J. Cell Set. 43, 59-74 (1980) Printed in Great Britain © Company of Biologists Limited igSo

TEMPERATURE-SENSITIVE PERIODS OF AFFECTING CELL DIVISION IN THERMOPHILA JOSEPH FRANKEL,* JYM MOHLERf AND ANNE KOOPMANS FRANKEL Department of Zoology, University of Iowa, Iowa City, Iowa 52242, U.S.A.

SUMMARY Temperature-sensitive periods were determined by application of temperature shifts and shocks to 3 temperature-sensitive cell division arrest (cda) mutants of Tetrahymena thermophila. A restrictive temperature, 36 °C, was found at which all 3 mutants are fully penetrant, yet other physiological effects are minimal. At this temperature, the temperature-sensitive period of cdaCl is a unique 5-min period in mid-division, that of cdaAi is a similarly brief period situated about 0-5 h prior to cell division, while the temperature-sensitive period of cdaHi is 20 to 30 min long and immediately precedes cell division. These periods either coincide with (cdaCi, cdaHi) or immediately precede (cdaAi) the onset of phenotypic abnormality at the restrictive temperature. Brief exposure to 36 °C during the temperature-sensitive period in any of these mutants brings about irreversible arrest of division furrows in progress or preparation. Mutant cells suffering such arrest can, however, divide again at a permissive temperature by forming new furrows at different sites.

INTRODUCTION One major advantage of studying temperature-sensitive mutations affecting developmental processes is that such mutations invite the use of temperature treat- ments to find out the time at which products exert their phenotypic effects. Changes in temperature are used to demarcate a temperature-sensitive period (TSP) by 2 primary operations: shifts from restrictive to permissive temperature to delineate its beginning, and shifts from permissive to restrictive temperature to identify its end (Esposito, Esposito, Arnaud & Halvorson, 1970; Suzuki, 1970). A double-shift from permissive to restrictive temperature and then back again (i.e. a shock) may also be used to pinpoint TSPs (Poodry, Hall & Suzuki, 1973). Various combinations of these operations have been applied to characterize TSPs for expression of lethality, male sterility, and organ-specific defects in Drosophila (A. Frankel, 1973; Martin, Martin & Shearn, 1977; Shellenbarger & Mohler, 1978) as well as TSPs of mutants affecting starvation-mediated developmental sequences in the microorganisms Bacillus subtilis (Szulmajster, Bonamy & Laporte, 1970; Leighton, 1974; Young, • Author to whom reprint request should be sent. t Present address: Room 16-720, Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, U.S.A. 5-2 60 J. Frankel, J. Mohler and A. K. Frankel 1976; Sumida-Yasumoto & Doi, 1977) and (Esposito et al. 1970). With one recent exception (Melero, 1979), studies on the effects of temperature- sensitive (TS) mutations on the eukaryotic have employed only shifts from permissive to restrictive temperature, and used these to characterize an 'execution point', defined as '.. .the time in the cell division cycle when the temperature- sensitive gene product completes its function at the permissive temperature' (Hart- well, 1971a) [also termed 'block point' (Howell & Naliboff, 1973); 'transition point' (Nurse, Thuriaux & Nasmyth, 1976) and 'shift-up point' (Ashihara, Chang & Baserga, 1978)].* This defines the end of the TSP. Execution points have been determined for a large number of TS cell division cycle mutants in Saccharomyces (Hartwell, Mortimer, Culotti & Culotti, 1973), Schizosaccharomyces (Nurse et al. 1976), (Orr & Rosenberger, 1976), and Chlamydomonas (Howell & Naliboff, 1973) and also for G1-arrest mutations in mammalian cell lines (Ashihara et al. 1978; Melero, 1979). TS mutations at 7 loci that bring about arrest at specific stages of cell division in the ciliate Tetrahymena thermophila (formerly T. pyriformis syngen 1) have been characterized genetically (Frankel, Jenkins, Doerder & Nelsen, 19766; Jenkins, un- published) and phenotypically (Frankel, Jenkins & DeBault, 1976 a; Frankel, Nelsen & Jenkins, 1977; Cleffrnann & Frankel, 1978; Frankel, unpublished). At 3 of these loci, cdaA, cdaC and cdaH, mutant alleles are available that bring about 100% arrest within the first cycle after the shift to restrictive temperature, and are therefore well suited for analysis of TSPs. The present communication is an analysis of the TSPs of mutations at these 3 loci, and also includes a demonstration of the local irreversibility of effects incurred during these TSPs.

MATERIALS AND METHODS Stocks and media All cells used in this study were of inbred strain B of Tetrahymena thermophila. The strain B was obtained from. Dr D. L. Nanney in 1972, when in its 18th generation of inbreeding. It was inbred a 19th time in our laboratory in 1975, and designated B-1975. A single wild type stock, of mating type II, was used. Four mutant stocks, homozygous for each of 3 TS cda (cell division arrest) , were employed. cdaAl (formerly moi") and cdaCz (formerly motf) were each represented by a single homozygous stock, IA-105 (mating type III) for the former and IA-123 (m.t. VII) for the latter. Two stocks homozygous for cdaHl were used: IA-150 (m.t. V) for analysis of silver-impregnated slides and furrowing times and IA-149 (m.t. II) for the remainder of the single-cell work. We will refer to stocks by the mutant alleles that they carry as homozygotes. The culture medium generally utilized contained 1 % proteose peptone plus o-i % Difco Bacto yeast extract (1 % PPY).

