Uniparental Inheritance of Mitochondrial Genes in Yeast: Dependence on Input Bias of Mitochondrial Dna and Preliminary Investigations of the Mechanism

UNIPARENTAL INHERITANCE OF MITOCHONDRIAL GENES IN YEAST: DEPENDENCE ON INPUT BIAS OF MITOCHONDRIAL DNA AND PRELIMINARY INVESTIGATIONS OF THE MECHANISM

C. WILLIAM BIRICY, JR.*t, CATHERINE A. DEMKO*, PHILIP S. PERLMAN*t, AND ROBERT STRAUSBERwt The Ohio Stale University, Columbus, Ohio 43210

Manuscript received September 26, 1977 Revised copy received March 17, 1978

ABSTRACT

In Saccharomyces cerevisiae, previous studies on the inheritance of mitochondrial genes controlling antibiotic resistance have shown that some crosses produce a substantial number of uniparental zygotes, which transmit to their diploid progeny mitochondrial alleles from only one parent. In this paper, we show that uniparental zygotes are formed especially when one parent (majority parent) contributes substantially more mitochondrial DNA molecules to the zygote than does the other (minority) parent. Cellular contents of mitochondrial DNA (mtDNA) are increased in these experiments by treatment with cycloheximide, alpha-factor, or the uvsp5 nuclear mutation. In such a biased cross, some zygotes are uniparental for mitochondrial alleles from the majority parent, and the frequency of such zygotes increases with increasing bias. In two- and three-factor crosses, the cupl, ery1, and oli1 loci behave coordinately, rather than independently; minority markers tend to be transmitted or lost as a unit, suggesting that the uniparental mechanism acts on entire mtDNA molecules rather than on individual loci. This rules out the possibility that uniparental inheritance can be explained by the conversion of minority markers to the majority alleles during recombination. Exceptions to the coordinate behavior of different loci can be explained by marker rescue via recombination. Uniparental inheritance is largely independent of the position of buds on the zygote. We conclude that it is due to the failure of minority markers to replicate in some zygotes, possibly involving the rapid enzymatic destruction of such markers. We have considered two general classes of mechanisms: (1) random selection of molecules for replication, as for example by competition for replicating sites on a membrane; and (2) differential marking of mtDNA molecules in the two parents, possibly by modification enzymes, followed by a mechanism that “counts” molecules and replicates only the majority type. These classes of models are distinguished genetically by the fact that the first predicts that the output frequency of a given allele among the progeny of a large number of zygotes will approximately equal the average input frequency of that allele, while the second class predicts that any

* Department of Genetics. t Developmental Biology Program. $ Present address: Department of Biochemistry, University of Texas, Health Sciences Center at Dallas, Dallas, Texas 75235.

Genetics 89: 015-651 August, 1078 616 c. WILLIAM BIRKY, JR. et al. input bias will be aniplified in the output. The data suggest that bias amplification does occur. We hypothesize that maternal inheritance of mitochondrial or chloroplast genes in many organisms may depend upon a biased input of organelle DNA molecules, which usually favors the maternal parent, followed by failure of the minority (paternal) molecules to replicate in many or all zygotes.

ITOCHONDRIAL and chloroplast genes are transmitted to progeny pre- Mdominantly or entirely by only one parent in a number of different organisms. This uniparental inheritance is generally through the maternal parent in organisms with oogamy, and in such cases may be attributable to the male gamete containing many fewer organelles than the egg, or to failure of the organelles from the male gamete to enter the egg (see BIRKY1976 for references). But in the geranium (Pelargonium), many crosses involving chloroplast mutants produce three classes of zygotes: those transmitting only maternal markers, those transmitting only paternal markers, and those transmitting both to vary- ing extents (TILNEY-BASSETT1975). In the algae Chlamydomonas reinhardtii (SAGER1975; GILLHAM1974) and C. eugametos (MCBRIDEand McBRIDE 1975), chloroplast markers are transmitted principally from one parent, but in both of these isogamous species the zygote receives the entire chloroplast of both parent gamctes. In these cases, it is clear that the failure of uniparental zygotes to transmit organelle markers from one parent must be due to the failure of those markers to replicate in the zygote and to their loss by dilution or enzymatic degradation, or to the conversion of the markers to the corresponding alleles from the other parent. At first sight, it may appear that mitochondrial genes in baker’s yeast (Sac- charomyces cereuisiae) are always inherited biparentally, since an examination of the mixed progeny of a large number of zygotes (random diploid analysis) always shows that mitochondrial genes from both parents can be recovered among the progeny (LINNANEet al. 1968; THOMASand WILKIE1968; COEN et al. 1970). When the progeny of individual zygotes were analyzed separately (zygote clone analysis), some zygotes were found that transmitted to their progeny mitochondrial genes from only one parent (COENet al. 1970; RANK and BECH-HANSEN1972; WILKIE1972; WILKIEarid THOMAS1973; WAXMAN, EATONand WILKIE1973; CALLEN1974; LINNANE,HOWELL and LUKINS1974). The number of such zygotes, however, was usually small, and they might have resulted from matings between spontaneous petite mutant cells that had lost the mitochondrial genes in question and a wild-type cell that retained them. BIRKY (1974, 1975a,b,c) first called attention to these zygotes and argued that they indicated uniparental inheritance analogous to that seen in other organisms, after finding that some yeast matings produced uniparental zygotes in substantial proportions: too many to be accounted for by petites, by sampling or scoring errors, or by zygotes that produce a single bud pure for one genotype and then die. Neither is it possible to explain uniparental zygotes by the transmission of only a very small number of identical genomes from the zygote to its buds; INHERITANCE OF YEAST MITOCHONDRIAL GENES 61 7 genetic and physical evidence argue against this (CALLEN1974; BIRKY1975c; DUJONand SLONIMSKI1976; BIRKYet al. 1977; SENA,WELCH and FOGEL1976). We now report further studies investigating the circumstances in which uniparental inheritance occurs and examining possible mechanisms. These studies show, for several different crosses, that uniparental zygotes are consistently produced when the two parents contribute different numbers of mtDNA molecules to the zygote (biased input) ; most uniparental zygotes transmit only mitochondrial alleles from the majority parent. Increasing input bias in a cross results in increasing frequencies of uniparental zygotes, whether the input bias is induced by treatment with cycloheximide or the mating hormone alpha-factor or by a nuclear gene. In a two-factor or three-factor cross, a given zygote may show uniparental inheritance of mitochondrial genes at only one locus, or at two or three loci. The most common class of uniparental zygotes are those uniparental at all loci studied. A possible explanation for uniparental zygotes is random loss and fixa- tion of alleles due to multiple rounds of random mating and gene conversion. We have ruled this out as a sufficient explanation by showing that the capl, eryl, and olil loci tend to become uniparental as a unit, although they are probably rarely or never included in the same gene conversion event, and that uniparental inheritance is not found in higher frequency among zygotes having a greater opportunity for recombination. Mitochondrial DNA (mtDNA) molecules are singled out for transmission or loss soon after the formation of the zygote; never- theless, markers may apparently be rescued by recombination from genomes destined to be lost. The effect of input bias may be explained by two different types of models: those invoking random selection of a small sample of mtDNA molecules for replication, and those in which the zygote differentially marks and “counts” mtDNA molecules from the two parents and replicates the majority type. Preliminary data relating input and output ratios of mitochondrial alleles favor the latter kind of model. Some of our data and conclusions were summarized earlier by PERLMANet al. (1976).

MATERIALS AND METHODS Stocks The stocks, their sources, and their genotypes are listed in Table I.For the sake of brevity we have used the symbols CR and Cs, ER and ES, OR and Os, PR and PS as abbreviations for the mitochondrial allele pairs caplr and capla, erylr and eryls, 0127 and OMs, par17 and par18 All crosses were homopolar (U+ x U+ or U- x U-), so that the phenomenon of polarity (DUJON, SLONIMSKIand WEILL1974) is not involved in any of our experiments or interpretations. Media YEPD, YEPG, and YEPGal contained 1% Difco yeast extract plus 2% Difco proteose pep- tone, and 2% dextrose (glucose), 4% glycerol, or 2% galactose, respectively. (All concentrations are w/v.) Comparable results were obtained with the semisynthetic media RD and RG, which contain 0.5% yeast extract, 0.1% (NH,)$O,, 0.1% KH,PO,, 0.05% MgSO,, and 0.05% NaCl at pH 6.5, plus carbon sources as above. These media were sometimes supplemented with amino TABLE 1 Strains used

Genotype Strain Nuclear w C EO P Abbreviation Source ~___~ NI23 a his ss NIB E. MOUSTACCHI NI23 UVSrho5 a his UVSp5 - ss P5 E. MOUSTACCHI D6 a arg met + ss D6 D. Y. THOMAS 2-3b a irp-I + RS 2-36 R. KLEESE 4810 a lys-I + ss 4810 R. KLEESE 48 10 diploid a/a lys-l/lys-I + ss 4810 diploid Diploidization of 4810 induced by cycloheximide n IL458-1A a: his + R32 1 R221 S S IL458 P. P. SLONIMSKI DPI-1 B/517 a his trp + R517 ss S DPI P. P. SLONIMSKI IL126-1 B a his R32l R221 S S IL126 P. P. SLONIMSKI 41-1 a: ade R321 R514 R4 R 41-1 C. A. DEMKO(spore from 4810 x ID4/1) ID4/1 a a& trp R32l R514 R4 S ID4/1 R. YOUNG(spore from (32)l-2/3 x DPl-lB/514) D243-4A-OR a ade lys + S S R4 S D243 R. CRIDDLE OEC1222 a ade-I his-I ihr-l + RI R3 RI S OEC K. SUDA 1-1/1 a ade-l - R321 S R4 S 1-1/1 P. S. PERLMAN(spore from D243 x IL126) 6-2/5 a: ade his + R32 1 S R4 S 6-2/5 P. S. PERLMAN(spore from IL468 x D243) LT70 a lys + R SRR LT70 L. TREAT(spore from 4810 x 6-2/5) 4'120E a leu + S RS S 4120 J. FORSTER ID41-6/152 a ade lys - R SRR COP/152 C. A. DEMKO(spore from 1-1/1 x 4810) ID41-6/19 a: ade lys R SRR COP/19 C. A. DEMKO(spore from l-l/l x 4810)

