Factors Determining the Frequency of the Killer Trait Within Populations of the Paramecium Aurelia Complex

Factors Determining the Frequency of the Killer Trait Within Populations of the Paramecium aurelia Complex

Wayne G. Landis Environmental Toxicology Branch, Toxicology Division, Chemical Research and Development Center, Aberdeen Proving Ground, Maryland, 21010-5423 Manuscript received March 3, 1986 Revised copy accepted September 15, 1986

ABSTRACT The factors maintaining the cytoplasmically inherited killer trait in populations of Paramecium tetraurelia and Paramecium biaurelia were examined using, in part, computer simulation. Frequency of the K and k alleles, infection and loss of the endosymbionts,recombination during conjugation and autogamy, cytoplasmic exchange and natural selection were incorporated in a model. Infection during cytoplasmic exchange at conjugation and natural selection were factors that would increase the proportion of killers in a population. Conversely, k alleles reduced the proportion of killers in a population,acting through conjugation and autogamy. Field studies indicate that the odd mating type is prevalent in P. tetraurelia isolated from nature. Conjugation and therefore transmission by cytoplasmic transfer would be rare. Competition studies indicate a strong selective disadvantage for sensitives at concentrationsfound in nature. Natural selection must therefore be the factor maintaining the killer trait in P. tetraurelia.

YTOPLASMIC inheritance is a widespread phe- The two hosts, P. tetraurelia and P. biaurelia, are C nomenon that includes sex and drug resistance sibling species within the P. aurelia complex as con- in bacteria, CO2 sensitivity in Drosophilia melanogaster structed by SONNEBORN(1975). Paramecia of this and possibly certain tumors in mammals. Indeed, the complex undergo two major sexual processes, conju- origin of the eukaryotic cell may have been dependent gation and autogamy. Conjugation occurs when re- on certain chloroplast and mitochondria-like inclu- active cells of one mating type are mixed with those sions similar to the bacterial endosymbionts found in of the complimentary mating type, resulting in first many present-day organisms. GILL and HAIRSTON cell clumping and then pairing. Paramecia undergoing (1972) have proposed that bacterial symbionts may conjugation do not feed and thereby do not take up mediate protozoan community structure. Fertility the toxin produced by the killers. Sensitives are there- barriers among mosquito species of the genus Culex fore protected and can mate with killers. The two are caused by a symbiotic rickettsia (YEN and BARR micronuclei undergo meiosis and the macronucleus 1973). However, little attention has been paid to the breaks down. Haploid nuclei are exchanged, fertiliza- factors affecting the inheritance of cytoplasmically tion occurs and new macronuclei and micronuclei inherited traits at the population level. develop, restoring the normal vegetative state. During The killer trait, a classic case of cytoplasmic inher- conjugation, a cytoplasmic bridge may form, enabling itance, observed within Paramecium biaurelia and P. the cytoplasm of the two partners to mix. Autogamy tetraurelia is caused by the bacterial endosymbionts of is a form of self-fertilization rendering the individual the genus Caedobacter, commonly called kappa homozygous at all loci. It occurs at intervals in lines if (PREER, PREER andJuRAND 1974). Paramecia contain- starved paramecia are not allowed to mate. The proc- ing the endosymbiont (killers) release a toxin that ess requires a fission to restore the normal nuclear causes the death of paramecia not maintaining the complement. For a detailed account of conjugation same bacteria (sensitives). Sensitives incur symptoms and autogamy, see SONNEBORN(1943, 1974) and specific to the type of endosymbiont involved. The BEALE(1954). presence of kappa within a paramecium depends on Although the killer trait was among the earliest the KK or Kk genotype of the host. When an organism cases of cytoplasmic inheritance discovered and has becomes kk, the kappa are lost (SONNEBORN1943). S been extensively studied by PREER,SONNEBORN and alleles at loci SI and Sz, discovered within stock 29 of others, its population genetics, ecological and evolu- P. tetraurelia, increase the probability of kappa loss tionary significance largely have been ignored. GILL (BALBINDER1959, 1961). The distribution of kappa (1972; GILL and HAIRSTON1972) has hypothesized is worldwide, with stocks originating from North that the superior competitive ability of P. biaurelia America, Central America, Europe, Japan and Aus- compared to P. triaurelia and P. pentaurelia within a tralia (PREER,PREER and JURAND 1974). seep near Ann Arbor, Michigan, may be due to an

