Cell, Vol. 8, 529-533,
August
1976,
The Selection Ovary Cells
Copyright01976
by MIT
of Virus-Resistant
Robert Taber, Veronica Alexander, and Norman Wald, Jr. Department of Medical Viral Oncology Roswell Park Memorial Institute and Graduate Faculty in Microbiology State University of New York at Buffalo (Roswell Park Division) Buffalo, New York 14263
We have selected Chlnese hamster ovary (CHO) cells resistant to infection by encephalomyocardltls (EMC) virus. Thus far, we have obtained five lines resistant to EMC, all of which manifest dlfferent phenotypes. Three of the five are not perslstently Infected with virus, while two lines produce infectious vlrus and grow in Its presence. The nonpersistently Infected lines exhibit different reslstance profiles to the other viruses we have tested, and they are stable in nonselective growth conditions. Their resistance appears to be due to a genetic alteration In the cell.
Virus resistance has long been used for selection of bacterial mutants. It was among the markers used in early studies on the nature of bacterial mutations (Luria and Delbruck, 1943; Lederberg and Lederberg, 1952). More recently, the virus resistance phenotype has been used to detect bacterial genes necessary for phage growth. Best characterized of these are three classes of E. coli genes which affect lambda phage RNA synthesis (gro-N), DNA synthesis (gro-P), and particle assembly (gro-E) (Georgopoulus, 1971; Georgopoulus and Herskowitz, 1972; Georgopoulus et al., 1973). The use of virus resistance as a marker in studies with animal cells has long presented an attractive possibility; one should be able to select useful mutants with defects similar to those described for bacteria. However, such apparently straightforward studies have previously proved to be complicated by three basic problems (Vogt, 1958; Vogt and Dulbecco, 1958; Morgan, Colby, and Hulse, 1973). The first of these problems was the lack of knowledge concerning the molecular mechanisms of viral replication. Early attempts to isolate such cells were frustrated in part by the inability to study the nature of the resistance in molecular terms. An example of this was the isolation of a HeLa cell line seemingly resistant to poliovirus at the level of penetration (Darnell and Sawyer, 1960). At the time such cells were isolated, it was impossible to study such cellular properties as protein composition of the
Chinese
Hamster
membrane, or to examine cellular submembrane structural components such as microtubules that may be involved in virus penetration (Luftig and Weihing, 1975). The second impediment to the study of virusresistant eucaryotic cells was the lack of genetically defined cultured mammalian cells. Only in the past few years have marked cell lines become available that would allow genetic studies of the defective cells to identify individual complementation groups. Chinese hamster ovary (CHO) cells in particular have their genetic characteristics (for example, mutation frequency, genetic stability, complementation, and so on) more thoroughly characterized than other cell lines (Thompson and Baker, 1973; Puck et al., 1971). The third problem has been the development of persistent infection during attempts to select for virus resistance (Vogt and Dulbecco, 1958; Morgan et al., 1973) and the lack of quantitative selection. While such infections certainly have inherent interest, they also create difficulties when one attempts to perform cell fusion experiments or analyses of a biochemical defect. A number of such persistently infected systems have been reported (Preble and Younger, 1975; Walker, 1964; Taber, Alexander, and Whitford, 1976). In these situations, one often finds that resistant cells arise only after several cycles of infection and reinfection, making quantitative selection frequency analysis difficult if not impossible (Morgan et al., 1973). In this paper, we describe a simple system using CHO cells and encephalomyocarditis (EMC) virus that apparently results in the selection of virusresistant cells at frequencies similar to those obtained by other selective agents. It appears that the system overcomes the difficulties described above since the molecular biology of EMC virus is well understood, the cell genetics and molecular biology of CHO cells is relatively well advanced, and a significant proportion of the resistant cells obtained are not persistently infected.
Results The most straightforward approach to the selection of virus-resistant cells is simply to infect the cells with virus at various multiplicities of infection. EMC virus was added at the input multiplicities shown in Table 1 A, and the infection was allowed to proceed. Observation of the cells with phase-contrast microscopy revealed extensive cytopathic effects l-3 days after infection with the time of initial appearance of cytopathic effect related to the multiplicity of infection. By the third day after infection, extensive scanning by phase-contrast microscopy of the flasks did not reveal obviously viable cells.
Cell 530
At this time and subsequently until staining, the titer of infectious EMC in the culture fluid was never less than 106 PFU/ml, and the titers from mutagenized and nonmutagenized samples were identical. At 3 weeks post-infection, the plates were stained and colonies counted. From the time of infection until staining, the flasks were not opened. The first column of Table 1A shows that colonies do appear in the presence of virus, and that the frequency of colony appearance is apparently independent of the multiplicity of infection. There are two obvious mechanisms by which such apparently resistant colonies might arise. The first of these is the establishment of some virusinduced cellular state that results in resistance. An example of this would be a stable defective virus replication that results in resistance to further infection without cell killing. The other mechanism is a genetic change in the cell. The results of Table 1A show that colony development frequency is not correlated with multiplicity of infection, suggesting that the basic defect is cellular rather than viral. Furthermore, the fact that very low multiplicities of infection were used considerably reduces the possibility that defective interfering virus particles are involved (Huang, 1973). We also mutagenized cells with ethyl methanesulfonate (EMS). It seems from the second column of Table 1A that the independence of colony number from multiplicity of infection holds with these cells as well. The second column also suggests that mutagenesis may increase the frequency of resistant cells. The fact that the number of colonies tend to vary widely in the unmutagenized sample suggests that spreading of cells may occur from an initial colony. In fact, results such as those in Table 1A are obtained only when flasks are selected for staining by the initial appearance of visible colonies rather than at a fixed time point. Table
1. Frequency
Multiolicitv
of Appearance
of Infection
of Resistant
Colonies
-EMS
+ EMS
0.01 0.001 0.0001 0.00001
0,18 0,0,3,0,1 l,l 39.0
39,43 38,24,41,24,32 35,36 44,36
(8) 0.001a
7/41
31/32
(A)
aThe multiplicity of infection for the unmutagenized cells was 0.002. (A) The numbers indicate the number of colonies of ceils arising at 2-3 weeks after virus infection of -107 cells at the indicated multiplicities. (B) The numbers indicate the fraction of the flasks containing resistant colonies appearing after infection in an experiment separate from that described in (A). The details are given in the text.
