The selfish yeast plasmid utilizes the condensin complex and condensed chromatin for faithful partitioning

Equipartitioning by chromosome association and copy number correction by DNA amplification are at the heart of the evolutionary success of the selfish yeast 2-micron plasmid. The present analysis reveals frequent plasmid presence near telomeres (TELs) and centromeres (CENs) in mitotic cells, with a preference towards the former. Inactivation of Cdc14 causes plasmid missegregation, which is correlated to the non-disjunction of TELs (and of rDNA) under this condition. Induced missegregation of chromosome XII, one of the largest yeast chromosomes which harbors the rDNA array and is highly dependent on the condensin complex for proper disjunction, increases 2-micron plasmid missegregation. This is not the case when chromosome III, one of the smallest chromosomes, is forced to missegregate. Plasmid stability decreases when the condensin subunit Brn1 is inactivated. Brn1 is recruited to the plasmid partitioning locus (STB) with the assistance of the plasmid-coded partitioning proteins Rep1 and Rep2. Furthermore, in a dihybrid assay, Brn1 interacts with Rep1-Rep2. Taken together, these findings support a role for condensin and/or condensed chromatin in 2-micron plasmid propagation. They suggest that condensed chromosome loci are among favored sites utilized by the plasmid for its chromosome-associated segregation. By homing to condensed/quiescent chromosome locales, and not over-perturbing genome homeostasis, the plasmid may minimize fitness conflicts with its host. Analogous persistence strategies may be utilized by other extrachromosomal selfish genomes, for example, episomes of mammalian viruses that hitchhike on host chromosomes for their stable maintenance.


Introduction
Eukaryotes, in contrast to prokaryotes, rarely harbor stably propagating extra-chromosomal DNA elements. However, circular plasmids have been identified in budding yeasts and in the slime mold Dictyostelium [1,2]. In addition, papilloma and gammaherpes viruses exist as episomes in infected mammalian cells during long periods of latency [3][4][5]. Eukaryotic nuclei also contain extra-chromosomal circular DNA molecules, called eccDNA that are excised from chromosomes [6][7][8]. Their formation, often associated with DNA replication/repair events, may be important in the evolutionary sizing and shaping of genomes. A subset of these circles has been associated with human diseases [9,10].
The multi-copy 2-micron plasmid of Saccharomyces cerevisiae, present as 40-60 molecules per haploid genome content, is the most well characterized among yeast DNA plasmids [11][12][13][14]. Current evidence suggests that plasmid molecules are organized into clusters of 3-5 foci that act as units of segregation during cell division [15][16][17]. The plasmid genome has a bi-partite functional organization, namely, a replication-partitioning module and a copy number amplification module. Plasmid molecules duplicated by the host replication machinery [18] are segregated equally (or nearly so) to daughter cells by the plasmid-coded Rep1 and Rep2 proteins acting in conjunction with the cis-acting STB locus [12][13][14]. The Flp site-specific recombination/amplification system housed by the plasmid compensates for copy number reduction resulting from rare missegregation events. Iterative copying of the plasmid template is triggered by a Flp-mediated recombination event coupled to bidirectional plasmid replication [19][20][21]. Mutually reinforcing regulatory mechanisms harbored by the plasmid [22][23][24][25] and imposed by the host [26-28] protect against any undue drop or runaway increase in plasmid population. Equal segregation and copy number optimization account for the evolutionary persistence of the 2-micron plasmid as a selfish DNA element that neither enhances nor significantly diminishes host fitness [29,30].
The current model for 2-micron plasmid propagation, supported by several lines of circumstantial evidence, posits plasmid segregation in physical association with chromosomes (hitchhiking) [16,31,32]. Furthermore, plasmid sisters formed by replication of a single copy STBreporter segregate in a one-to-one fashion, suggesting their association with sister chromatids [17,32]. This chromosome-like segregation is orchestrated by the Rep-STB system with the assistance of several host factors that also promote CEN-mediated chromosome segregation [17,31,[33][34][35][36][37][38][39][40]. It has been speculated that the unusually short, and genetically defined, point-CEN of budding yeast chromosomes and the plasmid-STB have diverged from a common ancestor that once directed both plasmid and chromosome segregation [37,41]. Extrapolation of the single-copy plasmid behavior to the native multi-copy plasmid foci suggests a highorder organization within each focus for coordinating plasmid replication with symmetric attachment of the replicated copies to sister chromatids.
Fluorescence-tagged STB-reporter plasmids have revealed their tendency to localize near CENs and spindle pole bodies in mitotic cells [15], and at or near TELs in meiotic cells [42]. Plasmid-TEL association is dependent on the meiosis-specific bouquet proteins Ndj1 and Csm1 [43][44][45][46][47]. The role of the mitotic spindle, the spindle-associated Kip1motor and the microtubule plus end binding Bik1 and Bim1 proteins in 2-micron plasmid segregation during mitosis [34,35,40] suggests spindle-mediated recruitment of the plasmid to chromosome sites in the vicinity of CENs. The association of the CEN-specific histone H3 variant Cse4 with STB [38], which is additionally detected at CEN-like regions (CLRs) [48], would also be consistent with the proximity between plasmid and CENs. The theme of plasmid segregation by attaching to chromosomes appears to be conserved during both mitosis and meiosis with variations in the mode of plasmid-chromosome association and in host factors responsible for such association as necessitated by the two distinct cell cycle programs.
The apparent differential plasmid localization on chromosomes in mitotic versus meiotic cells may, in principle, be reconciled by structural or organizational features common to CENs and TELs, and perhaps shared by other plasmid tethering sites on chromosomes. To address more definitively the still unresolved questions regarding chromosome locations utilized by the 2-micron plasmid for its association with chromosomes, we have now mapped the positions of a single-copy STB-reporter plasmid with respect to CEN V and TEL V (also TEL V and TEL VII) simultaneously. We find that the plasmid is located near CEN or TEL in � 65% of the mitotic cells (� 0.5 μm), with a clear TEL preference over CEN. Interestingly, 2-micron plasmid segregation is impaired when the disjunction of strongly condensed loci such as TELs or rDNA is blocked by Cdc14 inactivation. Consistent with this observation, induced missegregation of a long and highly condensin-dependent chromosome (Chr XII) has a stronger negative effect on equal plasmid segregation than the missegregation of a short and less condensin-dependent chromosome (Chr III). Genetic assays and chromatin immunoprecipitation (ChIP) analyses reveal the interaction of the condensin subunit Brn1 with the Rep-STB system, suggesting that condensin is an authentic plasmid partitioning factor appropriated from the host. Condensin requirement for plasmid segregation may be mandated by the need for multiple plasmid molecules within a focus to be condensed in order for them to co-segregate with a chromosome as one physical entity. Preferential localization at chromosome loci where condensin is enriched may reflect a plasmid strategy to satisfy this requirement. Alternatively, condensin acquisition by the clustered plasmid molecules, and presumably their condensation, may facilitate the tethering of plasmid foci to condensed chromosome locales. By associating with condensed, quiescent chromatin such as TELs, the plasmid may minimize potential deleterious effects on chromosome function and on the host's fitness.
as diploid cells enter the meiotic program [42,49]. These localizations were carried out primarily using fluorescence-tagged multi-copy STB-reporter plasmids [35,42], and a nearly single-copy reporter plasmid in one instance [40]. Visualization of the native 2-micron plasmid in mitotic cells using FISH also reveals a significantly higher number of plasmid foci unassociated with the nuclear membrane than those localizing with it [50]. Meiotic plasmid localization at the nuclear periphery increases from early meiosis to the pachytene stage, and requires the bouquet proteins Ndj1 and Csm4 [42]. In pachytene chromosome spreads, over half the plasmid foci are situated on chromosomes, predominantly at or near TELs [42]. The 2-micron plasmid may exploit chromatin features characteristic of CENs or TELs, or regions proximal to them, for its chromosome localization. Plasmid presence distal to these loci may be accounted for by similar chromatin features present at a subset of other chromosome regions or by alternative chromatin features recognized by the Rep-STB system. Differential plasmid localization with respect to the nuclear periphery during mitosis and meiosis may reflect changes in chromatin organization and/or altered spatial arrangements of chromosome locales between mitotic and meiotic cells.
In order to gain further insights into the chromosome-associated segregation of the 2-micron plasmid, we sought to map plasmid positions in mitotic cells at higher resolution by simultaneously fluorescence-tagging Chr V at CEN and the right arm TEL. Furthermore, to avoid any imprecision posed by multiple plasmid foci, we employed a previously described STB-reporter plasmid (pSG1) present at one copy (and as a single focus) in � 80% of the cells [17]. The segregation behavior of this plasmid is quite similar to that of an STB-reporter plasmid excised from a chromosome by site-specific recombination at G1, and therefore has a copy number of exactly one prior to DNA replication [16]. A unique advantage of pSG1, which also contains a GAL promoter-regulated CEN, is that it can be induced to behave as a CEN-plasmid, STB-plasmid or an ARS-plasmid (lacking partitioning activity) under appropriate experimental conditions (S1 Fig). When cells harboring pSG1 are grown in medium with glucose as the carbon source (to repress the GAL promoter), it is referred to as pSG1-CEN to indicate an active CEN. The CEN-mediated partitioning function of pSG1-CEN is not affected by the presence or absence of the native 2-micron plasmid in the host cells ([Cir + ] or [Cir 0 ], respectively). When present in [Cir + ] cells, and the carbon source is galactose, CEN is inactivated (by high level transcription from the GAL promoter), and the plasmid is referred to as pSG1-STB to denote its segregation by a functional Rep-STB partitioning system. When placed in [Cir 0 ] cells with galactose as the carbon source, the absence of the Rep proteins and lack of CEN function make the plasmid behave as pSG1-ARS in segregation (S1 Fig).
We distinguished CEN V and TEL V in cells expressing TetR-GFP by their distinct fluorescence intensities conferred by [TetO] 224 and [TetO] 448 arrays inserted at CEN V and TEL V loci, respectively [51,52] (S2 Fig). We verified that the two loci could be identified unambiguously at various stages of mitosis by marking their positions relative to the spindle pole body (SPB) tagged with Spc42-mCherry (S2 Fig). In S. cerevisiae, CENs remain congressed as a single cluster in close proximity to SPB during interphase as well as mitosis, while the rosette of three to six TEL clusters is positioned at the nuclear periphery distal to SPB in a Rabl-like pattern [53][54][55][56][57][58][59]. Consistent with this spatial organization, we observed partial overlap or close proximity of SPB to the fainter of the two GFP foci (CEN V), with the brighter GFP focus (TEL V) located away from it (S2 Fig). These three nuclear landmarks provided a frame of reference for mapping the spatial positions of reporter plasmids.