• We will continue to use 'execution point' as a synonym for the end of the TSP. Of the alternative expressions, 'block point' invites confusion between the end of the TSP and the time of phenotypic expression, while 'transition point' creates a potential for confusion between the ends of gene-specific TSPs and the more general 'physiological transition point' (c.f. Frankel, Mohler & Frankel, 1980). Temperature-sensitive periods of Tetrahymena mutations 61

Single-cell procedures Fifty- to 150-ml batches of medium in 250-ml conical flasks or 500-ml Fernbach flasks were inoculated at roughly 10 cells/ml from a 2-day-old tube culture and preincubated at 25 °C without shaking or aeration for about 20 h, to attain a density of about 1000 cells per ml when experiments were begun. A small portion of this culture was poured out into a glass Petri plate, for selection of single cells. In most experiments, the drop-culture procedure of Lovlie (1963) was used. Single non-dividing cells were selected at random and deposited in small drops of culture medium on the surface of plastic Petri dishes (Falcon no. 1008 or 3005), 25 drops to a dish. The drops were covered with a thin layer of light paraffin oil (viscosity 125/135) and maintained at 25 ± 1 °C. These drop cultures were examined at io-min intervals under a dissecting microscope and the state of cells with respect to cell division was noted. Cell divisions completed in each interval were recorded. After all cells in a newly established culture dish had divided, the dish, containing individually cultured pairs of sister cells of known interfission ages, was transferred to a waterbath set at a temperature ranging from 36'0 to 400 °C, according to the experiment. The temperatures in the drops were recorded using a thermistor probe connected to a tele-thermometer (Yellow Springs, model 47). The equilibrium temperature in the drop-cultures, about 1 -o °C below the bath temperature, was reached within 5 min after transfer. In the shift-up experiments, the culture dishes were re- moved from the waterbath after 3 h at the high temperature and scored as to whether or not a second division had taken place. In the shock experiments, the culture dishes were kept at the restrictive temperature for only 20 min, and then returned to 25 CC. They were observed under the dissecting microscope immediately after removal from the high temperature bath, and at io-min intervals thereafter. Calculation of the generation time of each cell was based on the total time elapsed between the completion of one division and that of the next, accurate to within ± 5 min. A variant of the above procedure was employed in shift-down experiments. Dividing cells were selected and placed in individual micro-drops. The exact time of division of these cells was recorded, and the culture dish was transferred to the waterbath, set 1 °C above the desired restrictive temperature, immediately after the division of the last cell (and within 10 min of the division of the first). The cells were kept at the high temperature for 70 min or more, and then removed from the bath and observed at frequent intervals at 25 °C to find out whether or not the subsequent division was successfully completed. In most experiments, one dish with mutant cells was maintained continuously at 25 CC, as a control, while other dishes of the same culture were given concurrent shocks or shifts. In certain experiments, wild-type cells were studied simultaneously with mutants in a 2x2 experimental design (wild type and mutant, each at 25 and 36 °C). In the culture-dish experiments, all cells were observed during the 2-3 min preceding the shift to high temperature, and, where possible, were classified according to visible stages related to cell division. The process of division furrowing (cytokinesis) was arbitrarily categorized into 5 stages, and one stage just prior to cytokinesis (B) could also be recognized. The B stage is characterized by a semi-rectangular (or 'boxy') shape, the E stage by a 'notch' on one side, the MB stage by a complete furrow less than one-third of the cell diameter deep, the M stage by a furrow one- to two-thirds of the cell diameter deep, the ML stage by a furrow more than two-thirds of the cell diameter deep but with a distinct waist still present, and the L stage by a figure-eight appearance with a waist of no distinguishable thickness as observed at 40 x magnification. The duration of each of these 6 stages was measured for 10 to 12 cells of each of the 4 genotypes by observations at i-min intervals made throughout the division process at 26 CC. The appearance of these stages and their timing are shown in Fig. 1. To allow a more precise assessment of the division stage at which cells are first exposed to the restrictive temperature, individual non-dividing cells were drawn into glass capillary pipettes (Prescott, 1957; Williams, 1964) and allowed to proceed through the cell cycle at 25 °C until a cleavage furrow was established. The furrowing stage was recorded, and the micropipette was then transferred to a waterbath set at 36-0 CC. Due to the narrow walled construction of the micropipette, we assume that the cell attained the bath temperature within a few seconds after transfer of the pipette. At o-5-h intervals afterwards, pipettes were temporarily removed from the high temperature bath and cells rapidly scored for completion of the furrowing process. 62 J. Frankel, J. Mohler and A. K. Franhel In all cases the temperatures referred to in the Results are those to which the cells them- selves had been exposed. Cytological analysis of division arrested cells Cells of mutant strains were grown in a mass culture that was shifted to restrictive temperature as described in the accompanying paper (Frankel et al. 1980). Samples were removed at 30-min intervals and prepared for silver impregnation, performed following the directions of Frankel & Heckmann (1968) with slight modifications reported elsewhere (Nelsen & DeBault, 1978). Two hundred to 300 cells were scored in each sample, according to a standard system of staging oral development (Frankel, 1962; Frankel & Williams, 1973). Data analysis Statistical procedures are those given by Sokal & Rohlf (1969). Standard parametric methods were employed in analyses of data obtained from wild type, cdaAi, and cdaCt. Non- parametric procedures based on ranking (Sokal & Rohlf, 1969, chapter 13) were used in analyses of data obtained from the highly variable cdaHi.