To provide more convincing evidence of the effect of mutagenesis on the development of virus resistance, we infected 32 25 cm2 culture flasks seeded with 5 x 105 mutagenized cells 1 day prior to infection (see Table 1 B). The inoculum was 103 PFU of EMC. We also infected 41 25 cm2 culture flasks seeded with 106 unmutagenized cells in a similar manner. At 14 days post-infection, the plates were examined for visible colonies. All the mutagenized dishes except one contained colonies (31/32). The number of colonies of >lOO cells ranged from 3 to 245 with a mean of 58. In the case of unmutagenized cells, only 7 of 41 flasks contained colonies. The number of colonies again was wide ranging, with numbers varying from 1 to over 200. The flasks that did not have colonies were incubated for 2 more weeks, during which no further colonies appeared. Such data indicate that mutagenesis significantly increases the probability of appearance of resistant cells independent of a colony reseeding phenomenon. They also indicate that it is quite possible that reseeding from an original colony can occur because of the wide variation in colony number in the unmutagenized cells. We then proceeded to grow up five of the resistant colonies from both mutagenized (VRE 3 and VRE 5) and nonmutagenized (VRE 1, VRE 2, and VRE 4) cells, and to determine whether they retained their phenotype in nonselective conditions. The unmutagenized resistant cells used (VRE 1, VRE 2, and VRE 4) arose from infection of three independent plates each containing 107 cells. In all cases, there were three or fewer colonies apparent on the plate, and one of these was picked for further growth. The mutagenized resistant cells (VRE 3 and VRE 5) were selected from two independent plates seeded before infection with 106 cells. All colonies picked were subcloned before analysis. Each colony was grown for 4 months. At that time, the cells were characterized as described in Table 2. The morphology was analyzed by phasecontrast microscopy. VRE 1 cells show a highly atypical morphology, being almost twice the size of wild-type CHO. This abnormal size is not due to tetraploidy, since the cells have a modal chromosome number of approximately 20 (the usual number). VRE 2 and VRE 3 do not exhibit consistent morphology over long periods in culture, but can generally be distinguished from wild-type CHO by their inability to form compact colonies. VRE 4 assumes an often spindle-shaped morphology and only rarely forms compact colonies. VRE 5 is indistinguishable from wild-type CHO on morphological grounds. Growth rate and plating efficiency were analyzed as described in Experimental Procedures. It is clear from the growth rates that VRE 1, VRE 3, and VRE 5 grow at or near wild-type doubling time of 13 hr.
Selection 531
of Virus-Resistant
CHO Cells
VRE 2 and VRE 4 grow far more slowly. VRE 2 doubles at about half the rate of wild-type cells, while VRE 4 doubles at about one seventh the rate of wild-type cells. The cloning efficiencies of all the lines vary widely and are generally reduced somewhat relative to wild-type CHO cells. Three of the lines were not persistently infected, since no infectivity could be detected in 1 ml of media obtained after the growth of 106 cells for 2 days in 25 cm2 flasks with 5 ml of media. The infectivity was tested at 33°C 37°C and 39°C. This assay does not rule out the presence of a noninfectious form of virus, and a cryptic infection of these cells remains a possibility. When VRE 2 and VRE 3 cells are grown under similar conditions, titers of 106 PFU/ml are obtained at 37°C. Resistance was assayed as described in Experimental Procedures. Briefly, 10 fold serial dilutions of virus were added to a constant number (105) of cells on 24 well plates. In the case of VRE 1, VRE 4, and VRE 5, multiplicities of infection of EMC varied from >l to tO.OOO1. In no case were the cells killed by the virus. The EMC used was retitered with wild-type CHO in a similar manner on the same plates used to test for resistance. VRE 1 proved partially resistant to other viruses. This was shown by the fact they could be killed by moi of 100 fold greater than was necessary to kill control cells. These assays have been performed at least 3 times for each cell line at roughly monthly intervals after the establishment of the line in nonselective conditions. It is obvious from Table 2 that we have obtained five distinct phenotypes. VRE 1 can be distinguished by its size alone. VRE 2 and VRE 3 are Table
2. Characteristics
of Cells
Selected
for EMC
VRE 1
both persistently infected, but VRE 2 grows at about one half the rate of VRE 3. VRE 4 is not persistently infected and grows extremely slowly. VRE 5 grows at a normal rate and is of normal size. While these lines exhibit different resistance profiles, they can be distinguished merely by the criteria listed above.