Mitotic localization of pSG1-CEN and pSG1-STB with respect to CEN V and TEL V
In order to follow plasmid localization in the nucleus, we introduced pSG1 into isogenic [Cir + ] and [Cir 0 ] strains containing the marked CEN V, TEL V and SPB and expressing cyan-LacI. Plasmid foci were visualized by repressor interaction with the plasmid-borne [LacO] 256 array [60]. As already noted, by using [Cir + ] and [Cir 0 ] backgrounds and manipulating the carbon source, we could induce pSG1 to behave as pSG1-CEN, pSG1-STB or pSG1-ARS (S1 Fig, [17]).
The localization of pSG1-CEN ([Cir + ]; glucose) at different stages of the mitotic cell cycle (Fig 1A and 1B) at or close to SPB (� 0.5 μm) was akin to that of chromosome CENs (S2 Fig), as expected. The spatial distribution of pSG1-STB ([Cir + ]; galactose), however, was different (Figs 2 and S3). In G1/S (Fig 2A and 2B) and in metaphase (Fig 2C and 2D) cells, 40-45% of the cyan foci were proximal to the TEL V (� 0.5 μm), 20-25% were close to CEN V (� 0.5 μm) and the rest showed no spatial linkage to either CEN V or TEL V (> 0.5 μm) (Fig 2A-2D). As the average copy number of pSG1 was only approximately one [17], not every cell in the population contained a single plasmid molecule prior to replication. Once replicated, pSG1-STB sisters would remain paired in most cells until anaphase onset, consistent with their bridging by the cohesin complex [16,31,39] and/or their symmetric attachment to cohesin-paired sister chromatids [32]. In 60-65% of metaphase cells, pSG1-STB was seen as a single focus of coalesced sisters. The cells scored for the distance estimates contained predominantly one pSG1-STB focus, but included a small number displaying two plasmid foci. A cell belonging to the latter subset was placed in the > 0.5 μm category even when only one focus showed this localization pattern (Fig 2C, row 3). All cells with more than two plasmid foci, indicating a copy number > 1, were excluded from the analysis.
As a more stringent test of the apparent tendency of pSG1-STB to remain TEL-proximal, we observed its localization in a selected subset of anaphase cells in which the CEN Vs were well separated from each other, and from the unsegregated TEL Vs that remained at the Taken together, the results from Figs 2 and S3 suggest that the 2-micron plasmid occupies TEL/CEN vicinities in well over half the cell population with a strong bias for TEL. At all stages of the cell cycle, at least a third of the cells showed no association of pSG1-STB with CEN V or , respectively. For those cells that showed two resolved plasmid foci, the � 0.5 μm criterion for proximity applied to both. Even when one plasmid focus failed this distance limit, the classification was > 0.5 μm. Bar, 5 μm.
https://doi.org/10.1371/journal.pgen.1009660.g002 TEL V (> 0.5 μm in Figs 2 and S3). In our assay, a plasmid mapped distal to TEL V may still be associated with a TEL cluster unmarked by the fluorescence tag. Conversely, due to TEL commingling, TEL V proximity of a plasmid represents its association with any one of the TELs present in the TEL V containing cluster. Localization distal to TEL/CEN would also be consistent with plasmid presence at other tethering sites distributed along chromosome arms.

