Synthetic peptides as artificial receptors towards proteins from genetically modified organisms.

Abstract

The aim of this work was the preparation of peptide ligands with good affinity and selectivity towards proteins from genetically modified organisms, namely neomycin phosphotransferase II (Npt II) and the endotoxin Cry1A. A 12 x 12 combinatorial solid phase synthesis in aqueous medium was performed to prepare peptide libraries. From this library, two dipeptides with binding properties towards the chosen ligands (Pro-Lys for Npt II, K(eq) 7.59 x 10(4)M(-1); Trp-Gln for Cry 1A, K(eq) 4.35 x 10(4)M(-1)) were selected as scaffolds for the synthesis of new tetrapeptide libraries. The equilibrium constants of the newly selected tetrapeptides increased slightly respect to the dipeptides (Pro-Lys-His-Phe for Npt II, K(eq) 7.88 x 10(4)M(-1); Trp-Gln-Ala-Phe for Cry 1A, K(eq) 5.65 x 10(4)M(-1)), but selectivity towards other proteins (wheat gliadins, bovine gamma-globulins, bovine serum albumin and chicken ovalbumin) became higher. It was demonstrated that selected tetrapeptides recognised well the ligands also in presence of very complex mixtures of potentially interfering proteins, such as whole cell lysates. This approach can be considered as a general method to obtain tailor-made reagents with antibody-like binding properties towards biomacromolecules.

Full text

Page 1

Biosensors and Bioelectronics 24 (2008) 493–497
Contents lists available at ScienceDirect
Biosensors and Bioelectronics
journal homepage: www.elsevier.com/locate/bios
Short communication
Synthetic peptides as artificial receptors towards proteins from
genetically modified organisms
Cinzia Tozzi, Laura Anfossi, Claudio Baggiani∗, Cristina Giovannoli, Gianfranco Giraudi
Laboratory of Bioanalytical Chemistry, Department of Analytical Chemistry, University of Torino, Via P.Giuria, 5, 10125 Torino, Italy
a r t i c l ei n f o
Article history:
Received 13 March 2008
Received in revised form 23 June 2008
Accepted 24 June 2008
Available online 3 July 2008
Keywords:
Combinatorial libraries
Genetically modified organism
Solid phase peptide synthesis
Artificial receptor
Molecular recognition
a b s t r a c t
The aim of this work was the preparation of peptide ligands with good affinity and selectivity towards
proteins from genetically modified organisms, namely neomycin phosphotransferase II (Npt II) and the
endotoxin Cry1A. A 12×12 combinatorial solid phase synthesis in aqueous medium was performed to
prepare peptide libraries. From this library, two dipeptides with binding properties towards the chosen
ligands (Pro-Lys for Npt II, Keq7.59×104M−1; Trp-Gln for Cry 1A, Keq4.35×104M−1) were selected as
scaffolds for the synthesis of new tetrapeptide libraries. The equilibrium constants of the newly selected
tetrapeptides increased slightly respect to the dipeptides (Pro-Lys-His-Phe for Npt II, Keq7.88×104M−1;
Trp-Gln-Ala-Phe for Cry 1A, Keq5.65×104M−1), but selectivity towards other proteins (wheat gliadins,
bovine ?-globulins, bovine serum albumin and chicken ovalbumin) became higher. It was demonstrated
thatselectedtetrapeptidesrecognisedwelltheligandsalsoinpresenceofverycomplexmixturesofpoten-
tially interfering proteins, such as whole cell lysates. This approach can be considered as a general method
to obtain tailor-made reagents with antibody-like binding properties towards biomacromolecules.
© 2008 Elsevier B.V. All rights reserved.
1. Introduction
The development of new therapeutic drugs, the growth of the
biotechnology industry and the results surrounding the impact of
genomics and proteomics on human health are a few examples
of fields that will require the development of new analytical and
preparative tools. These new approaches should permit the resolu-
tion and the characterisation of complex sets of molecule mixtures
in a high-throughput mode and the subsequent identification of
the target molecule. In the case of genetically modified organisms
(GMOs), their use in food has led to public demand for effec-
tive freedom of choice between GM and non-GM foods. European
legislation (Labelling Regulation EC/1139/98 and EC/49/2000), for
example, requires the compulsory labelling of foodstuffs produced
from GMOs. Moreover, several incidents involving GMOs in food
supplies have also served to raise awareness in USA (Baker and
Burnham, 2002).
The surveillance of food labelling requires the development
of valid analytical methods to detect and accurately verify the
presence of GMOs in a wide variety of raw and processed food.
Many analytical techniques have exploited the DNA determination
(Lockley and Bardsley, 2000; Ahmed, 2002), but the equipment
could be expensive and need skilled operators because of problems
∗Corresponding author. Tel.: +39 011 6707846; fax: +39 011 6707615.
E-mail address: claudio.baggiani@unito.it (C. Baggiani).
to obtain DNA in good and sufficient quantities. Other available
procedures determined the concentration of proteins expressed
by GMOs by using immunoassay techniques (Brett et al., 1999).
Antibodies can be very selective, and allow us to reach high sensi-
tivity levels. Anyway, immunoassays are rather expensive because
these ligands require purification, can be contaminated, show lot-
to-lotvariationsandcanbefragileandcostlytoproduce.Moreover,
difficulties are often experienced in translating the efficacy of the
ligands identified by laboratory screening assays into that dictated
by large scale production, different matrix samples and environ-
mentalconditionsofapplication.Allthesefactorshavecontributed
tothewidespreadopinionthatnewsyntheticsystems,whichcould
mimic recognition properties of natural ligands, had to be devel-
oped (Stevenson, 1996; Lowe et al., 2001).
Inthelastfewdecades,scientificresearchhastriedtosubstitute
the classic affinity devices with synthetic ligands, combining the
selectivity of the natural ones with the high capacity, durability
and cost-effectiveness of the synthetic systems (Li et al., 1998;
Meloen et al., 2000; Otto et al., 2002). Above all, the use of peptides
as synthetic ligands are becoming better recognised as peptides
were used successfully for the development of new synthetic
ligands (Verdolina et al., 2000; Tribbick, 2002; Tribbick and Rodda,
2002). Well-designed peptides have great potential as capturing
agents because of their varied chemical properties and functional
groups, their different physical properties and their well-known
and abundant synthetic approaches. Many peptides could be
easily purchased at low cost, even if some specific sequences
0956-5663/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.bios.2008.06.035

