Tetrahymena Macronuclear Genome Mapping: Colinearity of Macronuclear Coassortment Groups and the Micronuclear Map on Chromosome 1L

Corresponding author: Eduardo Orias, Department of Molecular, Cellular, and Developmental Biology, University of California, Santa Barbara, CA 93106., orias{at}lifesci.lscf.ucsb.edu (E-mail)

Communicating editor: S. L. ALLEN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION LITERATURE CITED

The genetics of the ciliate Tetrahymena thermophila are richer than for most other eukaryotic cells, because Tetrahymena possesses two genomes: a germline (micronuclear) genome that follows a Mendelian model of genetic transmission and a somatic (macronuclear) genome, derived from the micronuclear genome by fragmentation, which follows a different genetic transmission model called phenotypic assortment. While genetic markers in the micronucleus fall into classical linkage groups under meiotic recombination and segregation, the same markers in the macronucleus fall into coassortment groups (CAGs) under phenotypic assortment by the random distribution of MAC chromosome pieces. We set out to determine whether genomic mapping in the macronucleus by genetic means is feasible. To investigate the relationship between the micronuclear map and coassortment groups, we systematically placed into CAGs all of the markers lying on chromosome 1L that are also found in the macronucleus. Sixteen CAGs were identified, 7 of which contain at least two loci. We have concluded that CAGs represent a fundamental genetic feature of the MAC. The MIC and MAC maps on 1L are colinear; that is, CAGs consist exclusively of markers that map to a continuous segment in a given region of the micronuclear map, with no intervening markers from other CAGs. These findings provide a solid foundation for exploiting the MAC chromosome pieces to build a physical map of the Tetrahymena genome.

OUR lab is genetically mapping the genome of the ciliate Tetrahymena thermophila, a unicellular eukaryote having unique experimental advantages (reviewed by ORIAS 1998 ) and a history of importance as a model system. Notable advances using Tetrahymena include the discovery of ribozymes (e.g., KRUGER et al. 1982 ) and telomeres and telomerases (e.g., GREIDER and BLACKBURN 1985 ; YU et al. 1990 ). Tetrahymena has special advantages for the study of additional biological processes, including regulated secretion, phagocytosis, site-specific chromosome fragmentation and diminution, regulation of chromatin function, and cellular morphogenesis (see ORIAS 1998 ). Tetrahymena, and ciliates in general, are also particularly interesting from a fundamental genetic standpoint due to their genome organization, which differs substantially from those of more "familiar" eukaryotes.

The genetics of Tetrahymena are enriched by the possession of two distinct but related genomes within a single cell, each described by a different genetic segregation model. This is related to the phenomenon of nuclear dimorphism (reviewed in BRUNS 1986 ; ORIAS 1986 ; KARRER 1999 ), exhibited by Tetrahymena and other ciliates (reviewed in PRESCOTT 1994 ), in which cells contain two types of nuclei: a micronucleus (MIC), and a much larger macronucleus (MAC). The micronucleus is diploid and is silent in gene expression; it constitutes the germline of the cell. The Tetrahymena MIC has five pairs of chromosomes and behaves in a classical Mendelian fashion with respect to segregation of genetic markers at meiosis. The macronucleus, by contrast, is the site of all known gene expression in the cell (hence, it is often called the "somatic" nucleus). It is polyploid and behaves during vegetative reproduction according to a different model of genetic segregation known as phenotypic assortment, described below.

The MAC genome is derived from the MIC genome during sexual reorganization, which occurs when cells of different mating type conjugate (for a review of the Tetrahymena life cycle, see ORIAS 1986 ). A variety of DNA rearrangements and a large increase in ploidy occur in the developing MAC during this process, which is called macronuclear differentiation. One major type of DNA rearrangement is the site-specific fragmentation of chromosomes derived from the germline into much smaller pieces (ANDREEVA et al. 1979 ; PREER and PREER 1979 ). Each site of fragmentation is marked by a chromosome breakage site (Cbs), a 15-bp sequence that is necessary and sufficient to cause fragmentation at that site (YAO 1989 ; YAO et al. 1990 ). Telomeres are added de novo to both ends of all resulting macronuclear pieces, which have been called "macronuclear chromosomes" or autonomously replicating pieces (ARPs). We use the latter terminology here. This process is also associated with a large increase in MAC ploidy, which is ultimately maintained at a level of ~45 relative to the MIC (reviewed in PRESCOTT 1994 ). One MAC DNA molecule, the rDNA, is maintained at an especially high copy number: 9000 copies per MAC (YAO et al. 1974 ).

To be precise, ARP in our usage denotes a specific type of macronuclear DNA molecule, each one having been amplified in copy number from a specific segment of the MIC genome and having a defined size. There are 200–300 ARPs in the Tetrahymena MAC according to published estimates (ALTSCHULER and YAO 1985 ; CONOVER and BRUNK 1986 ), ranging in size from 21 kb for the rDNA to several megabase pairs in length. Note, however, that due to the high ploidy in the MAC, the total number of copies of all ARPs, or individual DNA molecules in the MAC, is >20,000. To avoid confusion, we refer to these MAC constituents as ARP copies to distinguish them from ARPs in the discussion below.

Another type of DNA rearrangement that occurs during macronuclear differentiation is the site-specific deletion of some internal sequences [internally eliminated sequences (IESs)], which account for ~10–20% of total genome complexity (YAO 1989 ; COYNE et al. 1996 ). About 6000 distinct deletion events are estimated to occur per haploid genome. We refer to genetic markers that fall in MIC IES segments as MIC-limited; they are absent from the MAC genome. Conversely, loci that fall in MIC segments that are represented in the MAC genome (the majority) we call MAC-destined; they are present in both genomes.

The macronucleus does not behave in a Mendelian fashion with regard to genetic segregation. The MAC does not contain observable kinetochores—or any other ultrastructure (DAVIDSON and LAFOUNTAIN 1975 ) that could be associated with regular partitioning of ARP copies during binary fission. Combined physical and genetic evidence strongly supports the idea that copies of a particular ARP are randomly partitioned at MAC division (L. WONG, L. KLIONSKY, S. WICKERT, V. MERRIAM, E. ORIAS and E. HAMILTON, unpublished results). Random partitioning would normally be expected to result in copy number fluctuations of each ARP among progeny and eventual loss of individual ARPs. However, a variety of observations suggest the existence of (as yet unknown) mechanisms that individually regulate the copy number of each ARP (reviewed by PRESCOTT 1994 ).

A key element of the transmission genetics of the MAC is how loci located on the same or different ARPs assort between daughter cells during vegetative propagation. Individual copy number control of ARPs and random partitioning of their copies between daughter cells have implications both for how loci behave when considered individually and when considered in combination with other loci.