Genetics 115: 197-205 (January, 1987) 198 W. G. Landis unidentified killer endosymbiont. GILLand HAIRSTON TABLE I further surmise that endosymbionts may play an im- The changes of the proportions of the five classes of paramecia portant role in protozoan community structure. in a mixed population of killers and sensitives This report describes the current revision of the model used during the field and laboratory studies of Class a b X Y I LANDIS(1981, 1986) on the role and evolution of the ~ ~~ KK, KK,no Kk, Kk,no M,no killer trait in natural populations of paramecia. As the Genotype kappa kappa kappa kappa kappa

~ fieldwork began in the mid 1970s, earlier versions of Proportion of each class A B X Y z this model posed certain important questions as to in the population at mating-type frequency, the effectiveness of the killer generation n trait at low densities of paramecia, and the roles of Relative number of each A’ B’ X’ Y’ Z’ conjugation and autogamy in the evolutionary strat- class in the population egy of paramecium. The importance of the model, as after natural selection, with most models of this genure, is its interaction and infection and loss have occurred influence upon the formation of theories and the precise design of experiments to test the predictions If no sexual process occurs, then the proportion of the classes in the next generation after natural selection, in- of the theory. fection and loss have taken place will be represented by

Af Bf Xf Yf Zf THE MODEL If conjugation takes place, the proportions of A,, Bf, etc., will be affected by recombination and cytoplasmic ex- Within a population of killers and sensitives of the change. The resulting proportions are same species there are five possible classes of individ- A, B, X, Y, Z, uals in regard to genotype and cytoplasmic characters, Since not all the members of the population will undergo ignoring the SI and S2 loci. These classes, along with each of the above processes during the same generation, most of the symbols used in this paper, are listed in the fraction that undergoes only a fission will be 7,conju- Table 1. Class a is homozygous K and contains kappa. gation, /3, and for autogamy, a. When properly manipu- It is killer. Class b is like class a, but lacks kappa and lated, the resulting proportions of the five classes can be is therefore sensitive. Classes x and y are like a and b, calculated. The proportions will be in generation n+ 1: respectively, but are heterozygous. Class z is KK, lacks An+, &+I &+I Y~+I Zn+1 kappa and is sensitive. The proportions within population of the five classes at generation, n, are desig- nated A, B, X, Y and Z. As explained below, these at each locus on the order of 1 or 2% (LEWONTIN proportions will be determined by the relative Dar- 1974). Therefore, the intents of this model are satis- winian fitnesses of sensitives and killers, the rate of fied by letting Ws remain constant within a range of spontaneous loss of kappa from the cytoplasm (L),the 0.95 to 1.0 times that of Wk. rate of infection of the endosymbiont from ingestion The spontaneous loss of kappa, occurring at rate L (C), cytoplasmic exchange (R)and the relative pro- each generation, will reduce the proportions of the portions of the nuclear genotypes produced at conju- kappa-bearing classes, adding individuals to the sen- gation and autogamy. sitive classes. Loss often occurs in the laboratory and The Darwinian relative fitness component, w,ex- may be due to a number of causes. The host outgrow- presses the effect of natural selection upon the various ing the endosymbiont, certain temperature regimes, phenotypes. The assumptions and validity of various antibiotics and other causes are known to eliminate selection models have been extensively discussed kappa from genetically competent hosts. (DOBZHANSKY1970; CROWand KIMURA 1970; WAL- Infection may occur in two ways. First, infection LACE 1968) and will not be mentioned here. Assuming may result from the ingestion of the infectious kappa that the only factors modifying the relative fitnesses particles by a competent host. However, in the labo- of the five classes are the effects of the presence or ratory such infection has been difficult to accomplish, absence of kappa, the sensitive classes can be assigned requiring a high density of kappa and competent the fitness W,, the killer W,. The relative fitness of sensitives in the presence of Ca+ and Mg’ ions (MUEL- both groups is frequency- and density-dependent, with LER 1963, 1964). In view of the conditions required, the fitness of the sensitive classes approaching zero the rate of infection due to ingestion, C, is thought to under conditions usually found in the laboratory. be low in nature. However, populations in nature seem very unlikely to The other parameter of infection is the rate of produce the enormous densities used in routine labo- infection (transmission) due to cytoplasmic exchange, ratory killing tests. Furthermore, very large differ- R. Cytoplasmic exchange may occur when a killer ences in fitnesses in nature are extremely rare, varying mates with a sensitive that is genetically competent Killer Paramecia 199 following the conjugation. R is a function of the rate TABLE 2 of cytoplasmic exchange and the probability of an Equations describing the inheritance of the killer trait in infectious particle invading and multiplying within a populations of paramecia competent host. This component is more fully