Molecular Biology of Plant Tumors

5-6.

Preface

Cancer is a particularly challenging problem both as a disease and as a phenomenon of abnormal cellular growth. For obvious reasons, the bulk of research has been devoted to study of animal and human cancers. As a scientific study object, however, plant cancers deserve our full attention.

Plant cancers are known to be caused either by viruses or by bacteria or by genetic imbalance, as is the case for the so-called genetic tumors that spontaneously arise on certain plant hybrids. Although a considerable amount of observations have been accumulated in relation to plant oncogenesis in general, very little was known about the underlying mechanisms until very recently.

Yet the study of the molecular mechanism responsible for the crown gall phenomenon (induced by Agrobacterium tumefaciens) has turned out to be extraordinarily rewarding and illustrates the point that the fundamental study of the cause(s) and mechanism(s) of abnormal growth might be one of the most efficient ways to understand cellular differentiation and the molecular basis of gene expression. The genetic structure responsible for the neoplasmic transformation of plant cells in crown galls is a bacterial plasmid (called Ti for tumor-inducing). Research described in this volume led us to realize that these Ti plasmids were designed by evolution as natural gene vectors with which some bacteria can introduce active genes into plants. These transferred genes are stably maintained by integration in the plant genome and their expression is directly or indirectly responsible for the tumorous growth pattern.

The oncogenes carried by Ti plasmids are therefore a unique model system to study the mechanisms by which some regulatory genes influence the expression of other genes and thereby control growth and differentiation. Indeed, these T-DNA linked oncogenes have been identified genetically, have been isolated, and can therefore be studied in great detail. Furthermore, because of the gene vector properties of Ti plasmids, genes can be modified experimentally, and the modified genes can be introduced in plants. Thus, one can hope to experimentally probe the relation between the molecular structure and the function of genes by measuring the influence of the alteration to the gene on its expression after introduction in plant cells.

It can be expected that the study of other types of plant cancers will similarly contribute to our understanding of plant molecular biology. It is our hope that this book will stimulate scientific interest in the study of abnormal growth in plants.

Günter. Kahl

Josef S. Schell

I

ABNORMAL PLANT GROWTH

Outline

Chapter 1: Habituation of Cultured Plant Cells

Chapter 2: Genetic Tumors: Physiological Aspects of Tumor Formation in Interspecies Hybrids

Chapter 3: Wound Tumor Disease

Chapter 4: Biology of Legume–Rhizobium Interactions in Nodule Formation

Chapter 5: Insect Galls

1

Habituation of Cultured Plant Cells

FREDERICK MEINS, JR.

Publisher Summary

This chapter discusses the habituation of cultured plant cells. When plant cells and tissues are propagated in axenic culture they sometimes lose their exogenous requirement for certain growth factors and they may be subcultured in the absence of the added factors. This type of cellular, heritable change is known as habituation. Cell division factor (CDF) habituation of tobacco, like cytodifferentiation in normal development, involves two phases, commitment of cells to a competent state followed by overt differentiation in response to an inductive stimulus. The overt differentiation phase results from the activation of a positive feedback loop in which CDF, and possibly auxin, increase their own production or decrease their own degradation. The mechanism for the more stable change in competence for habituation is unknown. One possibility suggested by a comparison of plants derived from cortex and leaf cells is that this involves directed genetic alterations that, depending upon cell type, do not always revert in the plant regeneration process.

I Introduction

II Stable Changes in the Growth Factor Requirement of Cultured Tissues

A Definition of Habituation

B Gradual versus All-or-None Changes

C The Physiological Basis for Autotrophic Growth

D Interactions in the Auxin and CDF Requirement of Cells

III The Nature of the Heritable Change in Habituation

A Epigenetic Changes

B Variation in the Competence of Cells to Habituate

C Positive Feedback Models for Habituation

D Genetic Changes

IV Habituation and Crown Gall Tumor Transformation

V Concluding Remarks

References

I INTRODUCTION

When plant cells and tissues are propagated in axenic culture they sometimes lose their exogenous requirement for certain growth factors, and, thereafter, they may be subcultured in the absence of the added factors. This type of cellular, heritable change, known as habituation (White, 1951), is of interest for two reasons. First, similar heritable changes occur in the normal development of plants and animals and are thought to be the basis for cell commitment and the progressive nature of the determination process (Meins and Binns, 1979). Second, habituation bears a striking formal similarity to tumor transformation in crown gall disease and thus may provide a clue to how the capacity for autonomous growth is maintained in populations of dividing tumor cells (Meins, 1977; Braun, 1978). In this chapter, I will describe the process of habituation for two classes of growth regulators, auxins and cell division factors (CDF), provide evidence that the habituated state is maintained by epigenetic mechanisms, and conclude with a discussion of how habituation bears on the problem of crown gall transformation.

II STABLE CHANGES IN THE GROWTH FACTOR REQUIREMENT OF CULTURED TISSUES

A Definition of Habituation

The proliferation of higher plant cells requires the concerted action of specific growth factors, of which the cell division factors (CDF) and auxins appear to be the most important (Jablonski and Skoog, 1954; Naylor et al., 1954; Everett et al., 1978). Thus, many plant tissues require CDF, usually provided as a synthetic cytokinin, kinetin, and auxin for continuous growth in culture on an otherwise complete medium (Gautheret, 1959; Street, 1969). In 1942, Gautheret reported that strains of carrot tissue in culture gradually lose their requirement for an exogenous source of auxin. Gautheret called this phenomenon accoutumance à l’auxine and later anergie à l’auxine (Gautheret, 1955a), which is usually translated as auxin habituation (White, 1951). It was soon recognized that auxin habituation occurs widely in cultured plant tissues and that similar changes are encountered for cytokinins and certain vitamins as well (Gautheret, 1955b; Street, 1969). The term habituation is now used more generally to describe the stable, heritable loss in the requirement of cultured plant tissues for growth factors (Butcher, 1977).