RESULTS Growth characteristics of mutant stocks at permissive temperatures Generation times of individual cells cultured at 25 °C in microdrops established from very early exponential phase cultures were substantially less than generation times observed in mass cultures at high densities (cf. Frankel et al. 1980). Generation times of wild-type, cdaAi and cdaC2 cells were similar, whereas cdaHi cells had a longer generation time (Table 1). In each stock, the distribution of generation times had 2 disjunct components: the majority of generation times belonged to a normal distribution, whereas a minority of generation times, that were longer than 140 min, belonged to a second distribution with a much greater variance, thus sub- stantially inflating the standard deviation estimate for the entire population. Except occasionally in cdaHi, these tardy cells did not appear unhealthy. The proportion of such tardy cells was fortunately low in the cdaC2 and cdaAi mutant stocks, surprisingly higher in wild type, and very high in the cdaHi mutant stock (Table 1). Owing to these differences among stocks in distribution of generation times, the generation time medians (reflecting the normally distributed majority class of cells only) are more similar to each other than are the means (Table 1). The means of generation times were also significantly inhomogeneous among experiments in all stocks except cdaC2. Since these inhomogeneities cause problems in interpreting the timing of temperature shifts, heat-treated cells were always compared to simultaneous controls derived from the same culture flask. Also, execution points, defined as the time when 50 % of the cells become able to divide following the shift to restrictive temperature, were compared to the interpolated median generation time (Table r, footnote%) of parallel controls. The latter provides the best estimate of the time when 50 % of the cells have divided.

Choice of a standard restrictive temperature T. thermophila can grow exponentially at temperatures up to 40 °C, and the mutants used in this study were originally selected at 39-5 °C. However, at 37-5 °C Temperature-sensitive periods of Tetrahymena mutations 63

Table 1. Statistics of single cells maintained in mtcrodrops at °C Genotype

Statistic Wild type cdaAi cdaCz cdaHx Grand mean* in) 128-3 (180) 122-3 (37°) H7'6 (176) I54'O (98) Standard deviation* ± 36-2 ± 282 ± 22-4 ± 62-6 No. of experiments! 9 [I] 19 [I] 9 [H] 5 [I] % in Normal distribution 79 91 95 64 Interpolated mediant 118 118 115 131 • Expressed in units of min. f H= means homogeneous; I = means inhomogeneous. X Based on linear interpolation, accomplished graphically, between the ends of the class intervals bracketing the 50% point on the cumulative distribution function. This is equivalent to the abscissa value corresponding to the 50 % ordinate value in cumulative distributions of cell division such as are shown in Figs. 2 and 3 (dashed Line). and above the effects of excess delay and of nonspecific division arrest may complicate the analysis of effects specific for mutant genes [see the accompanying paper, Frankel et al. (1980)]. We therefore analysed penetrance at different restrictive temperatures. Virtually all cdaAi and cdaHi cells underwent division arrest at the first division following a shift to 35 or 36 °C. About 65% of cdaC2 cells became arrested in div- ision following a shift to 35 °C, but over 90% underwent arrest after being shifted to 36 °C* Hence 36 °C was chosen as the standard restrictive temperature for analysis of the temperature sensitivity of mutant alleles.

Temperature-sensitive periods of mutants, analysed at 36 °C cdaC2. The temperature-sensitive period (TSP) of cdaC2 was characterized precisely by a combined application of temperature shift-up, shift-down and double shift (shock). Shift-up experiments carried out in micropipettes (Fig. 1, panel B) indicated that cdaC2 cells failed to divide if shifted from 25 to 36 °C at the M stage of division or earlier, and became capable of division if the shift was accomplished in the ML stage or later. The end of the TSP, or execution point, is thus near the end of the cell division process. The beginning of the TSP was assayed in shift-down experiments in which cdaC2 cells were transferred from 25 to 36 °C at the beginning of the cell cycle and returned to 25 °C near the time when the next division was scheduled to begin. Most cells underwent division arrest if they had attained the ME stage prior to the shift from 36 to 25 °C, and all cells became susceptible by the M stage (Fig. IB). The TSP of cdaC2 is about 5 min, beginning at the ME stage and ending at the M.L stage. • High penetrance at 36 CC was in fact the basis for choosing these particular alleles for analysis. Other alleles are available at all three loci (Frankel et al. 1976ft, 1977; Jenkins & Frankel, unpublished), but all except the ones under study here manifest incomplete pene- trance of division blockage at 36 °C (e.g. see Frankel et al. 19766, table 5, in which cdaA and cdaC are designated as mox and mo3 respectively). cdaA\ (moid of Frankel et al. 1977) under- goes almost total division blockage at 36 CC, but comparison at 35 °C revealed it to be a 'weaker' allele than cdaAi (Frankel & Jenkins, unpublished observations). 64 J. Frankel, J. Mohler and A. K.

66 J. Frankel, J. Mohler and A. K. Frankel

(14) 100

80

2'> ^

c SB 40 a.