* C, E, 0, and P are abbreviations for the chll, eryl, olil, and par1 loci. INHERITANCE OF YEAST MITOCHONDRIAL GENES 619 acids, adenine, or uracil, depending on the auxotrophic requirements of the strains being studied. MMD, MMG, and MMGal are minimal media after WICKERHAM(1946) but with 1 mg CaCl,/l. Antibiotics dissolved in methanol or ethanol were added to agar media (2% agar) at about 50”, as follows: chloramphenicol, 3mg/ml; erythromycin, 1 mg/ml; oligomycin, 2 pg/ml; paromomycin, 3 mg/ml (dissolved in water). All experiments were done at 30”. Cell growth and mating Glucose was the carbon source, unless otherwise indicated. Cells were grown to exponential phase in liquid or on plates, harvested, and mated two to three hours in the same medium with shaking. The mating mixture was usually begun with about IO7 cells of each mating type per ml. The mating mixture, containing zygotes that have not yet released their first buds and also unmated haploid parents, was plated on minimal medium with the same carbon source (prototroph selection plates), so that only the prototrophic zygotes and their diploid progeny could form colonies. In some experiments, cells were subjected before mating to one or more of the following treatments that have been reported to increase the mtDNA content of yeast: (1) the a parent culture was incubated for 90 minutes with the purified mating hormone “alpha-factor’’ (supplied by MICHAELDOUGLAS and DAVIDFINKLESTEIN), at a concentration that reduces the number of budded cells by about 50%; (2) cells were incubated with cycloheximide at 50 jtg/ml for three hours; (3) histidine-requiring cells were starved for histidine by incubation for three hours in minimal medium. All three treatments inhibit nuclear DNA synthesis and cell division, but permit mitochondrial DNA synthesis to continue (CRYERet al. 1973; GROSSMAN,GOLDRING and MARMUR1969). Mitochondrial DNA synthesis was inhibited in some experiments by incubating the parent culture(s) or the mating mixture with hydroxyurea at 25 mg/ml, or by incubating adenine-requiring strains for three hours in minimal medium followed by mating in minimal medium containing all auxotrophic requirements except adenine. Alpha-factor, cycloheximide, or hydroxyurea were removed by washing cells three times in water before mating. The effects of these treatments on the output of mitochondrial crosses are described by DEMKO(1975) and DEMKOand PERLMAN(in preparation), and will be characterized further in this paper. Extraction and density-gradient analysis of labelled DNA In experiments designed to verify the effects of alpha-factor and hydroxyurea on the synthesis of nuclear and mitochondrial DNA, cells were labelled for 150 minutes with 3H-adenine at 7 ,pCi/ml. Both mitochondrial and nuclear DNA are labelled. To provide a marker for mitochondrial DNA, cells of stock p5 were pre-incubated with cycloheximide (50 pg/ml) for 30 minutes, then labelled with 14C-adenine at 0.5 pCi/ml for 150 minutes in the presence of cycloheximide. Only mitochondrial DNA is labelled. Samples of these cells were added to other batches of “-labelled cells to provide *4C-labelled mitochondrial DNA as a marker in gradients. The mixed cells were converted to spheroplasts, and CsCl gradients were prepared and analyzed as described by PERLMANand MAHLER(1 971). Output ratios (a) Random diploid analysis: To determine the output ratios of mitochondrial genes from the “average” zygote, all cells were washed off prototroph selection plates containing colonies from several hundred or more zygotes. The cell suspension, in lOm~EDTA, was diluted and about 100 cells were plated on several MMD plates to make replica masters. After incubation, these plates were replica-plated onto YEPG or RG plus appropriate antibiotics to determine the mitochondrial genotypes of the colonies, and hence of the diploid cells washed off the prototroph selection plates. It is important to note that the original zygotes had undergone about 20 generations on the prototroph selection plates, so that most diploid cells were homoplasmic (pure) for one mitochondria2 genotype or another as a result of vegetative segregation of mitochondrial DNA molecules. (b) Zygore clone analysis: To determine the output ratios of mitochondrial genes from 620 c.WILLIAM BIRKY, JR. et al. individual zygotes, zygote clones were picked off prototroph selection plates and suspended separately in EDTA. Several replica master plates were then prepared for each zygote clone, incubated, and replica-plated as above. In either method, output ratios are expressed as the percent diploid cells having each of the possible parental and recombinant mitochondrial genotypes. The transmission of a mitochondrial allele is the percent of diploid cells expressing that particular allele; it is the total of the percentages of all genotypes that include the allele. The coordinate transmission seen for different loci in most of these crosses suggests that our results are not seriously affected by differential selection. Uniparental zygote identification methods (a) Zygote clone analysis: In a zygote clone analysis as described above, a variable number (usually several hundred) of cells from each zygote clone were individually scored; if all cells from a zygote clone carried the mitochondrial allele contributed by a particular parent, the clone was scored as uniparental for that locus. Clones pure for either the antibiotic-resistant or the antibiotic-sensitive allele could be identified in this way; i.e., clones uniparental for alleles from either parent. However, the method is relatively laborious and the sample size is consequently small; thus it is not possible to detect alleles transmitted by a zygote to less than about 1% of its progeny. More precisely, the 95% confidence interval for identification of a uniparental zygote may be estimated by 3.4/N, where N is the sample size; the range of N is about 100-500, so that this method may overlook 0.7-3.4% marker transmission. Consequently, a small fraction of the zygotes scored as uniparental may actually be biparental. (b) Dropping out: Samples of about IO4 to IO5 cells from each zygote clone were dropped directly onto antibiotic plates and scored for growth/no growth after incubation. No growth indicates that the zygote was uniparental for the antibiotic-sensitive allele, having transmitted the resistant allele to fewer than 0.01% to 0.001% of its progeny. Growth on the antibiotic plate means that the zygote was either biparental or was uniparental but transmitted only the antibiotic-resistance allele to its progeny. This technique can thus identify only one of the two possible classes of zygotes uniparental at a given locus, i.e., those uniparental for the antibiotic- sensitive allele. But for this class, the method provides a much more rigorous criterion of uniparental inheritance than does zygote clone analysis. (c) Replica plating: Individual zygote clones were picked from the prototroph selection plates and patched onto an MMD plate. After incubation overnight to increase the size of the colonies, these master plates were replica-plated onto antibiotic plates and scored for growth/ no growth. As with the dropping out test, this test permits the identification only of zygotes uniparental for sensitive alleles, but it is faster than either of the above methods and permits more zygotes to be scored from a given crsss. Tests showed that about 7.4 x IO4 to 3.8 X IO5 cells were transferred from a velvet to an antibiotic plate during replica plating, so that this technique should identify as uniparental those zygotes transmitting the resistant allele to less than about 11105 of their progeny. Nearly all of the frequencies of uniparental zygotes given in this paper were obtained by method (b) or (c). Two sources of nonrandom error are present in all three methods. First, parent cultures may contain mitochondrial petite (p-) mutants which have deletions of mitochondrial genes. A mating of such a petite by a wild-type cell (p+) will automatically produce a zygote that is uniparental for any genes in the p+ parent whose alleles are deleted in the P- parent. The frequency of p- cells in each parent culture is routinely determined by plating on YEPD a sample of several hundred cells taken from the cultures just prior to mating. After incubation, the colonies are overlaid with 1.5% agar containing buffered triphenyl tetrazolium chloride (OGUR,ST. JOHN and NAGAI1957); colonies formed by p- cells fail to stain, while pf colonies stain red. In most experiments reported here, the frequency of p- cells was too small to account for more than a few percent of uniparental zygotes, even if one makes the excessively conservative assumption that all P- cells are missing the loci being studied. The second source of error is that haploid cells, not growing but still viable, may be picked up off the prototroph INHERITANCE OF YEAST MITOCHONDRIAL GENES 621 selection plates along with the zygote clones. A few such cells from the resistant parent will then grow on antibiotic plates after dropping out or replica plating; if they are included in a zygote uniparental for the sensitive allele, that clone will be incorrectly scored as biparental. Such events are rare, and most of the data are not corrected for them. They tend to make our estimates of the frequency of uniparental zygotes more conservative. Apart from these general sources of error, clones can be misclassified by the replica plating test because (1) variations in the number of cells deposited from the velvet onto the drug plate can result in failure to detect rare resistant cells, and (2) occasional large clumps of sensitive cells deposited on the plate can be scored as resistant even though they are not growing. TO obtain an estimate of the frequency of niisclassification, zygote clones were first scored by replica plating and then scored again by the dropping out method. The frequency of resistant clones

sample 1 5 100 1 90

80 transmission

60 cn 50 8 40

30 uniparental 20

0 0 1 2 3 4 Time (hours) FIGURE1 .-Uniparental zygotes are not simply homoplastic first buds or their mothers. Cells of stock p5 were mated with stock ID4/1, with (open circles) and without (solid circles) prior treatment of p5 with alpha-factor. Aliquots of the mating mixture were plated on minimal medium after 0, 120, 150, and 210 minutes and colonies from these plates were used for random diploid analysis to determine the transmission of the Cs allele or for determination of the percent zygotes uniparental for the CY allele by the replica-plating method. Samples were removed from the mating mixture at five different times and examined microscopically for the presence of zygotes with and without large buds ready to separate from the mother. These samples showed: (1) no zygotes; (2) 0/17 zygotes with large buds; (3) 7/74 zygotes with large buds; (4) 46/101 zygotes with large buds; and (5) most zygotes with large buds, and some with second-generation buds. 622 c. WILLIAM BIRKY, JR. et al. (biparental or uniparental for the resistant allele) misclassified as uniparental sensitive was 6.5% and 18.8% for two different persons; the same two workers misclassified uniparental sensitive clones as resistant in 8.1% and 13.2% of the cases. A sample of zygote clones was retested by the dropping out procedure and the results were compared to those of the first dropping out test. This experiment showed that the dropping out test is very reproducible; most remaining errors were for clones that were nearly pure for drug sensitivity (a few resistant cells/l05 cells in the first or second analysis). Because uniparental sensitive clones are a minority of the total, these errors will cause us to overestimate their frequency. But estimates of the differencein the frequency of uniparental zygotes between two different matings will be slightly too low. Homoplasmic buds are rarely mistaken for uniparental zygotes Operationally, a uniparental zygote is defined as a colony, taken from a prototroph selection plate, which is pure for a mitochondrial allele from one parent. When a biparental zygote produces its first bud, that bud is also very often pure for one allele, due to vegetative segregation; mother cells are less often pure (see BIRKYet al. 1977 and PERLMAN,BIRKY and STRAUSBERG 1978 for references). It can thus be imagined that some colonies scored as uniparental zygotes actually arise, not from intact zygotes, but from buds that were formed and released from their biparental mother cells before the mating mixture was plated on minimal medium. To minimize this source of error, most matings were terminated after two hours, at which time first buds are rarely large enough to be released from the mother. In other cases, samples of the mating mixture were examined microscopically at intervals and the mixture was plated before significant numbers of large buds were seen (generally after about 150 minutes of mating). To demonstrate that this procedure was effective, a cross was done as described in the legend of Figure 1, with aliquots of the mating mixture being plated at intervals before and after large buds appeared. The frequency of “uniparental zygotes” remained essentially constant in samples plated after 0, 120, and 150 minutes of mating, then rose significantly at 210 minutes (zygotes formed after “0 minutes of mating” are actually due to mating events taking place on the prototroph selection plates). The rise at 210 minutes can be attributed to the release from biparental zygotes of pure first buds, which were scored as uniparental zygotes; such large buds were first seen in the mixture at 120 minutes, and increased in frequency until most zygotes at 190 minutes had buds ready for release. Determining zygoie budding patierns Mating mixtures containing unbudded zygotes were washed in 1 mM EDTA and sonicated to break up cell clumps, then streaked cn agar slabs. Individual zygotes were moved to marked positions with a micromanipulator and examined frequently. As each bud appeared, its position on the zygote was recorded. For example, C E, E, means the first bud came from the central neck of the zygote, the second from one end of the zygote, and the third bud came from the other end. It is not known which parent haploid cell gave rise to which end of the zygote.