discussed in the section on conjugation. Equations for fission Assuming, for simplicity, that the loss and infection A/ A’IT of kappa through the medium occurs only after selec- B/ B’IT tion has operated, these terms (wk, w,, c and L) X, = X’/T combined result in the following relative numbers Y/ Y‘/T after one vegetive cell generation: Zf = Z’/T Equations for conjugation A’ = AWk + BW,C - AWkL (1) ~~ ~~ ~~ A, = (A’2+ A’X’ + A’B’ + 0.5A’Y’ + 0.25X’2 + 0.25X’Y’ B’ = Bw, - Sw,c + AWkL (2) 0.5X’B’ + R(A’B’ + 0.25X’Y‘ + 0.5A’Y’ + 0.5X‘B’))IT B, = (B” + 0.25Y’2+ 0.5B’X’ + B’Y’ + 0.25X’Y’ + 0.5A’Y’ + x‘ = xwk + YW,c - XWkL (3) A’B’ - R(0.5B’X’ + 0.5A’Y’ + A‘B’ + 0.25X’Y’))/T Y’ = YW, - YW,c + XWkL (4) X, = (0.5X’2+ A’X‘ + 0.5A’Y’ + A’Z’ + 0.5B’X’ + 0.5X’Y’ + 0.5X‘Z’ + R(0.5A’Y’ + 0.5X’Y’ + A‘Z’ + 0.5X’Y’))/T Z’ = zw, (5) Y, = (0.5Y” + 0.5A’Y‘ + A‘Z‘ + 0.5B’X’ + B’Y’ + 2B’Z‘ + 0.5X’Z’ + Y’Z’ - R(0.5A’Y’ + A’Z’ + 0.5X’Y’ + O.SX‘Z‘))/T Each number is proportional to the absolute number of individuals of each classjust before the next fission. Z, = (2“ + X’Z’ + Y’Z’ f 0.25X” + 0.5X’Y’ + 0.25Y’2)/T The total is T = A’ + B’ + X’ + Y’ + Z’. After a Equations for autogamy fission without any sexual process taking place, the A, = (A’ + 0.5X‘)/T proportion of the a class will be Bo = (B‘ + 0.5Y’)IT Z, = (2’ + 0.5X‘ + 0.5Y’VT Af= A’/T. (6) Combination of equations for autogamy, conjugation and fission Similar equations give the proportions of the remaining classes (Table 2). Conjugation and autogamy, however, will alter the genome of the organism, rendering it competent for infection or producing the kk genotype- In order to predict the proportions of the five classes, it is necessary to calculate the proportions of Proportion of the KK killer = A‘ + 0.25X’ the possible matings in the population. Assuming ran- (7) AB 0.25 XY 0.5AY 0.5XB. dom mating, the proportions of the various possibili- + + + + ties will be (A + B + X + Y + Z)’ or A’ + B‘ + X2 + In order to incorporate the numerical changes of the Y2 + Z2 + 2AB + 2AX + 2AY + 2AZ + 2BX + 2BY + class ratios measured by Wk, W,, C and L, the expres- 282 + 2XY + 2x2 + 2YZ = 1, where A’ corresponds sions A’, B’, etc., may be manipulated exactly as the to the probability of a mating of a KK killer with a KK proportions above. However, the rate of transmission killer, 2AB is the proportion of the matings that are due to cytoplasmic exchange, R, must be incorporated KK killer with a KK sensitive, and so on. In order to into the expression representing the proportions of calculate the proportions of the five classes in the next each class in the next generation due to conjugation. generation following the conjugation of each member, R is more formally defined by letting K equal the the frequencies of the different kinds of organisms proportion of observed killers originating from a cross arising from each mating are sorted after one fission and J equal the expected proportion of killers origi- and then added. For example, the only matings that nating without cytoplasmic exchange, then may result in a KK killer clone are those in which at least one member is a or x. Not all of the exconjugants, however, will reside in the KK killer class. The x by x mating will produce the appropriate combination of In each mating of a killer with a sensitive, this factor cytoplasm and genes only one-quarter of the time must be taken into account. Since transfer of cyto- (Figure la). Similarly, the frequency of the a class plasm will not decrease the proportion of killers orig- from the a by y mating will only be 0.25 (2AY) or inating from a cross, K cannot be less than J. Although 0.5AY (Figure lb). Completing the process, the pro- each possible class combination may have a different portion of the KK killer class in the population after R value, for simplicity R is taken as a constant in the conjugation and a fission will be following computations. In other words, it is assumed 200 W. G. Landis maining classes in the next generation, if only conju- Kk. KILLER Kk, KILLER gation takes place, are similarly derived (Table 2). Autogamy renders each individual paramecium homozygous and can be regarded as a severe form of inbreeding (SONNEBORN1957). Assuming that the entire population passed through autogamy during the interfission period, all heterozygotes would disappear. In the case of x or Kk killer paramecia, the fission following an autogamy will result in one-half U. H Kk, KILLER of the progeny being KK killers, with the remaining being kk sensitives (Figure 1b). Using class a again, the proportion of this class if only autogamy took place would be Kk, SENSITIVE A, = (A’ + 0.5X‘)/T, (1 1) and so on for the remaining classes (Table 2). In the above derivations of the class proportions, each process, whether it be autogamy, conjugation or fission, was considered separately. To combine these % KILLER % KK, SENSITIVE KK. separate events, it is assumed that a proportion of the population undergoes each of these processes and that b. % Kk, KILLER % Kk. SENSITIVE these proportions remain constant. The fraction of the population that undergoes autogamy = a, conjugation = ,6 and fission = y is such that Kk, KILLER a+P+y=l. (12) In order to calculate the resultant proportion of the a class due to all three processes, A,+,, the terms above combined to give A,+1 = Af + A, + A,. (13) This process was continued to determine all the phenotypic and genotypic class frequencies (Table 2).