20

- A

25 45 65 85 105 125 145 165 185 Time after division, min Fig. 2. Percentage of cdaAi cells dividing following a temperature shift-up (A) or shock (O) as a function of time in the cell cycle at the initiation of exposure to the high temperature, compared to the cumulative division curve of control cells maintained continuously at 25 °C (•)• The abscissa is plotted according to the interfission age, while the ordinate represents the percentage of cells that completed division. The number of cells at each of the points involving exposure to high temperature is given in parentheses. A total of 9 reasonably homogeneous experiments are represented. temperature sensitivity to the execution point and the relative brevity of this period suggests that the true TSP of cdaAi is short. Those cdaAi cells that subsequently became arrested in division were generally in the B or E stages just after the end of the 36 °C shock. This suggests that the TSP does not precede the onset of cytokinesis by more than about 20 min of development at 36 °C [the equivalent developmental time at 25 °C is about 25-30 min (Cleffmann & Frankel, 1978)]. cdaHi. The results of shift-up experiments analysed according to stage indicate that cdaHi cells complete their temperature-sensitive function near the time of the beginning of division furrowing, the E stage (Fig. ic). When analysed according to interdivision age (Fig. 3), the execution point was estimated at 103 min after division, or 079 of the median generation time. This estimate is in fair agreement with an independent estimate of 111 min, or 0-85 of the median generation time, which was made by assuming (i) that execution at 36 °C occurs at the beginning of the E stage (Fig. 1 c); (ii) the beginning of the E stage is 20 min prior to completion of division in cdaHi (Fig. 1), and (iii) the median generation time is 131 min (Table 1, Fig- 3)- The results of experiments with 20-min shocks at 36 °C (Fig. 3) suggest that the TSP of cdaHi is moderately long. The range of interfission ages over which cells Temperature-sensitive periods of Tetrdhymena mutations 65 A 20-min exposure to 36 °C (shock) could induce permanent arrest of division furrows in cdaC2 cells. These cells were most susceptible to such arrest if the heat shock began in the first half of the cell division process (Fig. 1 B). Cells that became arrested following 36 °C shocks beginning at the B or E stages presumably did so because they continued to develop at the high temperature and entered the TSP in mid-division before the heat shock ended. Most of the few cells shifted at the M stage completed division despite the heat shock, presumably because the shock experiments were done on Petri plates, in which a few minutes is required for equilibration at the high temperature, thus allowing cells to reach the ML or the L stage before the temperature reached restrictive levels. cdaAi. The TSP of cdaAi was characterized by temperature shift-up and shock experiments. The majority of cdaAi cells shifted to 36 °C had already completed their temperature-sensitive function by the B stage (Fig. u). Hence, the cdaAi execution point could not be assessed in terms of stages visible in living cells. Instead, it had to be analysed in terms of mean interfission age, a method that is inherently less precise because of inescapable variation in rate of progression through the cell cycle. The results of such an assessment are shown in Fig. 2. The shift-up curve gives the time of execution, denned by the percentage of cells dividing following shifts to 36 °C at progressively later interfission times. This is roughly parallel to the control cumulative division curve, suggesting that each cell completes its temperature-sensitive function at a unique stage of development. The 50% execution time is 69 min after the previous division, or 0-58 of the median interfission time measured in simultaneous control cultures. The proportion of cdaAi cells dividing after 20-min 36 °C shocks is shown in Fig. 2. There was a sharp minimum in the percentage of cells dividing following a 36 °C shock, which preceded the execution point by about 15 min. Cells were thus most likely to be arrested in division by a 36 °C shock that began 15 min prior to the execution point and ended 5 min after the execution point. Interpretation of such information is subject to several complexities, including.the variation in stage at any given interdivision time, the relatively long duration of the shock, and the unknown minimal duration of exposure within the period of temperature sensitivity required to bring about division arrest. Nonetheless, the closeness of the period of

Fig. 1. Percentage of cells dividing following either a shift from 25 to 36 °C (shift-up, A), a shift from. 36 to 25 °C (shift-down, •), or a 20-min exposure to 36 °C (shock, O) as a function of the stage of cytokinesis at the time of the initial temperature shift. The abscissa is plotted according to the duration of each stage; the appearance of normally dividing cells at these stages is indicated in panel (A) (see also Materials and methods). The ordinate is the percentage of cells that complete division, A, cdaAi cells; B, cdaC2 cells; c, cdaHi cells. Open symbols (A, •, O) indicate that the data were obtained with the culture-dish technique, closed symbols (A) the pipette tech- nique, and half-closed symbols (A) that data from the 2 techniques were combined. Numbers of cells are indicated in parentheses at each point, while the mean duration of each stage (in min), is indicated beneath the corresponding stage designations. The estimates for the B stage are more approximate than the others, because pre- division cells assume the boxy shape gradually, and there is no objective basis for deciding when this stage begins. 66 J. Frankel, J. Mohler and A. K. Frankel „„ (14) —o—-^(12) U (16) (20) (6) n > (14) Shock / (28) (4) r--a— °— "" 0 (16),A / ,/ 80 - (29) y f | 60 h (17) f(32) '> \J / Shift-up cdaAi j Control (30) P fc 40 / / OL. / / / • / / 20 -