B Gradual versus All-or-None Changes

Habituation is a gradual process at the tissue level. By comparing the growth response of tissues on media with and without auxin, it was found that during serial propagation tissues often increase in degree of habituation, indicating that the process is progressive (Gautheret, 1955a). Two types of mechanisms, alone or in combination, could account for these gradual changes. First, it may be argued that habituation is an all-or-none phenomenon at the cell level. During prolonged culture, a few habituated cells arise in the tissue and then, because these cells proliferate more rapidly than nonhabituated cells, their relative abundance increases and, hence, the degree of habituation of the tissue increases as well. Second, changes at the tissue level may reflect gradual changes in degree of habituation of the individual constituent cells. Single-cell cloning experiments provide a means of distinguishing between these two hypotheses. Clones isolated from CDF-habituated tissues of tobacco vary widely in their relative growth rate on media with and without cytokinin (Tandeau de Marsac and Jouanneau, 1972; Binns and Meins, 1973; Meins and Binns, 1977). Thus, habituated tissues consist of an heterogenous population of cells, which vary in their degree of habituation. To find out whether or not this heterogeneity arises by gradual cellular changes, Meins and Binns (1977) monitored the degree of habituation of numerous subclones derived from slightly CDF-habituated, cloned tissues serially propagated for long periods of time. They found that the distribution of subclones gradually increased in degree of habituation with time in culture. Moreover, the average degree of habituation of subclones was approximately the same as the value obtained with tissue from which the subclones were obtained. Thus, it appears that changes in the exogenous CDF requirement of CDF-habituated tissues results from gradual, progressive changes at the cellular level.

C The Physiological Basis for Autotrophic Growth

There is evidence that the capacity for autotrophic growth of habituated cells results from changes in the production of specific growth factors rather than from a fundamental alteration in the cell’s requirement for these factors. In the case of CDF and auxin, direct measurements show that cells produce the growth factor for which they are habituated and in amounts sufficient to support the proliferation of nonhabituated cells (Kulescha and Gautheret, 1948; Fox, 1963; Wood et al., 1969; Dyson and Hall, 1972; Syōno and Furuya, 1972; Tandeau de Marsac and Jouanneau, 1972; Einset and Skoog, 1973; Pengelly and Meins, 1978; Pengelly, 1980). Indirect evidence linking CDF production with autotrophic growth comes from studies of temperature-sensitive cell lines. Binns and Meins (1979) found that CDF-habituated clones recently isolated from pith tissues of Havana 425 tobacco require an exogenous source of a CDF such as kinetin for growth at 16°C but not at 25°C, the standard culture temperature. Apparently, cold treatment affects the net production of CDF and not the affinity of cellular receptors for these growth factors because habituated clones at 16°C and nonhabituated sister clones at 16°C as well as 25°C require the same concentration of kinetin for optimal rates of growth. On prolonged culture, cold-sensitive cells give rise to cold-resistant variants that exhibit the habituated phenotype at both temperatures. The key observation is that cold-sensitive cells proliferate at the nonpermissive temperature when cold-resistant cells are present in the same tissue expiant, indicating that the variant cells produce a CDF to which the parent, cold-sensitive cells can respond. Thus, expression of the habituated states is correlated with growth factor production by cells in the same tissue. Since cold treatment appears to affect the net production of CDF and cold-sensitive habituated clones show an absolute requirement for CDF at nonpermissive temperatures, it is likely that habituation involves increased net production of CDF by the cell.

More detailed experiments aimed at measuring differences in growth factor production by habituated and nonhabituated cells are difficult to interpret. For example, in studies with Scorzonera hispanica tissues cultured on auxin-less medium, Kulescha and Gautheret (1948) found that the extractable auxin content of auxin-habituated tissue was higher than that of nonhabituated root tissue. As pointed out by Butcher (1977), the nonhabituated tissue does not grow under the conditions used, and, hence, the elevated auxin content of the autotrophic tissue may simply reflect the fact that it is growing. It is known that auxin is produced by actively growing regions of the plant (Sheldrake, 1973) and by proliferating, nonhabituated cells in culture. Nishinari and Yamaki (1976) have isolated a cloned line of Xanthi tobacco that requires the synthetic auxin, 2,4-dichlorophenoxyacetic acid (2,4D) for proliferation in suspension culture. They found that when cells in starved cultures are transferred to fresh, 2,4D containing medium they divide synchronously for at least four mitotic cycles. Under these conditions, the cells produce small amounts of the naturally occurring auxin, 3-indoleacetic acid, but only during the mitotic phase of the cell cycle. The results obtained with CDF are conflicting. Although still a controversial issue (cf. Miller, 1974, 1975; Wood et al., 1974), there are thought to be two classes of naturally occurring CDF: the cytokinins, which are 6-substituted purines (Skoog and Armstrong, 1970), and the cytokinesins, which appear to be substituted purinones (Wood et al., 1972; 1974). The cytokinin, isopentenyl adenosine, which has pronounced CDF activity, is present in the free nucleoside pool of CDF-habituated tobacco tissues but is not detectable in nonhabituated tobacco tissues actively growing on a medium containing kinetin (Dyson and Hall, 1972). On the other hand, Wood and Braun (1967) have reported that Vinca rosea tissues with an absolute requirement for CDF produce cytokinesins when cultured on a kinetin-containing medium.

In summary, there is evidence that habituation has as its basis alterations in the regulation of growth factor production and/or degradation. Although the issue is not settled, these alterations are likely to result in an increased net production of the growth factor for which the cell is habituated.

D Interactions in the Auxin and CDF Requirement of Cells

Cultured tissues of different plant species, or even different varieties within a species, vary widely in their exogenous requirement for growth regulators. Some tissues require both auxin and CDF, others only one of the two growth factors, and a few tissues neither of the growth factors (Gautheret, 1955b; Yeoman and Macleod, 1977). There is also evidence for complex interactions of auxin and CDF, provided as cytokinin, in regulating cell proliferation and growth factor production. The synthetic auxin, 2,4D, when provided at high concentrations, replaces both the auxin and CDF requirement of cultured soybean, sycamore, and tobacco cells (Witham, 1968; Simpkins et al., 1970). The naturally occurring auxin, indoleacetic acid, and another synthetic auxin, naphthaleneacetic acid, are ineffective. CDF-habituated pith tissues of Wisconsin 38 and Havana 425 tobacco require an exogenous source of cytokinin for growth on low-auxin medium but not on high-auxin medium (Einset, 1977; Binns, 1979), suggesting that auxin is required for expression of the CDF-habituated state. Conversely, CDF-habituated tissues cultured on media containing high concentrations of the cytokinin, kinetin, do not require auxin and produce this growth factor (Syōno and Furuya, 1972). Thus, under certain conditions there appears to be a reciprocal induction of auxin and CDF production in cultured cells. This might provide an explanation for variation of growth factor requirement in different plant species and for the observation that habituation for CDF and auxin are not independent events. There are numerous reports of tobacco tissues habituated for CDF alone, whereas, with one exception, all of the auxin-habituated lines that have been isolated are habituated for CDF as well (Syōno and Furuya, 1972), Apparently, CDF habituation is required as a prelude for auxin habituation.