The behavior of alleles at individual loci is well described by a genetic segregation model called phenotypic assortment. In 1958, Allen and Nanney discovered that if a cell starts out with a mixture of two alleles at a particular macronuclear locus, its binary fission descendants will assort into clonal lines that are pure for one allele or the other at that locus (ALLEN and NANNEY 1958 ; reviewed by DOERDER et al. 1992 ). This phenomenon, discovered in the segregation of differentiated alleles of the mating-type locus in the MAC, was later found to be general to all genes and was eventually named phenotypic assortment (SONNEBORN 1974 ). A detailed mathematical model that describes phenotypic assortment and agrees very well with experimental observations was proposed by SCHENSTED 1958 . This model assumes the stochastic partitioning of allele copies at MAC division (see Fig 1), and it makes detailed predictions about the kinetics of assortment and its limiting (steady-state) rate as a function of the number of assorting units. Indeed, the first estimates that the ploidy of the MAC is ~45 were calculated from the observed limiting rate of assortment under the model. Experimental support for the random partitioning assumption of the Schensted model was presented by ORIAS and FLACKS 1975 .

View larger version (25K): In this window In a new window Download PPT slide   Figure 1. Phenotypic assortment and coassortment in the macronucleus during vegetative propagation. (A) The circles represent macronuclei, shown over a period of two cell cycles. The "peanut shell" stages (2, 3a, and 3b) represent MACs dividing amitotically. The lines represent copies of a single ARP; straight and wavy denote the inbred strain B and C3 homologues, respectively. Only 3 out of 45 copies are shown for simplicity. The diamonds and circles represent two different loci; solid and open symbols denote the B and C3 alleles, respectively. Phenotypic assortment is illustrated by considering only the locus represented as a diamond (or only that represented by a circle). Note that the 2 C3 copies in dividing MAC 2b (or the 2 B copies in dividing MAC 3b) have equal probability of undergoing disjunction (i.e., one to each daughter MAC) or nondisjunction (i.e., both to the same daughter MAC) (ORIAS and FLACKS 1975 ). Phenotypic assortment therefore causes a gradual loss of allelic diversity with respect to the ARP copies, because once a daughter cell arises that has not received any copies of a given allele at cell division (MACs 2a and 3b1) due to random partitioning, the loss is irreversible, and all descendants of that cell will lack the allele. These assortants have become "pure" for the alternative allele, and all their vegetative descendants remain pure for the same allele. The kinetics of this stochastic process during vegetative propagation are described in detail by the Schensted model. With 45 copies, the generation of pure assortants from still mixed MACs proceeds, after reaching steady-state, at a rate of 1/2N, where N is the total G1 number of copies (SCHENSTED 1958 ). Coassortment is illustrated by considering jointly the diamond and circle loci. If there were no crossing in the MAC one would expect that every assortant pure for the B allele at the diamond locus would also be pure for the B allele at the circle locus, and likewise for the C3 alleles. This "assorting together" is called "coassortment." Note that all the assortants become pure for parental combinations. If two loci are in different ARPs, they are expected to assort independently, i.e., without a statistically significant excess of pure assortants of parental over recombinant phenotype. (B) Macronuclear crossing over between the two loci generates recombinant ARP copies, shown in the inset at bottom left. As described in the text, pure assortants of recombinant phenotype do occur, but are infrequent enough to allow easy distinction between coassortment and independent assortment. Note also that other genetic events can generate assortants of recombinant phenotype (see text).

Additional phenomena arise when one considers the assortment behavior of pairs or groups of loci. If the assumptions of the Schensted model are correct, if the assorting units are the ARP copies in the MAC, and if there were no crossing over in the MAC, then loci that are located together on the same ARP should assort together, while loci located on different ARPs should always assort independently from one other. Considered as a whole, the MAC genome should then be subdivided into groups of genetic markers, with each group corresponding to an ARP, whose members always assort identically to one another under phenotypic assortment as illustrated in Fig 1. These have been called "coassortment groups" (CAGs; LONGCOR et al. 1996 ).

Experimental confirmation of CAGs as a general feature of the MAC genome would validate the model as a complete fundamental description of macronuclear genetics in Tetrahymena and would open up many possibilities, beginning with MAC genome mapping by purely genetic approaches. Historically, the density of available genetic markers in Tetrahymena has been too low to allow experimental verification of coassortment, as the few loci investigated were found to assort independently. Recently, however, the high density of DNA polymorphisms generated by the Tetrahymena mapping initiative (http://lifesci.ucsb.edu/~genome/Tetrahymena) made possible the first observation of coassortment by LONGCOR et al. 1996 , who also presented physical evidence that ARPs are the physical basis of coassortment. Our lab will soon provide more rigorous physical and genetic evidence in support of this hypothesis (L. WONG, L. KLIONSKY, S. WICKERT, V. MERRIAM, E. ORIAS and E. HAMILTON, unpublished results).

In the work reported here, we sought to extend these previous findings and to establish that coassortment is a general feature of macronuclear genetics. We also sought to examine closely the relationship between CAGs and the micronuclear map. To accomplish these objectives, we mapped 40 loci on the left arm of chromosome 1 at high resolution in the MIC, using standard meiotic mapping (see WICKERT and ORIAS 2000 , this issue). In parallel with that effort, but using MAC genetics and phenotypic assortment, we also systematically placed all corresponding MAC-destined markers in chromosome 1L into CAGs. Results from the MAC mapping effort, and on the correspondence between MIC and MAC genomes, are presented here.

We find that (1) coassortment groups are indeed a general feature of Tetrahymena macronuclear genetics, and (2) loci in a given coassortment group map to a continuous segment of the MIC map, uninterrupted by loci destined to a different CAG. Thus, to the level of resolution investigated to date, there is a simple colinearity between MIC and MAC genetic maps. This work also validates the idea that MAC maps in Tetrahymena can be generated by purely genetic rationales and approaches, thus further increasing its advantages as a genetic system.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION LITERATURE CITED

Routine methods:
General methods for Tetrahymena culture, constructing panels of terminal assortants, PCR methods, and mapping loci to the MIC using meiotic segregants have been previously described (LYNCH et al. 1995 ; BRICKNER et al. 1996 ; LONGCOR et al. 1996 ; WICKERT and ORIAS 2000 ).

Delineating macronuclear coassortment groups:
We used a panel of 36 B/C3 heterozygous cell lines, which were subcloned after multiplying asexually for more than 500 fissions. [The panel and its derivation were described in LONGCOR et al. 1996 .] Members of this panel are considered "terminal assortants" because the probability of allele purity due to phenotypic assortment at any given locus is predicted to be >99.9% (DOERDER et al. 1975 ). Whole-cell DNA was extracted and purified as described in LARSON et al. 1986 . For every randomly amplified polymorphic DNA (RAPD), PCR was done (as described above) to determine which allele (B or C3) had become fixed in each terminal assortant (see also LONGCOR et al. 1996 ). Assortment patterns were compared between all pairwise locus combinations. The PCR primers used for each RAPD are listed in Table 1 of WICKERT and ORIAS 2000 . For a given locus pair, a statistically significant excess of parental types (LOD > 3) was taken as evidence of coassortment. MAPMAKER/EXP 3.0 software (LANDER et al. 1987 ) was used to speed up comparisons and to determine the level of statistical significance. MAPMAKER analysis used a 1:1 segregation model, expressing LOD scores against independent segregation as LOD scores against independent assortment, because coassortment is the MAC analog of meiotic linkage. Retrospective error detection took advantage of the rarity of assortants of recombinant type, because most scoring errors should turn parental into recombinant types. We repeated all RAPD reactions for terminal assortants of recombinant type within a coassortment group.

