Mapping the cellular distribution of labelled molecules by SIMS microscopy

Biol Cell (1992) 74, 81-88 © Elsevier, Paris

81

Original article

Mapping the cellular distribution of labelled molecules by SIMS microscopy Elif Hindie ~, Bernard Coulomb 2, Ren6 Beaupain 3,, Pierre Galle ~Laboratoire de Biophysique, SC27 de l'Inserm; 2 Unit~ 312 de l'lnserm, Facult6 de M~decine, 8, rue du G~ndrai Sarrail, 94000 Crdteil; ZLaboratoire d'Immunopharmacologie exp~rimentale, C N R S UPR 405 Paris, France (Received and accepted 25 November 1991)

Summary - We took advantage of one of the main possibilities of ion microscopy, ie isotopic analysis, to study the cellular distribu-

tion of molecules labelled either with carbon 14 or with stable isotopes of low natural abundance such as nitrogen 15 and deuterium. The surface of the sample is bombarded with an ion beam (Ce2, Cs ÷ etc). Secondary ions emitted from the sample are filtered by a mass spectrometer and the distribution of the labelling isotope is recorded. In this way, we obtained images showing the characteristic distribution of ~4C-thymidine and D-arginine in human fibroblasts, and of 15N-adenine in organotypic cultures of human breast cancer cells. The spatial resolution on the acquired images was close to 0.1/zm when using the UPS-ONERA ion microprobe. The sensitivity of the method for detecting carbon 14 is far greater than that of autoradiography and the technique is both fast and quantitative. On the other hand, the capacity of ion microscopy for studying the tissular distribution of molecules labelled with stable isotopes, opens the way for biological and pharmacological tracer studies of human diseases. SIMS microscopy / carbon 14 detection I stable isotopes detection / cellular distribution

Introduction

Determining the cellular distribution of molecules is an essential step towards understanding both biological and pharmaco-toxicological phenomena. Ion microscopy [5, 8, 14], a method capable of revealing both stable and radioactive isotopes is used here to m a p the distribution o f labelled molecules in biological tissues. The surface of the sample is b o m b a r d e d with an ion beam with an energy of a few KeV. Each primary ion (eg Cs ÷, O2 ÷, Ga ÷) impinging upon the surface triggers a cascade of atomic collisions which results in the ejection of one or several atoms from the sample. During the b o m b a r d m e n t there is a continuous sputtering o f the superficial atomic layers and the sample is progressively eroded. A t o m s ejected from the target can gain or loose an electron, giving rise to 'secondary ions' characteristic of the composition of the sample. These ions are directed towards a mass spectrometer which separates them according to their mass/charge ratio. The distribution o f the selected ions is recorded either directly or using a scanning procedure. The image obtained is a map showing the distribution of a specific isotope at the sample surface. To be detected by ion microscopy a molecule in a biological specimen must contain a specific a t o m not present in the surrounding medium; except for some molecules such as drugs containing a halogen atom, eg 5-fluorouracil, bromodeoxyuridine etc, isotopic labelling is therefore necessary. During primary ion b o m b a r d m e n t the molecules present in the tissue sample are broken up. The

* Present address: IOCMH H6pital Avicennes, Bobigny, France

labelling atoms are then liberated and can be identified according to their characteristic mass. The choice of the isotope for labelling is crucial. Radioactive isotopes are advantageous in that their natural concentration in the tissues is extremely low. Carbon 14 (half-life 5730 years) is preferred to tritium (half-life 12.3 years) since for a given molecular concentration, the radioactivity in the medium will be lower with carbon 14, thus reducing radiobiological effects. In addition, labelling with carbon 14 does not modify the pharmacokinetics of the studied molecule. Labelling should be done with stable isotopes when studying the tissular distribution of a molecule administered to human, although the specificity in detection can be limited by the natural presence o f these isotopes in tissues (table I). Using carbon 14 and two stable isotopes - deuterium and nitrogen 15 - we studied the distribution of ~4Cthymidine, D-arginine and ~SN-adenine in normal skin fibroblasts, together with that of 15N-adenine in organotypic cultures of h u m a n breast cancer cells.