THE COMPUTATIONS C. x KK, KILLER H kk, SENSITIVE FIGURE1 .-Expected combinations of genes and cytoplasm The above derivations help to identify the factors from typical sexual events. The first cross (a) involves two hetero- that theoretically can affect the ratio of killers to zygotes that are both killer. Note that one-quarter of the offspring are rendered sensitive. The second cross (b) is between a homozy- sensitives in populations at equilibrium (Table 3). It is gous KK killer and Kk sensitive. In the last example (c) a heterozy- obvious that natural selection could have a major gous killer undergoes autogamy. effect on the ratio. The spontaneous loss of kappa from killers or the infection of sensitives could also that the frequency of cytoplasmic exchange is constant affect the ratio. An increase in the proportion of for all crosses. In the example dealing with the KK killers by the transfer of kappa during cytoplasmic killer class, the number of killers originating because exchange is also important. Finally, it should be noted of cytoplasmic exchange is that kappa is eliminated from heterozygous killers (Kk) if they become kk. Therefore, higher frequencies of k = R(AB + 0.5AY + 0.25XY + 0.5XB); (9) alleles tend to decrease the proportion of killers, but therefore, the complete expression for the proportion the effect is influenced by the frequency of conjuga- of a in the next generation due to the process of tion and autogamy. The magnitude of these effects conjugation, including all the relevant parameters, is and the way in which they interact have been esti- mated by using the formulas derived in the preceding Ac = (A’2 A’X’ A‘B’ 0.5A’Y‘ 0.25X“ + + + + section. + 0.25X’Y’ + 0.5X’B’ + R(A’B’ + 0.5A’Y’ (IO) The equations derived in the previous section were transformed into a FORTRAN IV program. Varia- 0.25X’Y‘ 0.5X’B’))/T. + + bles could be changed individually. Gene frequencies The expressions for the frequencies of the four re- and initial genotype ratios were specified for the first Killer Paramecia 20 1

