Chromosome Morphology and Nomenclature

With few specialized exceptions, chromosomes from somatic cells of higher eukaryotes are visualized directly only during mitosis. Each mitotic chromosome consists of two sister chromatids that are held together at a waist-like constriction called the centromere. The portions of the chromosomes that are not in the centromere itself are called chromosome “arms” (Fig. 12-2).

Figure 12-2 anatomy of mitotic chromosomes from higher eukaryotes. Left., The principal structural features of chromosomes. Center, An electron micrograph of human mitotic chromosomes. Right, A diagram of the various classes of chromosomes. At mitosis, chromosomes of higher eukaryotes consist of sister chromatids held together at the centromeric region. Chromosomes are classified on the basis of the position of the centromere relative to the arms. In metacentric chromosomes, the centromere is located midway along the chromatid. In submetacentric chromosomes, the centromere is located asymmetrically so that each chromatid can be divided into short (P) and long (Q) arms. In acrocentric chromosomes, the centromere is located near the end of the arms. In telocentric chromosomes, the centromere appears to be located very near the end of the chromatid.

(Micrograph courtesy of William C. Earnshaw.)

One DNA Molecule per Chromosome

Each eukaryotic chromosome contains one DNA molecule that stretches between the telomeres at either end. Most prokaryotic and mitochondrial chromosomes are circular DNA molecules that lack telomeres, but naturally occurring eukaryotic nuclear chromosomes are generally linear DNA molecules with two telomeres. The clearest proof that each chromosome is composed of a single DNA molecule has been obtained for budding and fission yeasts, where intact chromosomal DNA molecules may be visualized by pulsed-field gel electrophoresis as a characteristic series of bands (Fig. 12-3). This technique can display the largest chromosome of fission yeast at 5,598,923 bp, but even the smallest human chromosome, which is about 40 million bp long, is too large to resolve in this way.

Figure 12-3 pulsed-field gel electrophoresis of budding yeast chromosomes. Intact cells embedded in a block of agarose are treated under very gentle conditions with proteases and detergents to free the chromosomal DNA from other cellular constituents. The DNA is then moved under the influence of an electrical field out of the agarose block and directly into an agarose gel. The technique uses a specialized gel apparatus in which the direction and strength of the electrophoretic field is varied periodically. This technique permits the separation of very long DNA molecules (of up to several million bp).

(Courtesy of P. Hieter, University of British Columbia, Vancouver, Canada.)

The Organization of Genes on Chromosomes

The first chromosome to be completely sequenced (in 1977) was that of the bacterial virus φx174 (Table 12-1). Starting in the 1990s much effort worldwide has been devoted to determining the complete sequences of the chromosomes of a wide variety of organisms (see Fig. 2-4). Sequencing efforts that have been completed to date have generated an enormous bank of data on the genetic composition of simple and complex organisms. For example, over 100 microbial genomes have been sequenced. One major goal of this effort—the sequence of the human genome—is now essentially complete.

Table 12-1 DNA CONTENT OF VARIOUS GENOMES

Note: In most higher eukaryotes, with the exception of some plants, the huge tracts of repeated DNA sequences in and around centromeres are poor in genes and beyond the limits of present technology to sequence. Thus, when statistics are given on chromosome sizes in descriptions of genome sequencing projects, these portions are generally omitted. Where possible, the genome size figures given here reflect the entire genome (sequenced and unsequenced).

* It appears that only 265 to 350 of these genes are essential for life.

Complex genomes that have been sequenced thus far range in size from 580,000 bp for Mycoplasma genitalium, which causes urinary tract infections in humans to 2,863,476,365 bp for humans themselves. Numbers of protein-coding genes identified range from 480 in M. genitalium to 20,000 to 25,000 for humans (Table 12-1). However, because gene prediction algorithms are still being perfected, only rough estimates of gene number are available, even for completely sequenced genomes.

As a rule of thumb, the bacterial genomes tend to make very efficient use of space, about 90% of the genome being devoted to coding sequences. The remaining 10% is mostly taken up by sequences involved in gene regulation. One notable exception to this is Rickettsia prowazekii, for which only 76% of the genome is devoted to coding sequences. Because this intracellular parasite derives many of its metabolic functions from the host cell, much of its noncoding DNA may be remnants of unneeded genes undergoing various stages of gradual loss from the genome.

The first eukaryote whose genome was entirely sequenced was the budding yeast Saccharomyces cerevisiae. The 14 million bp yeast genome is subdivided into 16 chromosomes ranging in size from 230,000 bp to over 1 million bp (Fig. 12-3). This genome has a dramatic history. Ancestral budding yeast apparently had eight chromosomes but at one point underwent a duplication of the entire genome. This event was followed by numerous small deletions that resulted in the subsequent loss of most of the duplicated genes, with about 10% remaining. As a result, the modern budding yeast genome contains about 5700 predicted genes, many of which are paralogs (genes produced by duplication that have evolved to take on distinct functions; see Box 2-1). As a result, only about 1000 of these genes are indispensable for life. About 5% of yeast genes are segmented, containing regions that appear in mature RNA molecules (exons) and regions that are removed by splicing (introns) (discussed in detail in Chapter 16). Exons occupy approximately 75% of the budding yeast genome, with the remainder in regulatory regions, repeated DNAs, and introns (Fig. 12-4).

Figure 12-4 COMPARISON OF THE DISTRIBUTION OF GENES OVER 90,000 BP OF THE CHROMOSOME OF A TYPICAL BACTERIUM (B. SUBTILIS), THE BUDDING YEAST S. CEREVISIAE, THE FRUIT FLY D. MELANOGASTER, AND HUMANS. To give a more accurate representation of the distribution of human genes, we also show a stretch of chromosome 21 spanning 500,000 bp. Arrows show the direction of transcription. Regions of genes encoding a product are shown as thick orange arrows. Intervening sequences (introns) are shown as thin lines.

(Courtesy of A. Kerr, University of Edinburgh, Scotland.)

Subsequent analysis of the fission yeast genome yielded some surprises. First, many more (43%) of the genes have introns. Second, despite the fact that the genome is about 15% larger than that of budding yeast, the number of genes is substantially less. People were very surprised to learn that a free-living eukaryote could “get by” with fewer than 5000 genes. An important point here is that this genome was not duplicated and later pared down, so it does not have so many sister (paralogous) genes. Although it has fewer genes than budding yeast, the variety of genes is actually greater. The biggest difference between the fission and budding yeast chromosomes is in the structure of their centromere regions (see later).

