Asymmetry and directionality in production of new cell types during clonal growth: the switching pattern of homothallic yeast

Cell, Vol. 17, 371-381,

June

1979,

Copyright

0 1979

by MIT

Asymmetry and Directionality in Production of New Cell Types during Clonal Growth: the Switching Pattern of Homothallic Yeast Jeffrey N. Strathern* and Ira Herskowltz Institute of Molecular Biology Department of Biology University of Oregon Eugene, Oregon 97403

Summary Homothallic Saccharomyces yeasts efficiently interconvert between two cell types, the mating types a and a. These interconversion8 have been proposed to occur by genetic rearrangement (“cassette” insertion) at the locus controlling cell type (the mating type locus). The pattern of switching from one cell type to the other during growth of a clone of homothallic cells has been followed by direct microscopic observation, and the results have been summarized as “rules” of switching. First, when a cell divides, it produces either two cells with the same mating type as the original cell or two cells that have switched to the other mating type. This observation suggests that the mating type locus is changed early in the cell cycle, in late GI or during S. Second, the ability to produce cells that have switched mating type is restricted to cells that have previously divided (“experienced cells”). Spores and buds (“inexperienced cells”) rarely if ever give rise to cells with changed mating type. A homothallit yeast ceil thus exhibits asymmetric segregation of the potential for mating type interconversionat each cell division, the mother, but not the daughter, is capable of switching cell types in its next division. Homothallic cells also exhibit directionality in switching: experienced cells switch to the opposite cell type in more than 50% of cell divisions. These results show that the process of mating type interconversion is itself controlled during growth of a clone of homothallic cells. By analogy and extension of these results, we propose that multiple cell types can be produced in a specific pattern during development of a higher eucaryote in a model involving sequential cassette insertion. Introduction During the development of multicellular eucaryotic organisms, cells of one type give rise to cells of other, differentiated types in a process ordered in time and space. The molecular mechanisms responsible for production of specific cell lineages are not understood. Although the yeast Saccharomyces cerevisiae is a lower (that is, single-celled) eucaryote, it exhibits behaviors which may be analogous to processes ocl Present address: Cold Spring Harbor Cold Spring Harbor, New York 11724.

Laboratory,

P. 0. Box 100,

curring during development in higher eucaryotes. In particular, homothallic strains of S. cerevisiae are able to switch between two complex cell types during mitotic growth by a process that is orderly with respect to cell lineage. The two haploid yeast cell types are the mating types a and LY. Cells of opposite mating type mate efficiently with each other. In addition, each type of cell produces a characteristic pheromone which acts specifically on the other cell type, causing arrest in the GI phase of the cell cycle (Duntze, MacKay and Manney, 1970; Bucking-Throm et al., 1973; Wilkinson and Pringle, 1974; reviewed by Manney and Meade, 1977). Cell type is determined genetically by alleles of the mating type locus, MATa for a cells and MATa for OLcells, which have been proposed to code for regulators controlling expression of other loci (MacKay and Manney, 1974; Hicks, 1975; Strathern, 1977). Yeast strains differ in the stability of their mating type. In strains carrying the ho allele (“heterothallic” strains), the MATa and MATa! alleles are stable during mitotic growth, with changes from MATa to MATa and from MATa! to MATa observed at low frequency (-1 O-‘1 (Hawthorne, 1963a; Hicks and Herskowitz, 1976). In some strains carrying the HO allele (“homothallic” strains), cells of one mating type switch frequently to the other mating type. For example, cells with a switched mating type are seen in more than 50% of the clones derived from a homothallic spore after two cell divisions (Hicks and Herskowitz, 1976). Additional switches in mating type are observed after further growth of cells with switched mating type (Oshima and Takano, 1971; Hicks, 1975). The change in mating type is due to a change in the mating type locus itself, and is stable when the HO gene is removed by genetic crosses. HO thus controls a mechanism for efficient interconversion of MATa and MATa alleles, and is not necessary for the maintenance of the new mating type. It has been proposed that the interconversion process occurs by genetic rearrangement mediated by the HO gene. Association of “controlling elements” coded by loci HMa and HMa with the mating type locus leads to an a or an a cell, respectively (Oshima and Takano, 1971; Harashima, Nogi and Oshima, 1974). Based on evidence indicating that existence of cryptic copies of the mating type loci, we have proposed that HMa and HMa are themselves unexpressed MATa and MATa information (informational “cassettes”), which become activated when copies of HMa or HMa are transposed into the mating type locus (Hicks ahd Herskowitz, 1977; Hicks, Strathern and Herskowitz, 1977a). Two features of S. cerevisiae have enabled us to determine the pattern of mating type interconversion which occurs during growth of a clone of homothallic

Cell 372

cells: a and a cells can be distinguished by their sensitivity to the mating pheromone a-factor (Duntze, MacKay and Manney, 19701, and mother and daughter cells can be distinguished since new cells are formed by budding. Two studies (Hicks and Herskowitz, 1976; Hicks, Strathern and Herskowitz, 1977b) showed that the switch in mating type can be observed at the four-cell stage but never at the two-cell stage; that a switch is observed in more than 50% of the clones at the four-cell stage; and that the switch affects a particular pair of cells-for example, cells S and D2 at the four-cell stage, or cells Dl and 01-2 at the eight-cell stage (Figure 1). This paper extends the analysis of the switching pattern beyond the second generation for all cells and presents the “rules” of switching. Our work shows that a clone of homothallic haploid cells is a mixed population of cells with different capacities for switching, and provides information on the rate and directionality of switching. We also discuss analogies between the switching process in homothallic yeast and aspects of development in higher eucaryotes. Results Sensitivity to the mating pheromone a-factor can be used to monitor changes in mating type in homothallic cells (Hicks and Herskowitz, 1976). In the presence of a-factor, a cells are arrested in the GI phase of the cell cycle and form pear-shaped cells, while (Y cell growth is unaffected (Duntze et al., 1970). Changes from a to a cell type are observed as changes from a-factor resistance to a-factor sensitivity, and reflect a switch from MATa to MATa. Because a cells are inhibited by a-factor, observations of changes of mating type performed in the presence of a-factor are limited to a single change from (Y to a. To monitor subsequent switches, a protocol was developed in which spores of known mating type were allowed to germinate in the absence of a-factor and tested for a-factor sensitivity after one or two generations. Switches of Mating Type in Successive Cell Divisions by MATa HO spores Homothallic MATa spores were identified in asci derived from sporulation of MATa/MATa HO/HO strain Xl O-l 6 as described in Experimental Procedures. Twenty two MATa spores were allowed to divide once and were grown to form two cells with small buds, at which time they were placed in the presence of (Yfactor. Because a-factor acts at the unbudded stage in the cell cycle, each budded a cell continued to develop after being transferred to the presence of (Yfactor and formed two cells. In all clones at the four-cell stage, the spore’s first daughter (Dl) and the cell derived from it (Dl-1) were arrested by a-factor (Figure 2). This result is expected