(28) ,' M22) (9) (14) (28) 0 - (20) a— -°~' i , i 1 1 1 1 1 f 1 I 5 25 45 65 85 105 125 145 165 185 Time after division, min Fig. 2. Percentage of cdaAi cells dividing following a temperature shift-up (A) or shock (O) as a function of time in the cell cycle at the initiation of exposure to the high temperature, compared to the cumulative division curve of control cells maintained continuously at 25 CC (D). The abscissa is plotted according to the interfission age, while the ordinate represents the percentage of cells that completed division. The number of cells at each of the points involving exposure to high temperature is given in parentheses. A total of 9 reasonably homogeneous experiments are represented. temperature sensitivity to the execution point and the relative brevity of this period suggests that the true TSP of cdaAi is short. Those cdaAi cells that subsequently became arrested in division were generally in the B or E stages just after the end of the 36 °C shock. This suggests that the TSP does not precede the onset of cytokinesis by more than about 20 min of development at 36 °C [the equivalent developmental time at 25 °C is about 25—30 min (Cleffmann & Frankel, 1978)]. cdaHi. The results of shift-up experiments analysed according to stage indicate that cdaHi cells complete their temperature-sensitive function near the time of the beginning of division furrowing, the E stage (Fig. ic). When analysed according to interdivision age (Fig. 3), the execution point was estimated at 103 min after division, or 079 of the median generation time. This estimate is in fair agreement with an independent estimate of in min, or 0-85 of the median generation time, which was made by assuming (i) that execution at 36 °C occurs at the beginning of the E stage (Fig. 1 c); (ii) the beginning of the E stage is 20 min prior to completion of division in cdaHi (Fig. 1), and (iii) the median generation time is 131 min (Table 1, Fig- 3)- The results of experiments with 20-min shocks at 36 °C (Fig. 3) suggest that the TSP of cdaHi is moderately long. The range of interfission ages over which cells Temperature-sensitive periods of Tetrahymena mutations 67

1 VJU -u— (12) (8) (4M6)(8\ ••(6) • (20v) ^v 80 - Shock /(15) (28)\ (19)

(12),/ ided 60 (12) (28) i - cdaHi i

it div / 1 g 40 _ / Control / 0- f (12) P Shift-up / 1 20 - (14) / J (14)/ (12) (8) (4) (6) (14) (22) r\ U II.II < i i i i i i i i i__i 25 45 65 85 105 125 145 165 185 205 Time after division, min Fig. 3. Percentage of cdaHi cells dividing following a temperature shift-up (A) or shock (O), as a function of time in the cell cycle, compared to cumulative division curve of controls (•)• All conventions are the same as in Fig. 2. All 3 curves are based on data obtained within 4 identically designed experiments. were subject to division arrest following heat shocks was substantially longer in cdaHi than in cdaAi. Furthermore, cdaHi cells that subsequently became arrested in division were at stages varying from non-dividing (ND) to mid-division (M stage) at the end of the shock. The cdaHi TSP thus begins at least 20 min before the onset of division furrowing, and ends at a time near the beginning of furrowing.

Further development in division-arrested cells Events following division arrest induced by 36 °C shocks were fundamentally similar in all 3 mutants, so they will be considered together. Cells that had been arrested in division as a consequence of a 20-min heat shock generally divided approximately one full cycle after the shock. In those cells in which the original arrested furrow sites remained clearly distinguishable (the majority), the arrested cells did not reactivate the original arrested furrow and eventually formed new furrows within each of the 2 presumptive division products. In the simplest cases the cells completed division along both of the new furrows and divided into 3 cells, anterior and posterior normal ones and a middle cell that still contained the arrested furrow site. More commonly, cells bent as they divided and one or occasionally both of the new furrows abutted on the old arrested furrow. When this happened, division was in general not successfully completed at the occluded site. The result was then typically 2 division products, a normal one at the anterior or posterior end, plus a contorted remainder that included both the original arrested furrow remnant and the new furrow that had become entangled with it. Further information about the development of cells arrested in division could be obtained from analysis of silver-impregnated slides made from samples of mass cultures that had been subjected to 30-min 36 °C shocks. This brief exposure to J. Frankel, jf. Mohler and A. K. Frankel

i-'t I*. ' •! .

V.VA-- 1 * • XV

' > '

,.. /• Temperature-sensitive periods of Tetrahymena mutations 69 36 °C induced the characteristic pattern of division arrest associated with each locus: a failure of ciliary meridians to undergo subdivision in cdaAi (Fig. 5), an elongate division arrest in cdaC2 (Fig. 6), and a distorted development of the fission zone in cdaHi (not shown). Subsequent to the division arrest, renewed oral develop- ment took place synchronously in both components of arrested cdaAi and cdaC2 cells and in the posterior component of arrested cdaHi cells. At later stages, new division furrows appeared and constricted anterior to the newly formed oral apparat- uses (Figs. 8, 9). The fully developed oral apparatus that remained posterior to the arrested furrow was generally not resorbed (Figs. 7—9), yet the fission zone anterior to it remained inactive while active furrowing took place elsewhere. The arrangement of ciliary meridians around the arrested fission zone was often (Fig. 8) but not always quite distorted. Thus, the high-temperature shock affected the developing furrow