Discusslon We have described in this paper a simple system for obtaining CHO cells selected for resistance to virus. The basic point we wish to establish is the feasibility of the quantitative selection of genetically stable, virus-resistant cell types which manifest a variety of phenotypes. We can argue that the cellular alteration that results in the virus resistance selected for is a genetic one on the following grounds. First, the appearance of virus-resistant cells is a rare event that is enhanced by mutagenesis of the cells. It occurs with a frequency similar to that obtained in selection for drug resistance (Baker et al., 1973). Second, the frequency of resistance obtained was increased by mutagenesis to an extent roughly similar to the increase observed for drug resistance markers (Thompson and Baker, 1973; Jones and Sargent, 1974). Third, the phenotypes derived from this selection proved to be stable over several passages in nonselective conditions. There are, apparently, at the very least five possible phenotypes that can result in the virus-resistant state. In the cells we have studied, we have yet to observe two colonies of the same phenotype, suggesting that we have not yet begun to saturate the possibilities. The question of just how many possi-
Resistance VRE 2
VRE 3
VRE 4
VRE 5
+
Mutagenesis
+ +
Morphology Doubling
Time
(Hr)
Cloning Efficiency (% of Wild-Type) Persistent (PFU/ml)
13
27
15
96
13
70
35
45
6
26
106
106
Infection
Resistance EMC
ND
ND
>5
Reovirus
>5 2
ND
ND
>3
>5
vsv
2
ND
ND
0
0
Vaccinia
0
ND
ND
0
0
0
(+) for morphology indicates the cells are indistinguishable from our wild-type CHO cells by microscopy. Persistent infection was determined as described in the text. The numbers for resistance indicate the increased orders of 10 virus multiplicities of infection needed to produce cell killing relative to wild-type cells as described in the text. For example, 2 indicates that 100 fold more virus is required to produce cell killing in a well.
Cell 532
bilities do exist is an interesting one, and we intend to expand our collection of resistant cells until we begin to obtain repeat phenotypes. The fact that we have thus far examined only five resistant colonies does not allow us to draw definitive conclusions about what fraction of the cells are persistently infected or if the ratio of nonpersistently infected cells to persistently infected cells is the same in the presence or absence of mutagenesis. In addition, we are as yet unable to estimate mutational frequencies because we are as yet unaware of the number of events we are dealing with. For instance, fluctuation analysis (Luria and Delbuck, 1943) should prove most enlightening when we have a sound estimate of the number of phenotypes we are dealing with. The question of the nature of the cellular defect that results in the resistance is, of course, a crucial one. The possibilities are numerous and include defects in the plasma membrane, the internal membranes, possible specific protein synthesis factors, possible host polymerase components, cellular proteolysis, and so on. EMC is a picornavirus whose well studied infectious cycle should facilitate such analysis. This analysis will also be facilitated by a knowledge of the viral resistance spectrum. For example, if a cell were resistant only to enveloped viruses, one might initially suspect a membrane defect in the cell, or if a cell were resistant only to (-) stranded viruses, a defect in a cellular component of transcription might be looked for. One obvious possibility is an absorption defect. It is of interest, however, that CHO cells selected for lectin resistance have highly altered surface glycoproteins and are not resistant to EMC (P. Stanley and C. Stanners, personal communication). It is also of interest that two of our three nonpersistently infected cells grow normally. This suggests that the cellular defect involved is not important to cellular integrity as measured by the growth of cells. It is possible that resistance could be due to constitutive interferon or constitutive antiviral protein (AVP) production (Ng and Vilcek, 1972). Either of these should result in a cell resistant to a wide range of viruses and, in the case of the former, the presence of detectable interferon in the media. We have thus far not been able to detect interferon in any of our cell lines. The detection of AVP should prove more difficult and require the use of “in vitro” protein-synthesizing systems. The phenotype one would expect of such cells would be resistance to all viruses tested. We have yet to observe such a phenotype. The persistently infected cells we observe should also prove to be of interest. However, we have obtained similar cells by a similar selection with reo-
virus and have chosen initially to study these cells in some detail, rather than VRE 2 and VRE 3, due to the availability of certain techniques and reagents that we have yet to develop for EMC. In addition, studies on the VRE 2 and VRE 3 are complicated by the fact the cells do not grow well as the confluent monolayers necessary for our standard virus assays. It is important to note that we have yet to obtain any nonpersistently infected cells after selection with reovirus for resistance. Of the five colonies we have studied, all have proved to be carrier cultures. This suggests that the ability of the EMC-CHO selection system to generate nonpersistently infected resistant cells may be fortuitous and somewhat unique. We have found that the cells persistently infected with reovirus are resistant to a degree to EMC, VSV, and vaccinia as well as reovirus (Taber et al., 1976). The resistance seems to resemble the heterologous interference phenomena described by Marcus and his co-workers. We suspect that the resistance of the VRE 2 and VRE 3 will prove to be similar. Preliminary results suggest that the frequency of persistently infected cells is increased by mutagenesis. This may mean that a cellular alteration results in the production of a possibly defective virus. It is conceivable that the different growth rates we observe in VRE 2 and VRE 3 are due to the “cycling” phenomena of virus production (Huang, 1973). The consistency of the phenotypic difference over long periods of time argues against this, but is not conclusive in itself. One might also speculate that the nonpersistently infected cells harbor a noninfectious virus that is inhibitory to an infectious virus. This possibility could be tested by high resolution molecular hybridization experiments. In conclusion, we have shown that it is possible to select virus-resistant cells in a genetically useful cell line at reasonable selection frequencies; that the virus-resistant cells are stable in their phenotype; that the alteration resulting in resistance is probably a mutational event; that a significant proportion of the cells are not persistently infected and consequently available for complementation studies; and that at the least, five distinct phenotypes are exhibited, with the probable appearance of several others. Experlmental