Localization of pSG1-ARS
To verify the role of the Rep-STB system in directing plasmids to preferred chromosome locales, we examined how the CEN V-TEL V-based map positions are affected for a control ARS-plasmid.
For the pSG1-ARS plasmid, the proximal (� 0.5 μm from CEN V or TEL V) and distal (> 0.5 μm from CEN V and TEL V) locations were roughly evenly divided among G1/S cells (Fig 3). There was no significant statistical difference (Chi-square test; p = 0.0666, α = 0.05) from the localization of pSG1-STB (~65% CEN V/TEL V-proximal; Fig 2A). Viewed in isolation, this finding might suggest that the Rep-STB system does not direct plasmid localization in the nucleus. Conversely, localization has no role in plasmid segregation. However, this interpretation would seem untenable in light of a variety of prior observations summarized below. Furthermore, pSG1-ARS positions mapped by live cell imaging were essentially random, spread over a wide range from CEN V or TEL V (S4D Fig).
Unlike STB-plasmids, ARS-plasmids do not physically associate with chromosomes [12,31]. A previous study showed pSG1-CEN, pSG1-STB and pSG1-ARS to be present in � 90%, 80% and <20%, respectively, of chromosome spreads prepared from G1 cells [32]. Furthermore, Rep1, Rep2 and pSG1-STB form coalesced foci in these spreads [32]. Multi-copy STBplasmids also associate with mitotic and meiotic chromosome spreads in a Rep1-Rep2-dependent fashion [31,42]. Higher resolution of the meiotic spreads revealed plasmid foci, a considerable fraction of which is present on chromosomes at or near their tips, to be colocalized with a subset of the Rep protein foci [42]. The evidence for the functional relevance of chromosome association to STB-plasmid segregation is also compelling [15,16,31,32,42] (see also 'Discussion').
Given the striking differences between STB-and ARS-plasmids in physical chromosome association and chromosome-associated segregation, the localization of pSG1-ARS seen in Fig  3 is best explained as random plasmid-chromosome encounters within the crowded nucleus. ARS-plasmid proximity or remoteness to CEN-TEL in the Fig 3 plots is qualitatively consistent with the tightly congressed single CEN cluster (~20% proximal to CEN V), the less tightly compacted 3-6 TEL clusters (~30% proximal to TEL V) and the remainder of the chromosome mass (50% distal to CEN V-TEL V).

Association of the 2-micron plasmid with TEL VII
Unlike CENs, which are congressed into one tightly knit cluster in mitotic cells, TELs form a rosette of 3-6 clusters [53,54]. The premise of our assays (Figs 2 and S3) is that TEL V is equally likely to associate with any of these clusters. In order to probe potential variations among TELs in plasmid association, we assayed plasmid localization with respect to two TELs, TEL V and TEL VII (on the right arm of Chr VII), differentially tagged by In sum, the plasmid localization data would be consistent with TEL/CEN association constituting a major segregation route for the 2-micron plasmid in the chromosome hitchhiking model. Preferential plasmid segregation as a TEL appendage observed during meiosis [42] appears to hold for mitosis as well, a fact that was not appreciated from prior published results. While Rep1 and Rep2 proteins are essential for mediating plasmid-chromosome association in both mitosis and meiosis, the host proteins involved in plasmid linkage to TELs are likely different between the two cell cycle programs [42].