III THE NATURE OF THE HERITABLE CHANGE IN HABITUATION

A Epigenetic Changes

Two types of mechanisms have been proposed to account for the heritability of the habituated state: genetic mutation (de Ropp, 1951; White, 1951) and epigenetic changes (Gautheret, 1955a; Braun, 1974; Meins and Binns, 1978). Genetic mutations most commonly involve rare, random alterations in genetic constitution, such as point mutations, deletions, and chromosomal rearrangements. Epigenetic changes are more difficult to define because their underlying mechanism is, as yet, unknown, and, except in unicellular organisms, it has not been possible to subject somatic cells to direct, genetic analysis. These changes, in contrast to classical mutations, involve directed alterations in cellular phenotype which occur at very high rates, are regularly reversible, result in the expression of pre-existent genetic potentialities, and leave the heritable altered cells totipotent (Nanney, 1958; Meins, 1972).

In principle, epigenetic changes and classical genetic mutations can be distinguished by a plant regeneration test. This involves, first, cloning the variant cells to ensure that the tissues used are of single cell origin. Second, plants are regenerated from a large number of recently isolated wild type and variant clones. These plants are allowed to develop and are tested for fertility to find out whether or not the cells from which they were derived are totipotent. Third, tissue expiants from the regenerated plants isolated from the same tissue type as the source of the original cultured tissues are returned to culture and assayed for the variant phenotype. If the tissues derived from variant clones regularly express the wild type phenotype, then it may be concluded that in some cases, at least, the process is potentially reversible — an important criterion for epigenetic changes. If, on the other hand, the tissues of variant cell origin do not revert to the wild type, then the regeneration test fails to distinguish between the two mechanisms. When these results are obtained, the regenerated plants are allowed to set seed, and the appropriate tissues from numerous progeny are scored for expression of the variant phenotype. If this phenotype is transmitted to the progeny in a regular fashion, then it is likely that the variation involves genetic changes.

Reversibility of a trait in the plant regeneration test does not exclude the possibility that variation involves rare, random mutations. For example, it may be argued that in the period between the isolation of the clone and the regeneration of plants a few variant cells arise by back mutation and that only the revertant cells form plants (Melchers, 1971). Thus, it is important to measure the rates at which cells undergo variation and establish that these rates are significantly greater than expected for somatic mutation.

There have been several attempts to apply tests of this type to the problem of habituation. Lutz (1971) and Lutz and Belin (1974) were able to induce the regeneration of plants and concomitant reversal of habituation with two cloned lines isolated from White Burley tobacco tissues habituated for both auxin and CDF. Most of the clones tested, however, were no longer capable of normal development, and there is now reason to believe that the tissues used by Lutz may have been of crown gall origin (Sacristán and Melchers, 1977). Sacristán and Wendt-Gallitelli (1970) isolated lines of Crepis capillaris habituated for both auxin and CDF. They were able to regenerate plants from the habituated and nonhabituated lines and found that tissues from plants of habituated tissue origin expressed the nonhabituated phenotype. The habituated cells and cells in plants from habituated tissues carried a specific chromosomal abnormality which was absent in the original nonhabituated line. Although plants were not regenerated from cloned tissues in this study, it is likely that they were of habituated cell origin, since the probability of the same chromosomal abnormality arising independently in two different cells is very low. The results of this type of experiment, while consistent with the hypothesis that habituation has an epigenetic basis, do not rule out somatic mutation, since rates of reversion were not measured.

Convincing evidence that habituation involves epigenetic changes comes from studies of pith parenchyma tissues obtained from Havana 425 tobacco (Meins and Binns, 1978). The choice of this experimental system was a fortunate one because Havana 425 pith cells habituate rapidly under inductive conditions, remain totipotent for many transfer generations in culture, and are readily cloned. Primary expiants of pith tissue habituate for CDF when incubated on auxin-containing medium at 35°C, about 10°C above the standard culture temperature (Meins, 1975a), or in the presence of trace amounts of kinetin (Meins and Lutz, 1979). Under these inductive conditions, the cells enlarge considerably and the tissues become translucent. Some of the expiants also form dense, white, hyperplastic nodules which are clearly discernible by transillumination. These nodules, but not the surrounding translucent tissue, are CDF habituated. Thus, the nodules provide a marker for the habituation event. If this event occurs at random, then the proportion of replicate expiants of the same size without nodules provides an estimate of the number of conversions to the habituated phenotype per expiant from which the rate of habituation may be calculated. This rate varied widely from experiment to experiment, with an average of ca. 5 × 10−3 habituation events per cell generation (Meins and Lutz, 1980a). The actual rate is undoubtedly much higher, since habituated cells arising late in the induction period probably form nodules too small to detect by eye. The important point is that the rate of habituation is 100 to 1000 times faster than the rates reported for somatic mutation in Nicotiana species (Sand et al., 1960; Carlson, 1974).

Cell division factor-habituated clones regularly formed fertile tobacco plants when placed on inductive medium, indicating that habituated cells have remained totipotent (Binns and Meins, 1973). Moreover, pith tissues from the regenerated plants were always nonhabituated and in this regard were indistinguishable from pith tissues from seed-grown plants or from plants of nonhabituated cell origin. Thus, at some point in plant regeneration at least some progeny of habituated cells revert to the nonhabituated state.

There is good evidence that the revertant cells do not arise by rare back mutation (Meins and Binns, 1978). First, recently isolated subclones of habituated and nonhabituated tissues form buds, the first step in plant regeneration, at roughly the same rate. Second, the rate of bud initiation calculated from the number of cell generations the progeny of single cells undergo before the first bud appears is 8.2 ± 5.8 × 10−3 initiations per cell generation (± standard error, n = 7), which is at least 1000 times faster than would be expected were bud initiation to depend upon back mutation. It appears, therefore, that the plants either develop directly from habituated cells indicating that habituation is regularly reversible or from revertant nonhabituated cells that arise in the cloned tissues at the high rates expected for epigenetic changes.