RESULTS AND DISCUSSION The bias effect Consider a generalized two-factor cross which may be symbolized a A“ B“ X a Ao,Ba, where A and B are two mitochondrial gene loci (e.g., A” and Aamight be the Cs and C” alleles, respectively). Assume that the cross is homopolar, or that A and B are nonpolar alleles in a heteropolar cross. According to the phage- analogy models of yeast mitochondrial genetics (especially DUJON,SLONIM- SKI and WEILL1974; reviewed by BIRKY 1975c), the output frequencies (transmission) of alleles from the two parents reflect the input frequencies of INHERITANCE OF YEAST MITOCHONDRIAL GENES 623 functional mtDNA molecules in the zygote. For example, an unbiased cross (yielding roughly 50% A", B", Am and Bm diploids) would be one in which the a and a: parents contributed approximately equal numbers of mtDNA molecules to the zygote, while a biased cross (e.g., 67% A" and B", 33% Aa and Bm)would be one in which the a parent contributed more mtDNA molecules than the a: parent, or vice versa (in the example, an input ratio of about 2:l would be suspected). There is substantial evidence, direct and indirect, for a positive correlation between input ratios of mtDNA molecules and output ratios of mitochondrial alleles (DUJONet al. 1974,1975; GOLDTHWAITE,CRYER and MARMUR 1974; GUNGE1975; DEMKO1975; PERLMANand DEMKO1974). BIRKY (1974, 1975a,b) showed that the cross p5 a his CsOsx 1-1/1 a: ade CRORproduced a high frequency of zygotes uniparental for alleles from the p5 parent, and random diploids with a strong output bias favoring p5, when the cells were grown and mated in glycerol; with glucoses as a carbon source, the cross was less biased in favor of p5, and there were few zygotes uniparental for p5, markers. Glycerol- grown cells mate poorly, and many of the observed zygotes may have been produced after the cells were plated on minimal medium where the p5 parent was starved for histidine and the 1-1/1 parent was starved for adenine. Histi- dine starvation would be expected to increase the mtDNA content of p5 cells by inhibiting cell division while mtDNA synthesis continues (GROSSMAN, GOLD- RING and MARMUR1969), and adenine starvation would decrease that of 1-1/1 by preventing the premating synthesis of mtDNA induced by mating hormones. To test the possibility that biased inputs might be responsible for the production of uniparental zygotes, the mtDNA content of the a parent was deliberately increased in several crosses by treating the a cells with alpha-factor prior to mating. Alpha-factor arrests cells of mating type a in the G1 phase of the cell cycle in preparation for mating; it inhibits nuclear DNA synthesis and cell division but mitochondrial DNA synthesis continues (CRYERet al. 1973), so that treated cells increase their content of mtDNA relative to untreated GI cells (which can also mate). In some cases, hydroxyurea was added simultaneously with the alpha-factor to inhibit mtDNA synthesis, providing an additional control. To see if alpha-factor and hydroxyurea were working as expected in our experiments, the incorporation of labeled adenine into mitochondrial and nuclear DNA was followed in cells of stocks NI23 and p5 from experiment UP12. Ninety minutes after the addition of alpha-factor or hydroxyurea to the cells, aliquots were labelled with 3H-adenine (labelling nuclear and mitochondrial DNA) or with 14C-adeninein the presence of cycloheximide (labelling mtDNA only), as described in MATERIALS AND METHODS. DNA from these cells werc? analyzed in CsCl gradients; the results are shown in Figure 2. Untreated control cells showed incorporation of adenine into both nuclear and mitochondrial DNA, or into mitochondrial DNA only when p5 was labelled in the presence of cycloheximide (Figure 2A). The peak of 14C counts serves as a marker for mtDNA in the gradient. Cells treated with alpha-factor (Figure 2B) showed incorporation into mtDNA only. These results confirm that alpha-factor, like 624 c.WILLIAM BIRKY, JR. et al.

100 0 A B 90 8 9 !I4 80 8

70 7

60 6

50 5

40 0I - 30 *T - 3- c) c) 0 0 20 2- Y E 10 la U U U I -0 ob, 5 D 5

4 4

3 3

2 2

1 1

0 C 10 20 30 40 0 10 20 30 40 Fraction number

FIGURE2.-CsCl gradient profiles of whole-cell DNA. Cells of p5 were labelled with 14C-adenine in the presence of 50 pg cyclo/ml for 150 minutes. An aliquot of those cells was added to cells labelled with "-adenine under the following conditions: (A) no treatment; (B) labelled in the presence of alpha-factor; (C) labelled in the presence of alpha-factor plus hydroxyurea (25 mg/ml); (D) labelled in the presence of hydroxyurea (25 mg/ml). The cells used in this experiment were taken from the same cultures used for the genetic analyses presented in Table 2 and similar data were obtained in the parallel experiment with strain NI23 (in place of p5). The cell mixtures were prepared for CsCl gradient analysis as described in MATERIALS AND METHODS. When the data are normalized to the number of 1% counts recovered (each cell mixture had the same number of 14C-labelled cells before the spheroplast procedure) the percent of control (A) mtDNA synthesis was calculated: B, 71%; C, 5.1%; D, 5.2%. Nuclear DNA synthesis was almost quantitatively inhibited by all treatments (including cyrlo) . 14C-data are unbroken lines and 3H-data are broken lines. INHERITANCE OF YEAST MITOCHONDRIAL GENES 625 cycloheximide, permits mitochondrial DNA synthesis to continue while inhibiting nuclear DNA synthesis and cell division. Hydroxyurea inhibited incorporation of adenine into nuclear DNA nearly completely, but only partially inhibited incorporation into mitochondrial DNA (Figure 2D). Cells treated with hydroxyurea and alpha-factor together (Figure 2C) had a labelling pattern like that of cells treated with hydroxyurea alone. Assuming that the extent of adenine incorporation into DNA is a measure of net DNA synthesis, some predictions can be made. First, treatment of cells with alpha-factor, which blocks nuclear DNA synthesis but permits mitochondrial DNA synthesis to continue, should result in an increase in the amount of mtDNA per cell. Second, since hydroxyurea blocks cell division and nuclear DNA synthesis completely but only partially inhibits mitochondria DNA synthesis in this strain, it should result in a smaller increase in the amount of mitochondrial DNA per cell, either alone or in conjunction with alpha-factor treatment. In a cross, we would expect to see an increase in the transmission of mitochondrial genes, i.e., an increased bias, from treated p5 cells, larger for cells treated with alpha-factor alone, smaller for cells treated with hydroxyurea or alpha-factor and hydroxyurea. Some further insights into the effects of hydroxyurea treatment were obtained (data not shown). It was found that the bias of the cross p5 x ID4/1 increases progressively with increased time of incubation in hydroxyurea (up to two hours) ; even after only 45 minutes of treatment, the output was distorted somewhat in favor of markers from the treated parent. This result is consistent with there being some continued synthesis of mtDNA in drug-treated cells, using endo- genous deoxynucleoside triphosphates. In these experiments hydroxyurea was shown to inhibit exogenous adenine incorporation into nuclear and mitochondrial DNA even during a 30-minute pulse of label following the addition of the drug; inhibition remained high during subsequent pulses at 45, 90, and 120 minutes of hydroxyurea treatment. When hydroxyurea is removed and the cells incubated in drug-free medium for two hours prior to mating, the bias decreased somewhat and both nuclear and mitochondrial DNA synthesis (using exogenous adenine) had resumed. Thus, even though hydroxyurea is not completely effective, it does appear to reduce the accumulation of mtDNA in cells treated with alpha-factor relative to that occurring in cells treated with alpha-factor alone and does show a partial inhibition of the genetic effects of alpha-factor treatment. Table 2 gives the effects of alpha-factor and hydroxyurea on both transmission and uniparental inheritance. In every experiment, treatment with alpha- factor increased the transmission of mitochondrial genes from the a parent, as previously demonstrated by DEMKO(1975), PERLMANand DEMKO(1974) and PERLMANet al. (1976). The increase was coordinate for the Cs,Es, Os, and Ps genes; the effects on transmission can be most easily seen by looking at the average transmission for these four loci. Zygotes that were uniparental for one or more of these genes were produced in substantial numbers in every cross where the transmission was biased in favor of the sensitive alleles being scored. TABLE 2 Effect, on transmission and uniparental inheritance, of treating parents with alpha-factor (A) and/or hydroxyurea (H), or cycloheximide (CY) m

Cross Percent transmission Percent uniparental zygotesf in

a parent ~~ Experiment aXa treatment CS ES OS PS XN c5 ES 05 PS XN UP12 N123 ID4/1 - 25.5 25.4 22.9 24.6 654 2.8 2.3 2.5 - 2.5 393 CXESOS CREROR A 38.1' 37.8' 33.3* 36.4 378 4.3 4.0 5.0 - 4.4 400 H 34.4* 34.4* 31.0' 33.3 552 1.3 1.3 2.8 - 1.8 386 AH 30.5 28.6 27.5 28.9 604 1.7 1.1 1.9 - 1.6 360 - 61.4 63.3 60.8 61.8 521 9.3 10.3 12.1 - 10.6 397 A 74.1* 74.9' 76.5' 75.2 459 23.0' 24.3 * 26.8' - 24.7 400 H 63.7 63.9 63.0 63.5 606 16.6* 18.6' 16.1 - 17.1 398 AH 66.9 64.9 61.1 64.3 583 15.9' 17.5* 18.5 - 17.3 395 NI23 COP19 - 14.9 - 16.2 14.9 328 3 - 22 2.3 100 n CSOSPS CRORPR A 23.4' - 19.2 23.0 380 1 - 1 1 1.o 100 3 H 17.6 - 17.4 19.9 597 0 - 1 1 0.7 100 E AH 16.0 - 12.3 16.1 45 7 - - - - - 100 - 51.5 - 58.2 54.4 361 5 - 544.7 100 $ A 71.8* - 67.7 67.1 291 15 - 19' 11 15.0 loo H 68.2' - 64.7 66.9 286 3 - 6 7 5.3 100 w -4H 58.1 - 61.7 61.8 227 7 - 11 7 6.3 100 9 -- - - - CY - 16 15 13 14.7 100 F DEMKO(1975) N123 1-1/1 A 83.7 - 84.0* 83.9 ? 34.4* - - - 34.4 96 m csos CROR AH 66.8 - 69.0 67.9 ? 6.6 - - - 6.6 91 p5 1-1/1 - 83.7 - 85.6 84.7 ? 9.9 - - - 9.9 141 g @OS CROR A 93.7 - 95.4' 94.6 ? 46.2' - - - 4.6.2 52 H 86.3 - 88.0 87.2 ? 7.9 - - - 7.9 114 AH 89.0 - 90.8 89.9 ? 15.0 - - - 15.0 147 ID4/1 - 38.8 41.2 44.8 41.6 502 3.1 2.7 2.7 - 2.8 521 CREEOR A 56.3* 57.3' 59.5' 57.7 597 12.0* 9.9' 9.0* - 10.3 424 ID4/1 - 61.7 61.0 61.7 61.5 397 8.4 7.1 8.4 - 8.0 438 CREEOR A 74.9* 75.6* 75.6* 75.4 4-46 34.7' 35.2* 32.6' - 34.2 196 ID4/1 - 67.7 68.2 68.7 68.2 534 21.7 19.5 18.4 - 19.9 609 CREROR A 83.6* 83.8' 83.2' 83.5 535 41.7* 37.9' 38.3' - 39.3 758 41-1 - 78.8 77.4 - 78.1 43 0 7.7 6.4 7.4 - 7.2 608 CREROR A 81.8 81.3 - 81.6 386 16.6' 14.4* 17.1' - 16.0 181

N - Sample size. * Significantly different from controls at 5% level. f Determined by replica-plating entire clones. INHERITANCE OF YEAST MITOCHONDRIAL GENES 627 And in every case, the frequency of such zygotes was increased by alpha-factor treatment. As expected from the labelling experiments, hydroxyurea alone consistently increased both the transmission bias and frequency of uniparental zygotes, but the effect was usually not statistically significant and was always smaller than that of alpha-factor. Hydroxyurea plus alpha-factor produced results similar to hydroxyurea alone, as expected from the incomplete blockage of mtDNA synthesis by hydroxyurea. The stock p5 is a mutant derived from stock N123; it is characterized by enhanced sensitivity to petite induction by UV (MOUSTACCHI,PERLMAN and MAHLER1976). It also shows a marked increase in transmission of mitochondrial genes relative to N123 (FRAENKEL1974; MOUSTACCHI,PERLMAN and MAHLER1976), and an increased frequency of uniparental zygotes ( BIRKY 1975b; DEMKO1975). These effects are verified and illustrated in Table 2 by crosses of N123 and p5 by strains ID4/1, COP19, and 1-1/1. The data are all consistent with p5 having a higher mtDNA content than N123, and this is verified by direct measurements showing that p5 has about twice as much mtDNA as N123 (MOUSTACCHI,PERLMAN and MAHLER1976). The correlation between input bias and the frequency of uniparental zygotes produced in a cross can be visualized by plotting the percent zygotes uniparental for a particular allele versus the transmission of that allele; Figure 3 summarizes, in this manner, the data from all crosses done in the course of this study. In general, crosses produced very few zygotes uniparental for an allele unless

m m 1 Y

30 L m

E =I 20 8 0. .' . 10 - .. . .: ... : ** e... **

% Tranamisolon FIGURE3.-Percent zygotes uniparental for markers from NI23 or p5 plotted against percent transmission for the same markers. Where more than one locus was studied in a cross, the plotted data point is the average for all loci. C. WlLLIAM BIRKY, JR. et d. TABLE 3 Correspondence between uniparental zygote frequency and transmission or genome ratio in crosses of N123 or pS(aCS) x l-l/l(aCR)

