SAMPLE PREPARATION

The quartz plate with the sample under investigation on its surface was washed with water to remove excess buffer salts and then placed in a recess of a homemade MALDI plate (Fig. 6). The surface was covered with matrix using a particular thin-layer preparation based on an air-brush method enabling a homogenous application of the matrix. The matrix layer was dried under a gentle stream of air. Sinapinic acid was found to be suitable for ConA determi-nation. CHCA could be applied for ConA as well as for the mannoside-bound phospholipid.

DISCUSSION

Both, the distribution and homogeneity of the ConA layer and the sugar-bound phospholipid were determined using the MALDI imaging technique. Therewith, the coating of the quartz sensor was analysed and the specific interaction of ConA and the mannosylated liposomes, as observed in the QCM experiment, could be confirmed. Direct mass spectrometric analysis of immobilised sub-stances is a new tool in surface analysis. Additionally, spe-cific interactions on quartz crystal biosensors can be proven. This kind of “molecular mass scanner” is not lim-ited to QCM crystals and can be adapted for the characteri-sation of other ionisable compounds on the surface of a range of materials.

Martin Luther UniversityHalle-Wittenberg

EXPERIMENTAL AND RESULTS

Concanavalin A, a lectin from Canavalia ensiformis, was immobilised on the gold surface of the quartz sensor. This lectin is able to bind á-D-mannopyranoside and á-D-glucopyranoside. P-Amino-phenyl-á-D-mannopyrano-side was bound to N-glutaryl-phosphatidylethanolamine after the phospholipid was incorporated in soy phospha-tidyl-choline (SPC) liposomes of about 100 nm in diame-ter. In QCM experiments liposomes specifically bound to the ConA layer [2].

Experiments were performed using a commercial QCM system LiquiLab 21 (ifak e.V., Barleben, Germany) equipped with a Reglo analog MS 2/12 pump (Ismatec, Wertheim-Mondfeld, Germany). The flow rate was 0.55 mL/min.The quartz sensor was rinsed with a Tris buffer until a con-stant frequency was reached. ConA solution (1 mg/mL) was pumped through a sample loop and immobilised onto the sensor by means of hydrophobic interactions. Then, a 5 mM liposome suspension was applied to the ConA layer allowing the specific interaction between the manno-pyranoside and the lectin.

QUARTZ CRYSTAL MICROBALANCE

Figure 1: Model of the mannosylated liposomes binding on a ConA layer immobilised on the sensor electrode

Figure 3: MALDI image mesh plot of ConA immobilised on a quartz electrode.

Figure 6:Homemade MALDI plate with a 10 MHz-QCM sensor

DATA ACQUISITION AND PROCESSING

Data acquisition was carried out using the Sequence Control Panel Software (Applied Biosystems). A post-macro was employed for noise removal and smoothing of each spectrum during the measurements.The 2500 resulting mass spectra were analysed by an homemade C++ program (available from the correspond-ing author), which performs a peak detection in a certain mass range in each MS file and creates an ASCII summary file containing the x and y positions as well as the respec-tive signal intensities. Thus, the image data can be easily imported into scientific graphing software and visualised by mesh or contour plots (Fig. 3 and 5).

Figure 5: The contour plot shows the distribution of the mannoside-bound phospholipid immobilised on the ConA layer (the circle indicates the effective measuring electrode).

Institute of Pharmaceutics and Biopharmaceutics, Wolfgang-Langenbeck-Str. 4, 06120 Halle (Saale), Germany

Corresponding author: schmelzer@pharmazie.uni-halle.de

C.E.H. Schmelzer, A. Huenerbein, K. Raith, and R.H.H. Neubert

REFERENCES

[1] A. Janshoff, C. Steinem, Sensors Update 9 (2001) 313-354

[2] A. Hildebrand, A. Schaedlich, U. Rothe, R. Neubert, J Colloid Interface Sci 249 (2002) 274

INTRODUCTION

Quartz crystal microbalance (QCM) is a technique used to characterise interactions such as enzyme/substrate, anti-gen/antibody, or carbohydrate/lectin interactions by mea-suring changes in resonance frequency and impedance [1]. For this purpose, the target or the selective structure is immobilised on the surface of a quartz sensor. Binding of the specific interaction partner causes a decrease in the res-onance frequency. The detection limit is in the range of a

2few ng per cm for a 10 MHz quartz. A profound knowl-edge of the molecular structure of the surface layers is required for the reliable application of QCM and the eval-uation and quantification of the results. Typical methods used in surface analysis are atomic force microscopy to characterise surface morphology and ellipsometry to determine the thickness of a layer. Fluorescence micros-copy only gives an estimate of the substance distribution on the surface provided that the compounds show fluores-cence. This has prompted us to develop a MALDI imaging technique for the two-dimensional mass spectrometric detection of the interaction partners in order to determine their distribution on the sensor surface. A specific interac-tion of pharmaceutical interest, namely the linkage between concanavalin A (ConA) and sugar-bound liposomes (Fig. 1), is presented to demonstrate the appli-cability of the method.

MATERIALS

• Concanavalin A from jack bean (Canavalia ensiformis) (Sigma-Aldrich, Taufkirchen, Germany)

• (N-Glutaryl ) -1 ,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine (Avanti Polar Lipids, Alabaster, AL, USA)

• p-Aminophenyl-á-D-mannopyranoside (Sigma-Aldrich)

• Sinapinic acid (Fluka, Buchs, Switzerland)• á-Cyano-4-hydroxycinnamic acid (CHCA) (Sigma-

Aldrich)• Tris buffer (20mM Tris, 1mM CaCl , 0.5mM MnCl 2 2

0.5M NaCl)

MALDI IMAGING

MALDI imaging experiments were carried out using a delayed extraction time-of-flight (TOF) mass spectrome-ter Voyager DE-Pro (Applied Biosystems, Foster City, CA, USA) equipped with a pulsed nitrogen laser (ë=337 nm, 3 ns pulse width, 20 Hz repetition rate).Each biosensor surface was characterised by 50 x 50

2points of a square area of 1111 points per cm . Characterisation of the ConA was performed in the linear

2+mode by measuring the [M+2H] ions whereas the mannoside-bound phospholipid was determined in the

-reflector mode. By isolating the respective [M-H] ions with the timed ion selector, an increase in S/N ratio was achieved.

Figure 2: Demonstration: Scanning MALDI image (left) and photograph (right) of the letter “A”, prepared by pipetting a concanavalin A (M »26 kDa) r

solution onto a quartz sensor (14 mm i.d.).

Figure 4: Single MALDI-RefTOF mass spectrum (position marked in Fig. 5 with ) and chemical structure of the mannoside-bound phospholipid.

Acknowledgments

We like to thank D. Reese and M. Utzig from the Technical Service Facility for the manufacture of the tailored MALDI plate. We gratefully acknowledge financial sup-port from the BMBF and the Kultusministerium Sachsen-Anhalt.

MALDI Imaging of Proteins and their Interaction Partners on Surfaces