An important source of scoring error in the assortment data was the appearance, for some of the RAPDs, of a weak band in assortants that had become pure for the band negative (band^-) allele (the allele that fails to generate a band). We attribute this weak signal to the lone copy of the band positive (band⁺) allele present in the still heterozygous micronucleus. The whole-cell DNA that we used to template the RAPD PCR reactions should contain a small percentage of MIC DNA relative to MAC DNA, based on the ploidy difference between MIC and MAC. If the RAPD primer binding sites do not bracket an IES or Cbs, both MIC and MAC DNA are expected to template polymorphic bands of identical size. Discrimination between MAC band⁺ and band^- assortants was possible, because almost always the MIC-derived band, if present, was uniformly weaker than that derived from a MAC assorted to the band⁺ allele, but very careful observation was required.

Identification of MIC-limited RAPDs:
Some of our genetic markers fall within IESs, and we call these "MIC-limited," because they are deleted from the macronucleus. RAPDs were identified as either MIC-limited or MAC-destined by testing nullisomic strains missing chromosome 1L (LONGCOR et al. 1996 ) or whole-genome B-C3 heterokaryon strains (E. ORIAS, E. V. MERRIAM, J. D. ORIAS and E. HAMILTON, unpublished results) MIC-limited RAPDs were excluded from consideration in the work reported here, because they are missing from the MAC.

Statistical support for colinearity between MIC and MAC maps:
Construction of the MIC map using MAPMAKER is described in detail in WICKERT and ORIAS 2000 . To compute the odds against split orders in the MIC for loci in a MAC CAG, full maximum-likelihood calculations using all MAC-destined markers were done using MAPMAKER. The calculations were done as follows: For each CAG having more than one member, a "window," in which marker order was allowed to vary, was placed around the markers in the CAG. This window included at least three markers outside the CAG at each end (except at the ends of the map). The MIC marker order on which the window was placed was the maximum-likelihood order of all MAC-destined markers (see WICKERT and ORIAS 2000 ). Within the window, maximum-likelihood maps were constructed for all possible marker orders, and their likelihoods, relative to the maximum-likelihood order, were computed (this procedure is automated by the MAPMAKER "compare" command). The calculations included all the MAC-destined markers outside the window, but their order was held fixed. For each CAG, the likelihood of the most likely split order, i.e., any MIC order in which the CAG members are not contiguous, was noted. In some cases, split orders involving the opposite ends of a CAG were noted separately, wherever reporting both values was more informative. Odds against split orders were computed as 10^-:1, where is the LOD (Log₁₀ of the ODds) score of the most likely split order, relative to the maximum-likelihood order.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION LITERATURE CITED

Macronuclear coassortment map of chromosome 1L:
Our aim was to construct a MAC coassortment map of MAC-destined loci that had previously been localized to MIC chromosome 1L and to relate MIC and MAC maps to one another. Of the 40 RAPD markers in the micronuclear map (see WICKERT and ORIAS 2000 ), 12 of them are MIC-limited (E ORIAS, E. V. MERRIAM, J. D. ORIAS and E. HAMILTON, unpublished results) and are therefore not destined for the macronucleus; they were excluded from consideration here. Because the two classical loci in 1L (PMR1 and mat) confer an observable phenotype expressed from the macronucleus, we were able to test them for coassortment with the RAPD loci as well.

We used a panel of 36 "terminal assortants," consisting of B/C3 heterozygous cell lines subcloned after asexual propagation for at least 500 fissions (see MATERIALS AND METHODS). The DNA from each panel member was PCR-amplified with the appropriate primers to determine which allele (B or C3) has assorted at each locus. The data were analyzed for independent vs. linked assortment (coassortment) of loci in all pairwise combinations as described in MATERIALS AND METHODS. Loci that showed coassortment at LOD > 3 were organized into coassortment groups. These were assigned the name of one locus in the group, without the leading 1, but prefixed by cag and the MIC chromosome arm to which the loci in the coassortment group map (example: cag1L-PM8). Where the context is clear, the MIC chromosome designator may be omitted.

Nineteen of the loci fell into seven coassortment groups of at least 2 loci (Table 1), and the remaining 9 loci assorted independently (Table 2 and below). Within any given coassortment group, most members of the panel were of parental type for all the loci: they assorted to the same alleles (B or C3) at each locus in the group (indicated by the "consensus pattern" at the top of the listing for each group in Table 1). A minority of panel members were of recombinant type: they had assorted to the B allele at some loci and to the C3 allele at others. The low frequency of assortants of recombinant type is fortunate, because it facilitates the distinction between coassorting and independently assorting loci with the relatively small panel we have used. On the other hand, the frequency of recombinants is generally too small to allow meaningful quantitation or independent determination of marker orders and distances within a coassortment group (see DISCUSSION).

Allele ratios in particular coassortment groups generally did not significantly deviate from 1:1 (Table 3). Highly significant biases (probability of ² < 1%) were observed in only 3 of the 16 assortment patterns (cag1L-rDNA, cag1L-BD6, and cag1L-1XS24). In the rDNA CAG, the PMR1 and 1EM10 loci assorted exclusively to the C3 allele (Table 1) and thus gave no MAC recombinants. PMR1 and 1EM10 also failed to give meiotic recombinants in the MIC (WICKERT and ORIAS 2000 ). PMR1 lies on the rDNA ARP (BRUNS et al. 1985 ; SPANGLER and BLACKBURN 1985 ) and the C3 homologue of this ARP is preferentially replicated/maintained with respect to the B homologue in mixed MACs (LARSON et al. 1986 ), which explains its assortment pattern. That the 1EM10 RAPD also lies on the rDNA ARP was demonstrated by specific hybridization of plasmid pD5H8 (YAO and YAO 1991 ), carrying the entire rDNA ARP, to the 1EM10 polymorphic band in RAPD PCR reactions (data not shown). Additional possible bases for the other examples of strongly biased assortment are listed in the DISCUSSION.

Pairs of loci from the 16 coassortment groups listed in Table 2 assorted independently of one another (LOD against independent assortment <1.7), with the lone exception of loci in the rDNA-BD6 coassortment group pair (LOD = 3.8). In contrast to 1EM10, the 1BD6 RAPD PCR band failed to hybridize with the identical rDNA probe, leading us to conclude that the rDNA and BD6 groups do not lie on the same ARP. We are treating the observed coassortment between PMR1 and 1BD6 as spurious, artificially caused by the strong C3 bias of the two patterns (see DISCUSSION).

