Cell Genetics and Hereditary Symbiosis

JOSHUA LEDERBERG

    Fmrr~ /he Depnrlfnenl of Genelics, I Unizwsity oj Wisconsin, Madison, Wisconsin
  ITCENT SYMPOSIA II.IVE been attended by controversies over the interpretation
w and scope of cytoplasmic inheritance, often centering on the diagnosis of
   virus 2's. plasmagene (5, 42, 43, 45, 48, 55, 197). This review is an attempt
to reconcile these views and, where possible, to look for further implications of the
synthesis. This account is (quite properly) limited by space allowances. Therefore,
objective reviews which already survey the experiments will be cited, when available,
in lieu of original references. The treatment of material which has not been condensed
previously is therefore exaggerated. An eclectic construction from the freely borrowed
views of many authors has been attempted, and few of the interpretations are
original.

The problems of cytoplasmic heredity have often merged with those of embry-
onic development and somatic differentiation (56, IIS, 125, 182, 183, 202, 203, 212,
220). The genetics of somatic cells has, however, been poorly studied. Their consti-
tution has been deduced from that of the zygote or the whole organism, except for a
few, but important, studies on somatic mutation (66, 119, 123) and somatic segrega-
tion (83, 88, 189). Cytological studies have taken some exceptions to the genetic
uniformity of somatic cells of the differentiated organism, but have not thereby con-
tributed affirmatively to a comprehensive theory of development (83, 121, 220).
Although the pertinence of extranuclear heredity for ontogeny has been generally
accepted, in the absence of extensive evidence for it, plausible theories of develop-
ment have been constructed which rely on the ultimate primacy of the nucleus
(73, 131, 193, 202, 220). For technical reasons, cell genetics is best studied in or-
ganisms whose germ and soma are not irreversibly differentiated, especially in the
microorganisms where vegetative proliferation is preeminent. Whether methodo-
logical obstacles with extranuclear heredity factors exaggerate the seeming dominant
role of the nucleus in the genetics of higher animals and plants has been debated
elsewhere (37, 55, 90, 131, 181, 220).

These discussions have left a plethora of terms adrift: pangenes, bioblasts,
plasmagenes, plastogenes, chondriogenes, cytogenes and proviruses, which have lost
their original utility owing to the accretion of vague or contradictory connotations.
At the risk of adding to this list, I propose plasmid as a generic term for any extra-
chromosomal hereditary determinant. The plasmid itself may be genetically simple
or complex. On occasion, the nuclear reference of the general term gene will be em-
phasized as cltromogetze (2 rgj.

This review is dedicated to the reconciliation of the attitudes that plasmids
are symbiotic organisms, and that they comprise part of the genetic determination
of the organic whole. The conflict may arise in part from fixed conceptions of the
scope of the organism. Heuristically, the taxonomic classification of plasmids as
viruses, symbionts, or plasmagenes should not obscure careful descriptions of their
function, hereditary or pathological, or both. This viewpoint need not prejudge

' Paper no. 497.

403


404                JOSHUA LEDERBERG             Volume 32

the evolutionary origin of different plasmids, although it may suggest possibilities of
wider connections.

GENETIC CONTINUITY OF CYT~PIASMIC ORMNELLES

  Various structures that are characteristic of almost all cells have been suspected
of hereditary functions from morphological evidence of their genetic continuity. The
history of the nucleus in gametogenesis and fertilization, and the constant number
and individuality of the chromosomes convinced many cytologists of their role in
heredity well before the rediscovery of Mendel's work (55, 21 I, 212). However, it
was not until the parallelism in the distribution of the chromosomes and the chromo-
genes was affirmed even for exceptional deviations (129) that their identification
convinced most biologists. No extra-chromosomal organelle has met this criterion
except in the most rudimentary way. Within the nucleus, the nucleolus has been
thought to be continuous but is now regarded as a metabolic product concentrated
at a special chromosome region (38, 122). The nuclear membrane is continuous
throughout mitosis in many protista, but has not exhibited any of the differentiation
needed fdr genetic analysis.

  The mitochondria have stimulated the greatest interest, but little has been
added to our conception of their genetic functions in animal cells since Wilson's
discussions over twenty-five years ago (135, 213, 214). Their development during
spermatogenesis in scorpions has furnished compelling suggestions but no proof of
their role in heredity. In Centrurus the mitochondria aggregate in the spermatocyte
to form a single large ring which is accurately partitioned at meiosis. In Opisthacan-
thus, twenty-four spheroids are formed which are equally distributed, six to a sperma-
tid. Despite his contrary conclusion, Wilson's statistics show that the distribution is
more precise than random. The small size and great number of the mitochondria
have made it very difficult to study their morphogenesis in somatic cells. The genetic
relationship of mitochondria to the submicroscopic microsomes is still controversial;
part of the life history of a mitochondrium may range beyond microscopic visibility
to be confused with de nova origin. Our appreciation of the chemical constitution and
enzymatic activity of cellular particulates as a whole (40, 75, 135, 168, 213) is more
advanced than their taxonomy. The distribution of properties among the particles
collectively labelled mitochondria is an urgent question for the geneticist.

  Many workers early described the origination of mitochondria from nuclear
fragments-the chromidial theory now largely discredited (135). Current opinion
emphasizes the continuous functioning of the energic nucleus (38, 73, 143, 190)
rather than an intermittent outpouring at karyokinesis or karyolysis. The chromidia
have a counterpart in gene-initiated but autonomous plasmagenes, postulated as
one mechanism of nuclear control (45, 219, 220), but this theory has not been sub-
stantiated to the exclusion of nuclear regulation (rather than origination) of pre-
existing plasmids.

  The genetic continuity of other cytoplasmic inclusions is even less well verified
than that of the mitochondria. The chloroplast, often regarded as a mitochondrial
derivative, will be reviewed in detail below.

  Special attention has lately been given to the fiber-organizing granules which
function as centrioles, blepharoplasts, parabasal bodies, basal granules etc. (I I 5))
for which the general name kinetosome has been suggested. Although different kineto-
somes may serve specialized functions in the same cell, several substitutions of func-
tion have been described. For example, the kinetochores of degenerating chromosomes


October 195-7     CELL GENETICS AND HEREDITARY SYMBIOSIS         40.5

of apyrene spermatids in viviparid snails become detached to function as accessory
blepharoplasts in the development of multiflagellated spermatozoa (145). However,
the spermatogenous mitoses of mosses and ferns, where centrioles and blepharoplasts
appear to arise cle UOYO (21 z), raise the question of the strict genetic continuity, or
at least the individuality of the kinetosomes generally. Although the organized
kinetosome system of a ciliate may function in morphogenesis as a community of
mutually regulatory, but individually autonomous parts (80, IIS, 202, 203), their
ultimate potential origin from a common indifferent reservoir is not excluded. The
best evidence of the individual autonomy of a kinetosome is the formation of apara-
basal trypanosomes when its division is impaired by trypaflavine or colchicine (144).
More work needs to be done on the details of this impairment, but the.loss appears
to be strictly irreversible. A possible direction for more detailed consideration of the
kinetosome (which is frequently reported to be Feulgen positive) as genetic units
is suggested by a Drosophila mutant in which kinetochore action is impaired, though
not abolished (9).

CYTOPLASMIC INHERITANCE

  The profusion of terms nearly equivalent to plasmid already points to the di-
vergence of the morphological and genetic exploration of the extra-nuclear territory.
The literature on cytoplasmic inheritance has been recently and critically reviewed
elsewhere (13, 37, 42, 43, 45, 55, 56, IIS, 125, 181, 182, 219, 220) so that only a few
of the most illustrative studies need to be restated here. Cytoplasmic and extranuclear
are misleading locatives. A resident of the nuclear sap (e.g. 199) would often be gene-
tically transmitted no differently from a cytoplasmic factor. In fact, supernumerary
chromosomes whose segregation is irregular, and may differ in oii- and spermatogene-
sis, might give genetic results readily mistaken for cytoplasmic transmission. The
concept of the plasmid is meaningful only in terms of regular, well-understood nuclear
behavior, and various claims of nonmendelian inheritance must be re-evaluated on
this criterion (106, 107, 162, 216).

CHLOROPLASTS: PLASTOCENES AND VIRUSES

  The most extensive literature on extranuclear inheritance concerns the chloro-
plasts of seed plants. Many mutants affect the chloroplasts and the synthesis of
chlorophyll, but most are recessive lethal or semi-lethal mendelian factors (76, 159,
172, 200). We shall deal here with variants that show maternal or matroclinous
inheritance, so that plasmids are implicated (157, 159).

  Morphological evidence of chloroplast continuity is inconclusive (135). Mature
plastids may divide as such, or may develop from mitochondrial primordia in meri-
stematic cells. The genetic duality of the mitochondria and pro-plastids is contro-
versial and not yet supported by morphological discrimination. Plastid genetics has
mainly dealt with plastid variegations, some of which resemble virus diseases (218).
Affected plants show a regular or irregular mosaic of normal and chlorotic tissue.
Seed from the white shoots will usually give moribund white seedlings, whereas the
pollen may function normally to give standard green progeny when crossed to normal
green. The demonstration of virus infection by mechanical or graft transmission to
healthy plants has disqualified a few cases originally thought to belong in this group
(19). The plastid variegations have been interpreted by many workers as indicating
a segregation of mutant white plastids, which arise from a highly unstable green
plastid condition. As a rule, however, the green and white plastids are not well de-


406                JOSHUA LEDERBERG

lineated in single cells, and there may even be a transition zone several cells deeu at
the margin of white patches in which the plastid condition is intermediate. Other
workers have therefore concluded that the variegations are not inherent in the plastids
but in another element of the cytoplasm. The term plaslogelfe was introduced to
describe the potentially mutable genetic unit of the plastid according to the first
hypothesis, but will be useful for present discussion to denote the determinant of
plastid behavior, whatever its form or location in the cytoplasm (84, 159, 200, 218).

To restate the problem, are the plastogenes confined to the plastids (hypothesis
of genetic autonomy) or distributed independently in the cytoplasm (a generaliza-
tion of the second hypothesis)? Are the plastogenes elementary units or would it
be profitable to investigate their genetic complexity by combination experiments?
If the plastogene is located on the plastid itself, is it also autonomous in function,
or is the phenotype of all the plastids in one cell regulated as a whole by the plasto-
gene aggregate? The last consideration weakens the force of the blending of plastid
phenotypes within single cells where this occurs (but see 218).

The stability as well as the function of the plastid system is under chromogenic
control as shown by recessive genes which initiate plastid variegation (S4, 159, 2 IS).
In maize, for example, homozygous iojap plants, obtained by selling normal green
heterozygotes, show a marked plastid instability that is reflected in a green and white
mosaic. By further crossings, the iojap can be replaced by wild type genes, but the
modified plastid character persists in maternal inheritance. Catnip (Ly~p~I~) may be
more favorable material for similar studies, because it can be kept indefinitely by
vegetative propagation, and self-inviable types can be grafted to green stocks. Woods
and duBuy have described a wide range of qualitatively distinguishable plastid types
which develop under the influence of a recessive chromogene analogous to iojap
(218). These authors have postulated hereditary modification of the plastid pri-
mordia (chondriogenes) for the origin of plastid variegations and transmissible
viruses. However, the effective plastogenes have not been conclusively located in
the affected plastids. In other terms, we might say that a previously latent virus
(with the remarkable property of being cell-bound but seed-transmitted) had been
activated by the iojup genotype so as to modify the character of the plastids. It is
feasible to take advantage of the occasional transmission of plastid primorida through
the pollen to show that green plastids will recur in seedlings from w/G/e x greeu puri-
fied and maintained by grafting on green. The white x white cross would serve as
the necessary control. The technical features of this material also permit the combina-
tion of different plasmid genotypes.

Chloroplast mutants have also been described in irradiated material, of which
fern prothallia have been most studied (120). Unfortunately, a genetic analysis has
not been reported, so that the site of the mutation is not known.

Experiments with the chloroplasts or chromatophores of certain algae support
their genetic autonomy. In Euglena species, certain conditions of cultivation in the
dark on rich media led to a diminution in the chromatophores, occasionally followed
by their irreversible loss (114). However, the affected clones may have been in-
trinsically unstable, and the conditions of culture merely permitted the survival of
the apochromatophoric clones (152).

