The toxic effect of spermidine on normal and transformed cells

Biological polyamines such as spermidine and spermine are widely distributed in bacteria and fungi as well as in mammalian cells (Herbst & Snell, 1949; Rosenthal & Tabor, 1956; Herbst, Weaver & Keister, 1957). It has been reported that these poly amines have a potent cytotoxic effect on tissue culture cells in the presence of calf serum, while no such effect is observed with horse or human serum (Alarcon, Foley & Modest, 1961; Alarcon, 1964; Bachrach, Abzug & Bekierkunst, 1967). The interpretation has been that polyamines are not cytotoxic by themselves but become so by the action of monoamine oxidase which is known to be present in calf serum but not in human or horse serum (Werle & Roewer, 1952; Hirsch, 1953; Tabor, Tabor & Rosenthal, 1954). The cytotoxic agents are probably oxidatively deaminated products of polyamines which have active aldehyde groups (Alarcon, 1964; Tabor, Tabor & Bachrach, 1964).

This communication describes a simple method for assaying the cytotoxic effect of oxidized spermidine on tissue culture cells. This assay method is then used for comparative tests on various tissue culture cell lines and on cells in various stages of the cell culture’s growth cycle and the general conclusions that can be drawn are discussed.

The Syrian hamster cell line BHK21/C13 (Macpherson& Stoker, 1962; Macpherson, 1963) and the recloned polyoma virus-transformed derivatives PyY (Stoker, 1962) were most com monly used. Another derivative, BHK21/C 13/rif 75, resistant to rifampicin up to 75 μg/ml (Subak-Sharpe et al. 1970) was made available by Prof. Subak-Sharpe and mouse cell lines C3, C10, polyoma virus-transformed rat cell line (RPT₁), and adenovirus type 2 transformed rat cell line (RT₁) were kindly supplied by Dr J. F.Williams and Mr S. Ustacelebi in this laboratory. Mouse L, HeLa, Hep-2 and BSC-1 were also used and were freshly prepared.

A monolayer of BHK21JC13 cells (1 × 10⁶ cells/50-mm Petri dish) was infected with 0·2 ml of polyoma virus suspension (around 10⁹ pfu) for 2 hat 37 °C. Then the cells were washed with Eagle’s medium, trypsinized and the monodisperse cells transferred into soft agar suspension cultures as described by Macpherson & Montagnier (1964). After 10 days incubation 11 single polyoma-transformed colonies growing in the agar suspension were isolated by Pasteur pipette. They were propagated by passaging twice a week and all showed the growth characteristics of transformed cells.

Using a small plaque variant derived from the Toronto strain of polyoma virus, a virus stock was prepared from secondary mouse embryo cultures by treating the infected cell debris with receptor-destroying enzymes (Crawford, 1962). The stock titre was 5 × 10⁶ pfu/ml.

All cell lines were propagated at 37 °C in 20-oz (570-ml) flat bottles or 80-oz (2·3-1.) Winchester bottles using modified Eagle’s medium (Eagle, 1959) supplemented with 10% tryptose phosphate broth and 10% unheated calf serum (10% ETC).

Confluent monolayers from the 20- or 80-oz bottles were trypsinized, washed twice with 10 % ETC and finally suspended in the same medium at a concentration of 10⁶ cells/ml; 50-mm plastic Petri dishes were seeded with 5 ml of this cell suspension and kept in a humidified CO₂ incubator at 37 °C. After appropriate time intervals cells were trypsinized and suspended in Eagle’s medium supplemented with 10 % tryptose phosphate broth and 1 % unheated calf serum (1 % ETC). An aliquot was taken for cell counting and the rest were subjected to the standard spermidine test.

The cell suspension prepared as above was centrifuged at low speed. The cell pellet was re-suspended in 1 % ETC at 10⁶ cells/ml, and distributed into 30-mm plastic Petri dishes in 2-ml aliquots. To each Petri dish 0·1 ml of spermidine trihydrochloride solution was added. The concentrations of spermidine solution covered the range from 5 to 150 μg/ml in steps of 5 μg/ml. The Petri dishes were incubated at 37 °C in a humidified CO₂ incubator and after 15–20 h incubation cultures were examined with an inverted microscope. The cytotoxic effect of spermidine was easily observable above a certain concentration of spermidine as shown in Figs. 7–10. The maximum amount of spermidine which did not show cytotoxic effect in this assay is defined as the spermidine index (SI).

Unless otherwise specified trypsinization was carried out in the following way. After removing the medium from the cell sheet, trypsin solution (0·25 % in tris-saline) was added then immediately removed. The cell sheet was left at room temperature until all cells had rounded off from the Petri-dish surface, and then the cells were suspended in fresh medium. All cell lines used gave single-cell suspensions by this method.

