Barbara McClintock’s Unsolved Chromosomal Mysteries: Parallels to Common Rearrangements and Karyotype Evolution

DEFICIENCY RECOVERY

What is deficiency recovery? Stadler reported that x-ray-induced or natural deficiencies that remove multiple genetic markers on a chromosome arm, as revealed by adjacent dominant alleles missing in most of the plant or endosperm, could have a small sector of the plant or endosperm in which the multiple dominant markers reappear (Stadler, 1930; Figure 1). It was clear from the phenotype that the deficiencies were missing parts of chromosomes, so how was it possible that the dominant markers could reappear in developmental sectors? This phenomenon was described before DNA was realized to be the genetic material and the nature of genes was known. Indeed, even after all that is understood today about chromosomes, it still seems mysterious.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Deficiency Recovery.

The kernel at the right illustrates an apparent spontaneous case of deficiency recovery. The kernel at the left has a recessive genetic marker for anthocyanin production in the endosperm (c1) and is representative of the maternal parent. The center kernel has the dominant allele for the pigment gene (C1) and is representative of the male parent. In the kernel at the right, the dominant marker is missing on most of the endosperm but a small sector with light pigmentation typical of a c1/c1/C1 endosperm is present. The dominant marker must have been delivered to the zygote but is missing in much of the endosperm and recovers only in a sector. (Photo by Zhi Gao.)

If we look back at the phenomenon now, can we imagine a basis for it? A potential explanation might come from a recent examination of other material from Stadler’s work. Stadler and Roman (1948) reported the discovery of a small chromosome called Duplication 3a. This chromosome arose in material in which pollen with dominant markers was UV-irradiated and crossed to a recessive a1 sh2 tester. Mosaicism for this pigment marker (a1) and starch content marker (sh2), which are closely linked, signaled the presence of an unstable chromosome. The instability of this chromosome led to the speculation that it might be a ring chromosome (Stadler and Roman, 1948).

However, a recent examination of this chromosome revealed that it is not in fact a ring, having telomeres on both ends, and that it has no detectable canonical centromere DNA (Fu et al., 2013). ChIP-seq studies using antibodies against the histone variant typical of active centromeres (CENH3) indicated that the chromosome possessed a de novo centromere (Fu et al., 2013). Other studies have revealed additional cases of de novo centromeres in plants (Nasuda et al., 2005; Topp et al., 2009; Zhang et al., 2013; Liu et al., 2015; Guo et al., 2016), indicating that their occurrence is not that rare (compared with spontaneous mutation, for example). Indeed, analysis of many lines of maize found great diversity for the exact positioning of CENH3, indicating a natural tendency for centromere inactivation (or deletion) with recovery via de novo centromere formation (Schneider et al., 2016).

Looking back at the phenomenon of deficiency recovery in light of these recent revelations might suggest that it results from de novo centromere formation on the fragment produced in the generation of the deficiency. If a chromosome is fractured by experimentally induced or natural causes, the remaining portion with a normal centromere will be inherited as usual to daughter cells. One could speculate that the acentric fragment could languish through some mitoses into one of two daughter cells, as documented by McClintock for acentric fragments produced from inversion heterozygotes, but upon formation of a de novo centromere at some point along its length, it would then subsequently be partitioned equally into both daughter cells in a portion of the remaining developmental lineages. The frequency of de novo centromere formation was realized only recently; previously, it was thought that acentric fragments were certainly destined to be lost. Can de novo centromere formation explain deficiency recovery? This possibility seems reasonable given the recent documentation of de novo centromeres on chromosomal fragments (Topp et al., 2009; Fu et al., 2013; Zhang et al., 2013; Liu et al., 2015). However, one might also postulate that the fragment chromatin becomes attached to a chromosome end (see below) after the early cell divisions of development, but this would need to occur before de novo telomeres were added to broken ends in the sporophyte (see Tsujimoto, 1993). With modern technologies, it might be possible to test for the presence of de novo centromeres in the recovered sectors via ChIP-seq with antibodies against CENH3 or to test whether the released fragments attach to nonhomologous chromosomes ends with imaging techniques, assuming the ability to recognize recovered developmental sectors amenable to such studies. Indeed, as McClintock noted in her July 30, 1931 letter to Burnham, it is difficult to find a recovered sector that can be analyzed cytologically. Until such time that the phenomenon is tackled with modern technologies, it remains a mystery. McClintock’s project that was suggested by Stadler went unsolved, although she was quite productive working on other material (Kass and Birchler, 2014), which, as McClintock noted, set her on the road to the discovery of transposable elements.