Parent Cell (“mother”)

Bud Cell (“daughter”) SPORE

0

S(1)

OL

/\ S(2)

ai

/“\ S(4) @

D7@\ D2W

I\

I\

OOOO@@@@ S(8) D3(8) Figure

D1#) @

@

1. Pedigree

Dl-l(4)

@

I\

D2(8)

02-l(8)

Analysis

Dl(8)

of Switching

I\

M-2(8)

M-1(8)

from MATa

M-l-1(8)

to MATa

Daughter cells are drawn to the right and mother cells to the left at each cell division. (S) spore ceil: (Dl), the spore’s first daughter; (D21, the spore’s second daughter; (Dl-I). the first daughter of cell Dl , and so on. Numbers in parentheses indicate the stage of growth of the clone (one-cell stage, two-cell stage and so on). a and a cells were distinguished by resistance and sensitivity to a-factor. respectively, as described in the text. Pattern A (2/22)

Pattern B (20/22)

S(4) D2C4) Dl(4) DlW

I\ I\

=34( T;

cxyoLo(a

Pattern

Figure 2. Switches MATa HO Spores

of Mating

81 W20)

Type

In Successive

Pattern

82 (17/20)

Cell Divisions

by

MATa spores were identified without exposure to a-factor as described in Experimental Procedures. Cells were allowed to grow in the absence of u-factor to the two-cell stage and to produce small buds, at which time they were transferred to the presence of a-factor. (‘1 Indicates that the cell gave rise to switched cells in its next division.

from the results of Hicks and Herskowitz (1976) since Dl and Dl-1 exhibit the same mating type (a-factor response) at the four-cell stage as did the original spore. This observation confirms that the spores were correctly identified as MATa. The other pair of cells at the four-cell stage, the oldest cell [S (411 and its second daughter [D2 (411, gave two types of response. In two clones, S (4) and D2 (4) were sensitive to afactor, indicating that no switch in mating type had occurred (Figure 2, pattern A). In the remaining 20

Switching 373

Pattern

of Homothallic

Yeast

clones, S (4) and 02 (4) were resistant to a-factor, and budded to give rise to D3 (8) and D2-1 (8) (Figure 2, pattern B). This behavior indicated that S (4) and D2 (4) had changed mating type from a to a in these clones. The timing and pattern of a second switch was obtained by further observation of ceils which exhibited pattern B-that is, those which had switched to (Y at the four-cell stage. The subsequent appearance of a-factor-sensitive cells indicated that a second switch in mating type (to MATa) had occurred. In 3 of 20 clones, S (81, D3 (81, 02 (8) and D2-1 (8) were insensitive to a-factor (pattern Bl). In the other 17 cases, S (8) and D3 (8) were sensitive to a-factor, while 02 (8) and D2-1 (8) were insensitive (pattern 82). The oldest cell and its third daughter thus showed the result of a second conversion of mating type. In summary, while the first mating type switch is invariably observed only after the second cell division of a spore, a second mating type switch can occur in the next division of the spore cell. It should be noted that although the fraction of clones which exhibited a change in mating type in the pair S (8) and D3 (8) was very high, no change was observed in the pair D2 (8) and D2-1 (8). In this respect, D2 (4) is analogous to Dl (2) and to S (l)-the cells produced at the next generation have the same mating type as the mother cell in all 62 divisions of these cells in the experiment shown in Figure 2. In contrast, S (2) and S (4) cells gave rise to cells of opposite mating type in 37 of 42 cell divisions. The common feature of cells which gave rise to a pair of cells with altered mating type is that they had previously budded. Switches of Mating Type in Successive Cell Divisions by MATcr HO Spores Spores from X10-1 B were identified as MATa by observing that two of the three other spores from the same ascus were sensitive to a-factor, while the third was insensitive. These MATa HO spores were allowed to germinate in the absence of a-factor on a slab of YEPD agar, and the cells were separated at each generation. The clones were allowed to develop into four budded cells and then transferred by micromanipulation to the presence of a-factor. The results of subsequent incubation are presented in Figure 3 for 18 such pedigrees. In each clone at the eight-cell stage, Dl-1 (8) and Dl-l-l (8) exhibited the a phenotype. We have invariably observed cells related to the spore through a series of “first buds” [cells Dl (2), Dl-1 (4), Dl-l-l (8) and so on] to have the same mating type as the spores; this relationship has been found for at least twelve generations of “first buds” from a MATa HO spore (our unpublished observations). The behavior of cells Dl -1 (8) and Dl-l-l (8) thus provides confirmation of the mating type assigned to the spore. The

Pattern

Figure 3. Switches MATa HO Spores

31),4,18~

of Mating

Pattern

Type

3E

in Successive

(3118:

Cell Divisions

by

MATa spores were identified without exposure to a-factor as described in Experimental Procedures. Cells were allowed to grow in the absence of e-factor to the four-cell stage and to produce small buds, at which time they were transferred to the presence of a-factor. (*) Indicates that the cell gave rise to switched cells in its next division.

other six cells at the eight-cell stage gave five different responses (patterns A, B, C, D and E in Figure 3). -Pattern 3A (2/18). Cells S, D3, D2 and D2-1 responded to e-factor, while Dl , Dl -2, Dl -1 and Dl -l1 did not do so. This pattern is interpreted to result from a switch in mating type from a to a during division of cell S (2), with no further switches in the next generation. Another explanation, that independent switches occurred in divisions by S (4) and 02 (41, is inconsistent with the results of the previous section, in which it was seen that D2 (8) and D2-1 (8) have the same phenotype as D2 (4) (see also Figure 4). -Pattern 38 (1 /18). Using the same argument, that the phenotype of D2 (8) and D2-1 (8) reflects the phenotype of D2 (4) and hence S (41, this pattern is seen to be the result of independent switches from a: to a during division of cells S (4) and Dl (4). Pedigrees of (Y HO spores performed in the presence of a-factor (Figure 4) support this interpretation. -Pattern 3C (8/18). Reasoning as above, this pattern can be explained as the result of two switches, one during division of cell S (2) and another during division of cell Dl (4). -Pattern 3D (4/18). The simple explanation of a switch from (Yto a in the progeny of cell D2 (4) is ruled out by the logic applied above. We believe that two successive switches have occurred, the first during division of cells S (2) and the second during division of cell S (4). -Pattern 3E (3/l 8). These clones are interpreted to have had successive switches by S (2) and S (4) as in pattern D, and a third switch during division of cell Dl (4). In summary, a switch from (Y to a occurred during

Cell 374

the cell division cycle of cell S (2) in 17 of the 18 pedigrees. In 7 of these 17 cases, the S (4) cell then gave rise to a! cells in its next division. This ability to switch mating type in successive generations is in marked contrast to the ability to change mating type in the first division of the spore or in the division of the cells related to the spore by a succession of “first buds.” Experienced and Inexperienced Cells The above results lead us to propose that cells within a clone differ in their ability to switch mating type. In particular, those cells which had previously budded (“experienced cells”) were capable of giving rise to two cells of altered mating type. In contrast, those cells which had not previously budded C’inexperienced cells”), such as the initial spore cell or a new daughter, gave rise to two cells with the same mating type as the parent. Further support for the distinction between experienced and inexperienced cells comes from experiments of the kind performed by Hicks and Herskowitz (1976), in which the growth of MATa HO spores was followed in the presence of a-factor for three or four generations. In 333 such pedigrees analyzed (Figure 41, a cells were never observed at the two-cell stage. This result conforms to the rule that the S cell at the one-cell stage was inexperienced and hence unable to give rise to switched cells. At the two-cell stage, the experienced a cell S (2) gave rise to a cells in 87% (289/333) of the cell divisions. All progeny of the 333 inexperienced Dl (2) cells remained as (Y. At the four-cell stage, the two experienced cells S and Dl gave rise to progeny cells with switched mating type 89% (35/44) and 54% (170/315) of the time, respectively. No switches were seen among the progeny of the inexperienced cell Dl-1 (O/31 5). The other inexperienced cell, D2 (41, gave rise to cells with switched mating type in 2 of 44 divisions. In summary, 494/692 (71%) of the divisions by experienced

cells

resulted

in switched

cells;

only

two

925 divisions of cells which had been defined as inexperienced. (It is not known whether these exceptions represent bona fide switches by inexperienced cells, or whether they are experimental errors in cell identipairs

of switched

cells

were

observed

among

the

fication.)

Switches of Cells Several Generations after Germination An a HO spore from Xl O-l 6 was allowed to divide to produce 16 cells, with care taken to order the cells. The cells (most with buds) were transferred to the presence

of a-factor

and

scored

for

the

production

of

The resultant pedigree is shown in Figure 5. The mating type of the cells prior to the 16-cell stage can be inferred from the pheno-

a-factor-sensitive

cells.

Cells Whtch Gave Rise To Swttckc Cells

Pattern A

2691333

Al

1481271

\ 5

/ & il u

A2 1231271

82

19/44

83

3/44

a P, /\

\ 21 I\ II

‘k

L2UNN.PI

a

/\

Lx

o(

D(4)

cd

“One


S(4). Dl(4)

NcYcslX‘XO(
“O”.CZ

B43/44

Nau‘xae‘Yu

m (4)

85

anaaaa

Figure Analysis a-factor.

2144

4. Switches

from MATa

was performed

Figure 5. Switches mination of a MATa

(Y

to MATa

as in Figure

of Mating HO Spore

Type

‘Y

by MATa

D2(4), Dl(4)

HO Spores

1, in the continuous

Several

Generations

presence

after

of

Ger-

A MATa spore was identified without exposure to n-factor as described in Experimental Procedures. Cells were allowed to grow in the absence of u-factor for a few generations. as indicated, and to produce small buds, at which time they were transferred to the presence of a-factor. Cells below the horizontal line developed in the continuous presence of a-factor. The mating type of the cells above the line was inferred; the mating type of the cells below the line was determined by resistance or sensitivity to a-factor.

types at the 32-cell stage, and from the rules that switches occur in pairs and are produced by division of experienced cells. This pedigree demonstrates that at least four changes in mating type (a + a + a + (Y -+ a) can occur in a cell lineage (this particular lineage involves cell D4). It is clear from this pedigree that the ability to change mating type is not confined to the