Chemotherapy is a promising new tool of plastid and plasmid research. Attempts
to remove bacterial contaminants from cultures of Euglemz gracilis with streptomycin
resulted instead in curing the flagellate of its chloroplasts (154). In some strains the
disappearance of the chromatophores was accompanied by that of the carotenoid


October rgjr     CELL GENETICS AND HEREDITARY SYMBIOSIS         407

eyespot or stigma; in others, the stigma persisted, but only so long as the culture was
illuminated. Once lost, neither chromatophores nor stigmata were restored, even upon
prolonged cultivation in the light. The success of this experiment depends, of course,
on the heterotrophic potency of the Euglena. Other algae are bleached, but cannot
survive without photosynthesis. In still further species, other functions are more
sensitive to streptomycin, and their inactivation precedes apochlorosis. The fate of
the pyrenoid and its relationship to the mitochondrial system of the bleached cells
are controverted (116, 152).

Unfortunately, sexual reproduction has not been described in Euglena, precluding
detailed genetic investigation, but other algae are under study. The similarity between
the artificial apochlorotic variants and naturally occurring Asfasia has given point
to hypotheses proposed for the origin of various colorless flagellates (114, 152).
Apochlorosis has also been achieved in some Euglena strains by temperature treat-
ments (I 53). The conditions suggest that the chromatophores are lost by not keeping
pace with cell division, rather than by their active destruction, as with streptomycin,

Some seed plants are also susceptible to streptomycin. As albino shoots can be
maintained in mosaics with or by grafts on green tissues, there are extensive possi-
bilities for genetic study with such material. Together with the plastid variegations,
these studies have already provided considerable evidence for the genetic autonomy
of the chloroplast. Such evidence is, however, by no means conclusive, and other
plasmids, not yet characterized, may be immediately responsible for the observed
effects.

Male sterility is maternally inherited in many plants. The pathology of pollen
abortion hints at a virus which might be transmissible, but despite its economic value
(59, 1$3), this has not been achieved. Unlike the plastid variegations, the male
sterility factor of maize does not sort out in somatic tissue, as shown by the absence
of chimaeras on the ears, and by a biometric study of the pollen of partly affected
plants (67, 158). Segregation does occur during or soon before microsporogenesis
(cf. the aggregation of scorpion chondriospheres, 214) so that some microspores
escape the plasmid and do not abort. The expression of male sterility and the main-
tenance of the plasmid are under rather complex chromogenic control (173, 89).
Two recurrences of the male-sterility plasmid have been reported in genetically
controlled maize cultures, one of them in an iojap genotype (160, 173). Its origin by
mutation of normal plasmids (mitochondria) has therefore been proposed, quite
plausibly. However, to the reviewer's knowledge, an infectious origin under field
conditions has not been explicitly disqualified.

CYTOPLXSNC MUTATION IX YEAST

Dwarf colony types had been sporadically observed in yeast cultures for many
years, but until Ephrussi's studies (57) little attention was paid to them. In the course
of experiments with chemical mutagens, it was noted that yeast suspensions exposed
to euflavine (2, %diamino-ro-methylacridine) produced large numbers of `petites
colonies' (PC). Despite the occasional spontaneous occurrence of the PC, a careful
population analysis, and later the direct examination of daughter buds from treated
cells showed unequivocally that the PC `mutation' was specifically, directly, and
prolifically induced by the acriflavine. The suspicion that the mutation recurred in
a cytoplasmic rather than nuclear particle was reinforced by the equal effect of
euflavine on haploid and diploid yeasts and verified by crosses of PC with wild type.
Such crosses regularly restore wild type zygotes whose further progeny is also nor-


408                JOSHUA LEDERBERG

mal, implving that the PC mutation occurs when a plasmid is irreversibly lost, as
acceleratei by acridine dyes. This depletion is not a direct destruction of the plasmid:
treated mother cells remained normal for some time after they had engendered a
series of I'C buds. Instead, the transmission of the plasmid is probably primarily
affected, although it may subsequently be directly modified in the mother cell. Since
euflavine is known to precipitate nucleic acids and agglutinate bacteria, it ma>
aggregate the plasmids or otherwise hinder their passage to the bud. Limited at-
tempts at artificial re-introduction of the wild type plasmids into PC cells have been
unsuccessful.

  The dwarf growth of the PC is a consequence of the lack of cytochrorne ositlase
and succinosidase associated with mitochondria. In the wild type yeast, some of the,
`mitochondria' give a positive indophenoloxidase (ru'adi) reaction; in the PC they
are all negative. This plasmid can therefore be plausibly considered as a genetically
self-determined set of mitochondria carrying the specified enzymes but, as Ephrussi
himself has cautioned, the evidence for the identification is inconclusive. The plasmitl
might correspond to a distinct cytoplasmic element necessary for the elaboration d
the oxidases, which are localized on various mitochondria. This need not be a gratui-
tous multiplication of determinants: The same phenotype is controlled by a chromn-
gene in a segregational PC mutant, but one does not identify the nuclear gene with
the oxidase system. In fact, the enzymes themselves are certainly not self-perpetuat-
ing as such, for dormant plasmids remain intact in the segregational PC mutant,
ready to co-act with a suitable gene substitution to redevelop the oxidase system
Thus, the copulation of a segregational PC with a vegetative PC mutant results in
a wild type zygote.

  An attractive theory of chemical carcinogenesis has been constructed on a similar
basis (148). The liver of normal rats contGns a protein that reacts with aminoazo-
benzene derivatives. Feeding on these dyes depletes the dye-binding protein, to-
gether with succinoxidase activity and the microscopically visible mitochondria.
The initiation of tumors is correlated with the total depletion of the reactive system.
Potter has speculated that a diversion of high-energy metabolites from regulated
physiological functions leads to unrestrained growth.

  Mitochondria are now a favored subject of biochemical research, with accumu-
lated evidence that biological oxidation sequences are coordinated in granules, cor-
responding to some or all of the mitochondria (40, 75, 135, 168). The metabolic
pattern of mitochondria differs from tissue to tissue in the same organism. \iThether
this differentiation involves the quality of individual granules, or the distribution of
diverse species is not known. The prospects for experimental study of genetic differ-
entiations of mitochondria are limited with material whose vitality depends on the
integrity of the mitochondrial system. The PC mutation would certainly be lethal
in fruit 0y or mouse and a similar argument might account for the paucity of other
plasmid demonstrations in higher forms. If induced hepatomas are viable cells analo-
gous to PC mutants, however, the door to such studies with animal cells is not closed,
and cytochemical specialists should be encouraged to correlate their work with
genetic studies.

  The slow adaptation of certain yeast strains to galactose straddles the per-
sistence of genetic variation and the transience of physiological response. The under-
standing of long-term adaptation has been confused by several different mechanisms
with superficially similar effects. The common datum is that various yeast strains,
which do not immediately ferment galactose, acquire the ability to ferment this sugar


Odober 19jr     CELL GENETICS AND HEREDITARY SVRfBIOSIS         409

over a period of several days. Adapted inocula, when maintained in the presence of
gnlactose, develop into promptly fermentin, (r cultures (184, 21 j). One yeast strain
becomes stably adapted after variable intervals of exposure to galactose. This
adaptation almost certainly consists of the selection of spontaneous mutants.

  \\-e are more concerned with other strains which respond to galactose after a
constant interval, and whose adaptation disappears abruptly when the cells are
eslwse(l to glucose. The de-adaptation has been analyzed both in mass cultures and
in single buds, with results that leave no doubt of the direct effect of the glucose, as
opposed to natural selection (1S4). But, unlike the I'C account, the mother cell is as
likely to be de-adapted as its bud. The kinetics agree with a model in which the
adapted cell carried about 100 plasmids, distributed at random to bud and mother.
A single plasmid will propagate in the presence of galactose until the normal level
is attained. \Vithout galactose, the plasmids remain dormant and are passively
apportioned to successive generations, until the average number per cell is too low
to assure that the nest generation will be uniformly infected. As might be surmised
irom the kinetic differences, these plasmids are distinct from those affected in the
I'(` mutants, and from the galactozymase complex itself, which disappears much
faster than adaptability. Other evidence also makes it very unlikely that an enzyme
is itself part of the mechanism of its adaptive formation (103, 128, 186).

  The chief inadequacies of this theory are that it does not encompass the original
mechanism of slow adaptation to galactose, and that this has not been resolxred in
terms of individual enzymes. In populations grown on glucose, about one cell per
thousand eventually responds to galactose. Competent plasmids might arise de
brow, i.e., from the nucleus or from other plasmids, but this ad Izoc assumption should
eventually be replaced by more precise testable specifications. The kinetic studies of
de-adaptation are, however, consistent with the plasmid-depletion theory which is
tenable in the absence of clearly formulated alternatives.