Percent zygotes a parent Percent CS Genome uniparental a parent treatment from a rativ for CS N P5 a 93.7 14.9 46.2 52 P5 cy + cyclo 90.1 9.1 23.2 125 P5 a+m 89.0 8.1 15.0 147 P5 HU 86.3 6.3 7.9 114 NI23 a 83.7 5.1 34.4 96 P5 83.7 5.1 9.9 141 P5 cyclo + HU 81.5 4.4 8.1 37 N123 a + cyclo 80.1 4.2 16.2 124 P5 cyclo 79.4 3.9 15.6 84 P5 * 73.8 2.8 6.5 1(48 P5 starved 73.6 2.8 5.9 102 P5 t 71.8 2.6 5.1 99 N123 a+HU 66.8 2.0 6.6 91

Samples of a culture of N123 or p5 were treated with alpha-factor (a), cycloheximide (cyclo), and/or hydroxyurea (HU), or starved for histidine, as described in MATERIALS AND METHODS, then mated to 1-1/1. Genome ratio = % CS/(lOO-% CS). Data from DEMKO(1975). * 1-1/1 starved for adenine. + Mating in minimal medium plus adenine and histidine. the output was strongly biased (>> 50% transmission) for that same allele. The exceptions may be significant, and will be discussed in detail later. Table 3 further illustrates the correlation between bias in transmission frequencies and the frequency of uniparental zygotes. Aliquots of a culture of N123 and of a culture of p5 were treated with various combinations of alpha- factor, cycloheximide, and/or hydroxyurea, or starved for histidine, then mated to aliquots from a culture of 1-1/1. Except for these treatments, and for the uvsp5 mutation, all zygotes produced in these crosses had identical nuclear and mitochondrial genotypes and were in as nearly identical physiological conditions as was possible. (Some of the results are shown in more detail in Table 2.) In- creasing transmission frequencies of the Cs allele from N123 or p5 correspond to increasing frequencies o€ zygotes uniparental for this same allele. The bias in each cross can be seen most clearly when expressed as the genome ratio = % Cs / (100 - % Cs) , which is simply the ratio of Cs to CR alleles in the random diploid output. On this scale, an unbiased cross (50% Cs) would have a genome ratio of 1. We conclude that a biased input of mtDNA molecules may be a necessary and sufficient condition for the production of uniparental zygotes in most yeast crosses, and that in general the greater the input bias the greater the frequency of such zygotes. Generality of the phenomenon Most of the data in this and previous papers (BIRKY 1974, 1975a, 1975b) INHERITANCE OF YEAST MITOCHONDRIAL GENES 629 have been obtained with only a few different strains and crosses. However, we have found substantial numbers of uniparental zygotes, correlated with transmission bias and mitochondrial DNA input, in crosses involving a large number of different stocks (and hence different nuclear genetic backgrounds) obtained from many different laboratories. Complete data will be published later; here we show only a few selected crosses to indicate that the phenomenon is quite general (Table 4). Our present data are also concentrated on the cap I, ery I, and oZi-I loci, which collectively cover about 15-20% of the mitochondrial genome according to physical map data (NAGLEYet al. 1976; SCHWEYENet d. 1976; SANDERSet al. 1976; MORIMOTOet aZ1976). However, as demonstrated in Table 2, we have also found zygotes uniparental for the par1 locus, which is on the opposite side of the genome, and the correlation with input holds for this locus as well. Unfortunately, our par-I allele shows noncoordinate transmission in some crosses, possibly because of nuclear modifier genes conferring partial paromomycin resistance; such genes have been identified in several of our stocks (M. WAXMAN,unpublished observations). A complete study of this locus, and extension of studies of uniparental inheritance to other loci, is in progress. Uniparental zygotes have been seen in crosses using glucose, galactose, glycerol, acetate, or ethanol as carbon sources, and thus in both glucose-repressed and derepressed cells. In many instances, a high transmission bias was induced by treating the parent of mating type a with alpha-factor, or by treating one parent or the other with cycloheximide. But uniparental zygotes were also seen in crosses where these special treatments were not used. Although most or all crosses do show increasing frequencies of uniparental zygotes with increasing input bias, the relationship between bias and uniparental frequency may differ from cross to cross. Figure 4 compares the relationship between uniparental frequency and genome ratio for the cross p5 X 1-1/1 (data from Table 3) with the cross p5 X ID4/1 (data from Table 2 and unpublished). It is clear that these crosses differ, the one with ID411 being more respon- sive to increasing genome ratio than the one with 1-1/1. The difference may be a consequence of differences in the nuclear or mitochondrial genomes, or both. The data from both crosses show an approximately linear relationship between genome ratio and uniparental zygote frequency; this linearity is not predicted by any simple model of the mechanism of uniparental inheritance

TABLE 4 Additional crosses producing uniparental zygotes

Transmission Unptalzygotes Cross Allele % N N D243 OR x 2-3b 0s OS 49.5 178 13.6 88 4810 PR x 6-2/5 PS PS 63.7 171 32.0 1083 4810 ER diploid x OEC ES ES 70.7 338 7.8 293 4810 ER diploid x DPI ES ES 70.1 3658 6.0 368 4120E ER x LT70 Es ES 73.8 221 12.0 100 630 c. WILLIAM BIRKY, JR. et al. 6or T 50 c

Genome ratio (CyCR)

FIGURE4.-Percent zygotes uniparental for the CS allele from p5 plotted against the genome ratio (% Cs/% CR in the random diploids) ; comparison of the relationship between genome ratio and uniparental frequency using 1-1/1 (open symbols) and ID4/1 (solid symbols) as the

(Y parent. Bars indicate 95% confidence intervals. and may reflect the fact that we have not studied the entire range of possible genome ratios. What is the fate of the missing markers? Uniparental zygotes, like biparental zygotes, are formed by the fusion of two haploid yeast cells. Both parents carry mitochondrial DNA and a complete set of markers; any parent cells in which one or more markers were deleted would be detected as petite mutants. The first question to be answered is, “What hap- pened to the missing markers?” Two possibilities exist. First, they may not be replicated, and after ten to 20 generations would be carried on such a small fraction of all mtDNA molecules ( to as to be undetectable, even if all were present in a single cell. Note that this would require virtually no replication during the approximately 20 generations required to produce a zygotic clone of the size usually analyzed. Consider a zygote containing 100 mtDNA molecules of each of two genotypes, ARand AS.If the ASmolecules were not replicated, in ten generations the ratio of the two types of molecules would be approximately lo5AR : lo2AS, or about lo3: 1. It is of course possible that the missing molecules are enzymatically destroyed in the zygote and/or its early progeny; for the moment we will treat this as a variant of the replication failure hypothesis. The second possible fate of the missing markers is that they are in some fashion transformed into their alleles. The only known mechanism for such a transformation is gene conversion: a “mating” in the zygote between a molecule carrying an AR allele and another carrying an AS allele, followed by recombi- INHERITANCE OF YEAST MITOCHONDRIAL GENES 631 nation at the A locus, could result in the conversion of the ARallele to AS,or uice wrsa, so that both molecules will be ARor both As. Evidence that gene conversion occurs in yeast mitochondria has been obtained by WILLIAMSONand FEN- NELL (1974) and by VAN WINKLE-SWIFTand BIRKY (in preparation); also DUJONet al. (1974,1975) argue that there are multiple rounds of random mating between mtDNA molecules in the yeast zygote. BIRKYand SKAVARIL(1976) showed by means of computer simulations that gene frequencies would drift in the zygote and its progeny due to multiple rounds of random mating and gene conversion, eventually leading to the loss of one allele or the other, i.e.,to a uniparental zygote. Although the simulations suggested that this drift would be too slow to produce large frequencies of uniparental zygotes, this hypothesis could not be ruled out. Moreover, it was clear a priori, and was verified by the simulation, that uniparental zygotes would be produced most often when the input of mtDNA molecules was highly biased; in such cases the loss of the minority allele occurs more rapidly and might conceivably be completed before the allele could be “rescued” by being segregated into a pure population in a bud. A test of this hypothesis depends upon the fact that gene conversion results from the repair of mismatched bases in heteroduplex regions of DNA molecules. The length of a heteroduplex region that is repaired as a single unit, and that consequently behaves as a single unit of gene conversion, is believed to be small [mode of several hundred base pairs in yeast nuclear chromosomes according to FOGELand MORTIMER(1969) ;less than three thousand base pairs in the phage according to WILDENBERGand MESELSON(1975)l. Thus “unlinked” markers such as the cap2 and olil loci or the eryl and olil loci in the yeast mitochondrial genome, which are about 11-15 x IO3 or 5-1 1 x IO3 base pairs apart (NAG- LEY et al. 1976; SCHWEYENet al. 1976; SANDERSet al. 1976; MORIMOTOet al. 1976) should rarely be included in the same heteroduplex repair tract. This expectation is borne out by the genetic studies of DUJON, SLONIMSKIand WEILL(1974). The two members of such pairs of loci should accordingly behave completely independently with respect to gene conversion and random drift, and should become uniparental in an independent manner. Specifically, if pl is the total frequency of zygotes uniparental for one allele at one locus, and pz is the total frequency of zygotes uniparental for the corresponding allele at the other locus, the frequency of zygotes uniparental for both alleles should be given by p1p2. The frequency of zygotes uniparental at neither locus is qlq2 where q = l-p, and so forth. The C and E loci, in contrast, are substantially closer (about 2-6 X IO3 base pairs) and are believed to be included in the same heteroduplex in some recombination events (DUJON, SLONIMSKIand WEILL1974; PERLMAN and BIRKY1974) ; these two loci might be expected to become uniparental in the same zygote more often than predicted from the product of their individual uniparental frequencies. At the other extreme, we might find that all loci become uniparental together in every zygote. In this case (“coincidence”), for example, the cross CSW X CRORwould produce zygotes uniparental for C8 and Os,or for CRand On,but never for Cs alone, or Os alone, etc. 632 c. WILLIAM BIRKY, JR. et al.