Figs. 4-9. Photographs of cdaAi and cdaC2 cells silver-impregnated by the Chatton- Lwoff technique. All photographs are printed with the anterior end of the cell uppermost and with the cell's right corresponding to the viewer's left. Figs. 4-8, x 1200; Fig. 9, x 1000. Fig. 4. A cdaAi cell in mid-division at a permissive temperature (28 °C). The arrows indicate the pre-existing oral apparatus (1) near the anterior end of the cell, and a newly developed oral apparatus (2) just posterior to the division furrow. Fig. 5. The left side of a cdaAi cell, fixed 05 h after the end of a 30-min 36 CC shock, manifesting the cdaA pattern of fission arrest. Both anterior (1) and posterior (2) oral apparatuses are fully developed but out of focus on the viewer's left margin of the cell. Note that ciliary meridians extend uninterrupted from the anterior to posterior end of the cell, with no discontinuity at the division site (marked by thin horizontal lines). Fig. 6. The left side of cdaC2 cell, fixed at the end of a 30-min 36 °C shock, mani- festing the cdaC pattern of fission arrest. The oral apparatuses are labelled as in Fig. 5. Note a complete zone of discontinuities in the ciliary meridians at the division site, marked as in Fig. 5. The presumptive division products are much more elongated than in any normally dividing cell. Fig. 7. A cdaAi cell, fixed 2 h after the end of a 30-min 36 °C shock, starting a new round of predivision oral development subsequent to fission arrest. The anterior and posterior oral apparatuses that had been present at the time of the heat shock are labelled as before, while 2 new stage-3 oral primordia that developed after the end of the heat shock are indicated by arrowheads. The arrested furrow site is indi- cated as in Fig. 5. Fig. 8. A cdaAi cell, fixed 2-5 h after the end of a 30-min 36 °C shock, in a late phase of the post-arrest cell division. There are now 4 complete oral apparatuses, labelled according to generation of origin as in Fig. 7. There are new division furrows within each of the 2 components of the original arrested cell, both located just anterior to the 2 newly developed oral apparatuses (arrowheads). The anterior constriction has virtually been completed (L stage), whereas the posterior new furrow is in the ML stage of constriction. The original blocked furrow remains inactive, with considerable distortion of ciliary meridians in its vicinity. Fig. 9. A cdaC2 cell, fixed 25 h after the end of a 30-min 36 °C shock, in a late phase of post-arrest cell division. Labelling is as in Figs. 7 and 8, with the anterior- most oral apparatus (1) invisible around the rim of the cell. The new division furrows are in stages similar to those depicted in Fig. 8, while the original furrow has completely regressed although the original posterior oral apparatus (2) remains fully normal. 70 J. Frankel, J. Mohler and A. K. Frankel site irreversibly, yet the cells could escape through the subsequent formation and constriction of a new furrow site.

DISCUSSION A characteristic of all 3 of the temperature-sensitive cell division mutants of Tetrahymena thermophila analysed here is that the temperature-sensitive period (TSP) immediately precedes, or coincides with, the diagnostic landmark. This landmark, defined as the first cell cycle stage that is detectably abnormal in the mutant at the restrictive temperature (Hartwell, 1974, p. 167; Pringle, 1978), is the formation of the fission zone in cdaAi and cdaH2 and the active middle phase of furrow con- striction in cdaC2. The phenotypic abnormalities appearing at these landmarks are failure of formation of the fission zone (i.e. absence of equatorial subdivision of ciliary meridians) in cdaAi, formation of a highly distorted fission zone in cdaH2, and an elongated arrest of cytokinesis in cdaC2. The way in which these phenotypic abnormalities arise, and their temporal relationships to the TSPs are indicated schematically in Fig. 10. The relationship between the TSP and the abnormal phenotype appearing at the diagnostic landmark is clearest in cdaC2, in which the 2 are manifested within exactly the same brief span of time. At a permissive temperature (25 °C) living cdaC2 cells are indistinguishable from wild type at all stages, while at a restrictive temperature (36 °C) they retain their normal appearance until the ME stage of division. They then undergo a rapid and extreme elongation, much greater than that which takes place in wild-type cells at the same stage. As the cdaC2 cells elongate, they stop furrowing. cdaC2 cells can be kept at 36 °C during the entire cell cycle and still divide if shifted back to 25 °C just prior to the time when this elongation occurs. Hence, the abnormal cdaC2 gene product either does not function before the time of gross manifestation of its abnormal phenotype, or else if it does function earlier any abnormality in that function can rapidly be repaired up to the very point of actual division arrest. A careful analysis of length/width ratios of several developmental stages in silver-impregnated samples of wild type and cdaC2 cells fixed during exponential growth at 28 °C and also 30 min following a shift to 36 °C revealed no evidence for differential elongation of cdaC2 cells before actual cell division (unpublished).* The most parsimonious interpretation is that the abnormal cdaA.2 gene product acts to generate division arrest by bringing about a sudden exaggerated elongation of the cell during the period of most active division furrowing, and that this action is both stage-specific and dependent solely on the temperature at the crucial moment. The available evidence indicates that the TSP of cdaAi is also short, and comes • Some pre-division elongation was reported earlier (Frankel et al. 1977, table 2). These earlier observations were made on a different allele (cdaCi, or moz") that manifests a very different spectrum of temperature sensitivity (Frankel et al. 19766, table 5), and involved a much longer exposure (3 h) to a higher restrictive temperature (39-5 °C). Temperature-sensitive periods of Tetrahymena mutations 71

