A move for the better.

02 tensions (Po2) were measured with microelectrodes in multicellular spheroids from EMT6/Ro and V-79-171B cells. The measurements were performed in spheroids kept in flowing growth medium that was equilibrated with 5% CO2 and air at a temperature of 37°C and contained 5-5 mm glucose. The recorded Po2 profiles are characterized by a diffusion-depleted zone surrounding the spheroids and by a steep drop in Po2 within the spheroids over mean distance of 220 and 188 um from the surface of EMT6/Ro and V-79-171 B spheroids respectively. Smaller spheroid exhibit parabolic Po2 profiles, larger ones show a central plateau. The region of the steep decrease in Po2 corresponds to the thickness of the viable rim: the plateau region is created by the absence of 02 consumption in the central necrotic area. Po2 in the centre of EMT6/Ro spheroids decreased from 66 mmHg at a diameter of 400 ,tm to 13 mmHg at a diameter of 1000 zm. Under the present conditions during growth and in the experiments, values below 5 mmHg were recorded only in spheroids > 1200 /am. Comparably low Po2 was recorded in V-79 spheroids with diameters of 650 um+. In spheroids of this cell type with a diameter of 400 ,am, Po2 was 42 mmHg. The findings provide evidence that necrosis may arise at average Po2 of 57 and 42 mmHg in EMT6/Ro and V-79-171B spheroids, respectively, grown under the conditions described. MULTICELL SPHEROIDS represent an in vitro tumour model in which the cancer cells are supplied by the diffusion of substrates from the surrounding growth medium. At constant substrate concentrations in the growth medium the efficiency of the nutritive supply to the tumour cells depends on the location of the cells within the spheroids. Cells in the inner part of the spheroid may be located beyond critical diffusion distances and may die from lack of nutrients. In agreement with these considerations, it is observed that spheroids are characterized by the development of central necrosis as they increase in size (Sutherland et al., 1971). Thus, by growing multicellular spheroids conditions for the tumour cells can be generated similar to the situation of cancer cells located in between nutritive vessels in solid tumours and being supplied by diffusion of substrates from the tumour capillaries. Restrictions in blood supply, as they may occur in solid tumours with increasing tumour weight (Vaupel, 1977), can lead to a decrease of the concentration of …

and air at a temperature of 37°C and contained 5-5 mm glucose. The recorded Po2 profiles are characterized by a diffusion-depleted zone surrounding the spheroids and by a steep drop in Po2 within the spheroids over mean distance of 220 and 188 um from the surface of EMT6/Ro and V-79 -171 B spheroids respectively. Smaller spheroid exhibit parabolic Po2 profiles, larger ones show a central plateau. The region of the steep decrease in Po2 corresponds to the thickness of the viable rim: the plateau region is created by the absence of 02 consumption in the central necrotic area. Po2 in the centre of EMT6/Ro spheroids decreased from 66 mmHg at a diameter of 400 ,tm to 13 mmHg at a diameter of 1000 zm. Under the present conditions during growth and in the experiments, values below 5 mmHg were recorded only in spheroids > 1200 /am. Comparably low Po2 was recorded in V-79 spheroids with diameters of 650 um+. In spheroids of this cell type with a diameter of 400 ,am, Po2 was 42 mmHg. The findings provide evidence that necrosis may arise at average Po2 of 57 and 42 mmHg in EMT6/Ro and V-79-171B spheroids, respectively, grown under the conditions described.
MULTICELL SPHEROIDS represent an in vitro tumour model in which the cancer cells are supplied by the diffusion of substrates from the surrounding growth medium. At constant substrate concentrations in the growth medium the efficiency of the nutritive supply to the tumour cells depends on the location of the cells within the spheroids. Cells in the inner part of the spheroid may be located beyond critical diffusion distances and may die from lack of nutrients. In agreement with these considerations, it is observed that spheroids are characterized by the development of central necrosis as they increase in size (Sutherland et al., 1971).