  Further study of the biochemical competences of individual cells would be fruit-
ful for embryolog& analogies, but very little is known of the physiological fluct-
uations in adaptability of microorganisms. The ability of Aerobacter aerogenes to
assimilate citric acid, presumably some sort of adaptive enzyme process, is physiologi-
cally variable from cell to cell, and responsive to nonheritable inactivation by ultra-
violet light (155). Similarly, the inhibition of enzymatic adaption of Pseudomonas
~~~u~~~.s~-PHs by ultraviolet light, and its restoration with visible light, might be inter-
preted in terms of an all-or-none response of individual cells (186). Unfortunately,
few systems are technically suitable for determining the adaptive status of a single
cell, although more use might well be made of microculture methods.

  (;enet ic studies with the relatively large animal, Parumecium aztrelia, are hardly
at all complicatccl by the populational selection difficulties sometimes encountered
with other material, but there is, unfortunately, a dearth of mutant chromogene
markers, especially for the biochemical characters which have been so fruitful with
the microthallophytes. Sonneborn and his co-workers have not permitted this diffi-
culty to obstruct their painstaking study of the nuclear and cytoplasmic hereditary
systems (178).

  The most distinctive cytological feature of most ciliates, including Paramecium,
is the dual nuclear system (179, 203). The micronucleus which alone is involved in
fertilization and meiosis, but is propagated mitotically during clonal growth, has


been compared with the germ line; the macronucleus which is derived anew from
the micronucleus after each fertilization, and whose fission is obscure, faith the soma
of metazoans. The somatic function of the macronucleus is seen by its regulation of
all gene-controlled functions during clonal proliferation in amicronucleate or allo-
karyotic animals. As a rule, the macronucleus is derived from the micronucleus at
the last preceding reorganization, and will be genetically equivalent to it. Under
certain conditions, or in the absence of a micronucleus, the macronucleus will be
regenerated from fragments of the previous macronucleus, irrespective of the geno-
type of the current micronucleus. The two nuclei may also come to deviate by mu-
tation. Paravneciwn is extraordinarily resistant to ionizing radiations, but s-rayed or
nitrogen mustard-treated lines may show a high mortality at the nest nuclear re;
organization. Much of this delayed effect may be accounted for by lethal mutations
or chromosome deletions whose phenotypic effect is masked by the macronucleus
as well as by dominance in themicronucleus. Finally, differentiationn-ithin themacro-
nucleus has been described in other ciliates as a response to organizing fields in the
cytoplasm (203).
  To the extent that the macronucleus may be regenerated from fragments of its
predecessor it becomes genetically independent of the micronucleus, and might
even be regarded as a giant plasmid. Macronuclear fragments have even been ob-
served, on occasion, to be transmitted to conjugant partners. As a rule, the micro-
nucleus retains its ascendance, but the persisting macronucleus is the most concrete
example of the gene-initiated plasmid. The hypothetical origin of plasmids from
macronuclear fragments recalls the chromidial theory but merits consideration on
its own merits.
  Clones of P. aurclia vary in the production of a poison, para7neii/l, and their
sensitivity to it. Experimental determination of its genetic control is facilitated by
the conjugation mechanism. The reciprocal exchange of gametes makes the two
conjugants identical in their chromogenes, but each animal retains the intact cyto-
plasm of its progenitor. 1Vhen separation of the conjugants is experimentally delayed,
however, some cytoplasm is exchanged. Two genetic requirements for the killer or
paramecin-producing character were discovered: I) that the animal carry one or two
doses of a gene K, i.e. that it be KK or Kk, and 2) that the K-bearing animal have
received cytoplasm from a killer animal, from which a plasmid h: is inferred. K was
once thought to be very intimately related to K, i.e. that it was either initiated by
K or participated with it in a bipartite unit in the macronucleus, but these hypo-
theses have since been abandoned as unnecessarily elaborate. The results are best
summarized as the dependence of the maintenance of Y on the chromogene I;. This
relationship is perhaps nutritional: The quantitative level of K is proportional to the
number of K genes in a given genotype, but the substitution (by- autogamy) of a
kk for a Kk macronucleus causes a depletion of I( only after a considerable interval.
  Paramecin and I( both contain desoxyribonucleic acid as an essential component.
However, KK and kk animals (sensitive for lack of K) are equally sensitive to para-
mecin, and no proliferation of the paramecin has been detected. The distinction of
K and paramecin bolsters the critique of premature identification of plasmids and
their products in other systems.
  Evidence of the genetic autonomy of h: has come from the study of mutants
affecting paramecin production. In a number of such mutants, variant L nas found,
affecting both the quality and amount of paramecin. Additional alleles of the chromo-
gene K have not, however, been described (so). Following upon the morphological


Oclober 1952     CELL GENETICS AKD HEREDITARY SYMBIOSIS          411

demonstration of K, its further study has followed a new course. It will therefore be
resumed under endosymbioses.

  A second system of cytoplasmic determination in Paramecium affects the ciliary
antigens (I&-182). \\:hen paramecia are injected into rabbits, antibodies are pro-
duced which paralyze the cilia and kill the paramecia if the treatment is too intense
or prolonged. The different antisera constitute a series of reagents A, B, C. . . which
react strongly with some animals, and not others, thus permitting a serotypic classi-
fication.

  The first evidence of extranuclear determination is the inter-transformability
of serotypes, which is especially dramatic in homozygous animals exposed to specific
reagents. For example, stock 51, rariely 4, shows three frequent serotypes, A, B,
and D, each of which is clonally stable. However, an animal exposed to liminal
concentrations of its antiserum becomes converted to one of the other types, establish-
ing a clone of equally stable serotype. The serotypic transformation is reversible
so that all types are mutually inter-transformable. Under `standard' conditions,
the existing type is perpetuated, unless disturbed with antiserum, and each type
has equal stability. Under other conditions of temperature and nutrition, however,
one type is preferred, and all others spontaneously transform to it. Even the anti-
serum effect may be indirect, and result in part from the temporary alteration of
growth habit induced by the antiserum (175). The cytoplasmic basis of the serotype
is verified by mating experiments: each exconjugant breeds true to its original sero-
type under standard conditions.

  The interplay of chromogenes with cytoplasm is revealed by comparisons of
stocks which differ principally in the range of serotypes detected in transformation
experiments. Thus stock 29 produced a serotype F not characteristic of stock 51.
Crossing experiments showed that the potentiality (or range of conditions) of trans-
formation to F was c:. iLtrolled by a dominant chromogene. Minor modifications of
the A antigen are also controlled by chromogenes. The genes thus discovered have
been termed `specificity genes' controlling the different antigens, but it may be more
circumspect to picture them operationally: the chromogenes control the conditions
under which the cell may produce a particular range of antigens.

  In ouriety I, genetically homologous antigens, i.e. antigens controlled by allelic
genes, may be serologically unrelated, and both alternatives displayed in the same
heterozygous animal whereas nonhomologous serotypes are mutually exclusive.
Thus, "only genes at one of the three loci are normally expressed at a given time

. . . depending on the state of the cytoplasm." (I 5). This does not mean that the other
genes are inert (a naive fallacy still voiced too frequently) but that the different
alleles do not alter the serotype of the preferred cytoplasmic state. The results with
r~uriety I show that it is the cytoplasmic states that are mutually exclusive, and that
the antigens are a product of the co-action of the chromogenes and.the cytoplasmic
state.

  The persistence of the existing cytoplasmic state may be interpreted in seemingly
different ways. Each state might correspond to the predominance of particular plas-
mids, the populations of which compete for nutrients. In order to account for the
reversibility of serotypic transformations, however, some means of replenishment
of different plasmid types would have to be postulated. Suggestions include re-initia-
tion from the macronucleus, the persistence of specific plasmids at low concentrations,
or a de WOZ'O origin by mutation of other plasmids. An alternative hypothesis dispenses
with particulate plasmids, but refers to anabolic steady states. Alternative reaction


sequences from different genes to tinal antigens are supposed to compete for metabo-
lites. The stability of the existing state is explained by regenerative feedback whereby
a given product facilitates its own synthesis, or inhibits alternatives.
These hypotheses are less distinciive than may appear at first sight. Any plasmid
behaves in much the same way. For example, K can be regarded as a gene product,
in view of its dependence on the K chromogene. K differs from the serotypic states
insofar as it can be irreversibly suppressed, whereas the serotypic states remain
mutually transformable. However, this may merely reflect the greater complexity,
and concomitant autonomy, of K which make it unlikely that it can be readily re-
evolved from non-K cell constituents. The elementary autocatalytic reactions of
growth that underly the reproduction of any organism are subsumed under the
steady-state hypothesis. Further investigation may furnish more details of the
cytoplasmic states, but are unlikely to decide between tautological hypotheses.
Because of its .simplicity, the antigen system may perhaps furnish information on
genetic conservation and growth that is less readily obtained from more comples
experimental materials. Cytoplasmic states clearly parallel the stable states of de-
velopmental modulations and differentiations. The mechanisms that may be involved
in embryology or neoplasia are best explored, at first, in this type of model. The actual
processes can only be told by looking at the embryo or tumor itself.
  A convergence of this model with original is furnished by an account of pattern
differentiation in spotted mammalian skin (17). Pigmented areas, at iirst sharply
delineated, tend to spread as the spotted guinea-pig or bovine matures. The spreading
is exaggerated when pigmented skin is autografted to white. It might be explained
by the migration of pigmented dentritic cells (melanophores), and their replacement
of unpigmented cells. Alternatively, pigmented cells are thought to inoculate a
melanogenic plasmid into unpigmented dendritic cells with whose terminal processes
they fuse. Homograft pigmentation also succeeds but only when the donor seeds are
so small as to avert iso-immune reactions (176). The experiments still do not entirely
disqualify the cell migration hypothesis, if we impute to the melanophores the same
low immunogenicity but high sensitivity to antibodies as claimed for the postulated
plasmids. Either melanophores or plasmids must differ from other tissue elements
which quickly provoke the histo-incompatibility reaction in homografts. Even on
the plasmid or cytocrine hypothesis, the identification of the plasmid with the visible
melanin granules (presumably mitochondrial derivatives) is tempting but so far
unsupported. If verified, the plasmid interpretation would be the first concrete justi-
fication of the general theories of somatic differentiation based on the assortment of
plasmagenes.
  The genetic basis of the hereditary, irreversible differentiations of somatic cells
is still obscure and experimentally difficult (125, 202). M'ith mammalian tissues,
there are few cellular characters that can be related to genotypic alterations observ-
able in crosses of intact animals. However, promising material is available in the histo-
compatibility patterns of transplantable tumors. A number of histo-compatibility
variations have been described in clones of transplanted tumors, and it should be
possible to relate them to mutations of Qecific loci by immunological tests (105,
119, 176, 192). Although their origination as spontaneous mutations is most plausible,
there are some obscurities which require further study to ensure that the histo-com-
patibility (or iso-antigenic) factors are directly involved, especially as virus contami-
nations may influence the growth of a transplant (194). The possibility of detecting
somatic segregation in tumor cells originating from heterozygous animals should not