THE X COMPONENT

In a lecture at the University of Missouri during the 1978 Stadler Genetics Symposium (in honor of L.J. Stadler), McClintock described results of chromosomal rearrangements produced from the BFB cycle (McClintock, 1978). It was in material undergoing the BFB cycle that she discovered activated transposable elements, to which she turned her research activities for several decades. However, in the 1978 article, she described some earlier studied BFB materials with a concentration on the types of chromosomal aberrations generated. Of particular focus was a chromosome that produced, repeatedly, specific types of aberrations. Her genetic analysis mapped the responsible region for this activity “close to or at the centromere of the initially isolated fragment chromosome 9.” McClintock dubbed this site the “X component,” which she postulated could restructure the genome in precise ways. The material carrying derivatives of this chromosome still exists but subsequently has never been studied systematically for chromosomal types produced but it does show aberrant meiotic behavior (Maguire, 1987). Below, we speculate on the basis of the specific types of aberrations generated in light of recent studies of chromosome dynamics and evolutionary genomics.

What are the types of chromosomal aberrations observed in material with the X component chromosome? The initial chromosome produced derivatives that were missing one arm resulting in a telocentric chromosome. Another type was the production of isochromosomes in which there is a mirror image duplication of one arm with loss of the other. A third major type was fusion of broken parts of this chromosome to one end or the other of the progenitor generating pseudo-isochromosomes or terminal reverse duplications. The initial chromosome was also prone to produce ring chromosomes in which the centromere was fused to one chromosome end or in which the two ends of the chromosome were fused.

In addition to rearrangement within its own structure, the fragment chromosome could become involved in rearrangements with other members of the complement. One type was the fusion of the centromere of the fragment chromosome with another centromere creating a centromere-centromere translocation. The other type of rearrangement was fusion of the fragments to the end of a chromosome arm. The site of breakage of the fragment chromosome was at the position of the X component, i.e., at or close to the centromere, and missing the fragment short arm. Of course, other chromosomal changes were found, but as McClintock noted, a pattern among them was not obvious and likely represents a background of changes. If the fragment chromosome attached to the end of another chromosome in such a manner that the fragment retained centromere activity, a dicentric would be produced that could potentially catalyze a chain reaction of rearrangements in the genome that might account for these less common aberrations.

Let us first consider the changes for the fragment chromosome itself. Because the centromere region was implicated in initiating the rearrangements, it is possible that the fragment chromosome centromere was prone to misdivision as a univalent in meiosis. Centromere misdivision results from the attachment of a centromere to opposing spindles with a resulting fission in which the two products can be capable of anaphase movement (Sears, 1952). It has been documented in several plant species but studied most prominently in maize and wheat (Triticum aestivum; Carlson, 1970; Carlson and Chou, 1981; Kaszás and Birchler, 1996, 1998; Sears, 1952; Lukaszewski, 1993).

Studies of centromere misdivision in maize (Carlson, 1970, 1973a, 1973b; Carlson and Chou, 1981; Kaszás and Birchler, 1996, 1998; Kaszás et al., 2002; Phelps-Durr and Birchler, 2004; Jin et al., 2005) reveal products such as telocentric chromosomes, isochromosomes, and rings as depicted in Figure 2. Fracture of the centromere can result in loss of one of the two chromosome arms. Alternatively, the broken centromere can fuse upon itself after replication to produce a mirror image isochromosome. Rings from misdivision result from the joining of the broken centromere to the end of the same chromosome arm and are a regular product of misdivision (Carlson, 1973b; Kaszás and Birchler, 1996, 1998). An example of a centromere-telomere fusion to form a ring chromosome is shown in Figure 3. Thus, the behavior of the X component is similar, if not identical, to that which would occur via centromere misdivision.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Diagrammatic Representation of Products of Centromere Misdivision.