Thus, by growing multicellular spheroids conditions for the tumour cells can be generated similar to the situation of cancer cells located in between nutritive vessels in solid tumours and being supplied by diffusion of substrates from the tumour capillaries. Restrictions in blood supply, as they may occur in solid tumours with increasing tumour weight (Vaupel, 1977), can lead to a decrease of the concentration of nutrients, such as 02 in the tumour capillaries (Vaupel et al., 1979), a situation which can be simulated in spheroids by lowering the 02 content in the growth medium. Which particular substrate may critically control the metabolic state, the cell cycle and the viability is difficult to answer. A profound analysis of this problem will require more experimental data than are currently available. Since spheroids exhibit almost ideal spherical geometry, it is possible to establish and solve the diffusion equation (Boag, 1969;Franko & Sutherland, 1979a) in its general form for substrates such as 02. However, the theoretical estimation of the actual 02 concentration in the spheroids is impeded by the fact that the metabolism of 02 is influenced by the glucose levels and vice versa (Crabtree, 1929;Golsalvez & Weinhouse, 1976;Vaupel & Thews, 1976). Furthermore, the possibility that 02 consumption at low [02] is reduced according to Michaelis-Menten kinetics must be considered (Froese 1962;Koch & Biaglow, 1978). Additional complications arise from the unknown role of factors in the interstitial milieu of spheroids (e.g. pH) which might also influence the tumour-cell metabolism. In particular, it has been demonstrated that the glucose consumption of tumour cells depends on the concentration of H+ ions in the surrounding medium (Zwartouw & Westwood, 1958;Paul et al., 1966;Bock & Frieden, 1976). Diffusion equations assuming consumption rates that are independent of the concentration of some of these substrates may, therefore, only yield a rough approximation to the actual distribution of these metabolites within spheroids.
An experimental approach to this problem is rendered feasible by the histological determination of the thickness of the viable rim in spheroids as a function of the [02] in the growth medium (Franko & Sutherland, 1979a,b;Sutherland & Durand, 1973 Sutherland, 1982) that convection in the growth medium is a decisive determinant of the oxygenation of spheroids, 02 tensions were recorded in spheroids under conditions they were exposed to during growth and during many previous experiments. These studies were mainly carried out on EMT6/Ro and V-79-17 1B spheroids. Therefore, spheroids of both cell lines in different stages of growth were subjected to measurement, in the present study.

MATERIALS AND METHODS
Spheroids of EMT6/Ro and V-79-171B cells were cultured in spinner flasks at 37°C as previously described (Freyer & Sutherland, 1980;Sutherland & Durand, 1976). The culture medium was Eagle's basal medium with 15% (v/v) and 500 (v/v) foetal bovine serum for the EMT6/Ro and the V-79-171B cells respectively. The medium was replenished daily. The spheroids used for experiments were removed from the flasks immediately before measurement. The diameters of all spheroids included in this study were determined in an inverted phase-contrast microscope. The geometric mean of 2 orthogonal diameters was taken as the mean spheroid diameter. 02 tensions in the spheroids were assessed by means of 02-sensitive microelectrodes. The probes w-ere constructed with outer tip diameters of 1-5 ,m, according to Whalen et al. (1967Whalen et al. ( , 1973a. The electrodes consist of micropipettes filled with Wood's metal (Fisher, Fair Law, NJ) electrolytically plated with gold, and covered with collodion (Merck, Rahway, NJ). The gold serves as an 02 cathode which is recessed 10-20 um from the micropipette tip, thus yielding a spatial resolution of a few microns (Schneidermann & Goldstick, 1978).
Since the experimental apparatus, the calibration of the electrodes and the measuring protocol have been published in a previous paper (Mueller-Klieser & Sutherland, 1982) only a brief description of the experimental set-up and the measuring procedure is to be given here. Using an average polarization voltage of 0 7 V the electrode signal is amplified (Transidyne, Ann Arbor, MI) and displayed on a chart recorder (Linseis, Princeton, NJ). Before and after each measurement, the electrode is calibrated at 3700 in growth medium with glucose and glucose oxidase (INC, Pharmaceuticals, Cleveland, OH) and in air-equilibrated medium. About every 5 experiments the linearity of the probes is checked using medium gassed with 8-91% 02 (remainder N2) as a third calibration point. Measurements were only considered for evaluation if the pre-and post-study calibrations did not differ more than 5%, which was usual.