TABLE 3 Theoretical factors affecting the equilibrium of the killer trait

Decrease in the proportion of killers

Factor Mode of action Loss of kappa Loss of kappa from genetically competent organisms brought about by ad- verse temperatures or harmful chemi- cals 0.1 - BASELINE Presence of alleles k alleles lower the frequency of compe- B = 0.05 k I I I I I tent hosts in the population and cause 100 200 300 400 500 600 700 800 loss of kappa when Kk killers pass GENERATIONS through autogamy or conjugation and (FISSIONS) become kk FIGURE2.-The effect of different rates of conjugation on the proportion of the killer phenotype in the population. The baseline Natural selection Selection operating against the symbiont, for example, by reducing fission rate rate of conjugation, /3 = 0.05 per generation, results in a sharp of symbiont bearing paramecia decrease in the proportion of killers. Rates of conjugation equal to 0.20 and 0.30 produce equilibria. Increase in the proportion of killers puted after successive fissions, a steady decline from Infection or sensitives incorporate kappa KK Kk 50% to no killers was observed (Figure 2, baseline). Cytoplasmic ex- Kappa transmitted at conjugation via cy- The effect of conjugation on the proportion of the change (conjuga- toplasmic exchange into a sensitive par- tion) amecium with a KK or Kk exconjugant killer trait was then investigated. The results can be seen in Figure 2. With a rate of conjugation of 0.05 Natural selection Selection operates against the sensitive phenotype per generation the killer phenotype becomes extinct, as pointed out above. After 500 fissions, the initial proportion of 0.50 dropped to 0.05. When the rate generation, but were subsequently allowed to vary. of conjugation was increased to 0.20 per generation The expected proportions for each generation were the proportion of killers approached an equilibrium calculated by iteration; the program was simply in- of 0.35. A further increase in the rate of conjugation structed to pass through the computations a specific to = 0.30 resulted in an equilibrium proportion of number of times. The frequency of each class as well killers of 0.46. as the total number of killers and sensitives was com- The mechanism through which conjugation acts to puted for each generation. In this way a recapitulation increase the number of killers in a population is by of the genotypic, phenotypic and allelic frequencies providing an opportunity for transmission by cyto- could be made during the course of the evolution of plasmic exchange to occur. In Figure 3, it is demon- the population. strated that, as the rate of transmission through cyto- The model was kept simple due to our lack of basic plasmic exchange, R, increases, the proportion of knowledge of the population biology and life history killers that can be maintained in the population is strategy of paramecium. Even with this simple model increased. Although a rate of R = 0.10 allows the assuming infinite populations and excluding demog- killers to become extinct, a value of 0.40 produces an raphy, almost no data were originally available by which to judge the values of many of the parameters in natural populations. a Consequently, a set of parameters was chosen to -1W O.? A start the computations. The absolute values of the Y 0.6 2 parameters used to start the computations, however, P 0.5 I- are not very important for they were varied one at a U ni time to ascertain the effects of each one on the pro- Q portions of the various types. E The values used for the initial computations were BASELINE autogamy, a = 0.05, conjugation, 6 = 0.05 and trans- R - 0.1 mission due to cytoplasmic exchange, R = 0.10. The 100 200 300 400 500 600 700 800 rate of infection, I, probably rare in nature, was given GENERATIONS a value of zero. The rate of loss, C, was set at 0.01. (FISSIONS) FIGURE3.-The effect of varying rates of transmission due to The fitness of the killers, W,, and sensitives, W,, was cytoplasmic exchange, R, on the proportion of the killer trait. Even Wk = W, = 1. When these values were put into the though the rate of conjugation remains at 0.05 per generation, equations and the proportions of killers were com- 42% of the population shows the killer phenotype if R = 0.40. 202 W. G. Landis

5 0.7 =! Y 0.6 z -0 0.5 I- e= 0.4 2 0.3 Q 0.2 0.1

a. GENERATIONS GENERATIONS (FISSIONS) (FISSIONS)