Blocking segregation of condensed chromosome loci causes STB-plasmid non-disjunction
In the budding yeast, TEL-proximal regions are rich in the condensin complex, and are organized as condensed, transcriptionally quiescent heterochromatin [61]. CENs act as cis-acting mediators to promote recruitment of condensin, cohesin and associated signaling molecules to pericentric regions, which have characteristic chromatin composition and organization [62][63][64][65]. Genome-wide mapping studies reveal quantitative differences in condensin occupancy along chromosome arms, as well as cell cycle dependent modulations in occupancy at individual loci [61]. In mitotic yeast cells, the rDNA gene cluster is also prominent in condensin enrichment. The non-random localization of an STB-plasmid on chromosomes, favoring the vicinity of TELs and CENs (Figs 2 and S3-S5), prompted us to examine a potential role for condensin and/or condensed chromatin in 2-micron plasmid segregation.
TELs and rDNA differ subtly from chromosomes as a whole in their segregation mechanism. In addition to early anaphase cleavage of cohesin that holds together sister chromatids in pairs, these loci require a second 'late' condensin-dependent step to complete unlinking and segregation [66]. The extra condensin recruitment is mediated by Cdc14 phosphatase, which is released from its sequestered state at the onset of anaphase by the FEAR (Fourteen Early Anaphase Release) network [67,68]. Upon inactivation of Cdc14, rDNA and TELs are delayed in disjunction relative to the rest of the chromosomes [69]. If TEL-proximity of the 2-micron plasmid is functionally relevant to its partitioning, perhaps by way of condensed chromatin, lack of Cdc14 may be expected to increase plasmid missegregation. We tested this prediction using a multi-copy STB-reporter plasmid pSV5-STB (Fig 4) that we had utilized in prior plasmid segregation assays [31]. The control reporter plasmids for this analysis were derivatives of pSV5 in which STB was either deleted (pSV6-ARS) or replaced by CEN (pSV7-CEN) (S6 and S7 Figs).
The pSV5 plasmid, fluorescence-tagged by [LacO]-[CFP-LacI] interaction, is organized into 2-5 foci per nucleus. Earlier experiments suggest that each plasmid focus is an independent unit in segregation [15]. As the number of plasmid molecules in individual foci is unknown, and occasionally foci tend to overlap, segregation estimates from counting foci numbers in mother and daughter nuclei are only semi-quantitative. Nevertheless, there is good agreement between data obtained from multi-copy and single-copy plasmid reporters in segregation assays [15][16][17].
In the cdc14-3 (Ts) strain at 33˚C (non-permissive), a significant fraction of the cells showed non-disjunction of TELs marked by [TEL V-TetO]-[TetR-GFP] (~35%; Type III + Type IV, Fig 4A) or the r-DNA locus marked by [rDNA-TetO]-[TetR-GFP] (~40%; Type III + Type IV, Fig 4B), as expected from prior studies [70]. In the wild type strain, these loci segregated normally in � 90% of the cells (Type I + Type II, Fig 4A and 4B). The corresponding equal segregation of~70% for pSV5-STB (Type I + Type III, Fig 4A and 4B), which is consistent with previous results for multi-copy and single-copy reporter plasmids [15,17], dropped to~50% in the cdc14 mutant (Type I + Type III, Fig 4A and 4B). The decrease in equal segregation due to Cdc14 inactivation, when normalized for the differences in the wild type host, was comparable between the STB-plasmid (~30%; Type I + Type III) and TEL V or rDNA (35-40%; Type I + Type II). Significantly, pSV5-STB and TEL V or rDNA were well correlated in their segregation, equal as well as unequal, in the wild type and mutant strains (Type I + Type IV >> Type II + Type III; Fig 4C and 4D). Note also that the fraction of Type II + Type III cells, representing plasmid segregation not correlated to TELV or rDNA segregation, was essentially unchanged by the cdc14 mutation. Or, the decrease in correlated equal segregation (Type I) in the mutant was matched nearly quantitatively by the increase in correlated missegregation (Type IV), as revealed in the near equality of Type I + Type IV between wild type and cdc14 in the plots in Fig 4C and 4D. In the majority of cells in which plasmid missegregation was coupled to the block in TEL V or rDNA segregation (Type IV, Fig 4A and 4B), the entire cluster of pSV5-STB foci, or a majority of the foci, was trapped in the mother or at the bud neck.
In contrast to pSV5-STB, pSV6-ARS [31], lacking STB, showed high missegregation in the wild type and cdc14-3 mutant strains (Type II + Type IV > Type I + Type III, S6A and S6B Fig). A characteristic feature of pSV5-STB, paucity of Type II + Type III cells (which signify contrary segregation between plasmid and TEL V or rDNA; Fig 4A and 4B), was not the case for pSV6-ARS (S6C and S6D Fig). In the wild type, Type II + Type III exceeded Type I + Type IV by~20%, and the difference was opposite in the cdc14 mutant. As non-disjunction tends to trap TEL V and rDNA in the mother nucleus, their segregation in the mutant approximates that of ARS-plasmids typified by their strong intrinsic mother bias [71]. Consequently, the shift from wild type to cdc14-3 switches the distribution towards Type I + Type IV at the expense of Type II + Type III. The histogram plots for the quantitative estimates of the cell types are shown below the respective cell type panels. The differences in plasmid segregation correlated with TEL V or rDNA segregation (Type I + Type IV) and uncorrelated with the segregation of these loci (Type II + Type III) are underscored in the plots at the bottom (C, D). The data were obtained from more than 100 and 160 cells in each assay set for the wild type and cdc14-3 cells, respectively. Statistical significance was estimated by Fisher's exact test. Bar, 5 μm. https://doi.org/10.1371/journal.pgen.1009660.g004

PLOS GENETICS
The pSV7-CEN plasmid segregated equally in most of the cells irrespective of the strain background (Type I + Type III >> Type II + Type IV; S7A and S7B Fig). As expected for normal CEN function, correlated segregation of pSV7-CEN with TEL V or rDNA was high in the wild type strain (Type I + Type IV >> Type II + Type III) (S7C and S7D Fig). Consistent with the adverse effect of Cdc14 inactivation on TEL V or rDNA segregation without disruption of general chromosome segregation, this strong correlation was absent in the cdc14 mutant (Type I + Type IV < Type II + Type III) (S7C and S7D Fig).
To further verify the apparent distinction between an STB-plasmid and an ARS-or a CENplasmid in segregation under Cdc14 inactivation, we followed the single copy reporter plasmids (pSG1-STB, pSG1-CEN and pSG1-ARS) with respect to Nop1, a nucleolar histone glutamine methyl transferase, as an rDNA surrogate in the cdc14 mutant strain (S8A and S8B Fig). Consistent with the results from Figs 4, S6 and S7, pSG1-STB alone showed the correlation with Nop1 in segregation (Type I + Type IV >> Type II + Type III) (S8C Fig; see also Fig 4). Segregation of pSG1-CEN was characterized by Type I +Type III (equal) >> Type II + Type IV (unequal), while the opposite was true for pSG1-ARS (Type II + Type IV > Type I + Type III) (S8D Fig).
Thus, Cdc14-assisted segregation of TEL V or rDNA also impacts plasmid segregation directed by the Rep1-Rep2-STB system. Neither CEN-mediated equal plasmid segregation (and by inference global chromosome segregation) nor high plasmid missegregation in the absence of a partitioning system is affected by the cdc14 mutation. These distinctions are best appreciated by the sharp differences in the p values for pSG1-STB in Fig 4C and 4D

