Content uploaded by Markus Stoeckli
Author content
All content in this area was uploaded by Markus Stoeckli on Jul 15, 2014
Content may be subject to copyright.
NATURE MEDICINE • VOLUME 7 • NUMBER 4 • APRIL 2001 493
NEW TECHNOLOGY
The molecular specificity and sensitivity of mass spectrometry
(MS) has been employed in a new technology for direct map-
ping and imaging of biomolecules present in tissue sections.
This technology has been developed using matrix-assisted laser
desorption/ionization MS (MALDI MS)
1
and has been initially
targeted for the analysis of peptides and proteins present on or
near the surface of tissue sections
2
. Imaging MS brings a new
tool to bear on the problem of unraveling and understanding
the molecular complexities of cells. It joins techniques such as
immunochemistry and fluorescence microscopy for the study of
the spatial arrangement of molecules within biological tissues.
Many previous experiments using MS to image samples have fo-
cused on the measurement of the distribution of elements and
small molecules in biological specimens, including tissue slices
and individual cells
3–5
. An extensive review on imaging by MS
can be found in the article by Pacholski and Winograd
6
.
Technological aspects
For the molecular image analysis, tissue samples can be pre-
pared using several protocols: direct analysis of fresh frozen sec-
tions
7–9
, individual cells or clusters of cells isolated by
laser-capture microdissection or contact blotting of a tissue on a
membrane target
10
. In a typical preparation procedure (Fig. 1),
we mounted a frozen section of tissue on a stainless steel target
plate, coated it with a solution of matrix (for example, sinapinic
acid), then dried and introduced into the vacuum inlet of the
mass spectrometer (Voyager Elite DE, Applied Biosystems,
Framingham, Massachusetts). The instrument was controlled by
MS imaging software written in our laboratory
11
. We created
molecular images from a raster over the surface of the sample
with consecutive laser spots (∼25 µm in diameter). The laser po-
sition was fixed and the sample plate was repositioned for con-
secutive spots. Each spot produced a mass spectrum obtained
from molecules present within the irradiated area. Typically,
each mass spectrum was the average of 50 laser shots acquired
using a fast transient recorder PC board (DP211, Acqiris,
Geneva, Switzerland). With a laser frequency of 20 Hz, the time
cycle was about 2.5 seconds per data point, including acquisi-
tion, data download to the computer, data processing and repo-
sitioning of the sample stage. A typical data array was
1,000–30,000 spots depending on the desired image resolution,
which contains the intensity of ions desorbed at each spot in a
molecular weight range of 500 D to over 80 kD. For most tissue
sections, we recorded over 200 protein and peptide peaks in the
mass spectrum from each spot ablated by the laser. We could
produce an MS image or molecular weight-specific map of the
sample at any desired molecular weight value. It is commonly
possible to generate individual maps to verify the presence, mol-
ecular weight and location of proteins. In the fullest extent,
from a single raster of a piece of tissue, imaging MS could pro-
duce hundreds of image maps each at a discrete molecular
weight value.
Application to mammalian tissue
We used imaging MS to study normal tissue sections from
mouse brain and human brain tumor xenograph sections. These
samples contained well-defined regions, many of which had
subsets of proteins and peptides in a unique distribution or
array. The bilateral symmetry of the brain provides an internal
confirmation of the localized distribution of proteins and the
homogeneity of the prepared tissue sections. An optical image
of the normal mouse brain section fixed on a metal plate and
coated with matrix is shown in Fig. 2a. We scanned the section
by acquiring 170 × 90 spots with a spot-to-spot center distance
of 100 µm in each direction. We recorded ions occurring in 82
different mass ranges and created images by integrating the
peak areas and plotting the relative values using a color scale.
For specific molecular images, we acquired data in a window de-
limited by two mass-to-charge (m/z) units on either side of the
molecular peak. Although many of the protein signals were
common to all areas of the brain, some were found to be highly
specific for a given brain region. For example, the protein de-
tected at m/z 8258 ± 1 (Fig. 2b) was present in the regions of the
cerebral cortex and the hippocampus; the protein at m/z 6716 ±
1 (Fig. 2c) was localized in the regions of the substantia nigra
and medial geniculate nucleus; and the peptide at m/z 2564 ± 1
was in the midbrain (Fig. 2d). These ions are [M+H]
+
species, and
the molecular weights of the compounds were obtained by sub-
tracting the weight of a proton, nominally 1 m/z unit from the
Imaging mass spectrometry: A new technology for the
analysis of protein expression in mammalian tissues
MARKUS STOECKLI, PIERRE CHAURAND, DENNIS E. HALLAHAN & RICHARD M. CAPRIOLI
Mass Spectrometry Research Center, Vanderbilt University School of Medicine, Nashville, Tennessee, USA
Correspondence should be addressed to R.M.C.; email: r.caprioli@vanderbilt.edu
Fig. 1 Methodology developed for the spatial analysis of tissue by
MALDI mass spectrometry. Frozen sections are mounted on a metal plate,
coated with an UV-absorbing matrix and placed in the mass spectrome-
ter. A pulsed UV laser desorbs and ionizes analytes from the tissue and
their m/z values are determined using a time-of-flight analyzer. From a
raster over the tissue and measurement of the peak intensities over thou-
sands of spots, mass spectrometric images are generated at specific
molecular weight values.