Procedures
Cells CHO-KI cells were obtained from the American Type Culture Collection (CCL 61). They were periodically subcloned All cell growth and virus infection experiments were done in Ham’s F12 buffered with 25 mM N-P-hydroxynethylpiperazine-N’-2-ethanesulfonic acid (Hepes) prepared by Gibco. This medium was supplemented with 10% fetal calf serum also obtained from Gibco. The cells were
Selection 533
of Virus-Resistant
CHO Cells
maintained at 37°C in Forma water-jacketed incubators in Corning 25100 flasks (25 cm2 surface) and Corning 251 IO flasks (75 cm* surface). Growth rates were measured by two methods. The first was to count the cells of 50 colonies on 4 successive days following seeding by phase-contrast microscopy. This technique was used to minimize the possibility of error from unusually sized cells and aggregation of mutant cells. The other technique was to seed 105 cells on four 25 cm2 flasks and to count the cells after trypsinization in one flask each day with an electronic particle counter (Particle Data Inc.) for 4 successive days. In all cases, the points generated a straight line on a semilog plot, and the doubling times were derived from the line. These results reproduced those obtained by colony size determination within 10%. Plating efficiency was determined by seeding ten 25 cm* flasks with 200 cells. The colonies were stained and counted after they had become visible (about one week after seeding, except for VRE 4). The data shown in Table 2 are the averages of 10 determinations for each cell type expressed as a percentage of a control efficiency of 60%. The plating efficiencies are presented as percentages of a wild-type CHO efficiency of 60-70%. Viruses Encephalomyocarditis virus was grown in our laboratory from a cloned stock provided by Dr. Roland Rueckert. Reovirus type Ill was grown from a stock provided by Dr. Wolfgang Joklik. Both these virus stocks have been cloned and are periodically passaged at very low multiplicity to reduce or eliminate defective interfering particles. Vesicular stomatitis virus was obtained from Dr. Judith O’Malley. Vaccinia virus was obtained from Dr. Robert Hughes. All viruses were titered by plaque assays. Virus Resistance Assays The virus resistance assays in Table 2 were obtained by the endpoint dilution method. 105 wild-type CHO cells per well were plated on Falcon 3008 plates. 1 day after plating, 10 fold serial dilutions of the virus were added to the wells without serum. After one half hour at 25’C, fetal calf serum was added to a final concentration of 10%. The cells and virus were then incubated for 3 or more days at 37°C. The plates were then washed, fixed, and stained with 0.1% crystal violet in 10% formalin. In the case of the resistance experiments shown in Table 2, infection was allowed to proceed to the endpoint-that is, one or more PFU of virus resulted in complete removal of staining capacity in the well. Mutagenesis Cells were mutagenized with 200 pg/ml of ethyl methanesulfonate (Eastman) at 37°C for 16 hr. This dose resulted in a reduction of plating efficiency to 15% of control in the experiments reported here. Acknowledgments We wish to thank Nick Hastie, William Held, Joel Huberman. and Kenneth Manly for valuable discussions. The work was supported by grants from the National Cancer Institute and the National Institute for Allergy and Infectious Disease. Received
January
12, 1976;
revised
April 26, 1976
References Baker, R. M., Brunette, D. M., Mankovitz, R., Thompson, Whitmore, G. F.. Siminovitch, L. and Till, J. E. 1974. resistant mutants of mouse and hamster cells in culture. 9-21. Darnell. J. E., and Sawyer, in susceptibility to poliovirus
L. H.. OubainCell 7,
T. K. (1960). The basis for variation in HeLa cells. Virology 7 1, 665-675.
Georgopoulus. G. P. (1971). Bacterial mutants in which the gene N function of bacteriophage lambda is blocked have an altered RNA polymerase. Proc. Nat. Acad. Sci. USA 68, 2977-2981. Georgopoulus, G. P., and Herskowitz, Lambda, A. Hershey, ed. (New York: ry), pp. 553-564.
I. (1972). The Bacteriophage Cold Spring Harbor Laborato-
Georgopoulus, G. P.. Hendrix, R. W., Casjens, A. D. (1973). Host participation in bacteriophage sembly. J. Mol. Biol. 76, 45-60. Huang, A. S. (1973). biol. 27, 101-117.
Defective
interfering
viruses.
Jones, G. E., and Sargent, P. A. (1974). Chinese hamster cells deficient in adenine ferase. Cell 2, 43-54. Lederberg, J., and Lederberg, indirect selection of bacterial Luftig, R., and Weihing, bules in vitro. J. Virol.
S. G., and Kaiser, lambda head asAnn. Rev. Micro-
Mutants of cultured phosphoribosyl trans-
E. M. (1952). Replica plating and mutants. J. Bacterial. 63. 399-406.
R. (1975). Adenovirus 76, 696-706.
binds to rat microtu-
Luria, S. E., and Delbruck, M. (1943). Mutations of bacteria virus sensitivity to virus resistance. Genetics 28, 491-500. Morgan, M. J., Colby, C., and Hulse, J. C. (1973). characterization of virus resistant embryo fibroblasts. 20, 377-385.
from
Isolation and J. Gen. Virol.
Ng, M. H.. and Vilcek, J. (1972). Interferons: physicochemical properties and control of cellular synthesis. Adv. Prot. Chem. 26, 173-241. Preble, 0. T., and Younger, J. S. (1975). Temperature-sensitive viruses and the etiology of chronic and inapparent infections. Infect. Dis. 131. 467-473.
J.
Puck, T. T., Wurthier, P., Jones, C., and Kao, F. T. (1971). Genetics of somatic mammalian cells: lethal antigens as genetic markers for study of human linkage groups. Proc. Nat. Acad. Sci. USA 68, 3102-3106. Taber, R., Alexander, V., and Whitford, W. (1976). Persistent reovirus infection of CHO cells resulting in virus resistance. J. Virol. 17, 513-524. Thompson, K. H., and Baker, Biology, 6, David M. Prescott, pp. 209-281.
R. M. (1973). In Methods ed. (New York: Academic
Vogt, M. (1958). A genetic change in a tissue plastic cells. J. Comp. Physiol. 52, (suppl. l), Vogt, M., and Dulbecco, R. (1958). Properties with increased resistance to poliomyelitis 425-434. Walker, D. L. (1964). The viral carrier Prog. Med. Virol. 6. 111-148.
state
in Cell Press),
culture line of neo271-285. of a HeLa cell culture virus. Virology 5,
in animal
cell cultures.
