Distinction of two different classes of small-cell lung cancer cell lines by enzymatically inactive neuron-specific enolase.

Neuron specific enolase (NSE) is widely used as a neuro-endocrine marker. However the presence of NSE in many non-neuroendocrine tissues has raised questions on the specificity of NSE. We have investigated NSE immunoreactivity (NSA-ag), gamma-enolase activity and total enolase activity in small cell lung cancer (SCLC) cell lines. During well-controlled exponential growth comparison of NSE-ag content and gamma-enolase activity with the doubling-time (Td) and NSE-ag content with gamma-enolase and total enolase activity led to a clear distinction of two types of cell line: variant cell lines plus part of the classic cell lines (type I) and the remaining classic cell lines (type II). The distinction was based upon both an abrupt 6-fold increase of gamma-enolase activity and an 18-fold increase of NSE-ag, which for the larger part was enzymatically inactive. Within each group the increase of NSE-ag content was significantly correlated with the increase of gamma-enolase activity and both NSE-ag content and gamma-enolase activity increased linearly with Td. It is concluded that gamma-enolase seems to be associated with the regulation of growth rate and that a compound with the gamma-enolase antigen but without enzyme activity can distinguish two different classes of SCLC cell lines. Furthermore the demonstration that NSE-ag can represent the active enzyme as well as an enzymatically inactive compound may explain why a controversy about neuron- or non-specificity of NSE exists.

Since the beginning of the eighties numerous continuously growing cell-lines from Small Cell Lung Cancer (SCLC) biopsies have been established. Investigators at the National Cancer Institute (Bethesda, USA) were the first to distinguish two types of cell-lines i.e. variant and classic cell-lines, characterised by the absence or presence respectively of the enzyme L-dopadecarboxylase. In comparison with classic cell lines variant cell lines were shown to have a higher growth rate, a higher cloning efficiency, a larger cell volume, a lower content of Neuron-Specific Enolase (NSE), amplification of c-myc, absence of gastrin releasing peptide (GRP) and neurotensin, and a decreased sensitivity to radiotherapy and chemotherapy (Bepler et al., 1987;Carney et al., 1985;Gazdar et al., 1985;Bepler et al., 1989a;Broers et al., 1985;Moody et al., 1985;Broers et al., 1988). Recently Bepler et al. (1989b) added a third class, so-called transitional cells, based on the presence of p64 c-myc in some of the classic cell lines. The addition of a third class of cell lines, is substantiated by intermediate levels of neuroendocrine markers, growth rate and cloning efficiency. NSE is widely used as a neuroendocrine marker. NSEimmuno reactivity is not only seen in neurons, but also in neuroendocrine cells present in endocrine glands and in the diffuse neuroendocrine systems of the lung, intestine, thymus and skin. NSE has been demonstrated in tumours, thought to arise from the neuroendocrine cell system, such as SCLC, neuroblastoma, carcinoid, pancreatic islet cell tumours and medullary thyroid carcinoma (Schmechel et al., 1978;Tapia et al., 1981;Wick et al., 1983). NSE is also present in erythrocytes, lymphocytes and platelets (Marangos et al., 1980b;Hullin et al., 1980) and in malignant lymphomas, testicular cancer, hypernephroma and non-small cell lung cancer (Takashi et al., 1989;Ariyoshi et al., 1983;Kuzmits et al., 1987;Pinto et al., 1989;Oka et al., 1989;Niehans et al. 1988). Such observations question the correlation between NSE and the neuroendocrine cell system (Schmechel, 1985). We  immunoreactivity and enolase-enzyme activity in SCLC-cell lines. It was found that immunoreactive NSE can represent the active enzyme y-enolase as well as an enzymatically inactive compound. The active enzyme was linearly correlated with the growth rate and the presence of the inactive compound distinguished two different classes of SCLC-cell lines.
Sample preparation A cell pellet, containing 1-3 x 106 cells, was obtained by centrifugation for 10 min at 250g. In order to remove dead cells and cell debris the pellet was washed once with PBS and treated with 1 ml of 0.05% Trypsin-0.02% EDTA (Flow Laboratories) for 3 min at 37°C. To inactivate trypsin culture medium supplemented with 10% foetal calf serum was added. Then DNAse (Sigma DN-25) was added to a final concentration of 0.1%. After mixing an aliquot of the single cell suspension was used for cell counting in a haemocytometer. The remaining cells were centrifuged at 800 g and the pellet was frozen at -70°C. After thawing the cell pellet was suspended in 0.5 ml of 50 mM Tris-HCI buffer pH 8.0 containing 100 mM KCI, 10 mM MgCI2, 2 mM dithiotreitol and 100 mM sucrose. After centrifugation for 10 min at 800 g the supernatant was used for measuring neuron specific enolase 'PI Macmillan Press Ltd., 1992Br. J. Cancer (1992), 66, 1065-1069 immunoreactivity, enolase enzyme activity, enolase isoenzyme composition and protein content. Neuron specific enolase immunoreactivity NSE-immunoreactivity (NSE-ag) was determined with the Pharmacia NSE-RIA, as previously described (Cooper et al., 1985).
Enolase isoenzymes Enolase isoenzymes were separated on cellulose acetate gel (Cellogel) in 20 mM sodium-phosphate buffer pH 7.0 and enolase activity as determined as described previously (Oskram et al., 1985).
L-dopa decarboxylase L-dopa decarboxylase (aromatic L-amino acid decarboxylase; ALAAD) was measured as described previously (Boomsma et al., 1986) and expressed as mU 10-6 cells. For measuring ALAAD activity a frozen cell pellet, containing a known number of cells, was dissolved and further diluted when necessary in bidistilled water containing 40 g I`of bovine serum albumin and 10gl-l of glutathione.
Purification of NSE and production of antiserum Human NSE was purified from postmortem human brain cortex and the antiserum was produced in rabbits (Haglid et al., 1973). This antiserum was extensively absorbed with human non-neuronal enolase until the resulting rabbit anti-NSE did not show any crossreactivity with a-enolase in an ELISA (Aurell et al., 1989). differences between the means of two groups was tested by the Student's t-test. When a P-value was less than 0.05 the difference was regarded significant.
NSE-immunoreactivity (NSE-ag), total enolase-and y-enolase enzyme activity As shown in Figure 2 the ratio between NSE-ag content and total enolase activity, measured in the same cell lines of the paragraph above, also distinguished the same two types of cell lines with the exception of GLC 1-13 and 16 (open squares). The mean ratio for type I cell lines was 635 ± 144 ng U ' and for the right group 2994 ± 350 ng U-' (P<0.0001). These data indicated the existence of an abrupt increase of NSE-ag without enzyme activity between the two types of cell lines. The ratios for the two exceptional cell lines were 1193 and 1294 respectively. Both cell lines had NSE-ag contents in earlier passages consistent with type II cell lines. Six and nine passages later the Td had decreased from 56 to 35 h and from 60 to 39 h respectively at which time the ratio Immunoblotting with anti-NSE Samples of the various SCLC-cell lines were sonicated (Branson, Sonifier, cell disruptor B15) in 1% SDS at 90°C until a clear solution was obtained, usually no longer than 60 sec. (Wang et al., 1990). Samples of sonicated SCLC-cell lines were run in SDS gel electrophoresis, using a 5-10% linear polyacrylamide slab gel. The proteins were transferred to a 0.45 ItM nitrocellulose membrance according to Towbin et al. (1979) except that 0.1% SDS was added to the transfer buffer. The electrophoretic blots were detected using rabbit anti-NSE serum (absorbed xix) (diluted 1:500 in TRISbuffered saline) as the first antibody and peroxidaseconjugated goat anti-rabbit IgG (diluted 1:200) as the second antibody Diaminobenzidine (0.5 mg ml-' in TBS) and H202 (0.03%) were used as the enzyme substrate for the colour reaction.

