Cloning of human lung cancer cells.

We have carried out a comparison of two different methods for cloning human lung cancer cells. The method of Courtenay & Mills (1978) generally gave higher plating efficiencies (PE) than the method of Carney et al. (1980). The number of colonies increased with incubation time in both methods and the weekly medium replenishment in the Courtenay method was advantageous for longer incubation times of several weeks. In the Courtenay method, the use of August rat red blood cells (RBC) and low oxygen tension were both found to be necessary factors for maximum plating efficiency. The usefulness of heavily irradiated feeder cells in improving PE is less certain; each cell type may have its own requirement.

We are interested in studying tumour cell heterogeneity in human lung cancer. The development of methods for cloning human tumour cells in agar (Courtenay & Mills, 1978;Hamburger & Salmon, 1977) has provided a means of selecting individual cell colonies and thereby establishing clonal subpopulations. However, using agarose instead of agar in their systems, Carney et al. (1980) and others  have shown that the plating efficiencies (PE) of human lung cancer specimens direct from the patient and early cell cultures derived from such specimens are usually < 1%. Such low PE may reflect a selectivity on clonal growth resulting in an unnecessarily narrow range of colony types. The aim of this study was to determine optimal cloning conditions which would promote the growth of as wide a range of colony types as possible. The methods of Courtenay & Mills (1978) and Carney et al. (1980) were used to examine the effect of various factors upon the clonogenicity of human lung cancer cells.

Materials and methods
The clonogenic assay methods used in these studies were those of Courtenay & Mills (1978) (the 'Courtenay method') and of Carney et al. (1980) (the 'Carney method').

The Courtenay method
The medium used in this method was modified Hams F12 supplemented with 15% foetal calf serum and with penicillin and streptomycin (all supplied by Gibco Biocult Ltd UK). Red blood cells (RBC) were obtained by cardiac puncture on August rats using preservative-free heparin, rinsed 3 Correspondence: G.A. Walls. Received 14 December 1984; and in revised form 1 July 1985. times with PBS and resuspended to the original blood volume in medium. The RBC suspension was then heated to 44°C for 1 h and stored at 4°C for up to one month. An 1/8 dilution in medium of the RBC was carried out immediately before use. A 6% solution of Agar Noble (Difco) was prepared in either distilled water or PBS and sterilized by boiling for 15 min. This was then diluted 1/10 in medium prewarmed at 44°C to give a final concentration of 0.6% and the solution kept at 44°C until required. Suspensions of the test cells in medium were prepared (as below) at 2.5 x the required final concentration and *kept at 37°C. Immediately prior to plating, 2.0ml of the test cell suspension was added to 0.5ml of RBC suspension followed by 2.5ml of 0.6% agar solution. Aliquots of 1ml of this suspension were then placed into each of 3 or 4 sterile plastic tubes with round bottoms (Falcon 2051). These tubes were allowed to stand in crushed ice until the agar set. They were then each gassed for 6s with a mixture of 90% N2, 5% 02 and 5% CO2 and the top of each tube 'snapped' closed. The tubes were placed in racks in plastic boxes which were then gassed with the same mixture for 10min before being sealed and incubated at 37°C in a walk-in warm room. After 7 and 14 days of incubation, 1 ml of medium was added to the agar plug in each tube and the tubes and boxes regassed. In those tubes where incubation continued for 4 or more weeks (i.e. in experiments with material other than established cell lines), 1 ml of medium was replaced with fresh medium at 21 days and weekly thereafter. At the end of the incubation period, the agar plug was tipped out from each tube into the inverted lid of a 5cm plastic petri dish. The base of the dish was then pushed down onto the plug so that the agar spread in a thin layer. Colonies containing >50 cells were counted under an inverted microscope. In some experiments, heavily-irradiated (100 Gy of 250kv X-rays) cells (HR) were included in the live cell suspensions. These HR cells were always of the same type as the live cells being tested. The gas mixture was also varied in some experiments in which tubes were gassed with either 95% air plus 5% CO2 or 95% N2 plus 5% CO2 as alternatives to the usual mixture. The effect of omitting RBC was also studied.