August
1976,
The Selection Ovary Cells
Copyright01976
by MIT
of Virus-Resistant
Robert Taber, Veronica Alexander, and Norman Wald, Jr. Department of Medical Viral Oncology Roswell Park Memorial Institute and Graduate Faculty in Microbiology State University of New York at Buffalo (Roswell Park Division) Buffalo, New York 14263
We have selected Chlnese hamster ovary (CHO) cells resistant to infection by encephalomyocardltls (EMC) virus. Thus far, we have obtained five lines resistant to EMC, all of which manifest dlfferent phenotypes. Three of the five are not perslstently Infected with virus, while two lines produce infectious vlrus and grow in Its presence. The nonpersistently Infected lines exhibit different reslstance profiles to the other viruses we have tested, and they are stable in nonselective growth conditions. Their resistance appears to be due to a genetic alteration In the cell.
Virus resistance has long been used for selection of bacterial mutants. It was among the markers used in early studies on the nature of bacterial mutations (Luria and Delbruck, 1943; Lederberg and Lederberg, 1952). More recently, the virus resistance phenotype has been used to detect bacterial genes necessary for phage growth. Best characterized of these are three classes of E. coli genes which affect lambda phage RNA synthesis (gro-N), DNA synthesis (gro-P), and particle assembly (gro-E) (Georgopoulus, 1971; Georgopoulus and Herskowitz, 1972; Georgopoulus et al., 1973). The use of virus resistance as a marker in studies with animal cells has long presented an attractive possibility; one should be able to select useful mutants with defects similar to those described for bacteria. However, such apparently straightforward studies have previously proved to be complicated by three basic problems (Vogt, 1958; Vogt and Dulbecco, 1958; Morgan, Colby, and Hulse, 1973). The first of these problems was the lack of knowledge concerning the molecular mechanisms of viral replication. Early attempts to isolate such cells were frustrated in part by the inability to study the nature of the resistance in molecular terms. An example of this was the isolation of a HeLa cell line seemingly resistant to poliovirus at the level of penetration (Darnell and Sawyer, 1960). At the time such cells were isolated, it was impossible to study such cellular properties as protein composition of the
Chinese
Hamster
membrane, or to examine cellular submembrane structural components such as microtubules that may be involved in virus penetration (Luftig and Weihing, 1975). The second impediment to the study of virusresistant eucaryotic cells was the lack of genetically defined cultured mammalian cells. Only in the past few years have marked cell lines become available that would allow genetic studies of the defective cells to identify individual complementation groups. Chinese hamster ovary (CHO) cells in particular have their genetic characteristics (for example, mutation frequency, genetic stability, complementation, and so on) more thoroughly characterized than other cell lines (Thompson and Baker, 1973; Puck et al., 1971). The third problem has been the development of persistent infection during attempts to select for virus resistance (Vogt and Dulbecco, 1958; Morgan et al., 1973) and the lack of quantitative selection. While such infections certainly have inherent interest, they also create difficulties when one attempts to perform cell fusion experiments or analyses of a biochemical defect. A number of such persistently infected systems have been reported (Preble and Younger, 1975; Walker, 1964; Taber, Alexander, and Whitford, 1976). In these situations, one often finds that resistant cells arise only after several cycles of infection and reinfection, making quantitative selection frequency analysis difficult if not impossible (Morgan et al., 1973). In this paper, we describe a simple system using CHO cells and encephalomyocarditis (EMC) virus that apparently results in the selection of virusresistant cells at frequencies similar to those obtained by other selective agents. It appears that the system overcomes the difficulties described above since the molecular biology of EMC virus is well understood, the cell genetics and molecular biology of CHO cells is relatively well advanced, and a significant proportion of the resistant cells obtained are not persistently infected.
Results The most straightforward approach to the selection of virus-resistant cells is simply to infect the cells with virus at various multiplicities of infection. EMC virus was added at the input multiplicities shown in Table 1 A, and the infection was allowed to proceed. Observation of the cells with phase-contrast microscopy revealed extensive cytopathic effects l-3 days after infection with the time of initial appearance of cytopathic effect related to the multiplicity of infection. By the third day after infection, extensive scanning by phase-contrast microscopy of the flasks did not reveal obviously viable cells.
Cell 530
At this time and subsequently until staining, the titer of infectious EMC in the culture fluid was never less than 106 PFU/ml, and the titers from mutagenized and nonmutagenized samples were identical. At 3 weeks post-infection, the plates were stained and colonies counted. From the time of infection until staining, the flasks were not opened. The first column of Table 1A shows that colonies do appear in the presence of virus, and that the frequency of colony appearance is apparently independent of the multiplicity of infection. There are two obvious mechanisms by which such apparently resistant colonies might arise. The first of these is the establishment of some virusinduced cellular state that results in resistance. An example of this would be a stable defective virus replication that results in resistance to further infection without cell killing. The other mechanism is a genetic change in the cell. The results of Table 1A show that colony development frequency is not correlated with multiplicity of infection, suggesting that the basic defect is cellular rather than viral. Furthermore, the fact that very low multiplicities of infection were used considerably reduces the possibility that defective interfering virus particles are involved (Huang, 1973). We also mutagenized cells with ethyl methanesulfonate (EMS). It seems from the second column of Table 1A that the independence of colony number from multiplicity of infection holds with these cells as well. The second column also suggests that mutagenesis may increase the frequency of resistant cells. The fact that the number of colonies tend to vary widely in the unmutagenized sample suggests that spreading of cells may occur from an initial colony. In fact, results such as those in Table 1A are obtained only when flasks are selected for staining by the initial appearance of visible colonies rather than at a fixed time point. Table
1. Frequency
Multiolicitv
of Appearance
of Infection
of Resistant
Colonies
-EMS
+ EMS
0.01 0.001 0.0001 0.00001
0,18 0,0,3,0,1 l,l 39.0
39,43 38,24,41,24,32 35,36 44,36
(8) 0.001a
7/41
31/32
(A)
aThe multiplicity of infection for the unmutagenized cells was 0.002. (A) The numbers indicate the number of colonies of ceils arising at 2-3 weeks after virus infection of -107 cells at the indicated multiplicities. (B) The numbers indicate the fraction of the flasks containing resistant colonies appearing after infection in an experiment separate from that described in (A). The details are given in the text.