Ocl(lbe; rgj.?     CELL GENETICS AKD HEREDITARY SYMBIOSIS         413

be overlooked. Although a microbiological approach to the genetics of tissue cells is
still technically difficult, demonstrations such as the spontaneous origin of mutations
for resistance to anti-leukemic chemicals (98) afford high promise. These mutations
are not developmental differentiations, but their study may be expected to lead to
the tools needed for the genetic study of development per se.

GFXETIC TRANSDUCTIONS (TRANSFOKM.ATIOKS) IK BACTERIA

  Genetic transfers in bacteria have been apprehended to infective heredity by
many authors including the reviewer (IO, 45, 102, 177). The first well-authenticated
transduction was the type transformation of the pneumococcus. Purified extracts of
capsulated cultures induce stable homologous capsulated strains from uncapsulated
variants. Analogies were soon suggested to the transmission of latent virus, or to
specific directed mutations. It is now generally agreed that the alteration is more
constructively considered as a restricted transfer of genetic material to the cell,
rather than as an induced mutation and the reviewer has proposed transductiotz
for such processes, to avoid the diverse connotations of transformation (IO, $3).
  So long as the transductions influenced only a single specialized trait, the capsu-
lar antigen, an infective plasmid furnished a plausible interpretation, especially by
analogy with latent virus. However, many transducible characters have been studied
in several bacteria (pneumococcus, IO, $3, 81; Hewophilus injluenzae, I ; Salmonella
f~~phinzurium, 222). In Salmonella these thirty or more traits include biochemical
syntheses (nutritional requirements), fermentations of sugars, resistance to anti-
biotics, and flagellar antigens: Most or all of the heredity of these bacteria can evi-
dently be transduced, but only one unit at a time. Thus, transduction is functionally,
and perhaps phylogenetically, a special form of sexuality, which follows a more famil-
iar, coordinated pattern in related bacteria, i.e. Escherichia coli (103). The possi-
bility of transduction of genetic factors in higher organisms has been suggested by
studies on the transmission of tumors by presumably cell-free extracts of chromatin
or mitochondria (225, 227). The likelihood of confusion or identification of the trans-
ducing fragments with tumor viruses is obvious.

The genotype is a priori unlikely to consist of a large number of mutually
unorganized autonomous units (102, 131, 219; also see discussions of $3, 81). U'ithout
incessant reduplication of each unit, the system could not long remain qualitatively
intact. In a complex organism, some mechanism must insure a meristic separation
of each hereditary unit at every fission, but it is difficult to conceive of such an
operation for an inordinate number of independent units. For the same reason that
amitosis is suspect in higher forms, we postulate that most of the genetic material
is coordinated in a definite structure. The only candidate whose availability is
certain is the nucleus, in which desoxypentosenucleic acid (DNA) has been cyto-
chemically demonstrated (21). Chemical and enzymatic studies have shown that
transductive competence in the pneumococcus and influenza bacillus is closely as-
sociated with DS.4. A single particle of DK.4 might escape unnoticed in the cyto-
plasm, but if every trait is referable to DNA, this must be located where it is visible,
that is in the nucleus.

The apparent fragmentation of the genotype in transduction, whereby individual
traits are separately transferred, thus need apply only to the transmission and not
to the internal organization of the genes either in host or recipient. \f;hether agents
of transduction are chromosome fragments or macromolecules containing DNA
(IO, 59, 103, 131) is entirely a matter of outlook. In either event, genetic material


414                JOSHUA LEDERBERG             Iblrme Jr

is transferred from the fragmented structure of the host cell, and reorganized into
the recipient. The characterization of this material as a plasmid rests squarely on
the hiatus of our present genetic knowledge of the bacterial nucleus. Until this is
much more thoroughly understood, infecti\:e transmission from cell to cell cannot
be relied upon as a criterion of extranuclear residence.
The chemical composition of transducing agents has important implications,
if they can be regarded as purified genes. The most active preparations from the
pneumococcus have been refined with the methods of nucleic acid chemistry, and
consist almost entirely of polymerized DKA with less than I per cent protein. Fur-
thermore, these preparations are rapidly inactivated by desoxyribonucleo-depoly-
merase, so there can be little doubt that DK.4 plays an essential role. iip*hether
protein has been completely excluded is a more difficult problem (127). The biologi-
cal activity of the preparations is less impressive when stated in molecular than in
millimicrogram units. There is ample room for an undetected active nucleoprotein
contaminant floating in an excess of purified, viscous, polymeric, inactive D?;A.
Preparations of isolated chromosomes from spermatozoa contain only a few per cent
of a residual structural protein. Hypothetically, the retention of activity in prepara-
tions fractionated to leave less than I per cent of this protein would not rule out its
possible biological importance.
The most plausible conclusion from existing studies with the pneumococcus is
that DNA is not only necessary but sufficient for transductive activity. However,
plausibility must be fortified by rigor before generalizations as to the overriding role
of DNA in genetic specificity can be acquiescently accepted. The biochemical studies
to date appear to have been designed to conserve the native nucleic acid, and this
might itself account for the trend of the experimental results.
The advantages of Salmonella for genetic studies are countered by the inaccessi-
bility of the transductive agent for chemical analysis, for it is apparently associated
with bacteriophage particles. The passive role of phage as a vehicle for the trans-
mission of genetic material should not be confused with the active determination of
certain traits by latent viruses, as reviewed below.

                   ENDOSYMBIOSIS
The plasmids of previous sections were first detected by genetic tests. After the
rules of their transmission had been worked out, cytoplasmic particles were often
identified with the plasmids, although the observed particles might often be equally
well understood as secondary products of plasmid action. There are many hereditary
endosymbioses which have been detected morphologically, and have not received
the attention they merit from geneticists. K, a plasmid in ParamecizAm aurelia
represents one instance that eventually would have been discovered by the morpholo-
gist. OUr XCOUnt Of K (See SeCtiOn, CyhIphT?ZiC i?ZheYifiZlZCe if2 PaYa??ZecbVZ) Inay
therefore be resumed where its history approaches that of other endosymbionts.
The all-or-none transmission of K in suitable conjugation experiments originally
suggested its particulate nature. The first estimates of its numbers came from kinetic
studies of its attenuation in rariety z, in which the killer trait had been observed to
be unstable. When the K-COntaining animals are grown with optimal nutrients, the
cells evidently multiply faster than the x particles, so that the latter become pro-
gressively diluted at each fission until some animals are entirely free of K. It was shown
kinetically that a single K particle could maintain the killer trait (if the animals were
then grown more slowly), and that the stable .Y population consisted of IO?--IO~


Oclober 1952     CELL GEXETICS AND HEREDITARY Sl-MBIOSIS         415

particles per cell (149). Later, it was found that temperature shocks and x-ray treat-
ments differentially inactivated the K. Its unexpectedly large radiosensitive cross-
section led to the microscopic search for and identification of K as a Feulgen-positive
particle about I k in diameter (ISO). Concurrently, Sonneborn succeeded in the
artificial infection of sensitive IX paramecia with K from concentrated breis of killer
animals (181). Killer cells can be chemically disinfected with nitrogen mustard and
with chloromycetin (26, 182). Despite superficial references to rickettsia, the taxo-
nomic position of K remains in doubt, but not its assimilation as an endosymbiont
(cf. 2, 5, 182).

Hereditary endosymbioses are probably more prevalent than many biologists
are accustomed to believe. Buchner (28) has collected the literature on plant-animal
symbioses in his magnificant book. Had he further emphasized the hereditary char-
acter of these associations, the present review would be superfluous. A complete re-
vision of Etzdosymbiose (29) has been announced, but was not available in time for
this n-riting. Most of the literature on symbiosis is morphologically oriented but nel\
nutritional knowledge is likely to make the biochemical basis of endosymbiotic
relationships far more meaningful. Too little emphasis has been placed on the oc-
currence and behavior of `disinfected' or aposymbiotic individuals, and on the cri-
terion of reinfection for the specificity and identity of the endosymbiotic microorgan-
ism.

Owing to the simplicity of their overt cell structure, cyanophyte (blue green
algae) endosymbionts have repeatedly been interpreted as intracellular organelles,
so that the `syncyanoms' are especially instructive (142). They include such complexes
as Geosiphorz (a phycomycete + *Yos/oc) and Gloeochaete (an apoplastidic chloro-
phyte + a cyanophyte). But the most interesting associations occur when the cya-
nelle (I'ascher's term for a cyanophytic endosymbiont) is regulated by the host.
Paz~liwlla clwomatoplzora is a testaceous rhizopod which regularly carries two sausage-
shaped blue chromatophores, so that Lackey (96) insisted they were cell organelles.
Functionally they are, for this animal exhibits a holophytic nutrition, contrary to
the voracious phagotrophy of related testaceans. The division of the rhizopod is ac-
companied by the segregation of the cyanelles, which later divide to restore the count
of two in each daughter cell,

Pcliuirra cyauea is a flagellate (chrysomonad or cryptomonad in its affinities)
which harbors from one larger to six smaller cyanella. Rare asymmetric fissions
yield colorless monads which produce oily reserves rather than starch like the Peliaitza
complex. Since free cyanophytes do not form starch either, the complex has achieved
a new function. The syncyanom must be of some antiquity, as there is no way that
the llngellate could have ingested the cyanella. Phagotrophic, amoeboid phases have
been described for some primitive flagellates, and the ingestion of a free-living cyano-
phyte by an amoeboid ancestor presumably initiated the symbiosis, to replace, at
least functionally, the initial plastid system.

Other types of algae are also involved in endosymbiotic complexes: A large part
of the phytoptankton of the sea may be bound as zooxanthellae and zoochlorellae
(221). Some of the symbiotic chlorellae may afford superior experimental material
for further work. The association of a chlorophyte with a flatworm, Colzz~olu/a roscof-
ferzsis, is the subject of a classical account (91), but the hereditary association is
marginal: The algae are transmitted in the egg capsule, rather than within the egg
(as in many sponges and the familiar Clzlorhydra cirissima (cf. also 69).
  Paramrcizm bursaria is one of the best known zoochlorella complexes, and has


416                JOSHU.1 LEDERBERG             Volrmre 32

also been genetically studied in other aspects (82, 141, ISI, 178). The symbiosis can
be maintained indehnitely in the light as a holophyte, whereas other paramecia re-
quire complex media with still undefined constituents. In the dark, however, the
chlorellae deteriorate and may subsequently be digested by the animal. Thus, aposym-
biotic animals are fairly easy to secure, and can be maintained like other species.
Free-living chlorophytes of various species were ingested but not stably established,
but zoochlorellae expressed from green animals were readily incorporated. Following
a series of negative reports, the isolation and culture of zoochlorellae has been de-
scribed, opening the door to studies of genetic variation of both components of the
complex (74, 109, 178). The chlorellae have also been removed with x-rays (207,
compare K). Symbionts of certain clones are toxic to others; it is not known wherher
this is related to the lethal mating reaction observed in certain combinations, or to
the K effect in P. aureha.

  In several chlorella-symbioses, particularly Peliairza and P. bursaria, the chroma-
tophores mediate a phototactic response, probably an indirect effect through CO2
utilization (142,151).

INSECT SYMBIOSES

  Symbioses of microorganisms with insects are very prevalent. Many are intra-
cellular and hereditary; others have equally effective techniques for persistent
occurrence. They have been most thoroughly examined by morphological techniques,
and very little physiological detail is known (28, 98, 187, 188, 208), although a nutri-