© 2001 Nature Publishing Group http://medicine.nature.com
© 2001 Nature Publishing Group http://medicine.nature.com
494 NATURE MEDICINE • VOLUME 7 • NUMBER 4 • APRIL 2001
NEW TECHNOLOGY
measured m/z value. Identification of the proteins can be done
through extraction, HPLC fractionation, proteolysis, mass spec-
trometric sequencing of one or more of the fragments and pro-
tein database searching. This procedure is illustrated below for
proteins in tumor sections.
Molecular imaging of tumor sections
One our aims is the molecular analysis and imaging of peptides
and proteins in brain tumors, specifically in human glioblas-
toma. Such an analysis would be an important if not essential
part of strategies designed to locate specific proteins that are
more highly expressed in tumors and those greatly diminished
in expression, relative to normal tissue. Currently, brain tumors
account for 2% of all cancer deaths, or about 11,000 deaths an-
nually in the United States. Gliomas account for 50% of all pri-
mary brain tumors, with glioblastomas compromising half of
those
12
.
Here, tumor-bearing tissues were generated
by subcutaneous implantation of human
glioblastoma cells (D54) into the hind limb of
a nude mouse. After tumors grew to about 1
cm in diameter, we surgically removed them
from the mouse and immediately froze them
using liquid nitrogen. For image analysis, we
cut the tumor tissue using a microtome in 12-
µm thick sections orthogonal to the point of
attachment to normal tissue. Frozen sections
were processed following the protocol de-
scribed above before image analysis by MS.
The optical image of a frozen human
glioblastoma section taken immediately fol-
lowing mass spectrometric imaging is shown
in Fig. 3a. The orientation in the figure is such that the actively
growing area of the tumor is at the top of the figure, and the
point where the tumor was attached to the healthy tissue at the
bottom. The fine line (cross-hatched) pattern on the optical
image was produced by laser ablation of the surface during the
scan. Mass spectrometric images were produced from a raster
over an area of 8.5 mm × 8 mm (image spots 100 µm apart on
center). During the scan, we recorded images of ions in 45 mass
ranges and the mass spectra were saved for further analysis.
Three mass spectrometric images of molecules present in dis-
tinct areas of the tumor are shown in Fig. 3b–d. In this figure,
color is used to represent different ions, with color saturation a
function of the relative intensity (see color reference bar).
Overall, we detected over 150 different proteins, with many
being present in all parts of the tissue. Individual selected pro-
teins were identified as described below. We took three different
Fig. 2 Mass spectrometric images of a
mouse brain section. a, Optical image
of a frozen section mounted on a gold-
coated plate. b, m/z 8,258 in the re-
gions of the cerebral cortex and the
hippocampus. c, m/z 6,716 in the re-
gions of the substantia nigra and medial
geniculate nucleus d, m/z 2,564 in the
midbrain.
Fig. 3 Selected protein images from a glioblastoma
section. a, Human glioblastoma slice mounted on a
metal plate, coated with matrix (the lines are from
ablation of matrix with the laser). b–d, Mass spectro-
metric images of proteins showing high concentra-
tion in the proliferating area of the tumor (d) and
other proteins present specifically in the ischemic and
necrotic areas (b and c).
ab
c
d
a
b
c
d
© 2001 Nature Publishing Group http://medicine.nature.com
© 2001 Nature Publishing Group http://medicine.nature.com
NATURE MEDICINE • VOLUME 7 • NUMBER 4 • APRIL 2001 495
NEW TECHNOLOGY
mass spectra from different regions of the glioblastoma during
the scan (Fig. 4). These spectra clearly show differences in pro-
tein expression in different parts of the tumor.
The proliferating area of the tumor was of particular interest
with many proteins being expressed at higher levels relative to
normal tissue. For example, the protein of molecular weight
4,964 (Fig. 3d) is localized only in the outer area of the tumor.