-de

,xU c -2 5 3 -d -_c .v wc $ 4 .Uc 3 "? :U I bnL S& 1 2s CFI)E 8% .>FI)Q .s -e8 2,- .2 E WE: 22 \D CL8 a 4-s i 23 4 w& B 3- $4ab .;; ._ 00 22 $5 E .$;. h% O2 8 .I? o'p -EUR .v Uc L .QH

UD % f \ " 2 E: $ E k INHERITANCE OF YEAST MITOCHONDRIAL GENES 633 The relevant data are given in Table 5. It can be seen that, in every case, the loci C and 0 or E and 0 become uniparental together more often than predicted on the hypothesis of independence. The same is true for C and E, but these two loci, as expected from their relatively close linkage, become uniparental together more often than the “unlinked” pairs. These results can be summarized by say- ing that there is a strong tendency for a zygote to become uniparental for whole mtDNA molecules rather than for individual loci. Additional evidence for this will be presented below and in a subsequent paper. Although these results do not say that random drift and gene conversion play no role whatsoever in uniparental inheritance in yeast, they do say that molecules of the minority genotype must fail to replicate. Genetic experiments alone cannot readily distinguish between failure of molecules to replicate followed by dilution, and outright enzymatic destruction. Hereafter, we will speak of molecules as being “lost” without prejudging the issue. Whatever the fate of the lost markers, it is appropriate to ask why they are not always lost as complete linkage groups (with predicted values for the coincidence hypothesis as shown in Table 5). The answer is almost certainly that some markers may be rescued by recombination from molecules destined to be lost, and transferred to molecules destined to be replicated. Closely linked markers such as C and E would be more likely to stay together, i.e., to be lost or transmitted together, on this hypothesis. The extent to which two markers do go uniparental together would then be a measure of their linkage. (Attempts to order the 0 locus relative to the C and E loci using uniparental zygote data have failed; this is not surprising since these three loci cannot be ordered by recombination mapping and 0 appears formally unlinked to C and E.) If this hypothesis is correct, we would expect that zygotes in which two of the three loci are uniparental for alleles from the majority parent should show very high transmission of the majority allele at the third locus, because only one or a few copies of the minority allele at the third locus would be rescued in any given zygote. Further, zygotes uniparental at only one locus should show very high transmission at the other two loci for the same reason. This expectation is borne out by the data in Table 6. In this experiment (p5 PESOS X ID4/1 CREROR),a number of zygotes were scored as uniparental at one or two loci. A sample of those that were uniparental sensitive at only one or two loci was tested by replica-plating to determine the frequency of each genotype, and of the sensitive allele at the remaining locus or loci; in every case, that frequency proved to be very high. This is further evidence that entire genomes tend to be lost or transmitted together in uniparental zygotes, i.e., that the uniparental mechanism acts on molecules rather than loci. If this is true, then the three loci studied here (capl, eryl, and OM)should become uniparental at approximately the same frequencies. That this is so can be seen by examining the data for individual crosses in Tables 2 and 5, or by looking at the mean frequencies of uniparental zygotes for a number of crosses shown in Table 7. Only crosses producing more than 10% uniparental zygotes are included in this table in order to be certain that matings between wild-type 0, 03 TABLE 6 P Complete zygote clone analyses of clones scored by the replica-plating test as uniparental at one or two loci

Output ratios (% diploid cells) Clones scored as as uniparental for CSESOS CEEsOs CsEROS CSEsOR CREROS CRE5OR @EROR CREEOR ZCs ZES ZOS N ESOs (8 clones) 96.5 3.5 96.5 100 100 433 100 100 100 100 348 100 100 100 100 332 100 100 100 1 010 391 100 494 99.6 0.4 99.6 100 P 87.1 12.9 87.1 100 100 41 0 100 100 100 1 0'0 ma 99.0 1.o 99.0 iao 100 2m E: Mean 97.8 2.2 97.8 100 100 z> CSOS (6 clones) 97.7 2.3 10.0 97.7 100 264 m, 95.0 5.0 100 95.0 100 574 ; 94.3 5.7 100 94.3 100 561 9 96.3 3.7 100 96.3 100 519 4 97.0 100 97.0 100 T 3.0 233 m 98.9 1.1 100 98.9 la0 535 n. Mean 96.6 3.4 100 96.6 100

CSES (8 clones) 87.0 13.0 100 100 87.0 21 6 99.5 0.5 100 100 99.5 22 98.9 1.1 100 100 98.9 371 100 100 100 100 403 99.2 0.8 100 1010 99.2 359 70.0 30.0 100 100 70.0 493 92.6 7.4 100 100 92.6 229 99.8 0.2 100 100 99.8 598 Mean 93.4 6.6 100 100 93.4 TABLE 6-Continued

Output ratios (% diploid cells) Clones scored as ~ as uniparental for CUES08 CRESOE CSEROS N __ OS (9 clones) 80.0 2.5 17.5 80.0 82.5 100 405 67.3 32.7 67.3 67.3 100 636 98.6 1.4 98.6 98.6 100 647 98.6 0.5 0.9! 99.1 98.6 99.1 216 94.5 3.5 2.0 98.0 94.5 100 4.57 100 100 100 100 435 99.0 1.o 99.0 99.0 100 7m 94.9 5.1 94.9 94.9 100 563 37.6 62.4 62.5 62.5 100 282 Mean 85.6 0.3 0.4 13.6 88.8 88.7 99.9 E* (4 clones) 86.0 8.9 4.9 0.2! 95.1 99.8 86.0 451 83.9 0.5 15.6 99.5 100 84.4 398 52.9 2.3 45.6 54.4 100 52.9 393 100 100 100 100 194 Mean 80.7 0.1 6.7 12.6 0.05 87.3 99.5 80.8 Cs (9 clones) 91 .o 0.3 8.5 0.3! 99.7 99.4 91.3 609 79.3 16.6 4.1 100 95.9 79.3 314 55.0 14.5 14.5! 16.0! 69.5 69.5 69.5 462 88.1, 8.51 8.5! 83.2 83.2 91.5 377 98.0 2.0 100 1010 98.0 150 96.6 3.4 100 100 96.6 2Q7 93.8 5.4 0.8! 10 100 93.8 129 88.7 0.9 10.4 100 99.1 88.7 337 99.5 0.5! 99.5 99.5 99.5 373 Mean 87.2 0.1 6.8 2.6 0.1 0.5 2.8 94.7 94.1 89.8

! Indicates resistant genotypes which should not have been found in the clones, according to the replica-plating test. These indicate an error either in the replica-plating test (most likely in the case of the two “CS” clones with large frequencies of unexpected genotypes) or in the zygote clone analysis (most likely in the other cases where only one or two cells had unexpected genotypes). See MATERIALS AND METHODS for further discussion of sources of error. 636 c. WILLIAM BIRKY, JR. et al. TABLE 7 Frequency with which different loci become uniparental, and transmission frequencies from same crosses

Mean percent Mean perFnt Lpci present Number uniparental zygotes transrmssion in crosses of crosses C E 0 C E 0 c, E 11 20.9 20.4 - 71.1 71.7 - E, 0 9 - 20.9 20.4 - 69.6 69.1 c, 0 11 22.2 - 23.5 70.1 - 69.8 c, E, 0 9 21.2 20.9 20.4 68.8 69.6 69.1

Data include only crosses involving NI23 or p5, with >10% uniparental zygotes. cells and petite mutant cells lacking markers do not influence the results, but the same relationship is seen in crosses with fewer uniparental zygotes. Thus, these three loci behave coordinately with respect to uniparental inheritance, as well as with respect to transmission frequencies in random diploids. Although this coordinate behavior has been demonstrated for only three loci, they include at least two functionally different regions of the genome and include regions far enough apart to behave independently in recombination. It would require ad hcx assumptions to believe that the region studied is not representative of the entire mtDNA molecule with respect to uniparental inheritance. Uniparental inheritance is independent of zygote budding patterns As illustrated in Figure 5, two haploid yeast cells that mate form the opposite

a- -Y

FIGURE5.-Illustration of the effect of bud position and incomplete mixing of mitochondria on mitochondrial gene recombination and transmission. See text for explanation. INHERITANCE OF YEAST MITOCHONDRIAL GENES 637 ends of the dumbbell-shaped zygote. The zygote will produce buds from the narrow neck (central buds), or from either end (end buds), or any combina- tion of these positions. Different crosses produce zygotes having different frequencies of end and central buds, and different zygotes from the same mating mixture may produce different bud patterns. STRAUSBERG(1976) showed that the mitochondria contributed by the two parents are not thoroughly mixed until some time after formation of the first bud. Prior to this time, substantial mixing occurs only in the neck of the zygote, so that first end buds contain low frequencies of recombinant genotypes and are very often pure for one parental genotype, while first central buds are less frequently pure for parental genotypes and tend to contain a high frequency of recombinant mtDNA molecules. Since the progeny of the first bud constitute about 50% of the cells of the entire zygotic clones, those zygotes with first central buds produce clones with a higher average frequency of recombination than zygotes with first end buds. A first end bud leaves the mother cell with a highly biased ratio of parental mitochondrial genotypes; such a bias reduces the frequency of detectable recombination events (DUJON,SLONIMSKI and WEILL1974). To study the effect of bud position on uniparental inheritance, we isolated by micromanipulation 93 zygotes from the cross p5 x ID4/1, which produces a high frequency of uniparental zygotes, and recorded the position of the first one, two, or three buds. After each zygote had produced a clone, the transmission of all three markers from the p5 parent (Cs, Es, and Os) was determined for each zygote. Nineteen zygotes were uniparental for one or more (usually all three) markers. The total frequency of zygotes uniparental at a locus, averaged over the three loci, was 17.6%, not significantly different from 20.2% for 553 zygotes from three control crosses analyzed by the usual method without manipulation; thus the manipulation and observation did not affect the operation of the uniparental inheritance mechanism. Table 8 compares the frequencies of different budding patterns in uniparental and biparental zygotes: x2 tests show no significant differences, whether one looks at the first bud only, the first two buds, or the first three buds. We conclude that the budding pattern of the zygote does not have a strong influence on whether or not the zygote will be uniparental. Three further conclusions follow: (1) Uniparental inheritance is not due to the death of a zygote following the production of one or two pure buds from the end of the zygote housing the majority mitochondrial genotype. Of the 19 uniparental zygotes, 12 produced three buds and six produced two buds during the period of observation. It is likely that all zygotes eventually produced more than three buds: STRAUSBERG(1976) and others (e.g.,FORSTER and KLEESE1975) have found that zygotes rarely die after producing only one or two buds. (2) It is a priori likely that the uniparental mechanism operates once, in the zygote, to identify the minority genotype and determine that the genotype will be eliminated. If this determination were made anew in each bud, it would be difficult for any zygote to become uniparental, as buds from the minority end would have a majority of such molecules, which would thus be rescued. Our 638 c.WILLIAM BIRKY, JR. et al. TABLE 8

Position of early zygote buds and uniparental inheritance

Bud position Uniparental zygotes Biparental zygotes 1st 2nd 3rd Number Percent Number Percent 7 25 21 75 12 19 53 82 1 100 0 0 6 22 21 78 9 19 38 81 3 18 14 82 1 20 4 80 2 50 2 50 5 23 17 77 3 19 13 81 1 14 6 86 0 0 1 100

Bud positions were scored as central (C) or end (E). The first end of the zygote to form a bud was called E,, and the opposite end E,. The position of the first three buds was noted for 55 zygotes; the first two only for 92 zygotes; and the first bud only for one zygote. data verify this, because they contain three uniparental zygotes in which the first bud came from one end and the second bud from the other end. In these zygotes either the first or the second bud must have come from the minority end. We conclude that the minority molecules are identified and irreversibly determined for loss at least before the production of the second bud, and cannot be rescued by incorporation into a bud. The frequency of this E,E, budding pattern is approximately the same in uniparental and biparental zygotes. (3) These data provide further evidence against the gene conversion/random drift mechanism for uniparental inheritance. According to that hypothesis, zygotes with the highest frequencies of recombination, i.e., zygotes with first central buds, should be more likely to become uniparental than zygotes with first end buds. This relationship is in fact observed, but the difference is small and is not significant. Again, we cannot rule out some contribution of gene conversion/random drift to the mechanism of uniparental inheritance. How are molecules determined for replication or loss? In most of the crosses we have studied, zygotes become uniparental for markers from only one parent, the one providing the majority input. To account for this observation, two general classes of models can be envisioned. In the first class, the mtDNA molecules from the two parents are marked in some fashion, e.g.. by modification of one or the other, or by modification of different sites in each, while they are still present in the parent cells. It is then necessary for the cell to ‘‘count” molecules in order to identify the minority genotype for loss. This could be done, for example, as shown in Figure 6, where loci closely linked to the mating type locus would produce a nuclease (analogous to restriction enzymes) in equal quantities but each one specific for the mtDNA molecules INHERITANCE OF YEAST MITOCHONDRIAL GENES 639 input ratio 6:3:2:1