To provide more convincing evidence of the effect of mutagenesis on the development of virus resistance, we infected 32 25 cm2 culture flasks seeded with 5 x 105 mutagenized cells 1 day prior to infection (see Table 1 B). The inoculum was 103 PFU of EMC. We also infected 41 25 cm2 culture flasks seeded with 106 unmutagenized cells in a similar manner. At 14 days post-infection, the plates were examined for visible colonies. All the mutagenized dishes except one contained colonies (31/32). The number of colonies of >lOO cells ranged from 3 to 245 with a mean of 58. In the case of unmutagenized cells, only 7 of 41 flasks contained colonies. The number of colonies again was wide ranging, with numbers varying from 1 to over 200. The flasks that did not have colonies were incubated for 2 more weeks, during which no further colonies appeared. Such data indicate that mutagenesis significantly increases the probability of appearance of resistant cells independent of a colony reseeding phenomenon. They also indicate that it is quite possible that reseeding from an original colony can occur because of the wide variation in colony number in the unmutagenized cells. We then proceeded to grow up five of the resistant colonies from both mutagenized (VRE 3 and VRE 5) and nonmutagenized (VRE 1, VRE 2, and VRE 4) cells, and to determine whether they retained their phenotype in nonselective conditions. The unmutagenized resistant cells used (VRE 1, VRE 2, and VRE 4) arose from infection of three independent plates each containing 107 cells. In all cases, there were three or fewer colonies apparent on the plate, and one of these was picked for further growth. The mutagenized resistant cells (VRE 3 and VRE 5) were selected from two independent plates seeded before infection with 106 cells. All colonies picked were subcloned before analysis. Each colony was grown for 4 months. At that time, the cells were characterized as described in Table 2. The morphology was analyzed by phasecontrast microscopy. VRE 1 cells show a highly atypical morphology, being almost twice the size of wild-type CHO. This abnormal size is not due to tetraploidy, since the cells have a modal chromosome number of approximately 20 (the usual number). VRE 2 and VRE 3 do not exhibit consistent morphology over long periods in culture, but can generally be distinguished from wild-type CHO by their inability to form compact colonies. VRE 4 assumes an often spindle-shaped morphology and only rarely forms compact colonies. VRE 5 is indistinguishable from wild-type CHO on morphological grounds. Growth rate and plating efficiency were analyzed as described in Experimental Procedures. It is clear from the growth rates that VRE 1, VRE 3, and VRE 5 grow at or near wild-type doubling time of 13 hr.
Selection 531
of Virus-Resistant
CHO Cells
VRE 2 and VRE 4 grow far more slowly. VRE 2 doubles at about half the rate of wild-type cells, while VRE 4 doubles at about one seventh the rate of wild-type cells. The cloning efficiencies of all the lines vary widely and are generally reduced somewhat relative to wild-type CHO cells. Three of the lines were not persistently infected, since no infectivity could be detected in 1 ml of media obtained after the growth of 106 cells for 2 days in 25 cm2 flasks with 5 ml of media. The infectivity was tested at 33°C 37°C and 39°C. This assay does not rule out the presence of a noninfectious form of virus, and a cryptic infection of these cells remains a possibility. When VRE 2 and VRE 3 cells are grown under similar conditions, titers of 106 PFU/ml are obtained at 37°C. Resistance was assayed as described in Experimental Procedures. Briefly, 10 fold serial dilutions of virus were added to a constant number (105) of cells on 24 well plates. In the case of VRE 1, VRE 4, and VRE 5, multiplicities of infection of EMC varied from >l to tO.OOO1. In no case were the cells killed by the virus. The EMC used was retitered with wild-type CHO in a similar manner on the same plates used to test for resistance. VRE 1 proved partially resistant to other viruses. This was shown by the fact they could be killed by moi of 100 fold greater than was necessary to kill control cells. These assays have been performed at least 3 times for each cell line at roughly monthly intervals after the establishment of the line in nonselective conditions. It is obvious from Table 2 that we have obtained five distinct phenotypes. VRE 1 can be distinguished by its size alone. VRE 2 and VRE 3 are Table
2. Characteristics
of Cells
Selected
for EMC
VRE 1
both persistently infected, but VRE 2 grows at about one half the rate of VRE 3. VRE 4 is not persistently infected and grows extremely slowly. VRE 5 grows at a normal rate and is of normal size. While these lines exhibit different resistance profiles, they can be distinguished merely by the criteria listed above.