The plant regeneration tests and rate measurements provide strong evidence that CDF habituation of pith parenchyma tissues of Havana 425 tobacco has an epigenetic basis. This is a directed process which occurs at high rates under inductive conditions, is reversible at rates much faster than expected for back mutation, and leaves the heritably altered cell totipotent. There is also indirect evidence for this hypothesis. Gautheret (1955a) believed that habituation involved enzymatic adaptation, an epigenetic mechanism, rather than somatic mutation because, at the tissue level, the process is gradual, progressive, and at least partially reversible. Meins and Binns (1977) confirmed Gautheret’s observations with CDF-habituated tissues and showed that these changes occur at the cellular level. Moreover, as pointed out earlier, habituated cells produce auxins and CDF of the type produced by certain tissues in the intact plant (Thimann, 1969; Torrey, 1976). Apparently, habituation involves the heritable expression of genes that are normally silent in cultured cells.

B Variation in the Competence of Cells to Habituate

Different tissue types from the same plant may vary in growth factor requirement and competence to habituate when placed in culture. For example, in the case of Havana 425 tobacco, primary pith expiants are usually CDF requiring and then rapidly habituate when placed on kinetin-containing medium (Meins and Lutz, 1979). In contrast, primary expiants of stem–cortex grow rapidly with or without added CDF in the culture medium — the tissues are initially habituated — and this phenotype persists when cells are cloned. A third, distinct phenotype is expressed by leaf tissues. Primary expiants of leaf lamella grow slowly or not at all unless CDF is added to the culture medium. Unlike pith, the nonhabituated state of leaf tissue is stable. Cloned leaf cells serially subcultured for at least ten transfers on a kinetin-containing medium continue to express the nonhabituated phenotype. These results bear directly on the problem of how the habituated state is maintained. First, they provide further evidence that CDF habituation does not involve somatic mutation. Were this the case, different cell types from the same plant would be expected to habituate at roughly the same rate. They do not. Second, the results suggest that cell types arise during development which differ in their competence to habituate under inductive conditions.

There is evidence that changes in competence, like the habituation process itself, have an epigenetic basis. Pith cells are totipotent and, hence, can give rise to leaf cells. This implies that during organogenesis some progeny of competent pith cells shift to a new, stable state in which cells are unable to respond to CDF by habituating. To test this hypothesis, Meins and Lutz (1980b) induced cloned, habituated pith tissues to form plants, incubated leaf tissues from these plants on kinetin-containing medium, and then assayed the leaf callus for habituation. Leaf tissues of pith cell origin did not habituate after kinetin treatment and, in this regard, were indistinguishable from leaf tissues from seed grown plants. In the reciprocal experiment, pith tissues of leaf cell origin exhibited the phenotype typical of cultured pith. Thus, cells can shift reversibly between alternative states of competence.

The regeneration experiments show that tobacco cells can undergo stable, potentially reversible changes in competence that leave the cell totipotent. Moreover, these changes are directed in the sense that they occur in response to specific developmental cues, in this case, the induction of organogenesis during plant regeneration. This suggests that the habituation process involves at least two types of epigenetic changes. First, cells may change in their competence to respond to CDF by habituating, and, second, competent cells may be induced to express the habituated state. Expression of the habituated state appears to be the less stable of the two epigenetic changes; competent cells in primary expiants of pith readily habituate in culture when triggered by CDF or 35°C treatment. Different states of competence, on the other hand, are extremely stable, even after many cell generations in culture, and appear to change only when triggered by events in organogenesis.

C Positive Feedback Models for Habituation

The fundamental question that arises is how can cells inherit different states of habituation when they appear to have the same genetic constitution. It has been proposed that heredity of this type may have as its basis the intrinsic properties of cellular regulatory mechanisms which can exist in alternative stable states (Delbrück, 1949; Kacser, 1957; Monod and Jacob, 1961; Rosen, 1973). Consider, for example, a metabolic pathway in which some intermediate, directly or indirectly, induces its own synthesis. Provided there is some way to degrade the intermediate, a metabolic pathway containing this type of positive feedback loop can exist in two alternative, self-perpetuating, stable states: one in which the concentration of the intermediate is zero; the other in which the concentration of the intermediate is equal to some steady state value greater than zero (Kacser, 1963). Although the kinetic properties of the system depend on the enzymes involved, and hence are specified genetically, the state expressed reflects the past history of the cell. Provided the positive feedback loop is not interrupted when the cells divide, two cells which are genetically identical could inherit different phenotypes.

Direct experimental evidence that positive feedback loops can generate alternative, epigenetic states has only been obtained in prokaryotes, for example, pseudomutation of the lac operon in galactoside-induced Escherichia coli (Novick and Weiner, 1957). Nevertheless, there are several examples of positive feedback relationships in plants which have the potential of generating alternative, heritable states. These include diverse developmental and physiological processes, such as the autocatalytic production of ethylene by flower tissues of morning glory (Kende and Baumgartner, 1974), of cyclic AMP by aggregating cells of Dictyostelium discoideum (Shaffer, 1975), and of auxin by primary pith expiants obtained from tumor-prone hybrids of Nicotiana glauca × langsdorffii (Cheng, 1972). Auxin also induces auxin-autotrophic growth of pith tissues from tumor-prone shoots regenerated from crown gall teratomas of tobacco (Braun and Wood, 1976; Turgeon et al., 1976). Finally, there is evidence that the induction of flowering has self-perpetuating properties in certain photoperiodic plants (Lang, 1965).

Meins and Binns (1978) have incorporated these types of regulatory mechanisms in an heuristic model which specifies that the CDF-habituated state is maintained by a positive feedback loop in which CDF induces its own synthesis or inhibits its own degradation. According to this hypothesis, competent cells in primary expiants of pith are poised to produce CDF, but require this factor above some critical, intracellular concentration to induce this process. When an exogenous source of CDF is provided above this critical concentration, the cells shift to the habituated state and remain in this state so long as the self-sustaining feedback loop is not interrupted.