Fig. io. Summary of developmental abnormalities and temperature-sensitive periods of cdaAi, cdaCz and cdaHi. The central horizontal sequence of diagrams represents the normal course of cell surface development in the terminal portion of the cell cycle as observed in wild-type cells and in the mutants at permissive temperatures. The diagonal sequences branching off from it represent abnormal sequences of development encountered in the mutants at 36 °C. The bars at the bottom represent estimates of the temperature-sensitive periods of the 3 mutants. The horizontal time axis is shown roughly to scale, and includes approximately the terminal 45 min of the cell cycle. well prior to cell division. The end of the TSP is roughly 45 min prior to the median time of completion of cell division (Fig. 2). On the basis of the known constancy of duration of late stages of development (Suhr-Jessen, Stewart & Rasmussen, 1977; Nelsen, Martel & Frankel, unpublished), the TSP is calculated to coincide with organization of oral membranelles (stages 3 and early-4 of Frankel & Williams, 1973), and to precede the appearance of the fission zone by about 10 min. The cdaAi TSP thus may reflect some discrete process in the construction of this zone, although it cannot be excluded that the process also occurs at earlier times and the TSP only reflects the time at which damage to that process becomes irreparable. cdaHi, unlike the other 2 mutant genes, has a fairly long TSP. The end of this period is close to the onset of division furrowing, but the precise stage of its beginning is more conjectural; it is not known whether it overlaps the TSP of cdaAi. In this 72 J. Frankel, J. Mohler and A. K. Frankel mutant, as in cdaC2, the TSP and the period in which phenotypic abnormalities become manifest appear to coincide. The cdaH mutant alleles generate major spatial dislocations in the fission zone (Frankel, unpublished, cf. Fig. 10). These dislocations, and the ensuing division-arrest, can be initiated over a long time span both before and after the formation of the fission zone. Temporal relationships such as those described here have also been found in mutants affecting cellular development in yeasts. TSPs have been ascertained by temperature shift-down and shift-up for 3 mutants affecting ascospore development in Saccharomyces (Esposito et al. 1970); the TSP of one of these (spo 3-1) is similar to that of cdaAi in that it is very short and precedes the diagnostic landmark by a brief interval, while that of another {spo 2-1) is like that of cdaHi in that it is longer and begins before yet ends after the stage at which phenotypic abnormality is first expressed (Moens, Esposito & Esposito, 1974). For mutants affecting cytokinesis, only the ends of the TSPs (execution points) have been determined, and these generally immediately precede cytokinesis (cell plate formation) in Schizosaccharomyces pombe (Nurse et al. 1976), whereas they appear to occur long before cytokinesis in Saccharomyces cerevisiae (Hartwell, 19716). More recently, it has been observed that in S. cerevisiae a ring of microfilaments appears at the prospective budding site early in the cell cycle (Byers & Goetsch, 1976a), and this ring is not formed in the mutants that prevent cytokinesis (Byers & Goetsch, 19766). Hence in this case also, the end of the TSP and the diagnostic landmark are closely juxtaposed. The Tetrahymena mutants allow a unique demonstration of local irreversibility not yet achieved with mutants in other organisms, cda mutant cells can recover and produce clones following return to permissive temperature after division arrest; however, the arrested division furrows do not recover. Instead, new division furrows develop elsewhere in the cell one generation after the arrest, while the original furrows remain inert. This may be a general characteristic of division furrows, as it was observed not only in earlier studies on the induction of division arrest in non- mutant cells of Tetrahymena pyriformis (GL) (Frankel, 1964, and unpublished), but also in cleaving octopus eggs that were briefly exposed to cytochalasin during cleavage (Arnold & Williams-Arnold, 1974). Presumably, a division furrow suffers irreparable damage whenever there is interference with its development. This might result in a biological amplification of relatively slight deficiencies in the formation or function of the division furrow caused by the effects of altered gene products.

The authors would like to thank Dr Stanley B. Kater for his assistance in measurement of temperature within the micro-drops, and Dr Joseph P. Hegmann for advice on statistical analysis. Valuable suggestions for improvement of the manuscript were provided by Drs Michael C. Newlon and Karl Aufderheide. This research was supported by grant no. HD-08485 from the U.S. National Institutes of Health.