Materials and methods

Biological models Normal skin fibroblast cells, obtained from 23-35-year-old women during plastic breast surgery, were propagated as monolayer cultures in Earle's modified Eagle's medium (EMEM; Boehringer) with 10070 fetal calf serum (SVF; Flow, Mc Lean, VA). After six passages, cells were detached from the plastic support using 0.050/0 trypsin and 0.02070 EDTA (Boehringer, Mannheim, Germany). Sixty #l of cell suspension (5 x 104 cells/ml) was deposited on l-cm 2gold strips. The strips were placed in Petri dishes and incubated in a 5070 CO2-95070 air atmosphere.

E Hindie et a!

82

Table I. The approximative natural concentration in tissues of some heavy stable isotopes. Main isotope concentration o tzmol/g of dry tissue b Hydrogen Carbon 12 Nitrogen 14 Oxygen 16 Oxygen 16

66000 40000 6000 11 000 11 000

Natural isotopic ratio heavy isotope (%) main isotope D/H 0.015 13C/12C 1.1 15N/14N 0.37 170/160 0.037 tsO/~60 0.2

Heavy isotope concentration izmol/g of tissue Deuterium 10 Carbon 13 440 Nitrogen 15 22 Oxygen 17 4 Oxygen 18 22

aThe first column is obtained from Bowen [4]; bthe molar concentration is given here for the monoatomic state of the element (H, C, N, O).

Twenty-four h after seeding, the fibroblasts were well attached, stretched onto the support and subconfluent. (Gold strips provide a pure and conductive support which favours the elimination of electrostatic charges which can build up during bombardment.) Four groups of three strips each were prepared. The first was incubated with fresh culture medium containing 1 nmol/ml of thymidine labelled with carbon 14 at all carbon positions (U-CI4 thymidine; Amersham, The Netherlands). The specific activity of the labelled molecule was 500/zCi/tzmol, giving a radioactivity concentration of 0.5/~Ci/ml of culture medium. The second was incubated with fresh culture medium containing 100 nmol/ml of arginine labelled with deuterium at seven positions (L-U-D7 arginine; CEA). The third was incubated with fresh culture medium containing 1/zmol/ml of adenine labelled with nitrogen 15 at the five nitrogen positions of the molecule (U-15-N adenine; CEA). In the fourth, the medium was renewed and the cultures were used as controls. After 24 h of incubation, the treated and control cultures were rinsed in a 0.1 M cacodylate buffer, fixed with 1070 glutaraldehyde in cacodylate for 20 min at 4"C, rinsed again in cacodylate, then rapidly rinsed in deionized water, and left to dry at 37°C before analysis. For the organotypic culture of cancer nodules we used a method described elsewhere [2]. The nodules were obtained from a cell line (MCF 7), derived from a human breast carcinoma [16] and were then maintained in continuous culture on a semi-solid agar medium consisting of RPMI 1640 with 0.5°70 bacto agar (Difco), 1007o fetal bovine serum, 4.5 mg/ml glucose and 0.04 IU/ml insulin. The nodules were subcultured every I0 days by dividing them in two with microsurgical scissors. The advantage of the organotypic cultures over monolayer cultures is that they maintain the 3-dimensional architecture of the cells and a higher degree * of differentmtaon. . . . For the. present experunent, three nodules were immersed in 1.5 ml liquid RPMI 1640 medium contalning 1/zmol/ml ~SN-adenine (CEA). After 48 h of incubation, the nodules were fixed with 1070glutaraldehyde in 0.1 M cacodylate buffer for 1 h at 4°C and set in Epon. One-/~m sections were deposited on a metal support for observation by ion microscopy.