Switching 375

Pattern

of Homothallic

Yeast

early divisions after germination (see also Klar and Fogel, 1977). Switching occurs very efficiently in the later generations (for example, 26 of 27 experienced cells in generations 6, 7 and 6 gave rise to switched cells), and was observed to occur as late as the twelfth generation. A Search for Undetected Switches Our conclusion that a population of homothallic cells contains both cells competent to switch and cells not competent to switch is based on assaying switching events by changes in cell type from (Y to a and from a to (Y. Viewing the switching pattern in the context of the cassette model leads us to consider whether we are failing to observe some mating type interconversion events. According to the cassette model, homothallic cells contain both silent MATa and silent MATa information. Hence it is possible that cells which do not manifest a change in cell type have nonetheless undergone an interconversion event-for example, replacing the MATa cassette at the mating type locus with another MATa cassette (“homologous cassette replacement”). We have therefore asked the following two questions. First, do experienced cells always undergo a mating type conversion event? J. Rine (personal communication) has found that a MATa strain with only silent MATa information (that is, of genotype HO hma HMa MATa) did not switch to MATa in 3 of 30 cell divisions by experienced cells. This result shows that experienced cells need not undergo an interconversion event during each cell division. Failure to observe mating type interconversion in this strain is clearly not due to homologous cassette replacement. Second, does homologous cassette replacement occur? In growth of a MATa HO HMa HMa spore, we have seen that cell Dl (2) does not give rise to afactor-sensitive cells. Have these cells sustained a homologous cassette replacement or has there been no transposition event at all? We can also ask whether or not homologous cassette replacement has occurred in cells S (4) and 02 (4) when these cells exhibit the same a-factor-resistance phenotype as S (2). To answer these questions, we marked the mating type locus genetically with the matd-5 mutation, which leads to defective mating. We then determined whether a matal - HO HMa HMa spore gives rise to MATa+ cells at the four-cell stage-specifically, cells Dl and Dl-1 (case A), and cells S and D2 (case B). Because both metal - and MATa’ cells are insensitive to a-factor, we have distinguished these cells by their mating proficiency. The matalmutation does not affect the switching process; matal -5 HO HMa HMa cells switch readily to MATa and then to MATa (Hicks and Herskowitz, 1977). -Case A (Figure 6A). If cells Dl (4) and Dl-1 (4) have undergone homologous cassette replacement,

A.

B.

S(2) /\ a

Dl(2) .--A--,

a

1L.-----E; cc r--------l I CL* L--------J

oc 1

Figure 6. Searching for Undetected ing by metal- HO Spores

r-------Yi LE---“c: Switches

by Analysis

of Switch-

(a-1 indicates that the cell carries the matal mutation and is defective in mating: (a+) indicates that the cell is MATa’ and is proficient in mating. (A) mat& HO spores were allowed to grow to the four-cell stage without separation of cells by micromanipulation. Zygotes were not formed between S (4) or D2 (4) and Dl (4) or Dl-1 (4). indicating that the latter pair of cells remained a-. (B) matal- HO spores were grown to the four-cell stage, with progeny cells separated by micromanipulation. Cells S (4) and D2 (4) were analyzed for mating proficiency by cell-to-cell mating with a cells. The results are described in the text.

then these cells would be MATa+ and able to mate with S (4) and D2 (4) cells which have switched to MATa. When matalHO spores are grown to the four-cell stage in the absence of a-factor without separation of mother and daughter cells, no zygotes formed between siblings were observed in 50 pedigrees (J. B. Hicks and J. Strathern, unpublished observations). In contrast, more than half the clones at the four-cell stage derived from MATa+ HO spores contained zygotes. These results indicate that cells Dl (4) and Dl-1 (4) do not undergo homologous cassette replacement, and thus that the Dl (2) cells are not competent to promote genetic rearrangement. -Case B (Figure 6B). In a pedigree derived from a matal - HO spore in which cells S (4) and 02 (4) are insensitive to a-factor, these cells would be MATa+ if homologous cassette replacement has occurred. We attempted to determine whether this event occurs in the following experiment. Seventy mata- (a-factorresistant) spores derived by sporulation of MATa/ mata- HO/ho were grown to the two-budded-cell stage. The progeny of the budded S (2) cell, S (4) and D2 (4), were then placed by micromanipulation next to a cells and examined for zygote formation. No zygotes were observed, indicating that switches from mata- to MATa did not occur. The number of possible switches of this type which might have been observed in this experiment was small, however, for the following reasons. First, only half of the mata- spores are expected to contain HO. Second, as indicated by other experiments, approximately 60% of cells S (4) and D2 (4) are expected to have switched to a. Third, the efficiency of cell-to-cell mating between a and a cells varies from 50-100% (our unpublished observations). Thus the maximum number of cells with a

Cell 376

possibility of having switched to MATa is approximately 7 [(70) (0.5) (0.211, of which 3-7 would be expected to have been detected by ceil-to-cell mating. These results suggest that mat&-5 does not efficiently switch directly to MATa in these strains. In summary, these results indicate that experienced cells need not undergo a mating type interconversion event. They also provide evidence against the occurrence of undetected mating type interconversion events due to homologous cassette replacement, and support our view that inexperienced cells do not undergo transposition events.

enced cells (for example, S and Dl ) for all generations [for example, S (2) and S (4)]? Is the probability of switching affected by whether the progenitors or siblings of a cell have switched? Table 1 summarizes our results on these questions. -Switching by S versus non-S cells (Table 1, part 1). In the pedigree experiment of Figure 4, the switching frequency of experienced S cells [S (2) and S (4)] was 0.86, compared with a frequency of 0.55 by the other experienced cells, which in this experiment were Dl (4). Experienced cells at the four-cell stage [S (4) and Dl (411 provide the best comparison of two different groups of experienced cells under the same conditions (Table 1, part 10; the switching frequency for S (4) is 0.80 and for Dl (4) is 0.55. This difference appears to be significant @hi* = 8.8; p c 0.01) and indicates that in this experiment, experienced S cells switched at higher frequency than experienced

Quantitative Analysis of Switching by Experienced Cells Do all experienced cells within a homothallic clone have a similar probability of switching? Specifically, is the probability of switching the same for all experiTable