Missegregation of Chr XII, but not Chr III, causes missegregation of the 2-micron plasmid
Faithful chromosome segregation in yeast, as in all eukaryotes, requires chromosome condensation, which facilitates compaction of chromosome arms, unlinking of catenated sister chromatids, and their organization into functional segregation units [72]. While condensation is easily visualized in the large chromosomes of plants and metazoans, more sensitive tools utilizing FRET or fluorescence quenching are needed to monitor condensation of the relatively short yeast chromosomes [64]. However, condensation of the rDNA locus on Chr XII (comprised of~150 copies of a 9.1 kbp repeat unit; 1.5 Mbp DNA) has been revealed by standard microscopy in conjunction with FISH [73]. As already pointed out, segregation of the rDNA and TEL loci requires a Cdc14-dependent extra step of condensin recruitment and action. Interruption of this step perturbs the segregation of these loci and of the 2-micron plasmid, while chromosome segregation overall proceeds almost normally.
When gross missegregation of yeast chromosomes is induced by a conditional mutation, the majority of foci formed by a fluorescence-tagged multi-copy STB-plasmid segregates to the nuclear compartment containing the bulk of the chromosome mass [15,31]. When the entire set of replicated sister chromatids is forced into either the mother or the daughter nucleus, a single-copy STB-plasmid is almost always localized in the chromosome-containing nucleus [32]. These observations are consistent with plasmid-chromosome interaction sites being distributed on more than one (perhaps all) chromosomes. However, the correlation between plasmid and Nop1 (rDNA) during segregation under Cdc14 inactivation suggests that missegregation of Chr XII, housing the rDNA array and critically dependent on condensin for segregation, is likely to affect 2-micron plasmid segregation more severely than missegregation of a smaller chromosome with less stringent dependence on condensin. We tested this hypothesis by following plasmid behavior in cells missegregating either Chr XII (one of the longest yeast chromosomes) or Chr III, which is roughly one-third as long.
The pSV5-STB plasmid was introduced into isogenic strains in which either CEN III or CEN XII could be inactivated conditionally by GAL promoter driven transcription through them [74]. The extent of induced missegregation was nearly the same for Chr III and Chr XII (S9 Fig). The expression of GFP-LacI and Nop1 (DsRed) in these strains permitted plasmid segregation and Chr XII segregation (by Nop1 as proxy) to be monitored simultaneously against chromosome segregation as a whole visualized by DAPI staining. Missegregation of Chr XII, but not of Chr III, resulted in a significant increase in plasmid missegregation (Type II + Type IV > Type I + Type III) (Fig 5A-5D). Furthermore, in cells showing missegregation of both pSV5-STB and Nop1 (Type IV; Fig 5A), the majority of plasmid foci was present in the nucleus containing Nop1.
The linkage of the 2-micron plasmid with Chr XII but not Chr III is consistent with differences between these two chromosomes in the amount of condensin complex and/or the degree of chromatin condensation needed for their segregation. In the hitchhiking model, such differences may lead to preferred plasmid association with Chr XII. A potential plasmid target on Chr XII is the highly repeated rDNA locus, whose condensin utilization exceeds that of most other chromosome loci, except perhaps TELs. The size advantage of chromosome XII over chromosome III should have only minimal effect on plasmid association/segregation, as it would be offset by the fifteen normally segregating chromosomes in each of the assays.

Absence of condensin function disrupts faithful STB-plasmid segregation
The potential functional relationship between condensed chromatin and the 2-micron plasmid in segregation suggests two possible roles, neither one incompatible with the other, for the condensin complex in plasmid physiology. The multiple plasmid copies present within a chromosome-associated plasmid focus, which appears to be the unit in segregation [14,15], may need to be condensed as a prerequisite for segregation. Or condensin may facilitate preferential plasmid association, and segregation in concert, with condensed chromosome loci. In support of the latter role, condensin is known to bridge DNA loci that are distal to each other on the genome. For example, it binds to tRNA genes dispersed throughout the chromosomes, and promotes their congregation into a cluster at the nucleolus [75]. In order to probe potential condensin requirement for 2-micron plasmid segregation, we compared the rates of missegregation of STB-plasmids with those of CEN-or ARS-plasmids following inactivation of the condensin component Brn1.
At 36˚C, the loss of the STB-plasmid (pRS422; ADE2) and the CEN-plasmid (pRS412; ADE2) [76] was elevated in the temperature sensitive brn1-60 mutant [77] over the wild type ( Fig 6A). The CEN-plasmid instability is consistent with the previously demonstrated role of condensin in CEN function [62]. The inherently low stability of an ARS-plasmid (YRp17, TRP1) [78] was not further worsened in the mutant background (Fig 6A). These stability results were further verified by assaying the segregation of fluorescence-tagged STB-reporter and CEN-reporter plasmids (pSV5-STB and pSV7-CEN, respectively) in the wild type and mutant strains (Fig 6B). A fluorescence-tagged ARS-plasmid (pSV6-ARS), by contrast, was not altered in its segregation by the loss of condensin function (Fig 6B).
These results demonstrate the requirement of condensin during STB-and CEN-mediated plasmid segregation. However, they do not address whether condensin plays similar or distinct roles in the segregation functions of STB and CEN.

Brn1 interacts with the 2-micron plasmid partitioning system
Since condensin is required for chromosome segregation, and the 2-micron plasmid is dependent on chromosomes for its own segregation, it is difficult to decipher whether condensin's effect on the plasmid is direct or indirect. Proximity of the plasmid to condensed chromosome loci may lead to incidental plasmid-condensin association, and likely plasmid condensation as a result. Alternatively, condensin acquisition by the plasmid (and perhaps condensation) may be functionally related to the localization of a plasmid focus at condensed chromatin and its chromosome-associated segregation. One reasonable criterion for the direct involvement of condensin in 2-micron plasmid segregation would be its interaction with the Rep-STB system. Such interactions by host factors that assist plasmid partitioning have been demonstrated in previous studies through genetic interaction, chromatin immunoprecipitation (ChIP), and affinity enrichment-mass spectrometry assays [31, 34, 36, 38, 40]. We have now subjected condensin to similar tests using the Brn1 subunit of the complex as its representative.

PLOS GENETICS
Interaction between STB and Brn1 was ascertained by the enrichment of STB DNA during chromatin immunoprecipitation (ChIP) in a [Cir + ] strain using anti-HA antibodies directed to Brn1-6HA (Fig 7A and 7B). The authenticity of Brn1-STB association was verified by the enrichment of CEN III, but not the tubulin locus TUB2, in the immunoprecipitate. TUB2 has been designated as weak or negative in condensin localization [61]. Consistent with the higher missegregation of STB-plasmids under Cdc14 inactivation (Figs 4 and S8), ChIP in the cdc14-3 mutant revealed a nearly complete absence of condensin at STB (Fig 7A and 7B). ChIP in a [Cir 0 ] strain (lacking Rep1 and Rep2) failed to bring down a copy of STB integrated into the . STB (as a potential target for Brn1 association) was provided by the endogenous 2-micron plasmid in the [Cir + ] strain. Brn1-6HA was immunoprecipitated using an antibody to the HA-tag. The ΔCT values between an experimental sample and the input DNA were derived from qPCR data after correcting for primer efficiency/amplification factor [109]. ChIP efficiency was quantitated as the fraction of the input DNA that was immunoprecipitated [ε^(-ΔCT)], where ε is the amplification factor). CEN3 and TUB2 loci were used as positive and negative controls for Brn1 binding, respectively. The gel images in (A) show ethidium bromide stained bands following fractionation of DNA amplified from immunoprecipitates by regular PCR. The graphs in (B) represent three independent experiments. The estimated p values are from two-tailed Student's t-test.
Rep1 and Rep2-mediated association of Brn1 with STB (Figs 7 and 8) suggests possible interaction of Brn1 with one or both of these partitioning proteins as implicated by unpublished observations mentioned in an earlier publication [15]. Results from two hybrid assays performed in a [Cir + ] strain are consistent with this notion (S10 Fig). In principle, interaction with either Rep1 or Rep2 should result in Brn1 recruitment at STB, as Rep1 and Rep2 interact

PLOS GENETICS
with each other [79][80][81][82] (S10 Fig). These two-hybrid results do not address whether Brn1 interaction with a Rep protein is dependent on the partner Rep protein or on host factor(s), including other component(s) of the condensin complex. Nevertheless, Brn1 interaction with at least one of the Rep proteins and the dependence of Brn1-STB association on both Rep proteins argue for condensin being an authentic host-coded 2-micron plasmid partitioning factor. The loss of Brn1 from STB upon Cdc14 inactivation and the rise in STB-plasmid instability upon Brn1 or Cdc14 inactivation bolster this possibility. The appropriation of the host's DNA compaction and/or high-order DNA organization factors (condensin and cohesin, for example) by the 2-micron plasmid would be consistent with the evolutionary attunement of plasmid segregation to chromosome segregation.