Cell suspensions were diluted to around 10⁴ cells/ml with phosphate-buffered saline (Dulbecco & Vogt, 1954) and counted with a Coulter counter.

Tests similar to those described above were carried out using [¹⁴C]spermidine except that twice the standard number of cells were used. After the incubation the medium was removed from each Petri dish with a Pasteur pipette, 2 ml of trypsin solution added, the cells therein suspended and then separated from the trypsin solution by low-speed centrifugaiion. The cell pellet was suspended in 2 ml of 5 % cold trichloroacetic acid (TCA) and the TCA-soluble fraction separated by a further centrifugation. The TCA-insoluble fraction was collected on Millipore filters (0·45 μm). The radioactivities of medium, trypsin-soluble and TCA-soluble fractions were measured in a Nuclear Chicago liquid scintillation counter after preparing the samples as follows: 0·1 ml of the solution was applied to a glass-fibre disk (Whatman GF/C, 21 mm), which was transferred after drying into vials containing toluene-based scintillation fluid. The TCA-insoluble fractions collected on Millipore filters were treated and counted like the glass-fibre disks.

Spermidine trihydrochloride and crystalline bovine serum albumin were purchased from Sigma Chemical Co. and Spermidine-C-14-trihydrochloride (specific activity, 10·7 mCi/mM) from The New England Nuclear Corporation.

The cytotoxic effect of spermidine on BHK2I/C13 cells is shown in Figs. 7–10. In the presence of 1·5 μg of spermidine all cells are rounded (Fig. 10) and with 1·0 μg some cells spread, but most remain rounded (Fig. 9). But at 0·5 μg spermidine, virtually all cells spread (Fig. 8). The cytotoxic effect is irreversible: once cells are rounded by the action of spermidine, they never spread again even after extensive washing with fresh medium, and no cell division is observed even with prolonged further incubation. In this standard spermidine assay the maximum amount of spermidine tolerated before an observable cytotoxic effect was 0·5 μg. This is a highly repeatable characteristic. Hence the SI of confluent BHK21/C 13 cells is 0·5 μg. As Tabor et al. (1954) reported, spermidine did not show any cytotoxic effect in the presence of horse or human serum which does not contain monoamine oxidase.

When the number of cells plated per dish was increased in the spermidine test system, more spermidine was required for cytotoxic effect. In Fig. 1 the SI values are plotted against the number of cells in the spermidine test system. This correlation suggests that the cells possess some definite capacity to neutralize the cytotoxic effect of spermidine. The possibility that spermidine being a strong base produced complexes with the cellular nucleic acids, thus avoiding oxidation by serum monoamine oxidase, can be ruled out, because the experiments using labelled spermidine suggest that very little radioactivity remains associated with the cells.

The distribution of radioactivity in cultures treated with labelled spermidine is shown in Table i. No cytotoxic effect appeared in the culture treated with 1·0 μg of labelled spermidine. In this culture 96% of the radioactivity was found in the medium, while only 4% was associated with cells. Increasing the amount of spermidine to 2·0 μg resulted in the unambiguous cytotoxic effect, but the distribution of the radio activity was unchanged; 98% was present in the medium and 2% in the cells. The radioactive compound in the medium was analysed by paper chromatography using butanol:pyridine: acetic acid: water (15:10:3:12) as solvent. The R_F value of spermi dine in this system was 0·15. The major radioactive compound found in the medium had an R_F value of 0·24 and no spermidine was detected. The chemical structure of the compound is not identified but the above strongly suggests that the spermidine was completely oxidized in both cultures. In a separate experiment it was found that up to 20 μg of spermidine were completely oxidized in the standard spermidine test system.

Another possibility is that the cells possess some capacity to neutralize the toxic substance produced by the oxidation of spermidine. To test this hypothesis the following experiments were performed. Spermidine (15·0, 12·5, 10·0, 7·5, and 5·0 μg) was incubated in a 30-mm Petri dish with 2 × 10⁶BHK21/C13 cells from a confluent monolayer suspended in 2 ml of 1 % ETC. Controls without cells were also run. After 16 h incubation at 37 °C the cytotoxic effect was clearly seen in the cultures containing more than 7·5 μg of spermidine, while 7·5 and 5·0 μg of spermidine failed to show any cytotoxic effect. The medium was taken from each Petri dish and tested for persistence of the cytotoxic effect against 2 × 10⁵ fresh cells in the standard test system. If cells, only while still alive, can neutralize the effect of spermidine, then it would be expected that the medium containing more than 10·0 μg of spermidine might retain cytotoxicity while the rest would not. It was found that all the media from cell-containing cultures had lost their cytotoxic effect. The controls which were incubated without BHK21/C13 cells showed that the cytotoxic substance produced by the oxidation of spermidine is stable at 37 °C for at least 20 h. Hence 2 × 10⁶BHK21JC13 cells are able to neutralize more than 15·0 μg of spermidine, although they do not survive in the presence of 10·0 μg or more of spermidine.