.1 output 4:l

Uniparental zygote

FIGURE&-A “counting” mechanism for uniparental inheritance. Genes at or near the mating type locus produce enzyme Maand Ma, which differentially modify mtDNA in the a and a parents. Nucleases Ra and R,, produced in the zygote by genes at or near a and a, degrade mtDNA not modified by Ma and Ma, respectively. Partial degradation produces a biparental zygote with amplified bias; further degradation produces a uniparental zygote. contributed by the other parent. Progressive action of these enzymes, at the same rate, would eliminate the minority molecules first and result in some cells being uniparental for the majority genotype. (The “counting” is done by the nucleases.) An important feature of this model, and probably of all members of this class of models, is that it would amplify the input bias so that the output was more extremely biased in favor of the majority parent. The alternative class of models is composed of those in which the parental molecules are unmarked. This means that they cannot be counted; in order to insure that only minority molecules are eliminated, it is necessary to postulate 640 c. WILLIAM BIRKY, JR. et al. input ratio 6:3-2:1

8 + Uniparental 27 zygote or output 310 12- I) Biparental 27 zygote or output 2:l -6 I) Biparental 27 zygote or output 112 1 I) Unipa ren ta I 27 zygote output 0:3 FIGURE7.-Example of a hypothetical uniparental inheritance mechanism involving replication of a small random sample of mtDNA molecules. Only molecules attached to the three sites on the cell membrane survive enzymatic degradation. Four classes of zygotes are produced; because a real zygote will have about 100 or more mtDNA molecules, the frequencies of the classes are given approximately by the binomial distribution. The random diploid output = (8/27) * (3/3) +(12/27). (2/3) +(6/27) .( 1/3) (2/3) = 2: 1. INHERITANCE OF YEAST MITOCHONDRIAL GENES 641 some sort of competition between molecules for the privilege of replication (or escape from random degradation). An example is illustrated in Figure 7; it is assumed that there are a limited number of sites (membrane attachment sites, or replicating enzymes) for which all mtDNA molecules must compete. Alter- natively, these sites might be required to protect molecules from enzyme degradation. A model of this sort will amplify the input bias only if the number of sites is greater than the number of molecules of the minority genotype. In the case of yeast, this would require so many replicating sites (ca. 50 or more) that no uniparental zygotes would be produced, unless the unit of replication is an entire, homoplasmic mitochondrion containing many mtDNA molecules. Experimentally, we can distinguish between these two models by determining whether the output bias is always or usually greater than the input bias. The input bias is known only if one knows the mtDNA content of cells of both mating types, a measurement that is difficult IO do on a large number of stocks. Alter- natively, one can mate many different tester stocks to two strains whose mtDNA contents differ by a known factor. We have done experiments comparing the mutant uvsp5 with its parent strain N123; log-phase cells of p5 have approximately twice as much mtDNA per cell as N123 with either glucose or glycerol as the carbon source (MOUSTACCHI,PERLMAN and ~HLER1976). If the output bias equals the input bias (no amplification), then the output using these two testers can be related as follows: Let n equal the output frequency (and hence the input frequency) of markers from N123; then 1 - n equals the output and input from the tester. In the uvsp5 mating, the zygote receives again 1 - n molecules from the tester parent, and 2n from the uvsp5 parent. The expected output from the cross with uvsp5 is then 2n/(2n + 1 - n) = 2n/(n 4- 1). The exact magnitude of the bias amplification predicted by a counting moldel cannot be specified at present, because it depends upon the details of the specific model and upon the value of several parameters. Table 9 summarizes the available data and the results of these calculations; it is clear that bias amplification is usually observed in crosses involving uvsp5. GOLDTHWAITE,CRYER and MARMUR(1974) worked with cultures which had markedly different mtDNA contents when grown on different carbon sources. Several crosses were strongly biased but the data show no consistent bias amplification. GUNGE(1975, 1976) has performed similar experiments using pairs of isogenic haploid and diploid strains; the diploids should have approximately twice the mtDNA content of the haploids (GRIMES,MAHLER and PERLMAN 1974). GUNGE’Sdata (1975, 1976) consistently show bias amplification when calculations are done as described above. We have preliminary data obtained with a different pair of haploid and diploid strains that show significant bias amplification with some tester strains, but not with others. These results are compatible with counting models as mechanisms for uniparental inheritance, and we conclude that such models are worthy of further examination. But the experiments do not conclusively demonstrate bias amplification; to do this, crosses must be done in such a way as to eliminate premating synthesis (e.g., by hydroxyurea treatment), which could change the effective 642 c. WILLIAM BIRKY, JR. et al. TABLE 9 Bias amplification: obserued transmission of mitochondrial alleles from p5, compared to transmission expected on basis of crosses of same tester strain to NI23 if bias is not amplified

Transmission Observed- Tester observed Expected expected Xa __~~ 1-1/1 0.585 0.516 0.069 2.30 0.8031 0.650 0.153 76.6 * 0.79211 0.625 0.167 63.9 * 0.843% 0.747 0.099 12.3 * 0.890% 0.801 0.089 0.766s 0.782 -0.016 0.64 ID-4/1 0.615 0.588 0.027 0.99 0.693% 0.593 0.100 15.2 * 0.618 0.395 0.223 108 * 41-1 0.588s 0.429 0.159 4.76** COP/152 0.84@$ 0.683 0.166 35.5 * COP/13 0.54+4 0.259 0.285 154 *

Data for 1-1/1 and COP/152 are pooled for the C and E loci; for ID-4/1 and 41-1, data are pooled for the C, E, and 0 loci; for COP/19, data are pooled for the C, 0, and P loci. * P < 0.001. ** P < 0.05. + Ethanol carbon source. 11 Glycerol carbon source. $ Data from DEMKO(1975). $ Data from RALPE KEIL (personal communication). input of mtDNA molecules if it occurred to different extents in different stocks. And bias amplification, if present, need not necessarily by associated with uniparental inheritance. For instance, bias would be amplified if the first buds produced by zygotes tended to come from the end of the zygote with the largest number of mtDNA molecules, hut received less than the diploid number of such molecules. which would then be replicated disproportionately in order to bring that bud up to the diploid level. Finally, we note that because only 1-10% of yeast cells actually mate in most crosses, the interpretation of these experiments assumes that those cells in cultures of differing mtDNA contents which actually do mate have mtDNA contents in the same ratio to each other as do the average cells in the two cultures. Yeast cells always mate at the same stage of the cell cycle (SENA et al. 1975a). But they grow and synthesize mtDNA continuously through the cycle (WILLIAMSONand FENNELL1974; SENAet al. 1975); moreover older cells that have produced more buds are larger and probably have more mtDNA than younger cells (LEEand JOHNSON1977). Thus, our assumption could be in error if cultures with different average mtDNA contents also consistently have differences in the relative lengths of different parts of the cell cycle, or if they had consistent differences in the age distributions of their cells. Multiple zygote classes BIRKY (1975a, b) showed that many crosses produce multiple classes of zygotes differing in their transmission of mitochondrial alleles from one parent INHERITANCE OF YEAST MITOCHONDRIAL GENES 643 or the other. Some crosses produce two uniparental classes: some zygotes transmit only mitochondrial alleles from the a parent, and other zygotes transmit only mitochondrial alleles from the a parent. Another example of a cross with multiple classes is shown in Figure 8. In this cross (p5 x ID4/I) there is a large class of zygotes uniparental for alleles from pS, a fairly distinct class transmitting p5 alleles to 71-80% of the progeny, and a small class of zygotes transmitting ID4/1 genes preferentially or exclusively (0-10% transmission of p5 alleles). Several crosses that produce a substantial frequency of uniparental zygotes are unbiased; no bias is seen in random diploids because zygotes uniparental for markers from one parent are balanced by zygotes uniparental for, or strongly biased in favor of, alleles from the other parent. Also, two crosses with a definite bias produced a substantial frequency of zygotes that were uniparental for markers from the minority parent. These observations must be dealt with in any hypothesis designed to explain uniparental inheritance. It is possible that there are multiple classes of zygotes differing in their inputs of mtDNA molecules (BIRKY 1975c), so that some zygotes are biased in favor of one or the other parent and become uniparental, while others are unbiased or weakly biased. The mechanism of uniparental inheritance per se could still be relatively simple, depending solely upon the presence or absence of an input bias. If, on the other hand, all zygotes have the same input bias, a site-competition model with a small number of sites could be applied. For example, the cross D6 x 6-2/7, which shows just three distinct classes (uniparental for a markers, uniparental for .o( markers, and biparental) could be explained by a two-site model. But not all unbiased crosses show these three classes, so that one would have to postulate that different crosses produce zygotes with different numbers of sites. A third possibility is that the uniparental mechanism is activated in only some zygotes in a cross. For example. in the hypothesis described above involving the action of modification and restriction enzymes, it might be supposed that the restriction enzymes were not produced in all zygotes, or that some zygotes produced restriction enzymes for only one of the two classes of mtDNA, possibly the minority class. The mechanism for this might be the activation (or inactivation) of alleles from the a and a parents, which produce the hypothetical restriction enzymes: either one, both, or neither of the alleles might be functional in a particular zygote and the probabilities of these three events might vary between crosses. A plausible analogy is facultative heterochromatization of the mammalian X chromosome, although in the case of yeast we would have to postulate that only a small chromosomal segment, perhaps closely linked to the mating-type locus, is involved. The analogy is particularly attractive because different mammals differ in the specificity of heterochromatization: in some tissues in marsupials, it is always the paternal X chromosome that is inactivated, while in eutherian mammals the paternal X chromosome is inactivated in some cells and the maternal X chromoaomal in others (reviewed by CATTA- NACH 1975). Another fruitful analogy may be allelic repression seen in hybrid plants and animals (e.g.,CASTRO-SIERRA and OHNO 1968; CASSIDYand BLACKLER 644 c.WILLIAM BIRKY, JR. et ai. 43