Discussion
We have characterized chromatin attributes exploited by the 2-micron plasmid for its localization and segregation during mitosis. Our findings suggest that condensed or condensinenriched chromatin and/or silent heterochromatin-like regions of chromosomes are preferred targets for plasmid-chromosome association. Furthermore, the condensin complex interacts with the plasmid partitioning system, and facilitates equal plasmid segregation. A role for condensin/condensed chromatin in plasmid localization and segregation is consistent with the hitchhiking model, in which the two processes are intimately interconnected.

2-micron plasmid localization on yeast chromosomes
The chromosome-hitchhiking model for 2-micron plasmid segregation [13,16,32,42] is based on a confluence of circumstantial, albeit persuasive, evidence. Strictly, plasmid association with a nuclear entity that segregates with the same characteristics as chromosomes cannot be ruled out. Direct demonstration of plasmid-chromosome association is impeded by the inability to resolve individual mitotic yeast chromosomes. Nevertheless, multi-copy and single-copy STB-reporter plasmids localize to yeast mitotic chromosome spreads in a Rep1-Rep2 dependent fashion, and colocalize with these proteins [31,83]. In mammalian cells, coexpressed Rep1 and Rep2 form merged foci on mitotic chromosomes collectively, and along the arms of individual chromosomes in metaphase spreads [32]. Based on Rep1-Rep2 association with STB [15,31,80], and assuming the mammalian system to recapitulate the native yeast system, it follows that the 2-micron plasmid is tethered to yeast chromosomes with assistance from the Rep proteins. Meiotic yeast chromosome spreads, with considerably higher resolution than mitotic spreads, reveal STB-plasmid foci to be coalesced with a subset of Rep foci localized on chromosomes [42]. The preferential plasmid localization at TELs or subtelomeric regions in these spreads differs from the reported CEN proximity of plasmid in mitotic cells [31]. Higher accuracy mapping has now localized the 2-micron plasmid near CENs and TELs in a majority of the cell population, with a clear bias towards TELs (Figs 2 and S3). In over a third of the population, the plasmid location is distal to CENs or a marked TEL. It is possible that these 'other' plasmid tethering sites on chromosomes (which may include TELs that are not clustered with the marked one) share organizational features with CEN and TEL, for example, chromatin architecture and/or transcriptional quiescence.

Functional link between STB-plasmid localization and segregation
Several observations suggest that Rep1-Rep2-assisted association of STB-plasmids precedes their chromosome-coupled segregation. Segregation of single-copy or multi-copy STB-plasmids is strongly correlated with that of chromosomes under gross chromosome missegregation induced by conditional mutations [15,31], total entrapment of paired sister chromatids in mother or daughter nucleus by blocking cohesin disassembly [32], or under forced co-segregation of paired sister chromatids by inappropriate expression of the meiosis-specific monopolin complex during a mitotic cell cycle [16]. This correlated chromosome-plasmid segregation is lost in the absence of the Rep proteins (when a reporter plasmid behaves functionally as an ARS-plasmid). Equal plasmid segregation during meiosis is dependent on Rep1-Rep2-STB and on the bouquet protein Ndj1, which promotes plasmid association with telomeres [42].
The present study shows that the relationship between localization with respect to CEN-TEL (Figs 2 and 3) and segregation is quite different for ARS-and STB-plasmids. For ARSplasmids, localization does not foster plasmid stability, which is quite low in wild type strains and remains low under treatments that alter chromosome segregation in specific ways (Figs 6  and S6). Blocking or delaying the disjunction of TEL DNA/rDNA by cdc14 does not grossly disrupt CEN disjunction and hence bulk chromosome segregation. However, compromising CEN function/chromosome condensation by brn1 impairs global chromosome segregation. In contrast to ARS-plasmids, STB-plasmids are sensitive to both cdc14 and brn1 (Figs 4 and 6). STB-plasmids also differ from CEN-plasmids whose segregation is affected by brn1 but is relatively insensitive to cdc14 (Figs 6 and S7).
The observed responses of STB-plasmids to distinct types of chromosome segregation defects, which set them apart from ARS-plasmids and also differentiate them from CEN-plasmids in certain respects, are consistent with their attachment to chromosome sites and such attachment being salient to plasmid segregation.

A role for chromatin organization in the selection of plasmid tethering sites?
The 2-micron plasmid partitioning system may recognize a class of chromosome regions, including TELs and CENs, by their heterochromatin signature(s). The non-canonical heterochromatin of S. cerevisiae promoted by Sir proteins [84][85][86][87] induces transcriptional silencing at TELs, rDNA and HML/HMR mating type loci, and enables CEN-mediated mitotic chromosome stability. Despite being genetically defined, the point CENs of S. cerevisiae manifest a certain degree of epigenetic character, which is typical of regional CENs. Established CENs are functionally propagated through many generations in budding yeast strains carrying kinetochore mutations that block de novo CEN establishment [88]. The TEL-and CEN-proximal localization of the 2-micron plasmid is consistent with a possible origin of CEN from TEL [89]. They do share organizational or functional features in certain biological contexts. In the fission yeast, CEN-centrosome contact can replace TEL-centrosome contact for successful meiosis [90]. The plasmid presence at CEN-TEL may be promoted by epigenetic mark(s) conserved during their evolutionary trajectories, which might also be retained within CEN-adjacent CEN-like regions (CLRs) [48].