Measurements were carried out in a special thermostatted measuring chamber (Mueller-Klieser & Sutherland, 1982) that has been designed to create the conditions to which the spheroids are exposed during growth and during numerous experiments with radiation and/or drugs. Temperature, 02 and C02 content, as well as pH in the medium flowing through the chamber can be controlled and maintained constant. The spheroid is put on to an 02 permeable membrane and held in its proper position by a micropipette vertically inserted into it. The positioning of the microelectrode in relation to the spheroid by a manual micromanipulator can be observed through a dissecting microscope and through a window in the front of the measuring chamber. The penetration of the electrode into the spheroid is controlled by a hydraulic microdrive. The electrode is stepwise driven from the medium into the spheroid on a track leading through the centre of the spheroid. The probe is kept at each step for min, by which time a steady-state reading has usually been reached. All measurements are carried out in a Faraday cage to prevent electrical interferences. -79-171B

RESULTS
A total of 399 steady-state measurements in 20 spheroids of EMT6/Ro cells and of 289 steady-state readings in 20 spheroids of V-79-171B cells were recorded. The. diameters of the EMT6/Ro spheroids ranged from 386 to 1900 ,um; the V-79-171B spheroids investigated were 376-1052 Km in diameter. Fig. 1 shows 3 representative Po2 profiles in EMT6/Ro (a) and in V-79-171B cells (b) at 3 different sizes. All profiles are characterized by a decrease of 02 in the medium directly surrounding the spheroid, thus lowering the Po2 at the surface of the spheroid (arrows in Fig. 1) considerably below that in the bulk of the medium. The shapes of the Po2 profiles within the spheroids were parabolic in small spheroids but the gradients were steep from the edge towards the interior with a central plateau in larger spheroids. The steep decrease of Po2 did not continue beyond 220 + 34 ,um from the rim in EMT6/Ro spheroids and beyond 188 + 31 ,um from the rim in V-79-171B spheroids. The extent of this region corresponds with the histologically determined thickness of the rim of viable cells. Fig. 2 shows histological thin-sections from EMT6/Ro (a) and V-79-171B (b) spheroids.
The Po2 values measured in the centres of EMT6/Ro (closed dots) and of V-79-171-B spheroids (open dots) are plotted as a function of spheroid size in Fig. 3. In both spheroid types, Po2 in the centre of the spheroids decreases with increasing spheroid size. The correlation between centre Po2 and size can be approximated by an exponential function within a certain size range. This is indicated in the legend to Fig. 3. Po2 in the innermost part of EMT6/Ro spheroids dropped from an average of 65 mmHg at a diameter of 400 [km to 10 mmHg at a diameter of 1100 ,um. Values between 0 and 5 mmHg were observed only in spheroids larger than 1200 ,um in diameter. In V-79-171B spheroids with a diameter of 400 ,um, an average Po2 of 42 mmHg was re- corded. Po2 in the centres dropped to an average of 2 mmHg in spheroids with a diameter of 680 ,um. Values between 0 and 5 mmHg occurred on average in spheroids larger than 600 ,um, though similar low values were detected in 2 spheroids larger than 500 pm in diameter. Both spheroid cell types showed a tendency to slightly greater central Po2 at the upper end of the diameter scale.

DISCUSSION
Oxygen partial pressures have been monitored in spheroids using the Whalentype 02 microelectrode. The performance characteristics of these probes have been experimentally (Whalen et al., 1973a,b) and theoretically (Schneidermann & Goldstick, 1978) analysed showing that they yield reliable Po2 measurements with a high spatial resolution, a low stirring sensitivity, a low 02 consumption and minimal tissue damage. According to Silver (1973) these characteristics are required to record proper tissue Po2.
With 02 microelectrodes constructed according to these criteria, 02 profiles were found in spheroids which closely correlated with their histological properties. A steep drop in Po2 within 220 and 188 pm in EMT6/Ro and V-79-171B spheroids (see Fig. 1), respectively, as the microelectrode penetrated the spheroids from the edge towards the centre, corresponded with the thickness of the viable cell rim, as demonstrated in Fig. 2. The viable cells in this part of the spheroid consume 02 as it diffuses from the bulk of the medium into the spheroid, thus lowering [02] towards the spheroid centre.