GENERATIONS IFISSIONS) FIGURE4.-The effect of the initial proportion of the z or kk sensitive class on the proportion of killers in a population, (a) An FIGURE5.-The effect of autogamy on the proportion of the killer trait. A dramatic increase in the rate of autogamy (a) has only increase in the z class increases the rate at which kappa is lost from a slight effect on the rate of decrease of the killer trait in the the population; (b) demonstrates the effect of the z or kk sensitive population. When an equilibrium is possible (b), an increase in the class on the proportion of killers in a population when an equilibrium is possible (the rate of conjugation has been set at 0.30). Values rate of antogamy depresses the equilibria of killers in a population. of z = 0.20 and z = 0.30 cause the killer trait to rapidly disappear from the population. 0.30). An increase in the rate of autogamy was found to depress the equilibrium proportion of killers attained by the population (Figure 5b). Although rate apparent equilibrium of 0.46. The above information a of 0.05 per generation resulted in an equilibrium indicates that, at sufficiently high rates of conjugation of and cytoplasmic exchange, the killer trait may be 0.46, a rate of 0.20 per generation depressed this maintained within a population in the absence of equilibrium value to 0.32. natural selection. The factor with the most dramatic effect is natural The magnitude of the effect of different frequen- selection. If the fitnesses of the killer and sensitive cies of the k allele within populations of paramecia phenotypes are equal, as they are under the initial was examined. Starting with the initial parameters, an conditions, the killer population becomes extinct (Fig- increase in the frequency of the k allele caused an ure 6). However, if the fitness of the sensitive phenotype, W,, is only l% less than that of the killer, an increase in the rate of extinction of the killer trait (Figure 4a). In order to study the effect when an equilibrium is reached that maintains 31% of the equilibrium is possible, the rate of conjugation was population as killer. A 2% reduction in the fitness of increased to = 0.30 per generation. With the propor- the sensitive phenotype produces a proportion of kill- tion of the z, kk sensitive, class set at 0.10, an equilib- ers equal to 0.60 that of the entire population. A rium of 0.46 was possible (Figure 4b). With Z = 0.20 selection differential of only 1 or 2% appears to be an equilibrium of 0.32 was attained. When Z = 0.30 capable of maintaining the killer phenotype in a pop- the killer trait became extinct. These results demon- ulation at relatively high levels. strate the importance of the frequency of the k. allele on the proportion of the killer trait in populations. DISCUSSION An increase in the rate of the autogamy, a, slightly The computer simulations have shown that a high increased the rate of extinction of the killer trait enough rate of conjugation with cytoplasmic exchange (Figure 5a). The rate of conjugation was again in- can theoretically maintain the killer trait within a creased so that an equilibrium would be possible (P = population. This finding is similar to that found by Killer Paramecia 203 TABLE 4 Rates of increase and relative fitness between killers and sensitives ws = 0.98