Materials An ion microscoPe consists of a primary ion source, a mass spectrometer and a detection device for counting and localizing the selected secondary ions. The sensitivity in detection is influenced by the kind of primary ion used for bombardment. In the case of carbon 14, deuterium and nitrogen 15-emitted in the form of negative secondary ions - bombardment with cesium ions produces the highest yield. According to the type of mass spectrometer used, the mass-resolving power (M/AM, where M i s the mass of the ion under study and A M the smallest measurable mass difference) can range from a few hundred to several thousand determining the specificicy in detection. The distribution of the selected secondary ions can be recorded either by a direct procedure using a large primary ion beam focussed on the whole area analysed or by a scanning procedure, where the primary ion

beam is reduced to a small diameter and rastered over the surface. The secondary ions are detected and recorded according to the position of the beam to form a digital image. The scanning mode of acquisition is slower than the direct mode but the spatial resolution is better. Three instruments were used in this work. Two were direct ion imaging microscopes (SMI 300 and IMS 3F) and the third, UPS-ONERA, is an experimental scanning ion microprobe.

SAIl 300 Cameca Analysis with the SMI 300 direct ion microscope uses a 10 KeV O2 + primary ion bombardment• Negative secondary ions emitted from the surface are accelerated into a mass spectrometer comprising an electrostatic sector and a magnetic sector; the assembly gives a low mass resolution ( M / A M = 300). The selected secondary ions are projected onto a metallic cathode with a flat surface, giving rise to electrons which are recorded on sensitive films. The position of the ions arriving on the detector corresponds to that of the sample surface to within 1/zm.

IMS 3F Cameca This second-generation instrument is equipped with a Cs ÷ primary ion source. Its double-focussing mass spectrometer, composed of an electrostatic and a magnetic sector, provides a mass-resolving power as high as 10000. The selected ions are projected either onto a microchannel plate detector coupled to a fluorescent screen for imaging, or towards an electron multiplier for ion counting and mass spectra acquisition. As for the SMI 300, spatial resolution on the image is limited to 1/~m.

UPS-Onera ion microprobe This new instrument [15] has a Cs ÷ primary ion beam of 10 KeV. The currently used beam diameter is 0.1 t~m dictating the spatial resolution (the diameter is that of a disk containing 80070 of the intensity of the beam). The probe is moved stepwise within a square frame composed of up to 512 x 512 pixels. The mass spectrometer, comprising an electrostatic and a magnetic sector, achieves high mass resolution. Electron multipliers positioned at different exits of the spectrometer permit the detection of up to four different ion species simultaneously. A computerdriven acquisition system records the different events at each position of the beam, giving up to four different ion images simultaneously.

Results F i b r o b l a s t cultures i n c u b a t e d with 14C-thymidine were o b s e r v e d with the S M I 300 ion m i c r o s c o p e u n d e r 0 2 + b o m b a r d m e n t . W i t h this i n s t r u m e n t c a r b o n 14 was best detected in a s s o c i a t i o n with n i t r o g e n in the f o r m o f negative 14CI4N- ions o f mass 28 a t o m i c mass unit (amu). T h e d i a m e t e r o f the areas a n a l y s e d was 250 tzm. A n i o n i m a g e is first r e c o r d e d for identification o f the structures. A n exa m p l e o f such m o r p h o l o g i c a l images is the 12Cl4N- ion i m a g e at m a s s 26 s h o w n in figure l a , d i s p l a y i n g the f i b r o blasts o n the gold s u p p o r t with the nuclei clearly visible d u e to their high n i t r o g e n c o n t e n t . A 14Cl4N- i m a g e was then r e c o r d e d f r o m the s a m e a r e a to d e t e r m i n e the distrib u t i o n o f c a r b o n 14. T h e i m a g e o b t a i n e d (fig l b ) shows t h a t three f i b r o b l a s t s were in the D N A synthesis p h a s e at the t i m e o f i n c u b a t i o n as their nuclei i n c o r p o r a t e d 14Ct h y m i d i n e , a l t h o u g h the intensity o f labelling was d i f f e r ent in the three nuclei. F i b r o b l a s t cultures i n c u b a t e d with D-arginine were o b served with the I M S 3F ion m i c r o s c o p e u n d e r Cs + b o m b a r d m e n t . D e u t e r i u m was d e t e c t e d in the f o r m o f 2Dions a n d analysis at m a s s r e s o l u t i o n 3000 s h o w e d the a b -

Mapping labelled molecules

83

L_ Fig 1. Secondary ion images of human fibroblast cells incubated with t4C-thymidine, a. 12CI4N- 'morphological' ion image. h. 14CI4N- image showing the distribution of carbon 14. Diameter of imaged area = 250 ~m-SMI 300 microscope-O2 + bombardment.