Page 2

494
C. Tozzi et al. / Biosensors and Bioelectronics 24 (2008) 493–497
could be obtained only by difficult syntheses that need expensive
reagents to prepare a sufficient amount of peptide. The elaboration
of a peptide library prepared by either chemical or biological
methods has become a format which is useful for the discovery of
peptides with specific recognition properties because of powerful
synthesising and screening approaches, and it is the first step for
preparing small synthetic recognition systems (Burbaum et al.,
1997; Hruby et al., 1997).
Recently, in our laboratory we have developed peptide ligands
for small-molecular-mass analytes, such as estrogens, aflatoxins
and ochratoxin A (Tozzi et al., 2002, 2003; Giraudi et al., 2007). In
those works, we exploited a novel combinatorial approach where
subsequent peptide libraries were prepared by performing combi-
natorial solid phase syntheses and peptides with the high binding
capacity and selectivity were obtained. The aim of this work was
the demonstration of the feasibility of this approach also towards
large dimensions analytes, such as proteins. In particular, we chose
two GMO proteins as analytes, the neomycin phosphotransferase
II (Npt II) and the endotoxin Cry 1A. The Npt II is used as a selec-
tion marker in genetically modified plants – for example, potato
and tomato – and can be a suitable target for GMO screening (Dai
et al., 2002; Sawahel, 2002; Cui et al., 2003). The second protein,
Cry 1A, increases the insect resistance of plants when expressed
(Jamesetal.,2003;Jurat-Fuentesetal.,2003).Themostwidespread
methods for determining these kinds of proteins are immunoassay
techniques(Brettetal.,1999),buttheyrequireexpensiveantibodies
that are also difficult to find on the market. Moreover, also pro-
tein standards are not so common and the quantities commercially
available are limited and very expensive.
Thus, we prepared synthetic peptide ligands with molecular
recognition properties towards these proteins, potentially useful in
developing cheap analytical methods for screening GMO samples
or as labelling probes coupled with more sophisticated analytical
approaches.
2. Materials and methods
2.1. Materials
All chemicals for buffers, organic solvents (HPLC analytical
grade),aminoacids(enantiomericpurityof99.9%)andreagentsfor
solidphasepeptidesynthesiswerefromVWRInternational(Milan,
Italy).Proteinsandcross-linkedpolystyrenebeadsforpeptidesolid
phase synthesis (Amberlite IRC-50, 10mmolg−1superficial car-
boxylic groups) were from Sigma–Aldrich (Milan, Italy). Coomassie
BluedyewasfromPierce(Rockford,IL,USA).Polystyrenemicrotiter
plates were from Biohit (Helsinki, Finland) and the 96-well filtra-
tion plates were from Millipore (Bedford, MA, USA). The filtration
device and the SPE 96-extraction plate were purchased from IST
(Mid Glamorgan, UK).
The immunoassay kit for Npt II (PathoScreen kit for Neomycin
Phosphotransferase II PSP 73000-LOQ 1ng/well, dynamic range
0.2–200ng/well) and Cry 1A (QualiPlateTMAP003SP-LOQ 0.1ng/
well, dynamic range 0.1–1500ng/well) were supplied by Agdia Inc.
(Elkhart, IN, USA) and Envirologix Inc. (Portland, MA, USA), respec-
tively. Whole cell lysate extracts, purified Npt II extracted from
AgrobacteriumtumefaciensandpurifiedendotoxinCry1Aextracted
from Escherichia coli were kindly supplied by Dr. Gribaudo of Cen-
tro Nazionale delle Ricerche of Torino and by Prof. Marzari of the
University of Trieste, respectively.
2.2. Solid phase synthesis of peptide libraries
