Proteomics 2007, 7, 0000–0000
1
DOI 10.1002/pmic.200601007
TECHNICAL BRIEF
Differential protein labeling with thiol-reactive infrared
DY-680 and DY-780 maleimides and analysis by
two-dimensional gel electrophoresis
Irène M. Riederer1 and Beat M. Riederer1, 2
1
2
Centre de Neurosciences Psychiatriques, Hôpital Psychiatrique, Prilly, Switzerland
Département de Biologie Cellulaire et de Morphologie, Université de Lausanne, Lausanne, Switzerland
Differential protein labeling with 2-DE separation is an effective method for distinguishing differences in the protein composition of two or more protein samples. Here, we report on a sensitive infrared-based labeling procedure, adding a novel tool to the many labeling possibilities.
Defined amounts of newborn and adult mouse brain proteins and tubulin were exposed to
maleimide-conjugated infrared dyes DY-680 and DY-780 followed by 1- and 2-DE. The procedure
allows amounts of less than 5 mg of cysteine-labeled protein mixtures to be detected (together
with unlabeled proteins) in a single 2-DE step with an LOD of individual proteins in the femtogram range; however, co-migration of unlabeled proteins and subsequent general protein stains
are necessary for a precise comparison. Nevertheless, the most abundant thiol-labeled proteins,
such as tubulin, were identified by MS, with cysteine-containing peptides influencing the accuracy of the identification score. Unfortunately, some infrared-labeled proteins were no longer
detectable by Western blots. In conclusion, differential thiol labeling with infrared dyes provides
an additional tool for detection of low-abundant cysteine-containing proteins and for rapid identification of differences in the protein composition of two sets of protein samples.
Received: December 6, 2006
Revised: January 30, 2007
Accepted: March 2, 2007
Keywords:
Immunoblots / Infrared dyes / Mouse brain / Thiol labeling
Applications that include protein labeling with sensitive
fluorescent dyes such as SYPRO Ruby and SYPRO Orange
or cyanine dyes (Cy2, Cy3, and Cy5) have been widely used in
DIGE for the detection of differences in the protein composition of various sets of proteins [1]. Here, we describe the
use of infrared DY-680 and DY-780 maleimides to label two
sets of proteins for rapid identification of differences in their
protein composition by 2-DE. The method is reproducible
and dyes can be obtained at a reasonable price. One drawback is that only cysteine-containing proteins are labeled,
Correspondence: Dr. Beat M. Riederer, Centre de Neurosciences
Psychiatriques, Hôpital Psychiatrique, 1008 Prilly, Switzerland
E-mail: Beat.Riederer@unil.ch
Fax: 141-21-692-51-05
Abbreviations: Cy, cyanine dyes; GFAP, glial fibrillary acidic protein; SCG10, stellar cervical ganglion protein 10
© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
thus excluding proteins without cysteines. As an example,
newborn and adult mouse brain proteins were labeled with
DY680- and DY780-maleimides (outlined in Fig. 1). These
stains have an Mr of 756.97 g/mol for DY-680 (C42H52N4O7S)
and 783.01 g/mol for DY-780 (C44H54N4O7S), a neutral pH,
and differ in their infrared light emissions by 100 nm due to
a difference in a vinyl residue (www.dyomics.com). Preliminary experiments with individual DY-680 or DY-780-labeled
proteins separated on 1-D gels showed no overlap in emission with the measuring wavelengths of 800 and 700 nm,
respectively. It should be noted that the maleimide dyes are
also available as NHS-conjugated dyes that could be applied
for labeling of all proteins via primary and e-amino groups;
such experiments are currently being performed.
All experiments were authorized by the local veterinary
office. Brains from deeply anesthetized newborn and adult
mice (C57Bl/6) were removed and kept at –807C until use.