Fig 2. Direct ion images of a single fibroblast, a. H - ion image. b. D- ion image showing the distribution of arginine. Bar = 20 ~ m - I M S 3F-Cs ÷ bombardment.

sence of mass interference. The distribution of hydrogen and of deuterium in a single fibroblast was recorded using a video imaging system which integrates the signal from the fluorescent screen. The H - ion image (fig 2a) shows a uniform distribution of hydrogen within the fibroblast, while the D - ion image (fig 2b) shows a much higher concentration in the nucleus than in the cytoplasm. This concords with the large contribution of arginine to the composition of nuclear proteins. Control cells gave a D signal of very low intensity corresponding to natural deuterium, and giving a D - / H - signal ratio of 0.012°70. This low background of naural deuterium was uniformly distributed, as was the case for hydrogen. Fibroblast cultures incubated with tSN-adenine were observed with the SMI 300 under O2+ b o m b a r d m e n t . Nitrogen 15 is not emitted as ~SN- because of the lack of electron affinity of this atom. However, it is intensely emitted in the form of 12CI5N-. Images were recorded from an area of 250 ~m in diameter. The 12CI5N- ion image (fig 3b) reflects the distribution of ~SN-adenine. Adenine is one of the bases shared by RNA and DNA. Figure 3b shows that ~SN-adenine was present in the nucleoli (arrow) and the cytoplasm, corresponding to newly synthesized RNA. Only one of the fibroblasts showed strong labelling of the whole nucleus, indicating that this cell was in a D N A synthesis phase. MCF 7 breast cancer nodules incubated with ~SNadenine were observed with both the SMI 300 microscope and the U P S - O N E R A microprobe. Images obtained from a section of the nodule with the SMI 300 under 02 + bombardment are shown in figure 4. The diameter of the area analysed was 250 ~zm. The morphological 12C14N- ion image (fig 4a) shows a large number of cells varying in size and shape, some surrounded by bright points which probably represent secretory activity. The ~2C~5N- ion image (fig 4b) obtained from the same area shows the distribution of lSN-adenine, the in-

84

E Hindie et al

corporation of which differs strongly from one cell to another. In some of the cells the nucleus is heavily labelled, indicating dividing activity. In other cells the molecule is mainly concentrated in the nucleoli and, to a lesser extent, in the rest of the nucleus and cytoplasm reflecting RNA synthesis. 15N-adenine was not incorporated by some of the cells, indicating very low or absent activity. The great disparity between the cells in the cultured cancer nodules reflects to some extent what happens in v i v o and stresses

the role of microanalytical methods in elucidating metabolic and pharmacotoxicological phenomena. The use of the UPS O N E R A ion microprobe illustrates the capacities offered by this new scanning instrument; ie high spatial resolution and simultaneous detection of several ions, with direct topographical correlation between the acquired numerical images down to the scale of the probe diameter. Using a 256 x 256 scanning process with a 0.1-/~m Cs ÷ ion microbeam, we analysed a 20-bLm x

Fig 3. Secondary ion images of fibroblast incubated with ~SN-adenine. a.

12CI4N- image, b. 12C15N- image showing the distribution map of ~SN-adenine (arrow shows one nucleolus). Diameter of the imaged area = 250 t~m-SMI 300 microscope-O2 + bombardment.