The Carney method The medium used in this method was RPMI 1640 supplemented with 20% heat-inactivated foetal calf serum and with penicillin and streptomycin (all supplied by Gibco Biocult Ltd. UK). A 5% solution of LGT Agarose ('Sea Plaque', Marine Colloids Inc. USA) in distilled water was prepared and sterilized by boiling for 15min. This was then diluted 1/10 in prewarmed medium (44°C) to give a final concentration of 0.5%, and 2ml aliquots of this agarose were pipetted into 35mm plastic petridishes (Falcon) and placed in the refrigerator to harden. A 6% solution of agarose in distilled water was also prepared, diluted 1/10 with medium to give a final concentration of 0.6% and kept at 44°C. Suspensions of the test cells in medium were prepared (as below) at 2 x the required final concentration and kept at 37°C. Immediately before plating, 2.5 ml of the test suspension was mixed with 2.5 ml of 0.6% agarose solution and aliquots of 1 ml were then pipetted into each of 3 or 4 of the 35mm petri dishes containing the agarose 'bottom layer' and recently removed from the refrigerator. The plates were placed into plastic boxes which were then gassed for 10 min with a mixture of 95% air and 5% C02 and incubated at 37°C in a walkin warm room, Plates were examined and colonies containing >50 cells were counted under the inverted microscope at 7-9, 12-16 or 21-22 days after plating. The agarose used in these studies is slightly different from the HGT agarose ('Sea Kem', Marine Colloids Inc.) used by Carney et al. (1980). A comparison was made between the PE of various cells obtained using these two types of agarose in the Carney method.

Preparation of cell suspensions
The established cell lines used in this study were NCI-H69, a small cell lung cancer (SCLC) line kindly supplied by Dr D. Carney; POC, a SCLC line kindly supplied by Dr M. Ellison; MOR, a lung adenocarcinoma line kindly supplied by Dr M. Ellison and COR-L23, a large cell anaplastic lung cancer line derived in our own laboratory. In addition, we used samples of cells derived from bone marrow trephine (COR-L31, COR-L32, COR-L42 and COR-L65) and lymph node (COR-L24 and COR-L47) specimens from patients with SCLC and maintained in vitro in our laboratory. These cells have been subsequently characterized as SCLC except COR-L65 which has been recently found to be B-lymphoblastoid (Baillie-Johnson et al., 1985).
All cell lines were maintained in vitro in RPMI 1640 supplemented with 10% FCS and with penicillin and streptomycin in either 75cm2 flasks or 500ml bottles (Techne). These were usually fed twice weekly and passaged when needed. Single cell suspensions were prepared from all of these lines by treatment for 15min with 0.4% trypsin and 0.2% versene in PBS and subsequently either pipetting or pulling the suspension in and out of a 10 ml syringe fitted with a 25 gauge needle. If any clumps were seen, the suspension was passed one or more times through tightly packed cotton gauze in a sterile glass funnel and re-examined. The pleural effusions (COR-L51 and COR-L90) from patients with SCLC were cleared of RBC by centrifugation on Ficoll (Pharmacia), passed through cotton gauze and pipetted. Routine pathological examination reported COR-L90 to be positive for tumour cells while COR-L,51 was reported to be negative. However, a cytospin preparation of COR-L,51 examined by Dr A Gazdar of the NCI (Navy Medical Oncology Branch, National Cancer Institute, USA) was reported to contain one clump of malignant cells and we have established a SCLC line from the effusion (Baillie-Johnson et al., 1985). The clinical samples of normal bone marrow (NBM 1-6) were obtained from hip-replacement specimens and cell suspensions prepared by agitation in medium containing neutral protease (Sigma Type IX) at 1 mg ml -1 and then cleared of RBC by centrifugation on Ficoll. These normal bone marrow specimens were used as a control for specimens from patients with lung cancer. Cloning of COR-L51 and normal marrow specimens was carried out on aliquots stored in liquid N2 and thawed immediately before use. These specimens were stored in liquid N2 as a matter of convenience in an attempt to maintain the cells in a state as close as possible to the source. A bone marrow trephine sample from a patient with squamous cell lung cancer (COR-L93) was agitated in medium containing neutral protease at 1 mg mland the RBC removed by centrifugation on Ficoll. Two bone marrow aspirate samples from patients with SCLC (COR-L91 and COR-L92) were centrifuged on Ficoll only. Histological examination by the hospital pathology department reported COR-L91 to be positive for tumour cells whilst COR-L92 and COR-L93 were both reported as negative. These 3 bone marrow samples and the pleural effusion COR-L90 were used directly in the Courtenay method and not stored in liquid N2 or maintained in vitro before use.