(r,h-’Y Density arame- Depression Killer Sensitive W,” ciafml

6g2 and 5g1 0.2 c \ 1 0.032 0.013 0.41 32 BASELINE 2 0.040 0.000 0.00 44 0.1 - Ws = 1.00 0.034 0.000 0.00 32 - I I I I I I I I 3 4 0.021 0.029 1.38 32 GENERATIONS 5 0.024 0.009 0.38 20 FIGURE6.-The effect of natural selection on the frequency of 6 0.034 0.019 0.56 48 the killer trait. If the fitnesses are equal, the killer trait goes to 7 0.036 0.030 0.83 54 extinction. However, small changes in their relative viabilities have 8 0.048 0.009 0.19 54 drastic effects, as only a 2% decrease in the fitness of the sensitive phenotype results in a proportion of killers equal to six of every ten 51k and 51s paramecia in the population. 1 0.028 0.024 0.86 32 2 0.030 0.000 0.00 32 L’HERITIER(1970) for CO2 sensitivity in D. melano- 3 0.026 0.015 0.58 22 gaster. In the case of COnsensitivity, infected flies will 4 0.029 0.021 0.72 34 predominate even at a mild selective disadvantage Table compiled from the data of LANDIS(1 98 1) describing the provided that females always and males occasionally competition experiments between killers and sensitives at densities transmit the sigma factor to their offspring. Natural comparable to those found in collections from the field. Stock 6gn selection can also maintain the killer trait, and small and 5g,, killer and sensitive respectively,were collected from Twin selective differences give quite drastic effects. The Lakes, Bloomington, Indiana. Stocks 51k and 51s are identical, except that 51k contains kappa and is killer. The medium was a relative importance of these factors in maintaining the dilute baked lettuce solution inoculated with Klebsiella pnuemoniae. killer phenotype in actual populations of P. tetraurelia ‘ Intrinsic rate of growth per hour calculated as rm(/a-’)= In/T, will now be considered. when T is the generation (fission) time (FINLAY1977). The host of Caedobacter taeniospiralis, P. tetraurelia, ’ Fitness of the sensitive paramecium. is classified as a severe inbreeder under laboratory conditions (SONNEBORN1957). The immature period trait in populations. It has been demonstrated that P. of the various stocks varies from zero to 25 fissions. tetraurelia killers have a selective advantage over sen- After approximately 25 fissions the members of the sitives at population densities found in collections population undergo autogamy if the food is exhausted taken from natural habitats. and no contact with a member of the opposite mating LANDIS(1981) performed a series of competition type is made. Laboratory conditions suggest that con- experiments using killers and sensitive P. tetraurelia. jugation may occur frequently. However, recent in- Two sets of stocks were used in these assays. Stocks formation that the proportion of the two mating types 5 1s (sensitive) and 5 1 k (killer) are identical except for of P. tetraurelia, even and odd, in nature is very the lack of kappa. Stocks 5g1(sensitive) and 682 (killer) unequal, drastically challenges this conclusion. were collected from Twin Lakes, Bloomington, Indi- The mating type of 25 stocks of P. tetraurelia orig- ana. No F2 mortality was observed in a testcross of inating from Twin Lakes and Skaters Pond, both near 6g2 and 51 indicating that the stocks are closely re- Bloomington, Indiana, was determined. All 20 of the lated (W. G. LANDIS,unpublished results). Densities stocks originating from Skaters Pond proved to be of of the paramecia were controlled to approximate nat- the odd mating type, and the five stocks from Twin ural populations by using dilute baked lettuce medium Lakes also proved to be of the odd type (M. inoculated with Klebsiella pneumoniae. In the experi- SCHNELLERand W. G. LANDIS,unpublished data). ments using Petri dishes, the proportion of killers This has proven to be the general situation in collec- increased from 0.40 to a range of 0.75 to 0.91 after tions of P. tetraurelia made during the past 40 yr. ’72 h. Densities ranged from 9 to 89 paramecium per With the even mating type of P. tetraurelia so rare in milliliter. Similar results were obtained for the exper- nature, the rate of conjugation and transmission via iments conducted in depression slides. However, it is cytoplasmic exchange must also be very low, much possible to calculate growth rates in these experiments lower than the values used in the simulations. Unfor- and thereby the fitness of the sensitives since each tunately, similar data on the proportions of mating organism in the depression was collected and its phe- types in killer P. biaurelia have not been available. notype determined. Table 4 lists the intrinsic rates of Several other lines of evidence support the impor- increase for the killers and sensitives along with the tance of natural selection in the continuation of the fitness of the sensitive phenotype. In all but one in- 204 W. G. Landis stance, the fitness of the sensitive is less than the killer portance of the various factors may vary from case to at densities similar to those found in nature. The case, the model demonstrates the interplay between fitness of the sensitives in many cases is sufficiently genetic, cytoplasmic and environmental factors in the low according to the calculations to ensure the contin- population genetics of cytoplasmically inherited traits. uation of the killer phenotype in the population. The ecologies of killers and sensitive species were I should like to thank JOHN R. PREER,JR. and his laboratory group at lndiana University for their support and comments. also found to be markedly different. Sensitives of P. ELAINETAUBER and GENADAWSON typed the numerous revisions primaurelia, P. triaurelia, P. biaurelia and P. pentau- of this report. This research was supported by Public Service grant relia were often found within the same microhabitat, PHS R01 6M 20038-05 to JOHN R. PREERand a CRDC In-house often within the same 5-ml sample. Conversely, killers Laboratory Independent Research Award to the author. rarely have been found with sensitive species of paramecium (LANDIS198 1). LITERATURE CITED The importance of the killer trait is also demon- BALBINDER,E., 1959 The genotypic control of kappa in Parame- strated by the conservation of the killing mechanism. cium aurelia, syngen 4, stock 51. Genetics 44: 1227-1241. Comparative morphology and DNA-DNA hybridiza- BALBINDER,E., 1961 Intraclonal variation in genotypically deter- tions have provided important clues into the phylog- mined mixed (killer-sensitive) clones of Paramecium aurelia, eny of the bacterial Caedobacteria. DNA-DNA hy- syngen 4. Physiol. 