Brain samples (25 mg brain tissue or 2.5 mg proteins) were
www.proteomics-journal.com
2
I. M. Riederer and B. M. Riederer
Figure 1. Summary of experimental set-up and procedure. Panel
A represents the common separation of the two differentially
labeled protein samples in 2-DE IEF pH 3–10 in the first dimension and by molecular weights in the second dimension (panel
A); and detection for wavelength specific emission at 700 nm
(newborn mouse brain, panel B) and 800 nm for adult brain proteins (panel C). For comparison unlabeled newborn (panel D) and
adult (panel E) CBB-stained brain proteins are shown for comparison. Arrows point to two stage-specific proteins, GFAP in
green and SCG10 in red (panel A). The molecular weights are
indicated to the right in panel A.
homogenized in 2.5 mL reducing buffer (0.1 M sodium
phosphate pH 6.0, 2.5 mM EDTA, 5 mL/mL protease inhibitors (Sigma, Buchs, Switzerland) and 6 mg 2-mercaptoaethylamine-HCL) and incubated for 90 min at 377C. In
addition, two samples of purified pig brain tubulin (0.5 mg)
[2] were labeled with each dye for a dilution series and for an
estimation of the LOD of a known protein. The reducing
© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Proteomics 2007, 7, 0000–0000
buffer was replaced by a D-salting Dextran desalting column
(Perbio Switzerland, Lausanne), previously equilibrated with
conjugation buffer (PBS, pH 7.5 and 1 mM EDTA). Subsequently 500-mL fractions were collected and protein amounts
were measured by the Bradford Protein Assay (BioRad,
Reinach, Switzerland) or the dot assay and by staining filters
with Amidoblack [2]. Protein-containing fractions were
pooled and concentrated to 500 mL by a centrifugal filter unit
(Amicon Ultra-4, Millipore, Zug, Switzerland). Infrared substances DY-680 and DY-780 maleimides were obtained from
Dyomics (Jena, Germany). Two hundred micrograms of DY680 maleimide in 20 mL DMF was added to 2.5 mg brain
proteins of newborn mice and 200 mg DY-780 maleimides in
20 mL DMF was added to adult brain protein samples and
kept overnight at 47C for saturation labeling of all cysteines.
Subsequently, 40-mg dye samples in 4 mL DMF were added to
tubulin samples and also kept over night at 47C. Samples
were desalted with D-salt Dextran columns and concentrated
by Amicon Ultra-4 centrifugation. A considerable amount of
proteins are lost during desalting and concentration procedures, but between 25 and 40% of the starting material is
obtained in dye-labeled form. For 2-DE, both labeled and
unlabeled proteins were treated with the 2-D clean-up kit
(Amersham, Otelfingen, Switzerland) and resuspended in
strip-rehydration buffer for IEF or in SDS-PAGE sample
buffer. Procedures for 1- and 2-DE as well as for silver nitrate
and CBB staining were previously described [3–5]. Briefly,
5 mg labeled proteins and 400 mg unlabeled proteins were
loaded on IPG ready strips of 11 cm length and pH 3–10
(BioRad) in the first dimension; in the second dimension,
12.5% or gradient 3.6–15% SDS-PAGE was applied. Gels
were scanned by the Odyssey infrared imaging system (LiCor, Bad Homburg, Germany) with detection channels of
700 nm (red) and 800 nm (green) at a sensitivity level setting
of 5, a scanning resolution of 169 mm, and at medium quality. Data sets were analyzed by computer software (Image
master, Amersham/GE Biosciences, Zurich). Proteins were
also transferred to NC filters and Western blots were
exposed to antibodies for tubulin, glial fibrillary acidic protein (GFAP), and neurofilament proteins [4, 5] and detected
with infrared-conjugated secondary antibodies IRD-700DX
or IRD800DX (Rockland, BioConcept, Allschwil, Switzerland). The minimal protein amount that can be used for
labeling is 200 mg proteins as starting material, yielding 40–
50-mg labeled proteins. This is sufficient for several 2-D gels,
while the same amount of unlabeled proteins is only sufficient for a single 2-D gel, CBB staining, and identification
of proteins.