Xenografts of NCI-H69, COR-L23, COR-L24, COR-L31, COR-L32, COR-L47 and COR-L51 were grown intramuscularly in the gastrocnemius muscle of the hind limbs of MF1 nu/nu mice. They were excised when 0.5 to 1.0g in weight. Tumours were disaggregated to a single cell suspension by mincing with scissors, agitating in medium containing neutral protease at 1 mg ml-1 for 2 h and filtering through cotton gauze. The yield of viable nucleated cells as determined by phasecontrast microscopy was 3-6 x 106 per tumour.

Comparison of clonogenic assays
Data for experiments comparing the two clonogenic methods are shown in Table I. The results of 2 experiments are shown so that the data in the left hand column for the Courtenay method is from the same experiment as the data in the left hand column for the Carney method. For the established cell lines, 4 x 102 and 2 x 103 viable cells were plated in both methods whereas for the recent in vitro specimens 4 x 103, 2 x 104 and 105 cells were plated. For samples direct from the patient, 105-106 cells were used in both methods. In all tables and figures, PE is expressed as the number of colonies counted as a fraction of cells plated; values are the means of triplicate cultures. A higher PE was found in the Courtenay method for all cell specimens tested except for 3 of the 6 normal bone marrow samples. Also, the colonies in the Courtenay method were generally larger and contained more cells.  Established cell linesa Small cell NCI-H69 4.6 x 10-1 3.9 x 10-1 1.9 x 10-1 8.5 x 10-3 POC 7.1 x 10-1 6.1 x 10-1 3.0 x 10-1 3.8 x 10-2 Large cell COR-L23 l.9x10-1 2.1x10-1 9.2x10-2 3.3x10-2 Adenocarcinoma MOR 6.8 x 10-2 1.1 x 10-1 5.2 x 10-2 1.7 x 10-3 Recent in vitro SCLC specimensb COR-L32 2.7 x 10-1 9.1 x 10-2 2.6 x 10-4 2.4 x 10-4 COR-L24 7.0 x 10-3 5.4   for cells of the NCI-H69 small cell line. Also, low 02 tensions produced higher PE than were seen when using 20% 02. Cells of the 2 non-small cell lines showed much less dependence on RBC but for each of them no RBC and high 02 tension was a particularly bad combination. A repeat experiment (not shown) generally confirmed these results. Results for the NCI-H69 xenograft cells were similar to those of the in vitro line, but the COR-L23 xenograft cells showed a greater dependence upon RBC. The importance of RBC in the cloning of NCI-H69 cells was investigated in one experiment (not shown). Increasing the dilution of RBC from 1/8 to 1/64 gradually reduced the PE from 0.72 to 0.36. In the absence of RBC, the PE was 0.30, about 40% of that obtained using the standard 1/8 dilution of RBC.
The effect of NCI-H69 HR feeder cells upon the PE of 3 x 102 cells of the NCI-H69 line in the presence or absence of RBC is shown in Figure 3. In the presence of RBC, there was no requirement for the feeder cells; in fact at 106 HR cells, the PE decreased presumably due to medium depletion. In the absence of RBC, 105 HR cells raised the PE almost to the level seen with RBC. A repeat of this experiment confirmed this effect of HR cells but showed a less dramatic effect of RBC. Table II shows the effect of 105 HR cells on the PE of cells derived from xenografts of 4 SCLC cell types when plated with RBC. Control cultures contained no HR cells. The PE of COR-L24, COR-L31 and COR-L32 was improved by the presence of HR cells while that of COR-L51 was not significantly  affected. Unlike the other cell types, COR-L24 has a PE that is both low and very variable. Other experiments (data not shown) have demonstrated that for this cell type, there is a non-linear relationship between the number of cells plated and colonies formed. For all cell types, in the absence of HR cells, there were many small clusters consisting of < 50 cells which were not scored.