Fig 4. SIMS images obtained under 10 KeV O,_+ bombardment from a human breast cancer nodule incubated 48 h with 1 tzmol/ml of ~SN-adenine. a. z2C~4N- image, b. ~zC~SN- image showing the distribution of ~SN-adenine. Image field = 250 ttm-SMl 300 ion microscope.

Mapping labelled molecules 20-/zm area containing one of the cells of the cancer nodule. The 12C14N- ion image shows fine details of the cell structure, with the nucleoli and the nuclear membrane clearly distinguishable. The 31p- ion image shows the distribution of organic phosphorous and its high concentration in the nucleus. The 32S- image shows the distribution of sulfur. The mitochondriae (arrow), which are rich in sulfur are visible. The 12CI5N- ion image shows the distribution of 15N-adenine, with the nucleus of the cell heavily labelled. The incorporation of N15 adenine is intense in only one of the two nucleoli which also appears to be rich in sulfur.

Discussion

The tissular localisation of labelled molecules has so far been analysed using microscopic autoradiography [3]. This widely used technique presents some well-known limitations. Labelling is done with/3- emitting radio-isotopes of long half-lives (eg tritium, carbon 14, iodine 125) which preclude the use of the technique in the exploration of human pathologies. Microautoradiography is also a lengthy procedure requiring several weeks of contact between the tissue section and a photographic emulsion. In addition, determining local concentrations with this technique is difficult. Ion microscopy can thus be applied to the localisation of labelled molecules in tissues. Using this method it is possible to obtain micrographs showing the specific distribution of molecules labelled with either stable or radioactive isotopes. The spatial resolution currently achieved allows subcellular localisation to be analysed and the technique is rapid and sensitive. Different instruments have been used during this study from the old SMI 300 direct ion microscope to the experimental UPS-ONERA ion microprobe. The introduction of spectrometers with high mass-resolving power, the use of scanning microbeams and especially cesium ion microbeams, are major improvements for the study of labelled molecules. Although still in full development, ion microscopy appears to possess the characteristics required for molecular microlocalisation, ie specificity, high spatial resolution and sensitivity. Specificity

Specificity in detecting labelling isotopes can be affected by the presence of polyatomic ions 'ion clusters' with a mass very close to that of the labelling isotope. During the present study and in previous work we have characterised the different mass interferences [11]. Deuterium is detected as D- ions and no mass interference is observed. Specific detection of carbon 14 in the form of t4C- ions requires a mass resolution of 2000 to eliminate the interfering ~3CH- and 12CH2- ions [9]. With the low-mass-resolution instrument SMI300, carbon 14 was detected at mass 28 in the form of 14C~4N- ions which provided the best signal/background ratio due to the low intensities of the interfering ions [9]. The highmass resolving power of the IMS 3F-IMS 4F and UPSONERA instruments permits direct analysis ol~14C- ions. In the same way, it is possible to detect nitrogen 15 specifically in the form of 12C15N- ions since a mass resolution of 4500 permits the separation of the interfering 13ClSN- ions [10]. In conclusion, all interference with the labelling isotopes can now be eliminated by the use of spectrometers with high mass-resolving power.

85

II. Some characteristics of the labelling isotopes used in this study. Table

Carbon 14 Nitrogen 15 Deuterium

Natural concentration 5 x 10-s ~mol/g of dry tissue) ~2000 Mass resolution M/AM required for detection Estimated useful yield 1 x 10 -3