In EMT6/Ro spheroids larger than 450 pm, and in V-79-171B spheroids larger than 400 ,um, central necrosis can be found (Franko & Sutherland, 1978). The absence of 02 consumption should result in constant Po2 levels across those areas. As demonstrated in Fig. 1, central plateaux were registered in 02 profiles from larger spheroids, corresponding with the central necrotic area.
In order to determine at which 02 levels the tumour cells in spheroids cease consuming 02 and disintegrate, the Po2 in the spheroid centre was plotted against the spheroid size for both EMT6/Ro and V-79-171B spheroids in Fig. 3. Assuming necrosis to develop at diameters of 450 and 400 ,um in EMT6/Ro and V-79-171B spheroids respectively (Franko & Sutherland, 1978), necrosis may arise at Po2 of 57 and 42 mmHg, respectively.
These levels are surprisingly high in comparison to values found in experimental tumours in rodents (Vaupel, 1977;Mueller-Klieser et al., 1981), in which measured tissue Po2 values are mostly < 10 mmHg. The experiments in solid tumours provide evidence that 02 is a critical factor in the control of cell viability in the investigated tumour type. However, in spheroids grown under the conditions described necrosis may develop at Po2 that is presumably much above any critical Po2 (values inducing cell death) (Froese, 1962;Koch & Biaglow, 1978). Even though previous experiments have demonstrated that 02 is involved in some way in the control of cell death (Franko & Sutherland, 1978, 1979a in V-79-171B spheroids, the present findings indicate that cells may not only disintegrate because of 02, deprivation, but also because of the restricted diffusive supply of some other nutrient, e.g. glucose.
The curves in Fig. 3 also show that cell death occurs at different Po2 as the spheroids increase in size, indicated by a decrease of the plateau Po2 values with increasing spheroid diameter.
One possible explanation for the decrease of Po2 in the centre of spheroids with increasing spheroid size is the deterioration of the diffusion conditions owing to a change in geometry. The diffusive flux of nutrients to the cancer cells is proportional to the spheroid surface area, according to Fick's law. The total consumption of a substrate in a spheroid may be assumed to be proportional to the total volume of consuming cells, i.e. the volume of the viable rim. The ratio of the surface area (S) to the volume of the viable rim (V) as a function of the spheroid radius R can be written as: The surface to volume ratio S/ V decreases with increasing spheroid size, thus restricting the area through which nutrients diffuse into the spheroids in comparison to the volume of the consuming cells. An indication that this assumption is effective in spheroids is given by the consideration of the Po2 at the spheroid surface as a function of spheroid size, as plotted in Fig. 4. The data show that, ,~~~~F ra. 4. Po2 at the surface of 21 EMT6jRo spheroids as a fulnction of sppheroid size. 18 smaller spheroids tend to have a higher Po2 at their surface. Although there is a considerable scattering of the data, a trend is apparent that can be explained by the decrease of the S/V ratio. The solid line in Fig. 4 represents a non-linear least-squares fit for the data, using a function of the type given in Equation (1). The scattering of the surface Po2 is mainly due to variable convection of the medium surrounding the spheroids. It has been pointed out in a previous investigation (Mueller-Klieser & Sutherland, 1982) that the diffusion-depleted zone and the surface Po2 are susceptible to flow changes in the spheroid environment, and that it is very difficult to create exactly the same conditions in terms of convection of the growth medium for each single spheroid measured with microelectrodes. Nevertheless, the data shown in Fig. 4 provide evidence that changes in geometrical properties with increasing spheroid diameter may impede the diffusion of 02 into the spheroid, and may lower the Po2 within the spheroid.
The same consideration can also be applied to the diffusion of glucose, leading to a decreasing glucose concentration in central regions with increasing spheroid size. A drop in glucose concentration presumably leads to an increase in 02 consumption (Crabtree, 1929;Golsalvez & Weinhouse, 1976) and consequently to a lowering of the cellular [02]. This may also contribute to the decline of Po2 in the centres of spheroids as they increase in size.
A slightly higher Po2 in the centre of large spheroids compared to medium spheroids is seen in both spheroid types (Fig. 3). This cannot be explained by the present data. It is questionable whether the small rise in Po2 in larger spheroids has any significance for the radiosensitivity or metabolism and cell cycle of the cells within the spheroids.