Colinearity of micronuclear and macronuclear maps:
A major goal of this work was to place the genetic data for coassortment patterns of MAC loci on 1L into the context of the MIC map (WICKERT and ORIAS 2000 ) for the same region. A striking feature of the MAC and the maximum-likelihood MIC maps of chromosome 1L in T. thermophila (Fig 2) is that they are colinear, in the sense that markers that fall into coassortment groups map to a continuous segment in the MIC map, with no intervening markers destined to other CAGs.

View larger version (27K): In this window In a new window Download PPT slide   Figure 2. Colinearity of MIC and MAC maps on chromosome 1L. The maximum-likelihood MIC map (see WICKERT and ORIAS 2000 for details) is represented by the thick solid line on the left (dotted line indicates continuity), and the corresponding coassortment groups (CAGs) in the MAC are shown to the right as hatched bars. Leading 1's on all RAPD marker names are omitted for clarity. Micronuclear map: Square brackets indicate uncertainty in marker relative order at LOD 3 (see the above reference). A small filled triangle to the left of a name indicates a MIC-limited marker, and this is also indicated by a small white rectangle at the corresponding location on the thick solid line. These markers are absent from the macronuclear map. Parentheses surround markers (1GM9, 1JO13R, and 1JO16) that could not be placed into a unique map interval at LOD 3 or better according to the criteria in WICKERT and ORIAS 2000 . A 10-cM scale bar is shown at top right. Macronuclear map: MAC-destined markers are shown in the coassortment groups (CAGs) into which they fall, indicated by the hatched bars. Only genetic information from macronuclear coassortment mapping is represented here, which can in general determine coassortment, but not genetic distances in the MAC (see main text). Markers are therefore shown in positions corresponding to their locations in the micronuclear map, except that some have been shifted slightly to better show their CAG assignments. The exact length of the hatched bars relative to genetic distance in the MIC is arbitrary.

We next examined the level of statistical support for colinearity in detail. Because not all MIC markers can be ordered at LOD 3 in our map (as indicated by square brackets in Fig 2; see legend), we compared the likelihoods of MIC marker orders that "split" a CAG with that of the maximum-likelihood order. All seven CAGs on chromosome 1L that have more than one marker were tested. The highest-likelihood split orders, along with the odds against them, are shown in Fig 3. Statistical support for colinearity is high, except in some cases where, for example, markers at the boundaries of adjacent CAGs are located very close together in the MIC map, making marker order difficult to resolve. These cases involve one or more markers in CAGs cag1L-YD19, cag1L-rDNA, cag1L-JO13R, and cag1L-PM8. We examine them individually below.

View larger version (28K): In this window In a new window Download PPT slide   Figure 3. Statistical support for colinearity of CAG markers in the micronuclear map. The most likely split orders and the odds against them are exhaustively listed for each of the CAGs on chromosome 1L that has more than one member. Split orders show only markers in the immediate vicinity of the split, and all other markers are in their maximum-likelihood order (see Fig 2) for purposes of calculation. Members of the named CAG are shown in boldface, and other MAC-destined markers, including the one that "splits" the CAG, are shown in italics. In comparing these split orders to the maximum-likelihood order shown in Fig 2, note that MIC-limited markers, which are lost from the MAC and do not appear in CAGs, are not shown here. They were also ignored in calculation of the odds for each order. aThe two members of this CAG, PMR1 and EM10, were shown physically to be on the same ARP, the rDNA. bNeither of the members of this CAG (GM9 and JO13R) placed on the map at LOD 3.

The region around cag1L-YD19 has an unusual cluster of markers (see WICKERT and ORIAS 2000 ). Four other markers (1JP34, 1MJ10aR, 1SN7a, and 1XS24) are coincident with 1YD19 in the MIC map, showing no micronuclear recombination at all among them. However, 1XS24 assorts independently in the MAC from the other members of the cluster (Table 2). The fact that these five markers are all coincident in the MIC map means that the position of 1XS24 cannot be resolved at all among the members of cag1L-YD19 (odds 1:1 against split orders). In addition, another independently assorting marker, 1JB10R, is very close to cag1L-YD19. Its maximum-likelihood position does not split cag1L-YD19 and is supported by odds of 2.4:1. Locus 1JO7 is located very close to the markers in cag1L-rDNA in the MIC map, but assorts independently. An unsplit order for cag1L-rDNA and 1JO7 is supported by only modest odds of 4.5:1 due to this proximity.

For cag1L-JO13R, the placements on the MIC map of both of its member markers are less certain than for most other markers [for details, see the legend of Fig 1 of WICKERT and ORIAS 2000 ], so the corresponding support against split orders is necessarily lower. Neither of the members of cag1L-JO13R could be placed on the map at LOD 3. Still, the maximum-likelihood order is unsplit, though supported by odds of only 4.6:1.

Finally, the large cag1L-PM8 shows a possible split order at each of its ends. At one end, the independently assorting marker mat is coincident with 1PM8, and their relative order cannot be determined. At the other end of the CAG, in a region of generally high marker density, an order unsplit by the independently assorting 1SN9 is supported by odds of 4.1:1.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION LITERATURE CITED

Coassortment groups are a fundamental genetic feature of the Tetrahymena macronuclear genome:
Coassortment is the MAC genetic analog of linkage in the germline, as indicated in Table 4, which also describes other genetic segregation analogies between the MIC and the MAC. Our present systematic analysis of coassortment in chromosome 1L has identified seven coassortment groups containing at least two loci. It generalizes the discovery of the first two coassortment groups in Tetrahymena (LONGCOR et al. 1996 ) and strongly suggests that coassortment groups represent a fundamental genetic feature of its macronucleus. The frequency of terminal assortants of recombinant type within coassortment groups found here is low, thus extending the original findings for the first two coassortment groups. This facilitates the distinction between coassorting and independently assorting RAPDs.

Independently assorting loci are viewed as representing coassortment groups in which so far only one locus has been identified. The consensus assortment patterns of the 7 coassortment groups and of the nine independently assorting loci altogether define 16 coassortment groups in chromosome 1L (see Table 2). Each assortment pattern can be viewed as the "signature" of each coassortment group in this particular panel of 36 terminal assortants. Table 2 also gives a striking glimpse into the extraordinary extent of intraclonal, hereditary phenotypic diversity generated from the allelic diversity contained in the heterozygous progenitor. The other side of the coin, of course, is the extreme decrease in allelic diversity in individual cells caused by phenotypic assortment, such that, functionally, a heterozygous clone ultimately becomes a collection of genetic haplotypes.