The next genome sequences to be completed were those of two very important “model” organisms that have been widely used by cell and developmental biologists: the nematode worm Caenorhabditis ele-gans and the fruit fly Drosophila melanogaster. These sequences revealed a number of important organizational differences from budding yeast. Although its ge-nome is eight times larger than that of budding yeast (97 million bp distributed in six chromosomes), the nematode has only about three times more genes. Surprisingly, the fly, despite its even larger genome and more complex body plan and life cycle, has about one third fewer genes than the worm. In fact, only about 27% of the C. elegans genome and 13% of the Drosophila genomic DNA code for proteins. Instead, the fly has much more noncoding repetitive DNA than the worm.

The “finished” sequence of the human genome, published in 2004, revealed an even lower density of genes. Humans have far fewer genes than had been predicted: about 20,000 to 25,000, in contrast to some earlier predictions of up to 100,000 (Table 12-1). Protein-coding regions occupy only about 1.2% of the chromosomes. In contrast, various repeated-sequence elements and pseudogenes appear to occupy about 50% of the genome, as is discussed in a later section. To put this all in perspective, every million bp of DNA sequenced yielded 483 genes in S. cerevisiae, 197 genes in C. elegans, 117 genes in D. melanogaster, and only 7 to 9 genes in humans. If the Escherichia coli chromosome were the size of chromosome 21, the smallest human chromosome at ˜40 × 10⁶ bp, it would have nearly 37,000 genes—more than the entire human complement! In fact, chromosome 21 is predicted to have only 225 genes.

Human genes range in size from a few hundred bp to well over 10⁶ bp, the average being about 28,000 bp. Most human protein-coding genes have introns separating an average of 9 exons averaging only 145 bp each. The average intron is a bit over 3000 bp in length, but the variability is enormous. Genes can have over 100 exons or only 1, and introns can be over 500,000 bp long. It is therefore not surprising that the discovery of new genes using the genomic DNA sequence is a complex art that is still in its infancy.

The distribution of protein-coding genes along chromosomes is also highly variable. For example, on chromosome 9, gene density ranges from 3 to 22 genes per 10⁶ bp. On chromosome 21 one region of 7 × 10⁶ bp, encompassing nearly 20% of the whole chromosome, has no identified genes at all. This region is almost twice the size of the entire E. coli chromosome! Approximately 25% of the genome is made up of regions of greater than 5 × 10⁵ bp that are devoid of genes and are termed gene deserts.

Much of this “noncoding DNA”—up to 40% to 50% in humans—is actually transcribed into RNA. The functions of these RNAs are unknown, but they could have important roles in chromosome structure and function.

Transposons Make Up Much of the Human Genome

Eukaryotic genomes contain large amounts of repetitive DNA sequences that are present in many copies (thousands, in some cases). By contrast, coding re-gions of genes (which are typically present in a single copy per haploid genome) are referred to as unique-sequence DNA.

Repetitive DNA shows two patterns of distribution in the chromosomes. Satellite DNAs are clustered in discrete areas, such as the centromeres. They are discussed in the next section. Other types of repetitive DNA are dispersed throughout the genome. In humans, most of this dispersed repetitive DNA is composed of transposable elements—small, discrete DNA elements dispersed throughout the genome—that either are now or were formerly capable of moving from place to place within the DNA. There are many types of these elements, but for purposes of simplicity, they are divided here into two overall classes. Transposons move via DNA intermediates, and retrotransposons move via RNA intermediates. Transposons generally move by a cut-and-paste mechanism, that is, the starting element cuts itself out of its location within the genome and inserts itself somewhere else. There is currently no evidence for active transposons in humans, but in Drosophila, transposition by transposons such as the P element accounts for at least half of spontaneous mutations.

Even though humans no longer have active transposons, we still use at least two functional vestiges of these elements. It has been known for years that one of the ways in which the diversity of the immune system is generated is by cutting and pasting portions of the genes that encode the variable regions of the immunoglobulin chains (see Fig. 28-10). This process involves moving bits of DNA around, and it now appears that the enzymes that accomplish this process were originally encoded by ancient transposons. In addition, CENP-B (centromere protein B; see Fig. 13-23), an abundant protein that binds to the α-satellite DNA repeats in primate centromeres, is closely related to a transposase enzyme encoded by one family of transposons.

Retrotransposons transcribe themselves into RNA, then convert this RNA into DNA as it is being inserted at another site in the genome. Retrotransposons move (transpose) from one place in the DNA to another through production of an RNA intermediate. Therefore, on completion of a transposition event, the original retrotransposon remains in its original chromosomal location, and a newly generated element (which may be either full-length or partial) is inserted at a new site in the genome. The copying of RNA into DNA is carried out by a specialized type of DNA polymerase called a reverse transcriptase. These enzymes were discovered in tumor viruses with RNA chromosomes, but human cells also have a number of genes encoding reverse transcriptases.

The best-known retrotransposons are LINES (long interspersed nuclear elements) and SINES (short interspersed nuclear elements). Reverse transcriptases encoded by LINES are responsible for movements of both LINES and SINES. The L1 class of LINES encodes two proteins, one of which has reverse transcriptase activity (Fig. 12-5). All DNA polymerases, including reverse transcriptases, work by elongating a preexisting stretch of double-stranded nucleic acid (see Chapter 42 for a discussion of the mechanism of DNA synthesis). L1 elements insert themselves into the chromosome by first nicking the chromosomal DNA, then using the newly created end as a primer for synthesis of a new DNA strand (Fig. 12-5). The template for this DNA synthesis by the reverse transcriptase is the LINE RNA, and the newly synthesized DNA is made as a direct extension of the chromosomal DNA molecule. Most LINES are only partial copies of the full-length element. Apparently, the reverse transcriptase is not very efficient (processive): It usually falls off before it completes copying the entire element.

Figure 12-5 mechanism of transposition of an l1 element. The element is transcribed by RNA polymerase II (see Fig. 15-4). Proteins encoded by the element nick the chromosome, promote base pairing of the L1 transcript with the target site, and reverse transcribe the RNA into DNA. The L1 DNA is synthesized as an extension of the chromosome. The mechanism of final closing up of the nicks and gaps is not yet fully understood.