Serum also neutralizes the cytotoxic effect of spermidine. This neutralizing or SI reducing activity of serum is retained in the dialysis bag, which indicates that it has high molecular weight. Calf serum was fractionated by gel filtration on a Sephadex G-200 column into 3 fractions as shown in Fig. 2. Si-reducing activity was only detected in the last peak, which mostly consisted of albumin. The neutralizing activity of purified serum albumin was therefore tested. In Table 2, Si-reducing activity of calf serum and bovine serum albumin are compared; 0·2 ml of calf serum neutralizes 1·5 μg of spermidine and 10 mg of albumin neutralize 2·0 μg. Since 0·2 ml serum contained about 8 mg of albumin (Schultze & Heremans, 1966), it can be concluded that all the Si-reducing activity of serum is probably due to serum albumin. The molar ratio between 10 mg albumin and 2·0 μg spermidine is 20:1, assuming the molecular weight of albumin is 6·9 × 10⁴ and that of spermidine 255. As 20 molecules of albumin appear to be required to neutralize one molecule of spermidine it appears unlikely that the neutralizing effect is due to spermidine molecules being adsorbed to large albumin molecules by their negative charges. If this were the case, many spermidine molecules would be adsorbed on to one molecule of albumin.

All experiments reported so far were done with BHK21JC13 cells derived from confluent cultures where the spermidine index is a reliable characteristic. However, the SI varies greatly if the cells are obtained at different stages in the formation of a monolayer.

BHK21/C13 cells were seeded in Petri dishes as described in standard growth cultures and harvested for testing at various stages up to the formation of a confluent monolayer. The growth curve and SI are shown in Fig. 3. There is an initial lag of more than 12 h, after which the cells number increases exponentially with a mean doubling time of around 12 h. The SI at 5 h is 0·5, which is the same value as for con fluent cells. It rises sharply at 12 h and reaches its maximum of 3·5 at 15 h. In the exponential phase it falls gradually and settles down to 0·5 when the monolayer has reached confluence.

Standard spermidine tests were carried out with various polyoma virus-transformed BUK2I/C13 lines. The growth curve and SI of one such line (clone 7) are shown in Fig. 4. The pattern of the SI changes during the growth stages in the formation of a monolayer as it does with BHK21JC13 cells, but the SI of transformed cells is much higher at any growth stage: 4-fold higher at confluence and 2-fold higher when the index is at its maximum. All 12 polyoma-transformed cell lines tested show such higher SI values. The results are shown in Table 3. BHK21/C13 cells always gave consistent results in repeated experiments, while the SI values of transformed cells, both from the same clone and between clones, showed greater variation but they were always well above that of BHK21/C13 cells.

The SI values of BHK21/C13/rf75 cells are also shown in Table 3. The morpho logy of this line is quite different from the original BHK21/C13 cells; it no longer has a typical fibroblast appearance and grows in suspension of soft agar like polyoma transformed cells. The SI of this line is very high at confluence, 6 times higher than that of BHK21/C13 cells, but the 15 h value is only 1·3 times that of BHK21/C13.

Sublines of BHK21/C13 have been isolated which have many properties of trans-formed cells but have never been in contact with polyoma virus (Montagnier, Mac-pherson & Jarrett, 1966). They emerged spontaneously during prolonged passage of ordinary BHK21/C13 cells. This does not necessarily rule out the possibility that transformation was induced by unrecognized viruses which were present in the cells, or through the cells becoming contaminated with unknown viruses during the passages. It is therefore possible that the transformed-like morphology of BHK21/C13/rif75 is not a consequence of adaptation to tolerate rifampicin. It is not known at present whether spontaneously transformed cells have a selective advantage in the presence of rifampicin.

In the previous section it has been shown that polyoma-transformed or spontaneously transformed BHK21/C13 derived lines have higher SI than the original line. It is of interest to test whether this is also true for other cell lines. Since most of the laboratory established permanent cell lines, including BHK21/C13, might possibly undergo transformation to some extent, mouse embryo primary culture cells were chosen as an example of normal cells. The growth curve and SI levels are shown in Fig. 5. Mouse embryo cells have a long lag phase of around 35 h. The SI of cells from confluent monolayers is 0·25; it then increases during the lag phase and reaches the maximum value of 4·0 at about 24 h. Shortly before the beginning of the exponential phase the index begins to fall gradually. This changing pattern and the actual value of the SI are very similar to those of BHK21JC13. In this respect therefore the BHK21JC13 cells retain the normal character. As expected, mouse L cells, which were established as a permanent line a long time ago, showed a quite different SI pattern. L cells start growing exponentially right after seeding with virtually no lag phase, although their generation time is almost identical with mouse embryo cells in the exponential phase (Fig. 6). At confluence they possess the characteristic high SI of transformed cells and the maximum value attained is slightly lower than that of the mouse embryo cells.

