live single-cell mass spectrometry

ANALYTICAL SCIENCES AUGUST 2009, VOL. 25 953

2009 © The Japan Society for Analytical Chemistry

E-mail: [email protected]

Live Single-cell Mass Spectrometry

Tsutomu MASUJIMA

Graduate School of Biomedical Sciences, Hiroshima University, 1-2-3 Kasumi, Minami-ward, Hiroshima 734–8551, Japan

The history from bio-imaging to live single-cell mass spectrometry (MS) is herein reviewed. The limitation of the current bio-imaging method is probing only known molecules, and a method for finding new molecules is needed for cells which, however, show individual behaviors even in the same incubation dish. Single-cell MALDI-TOF/MS has been developed, but it can detect only molecules that can be easily ionized, and not be exhaustive. Recently, the contents of a single cell have been sucked out by a nano-electro spray tip, and directly introduced into MS by nano-spray ionization. Thousands of molecular peaks have been successfully and exhaustively detected, and an extraction method for key molecules was also developed. This new method is now being widely applied to explore site- or state-specific molecules in various aspects of cell dynamisms.

(Received April 21, 2009; Accepted May 27, 2009; Published August 10, 2009)

1 Introduction

When we are watching a cell, it shows us various unexpected and interesting behaviors.1–3 We would thus like to know the reasons and mechanism of those fascinating phenomena of life. This is a real moment of the scientific study of living organisms. In such visualizations of cell behaviors, we frequently found that the behaviors of cells are not identical nor synchronized even in the same incubating dish.4 In allergy response, one rat mast cell pops its micro granules out, exocytosis, in rapid way on stimulation by a calcium ionophore, however, another cell pops very slowly, as shown in Fig. 1. We then came to doubt the results using many cells at once. In our current biochemical analyses, we usually prepare a sample by correcting many cells that are in a certain condition in order to make sufficient amounts of molecules to be detected at the current methodological sensitivity. The results should be considered as being the “averaged” one, and the conclusion is just for in total. However, under consideration of the before-mentioned cell’s individual behaviors, it should not be sufficient, and sometimes be incorrect in concluding about the mechanisms of life, and thus we decided to innovate a very sensitive method to analyze even a single cell with, ideally, direct observations of individual behaviors of cells and with direct detection of the molecules in charge.

2 Ideal Method for Cellular Analysis

The most ideal method for cell biology, we thought, is that can directly analyze the molecular mechanism of a live single cell by cell visualization, and then various molecular aspects of life can be disclosed in a quite rapid and direct way. Video imaging and bio-imaging3–7 of cells or single molecular imaging8–11 have been the leading methods for visualizing those molecular mechanisms. However, these imaging methods have a methodological limitation in that only known molecules can be probed or stained so as to be visualized. Since it is hard, of course, to visualize unknown or undiscovered molecules, we should find a new method to detect and identify many molecules including new ones, preferably, in a single cell.

In 1999, I proposed one method in the next millennium symposium, named a “video-mass spectroscope”,3 in which the cell behavior is monitored by a video-microscope and the cell contents can be sucked out by a glass capillary when the cell shows an interesting behavior, and the contents are fed to the mass spectrometer by an electro-spray ionization (ESI) method, as shown in Fig. 2. However, this idea using ESI was too poor in sensitivity and single-cell molecular detection has never succeeded for 7 years since then.

We used an resin-packed nano-spray tip to preconcentrate released histamine and serotonin from an allergy model cell line, Rat Basophilic Leukemia (RBL-2H3) cells upon stimulation.12 The releasing time course of these chemical

1 Introduction 9532 Ideal Method for Cellular Analysis 9533 Single Cell Analyses by MALDI-TOF/MS 9544 Video Mass Spectroscope 9555 Additional Method for Exploring the

Molecules in Charge 957

6 Application to Localization Analysis and Metabolomics of a Live Single Cell 957

7 Application to Classification or Grouping Analysis of Cells 957

8 Acknowledgements 9599 References 959

Reviews

954 ANALYTICAL SCIENCES AUGUST 2009, VOL. 25

mediators has been successfully detected, although analysis is needed of at least 10000 cells for one measurement.