2001. 34: 184-20 1. BEALE,G. H., 1954 The Genetics of Paramecium aurelia. Cam- bridizations of the genomes of C. taeniospiralis and C. bridge University Press, Cambridge, England. varicaedens have demonstrated the divergence of the CROW,J. H. and M. KIMURA,1970 An Introduction to Population two species (QUAKENBUSH1977, 1978). They even Genetic Theory. Harper Row, New York. have different species of Paramecium as hosts. Yet, DOBZHANSKY,T., 1970 Genetics of the Evolutionary Process. Colum- the killing mechanisms of both species are remarkably bia University Press, New York. FINLAY,B. J., 1977 The dependence of reproductive rate on cell conserved. Both species of kappa that kill without cell size and temperature in freshwater ciliated protozoa. Oecologia contact contain R-bodies, refractile ribbons of protein 30: 75-81. that are formed in the nonreproductive subpopulation GILL,D. E., 1972 Intrinsic rates of increase, saturation densities responsible for the killing phenotype. R-bodies are and competitive ability. I. An experiment with Paramecium. Am. Nat. 106: 461-471. implicated in the delivery of the toxin to the cytoplasm GILL,D. E. and N. G. HAIRSTON,1972 The dynamics of a natural of the sensitive (PREER,PREER and JURAND 1974). population of Paramecium and the role of interspecific com- The R-body, toxin and phage subunits are antigeni- petition in a community structure. J. Anim. Ecol. 41: 81-96. cally related across species lines (SINGLER-BASTIAANS LANDIS,W. G., 1981 The ecology, role of the killer trait, and interactions of five species of the Paramecium aurelia complex 1975). These similarities are in spite of the divergence inhabiting the littoral zone. Can. J. 2001. 59: 1734-1 743. demonstrated by restriction mapping for the genes LANDIS,W. G., 1987 The interplay among ecology, breeding located on the viral and extranuclear covalently closed system, and genetics in the Paramecium aurelia and Paramecium circular DNA that codes for the killing mechanism. bursaria complexes. In: Progress in Protistology, Edited by J. 0. Natural selection has apparently acted to maintain CORLISSand P. J. PATTERSON.Biopress Limited, Bristol. In press. intact the essentials of the killing mechanism. LEWONTIN,R. C., 1974 The Genetic Basis of Evolutionary Change. The field studies, laboratory competition assays, the Columbia University Press, New York. computer model and the comparative morphology of L'HERITIER,P., 1970 Drosophila viruses and their roles as evolu- the killing mechanism all support the hypothesis that tionary factors. Evol. Biol. 4: 185-209. the killer trait is maintained by natural selection and MUELLER,J. A., 1963 Separation of kappa particles with infective activity from those with killing activity and identification of the is a vital component in the life-history strategy of infective particles in Paramecium aurelia. Exp. Cell Res. 30 certain species. A great deal of research remains to 492-508. further refine this rather simple model and incorpo- MUELLER,J. A., 1964 Cofactor for infection by kappa in Parame- rate the interplay of ecology, breeding system and cium aurelia. Exp. Cell Res. 35: 464-476. PREER,J. R., JR., L. B. PREERand A. JURAND, 1974 Kappa and genetics into a meaningful construct. Obviously, the other endosymbionts in Paramecium aurelia. J. Cell Sci. 11: need for more elegant field research is crucial, but 581-600. that is dependent on a rapid means of identifying the QUAKENBUSH,R. L., 1977 Phylogenetic relationships of bacterial sibling species of P. aurelia. A complete understanding endosymbionts of Paramecium aurelia: polynucleotide sequence of the role of the killer trait and its influence on the relationships of 51 kappa and its mutants. J. Bacteriol. 129 895-900. evolution of the life-history strategies and genetic QUAKENBUSH,R. L., 1978 Genetic relationships among bacterial systems of Paramecium is still beyond the horizon. endosymbionts of Paramecium aurelia: deoxyribonucleotide se- The approach and findings of this model of the quence relationships among members of Caedobacter. J. Gen. population genetics of the killer trait also have a wider Microbiol. 108: 181-187. SINGLER-BASTIAANS,M. J., 1975 Studies with killer paramecia. applicability. The algorithm developed here should pp. 9 1-97. Ph.D. Thesis, Indiana University, Bloomington, be easy to transfer to other studies of the population Indiana. genetics of cytoplasmic inheritance. Although the im- SONNEBORN,T. M., 1943 Gene and cytoplasm. I. The determi- Killer Paramecia 205

nation and inheritance of the killer character in variety 4 of SONNEBORN,T. M., 1975 The Paramecium aurelia complex of 14 Paramecium aurelia. Proc. Natl. Acad. Sci. USA 29 329-338. sibling species. Trans. Am. Microsc. Soc. 94 155-178. SONNEBORN,T. M., 1957 Breeding systems, reproductive meth- YEN,J. H., and A. R. BARR,1973 The etiological agent of cyto- ods, and species problems in protozoa. pp. 155-324. In: The plasmic incompatibility in Culex pifliens. J. Invertebr. Pathol. Species Problem, Edited by ERNSTMAYER. American Association 22242-250. for the Advancement of Science, Washington, D.C. WALLACE,B., 1968 Topics in Population Genetics. W. W. Norton, SONNEBORN,T. M., 1974 Paramecium aurelia. pp. 469-594. In: New York. Handbook of Genetics, Vol 2, Chap. 20, Edited by R. C. King. Plenum Press, New York. Communicating editor: S. L. ALLEN