Two new cell lines have been established from mouse embryo cells in this laboratory by Dr J. F. Williams. Both lines came from the same batch of mouse embryo primary culture. During passage the C 3 line was always kept at low cell density while the C10 line was allowed to grow to high cell density. As a result of this selection procedure the C 3 line stops growing at low cell density, while the Cio line grows to form a thick monolayer (J. F. Williams, unpublished data). The SI values are shown in Table 4. The maximum SI of both lines are similar to mouse embryo primary culture cells or L cells. The C3 line shows a high SI at confluence comparable to that of L cells, while the C10 line is more similar to embryo primary culture cells. Hence by the criterion of SI the C3 line, which stops growing at very low cell density, behaves like transformed cell lines, while the C10 line, which forms a thick monolayer, behaves like normal cell lines.

In Table 4 data for various cell lines are shown. Primary embryo cultures as well as human WI-38 have low SI values. On the other hand those lines which have been maintained in laboratories for a long time show very high SI both at confluence and at 15 h.

The data presented in this paper indicate that normal cells grown to confluence are very sensitive to the cytotoxic effect of oxidized spermidine while polyoma virus transformed cells, adenovirus type 12 transformed cells or spontaneously transformed cells are considerably more resistant.

Both BHK21/C13 cells and embryo primary culture cells obtained from confluent cultures have a very long lag time before they enter exponential growth (Figs. 3, 5). This long lag time could be a consequence of their low SI at high cell density. The relationship between the growth curve and the SI curve might suggest that the growth of these cells is initiated only after the SI has attained a certain level. By contrast transformed cells or other permanent cell lines, which possess a higher SI to start with, grow virtually without any lag.

Since the mechanism of action of the cytotoxic agent produced by the oxidation of spermidine is not fully understood, it is difficult to know what kind of properties of transformed cells are associated with higher spermidine index. Changes in the cell surface have been postulated as the basis for the transformed phenotype (Abercrombie & Ambrose, 1962; Hakomori & Murakami, 1968). It has been reported that virus transformed cell lines react with an agglutinin, while under identical conditions their untransformed parent cell lines do not (Aub, Tieslau & Lankester, 1963; Burger & Goldberg, 1967). Since a short treatment with a low concentration of protease exposed agglutinin receptor sites in the parent cell line indistinguishable from the sites in transformed cells, it was proposed that tumour virus converts the membrane of the parent cells in such a fashion that the agglutinin receptor sites become exposed and available to the agglutinin (Burger, 1969). However, this explanation cannot account for the high SI because the levels reported were all obtained using trypsinized cells. Moreover, spermidine tests carried out on BHK21/C13 cells that were not trypsinized gave SI levels twice as high (author’s unpublished data).

A high SI of cells at confluence is associated with ability to grow in soft agar suspen sion. Polyoma transformed cells, BHK21/C13/rif75, L, HeLa and Hep-2 cells have been shown to grow in soft agar while primary embryo cells and BHK21/C13 cells do not grow under these precise conditions. However, BHK21/C13 cells can grow and form colonies in soft agar suspension, if the conditions are controlled so that a high SI is maintained. This can be achieved for example by inoculating large cell numbers (around 5 × 10⁵ cells per ml of soft agar medium), or by obtaining the cells from mono layers in lag or early lag phase, when the SI is highest: numerous colonies of BHK21J C/j cells are then observed in the agar after incubation for 7 days. Montagnier (1968) reported that a ‘conditioned medium’ obtained from confluent BHK21/C13 cultures helped the same cells to grow in agar suspension. My preliminary data suggests that a factor which neutralizes the cytotoxic effect of oxidized spermidine is present in the ‘conditioned medium’. Study of this factor is now in progress.

I am grateful to Professor J. H. Subak-Sharpe for his discussions and encouragement. I also thank Dr J. F. Williams for his suggestion to use mouse C3 and C10 lines. The excellent technical assistance of Mrs S. Robertson is thankfully acknowledged.

Figs. 7–10. BHK21/C13 cells treated with o (control), 0·5, 1·0, and 1·5 μg of spermidine, respectively. The photographs were taken after staining with Giemsa. In routine assays the cytotoxic effect was examined without staining, × 132.