tional function is almost self evident. A special organ, the mycetome or pseudo-vitel-
lus, harbors a high concentration of the microsymbionts in many insects; in others,
they occur in certain areas of the gut wall. The origin of the mycetome in Pseudo-
coccus (mealybug) from the polar bodies disputes its homology with endodermal sym-
biont-carrying cells in other insects (cf. 171, 188).

  The penetration of the symbionts into the egg occurs at different stages in
different organisms. This usually occurs early in the development of the oocyte; a
nutritional relationship between the ovary and the mycetome has been suspected
by several investigators (6, 71, 138). In many beetles, however, the egg surfaces are
contaminated as they pass through the ovipositor, and the larvae or nymphs are
infected only after they feed on the egg shells. (The extracellular symbiotic protozoa
of termites are similarly transferred by fecal ingestion, 82, 92.) In bostrychid beetles,
seminal transfer has been recorded. In Glossina and pupiparids, the milk-gland se-
cretions are implicated. Many of these processes have a counterpart in the transmis-
sion of agents primarily studied as plasmagenes.

  The taxonomic identification of the microsymbiotes awaits unequivocal proof
of their cultivation in ritro (or i~z oz'o), rarely accomplished. The rickettsia are the
most notorious symbionts; as many authors have pointed out, their inheritance in
arthropods contrasts with their virulence in mammalian hosts (187, 188). The bac-
teria-like symbionts have not been certainly cultivated (68, 174) ; if a preliminary
report that they lack Feulgen-positive nuclei is substantiated and generalized, it is
not likely that they can be identified with free-living microorganisms (97). Yeast-
like forms have beers cultured from several beetles, and probably do correspond to
their symbionts (132, 140).

  Koch's postulates provide reasonable criteria that should be met for any or-
ganism that can be cultured in &o, and of which the insect can be at least tempo-
rarily disinfected. Even without i/l vi/r0 culture, reimplantation and cross-implanta-
tion would provide important information, but this avenue has been little exploited.


Oclober 195.2     CELL GENETICS ASD HEREDITARY SUMBIOSIS         417

  Both biological and physiological studies of the symbioses are facilitated 1)~
techniques for aposymbiosis: the disinfection of the insect. This has been accom-
plished in several nays; surface sterilization of contaminated eggs, heat treatments,
surgical or centrifugal removal of the mycetome, and, more recently, antibiotic ther-
apy (6, 25, 27, 28, 71, 95, 134, 140, 187). Nutritional functions of the symbionts have
been clearly shown in several insects. In the reduviid (assassin bug) Rhodnius pro-
liws, aposymbiotic nymphs failed to moult after the fourth instar, despite repeated
blood feedings. The suspended nymphs resumed normal development when experi-
mentally re-infected with cultures of the symbiont, Streptomyces rhodnii. Attempts to
substitute other actinomycetes were not reported (25).

The flour beetles Stegobiwt (Sitodrepa) p aniceum and Lasioderma serricorue
carry distinct yeast-like forms, Saccharomyces alzobii Buchner. Heterologous and
homologous cultures were successfully re-implanted, but specitic adaptation to the
homologous host was indicated with respect to the availability of B vitamins syn-
thesized by the yeasts (94, 140).

Most aposymbiotic insects are developmentally or reproductively defective
(71, 187, 188, 208), but the granary beetle Oryzaephilzls surinamensis is exceptional
(95). It may be necessary to conduct careful nutritional experiments with aseptic
precautions to define the nutritional functions of some symbionts.

Carter's account of the Hawaiian pineapple mealybug, Pseudococcus breesipes,
is the closest rapprochement of these studies to genetics. The insect secretes a wilting
toxin which is associated with a green-spotting of the pineapple leaves. \Yhen type
colonies were transferred to panicum grass, the next and succeeding generations of
mealybugs lost the green-spotting potentiality. This loss was associated with the
disappearance of a rod-shaped symbiont from the mycetome, whereas other forms
persisted (36). The type culture is gray and obligately bisexual; the variant is pink
and parthenogenetic, only females (hermaphrodites?) developing (85).

The multiplication and hereditary transmission of a phytopathogenic virus in
its insect vector has been reported for the rice stunt, clover club leaf, wound tumor
and aster yellows viruses (18, 65). The first two of these have been maintained in the
insect vectors for many generations, where they are inherited matroclinously. Careful
cultural techniques, and the use of nutrient plants immune to the virus, have left
no doubt of the active proliferation of the `plant virus' in the insect. Inheritance
of the wound tumor virus is irregular, and has not yet been demonstrated for the
aster yellows, but the viruses can be propagated by serial inoculation of the insect
hosts. Genetic factors apparently play some role in the maintenance of the viruses
(18; cf. 191). Like so many plasmids, the viruses can often be removed by differential
heat treatment of the infected insects.

These viruses have had no detectable effect on their animal hosts, and might
therefore be readily overlooked, equally by insect pathologist or geneticist. Such in
fections may, however, have very subtle effects not discernible in ordinary cultures.
An infective but noncontagious hereditary agent, CT, has an increased susceptibility
to narcosis by CO1 as its sole effect on DrosojMa melanogaster (105). The plasmid is
inherited matroclinously, independently of any chromosome markers. Some seminal
transmission is also found, probably via polyspermic fertilizations. g is also trans-
missible by artificial inoculations of hemolymph or by organ transplantations from
infected donors. However, patroclinous transmission occurs only when the male
was already infected at the egg stage, in accord with the early separation of the germ
line in Drosophila.

u-infected, COZ-sensitive flies may be somatically or germinally cured by heat


418                JOSHUA LEDERBERG             1 ~olu?ne 3'

shocks of the larvae. Incomplete or delayed treatments give only temporary or
somatic cures. The distribution of the plasmid and its latency would seem to depend
on the stage at which the organism is infected: the developing egg (in an infected
mother), at fertilization (from an infected father), or later in development (by inocu-
lation).

  A mutant u has been described which is less readily transferred via the male
germ line. In addition, relapses following inadequate heat treatment of the larvae
give infections resistant to a second heat shock, perhaps by a modification of the fly,
or of the u by selection of a heat-resistant mutant.

Filtration experiments have indicated a particle size of about 0.1 p, but u has
not been further characterized. It has been transferred to some other Drosophila
species by inoculation, but not to other Diptera, nor has it been detected in them.
Because it is not contagious in ordinary culture, r has been described as a genoid,
intrinsic to the. organism. However, infection from food or ectoparasites has not been
disqualified. Until (T recurs, the problem of its unique origin will not be solved.

OTHER LATENT ~`IRUSES, IN PLANTS AND MALNALS

  he presence of a virus in an infected organism may be hidden in a number of
ways (5, II, 34, 48). In the most subtle form, the virus cannot be detected as an
infectious agent, and infection is inferred only from the previous or subsequent his-
tory of the organism. This category would include the masked Shope papilloma virus
in the domestic rabbit, swine influenza in the lungworm vector, and lysogenic bac-
terial cells as discussed below. In a second category, free virus is present, but patho-
logical effects may occur if at all only with very heavy infestation, or under special
environmental conditions, as in lysogenic bacterial cultures, the mouse hepatitis
virus, herpes simplex in man, and most of the latent X viruses of plants. Bacterial
cells are distinguished from cultures to emphasize the probability that individual
infected cells may be damaged by a latent virus, without detectable effects on the
whole organism.

  Pollen and seed transmission of infective viruses is exceptional in plants but
occurs sporadically in legumes. The special property that isolates the seed from virus
rampant in somatic tissues "remains one of the most interesting unsolved problems of
virus diseases" (II). The agents of the hereditary plastid variegations, on the other
hand, have not been artificially transmitted, although a persistent experimental
attack is perhaps still needed. The mode of transmission of these agents constitutes
a plausible but only empirical basis for classification as viruses, `proviruses,' or
plasmagenes (45).

  Several mammalian viruses have a quasi-hereditary transmission, including
murine hepatitis (72) and lymphocytic choriomeningitis (196), the exact route being
unknown. One etiological factor in the development of mammary tumors inmice,
previously characterized as a maternal influence, has been shown to be transmitted
primarily by the milk. The agent occurs throughout the tissues of infected animals,
and can also be transmitted by artificial inoculation and possibly via the semen and
in ziiero. Its pathological consequences and maintenance are under chromogenic
control. It has been propagated in chick eggs and in male mice, in which it remains
asymptomatic in the absence of hormonal stimulation. The viral and the genetic
properties of the milk agent have both been emphasized (4, 108, 130). The flourish-
ing interest in latent viruses along lines similar to those of this paper is exemplified
by a symposium (224) and review (226) which appeared while this was in press.


October 1952     CELL GENETICS AND HEREDITARY SYMBIOSIS         419

As a converse to the modification of host cells by viruses, there are several ex-
amples of virus alteration in different hosts. Host adaptation is probably most
often a consequence of spontaneous mutation and selection (33, 34, I 12), but pheno-
typic modification of a virus has also been considered (78) as well as genetic re-
combination of the invading virus with previous residents of the cytoplasm (35, 79,
164).

LYSOCENIC BACTERIA

Stable symbiotic associations of bacteria with viruses (bacteriophages) occur
very frequently, Because the symbiosis is detected by the lysis of a sensitive indicator
strain, the association has been called Zysogenicity. The lysis of an indicator reflects
the production of free virus only by few of the bacteria in a lysogenic culture. Most
of the cells perpetuate the potentiality of producing virus, although the virus itself
is rarely detectable in them. For this reason, such cells are considered as infected
with a provirus, a perpetuating, but immature and nonlytic agent. (The same ex-
pression has been used in other meanings. As used here, and by several authors,
24, 48, 117, prooirus is a formal designation for the nonlytic stage of development
of the symbiont; its application should not conceal our ignorance of the differences,
if any, between the hypothetical provirus and free virus particles.)

Lysogenicity has been used as an argument for the endogenous origin of bacterio-
phage (70, 217). However, since foreign viruses can enter into the same symbiotic
relationship that is displayed by bacteria lysogenic when first isolated, the argument
is inconclusive. On the other hand, symbiotic viruses which may be difficult to
detect as such occasionally mutate to virulent forms, simulating the de nova produc-
tion of phage from a normal bacterial culture (23, mo). Neither of these observations
contradicts the intracellular-parasite theory of phage, as enunciated by its discoverer
(33, 49), but this in turn does not rule out hereditary functions of the symbiont.

  Little detail is known of the initiation of the lysogenic symbiosis. For technical
reasons, recent phage studies have concentrated on atypically virulent strains,
particularly the `T phages' acting on E. coli B (46-48). Infection of a bacterial cell
with a single T phage particle promptly interrupts normal biosynthetic activity,
which appears to be almost completely diverted to the demands of the virus. After
a dark period of several minutes, during which the infecting phage is undemonstrable,
there is an interval of rapid increase of virus, as detected by chemical analysis or
artificial rupture of the bacteria (41, 52). The growth period culminates in bacterial
lysis, and the release of some dozens or hundreds of free virus particles.

  Certain phages, designated as temperate for particular hosts, may follow a
different course alternatively to the lytic pathway. The stages intervening between
infection and symbiosis are obscure but a bacterial clone develops in which a dark
period is indefinitely prolonged while normal growth proceeds. Later, descendants of
the infected parent may sporadically resume a lytic pathway and release free phage.
Unless this phage reaches a sensitive indicator stra.in, it is not detectable.