Top, left: Misdivision of the centromere occurs when the kinetochore attaches to the spindle from both poles and fractures the centromere of a univalent (Sears, 1952). The chromosome depicted is a univalent in meiosis I with sister chromatids of a chromosome with arms of different lengths. The different arms are shown in orange and blue. Top, right: The attachment of the chromosome from both poles breaks the chromosome at the centromere and the replicated chromatids of the different arms progress to opposite poles. Molecular studies indicate that centromere misdivision cleaves the underlying DNA sequences (Kaszás and Birchler, 1996, 1998; Jin et al., 2005). Bottom: Products of centromere misdivision include telocentric chromosomes derived from either arm (blue or orange) of the progenitor chromosome, isochromosomes derived from either chromosome arms but that are fused at the site of centromere breakage, and ring chromosomes that join the broken centromere to the end of the same chromosome (Carlson, 1970, 1973a, 1973b; Carlson and Chou, 1981; Kaszás and Birchler, 1996, 1998; Kaszás et al., 2002; Phelps-Durr and Birchler, 2004; Jin et al., 2005).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Ring Chromosome Showing Centromere-Telomere Fusion.

Carlson (1973b) documented that centromere misdivision of the supernumerary B chromosome was correlated with its nondisjunction property. The sister kinetochores apparently attach to the spindle from both poles but the tendency to nondisjoin causes rupture of the centromere. One chromosome recovered from TB-9Sb, a translocation between the B chromosome and the short arm of chromosome 9, was a ring chromosome (Carlson, 1973b). This chromosome is depicted in the pachytene stage of meiosis. On the left is fluorescence in situ hybridization (FISH) with the B-specific repeat that is concentrated at the centromere of the B chromosome in red and knob heterochromatin in green. The short arm of chromosome 9 has a knob at the very terminus of the chromosome. The FISH image reveals that the centromere is annealed to the knob to form the ring indicating that the broken centromere fused with the terminus of the chromosome. The other red signal is the terminal site of the B specific sequence on the 9-B chromosome as the reciprocal portion of the TB-9Sb translocation. Right, gray scale of the same image. Bar = 10 μm.

What is not explained are the end to end fusions or the centromere to chromosome end fusions. The fragments that were attached to the ends of other chromosomes were typical of products of centromere misdivision with the broken centromere attaching to a chromosomal terminus. We postulate that there is a failure of telomere capping that occurs, which could join the DNA double-strand break from centromere misdivision to the exposed chromosomal end. End-to-end fusion between different chromosomes would also result from such failure. The nature of telomere fusion is unknown but some fraction of telomeres in Arabidopsis thaliana are blunt ended (Kazda et al., 2012; Valuchova et al., 2017) and could serve as a substrate for nonhomologous ending joining repair if the protective proteins were disassociated from the chromosomal terminus. Telomeres with the canonical structure involving a single-strand overhang might require the action of an exonuclease to be available for fusion.

Indeed, mutational reduction of telomerase activity, which is involved with telomere maintenance, will result in chromosomal end to end fusion (Riha et al., 2001; Riha and Shippen, 2003; Heacock et al., 2004; Kazda et al., 2012; Valuchova et al., 2017). In McClintock’s initial characterization of the BFB cycle, she noted that broken chromosomes at meiosis would continue to fuse and break during the gametophyte generation (and endosperm) but were “healed” in the sporophyte. The healing occurs when a de novo telomere is added in the sporophytic generation but this process does not operate in the gametophyte (Wang et al., 1992; Werner et al., 1992; Tsujimoto, 1993; Chao et al., 1996). If centromere misdivision occurs in meiosis and telomere capping is compromised in the subsequent gametophyte, ring chromosome products from misdivision would result from the fusion of the broken centromere and the exposed telomere. Fusions of fragments to chromosome ends could also occur in the gametophyte and be delivered to the subsequent zygote without any chromosomal breakage.