The sample preparation and labeling process of adult
and newborn mouse brain proteins is demonstrated in
Fig. 1. The newborn DY-680-labeled proteins appear in red,
while the adult DY-780-labeled proteins (5 mg each) are seen
in green (Fig. 1A). The two emission patterns were transformed into black and white images, with newborn protein
composition represented in panel B and the adult protein
pattern in panel C. For comparison, unlabeled newborn and
www.proteomics-journal.com
Proteomics 2007, 7, 0000–0000
adult brain proteins (400 mg each) were separated individually and stained with CBB (ProtoBlue Safe, National Diagnostics, Chemie Brunschwig, Basel, Switzerland). The CBBstained proteins are quite useful for MALDI-TOF identification and for comparison with DY-labeled protein patterns.
Despite a certain similarity, it is apparent that a variety of
proteins are not present in the dye-labeled sample. Several
proteins do not contain cysteine residues [6] and are therefore not labeled, escaping detection in the infrared scanner.
Furthermore, there are development-related differences in
protein pattern, e.g. the GFAP (Fig. 1A, arrow pointing to
green protein) [5] is more prominent in adult brain while
stellar cervical ganglion protein 10 (SCG10; arrow pointing
to red spots) is more present in newborn brain. The amino
acid composition of these two proteins reveals a single cysteine residue for GFAP (Swiss-Prot no. P03995) and two
cysteines for SCG10 (P55821). Therefore, a successful labeling of proteins with only a minimal number of cysteines per
molecule is possible.
In Fig. 2, a dilution series of labeled tubulin demonstrates that the detection of labeled brain tubulin is possible
down to 1 fg (Fig. 2A). Given that tubulins contain seven to
eight cysteines, the sensitivity limit of the method is approximately 10 fg. A separation on 1-DE already demonstrated
that differences in protein labeling may occur (Fig. 2B). On
Western blots, a mAb for tubulin (Tu9b) clearly stained btubulin in the unlabeled adult brain protein sample (Fig. 2 C,
lane 2), while in the DY780-labeled sample no positive
detection was possible and staining appeared rather as a
nonspecific smear. A similar result was obtained with two
antibodies for NF-L and GFAP. This suggests that covalently
bound maleimide dyes may interfere with immunological
detection or alter epitopes. It remains to be seen whether
proteins that do not contain cysteines are still immunologically detectable. Regarding the influence of dyes on MALDITOF protein identification, corresponding and equal
amounts of samples from the tubulin region were cut out of
CBB-stained DY-labeled and non-DY labeled samples
(http://www.dyomics.com/48.html) [7] and tested for their
suitability to MALDI-TOF analysis by the Protein Analysis
Facility (PAF) in Epalinges, Switzerland. Tubulins a-1, b-2, b3, and b-5 were successfully identified in both samples with
MASCOT scores for unlabeled tubulins between 899 and
1017 and for DY-780 labeled tubulin between 285 and 677
(http://www.dyomics.com/48.html) [7]. This indicates that
tubulin identification is still possible, but given the lower
scores for dye-labeled proteins, one must suspect that thiollabeled peptide sequences may no longer allow identification
of the correct peptide mass. Tubulins contain between seven
and eight cysteines, and therefore several peptides may
become unavailable, thus explaining the lower MASCOT
scores.
Protein labeling with DY-680 and DY-782 maleimides
was shown to be a sensitive method for detecting minute
quantities of proteins by 2-DE and infrared laser scanning,
with a sensitivity in the femtogram range. The physical
© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Technology
3
Figure 2. The LOD of IR680-labeled tubulin is in the femtogram
range (A), the blank areas (asterisk) represent a saturated protein
concentration. One-dimensional SDS-PAGE with newborn (nb),
adult proteins (ad) or combined (mix) demonstrates that a 1-D
representation is by far sufficient to identify differences (B), the
molecular weights are indicated to the right. Panel C: An infraredlabeling of cysteines altered immunoreactivity of labeled (1) but
not unlabeled (2) tubulin. Note that the immunoreactivity
appears like a smear in lane 1 and not as specific reactivity with
tubulin in lane 2.
properties of infrared dyes did not alter the migration properties very much in the 2-DE separation, since the charge and
size of many proteins remained similar or the same. The
dye-labeled GFAP and SCG10 proteins are found more or
less at the same places as identified in the CBB-stained gel.