Interestingly, the key enzyme responsible for maintaining DNA sequences at telomeres, telomerase (see later), is a specialized form of reverse transcriptase, and its mechanism is closely related to that of the L1 reverse transcriptase.

LINES and SINES plus other remnants of transposable elements account for up to 45% of the human genome. LINES, with a consensus sequence of 6 to 8 kb, make up about 20% of the genome. (A consensus sequence is the average arrived at by comparing a number of different sequenced DNA clones.) About 79% of human genes have at least one segment of L1 sequence inserted, typically in an intron. The Alu class of SINES, with a consensus sequence of about 300 bp, constitutes about 13% of the total DNA—almost a million copies scattered throughout the genome. Alu elements are derived from the 7SL RNA gene, which encodes the RNA component of signal recognition particle (see Fig. 20-5). They are actively transcribed by RNA polymerase III (see Fig. 15-10) but are short and do not have enough coding capacity to encode for complex proteins. They therefore rely on the L1 machinery to move around. It is therefore somewhat paradoxical that SINES and LINES have quite different distributions along the chromosomes. LINES are concentrated in gene-poor regions of the chromosomes with a relatively higher content of A + T base pairs. In contrast, the Alu SINES are concentrated in gene-rich regions with a relatively higher content of G + C base pairs.

Transposition can be harmful, as along the way, genes can be disrupted, deleted, or rearranged. Because of their tendency to insert into gene-rich regions of chromosomes, Alu elements are one of the most potent endogenous human mutagens, with a new Alu insertion occurring once in every 100 births. In contrast, although LINES can cause genome instability when they move, several factors appear to minimize the damage that they cause. Despite the large fraction of the human genome that is derived from LINES, they cause only 0.07% of spontaneous mutations seen in humans, owing to several mitigating features: (1) Only about 60 to 100 L1 elements are active, and these appear to be active only in the germ line (i.e., during production of gametes). (2) LINE elements prefer to move into gene-poor areas of chromosomes. (3) Most LINE sequences are only fragments of the complete element. In contrast, mice apparently have many more active L1 elements (˜3000), and L1 transposition causes about 2.5% of spontaneous mutations in mice. One of the ancestral roles of the RNAi machinery (see Fig. 16-12) might have been to suppress the deleterious activity of transposable elements.

The physiological role, if any, of these elements is much debated. One long-favored possibility is that they do nothing advantageous and are analogous to an infection of the DNA that is tolerated as long as it does not move into genes that are essential for life. This is called the “selfish DNA” hypothesis. This notion has been challenged for Alu sequences, which are efficiently transcribed into RNA. Alu transcripts accumulate under conditions of cellular stress such as viral infection. This is interesting, because Alu transcripts can bind very efficiently to a protein kinase called PKR, which is induced by interferon as part of the cell’s antiviral protection pathways. The best-known function of PKR is phosphorylation of eukaryotic initiation factor 2a-subunit (eIF-2a; see Fig. 17-9). This profoundly inhibits protein synthesis. PKR is generally activated by dsRNAs, and this is presumably important for its antiviral role, as many viruses have RNA chromosomes. Alu transcripts at low levels activate PKR (i.e., suppress protein synthesis), but at higher levels, they inactivate the enzyme (i.e., promote protein synthesis). Thus, it has been suggested that Alu transcripts might be natural regulators of protein synthesis under conditions of cellular stress.

LINES can also modulate transcription of genes by influencing the behavior of RNA polymerase as it passes through them. Thus, they might have a role in the control of gene expression. As discussed at the end of this chapter, the structure of telomeres (the ends of chromosomes) is in part maintained by telomerase, a reverse transcriptase that is likely have originated as part of a retrotransposon.

Pseudogenes

One surprise that emerged from analysis of the human genome sequence was the number of pseudogenes, which may exceed the number of genes. Pseudogenes are derived from genes but are no longer functional. They arise in two ways, both involving transposable elements. Processed pseudogenes, the more common variety, are created by reverse transcription of mature mRNA sequences into DNA apparently by a LINE re-verse transcriptase that then inserts the copy back into the genome. Because these sequences come from ma-ture mRNA, they lack introns. They also lack sequences that regulate transcription initiation and termination (Chapter 15), so they are not expressed. Unprocessed pseudogenes are created either by reverse transcription of unspliced precursor mRNAs or by local duplications of the chromosome that generally occur as a result of recombination between transposable elements. Such duplications initially create bona fide functional gene copies that become pseudogenes as they accumulate mutations that render their transcripts nonfunctional. Because pseudogenes are not functional, mutation of their DNA is not selected against during evolution, as are harmful mutations in the coding sequences of genes. Thus, over time, pseudogenes become less and less recognizable, and they eventually are lost from recognition in the sea of noncoding DNA.

Segmental Duplications in the Human Genome

About 5% of the human genome is composed of regions of segmental duplication that have formed relatively recently in evolutionary time. Segmental duplications are regions of 1000 or more bp with a DNA sequence identity of 90% or more that are present in more than one copy but are not transposons. They are of considerable interest because they can have a significant impact on human health. Regions of highly related DNA sequence are able to base-pair with one another and can consequently undergo recombination with one another. Depending on how these regions are distributed on the chromosomes, this recombination can eliminate intervening regions of nonduplicated DNA. If the deleted region contains genes important for human health, then the end result can be human disease.

One example of this is found on chromosome 7, where deletion of a portion of the long arm is associated with Williams-Beuren syndrome, a complex developmental disorder associated with a highly variable range of symptoms that can include elfin-like facial features, defects in certain mental skills, and a wide range of physical problems (Fig. 12-6). These deletions, which typically remove about 1.6 × 10⁶ bp, occur because of large segmental duplications of blocks of >140,000 bp distributed across a region of 2 × 10⁶ bp. These duplications flank a unique sequence region of 1.15 × 10⁶ bp that is lost when recombination occurs between the regions of segmental duplication. Because of the highly complex organization of this region and the large size of the duplications, this turned out to be the most difficult region of chromosome 7 to sequence, and in fact, some ambiguities still remain.

