Gene transfer from cell to cell through cell division, i.e., asexual reproduction, is the fundamental process of genetics. Practically all of the procaryotes and some eucaryotic unicellular organisms have only this axis: they repeat horizontal inheritance indefinitely. In adverse environments, they can survive either by a makeshift mechanism (a transient stop of cell division usually followed by development of defensive structures such as spores) or by elimination of a part of the clonal population that failed to adapt to the environment.

Most multicellular organisms have another axis, vertical inheritance: gene transfer from one generation to the next generation. The core process of this axis is meiosis (2n®n) and fertilization (n®2n). It is only when this axis was introduced that clonal termination became possible (the horizontal axis came to be mortal).

Eucaryotic cells, which have an average diameter of about 10 µm, are 1,000 times as large in volume as procaryotic cells, which have an average diameter of about 1 µm.

Supposing that an ancestral procaryote with a single-copy genome increased the genetic material concomitantly with the cell volume, the increased DNA would require more time for its replication and become susceptible to more chance of mutations. Most mutations in the single-copy genome would result in the death of their possessors, unless they were protected by safety devices. Besides the nucleosome structure of the chromosomes stored in the nucleus, the following two may be of special importance as safety devices; 1) differentiation of the cells into the treasured "germ", which inherited the immortal nature of procaryotes and the "soma", which can utilize costly energy for work other than cell division, and 2) possession of genetic material in duplicate, i.e., genomic diploidization.

A possible scenario of the evolution of cellular lifespan
The scenario will be summarized as a proposal that the safety devices, i.e., "differentiation into germ and soma" and "diploidization", areÊcoupled to the interrelated brake systems to control replication in soma, to induce apoptotic degradation of soma, and to initiate meiosis in germ.
4.1 Differentiation into germ and soma The original immortal cells are alternatively called "germ" cells. It appears that germ and soma were, at the very beginning stage, identical to proliferative and non-proliferative, respectively, and the term germ has nothing to do with sexual reproduction. Differentiation into germ and soma is possible only in a multicellular (or multinuclear) state. Since procaryotes remain potentially immortal, although they can assume a multicellular state [4], this differentiation may have occurred after eucaryotes evolved. Since this differentiation is seen in haploid eucaryotes if they form a colony (e.g., Volvox) or aggregate to a multicellular form (e.g., Dictyostelium), this differentiation may have occurred before diploidization evolved in eucaryotes.

Later in the evolutionary process, probably after diploid cells evolved, the germ came to be associated with sexual reproduction, introducing the vertical genetic axis, and the soma came to have a regulated proliferative ability. Thus, both germ and soma are proliferative, but the proliferative ability of the soma is limited. This is what we call cellular lifespan.

In almost all ciliates, in which differentiation into the germ (diploid micronucleus) and soma (polygenic macronucleus) occurs at the nuclear level, and in which both the germ and soma are proliferative, the germ is produced from the germ and the soma from the soma during asexual reproduction. During sexual reproduction, both the germ and soma are produced from the germ passing through meiosis, mitosis, fertilization and mitosis (the old soma is destroyed).

In the karyorelictids (e.g., Loxodes), which are thought to belong to a primitive group of ciliates and in which the macronucleus has a diploid G₂ amount of DNA and is still non-proliferative, the soma is formed from the germ not only during sexual reproduction but also during asexual reproduction. The old soma is not destroyed, either during sexual reproduction or during asexual reproduction. During sexual reproduction in the heterotrichid ciliate Blepharisma, the soma is produced from the germ either by passing through a series of meiosis, mitosis, fertilization and mitosis or by direct differentiation of the germ [5]. Even in Paramecium, the soma can be produced directly from the germ if the germ is artificially placed in exconjugant cytoplasm in which the macronucleus is differentiating from the fertilized micronucleus [6]. These features suggest that for the production of the proliferative (to a limited extent) soma from the germ, what is essential is not an experience of "undergoing meiosis and fertilization" but one of "being placed in a spatio-temporally specific field" which is created through meiosis and fertilization. At least in ciliates, the mechanism for initiating meiosis in the germ seems to be coupled to the mechanism for destroying the soma (apoptotic degradation of the old macronucleus).
4.2 Alternation between haploid and diploid In enlarged eucaryotes, harmful effects of mutations are mostly avoided if diploidization takes place. However, if diploidization is only for the sake of avoidance of mutational effects, polyploidization would be even more beneficial. Diploidy instead of polyploidy would have been preferred for the selective advantage of the ease of reverting to the haploid state.