22

10

>4500

no interference

1 x

10 - 2

1 ×

10 - 4

After the mass separation step, the specific secondary ions reach a detector for measuring ion signal intensity or image formation. On ion microprobes such as the IMS 4F and UPS-ONERA the detector is an electron multiplier with a very low background noise (approximately one count per minute). This gives highly specific images since the average time required for the acquisition of one image is 10 min. Specificity in detecting a labelled molecule can also be affected by the natural presence of the isotope in the sample (table II). In the case of carbon 14, the natural background (0.05 pmol/g of dry tissue) is extremely low and there is no interference with the labelled molecule. The case is different with the heavy stable isotopes deuterium and nitrogen 15 which are present in appreciable amounts (10/zmol and 22 tzmol/g of dry tissue, respectively). However, the absence of toxic effects with stable isotopes means that high local concentrations of the labelled molecule can be achieved. Background subtraction is also possible when digital images of both the labelling isotope and its major isotope are recorded, as is the case for the 12CI5N- and 12CI4N- ion images in figure 5. Using the quantitative information contained in each pixel of the t2Cl4N- image, it is possible to subtract from the corresponding 12C15N- image the background due to natural nitrogen 15. This is done by multiplying each pixel of the 12CI4N- image by 3.7 x 10-3, ie the natural 15N/14N isotope ratio. The resulting matrix is then subtracted pixel by pixel from the 12CI5N- image matrix. Specific localisation also requires precautions to be taken during preparation of the specimen. We fLxedthe cells and nodules with glutaraldehyde which immobilizes macromolecules (nucleic acids, proteins, etc) in an inert phase. Molecules of 14C-thymidine, D-arginine and 15N-adenine which are not incorporated into these macromolecular structures are probably lost during the preparation of the sample. Freezing techniques are recommended when studying diffusible molecules. All steps, including cryosectioning and freeze-drying, should be done at very low temperatures [17]. Cryofracture of cultured cells has been used in an attempt to localise diffusible elements by means of ion microscopy [7]. A final aspect of specificity, which also concerns microscopic autoradiography, is that detection of the isotopes can also reflect the presence of a metabolic product. When assessing metabolic phenomena, labelling the molecules at different specific sites is therefore necessary. With ion microscopy this can be done in the same experiment by using several isotopes. Spat&l resolution

For all localisation techniques the spatial resolution is the capacity to attribute the studied molecule to specific cell

86

E Hindie et al

Fig 5. Ion images obtained from the same sample by a 256 × 256 scanning process with a 15 KeV Cs ÷ ion beam of 0. I t~m diameter. a. IEClaN- image, b. 31p- image, c. 32S- image (arrow = mitochondria) d. 12C15N- image. Area analysed 20 ~zm × 20 ~ m - b a r = 2 ~zm-UPS-ONERA ion microprobe.

structures. Using ion microscopy the ion image showing the distribution of the labelling isotope is correlated with an ion image showing the tissue structure. This ' m o r p h o logic' image can be obtained with elements such as sulfur, phosphorus and nitrogen which have different concentrations in the various cell structures, thereby providing contrast. The spatial resolution on the morphologic ion image depends upon the type of instrument used. Chromatic aberrations associated with the direct imaging ion microscopes SMI 300 and IMS 3F limit spatial resolution to approximately one micron. The new scanning ion microprobes such as the U P S - O N E R A provide much better resolution (about 100 nm). With this new resolution other intracellular structures and details become apparent (fig 5). Another advantage of the scanning instruments is the simultaneous acquisition of multiple ion species which provides direct spatial correlation between images. Superpo-

sition of digitised images is possible using a different colour code for each [6, 12]. Sensitivity

Sensitivity in detection by ion microscopy depends on the production of secondary ions. This ion yield is a function of the electronic characteristics of the atom but can be influenced strongly by the nature of the primary ions used for b o m b a r d m e n t [13]. Carbon 14, deuterium and nitrogen 15 produce mainly negative ions. Cesium, an alkaline electron donor, enhances the negative ion yield of these elements by a factor 100 to 1000 compared to oxygen bombardment. Instruments equipped with a cesium ion source are therefore much better adapted to the detection of labelling isotopes. Other instrumental factors which affect sensitivity are the capacity for collection, transmission and detection of the secondary ions at a given mass resolution. Using the UPS O N E R A ion microprobe, in which the