REFERENCES ARNOLD, J. M. & WILLIAMS-ARNOLD, L. D. (1974). Cortical-nuclear interactions in cephalopod development: cytochalasin B effects on the informational pattern in the cell surface. J. Embryol. exp. Morph. 31, 1-26. Temperature-sensitive periods of Tetrahymena mutations 73 ASHIHARA, T., CHANG, S. D. & BASERGA, R. (1978). Constancy of the shift-up point in two temperature-sensitive mammalian cell lines that arrest in Gi.J. cell. Physiol. 96, 15-22. BYERS, B. & GOETSCH, L. (1976a). A highly ordered ring of membrane-associated filaments in budding yeast, jf. Cell Biol. 69, 717-721. BYERS, B. & GOETSCH, L. (19766). Loss of the filaments in cytokinesis-defective mutants of budding yeast. J. Cell Biol. 70, 35a. CLEFFMANN, G. & FRANKEL, j. (1978). The DNA replication schedule is not affected in a division blocked mutant of Tetrahymena thermophila. Expl Cell Res. 117, 191-194. ESPOSITO, M. S., ESPOSITO, R. E., ARNAUD, M. & HALVORSON, H. O. (1970). Conditional mutant of in yeast. J. Bad. 104, 202-210. FRANKEL, A. W. K. (1973). The timing of temperature sensitivity of leaky fertility factor mutations in male . , Princeton 74, 115-132. FRANKEL, J. (1962). The effects of heat, cold and p-fluorophenylalanine on morphogenesis in synchronized Tetrahymena pyriformis GL. C. r. Trav. Lab. Carhberg 33, 1-52. FRANKEL, J. (1964). Morphogenesis and division in chains of Tetrahymena pyriformis GL. J. Protozool. II, 514-526. FRANKEL, J. & HECKMANN, K. (1968). A simplified Chatton-Lwoff silver impregnation procedure for use in experimental studies with ciliates. Trans. Am. microsc. Soc. 87, 317— 321. FRANKEL, J., JENKINS, L. M. & DEBAULT, L. E. (1976a). Causal relations among cell cycle processes in Tetrahymena pyriformis: An analysis employing temperature sensitive mutants. J. Cell Biol. 71, 242-260. FRANKEL, J., JENKINS, L. M., DOERDER, F. P. & NELSEN, E. M. (19766). Mutations affecting cell division in Tetrahymena pyriformis. I. Selection and genetic analysis. Genetics, Princeton 83. 489-506. FRANKEL, J., MOHLER, J. & FRANKEL, A. W. K. (1980). The relationship between the excess- delay phenomenon and temperature-sensitive periods in Tetrahymena thermophila. J. Cell Sci. 43, 75-91- FRANKEL, J., NELSEN, E. M. & JENKINS, L. M. (1977). Mutations affecting cell division in Tetrahymena pyriformis, syngen 1. II. Phenotypes of single and double homozygotes. Devi Biol. 58, 255-275. FRANKEL, J. & WILLIAMS, N. E. (1973). Cortical development in Tetrahymena. In The Biology of Tetrahymena (ed. A. M. Elliott), pp. 375-409. Stroudsburg: Dowden, Hutchinson and Ross. HARTWELL, L. H. (1971a). Genetic control of the cell division cycle in yeast. II. Genes controlling DNA replication and its initiation. .7. molec. Biol. 59, 183-194. HARTWELL, L. H. (19716). Genetic control of the cell division cycle in yeast. IV. Genes controlling bud emergence and cytokinesis. Expl Cell Res. 69, 265-276. HARTWELL, L. (1974). Saccharomyces cerevisiae cell cycle. Bact. Rev. 38, 164-198. HARTWELL, L. H., MORTIMER, R. K., CULOTTI, J. & CULOTTI, M. (1973). Genetic control of cell division in yeast. V. Genetic analysis of cdc mutants. Genetics, Princeton 74, 267- 286. HOWELL, S. K. & NALIBOFF, J. A. (1973). Conditional mutants in Chlamydomonas reinhardtii blocked in the vegetative cell cycle. I. An analysis of cell cycle block points. J. Cell Biol. 57- 76o-772. LEIGHTON, T. (1974). Sporulation-specific translational discrimination in Bacillus subtilis. J. molec. Biol. 86, 855-863. L0VLIE, A. (1963). Growth in mass and respiration rate during the cell cycle of Tetrahymena pyriformis. C. r. Trav. Lab. Carhberg 33, 377-415. J MARTIN, P., MARTIN, A. & SHEARN, A. (1977). Studies of i(3)c43** , a polyphasic, temperature sensitive mutant of Drosophila melanogaster with a variety of imaginal disc disorders. Devi Biol. ss, 213-232. MELERO, J. A. (1979). Isolation and cell cycle analysis of temperature-sensitive mutants from Chinese hamster cells. J. cell. Physiol. 98, 17-30. MOENS.P. B., ESPOSITO, R. E. & ESPOSITO, M. S. (1974). Aberrant nuclear behavior at meiosis and anucleate spore formation of sporulation-deficient {spo) mutants of Saccharomyces cerevisiae. Expl Cell Res. 83, 166-174. 6 CEL 43 74 J. Frankel, J. Mohler and A. K. Frankel NELSEN, E. M. & DEBAULT, L. E. (1978). Transformation in Tetrahymena pyriformis- description of an inducible phenotype. J. Protozool. 25, 113-118. NURSE, P., THURIAUX, P. & NASMYTH, K. (1976). Genetic control of the cell division cycle in the fission yeast Schizosaccharomyces pombe. Molec. & gen. Genet. 146, 167-178. ORR, E. & ROSENBERGER, R. F. (1976). Determination of execution points of mutations in the nuclear replication cycle of Aspergillus nidulans. J. Bact. 126, 903-906. 1 POODRY, C. A., HALL, L. & SUZUKI, D. T. (1973). Developmental properties of Shibire" , a pleiotropic affecting larval and adult locomotion and development. Devi Biol. 32, 373-386. PRESCOTT, D. M. (1957). Change in the physiological state of a cell population as a function of culture growth and age {Tetrahymena geleii). Expl Cell Res. 12, 126—134. PRINGLE, J. R. (1978). The use of conditional lethal mutants for temporal and functional sequence mapping of cell cycle events. J. cell. Pkysiol. 95, 393-406. SHELLENBARGER, D. L. & MOHLER, J. D. (1978). Temperature-sensitive periods and autonomy of pleiotropic effects of l(i)Nw, a conditional notch lethal in Drosophila. Devi Biol. 62, 432-446. SOKAL. R. R. & ROHLF F. J. (1969). Biometry. San Francisco: Freeman. SUHR-JESSEN, P., STEWART, J. M. & RASMUSSEN, L. (1977). Timing and regulation of nuclear and cortical events in the cell cycle. J. Protozool. 24, 299-303. SUMEDA-YASUMOTO, C. & Doi, R. H. (1977). Bacillus subtilis ribonucleic acid polymerase mutants conditionally temperature sensitive in various states of sporulation. J. Bact. 129, 433-444- SUZUKI, D. T. (1970). Temperature-sensitive mutations in Drosophila melanogaster. Science, N.Y. 170, 695-706. SzuLMAjSTER, J., BONAMY, C. & LAPORTE, J. (1970). Isolation and properties of a temperature- sensitive sporulation mutant of Bacillus subtilis. J. Bact. 101, 1027-1037. WILLIAMS, N. E. (1964). Induced division synchrony in Tetrahymena vorax. J. Protozool. 11, 230-236. YOUNG, M. (1976). Use of temperature-sensitive mutants to study gene expression during sporulation in Bacillus subtilis. J. Bact. lift, 928—936.

{Received 18 June 1979 - Revised 23 November 1979)