Mutations are resources for evolution, although harmful in most cases. Some mutations, if combined with others, would become harmless or even beneficial. In diploid cells, numerous combinations of mutations are stored in one genome set. However, recessive mutations are not expressed unless their possessors become haploid or are recessive homozygotes.

4.3 When do diploid cells become haploid?
Alternation between haploid and diploid constitutes the regular process of mitosis for haploid cells, but would later become a well-controlled process in diploid cells. The controls involve when and how haploidization occurs. In both paths of diploidization, i.e., genomic duplication in and homotypic cell fusion of haploid cells, the two sets of genomes would remain identical unless mutations were stored in one of them. In this situation, chromosomal exchange or crossing over does not produce genetic variations, so that there is no selective advantage to employing recombination at the DNA level so long as alternations between haploid and diploid are occurring as frequently as elimination of the accumulation of mutations.

In order to make full use of the strategy, "store sufficient mutations and then test their availability", the timing of haploidization should be controlled. The duration of the diploid state until the time of haploidization, i.e., the duration for sexual maturity, would have been subjected to natural selection.

In the diploid cells which differentiate into germ and soma, haploidization takes place in the germ line. For the diploid germ cells, to become haploid means that the continuity of their diploidy ends. This contradicts the immortality of the horizontal genetic axis in the germ line. On the other hand, the diploid soma cells do not have to stop dividing to become haploid, so that they can theoretically be immortal. This contradiction should be resolved in some way. My speculation is that the mechanism for the interruption of cell division in diploid germ cells to promote meiosis was utilized as the brake system to limit the division potential in diploid soma cells. This indicates that the mechanism of limiting the cellular lifespan in soma might have developed in close association with the mechanism of sexual reproduction in germ; in other words, the brake system to suppress cell division in soma might have coupled to the accelerator system to initiate meiosis in germ.

Paramecium as a model for the study of cellular lifespan
5.1 Sonneborn limits in ciliates
Paramecium was once thought to be immortal (able to multiply indefinitely), as represented by the "Methuselah" strain of one of the species of the P. aurelia complex (designated now as P. biaurelia), which had been cultured without loss of vigor for 33 years. The possibility of immortality was disproven by the finding of periodic occurrence of autogamy, the nuclear process accompanying meiosis and fertilization in a single, unpaired cell [1]. Also in P. caudatum, in which continuous cultivation without loss of vigor for 22 years had been reported, the alleged immortality was disproven [7]. The Sonneborn limits counted in the number of cell divisions are: 200-350 for the P. aurelia complex, including P. primaurelia, P. biaurelia, P. tetraurelia and P. octaurelia, 500-750 for P. caudatum, and longer than that for P. multimicronucleatum and P. bursaria [8].

The Sonneborn limit is also applicable to ciliates other than Paramecium, such as Tetrahymena, Euplotes, Stylonychia, Oxytricha, Spathidium and Tokophrya [9]. The sole exception is the amicronucleate strain GL of Tetrahymena pyriformis, which is genetically dead because of the lack of a germinal micronucleus and yet immortal in that it can multiply indefinitely [10].

5.2 Association of the clonal lifespan with other life cycle features
5.2.1 Clonal lifespan versus the length of sexual immaturity
There is a loose coupling between the length of lifespan and the length of development (species that have longer lifespans tend to have longer periods of sexual immaturity) in such remote groups of animals as mammals [11] and ciliates [12]. These observations suggest that the mechanism used for measuring the sexual maturation is also used for measuring the clonal lifespan.