The proportion of bacteria which follow the alternative pathways of lysis
(LP) or symbiosis (SE') varies with the hosts and viruses involved, and with the
conditions of infection. In some staphylococci, LP predominates (165, 209); in
enteric bacteria, SP may predominate to the extent that no overt lysis is observed
at all (205). In Salmonella typhimurium, high multiplicities of infection appear to
favor SP, suggesting a heterogenicity of the free virus population: hypothetical
temperate variants initiate lysogenicity and protect against lysis (24). The hy-


420                JOSHUA LEDERHERG             l'olwm 3'

pothesis of a reversible genetic differentiation between temperate and lytic free virus
should not be confused with the verified occurrences of stable, lytic variants from
temperate phages. The hypothesis has not received independent support but makes a
useful framework for further discussion. Multiple or iterated infection per se affects
development of lytic phage (51, 63) and might be directly involved in the prolonga-
tion of the dark period that culminates in lysogenicity. Alternatively, the LPISP
decision may rest on differentiation of the bacteria, for which genetic studies offer
some support.

The lysogenic symbiosis is, pima facie, plausibly regarded as the carriage of
provirus in the cytoplasm. To explain resistance to lytic attack by the provirus,
to free virus, and sometimes to related viruses, one may postulate the same kind
of interference as has been demonstrated in lytic systems, This simple picture has
not, however, been supported by crossing experiments with Escherichia coli K-12,
which point to a single chromogene as the major determinant of the sensitive,
lysogenic and .immune states (100). It is not settled whether lysogenicity results
from the selection of spontaneous mutants when sensitive cells are infected with the
virus, X or whether X more directly induces the alteration of the chromogene. If
the latter is supported by as high an efficiency of formation of lysogenic from sen-
sitive cells as appears at first sight, a chromosomal localization of the X provirus will
have to be given serious consideration. A special relationship between the nucleus
and some viruses may also be inferred from cytological and cytochemical analysis
of phage infected bacteria, and from the attachment to certain Salmonella phages of
genetically active material of the host cell (41, 113, 133, 222).

  The most prominent phenotypic effect of lysogenicity is the resistance acquired
against the symbiotic phage, and often against other intemperate phages. The pat-
tern of resistance to a standardized series of phages, the lysotype (44, 137), serves to
characterize individual strains of several bacterial pathogens for which epidemiological
tracers are needed. In staphylococci (165, 209), Salmonella lyphi Vi-types (,3), Salmo-
nella paratyphi B (137) and other Salmonellas (*IO), the lysotype is determined to
a large extent by the carriage of determinant phages. Such phages are as a rule closely
related to the typing phages themselves. In Salmonella trp/zi, the typing phages are
specifically adapted by growth on standard Vi strains, and will lyse only cells carry-
ing this antigen. The determinant, symbiotic phages show no Vi-specificity in their
own host range. The mechanism of adaptation of the typing phages has not been
closely studied, but may be either a selection of genotypic variants, or a phenotypic
modification of the typing phage imposed by the host cell (or its proviral symbionts
if these can be distinguished from the host).

Other effects of phage plasmids have been less extensively studied. Toxico-
genicity is induced in avirulent strains of the diphtheria bacillus by infection with
temperate phages (62). The El-Tor vibrio, distinguished by its hemolytic activity,
is claimed to be a lysogenic derivative of Vibrio cholerae; many bacteria release
hemolysins when lysed by phage (53, 167). Other `transformations' or filtrate effects
should be re-examined for the simple participation of phage (e.g. 64, reviewed IOI,
pp. 181-184).

The long mooted question whether phage was continuously secreted by lysogenic
bacteria has been conclusively answered. In Bacillw megalerium, the appearance of
free phage is always correlated with the lysis of one or more cells, under microscopic
observation (117). Similarly, free phage in replicate cultures of E. coli, strain
Lisbonne-Car&e, is distributed in discontinuous bursts. Bacteria may be lysogenic


Oclober rgyz     CELL GENETICS ASD HEREDITARS- SYAIBIOSIS         421

for two or more phages, and still support the LP growth of an additional virus. In
the experiments so far reported, individual bursts liberate one phage only (16, 201).

The resumption of LP occurs sporadically, but is also under experimental
control. Some lysogenic stra.ins of B. wzegateriwn are evidently poised at a precarious
balance, and exposure to small doses of ultraviolet light, x-rays, or reducing agents
will induce massive lysis of the culture. Ultraviolet light will also induce LP in
some, but not all, lysogenic associations of Escherichia coli. Consequently, infection
often leads to sensitization to ultraviolet light, if the cells are tested under ap-
propriate conditions (I 18). An antibiotic (colicin) producing strain of E. coli responds
to UV in the same way; whether the colicin is itself a modified phage, or whether
another undetected phage is concerned is uncertain (61, 86).

  Ko straightforward method is yet available for disinfecting lysogenic bacteria
(22, 32, 165), although sporadic successes have been reported (39, 54, 195). This
may simply attest to the close integration of host and plasmid. N'ithout such a
technique, or an indefinite range of potential indicators, it is impossible to assert
that any bacterial culture is free from the potentiality of producing a virus. In fact,
in certain groups of bacteria the incidence of proven lysogenic isolates approaches
one hundred per cent (23, 32, 60, 70, 124, 136, 169, 222). Conversely, lysogenicity
probably constitutes the principal reservoir of bacteriophage in nature (33).

HETEROKARYOSIS, HYPERPARASITISM AXD SEXUALITY IN THE FCKGI

The problem of delimiting the normal territory of the cell from included plasmids
is magnified when the components are of equal stature. Coenocytosis occurs only in
restricted tissues or phases in higher plants and animals (e.g. the insect blastoderm),
but is a regular feature in mycelial and plasmodial microorganisms (many fungi
and some algae and protozoa). So lon g as all of the nuclei of a coenocyte are
genetically identical (homokaryotic), no special problem arises. However, hetero-
karyosis may arise by mutation, cytoplasmic fusions, or special sequelae of sporo-
genesis which bring diverse meiotic products together (161). The genetic and
evolutionary consequences of heterokaryosis have been examined most closely in
ascomycetous fungi (146, 147). The implications of syncytia as exceptions to the cell
theory have been discussed recently (223).

  In the heterokaryon, several nuclei operate in a common cytoplasm undivided
by cell boundaries. The problem of delimiting the plasmid from the normal territory
of the cell is exaggerated here, where we must cope with reactants of equal stature,
the diverse nuclei. Ultraviolet light induces lethal mutations, probably chromosomal
deletions, distributed among the component nuclei of heterokaryotic conidia of
Seurospora. Homokaryotic segregates are therefore inviable, but different nuclei
sustain each other to give a stabilized or balanced heterokaryon, comparable to
balanced heterozygotes in Drosophila, or more particularly to obligate symbioses (7).
The substantial basis of the interaction is not precisely known for the lethal mu-
tations. However, nutritionally dependent (auxo-heterotrophic) mutants interact in
a similar way to give auxo-autotrophic heterokaryons (14). The auxoheterotrophic
mutation is a relative lethal, for it will survive only if its required growth factor is
supplied to it, whereas the autotroph synthesizes the growth factor for itself. The
number, location and kind of steps that lead from genetic determinant to growth
factor product are not known. The activities of the nuclei in a heteryokaryon might
be pooled at any intermediate extranuclear level, be it growth factors themselves,
biosynthetic enzymes, or less accessible, gene-controlled regulatory mechanisms.


422                JOSHUA LEDERBERG

Syntrophic interactions imply that the effective field of each nucleus is the entire
cytoplasm, so that the genotype of the entire coenocyte is a summary average of the
activities of the component nuclei.

  The composition of heterokaryotic populations raises a question of local indi-
viduality of the nuclear field (146). Unlike simple heterozygotes, in which gene
dosages are restricted to simple integral ratios, the proportions of a pair of alleles in a
heterokaryon may vary continuously. But how are the ratios determined? If the
individuality of the nucleus is submerged in the composite effects on the entire
coenocyte, direct selective effects on particular nuclei are precluded. This concept is
substantiated by many experiments in which auxo-heterotrophic mutant nuclei
were indefinitely propagated in association with wild type in heterokaryons grown
on minimal medium. Some exceptions must, however, be noted in which, for ex-
ample, leucine-independent nuclei were displaced by leucine-dependent, even on a
minimal medium which failed to support further growth of the latter, so that self-
extinction ensued (166). Phenotypic individuality of diverse nuclei in a heterokaryon
may therefore find expression in the determination of the nuclear ratios, of which
closer study is needed.

  Another field of action is the chromosome, as shown by position effects (43, 104).
Several examples have been found of chromogenes which do not interact in hetero-
karyons, but must be in juxtaposition on the same chromosome. A further possibility
has not yet been verified: a reaction limited by the nuclear membrane. Experimental
material for it is available in those fungi where heterozygous diploid mycelia can be
compared with heterokaryons (101, 163).

  The evolutionary place of heterokaryosis has been evaluated with respect to
sex and to parasitism (14). The heterokaryon permits the association of diverse
genotypes, whose adaptivity is perhaps signified by the high incidence of hetero-
karyosis in natural isolates. The instability of the heterokaryon and the limited
range of combinations it permits make it less productive than sexual recombination
for potentially adaptive variation. It has also been suggested that heterokaryosis
was an early phylogenetic step in the development of sexuality. However, the
prevalence of recombination processes even in viruses suggests that sexuality preceded
or coincided with the organization of the euploid nucleus.

  Heterokaryotic associations are not confined to intraspecific combinations. In
iVeuros@ra, the interspecific heterokaryons tend to be somewhat unstable, and can
be maintained only under selective pressure (146 and its discussion). Several cases of
hyperparasitism have been described, particularly in the Mucorales, in which the
parasite forms a heterokaryon with the host mycelium (31). It would be instructive
by appropriate manipulation of the medium, or the use of artificial nutritional
mutants, to test the mutualistic potentiality even of such parasitic heterokaryons.

SYNTROPHISM AND AUTOTROPHISM

  Almost all green plants, and a few special microorganisms are autotrophic, that
is are potentially self-sustaining in the absence of any organic nutrients. The remainder
of the living world is dependent on other organisms, ultimately the green plants
(93, 114, 170). Expressed otherwise, all heterotrophs are genitally insufficient, and
depend for their subsistence on the biosynthetically expressed genie functions of
other organisms (13). Thus it is possible to formulate a graded series of symbioses
as genie interactions, from the co-habitants of a single chromosome, through hetero-
karyons and plasmids, to extracellular ecological associations of variable stability


October rgjr     CELL GEXETICS Ah-D HEREDITARY SYMBIOSIS          423

and specificity (III). Other writers have commented on levels of cooperation in
societal rather than genetic terms, emphasizing the relationships of genetically
similar units such as cells, and some have drawn holistic conclusions (80, 156, 204).
Without furthering such extrapolations, our survey of symbioses should show that
the delineation of organic units, be they genes, plasmids, cells, organisms, genomes
or colonies is a tool of investigation and communication and not an absolute ideal.
For different purposes the units may be different but there need be no contradiction
between the Paulinella complex as a single chromatophoric organism (96) or as
rhizopod plus cyanophyte (92).

SELF-REPRODUCTION AND SELF-DEPENDENCE

  The concept of genetic autonomy or `self-reproduction' has been implicit
throughout this review, but has not been critically defined. When the capacity for
reproduction is self-contained, as in a free-living organism, no serious question arises,