1. The Switching

(1) Effect

Frequency

of Cell (S versus

(A)

S(2) and S(4) versus

(B)

S versus

Cells

Non-S) Dl(4)

non-S

(C) S (4) versus

(2) Effect

of Experienced

Dl (4)

of Generation

(Early

versus

Figure

4

switches S

Figure

5

switches -0 S

2 4

Figure

3

switches -= s (4)

6 iii

= o.44

Figure

4

switches -= s (4)

35 44

= OBO

Total (C)

switches -= S(4)

43 62

=

6 14

= g

switches Dl (4)

= g

= 0.55

switches non-S

35 = 43

= 0.81

switches D1o

12 = z

= 0.67

switches Dl (4)

= g

= 0.55

switches Dl (4)

= g

= 0.55

= 0.57

switches gen 6-12

=

29 33

= 0.66

= 0.91

~switches S(4)

=

_17 20

= 0.65

= 0.94

-

switches s (4)

=

6 14

= 0.57

=

35 44

= 0.60

60 78

= 0.77

72 ss

= o.75

=

0.86

= 0.5

0.69

Late after Germination)

(A) Generations generations

l-5 versus 6-l 2

Figure

5

switches gen l-5

=

(B)

S (4)

Figure

2

switches s

=

Figure

3

switches S(2)

I

Figure

4

switches s (2)

= E

= 0.67

switches s

Total (B)

switches = g s c-4

= 0.67

switches r(4)

Total’

switches s (2)

= E

= 0.67

switches S.

= E

= 0.67

switches s (4) x

35 = 44

=

25 = 42

=

switches pt-ns

42 = 53

= 0.79

switches Dl (4)~

24 = 44

= 0.55

(C)

S (2) versus

Two-cell stage versus S (4) and Dl (411

(3) Effect (A) (B)

of Switching

four-cell

by Progenitors

S (2) versus S (4) cells whose switch [S (4) x]

stage

[S (2) versus

did not

Cells whose parent switched in the previous generation [p&s] versus cells whose parent did not switch Dl (4) when S (2) switched S (2) did not switch [z]

’ Based on Figures 2-4. b Experienced parents. ’ Based on Figures 2-5.

17 16

=

and Siblings S (2) parents

Figure

4

switches s

(2)

switches -

TotalC

bt-nsP W

20 22

pt-s

[y] versus

Dl (4) when

Figure

4

switches = gf Dl (4) Y

0.60

= 0.55

0.60

Switching 377

Pattern

of Homothallic

Yeast

daughter cells. It should be noted, however, that experienced non-S cells can exhibit switching frequencies considerably higher than 0.55. For example, in the experiment in Figure 5, the switching frequency of experienced non-S cells was 0.80 (35/43). The reason for the difference between S and Dl cells in Figure 4 is therefore unclear. -Effect of generation (Table 1, part 2). The switching frequency of experienced cells remains high after the second cell division. The switching frequency of experienced cells at the four-cell stage [0.77 for S (4), or 0.75 for S (4) and Dl (4) together] is approximately the same as that at the two-cell stage (0.87). Furthermore, in the extended pedigree of Figure 5, the switching frequency was 0.88 (29/33) for experienced cells in generations 8-l 2. -Effect of genealogy (Table 1, part 3). The switching frequency of an experienced cell does not appear to be greatly affected by prior switching by a progenitor or by switching by siblings. In Figure 4, the frequency of switching by S (4) cells whose S (2) parent had not switched was approximately the same (0.80) as the frequency of switching by S (2) cells. Unfortunately, it was not possible in this experiment to determine the frequency of switching by S (4) cells whose S (2) parents had switched. Over all pedigrees (Figures 25), the frequency of switching by cells whose experienced parent had switched in the prior cell division was 0.60 (25/42), in comparison with a switching frequency of 0.79 (42/53) by cells whose experienced parents had not switched. Although this difference may be statistically significant (ch? = 4.9; p = 0.025), parental history does not greatly affect the frequency of switching. Similarly, the history of switching by one member of a homothallic clone [cell S (2)] does not affect the frequency of switching of its daughter Dl; the switching frequency of Dl (4) was 0.55 regardless of whether its parent S (2) had switched or not. Discussion The Switching Pattern Homothallic yeast cells switch between the two mating types in a specific pattern which can be accounted for by the following “rules” of switching. First, cells which have budded at least once (“experienced cells”) are capable of giving rise to cells of changed mating type. Cells which have not budded at least once (“inexperienced cells”) rarely, if ever, give rise to cells of changed mating type. Second, cells which exhibit changed mating type always occur in pairs (the experienced cell and its daughter). In addition to these two rules, we have observed the following specific features of mating type interconversion: switching is directional--that is, experienced cells switch more than 50% of the time; switching from a to (Y is approx-

imately as frequent as switching from (Y to a; the frequency of switching from (Y to a is not influenced by a-factor; and the probability of switching by individuals within a clone is not influenced by the history of switching within that clone. The switching pattern of homothallic yeast thus reflects an intrinsically determined program capable of producing a variety of cell lineages. Asymmetric Segregation of the Potential for Mating Type Interconversion Previous work (Hicks and Herskowitz, 1976) indicated that after germination of a homothallic spore, a switch in mating type was often observed at the second generation, but never earlier. In our work we have seen that switching from one mating type to another does not always require two generations. In particular, cells can switch mating type in successive generations (from MATa! to MATa to MATa and from MATa to MATa to MATa). The inability of the S cell to give rise to switched cells in its first division appears to be due to the requirement that cells must bud at least once before they can give rise to switched cells. Of 609 switches reported here, 607 were by cells which had alreadv budded at least once. Some event must occur during the first cell division cycle of an experienced cell which allows this cell to become competent to switch mating type in a subsequent division. The nonequivalence in the ability of mother and daughter cells to switch in the next cell cycle may reflect asymmetric distribution of a cytoplasmic component, differences in membrane constitution between mother and bud, or differences in the distribution of genetic information. Hartwell and Unger (1977) have observed that the daughter cell is smaller than its parent and has a longer cell division cycle. Perhaps a critical concentration or absolute amount of some factor must accumulate before mating type interconversion can occur. Another possibility is that there may be some modification of the genome required to attain competency to switch (reviewed by Holliday and Pugh, 1975). Such a modification might alter only one strand of the DNA and be segregated only into the mother cell. Cells competent to switch (“C cells”) and cells not competent to switch (“N cells”) can, in theory, exhibit either asymmetric cell divisions (N + C and N; C + C and N) or symmetric cell divisions (N or C + 2 N or 2 C). As noted above, asymmetric divisions are observed; furthermore, the mother is invariably C and the daughter is invariably N. Are symmetrical divisions ever seen? It is probable that N or C + 2 C does not occur, since it would have been detected due to production of, for example, four switched cells derived from a single progenitor. We do not know whether N or C cells ever give rise to two N cells, a finding which would indicate that an experienced cell is not neces-