What about centromere-centromere fusions? In wheat, when genotypes are produced that foster misdivision in two chromosomes (double monosomics) simultaneously, common products from these experiments are chromosomes in which the broken centromeres from nonhomologous chromosomes join to form a translocation with a junction at the respective centromeres (Lukaszewski, 1993, 2010; Vega and Feldman, 1998; Zhang et al., 2001; Wang et al., 2017). An example of a hybrid centromere joining wheat and rye (Secale cereale) chromosome arms produced by centromere misdivision is shown in Figure 4. With the production of double-strand breaks in two different chromosomes simultaneously as a result of misdivision, the two breaks can be repaired in such a configuration in which chromosome arms from nonhomologous chromosomes switch. If in McClintock’s materials the univalent fragment chromosome was regularly undergoing misdivision, any other centromere misdivision would generate the conditions for the recovery of the described chromosomes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Centromere-Centromere Translocation between Wheat and Rye Chromosomes Recovered from Centromere Misdivision.

A homozygous translocation (arrows) resulting from centromere misdivision that joins wheat chromosome 1B with rye chromosome 1. In this FISH image, the wheat centromere repeat is depicted in red and the rye centromere repeat is in green. The juxtaposition of the two signals indicates that the breaks were within the centromeric region and were joined together. Bar = 10 μm.

Chromosome end amendment by broken fragments has also been found in maize under other circumstances and thus is not unique to the X component material. Han et al. (2006) described the recovery of an inactive centromere of the supernumerary B chromosome of maize translocated to the short arm of chromosome 9. The B chromosome is an extra nonvital chromosome found in some lines of maize. Despite its dispensability, it is maintained by an accumulation mechanism. This process involves B nondisjunction at the second pollen mitosis so that one of the two sperm has two copies and the other has none (Roman, 1947). The sperm with the B chromosomes then preferentially fertilizes the egg as opposed to the polar nuclei in the subsequent fertilization event (Roman, 1948). The centromere of the B chromosome is obviously the site of nondisjunction, but this process requires specific other regions of the chromosome to be present in the nucleus, most notably the distal tip of the long arm (Ward, 1973). The chromosome 9 with the inactive B centromere does not undergo nondisjunction because the rest of the B chromosome is missing.

However, when B chromosomes are added to the genotype and thus supply the transacting factors for nondisjunction, the inactive B centromere containing chromosome 9 is induced to attempt nondisjunction (Han et al., 2007). Some nondisjoined chromosomes 9 were in fact recovered in the next generation, but, more frequently, the attempted nondisjunction caused chromosomal breakage of the short arm of chromosome 9 (Han et al., 2007). Released chromosomal fragments were found appended to the short arms of chromosomes 2, 7, and 8. The process of how these chromosomes were produced is shown in Figure 5. The chromosome 7 translocation was recovered and is homozygous viable. Thus, it must not be missing any vital genes arguing for an attachment at the end or very close to the end of 7S (Figure 6). Whether there is any resection of the chromosome ends of the recipient chromosome in these cases is not known. Also, McClintock (1941) in initial studies of the BFB cycle noted a rare case of an otherwise intact chromosome 9 with extra chromatin attached at the terminus of the short arm beyond the terminal knob heterochromatin. These products of chromosome rupture with amendment elsewhere in the genome are similar to those from the X component fragment chromosome described by McClintock (1978) but result from an independent process.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Diagrammatic Representation of Fragment Attachment to Chromosome Ends.