Other proteins, such as that of molecular weight 41,662 (Fig.
3b), were localized in the necrotic area. In addition, other pro-
teins were localized in the ischemic area between the necrotic
center and proliferating periphery, as shown for the protein
with a molecular weight of 11,639. To identify the mapped pro-
teins, we made an extract of the appropriate portion of the
glioblastoma tissue, and then fractionated the proteins by
HPLC. The UV chromatogram of such an extract is shown in
Fig. 5. The on-line mass spectrometric analysis (Ion Trap,
Finnigan Company, San Jose, California) performed using elec-
trospray ionization MS easily permitted localization of the frac-
tion containing the proteins of interest. For example, one of the
proteins of molecular weight 4,964 eluted at 28.35 min in the
chromatogram. We spotted a sample of this fraction onto a
MALDI target plate and performed an on-target digestion by
trypsin. We analyzed the digest by MALDI MS followed by a
database search in SwissProt using the software ‘MoverZ’
(ProteoMetrics, New York, New York). Thymosin β.4 (Tβ.4) was
found to match the digest data precisely. The sequence analysis
of the amino-terminal peptide confirmed the identification of
the protein as Tβ.4 in this human glioblastoma xenograft.
Increased expression of Tβ.4 has been reported in a variety of
different tumors
13
. The localization of Tβ.4 in the proliferating
area of the tumor correlates with previous findings of higher
levels of Tβ.4 in embryonic/neoplastic tissue compared with
normal/benign tissue
14
. One of the known activities of this im-
munoregulatory peptide is its ability to sequester cytoplasmic
monomeric actin
15
. Moreover, actin filaments have been shown
to change into clump formation in apoptosis induced by anti-
tumor drugs, a process thought to be the result of decreased Tβ.4
concentrations
16
.
We also observed the increased expression of Tβ.4 in other tu-
mors as well. For example, in some mouse models of prostate
cancer, high levels of this protein have been found using imag-
ing MS. To confirm the identification, we generated a fragment
ion spectrum (MS/MS analysis) using an electrospray quadru-
pole TOF mass spectrometer (Q-Star, Applied Biosystems/SCIEX,
Foster City, California) from one of the tryptic fragments. The
MS/MS spectrum of the N-terminal tryptic peptide obtained
from a similar digest of Tβ.4 purified from a mouse with prostate
cancer is shown in Fig. 6. Fragment ions were matched by iden-
tifying portions of the y and b ion series
17
, covering the com-
plete sequence of the peptide. This spectrum confirmed the
presence of Tβ.4 in mouse models of prostate cancer.
Furthermore, from the MS/MS spectrum, the presence of an
acetyl group at the N-terminal end of the Tβ.4 peptide was con-
firmed. The protein of molecular weight 11,639 ± 2 (Fig. 3c) was
similarly identified as S100 calcium-binding protein A4
(S100A4), and the protein of measured molecular weight 41,659
± 4 to be cytoplasmic actin.
Discussion
The identification of specific tumor markers, for example Tβ.4,
in the proliferating area of the tumors demonstrates the poten-
tial of this technique to be used in intra-operative assessment of
the surgical margins of tumors. Currently, frozen sections and
light microscopy are required for rapid decisions, but are, at
times, inaccurate
18,19
. There is presently a need to develop tech-
nology to improve the accuracy of such decisions
20,21
. For exam-
ple, cancer invasion into muscle indicates that more extensive
surgery or adjuvant therapy is needed
22,23
and intra-operative di-
agnosis of central nerve system neoplasia is required for surgical
management
24–26
. Clinical validation will determine the useful-
ness of imaging MS to demonstrate these pathologic criteria
accurately for more aggressive management of cancer.
Beyond the application of imaging MS to brain cancer research,
we are currently using this technology to study prostate and colon
cancer development and progression. In both cases, numerous
tumor-specific markers have been identified and specifically local-
ized within the tumors. Protein profiling and imaging MS are also
proving to be of prime importance in our current research aiming
at a better understanding of prostate development. Overall, imag-
ing MS can be a valuable molecular tool in a wide variety of studies
and applications involving animal tissues.
Fig. 4 MALDI mass spectra taken at different locations within a glioblas-
toma slice (Fig. 3). Over 150 different peaks could be detected, with some
of them having a distinct spatial distribution in the tissue. Top, distal and
most active area of tumor proliferation; middle, an ischemic area; bottom,
a necrotic area of the tumor. The inset shows an expanded portion of the
spectrum in the region of thymosin β.4.