Cell 378

sarily competent to switch. Although we know that experienced cells need not switch (27% did not switch in our experiments), we do not know whether these cells had acquired competence to switch. Timing of the Switch The observation that two cells produced by any cell division always have the same phenotype (sensitivity or resistance to a-factor) suggests that the change in the mating type locus occurs before DNA replication in the prior cell cycle, and that this alteration is segregated to both daughter nuclei. This suggestion is reinforced by studies of switching in homothallic MATcx/MATor and MATa/MATa diploids. In the case of MATa/MATor, the S (2) cell can give rise to either two MATa/MATa cells or to two MATa/MATa cells (Hicks et al., 1977b). Since both kinds of cells are produced, the switching event can apparently act independently on two mating type loci present in a diploid. If the switching mechanism could act independently on the two mating type loci present after DNA replication of a haploid, we would have expected to see instances in which daughter cells had different mating types. Similarly, if the switching mechanism could act independently on the four mating type loci present after DNA replication of a MATa/MATa diploid, we would have expected to observe cases in which one daughter cell was MATa/MATa and the other MATa/MATcx. In summary, the observation that changes in mating type always occur in pairs of haploid or diploid cells, but that not all MAT loci in a diploid cell need be changed, suggests that the alteration of the genome which results in a change in mating type occurs prior to replication of MAT. Another possible explanation for the observation that switched cells occur in pairs is that there is a one generation delay between the alteration of the mating type locus and the expression of the new phenotype. If this were true, then certain cells in a pedigree, particularly S (2) cells, should be phenocopies (cells in which the MAT allele has changed but which have the mating phenotype of their parent). Since 87% of the MATa spores from Xl O-l B gave rise to two MATa cells at the four-cell stage, the phenocopy model predicts that 87% of the S cells at the two-cell stage should be genotypically MATa but still express the a! phenotype. We observe, however, that zygotes formed between S (2) cells and heterothallic a cells do not behave as MATa/MATa diploids with respect to either sensitivity to a-factor or continued exhibition of mating type switches (of which MATa/MATa cells but not MATa/MATa cells are capable) (Strathern, 1977). We conclude that at the time of mating, S (2) cells are not phenocopies and still have the MATa locus. The observation that these S (2) cells are both genotypically and phenotypically (II indicates that mating type changes seen in cells S (4) and D2 (4) must have occurred after the point in the S (2) cell cycle at which cells mate and at which cells are arrested bv

mating factors. In summary, we believe that switching occurs after the time in the Gi phase of the cell cycle defined by pheromone arrest, and before or at the time of replication of the mating type locus. Directionality Experienced cells give rise to cells of changed mating type in more than 50% of cell divisions. 71% (31/44) of the experienced a cells switch to (Y, and 73% (576/ 791) of the experienced (II cells switch to a. Because the frequency of switching is greater than 50%. it is clear that the switching event is not the result of a random decision between two states, but rather a directed change of mating type, as indicated by Hawthorne (1963b). Two kinds of models can be envisioned to account for this directionality. In the first, the mating type locus codes for a product which directs its own removal. For example, in the context of the cassette model, the MATa locus might specify a function which leads to insertion of the silent MATa locus rather than the silent MATa locus, or which blocks some step of replacement by the silent MATa. A second, structural model proposes that the mating type locus contains recognition regions, perhaps analogous to prophage attachment sites, which can interact productively with recognition sites of one of the silent mating type loci, but not with the other. The observation that 73% of experienced cells switch to the opposite mating type can be interpreted in two ways: -The switching process is completely directional. All mating type interconversion events result in the substitution of heterologous information at the mating type locus. In some fraction of experienced cells (27% in our experiments), no interconversion event occurs. -The switching process is not completely directional. In some cases (73%), mating type interconversion events in experienced cells result in the substitution of heterologous information at the mating type locus; in the other cases, homologous information is substituted at the mating type locus. The observation that experienced cells need not undergo an interconversion event during each cell division leads us to favor the first hypothesis. Factors Affecting the Frequency of Switching by Experienced Cells As described above, certain aspects of the switching pattern are fixed -only experienced cells give rise to switched cells, and mother and daughter cells of a given generation always exhibit the same mating type. Pedigrees differ from each other, however, because not every experienced cell gives rise to cells with changed mating type; only 73% of all experienced cells switch. Is the probability that an experienced cell will switch constant or is it affected by the environment (for example, the nutritional medium or the presence of a-factor). the matina tvoe locus or the switchina