Among many trials in vain, we had been coming to recognize that electro spray ionization (ESI),13 which nebulizes the high-voltage charged liquids of samples with nitrogen sheath gas, is different from nano-electro spray ionization (nano-ESI)14–17 which generates very fine mist of a charged sample liquid induced by only a drop of applied high electro-voltage, and the ionization efficiency of nano-spray ionization seems to be much higher than that of ESI.

3 Single Cell Analyses by MALDI-TOF/MS

The MALDI-TOF/MS (matrix assisted laser desorption ionization–time of flight/mass spectrometry) method has been another challenge for single-cell analysis, and many trials have been made worldwide.18–22 After our challenges using this method, we, however, noticed the following problems in applications: 1) Cells were killed by the addition of a matrix solution, and it is hard to trace a tiny target single cell on the evaporation and crystallization process of the matrix.21 It is also hard to irradiate a laser beam to the point a target single cell in

Fig. 1 Time course of individual granule popping on calcium ionophore stimulation to rat mast cells. These cells are in the same video-microscopic view under observations, and number of poppings are analyzed by an image-subtraction method.

Fig. 2 Proposed concept and real images of the “Video-Mass Spectroscope” under development in 1999 shown in the millennium memorial symposium issue of Ref. 3. The molecules in and outside a single cell are sucked into a micro-pipette to introduce them to the electro-spray interface of the mass spectrometer (sited by courtesy of Elsevier, Amsterdam).


the congestion of micro-crystals of the matrix. These difficulties made this method tedious and not in “live” nor real time. 2) The number of peaks detected was very small even when using many cells, and not sensitive enough for a single cell at the usual size of about 10 μm. It seemed that only these molecules easy to be ionized were detected. Thus, this method was not exhaustive in molecular detection for so many components in a cell. This, we think, is also one limitation of the current MALDI/MS imaging of biological samples.23,24 3) Molecular identification was hard. Nowadays, we have atmospheric MALDI attachment25 and MS/MS analysis can be performed by MALDI-ion trap26 or -TOF/TOF27,28 attachments; however, we should know that it is still hard to identify the molecules even by MS/MS analysis when we have no related known MS/MS spectrum data as a reference, and sensitive MS/MS analysis at much higher resolution is needed to identify the “new molecules”.

4 Video Mass Spectroscope

After these challenges in vain for the video mass-spectroscopic method, we finally got an idea to use nano-spray ionization which seemed to be the most sensitive method with exhaustive molecular detection ability. Also, the nano-spray tip can be simultaneously used to insert it into a live cell and to suck out its contents under the “video-microscopic observation” because the tip is essentially a micro capillary.

The volume of a single cell at a 10-μm diameter is less than

1 pL, which is too tiny amount to be detected by our current detection limit of Q-TOF-type mass spectrometers. Single-cell content, such as cytosol or even an organelle, can be trapped only at the top point of a micro-capillary. We adjusted the nano-spray tip with the top bore to be 1 – 5 μm to insert and suck out the content of a live cell from its cytosol or organelle, and then set the same tip on a nano-ESI attachment of a mass spectrometer to feed the contents into the spectrometer by nano-electro spray ionization, as shown in Fig. 3.29 The nano-spray tip was improved for single-cell analysis by a cooperative company, Humanix, Co. Ltd., Japan, using our original know-how for making a steady and long lasting nano-spray. This tip can stably spray 1 μL of a standard solution for a maximum of 2 h.

However, when the contents of a live single cell were captured at the top of the nano-spray tip, they could not be directly sprayed to the mass spectrometer. A high viscosity of the cell contents prevented spraying. We thus added 1 μL of an ionization solvent into the tip, which is an organic solvent with a volatile acid or base, and finally found “thousand of peaks” in the mass spectrum for “a live single cell,” as seen in Fig. 4.29,30 The molecular contents of a cell, especially small molecules, can be extracted by a nl/min level stream of an organic solvent through the contents trapped by the tip, and sprayed out the molecular contents to the mass spectrometer. For the first 5 min or more, we could detect molecular peaks and then the peaks finally became only that of the solvent.

Fig. 3 Scheme of the new “Video-Mass Spectrometry”; now we call it as “Live Single-cell Mass Spectrometry (LiveSC/MS)”. The cell is observed by a video-microscope, at the moment of sampling, a nano-spray tip is inserted into a target site of a live single cell and the contents are sucked into the tip. After the addition of an ionization solvent into the tip, the tip is set to the nano-spray ionization source to feed the molecular contents into the mass spectrometer to obtain MS spectrum.