Correspondence between micronuclear and macronuclear maps:
We initially embarked on a project of higher resolution micronuclear mapping in Tetrahymena (WICKERT and ORIAS 2000 ) when preliminary data suggested that some CAGs might be "split" in the micronuclear map. In other words, some markers that fell into the same CAG under phenotypic assortment seemed not to be contiguous in the micronuclear map, but were apparently separated by markers located in other CAGs.

Our work here suggests that, at least on the left arm of chromosome 1, which may be about an eighth of the Tetrahymena genome (WICKERT and ORIAS 2000 ), the segments from which CAGs are derived are not split or rearranged in the micronuclear map on this gross length scale. Instead, there is a simple correspondence between the two maps, with CAGs sorting into uninterrupted clusters of markers in the micronucleus. If this colinearity is a general feature, continuity of a coassortment group will be a helpful criterion for tentatively choosing among alternative MIC map orders that cannot be statistically differentiated without higher resolution mapping.

Because the frequency of macronuclear recombinants is low, we generally cannot directly determine from macronuclear assortment data the order of markers within a CAG. A partial exception is the PM8 CAG. The recombinant frequencies (Table 1) divide the CAG into two subgroups (1PM8/1KF2 vs. 1JO9/1EO3R), supported at LOD > 3. This grouping is consistent (colinear) with the placement of these markers in the MIC map (Fig 2). The events responsible for subdividing the PM8 CAG are believed to be internal deletions (L. WONG, L. KLIONSKY, S. WICKERT, V. MERRIAM, E. ORIAS and E. HAMILTON, unpublished results), but this would not affect the preceding conclusion of colinearity.

The nature of apparent MAC recombinants:
Assortants are considered to have a recombinant phenotype if they show the presence of the inbred strain B or C3 allele at some but not all the loci of a coassortment group. At least two types of MAC events can generate apparent recombinants: crossing over/gene conversion and DNA rearrangements, such as deletions.

The readiest explanation for assortants of recombinant type is MAC somatic crossing over/gene conversion, requiring molecular exchange between copies of the B and C3 ARP homologues. Most of the observations in Table 1, including the presence of reciprocal recombinant types, can be accounted for by crossing over. Gene conversion may account for some recombinants with apparent multiple crossovers. A possible example is in the 1YD19 CAG: its three loci appear to be very close to one another in the MIC (no meiotic recombinants observed among roughly 200 segregants tested), and thus the genotype of clone 26 (Table 1) has a low probability of being the result of two independent crossovers.

A second basis for the generation of apparent recombinants is internal deletion. In cag1L-PM8, most, if not all, of the apparent recombinants can be accounted for by the post-MAC differentiation physical deletion of ~200 kb of DNA. This event generated assortants of apparent C3 genotype in our panel by deleting the 1PM8 and 1KF2 polymorphic loci from the B ARP homologue (L. WONG, L. KLIONSKY, S. WICKERT, V. MERRIAM, E. ORIAS and E. HAMILTON, unpublished results). Internal deletion was also considered the most likely basis for the loss of some of the MAC restriction fragments that hybridize with 5S rRNA gene clusters in old clones (ALLEN et al. 1985 ). Thus, physical evidence is ultimately required to identify the precise nature of any apparent recombinant event.

The frequency of MAC recombination:
The quantitation of the frequency of MAC crossing over is less simple and direct than for meiotic recombination. The latter occurs at a single time in the life cycle, between two homologous pairs of DNA molecules that are initially identical from nucleus to nucleus. In contrast, the generation of MAC recombinants is a highly stochastic event, leading to a high expected variance for the observed frequency of recombinants (DOERDER and DIBLASI 1984 ). The following key differences from meiotic recombination, dictated by high ploidy and random partitioning, can be highlighted and are summarized in Table 4:

1. For a given locus in the MAC, 45 DNA molecules are potentially capable of entering into molecular recombination with one another. However, the probability that newly generated recombinant DNA molecules will become fixed in a terminal assortant is only 1/45 (in the absence of differential replication/maintenance in the MAC or differential cell growth).

2. The ratio of the copies of the two homologous ARPs varies from MAC to MAC as a result of phenotypic assortment; on the average, the ratio in individual MACs becomes progressively more skewed as the fission age of the clonal population increases. The closer to 1:1 is the ratio of B and C3 homologous ARP copies in a given MAC, the higher is the probability that a given molecular exchange will be productive, in the sense of generating a phenotypically detectable recombinant type. Thus, the majority of productive crossing over will occur when the heterozygotes are young, i.e., have undergone relatively few fissions.

We attempted to quantitate the rate of recombination, exploiting the results of DOERDER and DIBLASI's (1984) Monte Carlo simulation. However, our results did not allow us to arrive at a meaningful estimate because the recombinants observed were too few, varied greatly in frequency among coassortment groups, and represent a heterogeneous collection of events. On the other hand, it seems unlikely that difficulties in accurately quantitating macronuclear recombination will be an obstacle to the usefulness of doing MAC genetics in Tetrahymena. Mere identification of coassortment, for the purpose of cloning mutant genes with interesting phenotypes, may well become the usual practical end point of genetic coassortment analysis. For this purpose, the small frequency of recombinants is an advantage, the nature of apparent recombinants is unimportant, and the panel size we have chosen is quite adequate.

Our data for chromosome 1L neither demonstrate nor rule out the existence of hot spots of recombination, such as one recently reported during MAC differentiation between two laboratory-induced mutations at the serH locus (DEAK and DOERDER 1998 ). Recombination hotspots would behave as "coassortment group breakers," i.e., they would have the effect of subdividing clusters of loci that reside in the same physical ARP into two or more genetic coassortment groups. A systematic physical correlation of coassortment groups with ARPs, now underway, should allow us to detect any cases where MAC recombination hotspots break up CAGs and to assess their frequency.

Strongly biased assortment:
In 3 of the 16 assortment patterns found in chromosome 1L (rDNA, cag1L-BD6, and 1XS24; Table 3), the allele ratio differed statistically in a highly significant way from 1:1 (probability of ² < 0.01). To understand the possible basis for a strongly biased allele ratio, it is important to distinguish the ratio of the two alleles at a locus in the MAC of a heterozygote immediately after its differentiation, which we call the "developmental input ratio" [to borrow from the terminology of NANNEY 1964 ], from the ratio of terminal assortants pure for either allele, or the "output ratio." One-to-one input and output ratios are expected if the two alleles (and the ARP homologues that carry them) are equally amplified in a differentiating heterozygous MAC, and if there is no differential replication or maintenance of copies of either ARP homologue and no differential growth rate of assortants of either genotype during subsequent vegetative multiplication. The earliest insights into these phenomena, developed long before there was a physical model of MAC DNA organization, were reviewed by NANNEY 1964 .

Biased output ratios are attributable to deviations from the preceding assumptions, which are listed below. In principle, these deviations are not mutually exclusive, should be allele-specific, could favor either allele, and could have any level of intensity.