Figure 12-6 a, Segmental duplications within a region of human chromosome 7 give rise to Williams-Beuren syndrome. Inappropriate recombination between duplicated sequences causes the deletion of a region of the chromosome. B, Williams-Beuren syndrome is a rare congenital disorder that is characterized by an outgoing personality, a characteristic elfin-like facial appearance, moderate to mild mental retardation, and a range of physical problems.

(Courtesy of the Williams Syndrome Association. To learn more, visit the association’s website at www.williams-syndrome.org/).

The Centromere: Overview

The centromere is at the heart of all chromosomal movements in mitosis and meiosis, as it is the region where the chromosome becomes attached to the mitotic spindle (the microtubule-based apparatus upon which chromosomes move; see Chapter 44). The centromere also has an important role in monitoring the attachment of the chromosomes to the spindle and controlling the progress of cells through mitosis. The centromere is a nucleoprotein structure, and both DNA and proteins are essential to its function.

In chromosomes of most higher eukaryotes, the centromere may be visualized directly as a waist-like stricture or primary constriction where the two sister chromatids are most intimately paired. Several abundant families of repetitive DNAs are concentrated in the centromeres of human chromosomes. The chromatin of human centromeres is entirely embedded within constitutive heterochromatin, a form of chromatin characterized by the presence of special proteins and special modifications of the histone proteins, that is generally nonpermissive for gene transcription and that remains condensed throughout the cell cycle (see Fig. 13-9). At least some of the repeated DNA elements are transcribed into dsRNA, and the cellular RNAi machinery is required for the assembly of heterochromatin. (For a discussion of heterochromatin, see Chapter 13.) At the surface of the centromeric heterochromatin is a button-like structure called the kinetochore, which directs chromosomal movements in mitosis (see Fig. 13-20). Kinetochores are packaged into a specialized form of chromatin called centrochromatin.

Variations in Centromere Organization among Species

Each chromosome has particular DNA sequences, called CEN sequences, that specify protein-binding sites required for assembly of the kinetochore. In budding yeast, CEN sequences are autonomous; if inserted into circular DNA molecules (plasmids), they can render them capable of interacting with the mitotic spindle and segregating during mitosis (Fig. 12-7). In other organisms, including the fission yeast Schizosaccharomyces pombe, CEN sequences appear to require an activation event in order to nucleate kinetochore formation. This event appears to involve some sort of modification of the DNA and/or chromatin (discussed later).

Figure 12-7 a, The budding yeast centromere is specified by a 125-bp sequence, with three conserved DNA elements (CDE I to CDE III). CDE I and CDE III bind proteins in a sequence-specific manner. CDE III has mirror symmetry: a central C (dot) is flanked by two regions of complementary DNA sequence (arrows). All that seems to be important about CDE II is its abundance of A and T nucleotides and its overall length. B–C, The assay for mitotic stability of a plasmid used to clone CEN DNA from most budding yeast chromosomes. The plasmid carries a gene encoding an enzyme involved in adenine metabolism. When the plasmid is present, colonies are white. If the plasmid is lost, the colonies become red as a result of the accumulation of a metabolic by-product. If the plasmid is capable of replication but lacks a centromere, the colonies will be mostly red, reflecting the inefficient segregation of the plasmid at mitosis (B). If the plasmid carries a functional centromere, the colonies will be white, as the plasmid will be successfully transmitted at nearly every division (C).

CEN sequences from all 16 chromosomes of budding yeast have a common organization based around three conserved sequence elements (Fig. 12-7). These are designated (in the 5′ to 3′ direction) CDE I (centromere DNA element I, 8 bp), CDE II (78 to 86 bp), and CDE III (25 bp). A 125-bp region spanning CDE I to CDE III is sufficient to direct the efficient segregation of a yeast chromosome, which can reach a size of more than 1 million bp. This type of centromere, in which the kinetochore is assembled as a result of protein recognition of specific DNA sequences, is known as a point centromere. Kinetochores assembled on point centromeres bind a single microtubule.

Even though the average size of S. pombe chromosomes is only fivefold larger than their counterparts in S. cerevisiae (4.6 mb versus 0.87 mb), fission yeast centromeres are 300- to 600-fold larger (Fig. 12-8). The smallest S. pombe centromere consists of 35,000 bp, whereas the largest spans 110,000 bp. Fission yeast centromeres are also much more complex than are their counterparts from budding yeast, containing a central core of unique-sequence DNA that is 4 to 7 kb long, flanked by complex arrays of repeated sequences. This type of centromere, where the kinetochore is assembled on a variable array of repeated DNA sequences, is known as a regional centromere. Kinetochores assembled on regional centromeres bind multiple microtubules (2 to 4 in the case of S. pombe).

Figure 12-8 Organization of the centromeric dnas of budding yeast, fission yeast, and fruit fly. A, The budding yeast point centromere is specified by a 125-bp sequence. B, The fission yeast regional centromeres all contain central core DNA flanked by complex arrays of repeated sequences. Embedded within these repeated sequences are a number of genes encoding transfer RNAs, not shown here. The minimum region required to construct a functional centromere in fission yeast artificial chromosomes is about 10 kb in length and includes the central core DNA plus a portion of the flanking repeated DNA. C, The fruit fly also has a regional centromere encompassing 420 kb. This is rich in satellite DNA and contains a number of transposable elements. Interestingly, the same satellite DNAs and transposons are also found at other, noncentromeric, regions of the chromosomes.

Studies of S. pombe centromeres revealed an additional level in the regulation of centromere function. S. pombe CEN DNA apparently must undergo an epigenetic activation event to function as a centromere. Epigenetic events are inheritable properties of chromosomes that are not directly encoded in the nucleotide sequence. They are thought to be explained either by enzymatic modification of the DNA (e.g., methylation of cytosine) or by modification of proteins that are stably associated with the DNA. There is increasing evidence that epigenetic mechanisms play an essential role in the assembly of centromeres in higher eukaryotes, including humans. What this means in practice is that (with the exception of budding yeast), no single DNA sequence can be put into cells and function directly as a centromere. If a piece of S. pombe centromeric DNA is introduced into cells, it must undergo a series of packaging events and modifications that turn it into a functional centromere. These events are so rare that when candidate DNA molecules with CEN sequences are introduced into S. pombe cells, only about one in 10⁵ assembles into a functional centromere.