In the description of their method, Courtenay & Mills (1978) used distilled water as a solvent for agar but more recently Courtenay (1983) describes using PBS for this purpose. Therefore, we compared PBS with distilled water as a solvent for the agar in the cloning of COR-L24 and COR-L32. For both cell cultures, PBS appears to give a PE 2-3 times higher than that produced when using distilled water (results not shown).
Also, Courtenay (1983) suggests an incubation period of 4 weeks. However, this length of time was sub-optimal for specimens from the patient which we found to form only small colonies of usually <50 cells after 4 weeks in culture. Table III shows the effect of incubation time on the PE of 3 bone marrow specimens and one pleural effusion specimen (104-105 cells plated) used directly from patients with SCLC and cells from xenografts of 3 recent in vitro cultures (103-105 cells plated). In general, for all the specimens tested, PE increases with incubation time up to 8 weeks except for COR-L92 and COR-L93 where there were essentially no colonies at any time. The lack of colonies formed from COR-L92 and COR-L93 indicate the absence of tumour cells in these specimens and thus confirmed the diagnosis based upon histological preparations of these cells. In the bone marrow specimen that was reported positive for tumour cells (COR-L91) and in the pleural effusion specimen (COR-L90) there was a considerable increase in PE between 4 and 6 weeks suggesting that an extended period of incubation may be necessary for specimens cloned directly from the patient.
The effect of incubation conditions on the PE of COR-L24 and COR-L32 (105 cells of each plated) in both the Courtenay and the Carney methods was investigated. Identical sets of Courtenay assay tubes were placed into 2 plastic boxes each gassed with a mixture of 90% N2, 5% 02 and 5% CO2 for 10 min and then sealed. One box was subsequently incubated at 37°C in a walk-in warm room. The other box was incubated at 37°C in an incubator gassed with 95% air and 5% C02' Identical sets of Carney assay dishes were placed into plastic boxes. One box was gassed with a mixture of 95% air and 5% C02, then sealed and incubated at 37°C in the walk-in warm room whilst the other box was left unsealed in the CO2 incubator. Colonies in the Carney method were counted 21 days after plating. The results are shown in Table IV. The PE of both COR-L24 and COR-L32 were higher when Specimens direct from the patient COR-L90 2.0 x 10-4 6.7 x 10-5 1.8 x 10-3 2.5 x 10-3 3.3 x 10-3 COR-L91 9.0 x 10-4 7.4 x 10-3 2.0 x 10-2 1.0 x 10-2 ND COR-L92 ND 0.0 0.0 0.0 ND COR-L93 ND 1.0x 10-5 2.0x 10-5 3.0x 10-6 ND Xenografts  3.l x 10-l 4.3 x 10-6.7 x 10-I ND ND COR-L24 2.9 x 10-3 5.0 x 10-3 4.8 x 10-3 8.6 x 10-3 ND COR-L32 7.5x 10-3 2.4x 10-2 1.7x 10-2.4x 10-ND ND, not done.  ('Sea Kem') in their studies but we found this to be less convenient to use than a low gel temperature agarose ('Sea Plaque'). In one experiment however, the 2 types of agarose were compared in the cloning of COR-L32 and COR-L65 cells (105 cells of each plated). The effect on PE of PBS and distilled water as solvents for the agarose was also examined. Colonies were counted 14 days after plating and the results are shown in Table V. For both cell samples, 'Sea Plaque' agarose produced higher PE than 'Sea Kem' agarose. There appears to be little difference in PE between PBS and distilled water when used as solvents for 'Sea Plaque' agarose and only a slight increase in clonogenicity of COR-L65 when PBS was used with 'Sea Kem' agarose. However, the combination of 'Sea Kem' agarose and PBS produced larger colonies than did the combination of 'Sea Kem' agarose and distilled water. There was no difference in the appearance of colonies produced between PBS and distilled water when used in combination with 'Sea Plaque' agarose.
