal amount of methyl and methylene groups, a protein

Many studies have provided evidence that infrared spec-
troscopy is a useful a nd powerful tool to screen cell and
tissue evolution occurring during c ancer progression. The
aim of this study was to show that among similar cancerous
cell lines, a subtle change such as expression o f MDR
phenotype can be identified by infrared spectrosc opy.
Fig. 4 . Compariso n of spectra and s pec tral features responsible for cell
type determination. (A) Mean spectrum of sensitive K562 cells. (B)
Mean spectrum of resistant K562 cells. (C) Difference spectrum (sen-
sitive/resistant) magnified by five. (D) Significant differences are
shaded (a ¼ 1% ). Discriminant vector built by LDA.
Table 2. R esults of LDA for spectra of the tes t se t w hen s pectra were
reduced by PCA. Overall a ccuracy o n t he training set was 10 0% and
overall accuracy on t h e test set i s 93%. Actual assignmen ts in columns,
LDA predicted assignments in rows.
K562/DNS K562/DNR Accuracy
K562/DNS 7 0 100%
A line
K562/DNS 4 2 67%
B line
K562/DNR 0 11 100%
A line
K562/DNR 0 6 100%
B line
Table 3. R esults of Student t-tests for equality of m ean between resistant
(22 spectra) a nd sensitive (26 spectra) K562 cells. Spectra were nor-
malized for e qual s urface b etween 1711 and 1585 cm
)1
. Lipid/P rotein:
area between 3000 a nd 2800 cm

In the second part of our work, we focused on the
biochemical information available i n infrared s pectra, and
we found that infrared spectra of resistant K562 cells are
significantly modified relative t o their sensitive counterpart
at all molecular levels.
The protein/nucleic acid ratio is significantly higher in
resistant cells. As discussed b efore, this can be caused
either by a real decrease in n ucleic acid content o r by a n
increase of the compaction state of DNA possibly related
to the cell cycle stage. Concerning the latter h ypothesis,
Boydston-White et al. reported that among nucleic acid
absorption bonds, RNA contribution can be prominent
when cells are i n G1 an d G2 ph ases, because h ighly
compacted DNA is opaque to infrared light [19]. In fact,
less condensed chromatin in K562 MDR cells in G1
phase (com pared to their sensitive c ounterpart in G1 cell
phase) has been previously reported [20], suggesting that
our experimental results can not be explained b y c hrom-
atin condensation. In favour of the former hypothesis,
flow cytometry studies on K562 MDR cell lines have
shown a relative de crease in DNA co ntent at con stant cell
phase distribution o f cells [12]. Our results a re therefore
consistent with a r eal decrease i n nucleic acid content i n
resistant K562 cells.
At the lipid level, we observed in r esistant cells a d ecrease
of the absorptions assigned to fatty a cids and phospholipids
relative to their protein content. This qua ntitative change
was accompanied by qualitative modification, as the
methyl/methylene CH
3

precludes definitive conclusions on the overall chemical
composition of the different cell lines.
In this study, w e conclude th at infrared spectroscopy is a
useful tool to identify K562 multiresistant cell lines. W e also
demonstrated that infrared spectroscopy is a powerful tool
for investigating the global biochemical modifications
related to the multiresistant phenotype.
ACKNOWLEDGEMENTS
Anthoula Gaigneaux is recip ient of a Televie Grant from the Fonds
National de l a Recherche Scientifique (Belgium). We thank A. Delforge
and C. D orval for giving us the second K562 ce ll line.
REFERENCES
1. Jackson, M., S owa, M.G. & Mantsch, H.H. ( 1997) Infrared
spectroscopy: a new frontier in med icine. Biophys. C hem 68,
109–125.
2. Le Gal, J.M., Morjani, H. & Manfait, M. (1993) Ultrastructural
appraisal of the multidrug r esistance in K562 and L R73 cell lines
from Fourier transform infrared spectroscopy. Cancer Res. 53,
3681–3686.
3. Rigas, B. & Wong, P.T. ( 1992) Human c olon adenocarcinoma cell
lines display in frared sp ectrosco pic fe atures o f m alignant c olon
tissues. Cancer Res. 52, 84–88.
4. Na um ann, D., Helm, D. & Labischinski, H. (1991) Micro-
biological characterization s by FT-IR spec troscopy. Nature 351 ,
81–82.
5. Schultz, C.P., Liu, K., J ohnston, J.B. & Mantsch, H.H. (1996)
Study of chr onic lymphocytic l euke mia cells b y FT-IR spectro-
scopy and cluster a nalysis. Leuk. Res. 20, 649– 655.
6. McIntosh, L.M., Jackson, M., Mantsch, H.H., Stranc, M.F.,
Pilavdzic, D. & C rowson, A.N. (1999) I nfrared spectra of basal

Appl. Spectrosc. 53, 148A –16 1A .
16. Wong, P .T., Wong, R .K., Caputo, T .A., Godwin, T .A. & Ri gas, B .
(1991) Infrared spectroscopy of exfoliated human cervical c ells:
evidence of extensive structural changes during carcinogenesis.
Proc. Natl Acad. Sci. USA 88 , 10988–10992.
1972 A. Gaigneaux et al. (Eur. J. Biochem. 269) Ó FEBS 2002
17. J ackson, M., Choo, L.P., Watson, P.H., Halliday, W.C. &
Mantsch, H.H. (1995) Beware of connective tissue proteins:
assignment and implications of collagen absorptions in infrared
spectra of h uman tissues. Biochim. Biophys. Acta 1270,1–6.
18. B enedetti, E ., Bramanti, E ., Papineschi, F . & Rossi, I . ( 1996)
Determination of the relative amount of nucleic aion of t he
relative amount of nucleic acids in leukemic and normal
lymphocytes by means of FT-IR microspectroscopy. Appl. Spec-
trosc. 51, 792–797.
19. B oydston-White , S., Gopen, T., Houser, S., Bargonetti, J . &
Diem, M. ( 1 999) In frared sp ec troscop y of human tissue. V .
Infrared spectroscopic s tudies of my eloid leukemia ( ML-1 ) cells at
different phases of t he cell c ycle. Biospectroscopy 5, 2 19–227.
20. D ufer, J., Millot-Broglio, C., Oum’Hamed, Z., Liautaud-Roger,
F., Joly, P., Desplaces, A. & Jardillier, J.C. (1995) Nuclear DNA
content a nd chromatin texture in multidrug-resistant human l eu-
kemic c ell lines. I nt. J. Cancer 60, 108 –114.
21. Nielsen, D. & Skovsgaard, T . (1992) P-glycoprotein as multidrug
transporter: a critical review of current multidrug resistant cell
lines. Biochim. Biophys. Acta 1139, 169–183.
22. W adkins, R .M. & Ro ep e, P.D. (1997) Biophysical a spects of
P-glycoprotein-mediated m ultidrug resistance. Int. Rev. Cytol.
171, 121–165.
23. Arsenault, A.L., Ling, V. & Kartner, N. (1988) Altered plasma