In both S. pombe and metazoans, these epigenetic changes involve the construction of a special chromatin environment. S. pombe kinetochores assemble into a spe-cialized form of chromatin containing the kinetochore-specific histone variant CENP-A (see Fig. 13-23). This flanking heterochromatin is important because it is required for binding a protein complex that regulates the pairing of replicated chromosomal DNA molecules during mitosis. Mutations in S. pombe that impair formation of heterochromatin also impair centromere function.

The best-characterized centromere of a metazoan comes from rice, in which the centromere of chromosome 8 has been completely sequenced. This was possible because this particular centromere contains relatively limited amounts of the rice centromeric satellite DNA (CentO), and this is dispersed in a number of blocks separated by transposons, retrotransposons, and fragments. All in all, 72% of this centromere is composed of repetitive sequences. The kinetochore, as defined by sequences associated with the kinetochore-specific histone H3 variant CENP-A (see Fig. 13-23), spans 750 kb and appears to be interspersed with regions of chromatin containing normal histone H3. The state of posttranslational modifications of that H3 indicates that it is packaged heterochromatin (a “closed” state of the chromatin usually associated with the absence of active gene transcription; see Fig. 13-9). It was therefore surprising that this centromere region contains at least four genes that are actively transcribed. The structure of this centromere appears to be intermediate between that of a canonical vertebrate centromere and a neocentromere (see later); however, rice centromere 8 is not a new variant but has had this organization for at least the last 10,000 years, since the indica and japonica cultivars of rice were separated.

The centromere organization of the fruit fly D. melanogaster shows important similarities and differences to the plant centromeres. The centromere of the fly’s X chromosome is contained within a stretch of roughly 420,000 bp (Fig. 12-8) that is composed mostly of simple-sequence satellite DNAs interspersed with transposable DNA elements. This resembles the situation in plants; however, in Drosophila, no sequences were found in this region that are unique to the fly centromeres; all sequences that are found at centromeres can also be found on the chromosome arms. Thus, it appears that something other than the DNA sequence alone must be responsible for conferring centromere activity on this region of the chromosome.

Most higher eukaryotes have regional centromeres, but a third variant is also observed in many insects as well as in the nematode C. elegans. Holocentric chromosomes have centromere activity distributed along the whole surface of the chromosome during mitosis. Thus, instead of having a tight bundle of 20 microtubules binding to a disk-like kinetochore at a centromeric constriction, as in humans, in C. elegans about 20 microtubules bind at scattered sites along the whole poleward-facing surface of the chromosome during mitosis. If a holocentric chromosome is fragmented, all the pieces have the ability to bind microtubules and segregate in mitosis. Perhaps surprisingly, the proteins of the holocentric kinetochore are the same as those found at disk-like regional kinetochores (see Chapter 13). At the moment, nothing is known about the DNA sequences, presumably interspersed throughout the genome, that direct the assembly of holocentric kinetochores. One possibility is that in these chromosomes, any chromatin can serve to nucleate kinetochore assembly—perhaps the requirement for special epigenetic marks has been relaxed.

Mammalian Centromere DNA

Vertebrate centromeres have proven extremely difficult to characterize in molecular detail, largely due to their large size and complex, highly repeated, organization. For example, the centromere of chromosome 21 (the smallest human chromosome at ˜40 million bp) has been estimated to encompass more than 5 million bp. This entire region appears to be composed of many thousands of copies of short DNA sequences that are clustered together in head-to-tail arrays. Such clustered DNA repeats are known as satellite DNA.

The major human centromeric satellite DNA, α-satellite, is a complex family of repeated sequences that constitutes approximately 5% of the genome. Monomers averaging around 171 bp long are organized into higher-order repeats (Fig. 12-9). Some of the monomers have a conserved 17-bp sequence (the CENP-B box), which forms the binding site for the centromeric protein CENP-B (mentioned earlier as having its origin in an ancient DNA transposon). The organization of higher-order repeats varies greatly from chromosome to chromosome, and numerous repeat patterns, comprising 2 to 32 monomers, have been described. Each chromosome has one or a few types of higher-order repeats of α-satellite DNA.

Figure 12-9 hierarchical organization of α-satellite dna at human centromeres. The numbers (1 to 4) indicate higher-order repeats of α-satellite DNA. These may contain from 2 to 32 monomers (indicated by A¹, B¹, and so on). DNA sequences of adjacent monomers within a repeat (e.g., A¹, B¹, C¹) may differ by as much as 40% from one another. DNA sequences of monomers occupying identical positions within the higher-order repeats (A¹, A⁴, an so on) are nearly 99% identical to one another. The red sequence shown at the bottom represents the binding site for centromeric protein CENP-B.

Human chromosomes also contain several other families of satellite DNA. Classical satellites I to IV, which together constitute 2% to 5% of the genome, are composed of divergent repeats of the sequence GGAAT. These satellites occur in blocks more than 20,000 bp long that are immediately adjacent to the centromeres of chromosomes 1, 9, and 16 and that may be found at much lower levels near most centromeres. The so-called pericentromeric region adjacent to the centromere of chromosome 9 apparently contains 7 to 10 million bp of satellite III sequence. The long arm of the Y chromosome also contains huge amounts of satellite III DNA (up to 40% of its total DNA).

What is human CEN DNA, that is, the DNA that nucleates kinetochore formation? The best candidate thus far is α-satellite, which occurs at all natural centromeres. The entire centromeric region of certain chromosomes may be composed of α-satellite monomers, apparently with little or no interspersed DNA of other types. The amount of α-satellite DNA at different centromeres varies widely: from as little as 300,000 bp on the Y chromosome to up to 5 million bp on chromosome 7. In addition, the α-satellite DNA content of a given chromosome can vary by more than a million bp between different individuals. Thus, whatever the function of α-satellite DNA may be, clearly, a wide variation in local organization is tolerated.

If α-satellite DNA arrays longer than about 50,000 bp are introduced into cultured human cells, they occasionally form tiny minichromosomes with functional centromeres. For this to work, the α-satellite DNA arrays must have a highly regular organization, and some of the monomers must contain binding sites for CENP-B. Formation of these mammalian artificial chromosomes is very inefficient, so it is clear that α-satellite DNA arrays cannot automatically function as CEN DNA—some type of epigenetic activation is required.

There is an interesting corollary of this role of epigenetic modifications in assembly of a functional centromere. Suppose a bit of noncentromeric DNA somehow acquired the right set of modifications. Could that now function as a centromere? The answer is yes. The formation of neocentromeres on noncentromeric DNA has been seen in fruit flies and humans and was first described in plants.