In the Carney method, colonies rarely appeared before day 7 of incubation and because no replenishment of the medium is carried out in this method, experiments were generally scored before day 21. During this period of incubation the PE of most cell specimens increased (Table VI). Cultures of all specimens contained small clusters of <50 cells which were not scored. In the case of NCI-H69, these small clusters were scored separately from the usual colonies. The ratio of colonies:clusters was, on day 9 of incubation, 0:33; on day 12, 7:94; and on day 15, 17:72. It seems likely that by day 12, all cells destined to produce small clusters would have done so; some of these growing to consist of more than 50 cells during the interval from day 12 to day 15 and therefore scored as colonies not clusters.

Discussion
The PE was higher in the Courtenay method than in the Carney method for all lung cancer specimens tested. The difference between the 2 methods was most noticeable when comparing the PE of early culture specimens suggesting an advantage of the Courtenay method for specimens of very low PE. In both methods the PE of the established cell lines was generally much higher than those of the early culture specimens. This finding is not surprising as established cell cultures consist of a more homogeneous population of replicating cells and therefore may contain a higher proportion of clonogenic cells than do early cell cultures. Carney et al. (1980) have shown that the PE of various fresh specimens containing SCLC cells is -1% of SCLC cells plated but only 0.02 to 0.25% of the total nucleated cells plated whilst the PE of 2 established cell lines was 1.0 to 5.6%. Interestingly Gazdar et al. (1980), using the Carney method, demonstrated that, of the established cell lines plated, the non-small cell lung cancer lines gave much higher PE (28 to 44%) than did the SCLC lines (0.03 to 5.2%). We have found the PE of our non-small cell and SCLC lines to be very similar using the Courtenay method. It appears that normal bone marrow specimens can form colonies giving rise to similar PE in both methods. These results are somewhat surprising as although it is well established that mouse bone marrow can give rise to both macrophage and granulocyte colonies in agar, the formation of these colonies depends upon the presence of a colony stimulating activity (Bradley & Metcalf, 1966;Metcalf et al., 1967;Bradley et al., 1971). Using the Carney method, Carney et al. (1980) reported no colonies being formed from 7 bone marrow specimens histologically negative for SCLC. Although the PE of our normal bone marrow specimens in the Carney method was low, they still gave large colonies consisting of >50 cells that ranged in number from 2 (NMB 4) to 20 (NBM 3) for 105 cells plated. The colonies formed from normal bone marrow were of 2 distinct types (1) diffuse groups of cells (not scored) and (2) tightly packed clusters of cells similar to colonies produced by SCLC specimens. Colonies of the latter type will make it difficult to select colonies from bone marrow specimens with SCLC infiltration if the specimen is cloned direct from the clinic.
There are several reasons which may explain why the Courtenay method produces higher PE than the Carney method. The two most critical factors appear to be the presence of rat RBC and low 02 concentration. Bradley et al. (1971) first demonstrated that washed rat RBC improved the clonogenicity of mouse marrow cells and produced larger colonies. RBC lysed before use were found to be equally effective as whole cells in promoting colony growth suggesting the release of a growth factor. However, using human tumour xenografts, Courtenay & Mills (1978) found that RBC lysed before use were less effective than whole cells and that the time of lysis in culture is important. RBC from August rats were superior to RBC from other strains of rat because they lysed over a period of 5-7 days, a critical time for colony growth. Also, the PE of various cell types seems dependent on an optimal number of RBC (Bradley et al., 1971;Courtenay & Mills, 1978;Courtenay, 1983). We found that there is a general requirement for rat RBC although the 2 non-small cell lines (COR-L23 & MOR) appear less dependent upon them than the NCI-H69 cell line. Indeed, NCI-H69 cells seem very sensitive to the concentration of RBC since the PE was found to be gradually reduced as the RBC were diluted from 1/8 to 1/64. In the absence of RBC, the PE of these cells was < 50% of the PE at the lowest dilution.