12×12 peptide libraries were prepared in according with the
solid phase synthesis approach previously published (Tozzi et
al., 2002, 2003; Giraudi et al., 2007) and reported in electronic
supplementary material.
2.3. Screening for peptides binding
For each library, 250?l of a solution of protein in phosphate
buffer 20mmoll−1(20mmoll−1phosphate, 0.13moll−1sodium
chloride, 1mmoll−1EDTA, pH 7.4), containing 0.4 and 0.5?gml−1
for Npt II and Cry 1A, 15?gml−1for the other proteins (wheat
gliadins, bovine ?-globulins, bovine serum albumin and chicken
ovalbumin), respectively, was added into the microplate wells. The
beads were incubated overnight at room temperature under con-
tinuous mixing by using a vibrating plate. They were filtered and
the supernatant was recovered in a microtiter plate placed below
by using the filtration device. Thus, 100?l of filtered protein solu-
tion was used to measure the unbound fraction (F) of protein. Npt
II and Cry 1A were measured by the specific immunoassay, while
theotherproteinsweremeasuredbycolorimetryinaccordingwith
the Coomassie Blue method (Bradford, 1976). Non-specific binding
towards the proteins was evaluated by replacing the beads func-
tionalised with the peptides with beads on which the spacer arm
was blocked with ethanolamine. The total amount of the protein in
solution (T) was determined by dispensing an equal volume of pro-
tein solution into four wells of the filtration plate and measuring
it directly. The bound protein (B) was calculated as T−F, and the
ratio B/F was used to express the affinity of each peptide towards
the proteins.
After the filtration step, to remove any residual protein from the
library, the beads were incubated for 30min at room temperature
under continuous mixing with the phosphate buffer containing
sodium dodecylsulphate 0.5% (v/v), repeating this procedure four
times. Cleaned beads were washed three times with phosphate
buffertoeliminatethedetergentandstoredsuspendedinthesame
buffer.
2.4. Determination of the equilibrium constants
For each peptide sequence of which the equilibrium constant
was measured, 250?l of a 60mg/ml bead suspension in phosphate
buffer was dispensed into the wells of a filtration microplate and
the buffer was removed by filtration. Then, 250?l of a protein solu-
tion in the same buffer was added in duplicate (concentrations
ranging between 1 and 30?moll−1). The beads were incubated
with the protein solution overnight under continuous mixing. The
supernatants were recovered by filtration and the protein con-
centration was determined by using a specific immunoassay for
Npt II and Cry 1A, otherwise through the Bradford method. The
total signal of each protein solution was determined by dispens-
ing an equal volume of each dilution into wells of the filtration
plate.
The free protein and the total protein concentrations were mea-
sured and the bound protein was calculated from this data for each
dilution. The equilibrium constants were then obtained by fitting
a linear regression through this data in according with Scatchard
(1949).
2.5. Extraction of GMO proteins from cell extracts
250?l of a 60mg/ml suspension in phosphate buffer of beads
supporting Npt II or Cry 1A binding peptides were dispensed in
triplicate into glass vials. Then, 250?l of a cell lysate extract con-
taining known amounts of proteins was added. The beads were
incubated 1h at room temperature under continuous mixing. The
supernatants were recovered after centrifugation, and Npt II or
Cry 1A residual concentrations were determined by using the spe-