This hypothesis leads to several predictions which can be tested experimentally. First, it should be possible to induce habituation by treating competent tissues with CDF. Meins and Lutz (1980a) incubated expiants of Havana 425 tobacco pith with increasing concentrations of kinetin and then scored the expiants for habituated nodules. They found that below ca. 10−9M kinetin, which is 1/100 to 1/1000 the concentration optimal for the growth of nonhabituated tissues, none of the expiants habituated. Above 10−9 M, the incidence of habituation increased with concentration to 100% at 10−7M. Thus, in agreement with the hypothesis, CDF above a critical threshold concentration induces CDF habituation.

A second prediction is that cells depleted of CDF by blocking CDF production momentarily should return to the nonhabituated state. To verify this, Meins and Binns (1978) grew cloned, CDF-habituated tissues for several transfers under nonpermissive conditions thought to prevent CDF production by the cells. The tissues were then shifted to permissive conditions and assayed for habituation. Since there is a carry over of CDF when tissues are subcultured, parallel experiments were run in which tissues were incubated under nonpermissive conditions with either an optimal concentration of kinetin or concentrations of kinetin that support slow growth but are below the critical threshold needed to induce habituation. CDF production was blocked in two ways: by incubating cloned, cold-sensitive tissue at 16°C (Binns and Meins, 1979) and by subculturing tissues on media containing low concentrations of auxin, conditions under which CDF-habituated tissues of tobacco require an exogenous source of CDF for growth (Tandeau de Marsac and Jouanneau, 1972; Einset, 1977, Binns, 1979). Although initial attempts to reverse habituation by low-auxin treatment gave erratic results (Meins and Binns, 1978), in later experiments reversal occurred regularly (Binns, 1979). The results obtained with low auxin and 16°C treatments were essentially the same. Tissues incubated on high-kinetin medium under nonpermissive conditions remained habituated when assayed under permissive conditions, whereas tissues incubated on low-kinetin medium shifted to the nonhabituated state. Moreover, tissues in this state regained their habituated character when incubated on high-kinetin media under permissive conditions.

A third prediction of the positive feedback hypothesis is that when tissues are induced to habituate by treatment with exogenous CDF, the tissues should produce more CDF than was added, i.e., CDF production should be autocatalytic. This prediction has not, as yet, been tested experimentally. There is evidence that some CDF-requiring tissue lines produce CDF in response to the added growth factor, while others do not; but, these studies were not performed with tissues competent to habituate rapidly in culture (Wood and Braun, 1967; Dyson and Hall, 1972).

There is evidence that auxin habituation also involves some sort of positive feedback mechanism. For example, pith expiants obtained from tumor-prone hybrids of Nicotiana require auxin for continuous growth on an otherwise complete medium (Cheng, 1972). These tissues shift to the auxin-autotrophic state when treated with auxin, and this shift is accompanied by an induction of 3-indoleacetic acid (IAA) production. Shifts between the auxin-requiring and habituated state in response to IAA have also been described for permanent tissue lines. The uncloned tissue line, T22, isolated from Bright Yellow tobacco is CDF habituated and can be serially transferred in culture on media containing 1 mg/liter of IAA as a source of auxin (Syōno and Furuya, 1971). These tissues do not grow on media without added auxin. Syōno and Furuya (1974) found that when T22 tissues were placed on a medium in which the auxin concentration was lowered to 0.1 mg/liter IAA, after 8 weeks — the optimum time for the effect — 56% of the expiants had shifted to the auxin-habituated state. These tissues were extremely stable; they have been serially propagated for over 5 years on medium without added auxin or CDF. Nevertheless, they regularly reverted to the auxin-requiring state when transferred onto medium containing 1 mg/liter IAA, the concentration needed to culture nonhabituated tissues. Moreover, it was found that tissues could be cycled through stable habituated and auxin-requiring states by sequential treatment with low and high concentrations of IAA. Similar results were obtained with a cloned, CDF-habituated tissue line, XD6S2, isolated from the Xanthi cultivar of tobacco. Although IAA was ineffective in this tissue line, the synthetic auxins, naphthaleneacetic acid (NAA) and 2,4D, when present at fairly high concentrations, regularly induced habituation. In this case, NAA was effective after only 1 week of treatment.

Although the details of induction and reversal of habituation for auxin and CDF differ, in both cases the experimental results suggest that the habituated phenotype is maintained by metabolic, regulatory systems capable of exhibiting alternative, stable states. The model proposed by Meins and Binns (1978) specifies positive regulation of growth factor production and is supported by indirect experimental evidence in the case of CDF habituation and the induction of auxin autotrophy in pith expiants isolated from tumor-prone shoots of tobacco. Similar models, which specify positive feedback loops involving the production of auxin and its conversion to inactive metabolites, could, in principle, account for the results of Syōno and Furuya (1971) as well.

D Genetic Changes

1 Chromosomal Mutation

Plant cells serially propagated in culture usually undergo progressive, heritable changes in gross chromosomal constitution (D’Amato, 1977). Fox (1963) proposed that habituation results from specific chromosomal changes of this type. His conclusion was based on a comparison of nonhabituated, Wisconsin 38 tobacco pith tissue with a variant habituated for both auxin and CDF. Whereas the chromosome number of the nonhabituated line varied widely from 48, the normal diploid number, to greater than 221, the distribution obtained with the habituated line was narrow; 80% of the cells were in the range 71–100. These results were confirmed first with uncloned (Sacristán, 1967) and later with cloned (Sacristán and Lutz, 1970) Ta tissue lines of White Burley tobacco that were auxin autotrophic, but are now thought to be of crown gall origin (Sacristán and Melchers, 1977). On the other hand, using different isolates of habituated and nonhabituated tissue lines from the same variety of tobacco, Dons et al. (1974) found that habituated cells had a higher nuclear DNA content, and presumably higher chromosome number, than nonhabituated cells.