Switching 379

Pattern

of Homothallic

Yeast

history within the clone? We have not yet found any external factor which affects the frequency of switching. The frequency of switching by a cells at the two-cell stage in the presence of a-factor (89%; 289 switches per 333 cells; Figure 4) is essentially the same as the frequency of switching in the absence of a-factor (95%; 17 switches per 18 cells; Figure 3). Furthermore, the frequency of switching, as judged by zygote formation among siblings, is approximately the same on minimal medium with glycerol or glucose as carbon source as on rich medium supplemented with glucose as carbon source (data not shown). Whether switching frequency by strains with a partial defect in the HO gene (“HO-slow”; L. Blair, personal communication) is increased on minimal medium has not yet been determined. Previous work (Hicks et al., 1977b) has shown that the mating type locus greatly influences the switching frequency-a/a diploids do not switch, whereas a/a and a/a diploids do switch. This difference is one of many between a/a and a/a or a/a cells with respect to sporulation, mating and associated processes, and presumably reflects a turn-off of the switching process by an as yet unknown mechanism. In our experiments, we found that the frequency of switching by experienced a cells is approximately the same as that by experienced a cells (71%, 31 switches per 44 cells, and 73%, 576 switches per 791 cells, respectively). The wild-type MATa and MATa loci thus do not affect the frequency of switching [although a variant of MATa, MATa-inconvertible, which switches only at low frequency and which may be an inefficient substrate for the switching machinery, does exist (Takano. Kusumi and Oshima, 1973)]. As noted above, the mating type locus does control the direction of switching, since switching from one mating type to the other occurs in more than 50% of experienced cells. Although there is some variability in the frequency of switching by experienced cells-notably in cells S (2) or S (4) and Dl (4) in Figure 4-the general rule appears to be that the switching frequency of an experienced cell is influenced little or not at all by the cell type (S versus non-S, a versus a), by its age (that is, distance from germination) or by the history of switching by its parent or siblings (see Table 1). The control of switching frequency therefore does not appear to be influenced, for example, by a factor stored initially in spores which is partitioned to daughter cells through cell division. Rather, the frequency of switching appears to be a function of the probability of some event(s) occurring during each cell cycle. Implications of Switching in Successive Cell Divisions The ability of a cell to switch mating type in successive generations has been seen in two ways: first, in the pedigrees of Figures 2 and 3, in which changes from MATa to MATa to MATa or from MATa to MATa to

MATa are inferred; and second, by the ability of matacells to switch to MATa and then to MATa by the eight-cell stage. Because both a genotypic and a phenotypic change in mating type can occur during each cell cycle, the cellular components conferring the mating phenotype-which include pheromones, the pheromone response system and agglutination factors-must be replaced within one cell cycle. We propose that the genotypic change occurs early in the cell cycle, at or before the time of replication of the mating type locus. The cell then has nearly a full cell cycle, until the next Gl, in which to express new mating type-specific functions. The old mating type functions may be removed either by spontaneous degradation or by a function specific to homothallic cells. Analogies with Development in Higher Eucaryotes In the development of multicellular eucaryotes, production of different cell types occurs in part through rigidly programmed cell lineages (as in Caenorhabditis elegans; see Sulston and Horvitz, 1977). The switching process of homothallic yeast also produces specific cell lineages, derived from the rules that only experienced cells give rise to cells of changed cell type, and that mother and daughter cells at any generation are of the same cell type. A clone of homothallic yeast cells is made up of two superimposed cell lineagescells of a or a mating type, and cells competent or incompetent to switch. Competent cells, which differ from incompetent cells in ways which are not yet understood, are analogous to determined cells-that is, they are poised for visible change. Segregation of yeast nuclei into the mother’s and daughter’s cytoplasms may be responsible for the mother cell acquiring competence to switch. In similar manner, segregation of embryonic nuclei into different cytoplasms as a result of cleavage may govern the developmental fate of these cells. a and a cells represent the differentiated cell types, exhibiting observable phenotypes as a result of activation of either a MATa or a MATa cassette by insertion into the mating type locus. Homothallic clones are similar to stem cell-differentiated cell populations in that the original cell type is always maintained, with differentiated cells produced during growth of the clone. For example, starting with an a spore, a “stem cell” line of first buds is maintained, and “differentiated” cells (a cells or a/a cells resulting from mating) are produced. By varying some parameters of the yeast switching process, it is possible to account for an important feature of higher eucaryotic development, sequential production of multiple cell types. This model, which invokes sequential cassette insertion, arises from the observation ttiat switching of homothallic cells from one type to the other is directed: MATa cells switch preferentially to MATa and vice versa. In yeast cells, the switch from one cell type to the other is proposed

Cell 380

T\ /-‘\ A /“\ /‘\ A wFiAi7A14A6

CCBBAAAABBAAAASS Figure 7. Production Sequential Cassette (S) stem competent division.

of a Cell Lineage Insertion

with Multiple

cell: (A, B. C) differentiated cells. to give rise to cells of a different

Cell Types

by

Underlined cells type in the next

are cell

Acknowledgments

to occur by activation of information from a “library” which contains only two cassettes, silent MATa and MATa. More complex cell lineages could be produced by having a stem cell (cell S; Figure 7) with a more extensive library-for example, containing silent cassettes CASA, CASB, CASC and so on, which determine cell types A, B, C and so on. As in yeast, the position of the cassette would determine whether it is expressed, with the active cassette being at a “playback” or “read-out” locus. Production of differentiated cells would occur by replacement of the active CA.33 in the S cell with another cassette. The time and order of appearance and the number of cells of each type could be controlled by intrinsic factors, a8 follows: -Cells with an active CASS might switch only to CASA, cells with an active CASA might switch only to CASB and so on. Hence differentiated cells would appear in the order A, B, C (Figure 7). -The efficiencies of switching from one cassette to another may differ. For example, the switch from CASA to CASB may be more efficient than the switch from CASB to CASC, resulting in a clone containing more B cells than A or C cells. Different numbers of cell types would also result if the growth rates of A, B and C cells differed. In addition to such intrinsic factors, extrinsic factors such as cell-cell contact or hormone action could also change cell lineages, for example, by affecting the probability of switching or the cell division rate, or by determining which silent cassette becomes activated. Experimental

Pedigree Analysis Asci from X10-1B were treated with a 1:20 dilution of Glusulase (Endo Laboratories) for 5 min at room temperature. Dissection of asci and separation of sibling cells was performed on thin slabs of 4% YEPD agar using a de Fonbrune micromanipulator at magnification 380X. as described by Hicks and Herskowitz (1976). a-factor was supplied by a streak of (1 cells (strain 70). In experiments in which we wished to determine the mating type of a spore without exposing it to a-factor, we used the following procedure. Three spores from an ascus (of Xl O-l B) were allowed to germinate in the presence of a-factor; the fourth spore was allowed to germinate in the absence of a-factor. The mating type of the fourth spore was inferred assuming 2 a (a-factor-resistant):2 a (a-factor-sensitive) segregation.