Fig. 4 One example of the MS spectrum for a live single granule in a cell. The insert is a picture on the moment of capturing the contents of a single granule by a nano-spray tip (triangular shadow). Several small but important peaks are magnified in the top.

Fig. 5 Tables of 66 granule-specific peaks with the m/z and the t-value (discriminated by the t-value more than 95%) and 8 cytosol-specific peaks (discriminated by the t-value less than –80%) out of a total of about 700 cell component peaks. The pictures show the trapping process and the sites where the content of a granule and cytosol have been trapped.


5 Additional Method for Exploring the Molecules in Charge

Among thousands of detected MS peaks, we can find molecular peaks that are related to a specific stage or function of a cell. Peak comparison and subtraction between two sets of obtained MS peaks were performed to extract the peaks that show a change in intensity between two corresponding stages of a cell.30,31 For validation, triplicate or more samples were measured. After this tedious manual data-processing trial, we adapted for this data a commercial analysis software, MarkerView (Applied Bio-systems, USA), which was originally developed for cancer marker finding or tracing of drug metabolites. The obtained MS spectra were subjected to a statistic analysis, t-test, for finding stage or site-specific molecular peaks, and multivariate analysis, such as the principal component analysis (PCA), are applied to check the raw data with contaminant and for assigning or grouping various related sets of data.

6 Application to Localization Analysis and Metabolomics of a Live Single Cell

Figure 5 shows the capturing moment of a live single granule and cytosol in a RBL-2H3 cell, and the result of a paired t-test between detected peaks of a single granule and those of cytosol showed 66 granule-specific peaks (t-value more than 95%) and 8 cytosol-specific peaks (t-value less than –80%).31 This t-value is useful as an index of data dependency to a group. In this case, 100% means that the peak can be found only in a granule, and –100% means that only in cytosol. The top columns for peak m/z 112 and 177 after a t-test analysis are shown more schematically in Fig. 6 which shows that only the peaks in 5 samples from single granules have m/z 112 peak intensities, while m/z 177 peaks have similar intensities. The peaks were subjected to MS/MS analysis, and each MS/MS spectrum resembles the histamine and serotonin standard to prove that only histamine is specifically found in granules in RBL-2H3 cells.31 As shown in Fig. 7, tryptophan was mainly found in

cytosol and its metabolite, serotoin, was found in both cytosol and granules.31 These results were totally understood by a metabolic pathway and transport of these amino acids. Histamine, a chemical mediator of allergy, is biosynthesized from histidine in a granule, but serotonin is synthesized from tryptophan in cytosol and then transported to a granule to have equal distribution in cytosol and granules.31

The senior editor of the Nature, Dr. Tanguy Chouard, gave our method a name, “Live Single-cell Mass Spectrometry (LiveSC/MS)” in his advising e-mail, and we have frequently used this name or its combination for our video-mass spectroscopic method since then.

With the same strategy and data analyses, the localization of many molecules, the live single-cell metabolomics,32 and even the live single-organelle metabolomics have been performed by our LiveSC/MS method. We are now using stable isotope- labeled compounds to trace the pathway of metabolism and the transport of various molecules at the single cellular level.

7 Application to Classification or Grouping Analysis of Cells

Principal-component analysis is the helpful tool for the discrimination or classification of data sets by finding the direction of maximum variance. Figure 8 shows MS spectra of 7 phenotypes of cells.30,31 Obtained spectra are apparently different from each other. In order to check the possibility to classify these cells by their spectral character, PCA was applied to these data sets. Figure 9 shows the results of a PCA analysis by a score plot to classify the data points into 7 groups of phenotypes of cells.31 The loading plot shows the peaks as dots and the peak points are scattered in two dimensions by two principal-component axes. The central zone is for common peaks, while the peripheral peaks are specific to each phenotype. One of the peaks at m/z 130.04 was identified as pyroglutamic acid as one of the characteristic molecules mainly found in the CAKI1 cell and the TIG-3 cell specifically.31 In combination of only two or three specific molecules found by this method, these phenotypes will be classified in the near future.