Although there is a correlation between growth factor autotrophy and chromosomal constitution, it is unlikely that these changes provide a mechanism for habituation (Sacristán, 1967; Butcher, 1977). First, the habituated state is not consistently correlated with a specific karyotype. Binns (1979) found no statistically significant correlation of degree of CDF habituation with chromosome number in different cloned lines of cultured Havana 425 tobacco tissues. Some highly habituated clones had a normal, diploid chromosome number, whereas other clones were polyploid and aneuploid. The same was true for nonhabituated clones. Second, auxin-autotrophic tissue lines induced to revert to the nonautotrophic state sometimes retain their characteristic chromosomal aberration. Thus, cells in plants regenerated from the habituated line, Ta, are still aneuploid but indistinguishable in their growth properties from cells cultured from plants of seed origin (Sacristán and Melchers, 1969). Similar results were obtained with bona fide auxin-habituated tissues of Crepis capillaris (Sacristán and Wendt-Gallitelli, 1971). In this case, cells in plants regenerated from habituated tissues with a specific chromosomal aberration, Unisat, retain this aberration but exhibit the nonhabituated phenotype. These experiments show that even when a specific change in karyotype is associated with habituation, this correlation is not sufficient to establish a causal relationship between the two events.

2 Nonclassical Mutations

While the directed, reversible nature of habituation excludes classical somatic mutation as a mechanism, it does not necessarily exclude conservative rearrangements in the sequential order of genes or increases in the number of gene copies. Nonclassical mutations of this type are particularly well documented in higher plants (Sand, 1961; Brink, 1973; Fincham, 1973; Fincham and Sastry, 1974; Nevers and Saedler, 1977). Genetic studies with maize provide evidence that transposable genetic elements can migrate to new locations on the chromosome and regulate the expression of the genes with which they are associated (McClintock, 1956). Genes regulated in this fashion can undergo directed, heritable changes during early development, and these changes persist in the germ line (Fowler and Peterson, 1978). Transposable genes have also been found to account for mating-type determination in yeast (Hicks et al., 1977, 1979). Finally, in animals, there is direct evidence for the rearrangement of DNA base sequences during the differentiation of antibody-producing cells in the mouse (Seidman and Leder, 1978) and for dramatic heritable changes in the number of dihydrofolate reductase genes in cultured lines of mouse and Chinese hamster cells (Alt et al., 1978; Nunberg et al., 1978).

Recent studies in my laboratory suggest that directed genetic changes may be involved in the habituation process. We have found that certain tissue-specific states of habituation are transmitted through the seed (Meins and Lutz, 1980b). Cortex cells isolated from Havana 425 tobacco plants are CDF habituated, leaf cells are not (Meins and Lutz, 1979). To find out whether or not these states are erased when cells are induced to form plants, we compared the kinetin requirement of leaf and cortex tissues from plants of leaf and cortex cell origin. Three of the four types of tissues exhibited the kinetin requirement of the comparable tissue from seed grown plants: cortex tissues of leaf and cortex origin were habituated; leaf tissues of leaf origin were not habituated. In striking contrast, leaf tissues of cortex cell origin were habituated. These results provide direct experimental evidence that plants regenerated from two different totipotent cell types — in this case leaf and cortex — are not necessarily equivalent. Moreover, because only leaf tissues of cortex origin were habituated, it appears that leaf cells remember the habituated character of the cortex cells from which they were derived.

Plants regenerated from four cortex clones and four leaf clones were allowed to flower, self-pollinate, and set seed. We then germinated the seed and assayed leaf tissues from 15 mature plants derived from each clone for habituation. Leaf tissues of progeny from leaf-derived plants were always nonhabituated. In contrast, leaf tissues of progeny from cortex-derived plants fell into two classes. Of the 60 progeny scored, 38 exhibited the abnormal habituated phenotype, and the remainder exhibited the nonhabituated phenotype. Thus, the habituated phenotype characteristic of leaves from cortex-derived plants is transmitted through seed and appears to segregate in the progeny.

Although more detailed breeding experiments are needed to establish the point, the results suggest that the two types of plants, and, hence, the original cortex and leaf cells, differ genetically. To date, our experiments have been performed with cells and plants derived from just one tobacco plant. Stable genetic chimeras affecting specific tissue layers are well documented in plants and can arise by somatic mutation in populations of cells in the apical meristem (e.g., Stewart, 1978). This type of explanation can only be ruled out when the experiments are repeated with a number of seed grown plants. The more exciting alternative is that the observed differences between cortex and leaf cells arise by a directed mutation which normally occurs in the course of development.

IV HABITUATION AND CROWN GALL TUMOR TRANSFORMATION

Tumors result when susceptible plant cells conditioned by wounding are exposed to a tumor-inducing principle (TIP) produced by the bacterium, Agrobacterium tumefaciens (Braun, 1947; Braun and Mandle, 1948). These tumors continue to exhibit their capacity for autonomous growth in the absence of the inciting bacterium and form transplantable tumors when grafted onto a suitable host plant, indicating that the bacterium induces a true neoplastic transformation (Braun and White, 1943; Camus and Gautheret, 1948; Braun, 1953).

There are striking formal similarities between habituation and crown gall transformation. The tumor phenotype, once established, persists when cells are cloned indicating that tumor transformation, like habituation, results in changes inherited by individual cells (Braun, 1959). Conversion of normal cells to fully autonomous tumor cells involves a change in their growth factor requirement for proliferation in culture. For example, whereas tobacco pith cells have an absolute requirement for an exogenous source of auxin and CDF, crown gall tumor cells derived from pith cells can grow on media without these growth factors (Braun, 1956). Moreover, as a result of transformation, it appears that cells acquire the capacity to produce both auxin and CDF in quantities sufficient to support the growth of normal tissues (Kulescha and Gautheret, 1948; Braun and Naf, 1954; Wood, 1970; Miller, 1974, 1975; Pengelly and Meins, 1978). Thus, habituated and crown gall cells exhibit similar capacities for autotrophic growth.

Crown gall tumor transformation is progressive, i.e., tissues increase in growth autonomy with increasing time of transformation. By interrupting the transformation process with high temperature, it is possible to obtain a graded series of Vinca rosea tumor tissues which maintain their characteristic rates of growth in culture (Braun, 1943, 1947). Studies of the nutritional requirements of these tissues provide good evidence that progression involves gradual and, in some cases, independent activation of several biosynthetic capacities of the host cell necessary for the growth of normal Vinca tissues at an optimal rate (Braun, 1958). Progressive, gradual changes of this type are encountered in habituation as well (Meins and Binns, 1977).