Fig. 5 UV chromatogram of a LC separation on a glioblastoma
xenograft extract. The analyte of molecular weight 4,964 was detected by
online electrospray mass spectrometry (inset shows mass spectrum) at a
retention time of 28.3 min.
© 2001 Nature Publishing Group http://medicine.nature.com
© 2001 Nature Publishing Group http://medicine.nature.com
496 NATURE MEDICINE • VOLUME 7 • NUMBER 4 • APRIL 2001
NEW TECHNOLOGY
Acknowledgments
We thank S. Schroeter, E. Sierra-Rivera, B. DaGue and Darell Bigner
for help with this study. This work was supported by NIH grants GM
58008 (to R.M.C.), CA 58506 (to D.H.) and C.A. 70937 (to D.H.).
1. Karas, M., Bachmann, D., Bahr, U. & Hillenkamp, F. Matrix-assisted ultravio-
let-laser desorption of nonvolatile compounds. Int. J. Mass Spectrom. Ion
Process. 78, 53–68 (1987).
2. Caprioli, R.M., Farmer, T.B. & Gile, J. Molecular imaging of biological sam-
ples: Localization of peptides and proteins using MALDI-TOF MS. Anal. Chem.
69, 4751–4760 (1997).
3. Valaskovic, G.A. & Morrison, G.H. Quantitative imaging ion microscopy — a
short review. Scanning Microscopy 6, 305–318 (1992).
4. Levisetti, R. et al. Imaging-SIMS (secondary-ion mass-spectroscopy) studies of
advanced materials. Scanning Microsc. 7, 1161–1172 (1993).
5. Todd, P.J., McMahon, J.M., Short, R.T. & McCandlish, C.A. Organic SIMS of
biological tissue. Anal. Chem. 69, 529A–535A (1997).
6. Pacholski, M.L. & Winograd, N. Imaging with mass spectrometry. Chem. Rev.
99, 2977–3005 (1999).
7. Li, K.W. et al. Direct peptide profiling by mass spectrometry of single identi-
fied neurons reveals complex neuropeptide-processing pattern. J. Biol. Chem.
269, 30288–30292 (1994).
8. Jimenez, C.R. et al. Direct mass spectrometric peptide profiling and sequenc-
ing of single neurons reveals differential peptide patterns in a small neuronal
network. Biochemistry 37, 2070–2076 (1998).
9. Moroz, L.L., Gillette, R. & Sweedler, J.V. Single-cell analyses of nitrergic neu-
rons in simple nervous systems. J. Exp. Biol. 202, 333–341 (1999).
10. Chaurand, P., Stoeckli, M. & Caprioli, R.M. Direct profiling of proteins in bio-
logical tissue sections by MALDI mass spectrometry. Anal. Chem. 71,
5263–5270 (1999).
11. Stoeckli, M., Farmer, T.B. & Caprioli, R.M. Automated mass spectrometry
imaging with a matrix-assisted laser desorption ionization time-of-flight in-
strument. J. Am. Soc. Mass Spectrom. 10, 67–71 (1999).
12. Nelson, D.F., McDonald, J.V., Lapham, L.W., Quazi, R. & Rubin, P. Central
nervous system tumors. in Clinical Oncology: A Multidisciplinary Approach for
Physicians and Students (ed. Rubin, P.) 617–644 (WB Saunders, Philadelphia,
1993).
13. Paciucoi, R. et al. Isolation of plasminogen activator, cathepsin H, and non-
specific cross-reacting antigen from SK-PC-1 pancreas cancer cells using sub-
tractive hybridization. FEBS Lett. 385, 72–76 (1996).
14. Hall, A.K. Differential expression of thymosin genes in human tumors and in
the developing human kidney. Int. J. Cancer 48, 672–677 (1991).
15. Sun, H.Q., Kwiatowska, K., Yin, H.L. Actin monomer binding proteins. Curr.
Opin. Cell Biol. 7, 102–110 (1995).
16. Iguchi, K. et al. Decreased thymosin β.4 in apoptosis induced by a variety of
antitumor drugs. Biochem. Pharmacol. 57, 1105–1111 (1999).
17. Roepstorff, P. Proposal for a common nomenclature for sequence ions in
mass spectra of peptides. Biomed. Mass Spectrom. 11, 601 (1984).
18. Turner, R.R., Hansen, N.M., Stern, S.L. & Giuliano, A.E. Intra-operative exam-
ination of the sentinel lymph node for breast carcinoma staging. Am. J. Clin.
Pathol. 112, 627–634 (1999).