We usually apply PCA first to check raw data to be exclusive

Fig. 6 Visualized t-test result for the peak m/z 112 and 177. No intensity for m/z 112 peak was observed for 5 samples of the cytoplasm, but the peak intensity was observed for 5 samples of granules (upper). However, similar intensities were observed for m/z 177 peak. The peak was subjected to MS/MS analysis to resemble the standard MS/MS spectrum of histamine and serotonin (lower).


Fig. 7 Localization and identification of histamine and serotonin related molecules by t-test and MS/MS analyses. Metabolic pathways of serotonin and histamine were clarified by specific distributions of molecules in a single cell.

Fig. 8 Pictures of randomly selected seven phenotypes of live single cells and those spectra of cytosol at interphase (Swiss 3T3, CAKI1, C007, TIG-3, HepG2, ACHN, P19).


or not by contaminants, such as cell medium, and then apply a t-test when we would like to find a site or state specific peaks, or apply PCA to to perform grouping or classification. We are totally performing LiveSC/MS with such a systematic procedure. Finding cell cycle-specific molecules, cancer-specific molecules, and differentiation control factors of iPS or stem cells is under investigation.

As much as we have applied LiveSC/MS to various live cells, this method has been quite useful to find state or site-specific molecules, especially small molecules, and it is rapid, simple and direct. We hope this new method will greatly speed up various studies of life sciences from now.

8 Acknowledgements

Our research to innovate LiveSC/MS has been conducted with many colleagues during the 20-year history of my laboratory in Hiroshima University. I would like to especially thank Dr. H. Mizuno and Dr. N. Tsuyama for developing LiveSC/MS and Prof. E. M. Eyring (Univ. of Utah), Mrs. K. Karasawa (Applied Biosystems, Japan) and Mrs. M. Kanai (Thermo Fisher Sci., Japan) for their kind advice and support for this study. Launching of the video-imaging research lab. with Dr. E. Suzaki and Dr. K. Ozawa made a start of this study and Dr. N. Ojima challenged single-cell MALDI-TOF/MS. I sincerely thank all staff members and students in this history of our research for the ideal of

bio-analyses. These works were supported by the Grants-in- Aid from the Ministry of Education, Science, Sports and Culture, and the Ministry of Economy, Trade and Industry, Japan.

9 References

1. B. Alberts, A. Johnson, P. Walter, and J. Lewis (ed.), “Molecular Biology of the Cell”, 5th ed., 2008, Garland Science, London.

2. YouTube Cell Press Channel at http://www.youtube.com/cellpressvideo.

3. T. Masujima, Anal. Chim. Acta, 1999, 400, 33. 4. T. Masujima, N. Tsuyama, and T. Hasegawa, Nanomedicine,

2006, 1, 331. 5. F. Brandizzi, S. L. Irons, J. Johansen, A. Kolzer, and U.

Neumann, J. Microscopy-Oxford, 2004, 214, 138. 6. R. Srinivasan, S. Q. Yao, and D. S. Y. Yeo, Comb. Chem.

High Throughput Screening, 2004, 7, 597. 7. M. Funovics, R. Weissied, and C. H. Tung, Anal. Bioanal.

Chem., 2003, 377, 956. 8. T. Funatsu, Y. Harada, M. Tokunaga, K. Saito, and T.

Yanagida, Nature, 1995, 374, 555. 9. S. Nie and R. N. Zare, Annu. Rev. Biophys. Biomol. Struct.,

1997, 26, 567.10. A. Ishijima, H. Kojima, T. Funatsu, M. Tokunaga, H.

Tanaka, and T. Yanagida, Cell, 1998, 92, 161.