Recently, Tveit et al. (1981a;1981b) showed that when cloning melanoma xenografts 5%°2 produces higher PE than does 20% 02. In fact, many different tumour cell types appear to have better PE at 02 concentrations of 5% or less (Richter et al., 1972;Courtenay, 1976Courtenay, , 1985Gupta & Krishan, 1982). Using the clonogenic assay of Hamburger & Salmon (1977), Gupta & Eberle (1984) demonstrated that although there was great variation between the PE of cells from the various human xenograft lines tested, the optimal concentration could be as low as 0.1 %. These reports suggest that the most effective 02 tension for improving the PE of cells in the clonogenic assay is in the range of 0.1-5%. Our data confirm these findings and show that gassing with 02-free N2 produces PE for several lung cancer cell types that are equal to or better than those obtained when 5% 02 was used. Although there may have been some residual 02 present after gassing, perhaps dissolved in the agar plug and/or plastic tube used in the assay method, it is unlikely to have been greater than 1 % and therefore would not have influenced the results. Courtenay (1983) recommends the use of HR cells to make up the number of cells plated to 104.
The function of HR cells is unknown but may be related to the production of growth factor(s) or the the consumption of 02 which would help to maintain a low 02 environment suitable for the formation of colonies. Although cells in culture derived recently from clinical specimens show an improvement in PE with HR cells, those from the well-established cell line, NCI-H69 do not. Also, in the absence of HR cells, most cell growth arising from recently derived cells is in the form of small clusters consisting of <50 cells. Thus it appears that cells which have been growing in culture for extended periods of time at elevated 02 levels may be less dependent on a low 02 concentration in the clonogenic assay than are cells more recently derived from tumours which may require an in vitro 02 environment more similar to that found in vivo. Factors influencing PE in both the Courtenay and Carney methods include incubation conditions and incubation time. It seems that an isotonic agar solution is advantageous for plating cells in the Courtenay method as agar prepared in PBS produced higher PE than agar prepared in distilled water. In the Carney method, there was no clear advantage of using PBS to prepare agarose. However, the use of 'Sea Plaque' agarose resulted in higher PE than was seen with 'Sea Kem' agarose regardless of the solvent used. Surprisingly, in both methods, PE were much higher in those cultures incubated in a walk-in warm room than in those incubated in the CO2 incubator. If there was a degassing effect on cell growth in the warm room, one would expect the opposite results. Although there is no apparent reason for the PE to be higher in cultures in the warm room, such variation in cell growth has been known to occur. Buckmeir et al. (1984) have observed large variations in clonogenic cell growth between replicate cell plates placed in several different gassing incubators. These authors suggest that the different PE may be related to differences in water loss from the agar which was found to occur during incubation.
Prolonged incubation periods in the Courtenay method were necessary for specimens cloned direct from the patient and for 1 of 3 xenografts of early culture specimens (COR-L32). For all specimens except COR-L92 and COR-L93 which were diagnosed negative for tumour cells, there was a considerable increase in PE between 4 and 6 weeks. In one case (COR-L90) the clonogenicity was still increasing at 12 weeks. The Courtenay method has a particularly useful advantage over the Carney method as it is possible to replensih the medium and thus maintain slow-growing colonies over long periods of time. In fact the weekly replenishment of the medium over the usual 3 to 4 week period may be responsible for the generally higher PE seen routinely with the Courtenay method. It is apparent that, for the cells studied, an incubation period in excess of 9 days is required for colony formation in the Carney method. Even NCI-H69 plated at 5 times the number of cells used in the Courtenay method failed to produce colonies during the first week. Although 4 specimens showed a continual increase in PE up to 3 weeks, the cultures could not be maintained longer than 3 weeks because in the Carney method, the agarose is not replenished.
In conclusion we have shown that the Courtenay method provides higher PE than the Carney method for human SCLC cells including specimens direct from the patient, early cell cultures and established cell lines. The improved clonogenicity in the Courtenay method is due primarly to the presence of RBC and low 02 tension although the replenishable nature of the assay allows for extremely long incubation periods often necessary for specimens cloned directly from the patient.