Statistical methods
The date are presented as the mean ± the standard deviation of the mean. Correlation coefficients were estimated by using a simple linear regression analysis. The significance of the  Figure 2, were due to a mixture of type I and II cell lines on their way to type I cell lines. In order to investigate this hypothesis a type II sample of GLC-16 with a Td of 60 h and a type I sample of GLC-16 with a Td of 28 h were cultured as a 1:1 mixture. At the start of the mixed culture the enolase/NSE-ag ratio showed an intermediate value. At each passage the Td and enolase/NSEag ratio decreased until after 5 passages a Td of 30 h and an enolase/NSE-ag ratio of 700 was reached and remained constant thereafter (data not shown). These data support the existence of an abrupt instead of a gradual change of NSE-ag without enzyme activity. In a separate experiment with 24 different passages of 8,11,14,19,28 and 34 the percentage of enzymatically active oaa, cry, 'y chains was measured. The percentage of enzymatically active y-chains was calculated by adding half of the percentage of ay-isoenzymes and two times the percentage of yy-isoenzymes. NSE-ag was determined in only 13 of these cell lines. The ratio between y-enolase activity and NSE-ag level again distinguished two types of cell lines, which were further characterised by a significant difference of the mean enolase activity, mean NSE-ag content and mean Td and an abrupt increase of NSe-ag without enzyme activity between both types (data not shown). In addition the mean y-enolase activity of the two groups was 33.5 ± 29 mU mg-' protein and 150 ± 45 mU mg' protein (P<0.0001) respectively and the mean percentage of enzymatically active 'y-chains 4.2 ± 2.3% and 25.9 ± 3.6% (P<0.0001) respectively. As shown in Figure 3 a significant correlation was found between the log y-enolase activity and log NSE-ag content in both types (y = 1.39 x -1.1 1, r=0.94, P=0.001 for type I; y= 1.28 x-1.83, r=0.97, P= 0.005 for type II). Interestingly the slope of both regression lines was almost identical. These data indicate that the logarithmic increase of y-enolase activity is significantly correlated with the logarithmic increase of NSE-ag content in an almost identical way in both types of cell lines but at a different NSE-ag level. The NSE-ag level shows an abrupt increase without enzyme activity between both types. The percentage of enzymatically active 'y-chains in type I cell lines was significantly correlated with the y-enolase activity. However no such correlation was observed in type II cell lines (Figure 4).