1. Differential replication/maintenance of copies of one ARP homologue over the other. In this case, the developmental input ratio should be essentially 1:1. This has been genetically as well as physically demonstrated in the case of the rDNA ARP, where the C3 homologue is preferentially replicated/maintained over the B homologue (LARSON et al. 1986 ; ORIAS and BRADSHAW 1992 ). This phenomenon accounts for the assortment of every panel member to the C3 allele at the PMR1 and 1EM10 loci (Table 1), both of which are located on the rDNA ARP (see RESULTS).

2. Differential amplification of one allelic ARP with respect to the other during MAC differentiation. This phenomenon can be inferred from observations regarding serH surface antigen (serotype) locus in heterozygotes obtained in crosses between different inbred strains (NANNEY et al. 1963 ). Identical rates of assortment to purity for either allele were demonstrated during subsequent vegetative multiplication, indicating the absence of differential replication/maintenance after MAC differentiation. Thus it is the developmental input ratio that can be inferred to be very skewed. This phenomenon has not yet been confirmed on a molecular level.

3. Cell growth advantage conferred by one allele. In a pure example of this type, the newly differentiated MAC starts out with a 1:1 input ratio. As assortment proceeds, heterozygous cells with MACs predominantly populated by the deleterious allele will be selected against. The stronger the selection differential, the stronger the expected allele bias among the terminal assortants. With the possible exception of RAPDs coassorting with the wild-type allele of drug resistance markers built into our panel, strongly biased assortment due to growth selection should be rare in our panel, because our heterozygotes were derived from crossing two otherwise wild-type inbred strains.

Additional tests are required if one wishes to determine the cause of the bias for any given coassortment group or independently assorting locus. Likewise, statistical evidence for coassortment between neighboring loci showing a strong allele bias in their assortment pattern must be considered tentative until confirmed by physical tests. If the loci lie in different ARPs (such as PMR1 and 1BD6), they are expected physically to assort independently and in reality to be in different coassortment groups. The observed assortment nonindependence must be spurious, attributable to the fact that just the strong bias in the same direction for two loci causes most assortants to be of parental type and to show statistically correlated genetic assortment.

Spurious biased assortment:
During this work we have encountered additional phenomena, listed below, that can give the misleading appearance of a strongly biased assortment ratio; they represent pitfalls to be recognized and avoided.

1. The RAPD is MAC-destined, but a fainter RAPD band of the same size is also amplified from the MIC. This problem was brought up under scoring errors (MATERIALS AND METHODS). The MIC contribution is inferred from the fact that the faint band is totally missing in initially heterozygous cell lines that have lost the corresponding chromosome arm in the MIC (E. ORIAS, E. V. MERRIAM, J. D. ORIAS and E. HAMILTON, unpublished results). Rarely, the MIC signal can be strong enough to make it appear as if virtually every cell line has assorted to the band⁺ allele. For example, subtle band intensity differences among the assortants (data not shown) were used to arrive at the assortment patterns of 1JB3 and 1SP9 (Table 2), which may still contain some errors.

2. The RAPD is MIC-limited. Because the MIC divides mitotically, both the B and C3 alleles of any RAPD are expected to be present in every terminal assortant. Thus every assortant for a MIC-limited RAPD should show the band⁺ phenotype, i.e., the RAPD should not assort in the panel. We have excluded all MIC-limited RAPDs from this study on the basis of genetic tests using heterokaryon strains (LONGCOR et al. 1996 ; E. ORIAS, E. V. MERRIAM, J. D. ORIAS and E. HAMILTON, unpublished results). This problem does not apply to gene mutations because they are all MAC-destined.

3. All the members of the panel lack DNA from some or all of either the B or the C3 homologue of a particular MIC chromosome due to a spontaneous loss in either parent. If so, every panel member will assort to the same allele (B or C3) at every locus in the segment in question. We have in fact observed this phenomenon in the panel of 36 terminal assortants among RAPDs located in chromosome 3R (E. ORIAS, K. SMYRNI, K. OZAR and S. L. ALLEN, unpublished observations). This problem is readily solved by constructing and testing a new panel of terminal assortants.

Extending coassortment mapping in Tetrahymena:
The assortment patterns shown in Table 2 will be useful for coassortment testing of future B-C3 DNA polymorphisms mapped to MIC chromosome 1L. Because coassortment mapping uses a purely genetic rationale, any genetic difference between the inbred strains B and C3 parents of our panel can be directly tested for coassortment by determining whether its assortment pattern matches one of these patterns. This is true regardless of whether the difference is detected as a molecular phenotype (e.g., an RFLP) or a biological phenotype (e.g., a surface antigen difference). We are currently constructing MAC coassortment maps of RAPDs previously mapped to other MIC chromosomes. This effort will yield the assortment pattern of each single- or multiple-locus coassortment group in our panel of 36 terminal assortants, as shown in Table 2 for 1L loci.

Laboratory mutations have generally been obtained within inbred strain B. To map these to coassortment groups, it is necessary to cross the mutant to a wild-type C3 strain and obtain a new panel of terminal assortants (HAMILTON and ORIAS 1999 ). Such a panel need only be tested with RAPDs representing each of the identified coassortment groups mapping, in the MIC, in the neighborhood of the mutant gene. Our experience (Table 3 and E. ORIAS, D. MENDINUETO, K. M. SPELLMAN, K. SMYRNI, M. RYAN, R. AVILA, K. OZAR, H. M. CHUN, M. WILLIAMS, J. D. ORIAS, M. JOHNSON, T. JOINER, A. MIRZAIAN, P. TSENG, V. MERRIAM and S. L. ALLEN, unpublished observations) suggests that 36 terminal assortants are sufficient not only to distinguish coassortment from independent assortment (for which purpose they are more than adequate) but also to generally ensure a unique assortment pattern for each coassortment group. (The uniqueness breaks down for the rare patterns with strong allele bias, but in this case a larger panel is much less useful than independent physical tests of ARP size, as discussed above). We have followed the preceding strategy to map the gal1-1 mutation, which confers resistance to 2-deoxygalactose (C. VAN SLYKE, M. WILLIAMS, K. SMYRNI, E. ORIAS and E. P. HAMILTON, unpublished results). Determining the size of the ARP that carries a mutant gene that coassorts with a mapped DNA polymorphism should facilitate its cloning by complementation because only a small fraction of the genome need be tested.

	CONCLUSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION LITERATURE CITED

Tetrahymena is a very useful research system for molecular and cell biology because it is one of a few microbial eukaryotes in which the capabilities of a well-developed genetic system can be combined with molecular analysis. The colinearity observed in this work highlights the power of Tetrahymena genetics: the ability to examine the correspondence between two genomes within the same cell, where one is a reconfiguration of the other by site-specific fragmentation, telomerization, amplification, and site-specific deletion. The two genomes behave according to different models of genetic transmission, both of which are mathematically well understood: classical Mendelian transmission for the micronuclear genome and phenotypic assortment for the macronuclear genome. That we are able to see such clear correspondence between the two genomes, using independent panels of cell lines generated by exploiting two vastly different genetic phenomena, strikingly illustrates the special capabilities and advantages of Tetrahymena genetics.