But what criteria are applicable to smaller units of or in the larger complex, by
which we can show that some are `self-reproducing,' while others are reproduced by
the organism as a whole? The problem is usually approached by abstracting the
unit in question from the organism. If it is then no longer produced by the organism
it is concluded to be self reproducing. This has unfortunately been taken to mean
that the unit is autosynthetic, but the proof shows only self-dependence, not self-
sufficiency. In this context, self-dependence describes a wide range of reactions, from
simple autocatalytic reactions which may direct the chemical evolution of intra-
cellular metabolites, to self-contained autotrophic organisms (2, 193, 219, 220). The
position of the gene in this hierarchy is not established.

  Experimentally, it may be very difficult to test a visible particle for self-de-
pendence. Surgical excision is rarely applicable. Chemotherapy or biological inci-
dence may indirectly remove a particle, but this does not preclude more remote ef-
fects. Only the most searching inquiry supporting the parallelism in distribution of a
visible particle with that of plasmid effects can rigorously justify their identification
with each other, as has been done with the chromogenes and chromosomes. When the
plasmid can be purified by chemical or microbiological procedures its successful re-
introduction will suffice as well.

  The need for some circumspection concerning viruses and self-reproduction has
been well illustrated by an `alkaligenic virus' described in 1925 (I IO). Heavy inocula
of E. coli on plates of a sugar-peptone-agar medium initially produce a uniform acid
reaction by glycolysis, inhibiting further growth and enzymatic action. On one plate,
alkaline reversal spread from its initial location, suggesting a growth process. When
a piece of the agar was planted on a second plate, the alkaline reversal was propagated,
suggesting the spread of an alkaligenic virus. In fact, reconstruction experiments
(the reviewer, unpublished) show that the so-called virus is the OH- ion. The local
neutralization of accumulated acid permits the oxidation of the peptone, from which
more alkali is released as ammonia. The simulated virus is a simple metabolic diver-
sion. To insist on the self-reproduction of OH- is a reduclio ad absurdurn.

  To avoid such difficulties, a second criterion has been suggested: the capacity
for propagable mutation. However, mutability may be merely a consequence of struc-
tural complexity, (as pointed out to the reviewer by C. C. Lindegren). All-or-none
deviations of the components or rearrangements of their pattern are translated into
mutations of the complex. The presence-absence concept as applied to mutation of
the chromogenes has been discredited by the discovery of multiple allelic series,


424               JOSHUA LEDERBERG             1'011tn2e jr

which suggested innumerable discriminations of gene forms. More recently, however,
a finer analysis of multiple alleles has revived the application of presence-absence
and positional relationships to units distinct from the conventional single chromo-
gene (73, 122, 185). A priori, the elementary units of genetic reduplication are likely
to consist of only a few species, from which complex specificity is derived by struc-
tural organization. The units themselves are self dependently reproduced in the cell,
but need not possess the mutability or individuality associated with the gene or or-
ganisms as a whole. On this basis, the elementary processes of reproduction may be
as simple as the alkaligenic virus.

Oversimplified analogies between crystallization and over-all biological growth
have been soundly criticized. But the selective deposition of homomorphic substit-
uents and the perpetuation of specifically oriented faces and habits in the crystalli-
zation of nonelectrolytes do fire speculative reasoning on these elementary proc-
esses. Unfortunately, we do not know much more about the growth of crystals than
of cells, but the immediate prospects are brighter (30).

The search for elementary processes of reproduction is urged on by speculations
on the origin of life. Recent writing has tended to assume the accidental production
of the first living protein molecule (13, 20, 139), discounting the improbabilities with
the stipulation that life has evolved. An approach more consistent with general
evolutionary theory would be to search for simpler units and processes (not neces-
sarily proteinaceous) which bridge the gap between the organic and inorganic by in-
numerable steps each oriented by natural selection. As a corollary, we may question
whether similar first steps toward abiogenesis do not occur frequently today, too
ineffective for our recognition, and with little hope of successful evolution in competi-
tion with the far more efficient creatures of the living world. Transposed to the cell,
this argument suggests that genetic autonomy may gradually arise in internal me-
tabolites initially devoid of self-directive capacities as well as by mutation of pre-
existent plasmids.

The complexity of genes, microorganisms, and visible plasmids makes them un-
promising material for the problem of their reproduction. Therefore, we must look
to simpler self-directive systems-perhaps including the ciliary antigens of protozoa,
or priming factors in polysaccharide synthesis (77)-for suitable experimental models.

EVOLUTION OF THE CELL AND ORGANISM

  Evidence from comparative cytophysiology and biochemistry supports the
monophyletic origin of contemporary life (93, 114), but the lines are not always
distinct. Speciation, especially in higher plants, has often been associated with
hybridization of previously divergent groups. Endosymbiosis on the one hand repre-
sents a cooperation of distinct forms, on the other verges on hybridization (cf. 126).

  If'e should not be too explicit in mistaking possibilities for certainties. Perhaps
the disrepute attached to some of the ideas represented in this review follows from
uncritical over-statements of them, such as the Famintzin-Merechowsky theory of the
phylogeny of chloroplasts from cyanophytes (28, 126) or the identity of mito-
chondria with free-living bacteria (198). Similarly, the origin of viruses from normal
constituents of the cell is likely to receive a more considered hearing from virologists
and geneticists if the concept is not overburdened with dogmatic but unsupported
generalizations as to the specific organelles from which they come. The generality
of the subject demands imaginative speculation checked by the most cautious crrtr-
cism.


October 1952     CELL GENETICS .~ND HEREDITARY swmrosrs         425

This review has contrasted the various forms of plasmid: the hereditary parasites
as against the functionally coordinated plasmagenes, with the mutualistic endosym-
bionts somewhere between. It should not be assumed that the evolutionary pathways
are undirectional. The latent virus has become a normal constituent of the cell. If
a pathogenic virus arises de IZOW tomorrow, shall we insist that the normal constituent
from which it arose was not a latent virus yesterday? The question seems almost
unanswerable, for the very reason that the plasmids may evolve. For example, it
has been proposed that the paracrinkle virus of potatoes is an intrinsic derivative,
since it is confined in nature, so far as we know, to the King Edward varietal clone,
and was thought to be transmissible only by artificial grafts (45). Even on these
premises, the argument is incomplete: The postulated normal precursor may have
been an imported latent virus whose imprisonment in the clone was a subsequent
adaptation (compare Peliaina supfa). This view has been strengthened by dis-
coveries of fact: I) the nonappearance of paracrinkle in the very rare King Edward
hybrid seedlings, 2) successful mechanical transmission with the help of abrasives,
and 3) completely symptomless infections of tomato by paracrinkle (12). The preva-
lence of virus latency in vegetatively propagated plants (potato, strawberry) is prob-
ably not accidental (II). The barriers to seed transmission of viruses would tend to
preclude slow adaptations of plant and virus towards a symbiotic association. LVhere
seed transmission of a latent virus has been accomplished, on the other hand, it is
less likely to be recognized as it becomes more ubiquitous. Plasmon inhibitions (37),
or genetic diseases simulating viroses (8, 159), might result from the recrudescence
of inherited latent viruses when genotypic alteration upsets the symbiotic equi-
librium.

  In a search for experimental support of his `viroplasm' theory, a predecessor to
recent restatements of virus-plasmagene relationships, Johnson inoculated indicator
plants with extracts of normal seedlings (87). New viruses were discovered but no
conclusion as to their origin could be reached. The donor seedlings may have carried
a seed-borne, symptomless virus. Whether such viruses are to be regarded as normal
constituents, especially if ubiquitous within the species, is a question that strains
our understanding of the normal.

  This discussion illustrates how readily the properties of plasmagenes may be
imputed to viruses, and vice versa. The general criteria that have been used to
decide the historical origin of particular plasmids are unverifiable, and such contro-
versies have therefore tended to be sterile. It would be far more constructive to focus
attention on plasmid functions which in terms of the whole organism may be now
adaptive, later pathological. Present evidence points to the nucleus as the predomi-
nant, if not quite exclusive, seat of hereditary factors in most organisms. Lf'hether
new techniques will alter this generalization has still to be seen. However, we learn
less, not more, if we ignore extranuclear agents as hereditary factors because they may
also simulate symbionts or parasites; such behavior is important but incidental to
their genetic functions.

  The cell or the organism is not readily delimited in the presence of plasmids
whose coordination may grade from the plasmagenes to frank parasites. Many geneti-
cists have pointed out further that the gene has displaced the cell as an ultimate
unit of life (131, 219). How then shall we choose the boundaries of the gene-complex
that constitutes an individual organism. ' If hierarchical definitions are to serve the
scientist, rather than the scientist servean Aristotelian category, different uses should
dictate different usages. The geneticist may well choose that entity whose reproduc-


426               JOSHUA LEDERBERG

Volume 32

tion is unified and hence functions as an individual in evolution by natural selection
(102). The microbiologist will focus his interest on the smallest units he can separate
and cultivate in controlled experiments, in test tubes, eggs, bacteria or experimental
animals, Genetics, symbiotology and virology have a common meeting place within
the cell. There is much to be gained by any communication between them which
leads to the diffusion of their methodologies and the obliteration of semantic bar-
riers.

REFERENCES

1. ALEXANDER, H. E. AND G. LEIDY. J. Expn. Med. 93: 345-359, 1951.
2. ALTENBURG, E. Am. Natur. 82: 252-264, 1948.
3. ANDERSON, E. S. J. Hyg. 49: 458-470, 1951.
4. ANDERVONT, H. B. AND T. DUNN. J. Nat. Cancer Inst. IO: 1157-1189, 1950.
5. ANDREWES, C. H. Proc. Roy. Sot., London, s. B. 139: 313-326, 1952.
6. ASCHNER, M. PafUSitOlOgy 20: 309-314, 1934.
7. ATWOOD,,K. C. Genet&&: 95-96, 1950.
8. ATWOOD, S. S. AND K. W. KREITLOW. iim. I. Bot. 3.3: 01-100, 1946.

9. ALTERBACH, C. Genetics 32: 3-7, 1947.
IO. AUSTRIAN, R. But. Rev. 16: 31-50, 1952.
II. BA~DEN, F. C. Plant Viruses and Virus Diseases. Waltham, Mass.: Chronica Botanica, 1950.
12. BA~DEN, F. C., B. KASSANIS AND H. L. NIXON. J. Gen. Microbial. 4: 21*219, 1950.
13. BEADLE, G. W. Chem. Rev. 37: 15-96, 1945.
14. BEADLE, G. W. AND V. COONRADT. Genetics 29: 291-308, 1944.
15. BEALE, G. H. Genetics 37: 62-74, 1952.
16. BERTANI, G. J. But. 62: 1951.
                  293-300,
17. BILLINGHAM, R. E. AND P. B. MEDAWAR. Heredity 4: 141-146, Igso.
18. BLACK, L. M. Ann. New York Acad. SC. In press.
19. BLAKESLEE, A. F. Genetics 11: 17-36, 1921.
20. BLUM, H. F. Time's Arrow and Evolution. Princeton, N. J.: Princeton, r9jr.
21. BOIVIN, A. Cold Spring Harbor Symp., Quant. Biol. 12: 7-14, 1947.
22. BORDET, J. AND P. BORDET. Ann. Inst. Pasteur 72: 161-173, 321-334, 1946.
23. Born, J. S. K. J. Path. & Bact. 52: 501-517, 1950.

24. BOYD, J. S. K. J. Path.& Bact. 63: 445-457, 1951.
2~. BFLECHER. G. AND V. B. WIGGLES~ORTH. Parasitolo.ey 35: 220-224, 1944.
26. BROWN, C. H. Nature, London 166: 527, 1950.
27. BRUES, C. T. AND R. C. DUNN. Science 101: 336-337, 1945.
28. BUCHNER, P. Tier und PJeanze in Symbiose. Berlin: Borntraeger, 1930.
29. BUCHNER, P. Endosymbiose. Basel: Verlag Birkhauser, 1952.
30. BUCKLEY, H. E. Crystal Growth. New York: Wiley, 1951.
31. BURGEFF, H. Zlschr. Bat. 23: 589-608, 1930.
32. BURNET, F. M. J. Path. h Bact. 35: 851-863, 1932.
33. BURNET, F. hf. Biol. Rev. g: 332-350, 1934.
34. BURNET, F. M. Virus as Organism. Cambridge, Mass.: Harvard, 194s.
35. BURNET, F. M. Bzrll. Johns Hopkins Hosp. 88: 119-180, 1951.
36. CARTER, W. Phylopathology 26: 176-183, 1936.