There is convincing evidence that certain crown gall tumor cells can lose their malignant properties and return to the normal state (Braun, 1959; Sacristán and Melchers, 1977). Crown gall teratomas of tobacco have a pronounced capacity for differentiation and for highly abnormal organogenesis. Under inductive conditions provided by grafting, progeny of individual teratoma cells eventually give rise to normal, fertile tobacco plants (Braun, 1959). This process apparently proceeds in two steps (Braun and Wood, 1976; Turgeon et al., 1976; Wood et al., 1978). First, during organized growth of shoots from teratoma cells, expression of the tumor state is gradually suppressed resulting in tissues, normal in appearance and growth regulation, that retain their neoplastic character in a covert form. Second, during meiosis, teratoma-derived cells undergo a second change and completely lose their capacity for autonomous growth. Thus crown gall transformation, like habituation is potentially reversible and leaves the heritably altered cell totipotent. The exact pathway by which reversal is achieved, however, differs for the two processes. In the case of CDF habituation, the pathway apparently depends on the origin of the cells. Habituated pith cells revert at some point during plant regeneration. Habituated cortex cells, on the other hand, give rise to leaf cells during plant regeneration, which, like crown gall cells, retain their capacity for autotrophic growth in a covert form. These cells give rise to revertants during meiosis or the formation of seed, probably as a result of genetic segregation.

The pathognomonic feature of plant tumor cells is their capacity to form large, non-self-limited growths when grafted onto a suitable host. Tissue lines habituated for auxin as well as CDF vary in their tumorigenicity. In experiments with tissues of European grape (Vitis vinifera L. var. Aramon), Braun and Morel (1950) found that grafts of habituated tissue increased in size somewhat, but none of the 51 successful grafts produced continuously growing tumors of the type obtained with crown gall tissue. Henderson (1954) has reported that habituated sunflower tissues give rise to small overgrowths on grafting which are intermediate in size between normal and crown gall tissues. Limasset and Gautheret (1950) grafted an habituated tobacco line described by Morel (1948), normal callus, and crown gall tissues of tobacco onto the side of the stem of young tobacco plants. They found that habituated and crown gall tissues formed large, tumorous overgrowths of comparable size, whereas normal tissues did not. Similar positive results were obtained with habituated tissues of tobacco and Nicotiana glutinosa by Hirth and Durr (1971) and Ackerman et al (1973). Lutz (1966, 1971) has reported that different clones isolated from Morel’s (1948) habituated tobacco line vary in tumorigenicity. One clone, S17, was highly tumorigenic and grew rapidly on basal medium suggesting that the cells were fully transformed. Another clone, S21, formed few tumors when grafted and grew with difficulty on basal medium, suggesting that it was partially transformed, while other clones were nontumorous and did not grow on basal medium. These experiments are difficult to interpret owing to what may have been a misidentification of the cell lines. Sacristán and Melchers (1977) point out that their habituated tobacco line, Ta, thought to be the same line used by Lutz (1966, 1971) is of crown gall origin. Ta from Melchers’ laboratory contains DNA sequences of Ti plasmid origin (Yang et al., 1981) and produces octopine (Wendt-Gallitelli and Dobrigkeit, 1973), a specific marker for crown gall cells, whereas other isolates of habituated, White Burley tobacco do not (Bomhoff et al., 1976). Binns* has recently confirmed Lutz’s experiments with habituated clones isolated from the Havana 425 variety of tobacco. A line habituated for both CDF and auxin, 156AH, was obtained by selecting for tissues able to grow on basal medium from the CDF-habituated clone, 156H (Binns and Meins, 1973). Binns grafted tissues from six clones of 156AH origin and found that four clones formed large overgrowths comparable to those obtained with crown gall tissues. Tissues of the remaining two clones fused with the host plant, indicating that the grafts had taken, but did not grow. These results, and those cited earlier, show that some, but not all, lines of habituated tobacco and N. glutinosa tissue are highly tumorigenic in grafting tests.

Comparison of the properties of habituated and crown gall cells leads to the conclusion that habituation is a form of neoplastic transformation involving heritable, progressive changes in cell phenotype that can result in autonomous growth. The significance of this conclusion lies in the fact that habituation occurs in the absence of a recognizable infectious agent and appears to have an epigenetic basis. This implies that tumor transformation in plants, and perhaps generally, does not necessarily involve either the introduction of foreign genetic information or somatic mutation that result in a permanent change in the cell genome (Braun, 1978). Direct confirmation of this important hypothesis depends upon demonstrating that cloned, habituated tissue lines are tumorigenic in grafting tests and that the same tissue lines satisfy the essential criteria for epigenetic changes.

Habituation also bears on the role of TIP in initiating and maintaining the tumorous state in crown gall. The capacity of crown gall bacteria to induce tumors is carried by plasmids (Van Larebeke et al., 1975; Watson et al., 1975). These Ti plasmids also specify which opine amino acid, e.g., octopine or nopaline, is produced by the tumor and catabolized by the inciting bacterium (Bomhoff et al., 1976). Permanent axenic lines of tumor cells, but not normal cells, contain DNA base sequences of Ti plasmid origin (Chilton et al., 1977). Thus, it appears that plasmid genes are transferred to the plant cell during tumor inception. Although the evidence is not, at present, conclusive, it is likely that DNA sequences of plasmid origin present in tumor cells are TIP or closely related to this factor. Braun (1958) proposed that the production of substances required for cell proliferation results in the autonomous growth of crown gall tumor cells. The relationship between the loss in requirement for exogenously supplied growth factors and tumor autonomy also exists in habituation and in tumorigenesis of tumor-prone genetic hybrids of Nicotiana, which do not involve infectious agents. This implies that the activation of growth factor production by enhanced synthesis, decreased production, or a combination of both processes is necessary and sufficient to account for the autonomous growth of plant tumor cells (Meins, 1974). If this is the case, then the primary site of TIP action in crown gall transformation would appear to be the production of growth factors or its regulation.