Rare individuals have a chromosome fragment that segregates in mitosis, despite the fact that the normal centromere has been lost. Such chromosomes have acquired a new centromere or neocentromere in a new location on one of the chromosome arms. Remarkably, neocentromeres are composed of the normal DNA that exists at that location on the chromosome arm and yet somehow has acquired centromere function. Neocentromeres are bona fide centromeres; for example, they bind all known centromeric proteins except for CENP-B, which requires specific sequences on α-satellite DNA for binding. Different neocentromeres need have no sequences in common. These observations strongly support models in which the centromere is specified by epigenetic markers rather than the exact DNA sequence per se. It could be that the linkage that is observed between α-satellite DNA and centromeres actually reflects a propensity of α-satellite DNA to acquire the epigenetic mark, rather than a sequence-specific mechanism as occurs in S. cerevisiae.

The epigenetic mark that defines an active centromere can be lost as well as gained. Thus, it is possible for a centromere to retain its normal DNA composition and yet lose the ability to assemble a kinetochore. This has been seen most clearly in naturally occurring human dicentric chromosomes. The example shown in Figure 12-10 arose through a breakage and fusion near the long arm of chromosome 13. Thus, it has two centromeres. As shown in the figure, one of these, even though it retains its α-satellite, has lost the ability to assemble a kinetochore.

Figure 12-10 epigenetic regulation of human centromere function. An unusual chromosome was discovered during prenatal screening of a fetus that sonography had indicated to be abnormal. This chromosome consisted of two copies of the maternal chromosome 13 linked end to end. It thus contained two centromeres and so was termed dicentric. Such dicentric chromosomes are normally very unstable during mitosis, as the two centromeres on one chromatid often become attached to opposite spindle poles. This causes the chromosome to be stretched between opposite spindle poles and ultimately break. In the case of this particular dicentric chromosome, one of the centromeres has been inactivated (presumably, it lost its epigenetic mark). This chromosome thus behaves perfectly normally in mitosis. When the distribution of centromere proteins at the active and inactive centromeres was compared, it was found that CENP-B was present at both but that CENP-C, a marker for kinetochores that can bind microtubules, was present only at the active centromere. A, Organization of the dicentric chromosome. B, Phase-contrast view of chromosomes from the amniocytes (left). Phase-contrast view taken with superimposed antibody staining for CENP-B (right). C, DNA stain of a different chromosome spread (left). Staining with antibody specific for CENP-C (right).

(Micrographs courtesy of William C. Earnshaw.)

Once a DNA sequence has acquired the proper epigenetic mark, it can assemble a functional kinetochore that can regulate chromosome behavior in mitosis. This involves the binding and function of a great many proteins, to be discussed in Chapter 13 (see Fig. 13-23).

The Ends of the Chromosomes: Why Specialized Telomeres Are Needed

The ends of chromosomal DNA molecules pose at least two problems that cells solve by packaging the chromosome ends into specialized structures called telomeres. First, it is essential that cells distinguish the ends of a chromosome from breaks in DNA. When cells detect DNA breaks, they stop their progression through the cell cycle and repair the breaks by joining the ends together (see Box 43-1). Telomeres keep normal chromosome ends from inducing cell cycle arrest and from being joined to other DNA ends by the repair machinery. Second, telomeres permit the chromosomal DNA to be replicated out to the very end (see later dis-cussion).

The Structure of Telomeric DNA

Telomeres in all eukaryotes tested to date (with the curious exception of flies such as Drosophila) are composed of many repeats of short DNA sequences. The sequence 5′ TTAGGG 3′ is found at the ends of chromosomes in organisms ranging from human to rattlesnake to the fungus Neurospora crassa. In the human, roughly 650 to 2500 copies of this sequence are found at the ends of each chromosome, yielding a total length of about 4000 to 15,000 bp (this varies in different tissues). Higher plant telomeres have the sequence TTTAGGG, and other variations of this repeat sequence have been noted in protozoans and yeasts.

The telomeric repeat is organized in a unique orientation with respect to the chromosome end. Thus, the end of every chromosome has one G-rich strand and one complementary C-rich strand. The G-rich strand always makes up the 3′ end of the chromosomal DNA molecule. Thus, the very 3′ end of the chromosome always has the following structure: ……… (TTAGGG)-OH. Furthermore, the end of the chromosome is not a blunt structure; the G-rich strand ends in a single-stranded overhang that is approximately 200 bp long. Recent evidence suggests that this single-stranded DNA might “invade” the double helix of telomeric repeats, causing the ends of chromosomes to form large loops, called T loops (see later discussion).

How Telomeres Replicate the Ends of the Chromosomal DNA

One role of telomeres is to prevent the erosion of the end of the chromosomal DNA molecule during each round of replication (for a more extensive discussion of DNA replication, see Chapter 42). All DNA replication proceeds with a polarity of 3′ to 5′ on the template DNA (5′ to 3′ in the newly synthesized DNA). Furthermore, all DNA polymerases (but not RNA polymerases) work by elongating a preexisting stretch of double-stranded nucleic acid. During cellular DNA replication, this is achieved by making a short RNA primer and then elongating the RNA: DNA duplex with DNA polymerase. The primer is subsequently removed, and the newly opened gap is filled by a DNA polymerase elongating from the next upstream DNA end (Fig. 12-11).

Figure 12-11 the dna replication problem at chromosome ends. DNA polymerases cannot initiate the formation of DNA on a template de novo; they can only extend preexisting nucleotide strands (see Chapter 42). In contrast, RNA polymerases can initiate synthesis without a primer. All replicating DNA chains start from a short region of RNA, which is used to “prime” DNA polymerase. A, DNA strand separation. B, RNA primer synthesis. Replication of DNA starts with the synthesis of an RNA primer complementary to a short sequence of DNA, which is extended by DNA polymerase. C, The RNA primer is degraded and the gap is filled in by DNA polymerase. This being true, how is the DNA underneath the very last RNA primer replicated?

If the terminus of the chromosomal DNA is replicated from an RNA primer that sits on the very end of the DNA molecule, it follows that when this primer is removed, there is no upstream DNA on which to put a primer. How, then, is the DNA underneath the last RNA primer replicated? Years of searching for a DNA polymerase that could operate in the opposite direction proved fruitless. The answer that ultimately emerged turned out to be both elegant and unexpected.