In the jumyo mutant of P. tetraurelia, which has an extremely short clonal lifespan (see the last section for detail), the length of autogamy immaturity is also short [13]. In wild type cells, the length of sexual immaturity (the number of cell divisions until conjugation becomes possible), the length of autogamy immaturity (the number of cell divisions until autogamy becomes inducible by natural starvation), and the length of clonal lifespan are all shortened by starting the new generation from a parent of more advanced clonal age. These carry-over effects of the parental age on the next generation may be related to the efficiency of a clock rewinding mechanism that may work during sexual reproduction.

5.2.2 Clock rewinding during sexual reproduction
The questions of why and how cells are rejuvenated during conjugation and autogamy remain unanswered. The generally accepted idea is that the mutations accumulated in the macronucleus are canceled by degrading it and constructing a new macronucleus from the germ micronucleus. Mutations are also accumulated in the micronucleus, but deleterious mutations in the micronucleus may be eliminated through recombination at meiosis, and the intact genome will be restored in the differentiating macronucleus. In this explanation, accumulation of mutations is the clock and their elimination through recombination is the clock rewinding.

In P. tetraurelia, cells of the age of around 20 cell divisions can be reset to time zero through autogamy. If they repeat autogamy every 25 cell divisions, they can live indefinitely. According to Sonneborn [14], the frequency of death at autogamy was only 0.7% on average in 50 successive autogamies induced every 25 cell divisions of parent clones. Note that autogamy is the process that occurs in a single cell and consists of meiosis, mitosis, fertilization and mitosis of the micronucleus to result in homozygosity in all genetic loci. If mutations do not occur in the micronucleus during 2 successive autogamies, the parent and the autogamy-progeny are identical in both their genetic and cytoplasmic compositions, and yet different in their clonal ages. The clock rewinding in this case has nothing to do with mutation canceling.

Then, what kind of mechanism for clock rewinding is possible? My speculation is that the clock rewinding in the nucleus will be done through interactions with the spatio-temporally specific cytoplasm. Although the mechanism is yet unknown, it should work sometime during nuclear construction during autogamy (or conjugation), including the period encompassing prezygotic and postzygotic micronuclear changes, development of the new macronucleus (anlage), and degradation of the old macronucleus. The cytoplasm as well undergoes dramatic changes spatially and temporally during this period. Since what is actually rejuvenated is the macronucleus, which will govern the phenotypes of the progeny, nucleo-cytoplasmic interactions via postzygotic micronuclear changes appear to be pivotal. The nuclear changes during this period are: the fertilized nucleus divides mitotically 2 times to produce 4 products, 2 of which are located in the posterior region of the cytoplasm and differentiate into macronuclear anlagen, and the other 2 of which are located in the anterior region of the cytoplasm and remain as micronuclei; the old macronucleus is broken into fragments (the pattern of nuclear differentiation differs from species to species).

The fertilized nucleus is not necessarily destined to differentiate into the macronucleus, because it behaves as an ordinary germ micronucleus if transplanted into a vegetative cell. What is special is the cytoplasm around this time, because every product of a fertilized nucleus, and even the micronucleus of a vegetative cell, can differentiate into a macronucleus if transplanted into the posterior region of the specific cytoplasm. The old macronucleus plays an important role in creating the specific cytoplasm, because no differentiation into macronuclear anlagen occurs if the old macronucleus is eliminated soon after the first division of the fertilized nucleus.

The hypothesis claiming that the time-space-specific cytoplasmic field created in the pre-development cell plays a key role in rewinding the cellular clock can be applied to explain the puzzling question of how the cloned sheep Dolly has overcome the Hayflick limit: in the experiment to create Dolly, the nucleus from an udder cell of a ewe was placed in an unfertilized egg, which is thought to be the spatio-temporally specific cytoplasm for clock rewinding. The signal to start development is then food supply in ciliates and electric pulse in the case of Dolly. In this context, it will be worth studying whether (and to what extent) exconjugant (or exautogamous) cells carrying a macronucleus derived from a transplanted micronucleus will be rejuvenated.