The ability to map the macronucleus by purely genetic means and to compare it to the micronuclear genetic map has relied on some of the special experimental capabilities of Tetrahymena genetics: (1) the ability to rapidly generate panels of heterozygous cells that have completed 500 fissions, exploiting the fast doubling time of Tetrahymena cells; (2) the rapid mapping of genetic markers to micronuclear chromosome arms using nullisomic strains (cells missing in their micronucleus one or both copies of a chromosome or chromosome arm; BRUNS and BRUSSARD 1981 ); (3) micronuclear linkage mapping using panels of meiotic segregants that are whole-genome homozygotes (LYNCH et al. 1995 ); and (4) the ability to quickly distinguish MIC-limited from MAC-destined markers using whole-genome heterokaryon strains (E. ORIAS, E. V. MERRIAM, J. D. ORIAS and E. HAMILTON, unpublished results). Many of the special genetic advantages of Tetrahymena rely on biological mechanisms that have coevolved with ciliate nuclear dimorphism, i.e., the separation of germline and soma within a single cell.

The genetic evidence presented here for the correspondence of the two genomes strongly suggests that no gross chromosomal rearrangements accompany the formation of the macronuclear chromosome pieces during macronuclear differentiation. This finding strengthens a view of the macronuclear genome as a natural, high copy number, large fragment, sequence-specific digest of the micronuclear genome. Because macronuclear chromosome pieces are separable (ALTSCHULER and YAO 1985 ; CONOVER and BRUNK 1986 ) and identifiable (LONGCOR et al. 1996 ), this finding encourages the use of these pieces to construct a physical map of the micronuclear genome.

	FOOTNOTES

¹ Present address: Protein Pathways, Inc., 1145 Gayley Ave., Suite 304, Los Angeles, CA 90024.
² Present address: Department of Molecular and Experimental Medicine, The Scripps Research Institute, 10550 N. Torrey Pines Rd., La Jolla, CA 92037
³ Present address: Quintiles CNS, 10201 Wateridge Cir., San Diego, CA 92121.

	ACKNOWLEDGMENTS

We thank Laura Wong for maintenance of the PCR supplies, Judy Orias for assistance with PCR, and Eileen Hamilton for important technical contributions and valuable comments on the manuscript. We also thank the National Institutes of Health for support of this work through grant RR 09231 and the UCSB College of Letters and Sciences for a Howard Hughes Undergraduate Research Fellowship to L.N. Portions of the work reported here are being submitted by S.W. in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Molecular, Cellular, and Developmental Biology at the University of California, Santa Barbara.

Manuscript received July 21, 1999; Accepted for publication November 12, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION LITERATURE CITED

ALLEN, S. L. and D. L. NANNEY, 1958  An analysis of nuclear differentiation in the selfers of Tetrahymena.. Am. Nat. 92:139-160.

ALLEN, S. L., P. R. ERVIN, T. C. WHITE, and N. C. MCLAREN, 1985  Rearrangement of the 5S ribosomal RNA gene clusters during the development and replication of the macronucleus in Tetrahymena thermophila.. Dev. Genet. 5:181-200.

ALTSCHULER, M. I. and M.-C. YAO, 1985  Macronuclear DNA of Tetrahymena thermophila exists as defined subchromosomal-sized molecules. Nucleic Acids Res. 13:5817-5831[Abstract/Free Full Text].

ANDREEVA, L. F., T. F. ANDREEVA, N. A. MERKULOVA, and S. N. BORKHSENIUS, 1979  Replicon size and rate of DNA replication in the macronucleus of Tetrahymena pyriformis.. Tsitologiia 21:318-326[Medline].

BRICKNER, J. H., T. J. LYNCH, D. ZEILINGER, and E. ORIAS, 1996  Identification, mapping and linkage analysis of randomly amplified DNA polymorphisms in Tetrahymena thermophila.. Genetics 143:811-821[Abstract].

BRUNS, P. J., 1986 Genetic organization of Tetrahymena, pp. 27–44 in Molecular Biology of the Ciliated Protozoa, edited by J. G. GALL. Academic Press, New York.

BRUNS, P. J. and T. B. BRUSSARD, 1981  Nullisomic Tetrahymena: eliminating germinal chromosomes. Science 213:549-551[Abstract/Free Full Text].

BRUNS, P. J., A. L. KATZEN, L. MARTIN, and E. H. BLACKBURN, 1985  A drug-resistant mutation in the ribosomal DNA of Tetrahymena. Proc. Natl. Acad. Sci. USA 82:2844-2846[Abstract/Free Full Text].

CONOVER, R. K. and C. F. BRUNK, 1986  Macronuclear DNA molecules of Tetrahymena thermophila.. Mol. Cell. Biol. 6:900-905[Abstract/Free Full Text].

COYNE, R. S., D. L. CHALKER, and M. C. YAO, 1996  Genome downsizing during ciliate development: nuclear division of labor through chromosome restructuring. Annu. Rev. Genet. 30:557-578[Medline].

DAVIDSON, L. and J. R. LAFOUNTAIN, JR., 1975  Mitosis and early meiosis in Tetrahymena pyriformis and the evolution of mitosis in the phylum Ciliophora. BioSystems 7:326-336[Medline].

DEAK, J. C. and F. P. DOERDER, 1998  High frequency intragenic recombination during macronuclear development in Tetrahymena thermophila restores the wild-type SerH1 gene. Genetics 148:1109-1115[Abstract/Free Full Text].

DOERDER, F. P. and S. L. DIBLASI, 1984  Recombination and assortment in the macronucleus of Tetrahymena thermophila: a theoretical study by computer simulation. Genetics 108:1035-1045[Abstract/Free Full Text].

DOERDER, F. P., J. H. LIEF, and L. E. DOERDER, 1975  Appendix: a corrected table for macronuclear assortment in Tetrahymena pyriformis, syngen 1. Genetics 80:263-265[Free Full Text].

DOERDER, F. P., J. C. DEAK, and J. H. LIEF, 1992  Rate of phenotypic assortment in Tetrahymena thermophila. Dev. Genet. 13:126-132[Medline].

GREIDER, C. W. and E. H. BLACKBURN, 1985  Identification of a specific telomere terminal transferase activity in Tetrahymena extracts. Cell 43:405-413[Medline].

HAMILTON, E. P., and E. ORIAS, 1999 Genetically mapping new mutants and cloned genes, pp. 263–278 in Tetrahymena thermophila (Methods in Cell Biology, Vol. 62), edited by D. J. ASAI and J. D. FORNEY. Academic Press, New York.