Discusslon We have described in this paper a simple system for obtaining CHO cells selected for resistance to virus. The basic point we wish to establish is the feasibility of the quantitative selection of genetically stable, virus-resistant cell types which manifest a variety of phenotypes. We can argue that the cellular alteration that results in the virus resistance selected for is a genetic one on the following grounds. First, the appearance of virus-resistant cells is a rare event that is enhanced by mutagenesis of the cells. It occurs with a frequency similar to that obtained in selection for drug resistance (Baker et al., 1973). Second, the frequency of resistance obtained was increased by mutagenesis to an extent roughly similar to the increase observed for drug resistance markers (Thompson and Baker, 1973; Jones and Sargent, 1974). Third, the phenotypes derived from this selection proved to be stable over several passages in nonselective conditions. There are, apparently, at the very least five possible phenotypes that can result in the virus-resistant state. In the cells we have studied, we have yet to observe two colonies of the same phenotype, suggesting that we have not yet begun to saturate the possibilities. The question of just how many possi-
Resistance VRE 2
VRE 3
VRE 4
VRE 5
+
Mutagenesis
+ +
Morphology Doubling
Time
(Hr)
Cloning Efficiency (% of Wild-Type) Persistent (PFU/ml)
13
27
15
96
13
70
35
45
6
26
106
106
Infection
Resistance EMC
ND
ND
>5
Reovirus
>5 2
ND
ND
>3
>5
vsv
2
ND
ND
0
0
Vaccinia
0
ND
ND
0
0
0
(+) for morphology indicates the cells are indistinguishable from our wild-type CHO cells by microscopy. Persistent infection was determined as described in the text. The numbers for resistance indicate the increased orders of 10 virus multiplicities of infection needed to produce cell killing relative to wild-type cells as described in the text. For example, 2 indicates that 100 fold more virus is required to produce cell killing in a well.
Cell 532
bilities do exist is an interesting one, and we intend to expand our collection of resistant cells until we begin to obtain repeat phenotypes. The fact that we have thus far examined only five resistant colonies does not allow us to draw definitive conclusions about what fraction of the cells are persistently infected or if the ratio of nonpersistently infected cells to persistently infected cells is the same in the presence or absence of mutagenesis. In addition, we are as yet unable to estimate mutational frequencies because we are as yet unaware of the number of events we are dealing with. For instance, fluctuation analysis (Luria and Delbuck, 1943) should prove most enlightening when we have a sound estimate of the number of phenotypes we are dealing with. The question of the nature of the cellular defect that results in the resistance is, of course, a crucial one. The possibilities are numerous and include defects in the plasma membrane, the internal membranes, possible specific protein synthesis factors, possible host polymerase components, cellular proteolysis, and so on. EMC is a picornavirus whose well studied infectious cycle should facilitate such analysis. This analysis will also be facilitated by a knowledge of the viral resistance spectrum. For example, if a cell were resistant only to enveloped viruses, one might initially suspect a membrane defect in the cell, or if a cell were resistant only to (-) stranded viruses, a defect in a cellular component of transcription might be looked for. One obvious possibility is an absorption defect. It is of interest, however, that CHO cells selected for lectin resistance have highly altered surface glycoproteins and are not resistant to EMC (P. Stanley and C. Stanners, personal communication). It is also of interest that two of our three nonpersistently infected cells grow normally. This suggests that the cellular defect involved is not important to cellular integrity as measured by the growth of cells. It is possible that resistance could be due to constitutive interferon or constitutive antiviral protein (AVP) production (Ng and Vilcek, 1972). Either of these should result in a cell resistant to a wide range of viruses and, in the case of the former, the presence of detectable interferon in the media. We have thus far not been able to detect interferon in any of our cell lines. The detection of AVP should prove more difficult and require the use of “in vitro” protein-synthesizing systems. The phenotype one would expect of such cells would be resistance to all viruses tested. We have yet to observe such a phenotype. The persistently infected cells we observe should also prove to be of interest. However, we have obtained similar cells by a similar selection with reo-
virus and have chosen initially to study these cells in some detail, rather than VRE 2 and VRE 3, due to the availability of certain techniques and reagents that we have yet to develop for EMC. In addition, studies on the VRE 2 and VRE 3 are complicated by the fact the cells do not grow well as the confluent monolayers necessary for our standard virus assays. It is important to note that we have yet to obtain any nonpersistently infected cells after selection with reovirus for resistance. Of the five colonies we have studied, all have proved to be carrier cultures. This suggests that the ability of the EMC-CHO selection system to generate nonpersistently infected resistant cells may be fortuitous and somewhat unique. We have found that the cells persistently infected with reovirus are resistant to a degree to EMC, VSV, and vaccinia as well as reovirus (Taber et al., 1976). The resistance seems to resemble the heterologous interference phenomena described by Marcus and his co-workers. We suspect that the resistance of the VRE 2 and VRE 3 will prove to be similar. Preliminary results suggest that the frequency of persistently infected cells is increased by mutagenesis. This may mean that a cellular alteration results in the production of a possibly defective virus. It is conceivable that the different growth rates we observe in VRE 2 and VRE 3 are due to the “cycling” phenomena of virus production (Huang, 1973). The consistency of the phenotypic difference over long periods of time argues against this, but is not conclusive in itself. One might also speculate that the nonpersistently infected cells harbor a noninfectious virus that is inhibitory to an infectious virus. This possibility could be tested by high resolution molecular hybridization experiments. In conclusion, we have shown that it is possible to select virus-resistant cells in a genetically useful cell line at reasonable selection frequencies; that the virus-resistant cells are stable in their phenotype; that the alteration resulting in resistance is probably a mutational event; that a significant proportion of the cells are not persistently infected and consequently available for complementation studies; and that at the least, five distinct phenotypes are exhibited, with the probable appearance of several others. Experlmental