5.2.3 Clonal lifespan versus cultural lifespan
When Paramecium cells are deprived of food resources, they stop dividing, and the starving cells (quiescent cells) undergo physiological changes called cultural aging. The cultural lifespan is referred to the period until starving cells die. The process of cultural aging may be viewed as the cellular response to the environmental stress of starvation, which has been studied extensively in bacteria. It is during this process that paramecia (and also other ciliates) initiate sexual reproduction (meiosis), or, in other words, switch the horizontal genetic axis to the vertical one.

There are some similarities between clonal aging and cultural aging in such respects as increasing morphological and physiological disorder and increasing UV sensitivity. The two processes may be more than superficially similar: we recently found that the length of autogamy immaturity became shorter when the new autogamous generation was started from the cells with more advanced cultural age. However, the relationship between these two processes remains to be defined.

Some relevant information from yeast cells appears to be worth mentioning. Kennedy et al. [15] isolated mutants of Saccharomyces cerevisiae that have long and short cellular lifespans, and found that those with long lifespans were also long in cultural lifespan in the sense that they were more resistant to stresses such as starvation or low temperature than those with short lifespans. The gene responsible for the long cellular lifespan and stress resistance was identified as the silencing gene SIR4, the product of which is a member of the protein complex called SIR (silent information receptor).

5.3 Coupling between the lifespan and sexual reproduction
Although the time-measuring mechanism is still unknown, it appears certain that the life-cycle staging of ciliates is genetically controlled. If so, alterations of the genetic program may alter the timing of the life cycle stages. Isolation of mutants with modified life cycle stages may lead to validation of this idea.

In both Tetrahymena thermophila and P. caudatum, mutants with short lengths of sexual immaturity have been isolated. They are all semidominant; dominant homozygotes have a shorter immaturity period than heterozygotes (in P. caudatum, for example, 20-25 cell divisions in dominant homozygotes, about 35 in heterozygotes, and about 50 in the wild type recessive homozygotes).

In P. tetraurelia, a mutant with an extremely short clonal life span (about one-tenth of the wild type) has been isolated in the author's laboratory. Breeding experiments revealed that recessive mutation in a single gene named jumyo was responsible. It has been shown that some previously isolated mutants of P. tetraurelia, such as am (abnormal macronuclear distribution to daughter cells) and some groups of nd (trichocysts non-discharging ) with macronuclear misdivision, have shorter lifespans than that of the wild type. The jumyo mutant is, however, normal in most of the aspects so far examined, including macronuclear division, trichocyst discharging, and the functions involved in sexual reproduction. The survival curve plotted against the clonal age shows a pattern of persistence of 100% survival followed by a sudden drop, not a linear regression.

An unusual characteristic found in the jumyo mutant is that it can undergo autogamy even when food is abundant. For some time after this mutant was isolated, the clonal termination was defined by either clonal death or autogamy; since then, cells terminating in autogamy have come to dominate the population of the stock culture. Also, in the wild type cells, autogamy occurs in nutrient rich conditions at late senescence. Clonal termination by autogamy does not practically differ from clonal termination by death, because autogamy at late senescence produces no viable progeny. In the jumyo mutant, however, autogamy occurs frequently under nutrient-rich conditions, and at very young clonal age (with short interautogamous intervals) to produce fertile progeny. These observations appear to support the speculation arguing that the sexual reproduction (haploidization system) is basically coupled to the cellular lifespan (brake system for stopping cell division).