Page 3

C. Tozzi et al. / Biosensors and Bioelectronics 24 (2008) 493–497
495
cific immunoassay. The recovery for each sample was calculated by
dividing the recovered protein concentration by the initial concen-
tration and by multiplying it by 100.
3. Results and discussion
3.1. Development of tetrapeptide libraries binding Npt II and Cry
1A
We have reported previously a combinatorial approach to
developpeptiderecognitionsystemstowardssmallmolecularmass
analytes (Tozzi et al., 2002, 2003; Giraudi et al., 2007). Compared
withotherwellknownscreeningmethods–e.g.phagedisplaytech-
nology–ourapproachisbasedontheuseofaverysimpleandcheap
chemistryandonarecursivescreeningofthebinding,whichallows
selecting the “best binding peptide” at each level of the developing
library,thuslimitinggreatlythenumberofpeptidestobescreened.
The aim of this work was to demonstrate the feasibility of this
approach to the development of peptide with molecular recogni-
tion properties towards biomacromolecules, such as proteins from
genetically modified organisms.
The binding pattern for Npt II and Cry 1A of the 144 dipeptides
synthesised is reported in Fig. 1. The sequence Pro-Lys and Trp-Gln
showthebestbindingpropertiestowardsNptII(B/F=0.77)andCry
1A (B/F=0.44). We must underline that these results were repro-
ducible as the test was repeated eight times on the same library
and the coefficients of variation were lower than 3%. Moreover,
the synthesis sequence for obtaining the peptide was repeated in
five different experiments and the binding capacity towards Npt II
and Cry 1A was checked each time. The coefficient of variation was
less than 10%. As lysine was present in the amino acid sequence
which showed the highest binding capacity towards the Npt II, in
this case two different peptides (bound either via an ?- or an ?-
amino group) were synthesised on the solid phase and the binding
properties were probably defined by both these contributions as,
statistically,thepercentageoflysineimmobilisedviaan?-aminoor
an ?-amino group would be 50% for each. However, amino acid ori-
entation was not considered in this work on the peptide–protein
interactions, because the synthesis was reproducible and thus
the same amino acid orientation was statistically obtained each
time.
For the sequences Pro-Lys and Trp-Gln, the binding properties
towardsseveralproteinsweremeasured.Theselectivityofthepep-
tides towards other proteins than Npt II or Cry 1A (wheat gliadins,
bovine ?-globulins, bovine serum albumin, chicken ovalbumin)
were calculated by dividing the peptide equilibrium constant
towards the other protein by the peptide equilibrium constant
towards the GMO protein and then by multiplying it by 100. As
reported in Table 1, the value of the equilibrium constant towards
Npt II is relatively higher considered that are only two amino acids
but also the other proteins are recognised, even if in a limited
extent. The equilibrium constant of peptide Trp-Gln towards Cry
Fig. 1. Npt II and Cry 1A binding measured as bound-to-free ratio (B/F, mean value
on eight replicates, mean coefficient of variation <3%) for the 144 dipeptides of the
first-generation library.
1A is lower than the one showed by Pro-Lys for Npt II, but it is
always satisfactory. Trp-Gln can be considered as a less selective
sequence than Pro-Lys, as it also bounds Npt II in the same way as
the Cry 1A. So, this sequence can be considered as a ligand for the
Npt II also.
Thesedipeptidesequenceswereusedasscaffoldstopreparetwo
distinct 12×12 tetrapeptide libraries, one for the Npt II and one for
the Cry 1A. The binding towards the target proteins was checked
Table 1
Equilibrium constants (Keq) and selectivity (˛) of the peptides Pro-Lys, Trp-Gln (first level-library) and Pro-Lys-His-Phe and Trp-Gln-Ala-Phe (second level-library) towards
Npt II, Cry 1A, wheat gliadins (GLI), bovine ?-globulins (BGG), bovine serum albumin (BSA) and chicken ovalbumin (OVA)
Pro-Lys Trp-GlnPro-Lys-His-PheTrp-Gln-Ala-Phe
Keq(×10−4M−1)
7.59 ± 0.53
1.71 ± 0.40
1.16 ± 0.37
1.46 ± 0.52
1.55 ± 0.45
0.84 ± 0.51
˛
Keq(×10−4M−1)
4.04 ± 0.44
4.35 ± 0.45
1.47 ± 0.35
0.57 ± 0.24
2.10 ± 0.53
0.53 ± 0.12
˛
Keq(×10−4M−1)
7.88±0.68
2.24±0.22
0.30±0.11
<0.01
0.50±0.12
0.31±0.09
˛
Keq(×10−4M−1)
<0.01
5.65±0.47
0.38±0.13
0.64±0.25
0.64±0.23
0.50±0.19
˛
Npt II
Cry 1A
GLI
BGG
BSA
OVA
100
22.5±5.3
15.3±4.9
19.2±6.8
20.4±5.8
11.1±6.7
92.9±10.1
100
33.9±8.1
13.1±5.5
48.2±12.0
12.3±2.8
100
28.5±2.8
3.8±1.4
<0.001
6.3±1.5
4.0±1.2
<0.001
100
6.8±2.3
11.3±4.4
11.4±4.1
8.8±3.3