2-micron plasmid missegregation upon Cdc14 inactivation
The functional relevance of TEL proximity of the 2-micron plasmid on chromosomes is upheld by the increase in plasmid missegregation when TEL (and rDNA) disjunction is impeded by inactivating Cdc14. However, total missegregation of all plasmid foci in tandem with TEL or rDNA (represented by Nop1) is rarely seen (Fig 4). This is the expected outcome if plasmid tethering sites were present at chromosome locales other than TELs, including CEN-proximal regions, as revealed in the localization assays. According to a prior study, when the entire set of chromosomes is confined to either the mother or daughter nucleus, there is near perfect correlation between an STB-plasmid and chromosomes in their localization [32].
Taken together, these results support a non-random distribution of plasmid tethering sites on chromosomes, TELs or nearby loci being high density or high affinity sites. Cdc14 inactivation has pleiotropic consequences, one of which is compromised spindle integrity [91,92]. The mitotic spindle and spindle-associated proteins play a role, even if indirect, in 2-micron plasmid segregation [31,34,35]. Hence, potential contribution of spindle defect to plasmid missegregation under loss of Cdc14 function cannot be ruled out. However, such an effect is likely modest, as overall chromosome segregation (revealed by DAPI staining) appears to be normal in our assays. Furthermore, STB plasmid missegregation can also be induced by inactivating the condensin component Brn1, which is not known to be involved in spindle function.

A role for the condensin complex in 2-micron plasmid segregation
As noted, Cdc14-promoted condensin recruitment late in the cell cycle at rDNA and TELs is a pre-requisite for their proper segregation [68,70]. The possibility that this instalment of condensin resolves rDNA sisters independently of chromosome condensation and Topo II activity has been raised [70]. However, full condensation of rDNA, approximately equal to that of condensed euchromatin, is completed only in anaphase [68]. Resolution of the interlinks between rDNA sisters by recruitment of Topo II in a condensin dependent step, or remodeling of the interlinks by condensin to a configuration favored by Topo II, has not been ruled out [68,93]. The increased missegregation of the 2-micron plasmid in a condensin mutant (Fig 6) may result from a direct effect of condensin on the plasmid itself, or an indirect effect manifested through chromosomes on which it hitches a ride.
By being a chromosome appendage, the plasmid becomes subjected to the same physical and conformational constraints that a chromosome experiences during segregation. Chromosome condensation and plasmid condensation may thus be coincidental events. However, the interaction of Brn1 with the Rep proteins and its localization at STB with Rep protein assistance (Figs 7,8 and S10) would be consistent with a more direct role for condensin in plasmid segregation. Interestingly, Brn1 association with STB is reduced by Cdc14 inactivation (Fig 7). Perhaps Cdc14 dependent condensin assembly is shared by repeated loci that include rDNA and TELs on chromosomes and the extra-chromosomal (but chromosome-associated) 2-micron plasmid. Cohesin-mediated pairing of plasmid sisters [31, 39], in conjunction with possible condensin-assisted DNA compaction, may promote the organization of replicated plasmid molecules into two equally populated sister clusters that tether symmetrically to sister chromatids. Such a mechanism is consistent with the segregation pattern observed for singlecopy STB plasmids whose sister copies segregate one-to-one in most cells [12,17,32]. Condensin associated with STB may also serve to recruit Topo II to the plasmid and/or assist decatenation of plasmid sisters. Topo II inactivation causes an accumulation of replicated 2-micron plasmids as catenanes [94].

Chromosome association by selfish DNA elements
The high fidelity of chromosome segregation is likely the driving force behind the convergence by the yeast plasmid and viral episomes, whose respective hosts diverged~1.5 billion years ago, on exploiting chromosome segregation for self-preservation [3,14,[95][96][97]. The TEL-and CEN-proximal plasmid localization in mitotic cells suggests a propensity for plasmid tethering sites to be sparsely populated by genes or organized into silent chromatin. It is noteworthy that the portion of STB proximal to the 2-micron plasmid replication origin, and critical to plasmid partitioning, resides in a transcription-free zone [98]. The origin-distal portion of STB contains a 'silencer box', and is important in protecting STB-distal from readthrough of neighboring transcription. Furthermore, STB can functionally replace the chromosomal silencer sequences adjacent to the HML yeast mating type locus [99]. Strikingly, the viral tethering sites on mammalian chromosomes also appear to be enriched for condensed/repressive chromatin that include pericentromeric, peritelomeric and rDNA loci [95,[100][101][102][103][104][105][106]. Such localization may be advantageous to a selfish DNA element by lowering the prospect of its dislodgement from chromosomes due to gene activity and by protecting the host against perturbations of gene functions.
The present studies, however, do not rule out other chromatin features that might be utilized by the 2-micron plasmid for chromosome tethering. A potential additional plasmid strategy for targeting more readily accessible chromatin may account for TEL/CEN-distal plasmid localization. Maximizing its options to remain chromosome-associated would be beneficial to the plasmid, and to any extra-chromosomal element, that propagates itself by chromosome hitchhiking.

Yeast strains and plasmids
The yeast strains used in this study (W303 background) and their relevant genotypes are listed in S1  [107]. Gene fusion and expression of the tagged proteins were verified by PCR using total extracted DNA as template, and by immunostaining chromatin spreads with HA-antibody, respectively.
The plasmid constructs employed in this work and their salient features are summarized in S1 Table. The plasmid pRS402CFP-LacI (ADE2) was obtained by replacing the coding region for GFP in pRS402GFP-LacI [15] with that of CFP. as template) flanked by chromosome homology at either end was inserted between IMA1 and MAL13 genes on the right arm of Chr VII (~22 kbp from the end). Viable colonies obtained following transformation were screened by fluorescence microscopy to identify successful integrants. Based on the sizes of PCR products amplified using isolated total DNA, the number of [TetO] copies was � 50, and varied in individual transformants. One transformant showing two green fluorescent foci (bright TEL V and faint TEL VII) in most individual cells was saved, and used for further analyses.

Cell cycle synchronization
Cells were arrested in G1/S with α-factor, and released into the cell cycle by washing off the pheromone. Reporter plasmids were localized in the arrested cells and following their release. Cells harvested at intervals during cell cycle progression were examined by microscopy. Metaphase and anaphase cells were identified by their characteristic morphological features, as described previously [17,40].

Plasmid stability assays
Cultures of transformant colonies containing a reporter plasmid were grown overnight in selective liquid medium, and were diluted into YPDA medium (non-selective). The inocula were grown for 'n' generations in non-selective media (n varied from 5 to 10, depending on the experiment). The fractions of plasmid containing cells in the population at the start (f 0 ) and at the end (f n ) of growth in the non-selective media were estimated by plating equal aliquots of the cultures on selective and non-selective plates. The plasmid loss rate per generation was estimated as 'i' = (1/n) x ln(f 0 /f n ) [98].