The 9-Bic-1 (9-B inactive centromere-1; a translocation of an inactive B chromosome centromere on the tip of chromosome arm 9S) chromosome has an inactive B centromere present on the tip of the short arm of chromosome 9 (Han et al., 2006). Chromosome 9 is depicted in blue, B chromosomes are depicted in purple, and B centromeres are depicted in red. In the absence of a normal B chromosome, 9-Bic1 disjoins normally (not shown). However, when a normal B chromosome is added to the genotype, it supplies the transacting factors for nondisjunction to itself and to the inactive B centromere on 9S (depicted on the left). The arrows indicate nondisjunction of 9-Bic-1 and the B chromosome. Thus, 9-Bic-1 now undergoes nondisjunction or, more often, is broken in the anaphase of the second pollen mitosis (Han et al., 2007) shown in the center. The B centromere has a specific repeat that allows it to be followed by FISH (depicted in red). In the progeny of crosses of 9-Bic-1 + B chromosomes, broken fragments of 9S with the B centromere were found attached to the ends of other chromosomes (Han et al., 2007; right). Red circles indicate the inactive B centromere attached to chromosome 2, 7, or 8.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Fragment Attachment to the Terminus of the Short Arm of Chromosome 7.

FISH image of 7-Bic-1 in which the inactive B centromere (red) was broken from 9-Bic-1 (see Figure 5) and became attached to the terminus of the short arm of chromosome 7 (arrows). Knob heterochromatin is labeled in green. The chromosome is homozygous viable and thus must not be missing any vital genes. Inset shows the 7-Bic-1 chromosome with the merged image, with the B repeat indicating the inactive B centromere and the knob heterochromatin labels with the gray-scale 4′,6-diamidino-2-phenylindole (DAPI). Bar = 10 μm.

Furthermore, genome sequences of related plant species coupled with comparative evolutionary studies have revealed that some of the most common rearrangements involve chromosome end-to-end fusion and the insertion of one chromosome into another at or near the centromere (Lysak et al., 2006; Luo et al., 2009; Murat et al., 2010; Schubert and Lysak, 2011; Wang and Bennetzen, 2012; Wang et al., 2015; Hoang and Schubert, 2017; Luo et al., 2017) (Figure 7). If telomere capping is compromised in the gametophyte generations as suggested by the nature of the BFB cycle described above (McClintock, 1941), end-to-end fusions together with inactivation of one centromere could be transmitted to the subsequent generation. If a centromere fracture occurred in meiosis for one chromosome, another chromosome with compromised telomeres could insert into this single chromosomal break. If the centromere split results from misdivision, the insertion would have to occur before separation of the breakage products, but they might also occur via other circumstances of centromere fracture. Indeed, centromeres appear to be relatively susceptible to DNA damage due to spindle defects that foster centromere fission (Guerrero et al., 2010; Schneider et al., 2016). These evolutionary events would require zero double-strand breaks (end to end fusions) or only one (centromere break) and therefore have a higher probability of occurrence than other types of chromosomal aberrations that require two simultaneous double-strand breaks (deletions, duplications, inversions, and translocations). While some cases of chromosome capture appear to be near to as opposed to within centromeres, the exact position of the active centromere shifts across a narrow range in different lines due to centromere deletion/inactivation and de novo formation over relatively short evolutionary timeframes (Schneider et al., 2016; Zhao et al., 2017), potentially making such determinations only an approximation. Subsequent inversions in centromere regions (Lamb et al., 2007) also contribute to an ambiguity to the precise prediction of the site of chromosome capture within the vicinity of the centromere (Schneider et al., 2016). These chromosomal events over evolutionary time appear related to those documented experimentally by McClintock (1978) and others (Carlson, 1970, 1973a, 1973b; Kaszás and Birchler, 1996, 1998; Phelps-Durr and Birchler, 2004; Han et al., 2006, 2007; Gao et al., 2011).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Diagrammatic Representation of Nested Chromosomal Insertion and End-to-End Fusions: Common Chromosome Number Reduction Events in Evolution.