KARRER, K. M., 1999 Tetrahymena genetics: two nuclei are better than one, chapter 3 in Tetrahymena thermophila (Methods in Cell Biology, Vol. 62), edited by D. J. ASAI and J. D. FORNEY. Academic Press, New York.

KRUGER, K., P. J. GRABOWSKI, A. J. ZAUG, J. SANDS, and D. E. GOTTSCHLING et al., 1982  Self-splicing RNA: autoexcision and autocyclization of the ribosomal RNA intervening sequences of Tetrahymena. Cell 31:147-157[Medline].

LANDER, E., P. GREEN, J. ABRAHMSON, A. BARLOW, and M. DALEY et al., 1987  MAPMAKER: an interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1:174-181[Medline].

LARSON, D. D., E. H. BLACKBURN, P. C. YAEGER, and E. ORIAS, 1986  Control of rDNA replication in Tetrahymena involves a cis-acting upstream repeat of a promoter element. Cell 47:229-240[Medline].

LONGCOR, M. A., S. A. WICKERT, M.-F. CHAU, and E. ORIAS, 1996  Coassortment of genetic loci during macronuclear division in Tetrahymena thermophila.. Eur. J. Protistol. Suppl. 1:85-89.

LYNCH, T. J., J. BRICKNER, K. J. NAKANO, and E. ORIAS, 1995  Genetic map of randomly amplified DNA polymorphisms closely linked to the mating type locus of Tetrahymena thermophila.. Genetics 141:1315-1325[Abstract].

NANNEY, D. L., 1964 Macronuclear differentiation and subnuclear assortment in ciliates, pp. 253–273 in Role of Chromosomes in Development (23rd Symposium, Society for the Study of Development and Growth), edited by M. LOCKE. Academic Press, New York.

NANNEY, D. L., S. J. REEVE, J. NAGEL, and S. DEPINTO, 1963  H serotype differentiation in Tetrahymena. Genetics 48:803-813[Free Full Text].

ORIAS, E., 1981  Probable somatic DNA rearrangements in mating type determination in Tetrahymena thermophila: a review and a model. Dev. Genet. 2:185-202.

ORIAS, E., 1986 Ciliate conjugation, pp. 45–84 in Molecular Biology of the Ciliated Protozoa, edited by J. G. GALL. Academic Press, New York.

ORIAS, E., 1998  Mapping the germ-line and somatic genomes of a ciliated protozoan, Tetrahymena thermophila.. Genome Res. 8:91-99[Abstract/Free Full Text].

ORIAS, E. and A. D. BRADSHAW, 1992  Stochastic developmental variation in the ratio of allelic rDNAs among newly differentiated, heterozygous macronuclei of Tetrahymena thermophila. Dev. Genet. 13:87-93[Medline].

ORIAS, E. and M. FLACKS, 1975  Macronuclear genetics of Tetrahymena. I. Random distribution of macronuclear gene copies. Genetics 79:187-206[Abstract/Free Full Text].

PREER, J. R., JR. and L. B. PREER, 1979  The size of macronuclear DNA and its relationship to models for maintaining genic balance. J. Protozool. 26:14-18.

PRESCOTT, D. M., 1994  The DNA of ciliated protozoa. Microbiol. Rev. 58:233-267[Abstract/Free Full Text].

SCHENSTED, I. V., 1958  Appendix: model of subnuclear segregation in the macronucleus of ciliates. Am. Nat. 92:161-170.

SONNEBORN, T. M., 1974 Tetrahymena pyriformis, pp. 433–467 in Handbook of Genetics, edited by R. C. KING. Plenum Press, New York.

SPANGLER, E. A. and E. H. BLACKBURN, 1985  The nucleotide sequence of the 17S ribosomal RNA gene of Tetrahymena thermophila and the identification of point mutations resulting in resistance to the antibiotics paromomycin and hygromycin. J. Biol. Chem. 260:6334-6340[Abstract/Free Full Text].

WICKERT, S. and E. ORIAS, 2000  Tetrahymena micronuclear genome mapping: a high resolution meiotic map of chromosome 1L. Genetics 154:1141-1153[Abstract/Free Full Text].

YAO, M.-C., 1989 Site-specific chromosome breakage and DNA deletion in ciliates, pp. 715–734 in Mobile DNA, edited by D. E. BERG and M. M. HOWE. American Society for Microbiologists, Washington, DC.

YAO, M.-C. and C. H. YAO, 1991  Transformation of Tetrahymena to cycloheximide resistance with a ribosomal protein gene through sequence replacement. Proc. Natl. Acad. Sci. USA 88:9493-9497[Abstract/Free Full Text].

YAO, M. C., A. R. KIMMEL, and M. A. GOROVSKY, 1974  A small number of cistrons for ribosomal RNA in the germinal nucleus of a eukaryote, Tetrahymena pyriformis. Proc. Natl. Acad. Sci. USA 71:3082-3086[Abstract/Free Full Text].

YAO, M.-C., C. H. YAO, and B. MONKS, 1990  The controlling sequence for site-specific chromosome breakage in Tetrahymena.. Cell 63:763-772[Medline].

YU, G. L., J. D. BRADLEY, L. D. ATTARDI, and E. H. BLACKBURN, 1990  In vivo alteration of telomere sequences and senescence caused by mutated Tetrahymena telomerase RNAs. Nature 344:126-132[Medline].

This article has been cited by other articles:

				D. Cassidy-Hanley, Y. Bisharyan, V. Fridman, J. Gerber, C. Lin, E. Orias, J. D. Orias, H. Ryder, L. Vong, and E. P. Hamilton Genome-Wide Characterization of Tetrahymena thermophila Chromosome Breakage Sites. II. Physical and Genetic Mapping Genetics, August 1, 2005; 170(4): 1623 - 1631. [Abstract] [Full Text] [PDF]

				N. D. Chilcoat, N. C. Elde, and A. P. Turkewitz An antisense approach to phenotype-based gene cloning in Tetrahymena PNAS, June 28, 2001; (2001) 151243498. [Abstract] [Full Text] [PDF]

				L. Wong, L. Klionsky, S. Wickert, V. Merriam, E. Orias, and E. P. Hamilton Autonomously Replicating Macronuclear DNA Pieces Are the Physical Basis of Genetic Coassortment Groups in Tetrahymena thermophila Genetics, July 1, 2000; 155(3): 1119 - 1125. [Abstract] [Full Text]

				S. Wickert and E. Orias Tetrahymena Micronuclear Genome Mapping: A High-Resolution Meiotic Map of Chromosome 1L Genetics, March 1, 2000; 154(3): 1141 - 1153. [Abstract] [Full Text]

				N. D. Chilcoat, N. C. Elde, and A. P. Turkewitz An antisense approach to phenotype-based gene cloning in Tetrahymena PNAS, July 17, 2001; 98(15): 8709 - 8713. [Abstract] [Full Text] [PDF]