-100 -50 0 50 100PC1 Score

-50

0

50

100

150

Swiss

HepG2HepG2

C007ACHN P19

CAKI1

TIG-3

PC

2S

core

Scores for PC1 versus PC2

-0.10 0.00 0.10 0.20

PC1 Loading

-0.10

0.00

0.10

120.08227.12

62.9384.94

192.15

64.93

284.27174.13

256.26320.22

304.22

341.27

118.08203.06

131.00 313.2456.96122.03

79.03

61.01117.00

228.17282.14 271.50215.13

77.02 150.09

272.16285.23

564.39

267.21739.56

60.04 105.96429.19

133.96221.09

202.18409.17

105.0155.03

359.29

381.29131.98

382.27

161.01149.02

217.10

353.26

258.22

740.57

130.04

137.02338.30

299.10

271.18

301.11175.07

209.06104.00

560.06380.70123.05

PC2

Load

ing

Loadings for PC1 versus PC2

130.04

NH

OO

OH

Pyroglutamic acid

m/z 130.04

3 1 3 3 2 2 1 3 2 3 10

500

1000

21 212 1 1

ACHN C007 CAKI1 Hep G2 P19 Swiss TIG-3

32 3

Peak

Inte

nsity

Fig. 9 Single-cell grouping of seven phenotypes of cells (n = 3) by a score plot (upper left) of PCA analysis applied to the spectra in Fig. 8. A loading plot (upper right) shows the peak (shown in dot) dispersion along the principal-component axis and the cell type dependent distribution of the peak m/z 130.4 which was identified as pyroglutamic acid by MS/MS analysis, is shown in the bottom figure.


11. S. Allen, S. M. Rigby-Singleton, H. Harris, M. C. Davies, and P. O’Shea, Biochem. Soc. Trans., 2003, 31, 1052.

12. A. Sirikatitham, T. Yamamoto, M. Shimizu, T. Hasegawa, N. Tsuyama, and T. Masujima, Rapid. Commun. Mass Spectrom., 2007, 21, 385.

13. M. Yamashita and J. B. Fenn, J. Phys. Chem., 1984, 88, 4451.

14. J. Zeleny, Phys. Rev., 1917, 10, 1.15. G. I. Taylor, Proc. Royal Soc. London A, 1964, 280, 383.16. P. Kebarle and L. Tang, Anal. Chem., 1993, 65, 972A.17. M. S. Wilm and M. Mann, Int. J. Mass Spectrom. Ion.

Proc., 1994, 136, 167.18. P. A. van Veelen, C. R. Jiminez, K. W. Li, W. C. Wildering,

P. M. Geraerts, U. R. Tjaden, and J. van der Greef, Org. Mass Spectrom., 1993, 28, 1542.

19. C. R. Jimenez, P. A. van Veelen, K. W. Li, W. C. Wildering, W. P. Geraerts, U. R. Tjaden, and J. van der Greef, J. Neurochem., 1994, 62, 404.

20. E. S. Yeung, Anal. Chem., 1999, 71, 522A.21. M. Shimizu, F. Levi-Schaffer, N. Ojima, T. Shinagaki, and

T. Masujima., Anal. Sci., 2002, 18, 107.22. S. S. Rubakhin and J. V. Sweedler, Nat. Protoc., 2007, 2,

1987.

23. R. M. Caprioli, T. B. Farmer, and J. Gile, Anal. Chem., 1997, 69, 4751.

24. R. W. Garden and J. V. Sweedler, Anal. Chem., 2000, 72, 30.

25. V. V. Laiko, M. A. Baldwin, and A. L. Burlingame, Anal. Chem., 2000, 72, 652.

26. N. Ojima, K. Masuda, K. Tanaka, and O. Nishimura, J. Mass Specrom., 2005, 40, 380.

27. K. F. Medzihradszky, J. M. Campbell, M. A. Baldwin, A. M. Falick, P. Juhasz, M. L. Vestal, and A. L. Burlingame, Anal. Chem., 2000, 72, 552.

28. D. Suckau, A. Resemann, M. Schuerenberg, P. Hufnagel, J. Franzenm, and A. Holle, Anal. Chem., 2003, 376, 952.

29. T. Masujima et al., Proceedings of the 56th Annual Meeting of the Japan Society for Analytical Chemistry, Tokushima, Japan, Sep. 21, 2007 (C3008 – C3011).

30. N. Tsuyama, H. Mizuno, E. Tokunaga, and T. Masujima, Anal. Sci., 2008, 24, 559.

31. H. Mizuno, N. Tsuyama, T. Harada, and T. Masujima, J. Mass Spectrom., 2008, 43, 1692.

32. H. Mizuno, N. Tsuyama, T. Harada, and T. Masujima, Anal. Sci., 2008, 24, 1525.

live single-cell mass spectrometry

Documents