Plasmid segregation under induced missegregation of chromosomes
For inducing Chr III or Chr XII to missegregate in galactose medium, the GAL1 promoter was inserted immediately upstream of CEN III or CEN XII as described previously [74]. Cells harboring the GAL-CEN cassette were grown till mid-log phase in SC (-Leu, -Trp, raffinose) medium at 30˚C before being transferred to SC (-Leu,-Trp) medium supplemented with either 2% galactose or 2% glucose for 5 hr at 30˚C. Cells were harvested, stained with DAPI, and reporter plasmids ([LacO]-[GFP-LacI]) and rDNA (DsRed-Nop1) were observed by fluorescence microscopy to follow the segregation pattern of plasmid with respect to rDNA. In strains lacking a reporter plasmid, segregation of Chr III was assayed by fluorescence-tagging at the LEU2 locus using [LacO]-[GFP-LacI]; that of Chr XII was followed by tracking rDNA marked with DsRed-Nop1.

Di-hybrid assay
The strain for di-hybrid assays contained the HIS3 reporter gene regulated by the GAL1 UAS (upstream activation sequence) [108]. The bait and prey proteins to be tested were fused to the Gal4 DNA binding and Gal4 activation domains, respectively, and their expression was controlled by the ADH1 promoter. Positive bait-prey interaction resulting in enhanced HIS3 transcription was revealed by colony growth on medium lacking histidine and supplemented with 0.25 mM 3-AT (3-amino-1,2,4-triazole) to inhibit His3 activity resulting from basal expression. A positive response by a pair of test proteins was verified by reciprocally swapping the DNA binding and activation domains between them. Expression of the binding or activation domain alone (without fusion to the bait or prey) provided the negative controls. Interaction between Rep1 and Rep2 was used as positive control.

Chromatin Immunoprecipitation assays
Chromatin Immunoprecipitation (ChIP) assays were performed as described previously [109,110]. The STB from the native 2-micron circle provided the potential target for Brn1 association in the [Cir + ] strain. In the [Cir 0 ] strain used for ChIP, a copy of STB was integrated into the genome at the ADE2 locus. Derivatives of this strain were constructed to express either Rep1, Rep2 or Rep1 and Rep2. Expression was driven by the ADH1 promoter from cassettes integrated at the TRP1 locus (for REP1) and URA3 locus (for REP2 or REP1 and REP2). Brn1 fused to 6HA at the carboxyl-terminus was immunoprecipitated using a polyclonal antibody to HA (Rabbit polyclonal HA11, Abcam, UK). The specificity of the association was verified using a chromosomal locus 'TUB2' designated to be weak or negative in condensin association [61]. Brn1 with no tag and mock immunoprecipitation (no antibody) provided additional negative controls. The relative amounts of immunoprecipitated DNA were quantitated using qPCR [110]. Corrections were applied to account for primer efficiencies below 100% (amplification factors < 2.0) using standardization graphs of CT values against dilutions of the input DNA. The amplification factor ε was estimated as 10^(-1/slope) of the regression line, and the primer efficiency E (%) as {[10^(-1/slope)]-1} x 100. The fraction of immunoprecipitated DNA in a ChIP sample relative to the input DNA was calculated as ε^(−ΔCT), ΔCT = CT (ChIP)-[CT (Input)-log ε (Input dilution factor)]. The primers used for PCR amplification are listed in S2 Table under 'Supporting Information'.

Fluorescence microscopy and distance measurement
Fluorescence microscopy was performed in live or mildly fixed cells. Fixing was performed in p-formaldehyde (4% v/v) by incubating on ice for 5-10 min. After two washes with 0.1 M phosphate buffer (pH 7.4), cells were imaged using a Zeiss Axio Observer Z1 fluorescence microscope and the Axiovision software. Images were acquired using the z-stack mode, with a 0.20 μm interval between two successive planes [111]. They were opened in Imaris software (Bitplane, Imaris 8.0.2), and 'slice tool' was utilized to measure the spacing between the centroids of two fluorescent foci [112]. The distances were grouped into two types, � 0.5 μm and > 0.5 μm.

Segregation of fluorescence-tagged multi-copy reporters
The method for estimating segregation of multi-copy fluorescence-tagged reporter plasmids (Figs 4, 5 and S6-S8) was similar to that described previously [15]. Plasmid foci in the two nuclear compartments of an anaphase cell were counted by scanning stacks along the Z-axis, and were categorized as 'equal' or 'unequal'. These estimates are subject to some intrinsic error from the potential overlap of foci and variations in plasmid copy number among foci. The plots in (C) highlight the distinctions between pSG1-STB and pSG1-CEN or pSG1-ARS in Nop1-correlated segregation (Type I + Type IV) and Nop1-uncorrelated segregation (Type II + Type III). In (D), overall plasmid segregation, Nop1-correlated or not, is displayed equal (Type I + Type III) or unequal (Type II + Type IV). For each plasmid type, over 200 cells were analyzed. Pairwise comparisons of pSG1-STB with pSG1-ARS or pSG1-CEN (C) by Fisher's exact test showed statistical significance in each case (p < 0.0001). This was not the case for the pSG1-ARS and pSG1-CEN pair, p = 0.392. The Type I-Type IV distributions were statistically indistinguishable for the pSG1 plasmids (this figure)  In the experimental strains, lacking a fluorescence-tagged reporter plasmid, the GAL1 promoter-controlled CEN III or CEN XII was kept active or inactive in glucose or galactose, respectively. The strain with the conditionally active CEN III was engineered to express GFP-LacI, and contained a [LacO] 256 insertion within Chr III. The strain with the conditionally active CEN XII was engineered to express fluorescence-tagged Nop1 (DsRed). The images show equal segregation and missegregation of Chr III and Chr XII, as followed by green and red fluorescence, respectively. B. The data from the analysis of at least 120 cells for each assay are plotted. Bar, 5 μm. (TIF) S10 Fig. Brn1 interacts with Rep1 and Rep2 in dihybrid assays in [Cir + ] strains. The dihybrid system based on the HIS3 reporter [108] is schematically diagrammed at the top (AD, activation domain; BD, binding domain). The expression vectors for ADH1 promoter-controlled AD-and BD-fusions were maintained by selection for the LEU2 and TRP1 markers, respectively, that they harbored. The indicated protein-protein interactions were assayed on SC-Leu-Trp-His and SC-Leu-Trp-His plates with or without added 3-AT. The interaction between Rep1 and Rep2 served as the positive control (last two rows of the top panel). The empty AD and BD vectors are indicated by '-'. (TIF) S1