Page 4

496
C. Tozzi et al. / Biosensors and Bioelectronics 24 (2008) 493–497
Fig. 2. Npt II and Cry 1A binding measured as bound-to-free ratio (B/F, mean value
on eight replicates, mean coefficient of variation <3%) for the 144 dipeptides of the
second-generation library.
again, as reported in Fig. 2, and the best recognition properties
towards the Npt II and Cry 1A resulted Pro-Lys-His-Phe (B/F=1.06)
and Trp-Gln-Ala-Phe (B/F=0.76), respectively.
The binding of these peptides towards other proteins than Npt
II and Cry 1A was checked again by using bovine serum albumin,
bovine ?-globulins, chicken ovalbumin and wheat gliadins as test
analytes.Interestingly,theequilibriumconstantstowardsNptIIand
Cry 1A do not increase significantly, but peptide affinity for the
other proteins becomes very low. The Npt II binding sequence (Pro-
Lys-His-Phe) shows a selectivity similar to the one shown by the
related dipeptide Pro-Lys, while the Cry 1A binding sequence (Trp-
Gln-Ala-Phe) results to be selective for that protein, with a marked
difference respect to the related dipeptide Trp-Gln, which bounds
Npt II in the same way as the Cry 1A.
3.2. Extraction of GMO proteins from cell extracts
As these preliminary results were promising, the binding prop-
erties of the tetrapeptide sequences were checked on real samples.
Npt II was selectively extracted by a whole lysate of A. tumefaciens
cells, with a Npt II amount of 6.0±0.40ng (ELISA assay) and a total
protein amount of 120±7?g (Bradford method), while Cry 1A was
selectively extracted by a whole lysate of E. coli cells, with a Cry
1A concentration of 2.5±0.12ng (ELISA assay) and a total protein
concentration of 100±8?gml−1(Bradford method). It should be
noted that both whole lysate samples contained GMO proteins as
minor components, with a difference of concentration greater than
104times respect to the remaining proteins.
After incubating the samples in the presence of the beads with
the immobilised peptides, the result was that beads supporting the
tetrapeptide Pro-Lys-His-Phe were able to bind 5.0±0.05ng of Npt
II in phosphate buffer and 4.2±0.5ng of Npt II in whole lysate of
A. tumefaciens cells, with recovers of 83% and 77%, respectively. The
same beads were able to bind 2.0±0.04ng of Cry 1A in phosphate
buffer and 1.9±0.7ng of Cry 1A in whole lysate of E. coli cells, with
recovers of 80% and 76%, respectively. When beads supporting the
tetrapeptide Trp-Gln-Ala-Phe were used, no adsorption of Npt II
was detected, while the adsorption of Cry 1A was 1.8±0.03ng in
phosphate buffer and 1.7±0.03ng in whole lysate of E. coli cells,
with recovers of 72% and 68%, respectively.
These preliminary results show that beads supporting the pep-
tides Pro-Lys-His-Phe and Trp-Gln-Ala-Phe have the same binding
behaviour described during the development of the combinatorial
libraries when Npt II and Cry 1A are present in complex samples,
such as whole cell lysate extracts, and that they are able to effi-
ciently extract these proteins.
4. Conclusions
This work confirms the feasibility of peptide combinatorial
libraries as synthetic receptors with binding capacity towards
large molecules, such as proteins. The experimental approach here
described allowed to select binding amino acid sequences in a
relatively quick and easy way with minimal screening work. More-
over, this approach did not need a great quantity of analyte, and
in the case of GMO proteins it was important as these proteins are
very expensive and difficult to find on the market. The preliminary
results obtained on real samples represented by whole cell lysate
extracts are very promising and show that the selected sequences
areabletoselectivelybindtheanalytesinacomplexmatrixinpres-
enceofveryhighconcentrationsofpotentiallyinterferingproteins.
Acknowledgement
This research was supported by the Italian Ministry of
Instruction, University and Research (MIUR project COFIN01-
2001032971).
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.bios.2008.06.035.
References
Ahmed, F.E., 2002. Trends Biotechnol. 20, 215–223.
Baker, G., Burnham, T.A., 2002. Int. Food. Agribus. Manag. Rev. 4, 351–360.
Bradford, M.M., 1976. Anal. Biochem. 72, 248–254.
Brett, G.M., Chambers, S.J., Huang, L., Morgan, M.R.A., 1999. Food Control. 10,
401–406.
Burbaum, J.J., Sigal, N.H., Nolan, H., 1997. Curr. Opin. Chem. Biol. 1, 72–78.