In principle, two alternative types of hypotheses could account for changes in growth factor production induced by TIP. Crown gall bacteria have an exceptionally wide host range; numerous dicotyledonous and some monocotyledonous species in at least 331 genera are susceptible to transformation (DeCleene and DeLey, 1976). These species, in some cases at least, differ in the requirement of their cultured tissues for growth factors. For example, whereas tobacco pith tissues require auxin and CDF for growth, normal Vinca rosea tissues require, in addition, myo-inositol, amide amino acids, purines, and pyrimidines for optimal rates of growth (Braun, 1958). The important point is that transformation to the fully autonomous state in both cases involves a loss in the exogenous requirement for all of the growth factors that normal tissues of the same species require (Braun, 1956, 1958). To account for this, one may propose that the various biosynthetic activities required for cell proliferation are controlled by a master switch with a similar susceptibility for TIP in the different host species. The alternative hypothesis is based on the notion that tumor initiation and progression involve chance events followed by cell selection (Comings, 1973; Cairns, 1975; Nowell, 1976; Watson, 1977). According to this hypothesis, normal cells are perturbed during tumor inception and give rise to an heterogenous population of heritably altered cells. Some of these cells produce growth factors required for proliferation. When, by chance, cells arise that produce all of the factors for growth they have a selective advantage, proliferate, and form a clone of autonomous tumor cells. These hypotheses lead to specific predictions that can be experimentally verified. If tumor initiation involves the chance activation of several different biosynthetic capacities in the same cell, then it should be possible to find cells arising during tumor inception that produce less than the full complement of growth factors and are incapable of autonomous growth. If, on the other hand, tumor initiation involves a master switch mechanism, then none of the cells incapable of autonomous growth should produce the tumor-specific growth factors.

Perhaps the simplest interpretation of tumor transformation is that TIP is a self-replicating element which persists in the transformed cells and contains genes specifying growth factors required by the plant cells. This is unlikely, since it does not account for the fact that the same strain of crown gall bacterium can effectively transform plant species with different growth factor requirements for cell proliferation. It is also difficult to reconcile with the observation that certain attenuated strains of bacteria only initiate tumors when the host plant is provided with an exogenous source of auxin (Braun and Laskaris, 1942). Another possibility, which is consistent with the master switch hypothesis, is that TIP-associated genes code for specific inducers that act at a common site in the different plant species to regulate growth factor production.

The perturbation–selection hypothesis leads to a very different interpretation of TIP action. In this case, each transformation event may well be unique. For example, stable transformation could involve the insertion of one or more genetic elements into different sites in the genome. Whether or not these elements are transcribed, their physical presence may, by chance, perturb the regulation of pathways involved in growth factor synthesis and lead to the persistent expression of the tumor phenotype.

Whatever the exact molecular mechanism for TIP action, it is likely that expression of the tumor phenotype has an epigenetic basis. This is the case for certain animal tumors in which viruses are the etiologic agent and the continued presence of viral genes are required for autonomous growth. For example, mammalian cells transformed by the oncogenic DNA viruses, SV40 and polyoma, occasionally revert to normal type in culture but retain the viral genes (Pollack et al., 1968; Rabinowitz and Sachs, 1968, 1970). Mouse blastocyst embryos microinfected with SV40 DNA develop into adult mice with SV40 DNA sequences in some tissues. Nevertheless, these animals do not develop tumors (Jaenisch and Mintz, 1974). When mouse embryos are inoculated with Moloney leukemia virus, the viral genome is introduced into many cell types, including the germ line; however, only the appropriate stem cell lines are transformed (Jaenisch, 1976).

With regard to crown gall, the strongest evidence for this type of epigenetic regulation comes from the discovery of the suppressed tumor state in tobacco (Braun and Wood, 1976; Turgeon et al., 1976). Cloned teratoma tissues, when grafted onto the cut stem of tobacco plants, from which the axillary buds have been removed, give rise to shoots which gradually become more normal in appearance. Near the top of such shoots, the leaves are perfectly normal in gross morphology, physiology, and histology. Nevertheless, when expiants of these leaves are placed in culture on a basal medium that does not support the growth of normal tissues, the cells at the cut margin proliferate and give rise to typical teratoma tissues. Leaves with this potentiality for tumorous growth also retain the capacity to produce nopaline, indicating that genes of plasmid origin are present and expressed in cells which do not express the tumor phenotype (Wood et al., 1978). Leaves arise ontologically by cell division and enlargement from a small number of initials in the apical meristem (Esau, 1977). Although it is not known, at present, what proportion of teratoma-derived leaf cells are potentially tumorous, the fact that these cells retain plasmid genes but remain suppressed when they arise by proliferation of initials provides evidence that expression of growth autonomy depends upon the epigenetic state of the cells.

Hints as to the nature of the interaction of epigenetic and genetic factors in crown gall transformation are provided by comparing the physiology of habituated cells and crown gall cells proliferating under similar conditions in culture. For example, CDF-habituated tissues revert to the nonhabituated state when incubated under conditions believed to temporarily block growth factor production (Meins and Binns, 1978). A similar approach has been used to promote the normalization of crown gall teratoma cells in culture. Cloned teratoma line B of Turkish tobacco origin requires an exogenous source of auxin for proliferation at 35°C but not at 25°C, the normal growth temperature (Meins, 1975b). In the original experiments begun in 1972, preincubation of line B tissues at 35°C on auxin-less medium resulted in decreased growth autonomy which persisted at 25°C but was slowly lost after several successive transfers. These results suggest that the same type of epigenetic control mechanism may be operating in crown gall cells as in habituated cells. Unfortunately, the experiments are equivocal. Over the years, the teratoma line has changed dramatically in growth factor requirement (Pengelly, 1980), and recent experiments have shown that this line, which is now only slightly temperature sensitive, no longer reverts at 35°C*

Indirect evidence for the similarity of growth regulation in habituated and crown gall cells is provided by studies with antimetabolites. The thymidine analog, 5-bromodeoxyuridine (BUdR) selectively prevents the expression of differentiated functions in diverse types of animal cells (Rutter et al., 1973). BUdR also specifically blocks expression of the CDF-habituated phenotype by tobacco cells (Meins, 1976). This effect is reversible, is accompanied by incorporation of 5-bromouracil into cellular DNA, and is prevented by thymidine treatment. Vyskot et al. (1977) confirmed that CDF-habituated tissues are BUdR sensitive, but found that two different lines of crown gall tissue and tissues habituated for both auxin and CDF were BUdR tolerant. It appears, therefore, that the transition of tobacco cells to the fully autonomous state, whether by habituation or crown gall transformation, involves a shift to a BUdR-insensitive type of growth