35 // 25 i

n 2 20 z0 N

Y 0 L n4) 5 15 Z 1

11 m 21 31 41 51 61 81 9 -10 IES- 20 -30 -40 -50 -60 70 -90 -100 -

FIGURE8.-Transinission frequencies of the mitochondrial alleles in 99 individual zygote clones from the cross CSESOS x CREEOR (p5 x ID4/1). Bars represent the number of zygotes transmitting the C, E, or 0 alleles from the p5 parent to various percentages of their diploid progeny, in intervals of 10%. In the O-lO% and 91-100% intervals, those portions of the bars above the horizontal breaks represent zygotes scored as uniparental by replica-plating. INHERITANCE OF YEAST MITOCHONDRIAL GENES 645 1974), where specific alleles from only one parental species are transcribed or expressed. In this connection, we note that laboratory stocks of SaccharomycPs cereuisiae were originally derived from hybrids between several yeast species (SHERMANand LAWRENCE 1974), and crosses between stocks may sometimes be essentially like interspecific crosses at one or more loci. Compzrison of uniparental inheritance in Saccharomyces and other systems. It has long been recognized that uniparental (principally maternal) inheritance distinguishes the inheritance of genes in mitochondria and chloroplasts from that of genes in the nucleus, and although not universal, is so common a phenomenon that it can be taken as a general rule of organelle gene inheritance (BIRKY1976). Our finding that many, and perhaps all, yeast crosses can produce zygotes uniparental for mitochondrial genes invites a comparison with other systems and an attempt to identify common mechanisms. The yeast data provide the clearest evidence that uniparental inheritance involves the loss of entire organelle DNA molecules rather than individual loci: physically linked genes tend to be lost or transmitted together, not independently, and exceptions can probably be accounted for by marker rescue through recombination. This is probably also true for chloroplast genes in Chlamydo- monas. SAGERand RAMANIS(1963) performed a two-factor cross using the allele pairs sd/ss and ac-/ac +. Unirradiated zygotes were plated on media that would select for zygotes biparental for one locus or for both; data were not given, but the authors stated that zygotes biparental at both loci were more frequent than expected if these loci become uniparental independently. Additional data from zygote clones obtained in two- and three-factor crosses have been analyzed by the method shown in Table 5 of this paper, with results similar to the yeast case (BIRKYet al. in preparation). The outside markers in these crosses show 5-12% recombination, but it is not possible to estimate their physical distance; it is possible that they could all be included in a single heteroduplex. In other systems where mitochondrial or chloroplast genes show uniparental inheritance, such as Neurospora, Oenothera or Pelargonium, no analysis is possible because only one organelle gene locus has been examined in each cross. It is of special interest to note that uniparental inheritance of mitochondrial genes in yeast occurs principally when there is a biased input of mtDNA molecules from the two parents, and that the frequency of uniparental zygotes increases with increasing input bias. In most other systems, the input of organelle genes into the zygote is not even approximately known, but a strong maternal bias is suspected in many cases. In higher plants and animals, the cytoplasmic volume of the egg is much greater than that of the sperm, and this very likely indicates a bias in terms of the numbers of organelle DNA molecules. In Xeno- pus, DAWIDand BLACKLER( 1972) have demonstrated maternal inheritance of mtDNA, and interpret this in terms of their finding of about lo8and lo2mtDNA molecules in the egg and sperm, respectively. In plants, light and electron micro- scope observations suggest a relative paucity (or complete absence) of proplas- tids in the pollen or pollen tube of some species (reviewed by JENSEN 1973). The 646 c. WILLIAM BIRKY, JR. et al. conidia of Neurospora contain mitochondria, but the total volume of conidial cytoplasm is much less than that of the protoperithecium. The transformation of a biased input of organelle DNA molecules into a uniparental output may be a general feature of organelle transmission genetics. We have proposed that models for the mechanism of uniparental inheritance can be divided into two classes: (1) random selection of a small sample of molecules for replication, where the probability that a particular genotype will be replicated is proportional to its frequency (input) in the zygote; and (2) differential marking of molecules from the two parents, followed by a counting mechanism that ensures that molecules of the majority type will be preferentially replicated and molecules of the minority type will be preferentially lost. We are unable to decide conclusively between these two models in yeast, but we favor the second for two reasons. First, it is difficult to devise random selection models that accurately predict the relationship between the percent uniparental zygotes and the percent transmission, or the frequency distributions of transmission percentages in zygote clones. Second, the data showing amplification of the input bias argue in favor of the counting model. Specific examples of the first kind of models have been proposed in the literature for other systems. GILLHAM,BOYNTON and LEE (1974) and ADAMS(1975; ADAMS,GILLHAM and BOYNTON1975) have proposed random sample models for Chlamydomonas zygotes after UV irradiation of mt+ gametes, postulating that only a fraction of the total number of chloroplast DNA molecules are able to replicate because they are successful in competing for membrane attachment sites analogous to those proposed for bacterial and viral DNA replication. TILNEY- BASSETT(1970) discussed models formally similar to the random sample model in connection with his observations on plastid inheritance in the geranium (Pelargonium). In contrast, SAGER(1975) interprets uniparental inheritance of chloroplast genes in Chlamydomonas in terms of a model in which the chloroplast DNA molecules are differentially marked, but only DNA from the mt- parent is destroyed. For Chlamydomonas crosses where the mt+ gametes are not irradiated, the models of GILLHAM,BOYNTON and LEE (1974) and ADAMS(1975; ADAMS,GILLHAM and BOYNTON1975) also postulate differential marking of genomes from the nt+ and mt- genomes; only the former occupy replication sites. These models are analogous to our “counting” model in requiring differential marking, but the decision to replicate and transmit genomes or not is based on the mating-type allele of the parent, rather than upon the relative numbers of genomes. At first sight, uniparental inheritance of chloroplast genes in Chlamydomonas appears to differ markedly from uniparental inheritance of mitochondrial genes in Saccharomyces. In yeast, the direction of uniparental inheritance is not determined directly by the mating-type locus or other nuclear genes, but rather by the input bias. In Chlamydomonas, on the other hand, the majorityofuniparental zygotes transmit chloroplast genes from the mt+ parent in most crosses. But the two systems may be basically very similar. SAGER(1975) and co-workers find that chloroplast DNA from the mt- parent is degraded in the zygote in crosses INHERITANCE OF YEAST MITOCHONDRIAL GENES 647 with a high frequency of zygotes uniparental for mt+ markers (“maternal” zygotes). CHIANG(1976) reported that the chloroplast DNA from both parents is degraded extensively in zygotes, but that DNA from the mt- parent is degraded more rapidly at first, so that young zygotes may still have substantially more chloroplast DNA molecules from the mt+ than from the mt parent. GILL- HAM’S and BOYNTON’Sgroup (GILLHAM,BOYNTON and LEE1974; ADAMS1975; VANWINKLE-SWIFT 1976a) have obtained genetic evidence that those Chlamy- domonas exhibiting biparental inheritance of chloroplast genes contain functional copies of chloroplast DNA molecules from both parents, with a strong bias in favor of the mt+ parent; this might result from the destruction of most, but not all, mt- chloropast DNA. (See, however, SAGERand RAMANIS1976 for some contradictory evidence.) Moreover, GILLHAM( 1969) and VANWINKLE- SWIFT(1976a) have found that an attempt to reduce this input bias by mating haploid mt+ gametes with diploid mt- gametes was successful in reducing the frequency of maternal zygotes and increasing the frequency of zygotes transmitting paternal chloroplast alleles. This result is analogous to those from haploid x diploid crosses, which we have performed with yeast, to be reported later. These results could be interpreted in terms of a two-step mechanism for uniparental inheritance in Chlamydomonas: first, preferential degradation of mf- chloroplast DNA molecules in the zygote, controlled by the mating-type locus, results in a population of molecules strongly biased in favor of the mt+ parent. The bias then sets into motion a second mechanism, possibly the same as in yeast, which results in the preferential replication of chloroplast genomes from the mt+ parent. The zygotes of Chlamydomonas may thus differ from those of yeast only in having an obligatory “input” basis for organelle genes. The preceding discussion refers to the majority of zygotes in Chlamydomonas, those which undergo meiosis immediately. VANWINKLE-SWIFT (1976a, b) has recently done an extensive analysis of chloroplast inheritance in “diploids,” i.e., zygotes that behave like yeast in that they do not undergo meiosis but instead produce colonies of diploid cells by mitosis. Uniparental diploids are found but the frequency of uniparental inheritance is lower than among meiotic zygotes produced in the same cross. Uniparental inheritance is much less closely controlled by the mating-type locus: although in most crosses the majority of uniparental diploids transmit alleles only from the mt+ parent, there are always a significant number that are uniparental for alleles from the mt- parent, and some crosses produce a majority of such paternal diploids. No direct evidence is available for or against an input bias of chloroplast genomes in the case of diploids, but the frequency of uniparental diploids is correlated with transmission frequencies, just as it is for meiotic zygotes or for yeast. These and other simi- larities lead us to hope that the laws of transmission genetics will prove to be very similar for chloroplast genes in Chlamydomonas and mitochondrial genes in Saccharomyces.

We are grateful to MARYTOWNSEND, ANN GLAMM,and ROBBIN PALMERfor their skillful technical assistance; to MICHAELDOUGLAS and DAVIDFINKLESTEIN for the gift of alpha-factor; 648 c. WILLIAM BIRKY, JR. et al. to HARRYM. CROOKS,JR. (Parke, Davis and Company) for the gift of paromomycin; and to KAREN VAN WINKLE-SWIFT,JOHN BOYNTON, MICHAEL WAXMAN, and RALPH KEIL for sharing their unpublished data. We thank KARENVAN WINKLE-SWIFTfor many stimulating and critical discussions and reading of the manuscript, and NICHOLASGILLHAM, JEAN FORSTER, G. M. W. ADAMS,and other reviewers for helpful comments. This research was supported by research grant GM19607 from the Public Health Service. R. L. STRAUSBERGwas a predoctoral trainee, Public Health Service Training Grant HDO431.