Page 5

C. Tozzi et al. / Biosensors and Bioelectronics 24 (2008) 493–497
497
Cui, M.L., Handa, T., Ezura, H., 2003. Mol. Genet. Genomics 270, 296–302.
Dai,N.,German,M.A.,Matsevitz,T.,Hanael,R.,Swartzberg,D.,Yeselson,Y.,Petreikov,
M., Schaffer, A.A., Granot, D., 2002. Plant Sci. 162, 423–430.
Giraudi, G., Anfossi, L., Baggiani, C., Giovannoli, C., Tozzi, C., 2007. J. Chromatogr. A
1175, 174–180.
Hruby, V.J., Ahn, J.M., Liao, S., 1997. Curr. Opin. Chem. Biol. 1, 114–119.
James, D., Schmidt, A.M., Wall, E., Green, M., Masri, S., 2003. J. Agric. Food Chem. 51,
5829–5834.
Jurat-Fuentes, J.L., Gould, F.L., Adang, M.J., 2003. Appl. Environ. Microbiol. 69,
5898–5906.
Li, R.X., Dowd, V., Stewart, D.J., Burton, S.J., Lowe, C.R., 1998. Nat. Biotechnol. 16,
190–195.
Lockley, A.K., Bardsley, R.G., 2000. Trends Food Sci. Technol. 11, 67–77.
Lowe, C.R., Lowe, A.R., Gupta, G., 2001. J. Biochem. Biophys. Methods 19, 561–574.
Meloen, R.H., Puijk, W.C., Slootstra, J.W., 2000. J. Mol. Recognit. 13, 352–359.
Otto, S., Furlan, R.L.E., Sanders, J.K.M., 2002. Science 297, 590–593.
Sawahel, W.A., 2002. Cell. Mol. Biol. Lett. 7, 19–29.
Scatchard, G., 1949. Ann. N.Y. Acad. Sci. 61, 660–672.
Stevenson, R., 1996. Am. Biotechnol. Lab. 14, 6–12.
Tribbick, G., 2002. J. Immunol. Methods 267, 27–35.
Tribbick, G., Rodda, S., 2002. J. Mol. Recognit. 15, 306–310.
Verdolina, A., Basile, G., Fassina, G., 2000. J. Chromatogr. B 749, 233–242.
Tozzi,C.,Anfossi,L.,Baggiani,C.,Giovannoli,C.,Giraudi,G.,2003.Anal.Bional.Chem.
375, 994–999.
Tozzi, C., Anfossi, L., Giraudi, G., Giovannoli, C., Baggiani, C., Vanni, A., 2002. J. Chro-
matogr. A 966, 71–79.