Mapping labelled molecules different aspects of sensitivity have been optimized, Slodzian et al [15] measured the useful yield of carbon in a graphite sample (the useful yield is the ratio of the number of carbon ions detected to the number of carbon atoms sputtered from the sample). This useful yield is 5 x 10 -3 for C - when the mass resolving power is 3000 at 2°70 valley between equal height peaks. In biological samples, the useful yield of carbon in the form C - may be a little lower due to the abundance of polyatomic ions ( C N - , C 2 H - , etc) which compete with C - emission. Considering this factor, the approximate useful yield of carbon 14 in the form of 14C- ions is I × 10 -3. This value is much higher than that obtained by carbon 14 microautoradiography. In ultrastructural microautoradiography one visible silver grain is obtained for 20 disintegrations [1]. Given the half-life of this isotope, after an exposure period of two months only one in 106 o f the carbon 14 atoms present in the sample would be detected. The sensitivity of carbon 14 detection provided by ion microscopy is therefore a thousand times better than that provided by microautoradiography.

87

elements hydrogen and nitrogen (table II) relative to that of carbon (I x 10-3). This can be done by measuring the intensity of the signal for ~2C-, H - , and 12Cl4N- taking into account the relative concentrations of these elements in the analysed tissue and the mass resolution necessary for a specific detection of the labelling isotope. Using these approximations the useful yield for deuterium seems to be about ten times lower than that of carbon (1 x 10-4), while that of nitrogen 15 is ten times higher (1 x 10-2). Due to its very high ion yield, nitrogen 15 may be preferred to deuterium for ion microscopy studies. The need to subtract the background noise due to the natural presence means that the detection limit of a molecule labelled with nitrogen 15 or deuterium will be approximately a hundred times higher than that o f a molecule labelled with carbon 14 [11]. Carbon 13, the heavy stable isotope of carbon, has limited value in ion microscopy due to its high natural concentration in tissues (440 ~mol/g). In contrast, oxygen isotopes, especially oxygen 17, have attractive characteristics. Molecules labelled with these isotopes are, however, difficult to obtain at the present time.

L o w e r detectable limit

Translating the useful yield of carbon 14 in terms of the lower detectable limit of a 14C-labelled molecule will depend essentially upon the spatial resolution required on the image of distribution. Let's consider the use o f a 256 x 256 scanning procedure to analyse first a large area, then a small area of the sample. In the first situation, using a relatively large cesium microbeam (1/~m in diameter) a 250 ~m x 250 ~m area of the sample is scanned until 1 ~m of the sample has been eroded. In this case the total volume analysed will be 250 × 250 x 1 = 62 500 ~m 3, ie 60 ng of biological tissue. If approximately, 300 ~4Cpoints on the imaged surface are necessary to interprete the distribution of the labelled molecule (to identify the nuclei which have incorporated ~4C-thymidine for example), with a useful yield of 1 × 10 -3, this corresponds to a sputtering of 3 x 105 atoms of carbon 14. If the molecule is labelled at one site only with carbon 14, it should then have an average concentration in the analysed volume of 10 pmol/g of dry tissue. In the second situation, a spatial resolution of about 100 nm is necessary to obtain finer details of the distribution of the labelled molecule inside a specific cell. This can be done by reducing the diameter of the ion beam to less than 100 nm, limiting the analysed area to 25/zm x 25/zm and eroding no more than 0.1 tzm. In this case, only 0.06 ng of the sample will be sputtered. To obtain 300 ~4C- points on the imaged surface, the concentration of the labelled molecule should be much higher (about 10 nmol/g of dry tissue). These examples show that a compromise has to be made between high spatial resolution and the local concentration of the labelled molecule. The very low natural abundance of carbon 14 in living tissues, together with its relatively high useful yield, make it the isotope of choice for ion microscopy studies. The majority of biological molecules so far synthesized can be obtained in a 14C-labelled form. Labelling with this isotope is reliable and does not modify the biological or metabolic properties of the molecule. The stable isotopes deuterium and nitrogen 15 can be useful when there are restrictions on the use of the carbon 14, as is the case for pharmacocellular studies in human pathologies. They are also of interest in studies using multiple isotope labelling. It is possible to estimate the relative useful yield of the

Conclusion Ion microscopy is a new approach to study the cellular and subcellular distribution of labelled molecules. Its sensitivity, rapid image acquisition and capacity to detect stable isotopes are all marked advantages over the classical method, microautoradiography. Ion microscopy can also be used to measure the concentration of a labelled molecule [11] on the basis of local isotope ratios 14C/12C, D / H , 15N/14N) and could become a method of choice for pharmacological studies at the cellular level.