19. Gibbs, J.F., Huang, P.P., Zhang, P.J., Kraybill, W.G. & Cheney, R. Accuracy of
pathologic techniques for the diagnosis of metastatic melanoma in sentinel
lymph nodes. Ann. Surg. Oncol. 6, 699–704 (1999).
20. Richter, T., Nahrig, J., Komminoth, P., Kowolik, J. & Werner, M. Protocol for
ultrarapid immunostaining of frozen sections. J. Clin. Pathol. 52, 461–463
(1999).
21. Wang, H.W., Willis, J., Canto, M.I.F., Sivak, M.V. & Izatt, J.A. Quantitative
laser scanning confocal autofluorescence microscopy of normal, premalig-
nant, and malignant colonic tissues. IEEE Trans. Biomed. Eng. 46, 1246–1252
(1999).
22. Cheng, L., Neumann, R.M., Weaver, A.L., Spotts, B.E. & Bostwick, D.G.
Predicting cancer progression in patients with stage T1 bladder carcinoma.
J. Clin. Oncol. 17, 3182–3187 (1999).
23. Longacre, T.A. & Hendrickson, M.R. Diffusely infiltrative endometrial adeno-
carcinoma — an adenoma malignum pattern of myoinvasion. Am. J. Surg.
Pathol. 23, 69–78 (1999).
24. Shah, A.B., Muzumdar, G.A., Chitale, A.R. & Bhagwati, S.N. Squash prepara-
tion and frozen section in intra-operative diagnosis of central nervous system
tumors. Acta Cytologica 42, 1149–1154 (1998).
25. Silbergeld, D.L. & Chicoine, M.R. Isolation and characterization of human
malignant glioma cells from histologically normal brain. J. Neurosurg. 86,
525–531 (1997).
26. Firlik, K.S., Martinez, A.J. & Lunsford, L.D. Use of cytological preparations for
the intra-operative diagnosis of stereotactically obtained brain biopsies: A 19-
year experience and survey of neuropathologists. J. Neurosurg. 91, 454–458
(1999).
Fig. 6 The mass spectrometric analysis by electrospray MS/MS of the N-
terminal tryptic fragment of Tβ.4. The complete sequence of the frag-
ment was confirmed from the mass spectrometric data.
Methods
Tissue preparation. 12-µm sections were cut from a frozen
mouse brain on a Leica CM 3000 cryostat at –15 °C and di-
rectly picked up onto a gold-coated stainless steel plate. The
sections were immediately transferred to a cold room (4 °C),
where 10 µl of matrix (sinapinic acid, 10 mg/ml in acetoni-
trile/0.05% trifluoroacetic acid 50:50) were deposited with a
pipette in a line adjacent to the tissue and mechanically
spread over the tissue using a small plastic spatula. After air-
drying for 45 min, the sections were dried for 2 h in a desic-
cator before mass spectrometric analysis. This application
technique results in formation of crystals of the organic ma-
trix on the surface of the tissue while minimizing the spread-
ing of sample molecules.
Glioblastoma extraction and protein fractionation by
HPLC. A portion of the glioblastoma (82 mg) was immersed
in 500 µl extraction buffer (0.25 M sucrose, 0.01 M Tris-HCl
and inhibitor mix; (Roche Molecular Biochemicals,
Switzerland), homogenized using a Duall homogenizer and
centrifuged 3 times (10 min at 680g, 10 min at 10,000g and
1 h at 55,000g), each time transferring the soluble fraction
to a new tube. The final fraction (50 µl) was separated over a
C4 microbore column (Vydac, Hesperia, California), samples
were collected and the separation run was recorded with a
UV detector set at 214 nm. Solvent A was 0.1 trifluoro acetic
acid and solvent B was 95% acetonitrile, 5% water and 0.1%
trifluoro acetic acid. A flow rate of 200 µl/min was used with
a gradient of 5 min at 5% B, then in 55 min to 60% B, then
in 10 min to 100% B, and finally 5 min at 100% B.
On-target digestion by trypsin. For this procedure, the
sample (2 µl) was placed on the target and allowed to dry
before adding digest solution (2 µl, 20 nM bovine trypsin,
sequencing-grade, (Roche Molecular Biochemicals), and 50
mM ammonium hydrogen carbonate). The plate was kept at
37 °C for 30 min while adding water to maintain the vol-
ume. After drying the sample, 2 µl of a saturated α-cyano-4-
hydroycinnamic acid (Sigma) solution in 50:50 acetonitrile
and 0.1% trifluoro acetic acid (2 µl) was added as a MALDI
matrix.
© 2001 Nature Publishing Group http://medicine.nature.com
© 2001 Nature Publishing Group http://medicine.nature.com