One solution to this problem was taken by dipterans such as Drosophila melanogaster, in which the ends of the chromosomes are composed of transposable elements. In the fly, a few bp are lost from the end of the chromosome at every round of replication. This erosion of the chromosome ends is remedied by the occasional transposition of a transposable element to the chromosome end. Thus, this appears to be an example of an originally “selfish DNA” that has become recruited for an essential cellular function.

Most other organisms have an enzymatic activity whose specific function is to lengthen the 3′ end of the chromosomal DNA. This activity is referred to as telomerase. Telomerases are enzymes that contain both protein and RNA subunits. The sequences of these components provide an essential clue to how this enzyme works.

The sequence of the human telomere is TTAGGG. When the RNA component of human telomerase was characterized, it was found to contain the sequence AUCCCAAUC, which could base-pair with the TTAGGG repeat at the end of the chromosome. This observation led to a proposal for the mechanism of telomerase action (Fig. 12-12). In brief, the enzyme uses its own RNA as a template for the synthesis of DNA, which it attaches to the end of the chromosome. This hypothesis has been confirmed by studies that show that alterations in the sequence of the telomerase RNA lead to alterations in the telomere sequence at the end of the chromosome.

Figure 12-12 telomerase provides a special mechanism for lengthening chromosomal ends. a–b, Normal mechanisms of DNA replication are unable to replicate the very 3′ end of the chromosomal DNA. C, Telomerase solves this problem by providing its own template in the form of an intrinsic RNA subunit. This RNA subunit contains a sequence complementary to that found at the chromosome terminus on the 3′ strand. This sequence is able to base-pair with the DNA at the chromosome terminus and act as a template for DNA synthesis. In this case, the primer is the 3′ end of the chromosomal DNA, and the template is the RNA of the telomerase enzyme. Thus, the process of telomere elongation is a specialized form of reverse transcription (copying RNA into DNA), a process similar to that occurring during transposition of LINE elements (see Fig. 12-5), and most well known during the life cycle of certain RNA-containing tumor viruses. The telomerase enzyme releases and rebinds its template after each 6 to 7 bp of new DNA has been synthesized. Up to several hundred bp may be added to the telomere in this way. D, In most cells, the cycle of telomerase activity and subsequent DNA replication leaves a single-stranded G-rich strand about 200 nucleotides long.

According to this model, the telomerase actually synthesizes DNA from an RNA template. Thus, telomerase is a reverse transcriptase similar to that involved in the movement of the LINE retrotransposons (see Fig. 12-5). Interestingly, the L1 family of LINE retrotransposons appear to insert themselves into the chromosome by a similar mechanism, in which a DNA end created at a nick in the chromosome is used to prime synthesis of a DNA strand using the LINE RNA as template, the newly synthesized DNA being a direct extension of the chromosomal DNA molecule.

Telomerases have been characterized from budding yeast, mice, humans, and two ciliated protozoans. The enzymes consist of at least two protein subunits complexed to the RNA. The telomerase RNA varies in size and sequence between species, the human RNA, hTR, being 450 nucleotides long. hTR is complexed with two polypeptides: TP1 (a 240-kD protein that specifically binds to the RNA) and hTERT (a 127-kD protein that is the telomerase reverse transcriptase). Active human telomerase can be reconstituted in vitro from purified hTR and hTERT in the presence of a cell-free lysate from reticulocytes (which appears to provide essential protein-folding factors).

Telomerase is subject to tight biological regulation, active enzyme being detected in only a few normal tissues of adult humans. These include the intestine and testis. In addition, about 85% of cancer cells express active telomerase. In cells that lack telomerase, an alternative pathway, thought to involve DNA recombination, can help to maintain the telomeric repeats at chromosome ends. Paradoxically, hTR and TP1 are not tightly regulated. They are detected in many tissues, most of which lack telomerase activity. By contrast, the expression of hTERT correlates tightly with telomerase activity. Furthermore, introduction of a DNA-encoding hTERT into telomerase-negative cells is sufficient to convert those cells to telomerase-positive. This can have extremely important consequences for the proliferation of the cells (Fig. 12-15).

Figure 12-15 Introduction of hTERT, the human reverse transcriptase subunit of telomerase, into normal cells is sufficient to overcome the senescence limit and immortalize the cells. These cells are not transformed into invasive cancer cells; instead, they act like normal cells that can now grow indefinitely.

Structural Proteins of the Telomere

Telomeres provide special protected ends for the chromosomal DNA molecule, in part by coating the end of the DNA molecules with protective proteins and by adopting a specialized DNA loop structure. In organisms in which telomeric DNA sequences are relatively short, these sequences are packaged into a specialized chromatin structure. In mammals, in which the telomeric sequences are much longer, the bulk of the telomeric DNA is packaged into conventional chromatin (see Chapter 13).

Three types of proteins associate with telomeres (Fig. 12-13). The first binds in a sequence-specific manner to the double-stranded telomeric repeats. In budding yeast, the best known such protein is Rap1p. Mammals have two essential proteins of this type: TRF1 and TRF2 (telomere repeat factor). TRF1 and its associated factors regulate telomerase activity, thus helping to maintain the proper length of telomeres. A similar function has been proposed for Rap1p in budding yeast, where the telomeres appear to elongate only until they create a threshold number of binding sites for this protein. TRF2 and its associated factors protect the chromosome ends; interference with the binding of this protein to telomeres results in a loss of the G-strand overhangs and a dramatic increase in the tendency of chromosomes to fuse end to end. This may be because TRF2 can promote the formation of a special looped configuration of DNA in which the single-stranded G-strand overhang is base-paired with “upstream” DNA (Fig. 12-13B). Figure 12-14 shows the phenotype of a Drosophila mutant that lacks a protein essential for the assembly of the proper protective structure at telomeres and therefore undergoes extensive chromosome fusion.