Two of the most common chromosomal rearrangements during karyotype evolution are end-to-end fusions and nested chromosomal insertions (Lysak et al., 2006; Luo et al., 2009, 2017; Murat et al., 2010; Schubert and Lysak, 2011; Wang and Bennetzen, 2012; Wang et al., 2015; Hoang and Schubert, 2017). At the left, two chromosomes are depicted in orange and blue with the centromere depicted as gray ovals. End-to-end fusions result from the joining of the termini of the two chromosomes, which is likely associated with deletion or inactivation of one of the two centromeres such that the result is not functionally dicentric, which would lead to its fracture. Experimentally, centromere inactivation can occur within the span of a cell cycle, or at least a few, and then the inactive state is perpetuated over generations (Han et al., 2009; Gao et al., 2011). Nested chromosomal insertion occurs when one chromosome inserts at or near the centromere of another chromosome. Again, only one centromere would retain function for the integrity of the new chromosome to be maintained.

One potential explanation for these chromosomal changes is that some type of recombination ties together nonhomologous chromosomes at their ends. This scenario would require identical inverted tandem repeat sequences at the ends of all chromosomes in order to align sequences in the proper register for crossing over to occur. We are not aware that such structures, which would have to permit recombination, are universally present near the ends of plant chromosomes. Moreover, the parallels to experimentally induced rearrangements involving chromosomal ends that do not involve recombination would suggest otherwise. Furthermore, while it is possible that end to end fusions and nested chromosomal insertions might be common because other types of aberrations are selected against to some degree, the prevalence of related structures in unselected materials from experimental work and the lower order kinetics for their formation argues against such a scenario. Indeed, genome sequencing of Aegilops tauschii revealed the remnants of telomere sequences at the junction of an evolutionary end to end chromosomal fusion (Luo et al., 2017).

McClintock noted that heterochromatic knobs could be preferred sites for chromosomal rearrangements in addition to centromeres and end to end fusions. In this regard, Rhoades and colleagues (Rhoades et al., 1967; Rhoades and Dempsey, 1972) described a genotype (called Hi Loss) that fostered chromosomal breaks at knob sites at the second pollen mitosis. Knobs are visible landmarks on maize chromosomes that are composed of thousands of copies of a unit repeat (Peacock et al., 1981). They are typically highly condensed chromosomal sites that are late replicating in the cell cycle (Pryor et al., 1980). In the Hi Loss background, the presence of B chromosomes would induce knobbed chromosome arms to break at the second pollen mitosis in the gametophyte generation. Although never examined experimentally to our knowledge, such knob breakage in the presence of a misdivided centromere or other broken fragments is likely to produce fusions at knob sites. The relationship of the B chromosome nondisjunction and knob breakage at the same specific mitosis was noted (Pryor et al., 1980) with the suggestion that both result from delayed replication of the respective structures beyond the timeframe of the initiation of anaphase. Such delayed replication for the B centromere would result in nondisjunction but chromosomal breakage for the interstitial knobs. Because the B chromosome can occasionally nondisjoin in the endosperm and other tissues (Carlson, 1973b), it is possible that knob breakage might also occur occasionally under other circumstances.

In conclusion, we speculate that there is a convergence of observations from experimental studies of chromosomal breakage about the preponderance of broken centromeres and fusions to the ends of chromosomes that explain several common karyotypic changes revealed by comparative genomics. Chromosomal breakage initiated in meiosis, such as centromere misdivision, would occur immediately prior to the gametophyte generation that produces the gametes. If telomere capping of the fractured chromosome fails to protect the break from DNA repair pathways, as suggested by the behavior of the BFB cycle, the most common chromosomal changes noted by McClintock would be formed and directly introduced into the next generation. As opposed to chromosomal changes in somatic cells, this progression of events will lead to heritable genomic rearrangements.

Here, we have reexamined problems studied by Barbara McClintock that have all but been forgotten in the background of her many other brilliant contributions. Many aspects of these topics deserve further experimental investigation such as determining the frequency and developmental timing of the apparent occasional failure of telomere capping. It is often a useful exercise to place recent discoveries such as centromere epigenetics and comparative genomics in the context of the historical literature, in which hidden mysteries are clarified and a greater synthesis can potentially emerge.