Immunoblotting
An immunoblotting assay with polyclonal rabbit antibodies against human --enolase was performed on extracts of a variant (GLC-2), a transitional (GLC-8) and a classic (GLC-34) cell line. Purified human yy-enolase and human brain cortex grey matter were used as references. All cell lines

Discussion
The relevance of the distinction between classic, transitional and variant cells in vitro has not been demonstrated in vivo. This is mainly due to the heterogeneity of the tumour and difficulty to obtain representative and sufficient tumour material from patients. Therefore our attention focused on tumour markers as a possible source of information about the composition of SCLC-tumours in vivo. Neuron-specific enolase (NSE) is one of the most widely used tumour markers in SCLC. It has been shown to be a clinically reliable tool to monitor the course of the disease (Cooper et al., 1985;Splinter et al., 1987a;Splinter et al., 1989), the doubling-time of NSE at relapse was highly significantly correlated with survival from time of relapse (Splinter et al., 1987b) and pretreatment-values were shown to have prognostic value by some (J0rgenson et al., 1988), but not by others (Van der Gaast et al., 1991). Moreover NSE is a good marker for neuronal differentiation and maturation (Schmechel et al., 1980). However the presence of NSE in many non-neuronal and non-neuroendocrine tissues together with a lack of understanding how a relatively unimportant glycolytic enzyme might play such a distinct role in differentiation raised the question whether NSE was not neuronspecific but nonspecific (Schmechel, 1985). Therefore we started to investigate the relationship between NSE and biological behaviour in SCLC-cell lines. Two types of cell lines could be distinguished by significant differences of NSEag content, total enolaseand y-enolase enzyme activity, Td and an abrupt appearance or increase of NSE-ag without enzyme activity, which was reflected in significantly different NSE-ag/enolase ratios and NSE-ag/y-enolase ratios. In both types the NSE-ag content and y-enolase activity (data not shown) were linearly correlated with the Td, albeit at two different levels. It is therefore concluded that y-enolase activity seems to be associated with the regulation of growth rate. In addition in type II cell lines, but not in type I cell lines, a dissociation was observed between the percentage of enzymatically active y-chains and y-enolase activity. This may indicate that in type II cell lines either the production of enzymatically active y-chains is dissociated from the production of a-chains or active y-chains are inactivated by a mechanism, which acts independent of the regulation of production. Recently it was shown that V-src could induce phosphorylation of glycolytic enzymes, such as enolase and especially -y-enolase (Cooper et al., 1983). Phosphorylation of y-enolase led to partly inactivation of the enzyme, accompanied by an increase of the total amount of the enzyme (Eigenbrodt et al., 1983). Moreover c-src expression in neuroblastoma-and SCLC-cell lines correlated with neuroendocrine differentiation (Mellstrom et al., 1987) and c-src is connected to neurogenesis and neuronal differentiation (Brickel et al., 1991), as is the switch from a to y-enolase (Schmechel et al., 1980). In this connection it is very interesting that Wevers et al. (1988) found in cerebrospinal fluid but not in serum from healthy individuals that 50% of NSE-ag had no enzyme activity. These data suggest that NSE-ag without enzyme activity may arise from inactivation of y-chains and may be correlated with neuronal or neuroendocrine differentiation. However, it is also possible that inactivation of y-enolase by phosphorylation merely reflects protein kinase activity, leading to different changes in different cell types. With polyclonal rabbit antibodies against yenolase we could not demonstrate the presence of a compound which migrated differently from y-enolase. Whatever the explanation, characterisation of NSE-ag without enzyme activity, investigation of the regulation of its production and of the production of active 'y-enolase may produce more information about the growth rate and differentiation of SCLC-cell lines.
The data presented in this paper support the distinction of two new classes of SCLC-cell lines, type I and II, with different biological characteristics. Possibly the classic cell lines, belonging to type I, are similar to the transitional cell lines, characterised by the presence of p64 c-myc (Bepler et al., 1989b). Whether such a distinction in vitro has any relevance in vivo should be and possibly can be investigated by measuring the ratio between NSE-ag and y-enolase activity in serum samples from patients with SCLC. A comparison of enzyme activity and immunoreactivity by Sorensen et al. (1988) in serum samples from five SCLC-patients showed that at least in some samples the amount of NSE-ag and y-enolase activity were significantly different.
Finally, the demonstration that 'neuron-specific enolase' measured with an antibody, can be the enzyme NSE or an enzymatically inactive compound emphasises that further investigations about neuron-specificity of NSE needed. It may be that NSE immunoreactivity in neuroendocrine cells is different from the one in non-neuroendocrine cells.