LITERATURE CITED

ADAMS,G. M. W., 1975 An investigation of the mechanisms of inheritance of chloroplast genes in Chlamydomonas reinhardtii. Ph.D. Thesis, Duke University. ADAMS,G. M. W., N. W. GILLHAMand J. E. BOYNTON,1975 A probabilistic model for the inheritance of chloroplast genes in Chlamydomonas reinhardtii. J. Cell Biol. 67: 2a (abst.) . BIRKY,C. W. JR., 1974 Uniparental inheritance of organelle genes in yeast and in general. Genetics 77: s5 (abst.). -, 1975a Effects of glucose repression on the transmission and recombination of mitochondrial genes in yeast. Genetics 80: 695-705. -, 1975b Zygote heterogeneity and uniparental inheritance of mitochondrial genes in yeast. Molec. Gen. Genet. 141: 41-58. __, 1975c Mitochondrial genetics in fungi and ciliates. pp. 182-224. In: Genetics and Biogenesis of Mitochondria and Chloroplasts. Edited by C. W. BIRKY,JR., P. S. PERLMANand T. J. BYERS.Ohio State University Press, Columbus. -, 1976 The inheritance of genes in mitochondria and chloroplasts. Bioscience 26: 26-33. BIRKY,C. W. JR. and R. V. SKAVARIL,1976 Maintenance of genetic homogeneity in systems with multiple genomes. Genet. Res. 27: 249-265. BIRKY,C. W. JR., R. L. STRAUSBERG,P. S. PERLMANand J. L. FORSTER,1977 Estimating the parameters of random vegetative segregation of mitochondrial genomes in yeast. Molec. Gen. Genet. 158: 251-261. C4LLEN, D. F., 1974 Segregation of mitochondrially inherited antibiotic resistance genes in zygote cell lineages of Saccharomyces cereuisiae. Molec. Gen. Genet. 143: 65-76. CASSIDY,D. M. and A. W. BLACKLER,1974 Repression of nucleolar organizer activity in an interspecific hybrid of the genus Xenopus. Dev. Biol. 41 : 84-96. CASTRO-SIERRA,E. and S. OHNO,1968 Allelic inhibition at the autosomally inherited gene locus for liver alcohol dehydrogenase in chicken-quail hybrids. Biochem. Genet. 1 : 323-335. CATTANACH,B. M., 1975 Control of chrommome inactivation. Ann. Rev. Genet. 9: 1-18. CHIANG,K. S., 1976 On the search for molecular mechanisms of cytoplasmic inheritance: past controversy, present progress, future outlook. pp. 305-312. In: Genetics and Biogenesis of Chloroplasts and Mitochondria. Edited by T. BUCHER,W. NEUPERT,W. SERALDand S. WERNER.Elsevier/North-Holland Biomedical Press, Amsterdam. COEN,D., J. DEUTSCH,P. NETTER,E. PETROCHILOand P. P. SLONIMSKI,1970 Mitochondrial genetics. I. Methodology and phenomenology. Symp. Soc. Exp. Biol. 24: M9-496. CRYER,D. R., C. D. GOLDTHWAITF.,S. ZINKER, K.-B. LAM,E. STORM,R. HIRSCHBERG,J. BLAMIRE, D. B. FINKELSTEINand J. MARMUR,1973 Studies on nuclear and mitochondrial DNA of Saccharomyces cereuisiae. Cold Spring Harbor Symp. Quant. Biol. 38: 17-29. DAWID,I. B. and A. W. BLACKLER,1972 Maternal and cytoplasmic inheritance of mitochondrial DNA in Xenopus. Develop. Biol. 29: 152-161. DEMKO,C. A., 1975 The effects of gene dosage on mitochondrial marker transmission in Sacchrcmyces cereuisiae. M.Sc. Thesis, The Ohio State University, Columbus. INHERITANCE OF YEAST MITOCHONDRIAL GENES 649 DUJON,B., A. KRUSZEWSKA,P. P. SLONIMSKI,M. BOLOTIN-FUKUHARA,D. COEN, J. DEUTSCH, P. NETTERand L. WEILL, 1975 Mitochondrial genetics X: effects of UV irradiation on transmission and recombination of mitochondrial genes in Saccharomyces cereuisiae. Molec. Gen. Genet. 137: 29-72. DUJON,B. and P. P. SLONIMSKI,1976 Mechanisms and rules for transmission, recombination and segregation of mitochondrial genes in Saccharomyces cereuisiae. pp. 393-4Q4. In: Genetics and Biogenesis of Chloroplasts and Mitochondria. Edited by T. BUCHER,W. NEUPERT,W. SEBALDand S. WERNER. Elsevier/North-Holland Biomedical Press, Amsterdam. DUJON, B., P. P. SLONIMSKIand L. WEILL, 1974 Mitochondrial genetics IX: a model for recombination and segregation of mitochondrial genomes in Sacchromyces cereuisiae. Genetics 78: 415-437. FOGEL,S. and R. K. MORTIMER,1969 Informational transfer in meiotic gene conversion. Proc. Nat. Acad. Sci. US. 62: 96-103. FORSTER,J. L. and R. A. KLEESE, 1975 The segregation of mitochondrial genes in yeast. 11. Analysis of zygote pedigrees of drug-resistant x drug-sensitive crosses. Molec. Gen. Genet. 139: 341-355. FRAENKEL,A. M., 1974 The effects of nuclear mutations of recombination and repair functions and of caffeine on mitochondrial recombination. Ph.D. Thesis, The Ohio State University, Columbus. GILLHAM,N. W., 1963 Transmission and segregation of nonchromosomal factor controlling streptomycin resistance in diploid Chlamydomonas. Nature 200: 294. -, 1969 Uni- parental inheritance in Chlamydomonas reinhardii. Am. Naturalist 103 : 355-388. -, 1974 Genetic analysis of the chloroplast and mitochondrial genomes. Ann. Rev. Genet. 8: 347-391. GILLHAM,N. W., J. E. BOYNTONand R. W. LEE, 1974 Segregation and recombination of non- Mendelian genes in Chlamydomonas. Genetics 78: 439-46 7. GOLDTHWAITE,C. D., D. R. CRYERand J. MARMUR,1974 Effect of carbon source on the replication and transmission of yeast mitochondrial genomes. Molec. Gen. Genet. 133: 87-104.. GRIMES,G. W., H. R. MAHLERand P. S. PERLMAN,1974 Nuclear gene dosage effects on mitochondrial mass and DNA. J. Cell Biol. 61 : 565-574. GROSSMAN,L. I., E. S. GOLDRINGand J. MARMUR,1969 Preferential synthesis of yeast mitochondrial DNA in the absence of protein synthesis. J. Mol. Biol. 4.6:367-376. GUNGE,N., 1975 Genetic analysis of unequal transmission of the mitochondrial markers in Saccharomyces cereuisiae. Molec. Gen. Genet. 139: 189-202. -, 1976 Effects of elevation of strain-ploidy on transmission and recombination of mitochondrial drug resistance genes in Saccharomyces cereuisiae. Molec. Gen. Genet. 146: 5-16. JENSEN,W. A., 1973 Fertilization in flowering plants. Bioscience 23: 21-27. LEE, E. H. and B. F. JOHNSON,1977 Volume-related mitochondrial deoxyribonucleic acid synthesis in zygotes and vegetative cells of Saccharomyces cereuisiae. J. Bacteriol. 129 : 1066-1071. LINNANE,A. W., G. W. SAUNDERS,E. B. GINGOLDand H. B. LUKINS,1968 The biogenesis of mitochondria. V. Cytoplasmic inheritance of erythromycin resistance in Saccharomyces cereuisiae. Proc. Nat. Acad. Sci. U.S. 59: 903-910. LINNANE,A. W., N. HOWELLand H. B. LUKINS,1974 Mitochondrial genetics. pp. 193-213. In: The Biogmesis of Mitochondria. Edited by A. M. KROONand C. SACCONE.Academic Press, New York. MCBRIDE,A. C. and J. C. MCBRIDE,1975 Uniparental inheritance in Chlamydomonas eugametos. J. Phycology 11 : 343-344. 650 c.WILLIAM BIRKY, JR. et al. MORIMOTO,R., A. LEWIN, S. Mmrm and M. RABINOWITZ,1976 Restriction endonuclease mapping and analysis of grande and mutant yeast mitochondrial DNA. pp. 519-524. In: Genetics and Biogenesis of Chloroplasis and Mitochondria. Edited by T. BUCHER,W. NEUPERT,W. SEBALDand S. WERNER. Elsevier/North-Holland Biomedical Press, Amsterdam. MOUSTACCHI,E., P. S. PERLMANand H. R. MAHLER,1976 A novel class of Saccharomyces cereuisiae mutants specifically UV-sensitive to “petite” induction. Molec. Gen. Genet. 148: 25 1-261. NAGLEY,P., K. S. SRIPRAKASH,J. RYTKA, K. B. CHOO,M. K. TREMBATH,H. B. LUKINSand A. w. LINNANE,1976 Physical mapping of genetic markers in the yeast mitochondrial genome. pp. 231-242. In: The Genetic Function of Mitochondrial DNA. Edited by C. SACCONEand A. M. KROON.North-Holland Publishing Company, New York. OGUR,M., R. ST. JOHNand S. NAGAI,1957 Tetrazolium overlay technique for population studies of respiration deficiency in yeast genetics. Science 125 : 928-929. PERLMAN,P. S. and C. W. BIRKY,JR., 1974 Mitochondrial genetics in bakers’ yeast: a molecular mechanism for recombinational polarity and suppressiveness. Proc. Nat. Acad. Sci. U.S. 71: 4612-4616.

PERLMAN,P. S., C. W. BIRKY,JR., C. A. DEMKOand R. L. STRAUSBERG,1976 Confirmations and exceptions to the phage analogy model: input bias, bud position effects, zygote heterogeneity, and uniparental inheritance. pp 405-414. In: Genetics and Biogenesis of Chloro- plasts and Mitochondria. Edited by T. BUCHER,W. NEWPERT,W. SEBALDand S. WERNER. Elsevier/North-Holland Biomedical Press, Amsterdam.

PERLMAN,P. S., C. W. BIRKY,JR. and R. L. STRAUSBERG,1978 Segregation of mitochondrial markers in yeast. Methods in Enzymology. Vol. 56, in press. PERLMAN,P. S. and C. A. DEMKO,1974 Effects of gene dosage on mitochondrial marker transmission in Saccharomyces cereuisiae. Genetics 77: s50-51 (abst.) . PERLMAN,P. S. and H. R. MAHLER,1971 Molecular consequences of ethidium bromide mutagenesis. Nature New Biol. 231: 12-16. RANK,G. H. and N. T. BECH-HANSEN,1972 Somatic segregation, recombination, asymmetrical distribution and complementation tests of cytoplasmically-inherited antibiotic-resistance mitochondrial markers in S. cereuisiae. Genetics 72: 1-15. SAGER,R., 1975 Patterns of inheritance of organelle genomes: molecular basis and evolutionary significance. pp. 252-267. In: Genetics and Biogenesis of Mitochondria and Chloroplasts. Edited by C. W. BIRKY,JR., P. S. PERLMANand T. J. BYERS.Ohio State University Press, Columbus. SAGER,R. and Z. RAMANIS,1963 The particulate nature of nonchromosomal genes in Chlamy- domonas. Proc. Nat. Acad. Sci. US. 50: 260-268. SAGER,R. and Z. RAMANIS,1976 Chloroplast genetics of Chlamydomonas: I. Allelic segregation ratios. Genetics 83: 303-321. SANDERS,J. P. M., C. HEYTING,A. DIFRANCO,P. BORSTand P. P. SLONIMSKI,1976 The organization of genes in yeast mitochondrial DNA. pp. 259-272. In: The Genetic Function of Mitochondrial DNA. Edited by C. SACCONEand A. M. KROON.North-Holland Publishing Company, New York. SCHWEYEN,R. J., B. WEISS-BRUMMER,B. BACKHAUS and F. KAUDEWITZ,1976 Localization of seven gene loci on a circular map of genetic markers in the yeast mitochondrial genome. pp. 251-258. In: The Genetic Function of Mitochondrial DNA. Edited by C. SACCONEand A. M. KROON.North-Holland Publishing Company, New York. INHERITANCE OF YEAST MITOCHONDRIAL GENES 65 1

SENA,E. P., D. N. RADIN,J. WELCHand S. FOGEL,1975a Synchronous mating in yeast. Vol. 11, pp. 71-88. In: Methods in Cell Biology. Edited by D. M. PRESCOTT,Academic Press, New York. SENA,E., J. WELCHand S. FOGEL,1976 Nuclear and mitochondrial DNA replication during zygote formation and maturation in yeast. Science 194: 433-435. SENA,E. P., J. W. WELCH,H. 0. HALVORSONand S. FOGEL,1975b Nuclear and mitochondrial deoxyribonucleic acid replication during mitosis in Saccharomyces cereuisiae. J. Bacteriol. 123: 497-504. SHERMAN,F. and C. W. LAWRENCE,1974 Saccharomyces. pp. 359-393. In: Handbook of Genetics. Edited by R. C. KING.Plenum Press, New York. STRAUSBERG,R. L., 1976 Cellular and partial pedigree analysis of mitochondrial inheritance in Saccharomyces cereuisiae. Ph.D. Thesis, The Ohio State University, Columbus. THOMAS,D. Y. and D. WILKIE, 1968 Inhibition of mitochondrial synthesis in yeast by erythromycin: cytoplasmic and nuclear factors controlling resistance. Genet. Res. 11 : 33-41. TILNEY-BASSETT,R. A. E., 1970 The control of plastid inheritance in Pelargonium. Genet. Res. 16: 49-61. -, 1975 Genetics of variegated plants. pp. 268-308. In: Genetics and Biogenesis of Mitochondrial and Chloroplasts. Edited by C. W. BIRKY,JR., P. S. PERLMAN and T. J. BYERS.Ohio State University Press, Columbus. VAN WINKLE-SWIFT,K. P., 1976a The transmission, segregation and recombination of chloroplast genes in diploid strains of Chlamydomonas reinhardtii. Ph.D. Thesis. Duke University. __, 197613 Variability in chloroplast gene transmission in mitotic zygotes of Chlamy- domonas reinhardtii. Genetics 83: s78-79 (abst.) . WAXMAN,M. F., N. EATONand D. WILKIE,1973 Effect of antibiotics on the transmission of mitochondrial factors in Saccharomyces cereuisiae. Molec. Gen. Genet. 127: 277-284.

WICKERH4M, L. J., 1946 A critical evaluation of the nitrogen assimilation tests commonly used in the classification of yeasts. J. Bacteriol. 52 : 293-301. WILDENBERG,J. and M. MESELSON,1975 Mismatch repair in heteroduplex DNA. Proc. Nat. Acad. Sci. U.S. 72: 2202-2206. WILKIE,D., 1972 Genetic aspects of mitochondria. Proc. Eighth Meeting Fed. Europ. Biochem. Soc. 28: 85-94. WILKIE,D. and D. Y. THOMAS,1973 Mitochondrial genetic analysis by zygote cell lineages in Saccharomyces cereuisiae. Genetics 73: 368-377. WILLIAMSON,D. H. and D. J. FENNELL,1974 Apparent dispersive replication of yeast mitochondrial DNA as revealed by density labelling experiments. Molec. Gen. Genet. 131: 193-207. Correpsonding editor: N. W. GILLHAM