Figure 12-13 models for telomere structure. A., The proposed structure of the end of a human chromosome. TRF1 and TRF2 bind to the double-stranded (TTAGGG)_n repeats at telomeres. TRF1 somehow regulates the action of telomerase. TRF2 is responsible for protecting the integrity of chromosomal ends. If it is lost, chromosomes fuse with one another, and many abnormalities are seen. In yeast, a protein called Ku binds to the ends of the DNA. (Pot1: 1QZG.pdb. Ku 70/80: 1JEY.pdb. Rad 50: 1L8D.pdb.) B, Alternative loop model for vertebrate telomeres. Chromosomal ends may form a T-loop structure when a single-stranded G-rich 3′ end of the chromosome “invades” a double-stranded portion of the telomere, base-pairing with one strand and displacing the other strand (D loop). TRF2 can promote formation of T loops in vitro. Inset, A T loop excised together with its chromatin proteins from a chicken erythrocyte chromosome.

(Inset, From Nikitina T, Woodcock CL: Closed chromatin loops at the ends of chromosomes. J Cell Biol 166:161–165, 2004.)

Figure 12-14 disruption of the protective complex at telomeres results in chromosome fusion. A, The chromosomes of a wild-type female Drosophila melanogaster seen at mitotic metaphase (see Chapter 44). B, The caravaggio mutant is characterized by a “train” of chromosomes generated by telomere-telomere fusions. (Caravaggio is the name of an Italian train.) C, The cav gene encodes HP1/Orc2 Associated Protein (HOAP), which specifically localizes at all Drosophila telomeres.

(Images courtesy of Gianni Cenci and Maurizio Gatti, University of Rome, Italy. B, From Cenci G, Siriaco G, Raffa GD, et al: Drosophila HOAP protein is required for telomere capping. Nat Cell Biol 5:82–84, 2003. C, Part of montage on Nature Cell Biology cover.)

Telomeric proteins of the second class bind to the single-stranded DNA of the G-strand overhang. One such protein, called Pot1, is conserved from humans to yeasts. Pot1 binds to single-stranded telomeric DNA and controls telomere length. The budding yeast homolog of Pot1 binds to the G-strand overhang and protects the end of the recessed C-rich strand at telomeres. In mutants lacking this protein, the C-rich strand is rapidly degraded, with lethal consequences for the cell.

Proteins of the third class pose a conundrum. At telomeres, they are required to protect the chromosome ends and prevent them from fusing with the ends of other chromosomes. Elsewhere on chromosomes, these same proteins function in the repair of DNA breaks by joining bits of broken DNA together, a pathway known as nonhomologous end joining (NHEJ). This appears to be exactly the opposite of their role at telomeres. The proteins involved are the Ku70/80 complex, which can recognize DNA ends, and a complex called MRN (MRE11, RAD50, NBS1). These proteins are highly conserved components of telomeres—from yeast to human—and if they are inactivated by mutations, telomeres frequently fuse together. It thus appears that chromosome ends are recognized by the breakage repair machinery but that some aspect of the telomeric structure changes the function of this machinery so that it assumes an end-blocking protection role rather than an end-joining role.

Telomeres may also direct chromosome ends to their proper location within the cell. In budding yeast (and many other species), telomeres prefer to cluster together at the nuclear periphery. Mutants in the telomere-binding SIR proteins, or in regions of the histones with which they intersect, disrupt this clustering in yeast. This results in activation of genes that are normally silenced when located in close proximity to telomeres. Thus, positioning of the telomere within the nucleus may be used to sequester genes into compartments where their transcriptional activity is repressed.

Telomeres, Aging, and Cancer

Although the average length of telomeric repeats in humans is about 4000 bp, this length varies. Chromosomes of older individuals have shorter telomeres, and gametes have longer telomeres. This suggested the interesting possibility that chromosomes might lose telomeric sequences during the life of an individual.

The relationship between telomere length and aging can be studied in cultured cells. Normal cells in culture grow for only a limited number of generations (often called the Hayflick limit) before undergoing senescence (this involves cessation of growth, enlargement in size, and expression of marker enzymes, such as β-galactosidase). Because normal somatic cells lack telomerase activity, their telomeres shorten by about 50% before the cells senesce. Senescent cells stop dividing before their telomeres become critically short. In some cases, it is possible to force the senescent cells to resume proliferation (e.g., by expressing certain viral oncogenes). These “driven” cells continue to divide and their telomeres continue to shorten until a crisis point is reached. In crisis, cells suffer chromosomal instability (chromosomal fusions and breaks can occur) and cell death. In populations of human cells in crisis, very rarely (in about 1 in a million cases), cells appear that once again grow normally. These cells now express telomerase. These observations with cultured cells led to the suggestion that senescence might occur in cells when the telomeric repeats of one or more chromosomes are reduced to some critical level.

If this model is correct, it suggests very interesting (and controversial) implications for the regulation of cell life. Suppose that telomerase is active only in the germ line, so that all gametes have long telomeres. Now, if the enzyme were inactivated in somatic cells, this would effectively provide every cell lineage with a limitation on how many times it could divide before loss of telomeric sequences caused it to become senescent. Provided that the starting telomeres were sufficiently long and that telomerase was expressed in unusual tissues, like testis and intestine, in which rapid division occurs throughout the life of the individual, this would have no deleterious effect on the life span of the organism. In fact, such a mechanism might provide an important advantage by minimizing the chances that a clone of cells would escape from the normal regulation of growth control and become cancerous.

This model has been tested in two ways. First, mice were prepared in which the gene coding for the RNA component of telomerase was disrupted. These mice were healthy and fertile for six generations in the complete absence of telomerase but then abruptly became sterile as a result of cell death in the male germ line. The abrupt cell death presumably occurred when the telomeres shortened below a critical threshold. Having telomeres about three times longer than humans might have contributed to their initial survival through several generations. This experiment thus showed that telomerase is not essential for the day-to-day life of a mammal, but clearly, it is needed for the survival of the species.

In a second experiment, the hTERT reverse transcriptase subunit of telomerase was introduced into normal cells growing in culture. This caused an increase in the level of active telomerase with dramatic results. Instead of undergoing senescence, these cells kept dividing in culture, apparently indefinitely (Fig. 12-15). However, unlike cancer cells, which are also immortal, these cells did not acquire the ability to cause tumors. Thus, this experiment showed convincingly that telomeres are part of a mechanism that regulates the proliferative capacity of somatic cells.

ACKNOWLEDGMENTS

Thanks go to Robin Allshire, Michael Ashburner, Maurizlo Gatti, Ursula Goodenough, Paul Hieter, and Alastair Kerr for their suggestions on revisions to this chapter.