Proteins without cysteine residues are obviously not labeled
and therefore escape detection in the Odyssey infrared imaging system.
The high sensitivity of the detection system allows the
use of only a few micrograms of proteins per 2-D gel, but the
protein amounts of most protein spots in such a 2-D gel are
far from sufficient for a positive MALDI-TOF identification,
requiring either higher amounts of DY-labeled proteins or
the addition of unlabeled proteins and a CBB or silver staining for the identification of matching proteins by MS. It has
been shown that DY-680- and DY-780-labeled proteins can
also be stained with CBB and silver nitrate [7], but it is
recommended to image proteins first with an infrared scanner, since CBB-stained and silver nitrate-stained proteins
interfere with the infrared imaging. Since detection with
CBB and silver nitrate staining is possible, other noncovalent
www.proteomics-journal.com
4
I. M. Riederer and B. M. Riederer
stains such as SYPRO Ruby, Deep Purple or Lightning Fast
may also work for the detection of DY-labeled proteins. It
should also be noted that on Western blot filters the proteins
are brought to the same focal plane. This favors a better
detection by the infrared imaging system and may also allow
detection with a variety of noncovalent protein stains [1, 2].
Infrared labeling and DIGE provides an additional, inexpensive tool for identifying differences in cysteine-containing
proteins between two developmental stages, two experimental conditions, or between a control and a pathological
state [2]. By using five or more dyes and an imaging system
with appropriate filters, the separation of many samples
within a single 2-D gel seems feasible.
In the present protocol, all thiol groups were reduced and
the oxidation-sensitive modifications were lost and no longer
distinguishable by 2-D DIGE. It is known that protein oxidation plays a crucial role during aging and a variety of diseases [8]. Therefore, the labeling protocol could be modified
so that a given sample is labeled with one color prior to the
reduction of oxidized cysteines, followed by reduction of oxidized proteins and labeling of the reduced cysteines with
another color [2]. The identification of oxidized proteins may
find a wide application in the study of different pathologies
and during normal aging.
© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Proteomics 2007, 7, 0000–0000
The authors thank Dr. F. Lehmann (Dyomics GmBH) for
his helpful suggestions and support. This work was supported by
an FNRS grant (31-067201.01).
References
[1] Patton, W. F., J. Chromatogr. B: Anal. Technol. Biomed. Life
Sci. 2002, 771, 3–31.
[2] Bondallaz, P., Barbier, A., Soehrman, S. et al., Cell Motility
Cytoskeleton 2006, 63, 681–695.
[3] Porchet, R., Probst, A., Bouras, C. et al., Proteomics 2003, 3,
1476–1485.
[4] Porchet, R., Probst, A., Dráberova, E. et al., Neuroreport 2003,
14, 929–933.
[5] Shaw, M., Riederer, B. M., Proteomics 2003, 3, 1408–1417.
[6] Patton, W., Electrophoresis, 2000, 21, 1123–1144.
[7] Riederer, I. M., Riederer, B. M., in: Palagi, P. M., Quadroni, M.,
Rossier, J. S., Sanchez, J. C., Stöcklin, R. (Eds.), Proceedings
of the Swiss Proteomics Society, Fontis Media, Bern, Switzerland, 2004, pp. 183–185.
[8] Paget, M. S. B., Buttner, M. J., Annu. Rev. Genet. 2003, 37, 91–
121.
www.proteomics-journal.com