Although the somatic cells of multicellular organisms which have cellular lifespans seems to have nothing to do with sexual reproduction, a cryptic relationship between the brake system for suppressing cell division in soma or even for destroying the soma, and the accelerator system for promoting meiosis in germ, sounds natural in ciliates, in which evolutionarily primitive systems are thought to have been preserved more than in mammals. During conjugation and autogamy in ciliates, the germinal micronucleus undergoes meiosis and fertilization, while the somatic macronucleus shrinks or is degraded to fragments and finally disappears. The commitment to macronuclear degradation occurs almost concomitantly with the commitment to micronuclear meiosis. Note that the life-managing functions in ciliates are carried out by the macronucleus, including the initiation, commitment and prosecution of sexual reproduction. It is noteworthy that the macronuclear degradation is suicidal (apoptotic), because apoptotic DNAs of ~200 bp and multiples thereof are detectable on agarose gel electrophoresis [16,17], and because macronuclear degradation is inhibited by inhibitors of gene expression such as actinomycin D and cycloheximide and by nuclease inhibitors [17,18]. The speculation of cryptic association between the brake system in soma and the accelerator system in germ leads me to speculate further: "Destroy the circuit of sexual reproduction, and you will get an eternal asexual life." This is testable in Paramecium by looking for an immortal mutant of P. tetraurelia by screening the cells that are unable to undergo autogamy.

References

1. Sonneborn TM (1954) The relation of autogamy to senescence and rejuvenescence in Paramecium aurelia. J. Protozool., 1: 38-53.
2. Hayflick L, Moorhead PS (1961) The serial cultivation of human diploid cell strains. Exp. Cell Res., 25: 585-621.
3. Takagi Y (1993) Organimic lifespan and cellular lifespan: from a viewpoint of Paramecium (in Japanese), Heibonsya, Tokyo
4. Kolter R, Losick R (1998) One for all and all for one. Science, 280: 226-227.
5. Miyake A, Rivola V, Harumoto T (1991) Double paths of macronucleus differentiation at conjugation in Blepharisma japonicum. Eur. J. Protistol., 27: 178-200.
6. Mikami K, Ng SF (1983) Nuclear differentiation in Paramecium tetraurelia. Transplantation of vegetative micronuclei into early exconjugants. Exp. Cell Res., 144: 25-30.
7. Takagi Y, Yoshida M (1980) Clonal death associated with the number of fissions in Paramecium caudatum. J. Cell Sci., 41: 177-191.
8. Takagi Y (1988) Aging. In: Gšrtz H-D (ed) Paramecium. Springer- Verlag, Berlin, pp 131-140.
9. Smith-Sonneborn J (1985) Aging in unicellular organisms. In: Finch CE, Schneider EL (eds) Handbook of the biology of aging, second edition. Van Nostrand Reinhold, New York, pp 79-104.
10. Nanney D (1974) Aging and long-term temporal regulation in ciliated protozoa: a critical review. Mech. Ageing Dev., 3: 81-105.
11. Cutler RG (1978) Evolutionary biology of senescence. In: Behnke JA, Finch CE, Moment GB (eds) The biology of aging. Plenum Press, New York, pp 311-360.
12. Smith-Sonneborn J (1981) Genetics and aging in protozoa. Int. Rev. Cytol., 73: 319-354.
13. Takagi Y, Suzuki T, Shimada C (1987) Isolation of a Paramecium tetraurelia mutant with short clonal life-span and with novel life-cycle features. Zool. Sci., 4: 73-80.
14. Sonneborn TM (1974) Paramecium aurelia. In: Mayr E (ed) Handbook of genetics, vol 2. Plenum, New York London, pp 469- 594.
15. Kennedy BK, Austriaco Jr NR, Zhang J, Guarente L (1995) Mutation in the silencing gene SIR4 can delay aging in S. cerevisiae. Cell, 80: 485-496.
16. Davis MC, Ward JG, Herrick G, Allis CD (1992) Programmed nuclear death: apoptotic-like degradation of specific nuclei in conjugating Tetrahymena. Dev. Biol., 154: 419-432.
17. Mpoke S, Wolfe J (1996) DNA digestion and chromatin condensation during nuclear death in Tetrahymena. Exp. Cell Res., 225: 3357-3365.
18. Mikami K (1996) Repetitive micronuclear divisions in the absence of macronucleus during conjugation of Paramecium caudatum. J Euk. Microbiol., 43: 43-48.