37. CASPARI, E. Advances Genetics 2: r-66, 1948.
38. CASPERSSON, T. Cell Growth and Cell Function. New York: Norton, 1950.
39. CLARKE, N. A. J. Bact. 63: 187-192, 1952.
40, CLAUDE, A. Harvey Lect. 43: 121-164, 1947.
41. COHEN, S. S. Bact. Rw. IS: 131-146, 1951.
42. Cold Spring Harbor Symp., Quanl. Biol. Vol. II, Heredity and Variation in Microorganisms.

Long Island Biol. Assn., 1946.

43. Cold Spring Harbor Symp., Quant. Biol. Vol. 16, Genes and Mutations. Long Island Biol.

Assn., 1951.

44. CRAIGIE, J. AND A. FELIX. Lancet 1: 823, 1947.
45. DARLINGTON, C. D. AND K. MATHER. Genes, Plants and People. London: Allen & Unwin,

1950.

46. DELBRUCK, M. Harvey Lect. 41: 161-187, 1946.


October 1952     CELL GENETICS AND HEREDITARY SYMBIOSIS         427

47. DELBRUCK, M. B&l. Rev. 21: 3+40, 1946.
48. DELBRUCK, M. (cd.) I'irr~ses rg;o. Pasadena: Biol. Div., Calif. Inst. Tech., 1950.
49. D'HERELLE, F. Tfre Bacteriophage and its Beftaoior (G. H. Smith, transl.). Baltimore: Williams

& Wilkins, 1926.
50. DIPPELL, R. V. Heredity 4: 165-788, 1950.
gr. DOERMANN, A. H. 1. Bacl. 55: 7.57-276, 1948.
52. DOERYANN, A. H. Federation Proc. IO: 591-594, 1951.
53. DOORENBOS, W. .4nn. Inst. Pasteur 48: 457-469, 1932.
54. DUNBAR, J. M. Nalure, London 162: 851, 1948.
55. DUNN, L. C. (ed.) Genetics in the Twentietlt Century. New York: Macmillan, 1951.
56. EAST, E. ?vi. Am. Natur. 68: 289-303, 402-439, 1934.
57. EPHRUSSI, B. AND H. HOTTINGUER. Cold Spring Harbor Symp., Quanl Biol. 16: 75-85, 1951.
58. EPHRUSSI-TAYLOR, H. Cold Spring Harbor Symp., Quant. Biol. 16: 445-456, 1951.
59. EPHRUSSI-TAYLOR, H. Exper. Cell Res. 2: 589-607, 1951.
60. FISK, R. T. J. Inject. Dis. 71: 152-160, 1942.
61. FREDERICQ, P. Rev. beige path. et mti. ezper. 19 (Suppl. 4): r-107, 1948.
62. FREEMAN, V. J. AND I. U. MORSE. I. Bact. 63: 407-414, 1951.
63. FRENCH, R. C., S. M. LESLEY, A. F. GRAHAM AND C. E. VAN ROOYEN. Canad. J. Med. SC.

  29: 144-148, 195r.
64. FROBISHER. M. AND T. H. BROWN. Bull. Johns Hopkins Hosp. 41: 167-173,
6;. F~KUSHI, ?. J. Fat. 2gr. Hokkaido Imp. Univ. (S&poro) 4;: 83-154, 1940. 1927.

66. FURTH, J., M. C. BOON AND N. KALES. Cancer Research 4: I--IO, 1944.
67. GABELMAN, W. H. Proc. Nat. Acad. SC. 35: 634-640, 1949.
68. GIER, H. T. J. Bact. 173-189, 1947.
                53:
69. GILBERT, P. W. Ecology 23: 215-227, 1942.
70. GILDEMEISTER, E. AND I. AHLFELD. Zentralbl. f, Bakt. 147: 417-437, 1941.
71. GLASER, R. W. J. Parasitol. 32: 483-489, 1946.
72. GLEDHILL, A. W. AND C. H. ANDREWES. Brit. J. Exper. Path. 332: 559-j68, 1951.
73. GOLDSCHMIDT, R. Physiological Genetics. New York: McGraw-Hill, 1938.
74. GRANICK, S. Harvey Lect. 44: 22~~245, r949. In FRANCK, J. AND W. E. LOOMIS. Photosynthesis

in plants. Ames: Iowa State College, 1949, pp. 113-132.
75. GREEN, D. E. Biol. Rev. 26: 410-455, 1951.
76. HAAN, H. DE. Bibliographica Genetica IO: 357-412, 1933.
77. HASSID, W. Z., M. DOIJD~ROFF AND H. A. BARKER. In SUMNER, J. B. AND K. MYRBKCK (eds.).
The Enzymes. New York: Academic Press, 1951, vol. I, chap. 31, pp. 10x4-1039.
78. HERSHEY, A. D., C. ROESEL, M. CHASE AND S. FOUNAN. Carnegie Inst. Wash. Yrbk. so: 195-

  200, 1951.
79. HERSHEY, A. D. AND R. ROTMAN. Genetics 34: 44-71, 1949.
80. HOLXES, S. J, Organic Form and Related Biological Problems. Berkeley, Calif.: Univ. California

Press, 1948.
81. HOTCHKISS, R. D. Cold Spring Harbor Symp., Quant. Biol. 16: 457-462, 1951.
82. HUNGATE, R. E. Ann. Rev. Microbial. 53-66,
                         4: 1950.
83. HUSKINS, C. L. Am. Scientist 39: 688-699, 1951.
84. IMAI, Y. Cylologia (Fujii Jubilee Volume): 934-947, 1937.
8s. ITO, K. J. Econ. Entom. 31: 291-298, 1938.
86. JACOB, F., L. SIMINOVITCH AND E. WOLLMAN. Corn@. rend. Acad. d. SC. 233: 15oo-1502, 1951.
87. JOHNSON, J. J. Agti. Research 64: 443-454, 1942.
88. JONES, D. F. Genetics 29: 426427, 1944.
89. JONES, D. F. Genetics 35: 507-512, 1950.
90. JONES, D. F. Proc. Nat. Acad. SC. 37: 40%410, 1951.
91. KEEBLE, F. Plant Animals: A Study in Symbiosis. London: Cambridge Univ. Press, 1912.
92. KIRBY, H. In CALK~NS, G. N. AND F. M. SUM~~ERS. Protozoa in Biological Research. New

York: Columbia Univ. Press, 1941, pp. roo&r113.
93. KNIGHT, B. C. J. G. Vilamins &Hormones 3: Iog-2z8b, 194s.
94. KOCH, A. &Of. Zmtralbl. 53: 19~203, 1933.
95. KOCH, A. Ztscfrr. f. Morphol. okol. T&e 32: 137-180, 1936.
96. LACKEY, J. B. Biol. Bull. 71: 492-497, 1936.
97. LANHAN, U. N. Science IIS: 4sg-460, 1952. 116: 332-333, 1952.
08. LAW, L. W. Nature, London 169: 628-629, 1952.
99. LEACH, J. G. Insect Transmission of Plant Diseases. New York: McGraw-Hill, 1940.


Volume 3.2

428               JOSHUA LEDERBERG

Ioo. LEDERBERG, E. RI. AND J. LEDERBERG. Geaelics. In press.
101. LEDERBERG, J. Heredity 2: 145-198, 1948.
Io2. LEDERBERG, J. .Inn. ROD. Jlicrobiol. 3: 1-22, 1949.
103. I.EUEKBERG, J., E. M. LEDERBERG, -N. D. ZINDER AND E. R. LI\%LY. Cold Spring Harbor

104.
105.
106.
10;.
108.
`09.
I IO.
III.
112.
113.
II.+.

115.
116.
117.
118.
119.
120.
121.
122.
123.
124.
125.
126.
127.
128.
129.

130.

131.
132.
`33.
134.
135.
136.
`37.
138.
139.
140.
141.
142.
143.
144.
14.5.
146,
147,
148

`49,
150.
151.
152
`53,

L'HERITIER, P. Cold Spring Harbor Symp., Quant. Biol. 16: 99-112, 1951.
LINUEGREN, C. C. The I-east Cdl. St. Louis: Educ. Pub]., 1949.
LINDECREX; C. C. AXD G. LINDEGREN. J. Gen. Jficrobiol. 5:.885-893, 1951.
LITTLE, C. C. Biol. Rm. 22: ,315-343, 1947.
LOEFER, J. Aw Nalw. 70: 184-188, 1936.
LBFFLER, E. 2entraZbZ.f. Bakt. 96: 398-402, 1925.
LUCAS, C. E. Sot. Exper. Biol. Symp. 3: 33635'6, 1949.
LURIA, S. E. Genetics 30: 84-99, 1945.
LURIA, S. E. A? M. L. HUMAN. J. But, 59: 551-560, 1950.
LWOFF, A. L'Evolution Physiologique: Et&e des perfes de jonctions ckez les tnicroorganisnles.
Paris: Hermann, 1944.
LWOFF, A. Problems of Morphogenesis in Ciliutes. New Yorlr: Wiley, 1950.
LWOFF, A. New Phytol. 49: 77-80, 1950.
LWOFF. A. AND A. GUTMANN. Ann. Inst. Pasteur 78: 711-739, 1950.
LWOFF, A., L. SIXINOVITCH AND N. KJELGAARD. Ann. Inst. Pus&w 79: 815-858, 1950
MACDOWELL, E. C. Cold Spring Harbor Syrup., @ant. Biol. II: I+176, 1946.
M.~LY, R. Ztschr. indukt. Abstammungs- u. Vererbwzgslehre 83: 447-478, 1951.
MATHER, K. SOG. Exper. Biol. Symp. 2: 196-216, 1948.
MCCLINTOCK. B. Ztsclw. f. Zellforsch. u. mikr. Anat. 21: 294-328, 1934.
MCCLINTOCK; B. Cold S$rin.g Harbor Symp., Quant. Biol. ;6: 13-47,.1951
McKIE, M. Australian J. Exper. Biol. & M. SC. 12: 169-175, 1934.
MEDAWAR, P. B. Biol. Rmv. 22: 360-389, 1947.
MEYER-ABICH, A. Bibliotheca Biotheoretica 5: 1-206, 1950.
MIRSKY, A. E. Science 101: 15-16, 1945.
MONOD, J. AND M. COHN. Advances Enzymol. 13: 121-149, 19jZ.
MORGAN, T. H., A. H. STURTEVANT, H. J. MULLER AND C. B. BRIDGES. The Mechanism
of Mendelian Heredity. New York: Holt, 1923.
MOULTON, F. R. (ed.) Symposium on Mammary Tumors in Mice. A.A.A.S. Publication No.
22, 1945.
MULLER, H. J. Proc. Roy. SOG., London, s. B. 134: 1-37, 1947.
MULLER, W. Arch. f. Mikrobiol. 5: 84-147, 1935.
MURRAY, R. G. E., D. H. GILLEN ANLI F. C. HEAGY. J. Bact. 59: 603-615, 1950.
MUSGRAVE, A. J. AND J. J. MILLER. Canad. Entom. 83: 343-345, 1951.
NEWCOMER, E.H. Bot.Rev. 6: 85-147, 1940; 17: 53-89, I9jI.
NICOLLE, P., G. GRABAR AIVD P. GILBERT. Ann. Inst. Pasteur 72: 8x-88, 1946.
NICOLLE, P. AND Y. HAMON. Aw. Inst. Pasteur 81: 614-630, 1951.
NOLAND, J. L. AND C. A. BAUXANN. Ann. Entnm. Sot. Am. 44: 184-188, I9jI.
OPARIK, A. &igin of L;fe. New York: Macmillan, 1938.
PANT, N. C. AND C. FRAENKEL. Science 112: 498-509, 1950.
PARKER, R. C. J. Exper. Zool. 46: 1-12, 1926.
PAWHER, A. J&b. w&s. Bot. 71: 386-462, 1929.
PETER, K. Ztschr. Zellforsch. u. mikr. Anal. 30: 721-750, 1940.
PIEKARSIX, G. Zentralbl. j. Bokt. 153: 109-115, 1949.
POLLISTER, A. W. AND P. F. POLLISTER. Aw. New York Acad. SC. 45: 1-48, 1943.
PONTECORVO, G. Cold Spring Harbor Sump., Qualtt. Biol. II: 193-201, 1946.
PONTECORVO, G. Sot. Exper. Biol. Symp. In press.
POTTER, V. R., J. M. PRICE, E. C. MILLER AND J. A. MILLER. Cancer Research IO 28 .%5,
1950.
PREER, J. R. Genetics 33: 349-401, 194s.
PREER, J. R. Genetics 35: 344-362, 1950.
PRINGSHEIM, E. G. Arch. .f. Protistenk. 64: 289-418, 1928.
PRINGSHEIN, E. G. AX;D R. HO~ASSE. New Phytol. 47: 52-87, 1948.
PRINGSIIEIJI, E. G. AND 0. PRINGSHEIM. Nea Phytol. 51: 65-76, 1952.


Ocfober rgjz     CELL GENETICS XXD HEREDITARY SY1lBIOSIS         429

154.

'55.
156.

157.
Ij8.
159.
160.
161.
162.
163.
164.
165.
166.
167.
168.
169.
170.
171.
172.
173.
174.
175.
I 76.
177.
I 78.
179.
180.
181.
182.

183.
184.
185.
186.
187.
I 88.
189.
rgo.
191.
192.
`93.

PROVASOLI, L., S. H. HUTNER AND I. J. PINTNER. Cold Spring Horbor Symp., Quani. Biol.
16: 113-120, 1951.
RAVIN, A. W. AND A. NORMAN. Erperientia 8: 108-110, 1952.
REDFIELD, R. (ed.) Le& o/ In@rafion in Biological and Social Sysiems. Biol. Spmp. I 2, Lan-
caster, Pa.: Jacques Cattell Press, 1942.
RENNER, 0. Flora 30: 218-290, 1936.
RHOADES, M. &I. J. &ti&S 27: 71-93, 1933.
RHOADES, M. M. Cold Spring Harbor Symp., Qua&. Biol. I I: 202-207, 1946.
RHOADES, hl. M. Proc. Nat. tlcad. SC. 36: 634-635, 1950.
RIZET, G. AND E. ENGELMANN. Rev. Cytol. Biol. vEg. II: 201-303, 1949.
RO~IAN, H., D. C. HAIVTHORNE ohm H. C. DOUGLAS. Proc. &Vat. Acad. SC. 37:`79-84, 1951.
ROPER, J. A. Experientia 8: 14-15, 1952.
ROSE, S. M. AND F. C. ROSE. Cancer Research 12: 1-12, 1952.
ROUNTREE, P. M. J. Gen. Microbial. 3: 153-163, 1949.
RYAN, F. J. AND J. LEDERBERG. Proc. Nat. Acad. SC. 32: 163-173, 1946.
SCHIFF, F. ANJ S. BORNSTEIN. J. Immunol. 39: 361-367, `940.
SCHNEIDER, W. C. AND G. H. HOGEBOOM. Cancer Research II: 1-22, 1951.
SCHOLTENS, R. T. Antonie van Leeuwenhoek 16: 256268, 1950.
SCHOPFER, W. H. Plants and Vitamins. \Valtham, Mass.: Chronica Botanica, 1943.
SCHRADER, F. Viol. Bull. 4j: 229-302, 1923.
SCHWARTZ, D. Bot. Gaz. III: 123-130, 1949.
SCHw'ARTZ, D. Genetics 36: 676-696, 1951.
SCHWARTZ, w. Arch. f. Mikrobiol. 6: 369-460, 1935.
SKAAR, P. D. Ph.D. Thesis. Indiana University, 1952.
SNELL, G. D. J. Genetics 49: 87-108, 1948; Proc. 1st U. S. Natl. Cancer Con/. 28-33. 1949.
SONNEBORN, T. M. Proc. Nat. Acad. SC. 29: 329-343, 1943.
SONNEBORN, T. M. Advances Genetics I: 263-358, 1947.
SONKEBORN, T. M. Ann. Rev. Microbial. 3: 55-80, 1949.
SONNEBORN, T. M. Harvey Lect. 44: 145-164, 1949.
SONXEBORN, T. M. Heredity 4: 11-36, I9jo.
SONKEBORN, T. M. In BAITSELL, G. A. (ed.). Science in Progress. 7th series. New Haven:
Yale, 1951, pp. 167-203.
SPIEGELMAK, S. Sot. Eaper. Biol. Symp. 2: 286-325, 1948.
SPIEGELMAN, S. Cold Spring Harbor Symp., Qzdanl. Biol. 16: 87-98, 1951
STADLER, L. J. Cold Spring Harbor Symp., Quant. Biol. 16: 49-63, 1951.
STANIER, R. 2. Ann. kev<Wicrobiol.~~: 35-56, 1951.
STEINHAUS. E. A. Insect Microbioloav. Ithaca. N. Y.: Comstock, 1046.
STEINHA& E. A. Principles of I&t Pathoiogy. New York: &&awHill, 1949.
STERN, C. Genetics 21: 625-730, 1936.
STERN, C. ilm. Nalur. 72: 3je357, 1938.
STOREY, H. H. BOl. Rev. 5: 24-272, 1939.
STRONG, L. C. Genetics II: 294-303, 1926.
STURTEVANT, A. H. In DUNN, L. C. (ed.). Genetics in tire Twentieth Century. Pu'ew York: Mac-
millan, 1951, pp. IOI--110.
TAYLOR, M. J. AND E. C. MACDOWELL. Cancer Research g: 144-149, 1949,
THOMAS, R. C. Ohio J. SC. 48: 102-106, 1948.

`94.
`95.
,

190. TRAUB, E. J. Exper. Med. 68: 229-250, x938.
r97. C'nitts biologiques do&es de con&zu& gtnt+he. Centre National de la Recherche Scientifique,

  colloques internationaux No. 8. Paris, 1949.
198. WALLIN, J. E. Symbionticism and the Origin of Species. Baltimore: Wilhams & Wilkins, 1927.
199. WEBB, J. L. ParaSitOlOgy 32: 355-360, 1940.
zoo. WEIER, T. E. AND C. R. STOCKING. Bot. Rey. 18: 14-75, 1952.
201. WEIGLE, J. J. AND M. DELBRUCK. J. Bacl. 62: 301-318, 1951.
202. WEISS, P. Quart. Rev, Biol. 25: 177-198, 1950.
203. `1\IEISZ, P. B. Am. Natur. 85: 293-317, 1951.
204. WHEELER, W. M. Essays in Philosophical Biology. Cambridge, Mass.: Harvard, 1939.
2Oj. WHITE, P. B. J. Path. & Bact. 44: 276-278, 1937.
206. WHITE, P. R. Quart. Rev. Biol. 26: 1-16, 1951.
207. WICHTERMAN, R. dill. gull. 81: 304-305, 1941 and 94: II3-`27, 1948.


430               JOSHUA LEDERBERG             Volume 3.2

208. WIGGLEWORTH, V. B. Principles oj Insect Physiology (4th ed.). New York: Dutton, Igsr.
209. WILLIAMS-SMITH,H.J.H~~. 46: 7&3g,Ig&3.
210. WILLIAMS-SXITH, H. J. Cert. Microbial. 5: 458-471, zgsr.
211. WILSON, E. B. The Cell in De&opv:ent and Zwkerilance. New York: Macmillan, 1896.
212. \VILSON, E. B. T/&e Cell in Dmelopment und Heredity (3rd ed.). New York: Macmillan, 1925.
213. WILSON, E. B. Am. Natur. 59: 481-486, 1925.
214. WILSON, E. B. J. Morphol. 52: 429-483, 1931.
215. WINCE, 0. AND C. ROBERTS. Compt. rend. d. trm. du lab. Carlsberg sirie physiol. 24: 263-315,
  '948.
216. JVINGE,O. It'allerslein Labs. Comntwt. 15: 21-42, 1952.
217. WOLLMAN, E. AND E. WOLLYAN..~~~. Inst. Pasteur 60: 13-57, 1938.
218. WOODS, M. W. AND H. G. DuB~Y. J. Nat. Comer Znsf. II: frog-1151, I9jI.
219. WRIGHT, S. Pkysiol. Rezl. 21: 487-527, 194~.
220. WRIGHT, S. ilm. N&w. 79: 289-303, 1945.
WI. YONGE, C. M. Biol. Rex 19: 68-80, 1944.
222. ZINDER, N. D. AND J. LEDERBERG. J. But. In press.
223. BAKER J. R. Quart. J. M&r. SC. 93: 157-rgo, 1952.
224. Conference on Virtrses as Cansdioe Agents in Cancer. Ann. New York Acad. SC. 54, 1952.
225. LETTR~,, H. Naturwissensch. 37: 335-336, 1950.
226. SMITH, K. M. Biol. Rm. 27: 347-357, 1952.
227. STASNEY, J.,.;i. CANTAROW AND K. E. PASCHKIS. Cancer Research IO: 775-782, 1950.