Acknowledgments The authors wish to thank Mr G Slodzian, Mr B Daigne, Mr F Girard, Mr F Boust and Mr F Hillion for the ion images recorded on the UPS-ONERA ion microprobe and Mrs F Escaig for the ion images recorded on the SMI 300.

References l Bachmann L, Salpeter M (1967) Absolute sensitivity of electron microscope radio-autoradiography. J Cell Biol 33, 299-305 2 Beaupin R, Bilard C, Falcoff E (1986) Effects of human recombinant interferons-~, fl and "r on growth and survival of human cancer nodules maintained in continuous organotypic culture. Eur J Cancer Clin Oncol 22, 141-149 3 Belanger LF, Leblond CP (1946) A method for locating radioactive elements in tissues by covering histological sections with a photographic emulsion. Endocrinology 39, 8-13 4 Bowen HIM (1966) Trace elements in biochemistry. Academic Press, London and New York 5 Castaing R, Slodzian G (1962) Premiers essais de microanalyse par ~mission ionique secondaire. CR Sdances Acad Sci Paris 255, 1983 6 Cavellier JF, Escaig F, Boumati P, Gaume P, Hallegot P (1988) Numerisation and digital processing of images in secondary ion microscopy. Secondary Ion Mass Spectrometry, SIMS VI. (Benninghoven A, Colton RJ, Simons D, Werner HW, eds) Springer-Verlag, NY 385-388 7 Chandra S, Morrison GH (1985) Imaging elemental distribution and ion transport in cultured cells with ion microscopy. Science 228, 1543-1544

88

E Hindie et al

8 Galle P (1985) La microscopie ionique analytique des tissus biologiques. Ann Phys Fr 10, 287-305 9 Hindie E, Hallegot Ph, Chabala JM (1988) Ion Microscopy: A new approach for subceilular localization o f labelled molecules. Scanning Microscopy International 2, 1821-1829 10 Hindie E, Blaise G, Galle P (1990) Origin of the CNsecondary ions emitted from biological tissue under ten KeV Cs + bombardment. SIMS VII, Monterey, 3-8 Sept 1989 (Benninghoven HA, Evans CA, McKeegan KD, Storms HA, Werner HWW, eds) John Willey and Sons, 335-338 11 Hindie E (1990) Apport de la microscopie ionique en biologie: localisation intracellulaire et appr6ciation de la concentration locale de mol6cules marqu6es. Th6se es Sciences, Universit6 Paris Val de Marne 12 . Kahn E, Olivo JC, Marque I, Halpern S, Larras-Regard E, Fragu P, Di Paola R (1986) Image processing in ion microscopy by the PC-AT computer. Secondary Ion Mass

13 14 15

16 17

Spectrometry, SIMS VI (Benninghoven A, Colton R J, Simons D, Werner HW, eds) Springer-Verlag NY, 381-384 Krohn VE (1962) Emission of negative ions from metal surfaces bombarded by positive cesium ions. J Appl Phys 33, 3523 Slodzian G (1987) Basic aspects in ion imaging with secondary ions. Scanning Microscopy Suppl 1 1 Slodzian G, Daigne B, Girard F, Boust Hillion F (1990) Cartographie paraU/;le de plusieurs 616ments ou isotopes par balayage avec une sonde ionique submicronique: premiers r6sultats. CR Acad Sci Paris 311, s6rie II, 57-64 Soule HD, Vazquez J, Long A, Albert S, Brennan MA (1973) A human cell line from a pleural effusion derived from a breast carcinoma. J Natl Canc Inst 51, 1409 Terracio L, Schwabe KG (1981) Freezing and drying of biological tissues for electron microscopy. J Histochem Cytochem 29, 1021