Procedures

Yeast Strains X10-1B. MATa/MATa HO/HO HMa/HMa HhWHMol (Hicks and Herskowitz. 1978); 70. MATrr thr3-10 (F. Sherman): 227, MATa lysl (F. Sherman) were used. Media Cells were grown on rich medium containing 1% yeast extract, 2% peptone, 2% agar and 2% glucose (added after autoclaving). Cells were induced to sporulate on plates containing 1.5% potassium acetate, 0.25% yeast extract, 0.1% glucose and nutritional supplements.

We would like to thank Robert Horvitz and the members of our laboratory for many discussions; Jasper Rine and Jim Hicks for personal communications; and Kerrie Rine for the preparation of the figures. This work has been supported by a research career development award and a research grant from the NIH to I. H.. and a molecular biology training grant to the Institute of Molecular Biology. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received

November

14, 1978;

revised

March

15. 1979

References Biicking-Throm. E., Duntze. W.. Manney. T. R. and Hartwell. L. H. (1973). Reversible arrest of haploid yeast cells at the initiation of DNA synthesis by a diffusible sex factor. Exp. Cell Res. 76. 99-l 10. Duntze, W., MacKay, V. and Manney, T. R. (1970). Saccharomyces cerevisiae: a diffusible sex factor. Science 768. 1472-1473. Harashima. controlling 639-650.

S.. Nogi. Y. and Oshima. Y. (1974). The genetic system homothallism in Saccharomyces yeasts. Genetics 77,

Hartwell. L. H. and Unger. M. W. (1977). Unequal division in Saccharomyces cerevisiae and its implications for the control of cell division. J. Cell Biol. 75, 422-435. Hawthorne, D. C. (1963a). A deletion in yeast and its bearing structure of the mating type locus. Genetics 48. 1727-l 729.

on the

Hawthorne, D. C. (1963b). Directed mutation of the mating type alleles as an explanation of homothallism in yeast. Abstracts of the Proceedings of the Eleventh International Congress of Genetics 1, 34-35. Hicks, thesis,

J. B. (1975). Interconversion of mating University of Oregon, Eugene, Oregon.

types

in yeast.

Ph.D.

Hicks, J. B. and Herskowitz, I. (1976). Interconversion of yeast mating types. I. Direct observation of the action of the homothallism (HO) gene. Genetics 83. 245-258. Hicks, J. B. and Herskowitz, I. (1977). Interconversion of yeast mating types. II. Restoration of mating ability to sterile mutants in homothallic and heterothallic strains. Genetics 85. 373-393. Hicks, J. B., Strathern, J. N. and Herskowitz. I. (1977a). The cassette model of mating type interconversion. In DNA Insertion Elements, Plasmids. and Episomes. A. Bukhari. J. Shapiro and S. Adhya. eds. (New York: Cold Spring Harbor Laboratory), pp. 457-462. Hicks, J. B.. Strathern. J. N. and Herskowitz. I. (1977b). Interconversion of yeast mating types. Ill. Action of the homothallism (HO) gene in cells homozygous for the mating type locus. Genetics 85. 395405. Holliday. R. and Pugh, J. E. (1975). DNA modification mechanisms and gene activity during development. Science 787. 226-232. Klar. A. J. S. and Fogel. S. (1977). in Saccharomyces diploids during

The action of homothallism genes vegetative growth and the equiva-

Switching 381

Pattern

of Homothallic

lence of hma and /Ma

Yeast

loci functions.

Genetics

85, 407-418.

MacKay. V. L. and Manney, T. R. (1974). Mutations affecting sexual conjugation and related processes in Saccharomyces cerevisiae. II. Genetic analysts of nonmating mutants. Genetics 76. 273-288. Manney. T. R. and Meade. J. l-l. (1977). Cell-cell interactions during mating in Saccharomyces cerevisiae. In Microbial Interactions. 3. Receptors and R.ecognition. Series B, J. L. Reissig. ed. (London: Chapman and Hall), pp. 281-321. Oshima. Y. and Takano. I. (1971). Mating types in Saccharomyces: their convertibility and homothallism. Genetics 67. 327-335. Strathern, cerevisiae.

J. N. (1977). Ph.D. thesis,

Regulation of cell type in Saccharomyces University of Oregon, Eugene, Oregon.

Sulston, J. E. and Horvitz, H. R. (1977). Post-embryonic cell lineages of the nematode, Caenorhabditis elegans. Dev. Biol. 56, 11 O-l 58. Takano. I.. Kusumi. T. and Oshima. Y. (1973). An a mating-type allele insensitive to the mutagenic action of the homothallic gene system in Saccharomyces diastaticus. Mol. Gen. Genet. 126, 19-28. Wilkinson, L. E., and Pringle. J. R. (1974). Transient Gl arrest of S. cerevisiae cells of mating type LY by a factor produced by cells of mating type a. Exp. Cell Res. 89, 175-l 87. Note

Added

in Proof

It has been brought to our attention by Urszula Hibner and Magdalene So that the series of switches from one antigenic type to another exhibited by pathogens such as trypanosomes and Borrelia can also be explained by a sequential cassette insertion mechanism.