Procedures
Cells CHO-KI cells were obtained from the American Type Culture Collection (CCL 61). They were periodically subcloned All cell growth and virus infection experiments were done in Ham’s F12 buffered with 25 mM N-P-hydroxynethylpiperazine-N’-2-ethanesulfonic acid (Hepes) prepared by Gibco. This medium was supplemented with 10% fetal calf serum also obtained from Gibco. The cells were
Selection 533
of Virus-Resistant
CHO Cells
maintained at 37°C in Forma water-jacketed incubators in Corning 25100 flasks (25 cm2 surface) and Corning 251 IO flasks (75 cm* surface). Growth rates were measured by two methods. The first was to count the cells of 50 colonies on 4 successive days following seeding by phase-contrast microscopy. This technique was used to minimize the possibility of error from unusually sized cells and aggregation of mutant cells. The other technique was to seed 105 cells on four 25 cm2 flasks and to count the cells after trypsinization in one flask each day with an electronic particle counter (Particle Data Inc.) for 4 successive days. In all cases, the points generated a straight line on a semilog plot, and the doubling times were derived from the line. These results reproduced those obtained by colony size determination within 10%. Plating efficiency was determined by seeding ten 25 cm* flasks with 200 cells. The colonies were stained and counted after they had become visible (about one week after seeding, except for VRE 4). The data shown in Table 2 are the averages of 10 determinations for each cell type expressed as a percentage of a control efficiency of 60%. The plating efficiencies are presented as percentages of a wild-type CHO efficiency of 60-70%. Viruses Encephalomyocarditis virus was grown in our laboratory from a cloned stock provided by Dr. Roland Rueckert. Reovirus type Ill was grown from a stock provided by Dr. Wolfgang Joklik. Both these virus stocks have been cloned and are periodically passaged at very low multiplicity to reduce or eliminate defective interfering particles. Vesicular stomatitis virus was obtained from Dr. Judith O’Malley. Vaccinia virus was obtained from Dr. Robert Hughes. All viruses were titered by plaque assays. Virus Resistance Assays The virus resistance assays in Table 2 were obtained by the endpoint dilution method. 105 wild-type CHO cells per well were plated on Falcon 3008 plates. 1 day after plating, 10 fold serial dilutions of the virus were added to the wells without serum. After one half hour at 25’C, fetal calf serum was added to a final concentration of 10%. The cells and virus were then incubated for 3 or more days at 37°C. The plates were then washed, fixed, and stained with 0.1% crystal violet in 10% formalin. In the case of the resistance experiments shown in Table 2, infection was allowed to proceed to the endpoint-that is, one or more PFU of virus resulted in complete removal of staining capacity in the well. Mutagenesis Cells were mutagenized with 200 pg/ml of ethyl methanesulfonate (Eastman) at 37°C for 16 hr. This dose resulted in a reduction of plating efficiency to 15% of control in the experiments reported here. Acknowledgments We wish to thank Nick Hastie, William Held, Joel Huberman. and Kenneth Manly for valuable discussions. The work was supported by grants from the National Cancer Institute and the National Institute for Allergy and Infectious Disease. Received
January
12, 1976;
revised
April 26, 1976
References Baker, R. M., Brunette, D. M., Mankovitz, R., Thompson, Whitmore, G. F.. Siminovitch, L. and Till, J. E. 1974. resistant mutants of mouse and hamster cells in culture. 9-21. Darnell. J. E., and Sawyer, in susceptibility to poliovirus
L. H.. OubainCell 7,
T. K. (1960). The basis for variation in HeLa cells. Virology 7 1, 665-675.
Georgopoulus. G. P. (1971). Bacterial mutants in which the gene N function of bacteriophage lambda is blocked have an altered RNA polymerase. Proc. Nat. Acad. Sci. USA 68, 2977-2981. Georgopoulus, G. P., and Herskowitz, Lambda, A. Hershey, ed. (New York: ry), pp. 553-564.
I. (1972). The Bacteriophage Cold Spring Harbor Laborato-
Georgopoulus, G. P.. Hendrix, R. W., Casjens, A. D. (1973). Host participation in bacteriophage sembly. J. Mol. Biol. 76, 45-60. Huang, A. S. (1973). biol. 27, 101-117.
Defective
interfering
viruses.
Jones, G. E., and Sargent, P. A. (1974). Chinese hamster cells deficient in adenine ferase. Cell 2, 43-54. Lederberg, J., and Lederberg, indirect selection of bacterial Luftig, R., and Weihing, bules in vitro. J. Virol.
S. G., and Kaiser, lambda head asAnn. Rev. Micro-
Mutants of cultured phosphoribosyl trans-
E. M. (1952). Replica plating and mutants. J. Bacterial. 63. 399-406.
R. (1975). Adenovirus 76, 696-706.
binds to rat microtu-
Luria, S. E., and Delbruck, M. (1943). Mutations of bacteria virus sensitivity to virus resistance. Genetics 28, 491-500. Morgan, M. J., Colby, C., and Hulse, J. C. (1973). characterization of virus resistant embryo fibroblasts. 20, 377-385.
from
Isolation and J. Gen. Virol.
Ng, M. H.. and Vilcek, J. (1972). Interferons: physicochemical properties and control of cellular synthesis. Adv. Prot. Chem. 26, 173-241. Preble, 0. T., and Younger, J. S. (1975). Temperature-sensitive viruses and the etiology of chronic and inapparent infections. Infect. Dis. 131. 467-473.
J.
Puck, T. T., Wurthier, P., Jones, C., and Kao, F. T. (1971). Genetics of somatic mammalian cells: lethal antigens as genetic markers for study of human linkage groups. Proc. Nat. Acad. Sci. USA 68, 3102-3106. Taber, R., Alexander, V., and Whitford, W. (1976). Persistent reovirus infection of CHO cells resulting in virus resistance. J. Virol. 17, 513-524. Thompson, K. H., and Baker, Biology, 6, David M. Prescott, pp. 209-281.
R. M. (1973). In Methods ed. (New York: Academic
Vogt, M. (1958). A genetic change in a tissue plastic cells. J. Comp. Physiol. 52, (suppl. l), Vogt, M., and Dulbecco, R. (1958). Properties with increased resistance to poliomyelitis 425-434. Walker, D. L. (1964). The viral carrier Prog. Med. Virol. 6. 111-148.
state
in Cell Press),
culture line of neo271-285. of a HeLa cell culture virus. Virology 5,
in animal
cell cultures.