The difference in Po2 between EMT6/Ro and V-79-171B spheroids can be qualitatively explained by differences in the packing density of the cells. It can be seen in histological thin sections that the intercellular space is much more extended in EMT6/Ro than in V-79-171B spheroids (Sutherland, unpublished), producing a higher density of 02-consuming sites in the latter case and lower tissue Po2. Differences in metabolism of both cell lines cannot be excluded as a further explanation for the differing 02 levels in their spheroids. However, more data are required to make detailed statements in this regard.
Investigations of cell viability and radiation sensitivity in EMT6/Ro and V-79-171B spheroids generally agree with the findings reported here. Sutherland et al. (unpublished) showed that radiobiological hypoxia did not occur in EMT6/Ro spheroids smaller than 1200 [km in diameter grown in air-equilibrated medium with 5.5mM glucose, but may occur in spheroids larger than 1400 [m.
Assuming that a Po2 of 3 mmHg halves the radiosensitivity of tumour cells relative to a Po2 of 20 mmHg or more (Tannock, 1972) the results of the present study suggest that radiobiological hypoxia in EMT6/Ro spheroids of 1200 pm in diameter.
Investigations in V-79-171B spheroids of the correlation between thickness of the viable cell rim and [02] in the medium generated results (Franko & Sutherland, 1979a) which could be explained, among other interpretations, by cell death occurring at 500 (v/v) in the equilibrating gas phase, corresponding to a Po2 of 35 mmHg. Even though this explanation was rejected as unlikely, the results of the present investigation yield strong support for it. It can be deduced from Fig.  3 that necrosis in V-79-171B spheroids of 400,um diameter arises at an [02] of 6% (v/v) in the equilibrating gas phase.
The development of necrosis at [021 levels higher than any possible critical [02] is in agreement with P02 measurements in spheroids by Carlsson et al. (1979). These authors found a different correlation between central Po2 values and diameters in V-79-171B spheroids from that found in the present investigation. However, this can be attributed to differences in the growth conditions, particularly in the convection of the culture medium during growth. It has been shown (Mueller-Klieser & Sutherland, 1982) that the oxygenation of spheroids is susceptible to convection in the growth medium, which may also apply to other nutrients. Thus, metabolism and growth pattern are very likely to be different in spheroids growing on agarose gel (Carlsson et al., 1979) from spheroids cultured in spinner flasks. A further apparent difference in oxygenation of the spheroids presumably arises from different recording techniques. While the spheroids in this study were measured immediately after removal from the spinner flask, Carlsson et al. (1979) allowed the spheroids to attach to a glass surface for 6-12 h. This not only impedes the uniform 02 diffusion into the spheroid, as previously demonstrated (Mueller-Klieser & Sutherland, 1982) but may also influence consumption. For example, Fig. 5 shows Po2 profiles in a spheroid immediately after removal from the spinner flask (solid curve) and in the same spheroid attaching to the supporting O2-permeable membrane for 8 h (dashed curve). From the different slopes of both profiles it can be concluded that the 02 consumption is lower in the attached than in the freshly investigated spheroid. This may be due to restricted supply of substrates and inadequate removal of metabolic waste in the attached spheroid (which is suspended in static medium for several hours). Finally, it cannot be excluded that the V-79 strain used by Carlsson et al. (1979) had an 02 consumption different from the cell line used in this study. However it can be stated that, despite those differences in measured Po2, necrosis develops in V-79 spheroids in the 2 laboratories under both experimental conditions, where the Po2 is not severely low.
In the experiments reported here Po2 was measured under conditions similar to those of the spheroids during growth and numerous previous experiments. Under these conditions, surprisingly high [02] tensions were found in small and medium size spheroids with necrosis in the centre. These findings suggest that 02 lack may not be the only cause of cell death; cell necrosis in the investigated spheroid types under the present growth conditions may be due to the diffusion limitation of another substrate and/or a metabolic waste product.
An attempt to simulate the conditions in the vascular network of solid tumours requires the reduction of the Po2 in the in the growth medium to 35-40 mmHg, i.e. well below the 140 mmHg used in this study. The study of variations in glucose concentration in the medium should provide additional information useful in understanding the microenvironmental and metabolic peculiarities of cancer cells in solid tumours. This will be the subject of our future investigations with spheroids.