Mobile DNA in Health and Disease

The completion of the human genome project¹ has enabled the identification of many genes, including variants that cause disease.^2–4 An important discovery, although less celebrated, is that more than half the human genome is derived from mobile pieces of DNA called transposable elements (colloquially known as “jumping genes”). Transposable elements were discovered in corn by Barbara McClintock more than 60 years ago,⁵ but few people would have guessed that the aggregate length of these sequences exceeds that of protein-coding exons by a factor greater than 40 (Fig. 1).¹ Although the bulk of transposable element–derived DNAs are remnants of their former selves and cannot transpose, some retain the ability to mobilize.^11,12 The insertion of mobile elements into the DNA of gametes or the early embryo can disrupt genes, leading to sporadic cases of disease,¹³ and their insertion into the DNA of somatic cells may contribute to cancers and neuropsychiatric disease.^9,14 Clearly, mobile DNA has been instrumental in shaping the structure, function, and evolution of the human genome. Here we discuss the biology of mobile DNA, emphasizing key discoveries that illustrate how it contributes to human disease.

Figure 1. Transposable Elements in the Human Genome.

Panel A shows protein-coding exons and classes of human transposable elements as a percentage of the genome.¹ As shown in Panel B, DNA transposons encode transposase, an enzyme that binds at or near inverted repeats flanking the element (white arrows flanking transposase) to promote mobility by means of a cut-and-paste mechanism.⁶ Retrotransposons mobilize by means of a copy-and-paste mechanism through an RNA intermediate.⁷ Human endogenous retroviruses (HERVs) contain sequences analogous to the gag, pol, and env genes of retroviruses, but most HERVs, if not all of them, are not active in humans.⁸ Elements are considered to be autonomous when they encode the protein activities necessary for retrotransposition (e.g., an endonuclease [EN] and a reverse transcriptase [RT]). The arrow at the 5′ end of poly(A) retrotransposons denotes the transcription start site from their internal promoter. Nonautonomous elements do not encode proteins; their retrotransposition depends on the proteins encoded by autonomous elements.⁹ Nonautonomous elements include Alu elements and SVAs. Alu elements have a size of approximately 280 bp and consist of two monomers separated by an adenosine-rich (AR) linker; the left Alu monomer has an internal RNA polymerase III promoter (bars labeled A and B). SVAs are composite elements with a portion of a HERV (SINE-R [short interspersed element of HERV origin]), a variable number of tandem repeats (VNTRs), and a backward Alu, as well as a CCCTCT multimer. Processed pseudogenes are copies of cellular messenger RNAs that have been reverse-transcribed into DNA and lack introns. C denotes cysteine-rich domain (encoded by ORF2), HERV-K a HERV of the K type, LINE long interspersed element, LTR long terminal repeat, ORF1 open reading frame 1, ORF2 open reading frame 2, and UTR untranslated region. The diagram in Panel B is adapted from Beck et al.¹⁰

MOBILE DNA IN HUMANS

DNA TRANSPOSONS

DNA transposons are a major class of mobile elements (Fig. 1). They are active in many lower organisms, including bacteria,⁶ but have been inactivated by mutations and can no longer transpose in humans.¹ However, transposon-derived sequences have been repurposed over evolutionary time and affect human biology. For example, the recombination-activating genes RAG1 and RAG2 were probably derived from an ancient DNA transposon and are critical for adaptive immunity because they encode enzymes that generate diverse immunoglobulin proteins.^15,16 Moreover, DNA sequences derived from ancient DNA transposons have served to rewire endometrial-cell gene expression in eutherian mammals and may have influenced the evolution of pregnancy.¹⁷

RETROTRANSPOSONS

The second major class of mobile DNA is retrotransposons (see the Glossary).¹⁰ The DNA of a retrotransposon is copied into RNA, which is then copied back into DNA (the “retro” step) by a reverse-transcriptase enzyme encoded by the retrotransposon. The reverse-transcribed DNA is then integrated into the genome.⁷

Retrotransposons fall into two classes: the human endogenous retroviruses (HERVs; also known as long-terminal-repeat [LTR] retrotransposons, owing to the long repeat at each end) and the poly(A) retrotransposons. Much like DNA transposons, almost all HERVs are mutated and cannot retrotranspose in humans.^10,18 However, sequences within HERVs influence host gene expression in the early embryo,¹⁹ and HERV-derived proteins are important for placental development.²⁰ Moreover, stretches of DNA derived from endogenous retroviruses have shaped, over evolutionary time, a transcriptional network involved in the interferon response of innate immunity.²¹ HERV expression (i.e., the level of HERV messenger RNA [mRNA]) is elevated in the affected tissues of persons with rheumatoid arthritis, multiple sclerosis, or amyotrophic lateral sclerosis, and the expression of HERV-encoded accessory proteins may be involved in the development of certain cancers.^8,22 However, these are only associations; it is not known whether HERV expression is pathogenic.

The retrotransposition of LINE-1 (long interspersed element 1) poly(A) retrotransposons (which can autonomously replicate),²³ as well as the non-autonomous retrotransposon RNAs (e.g., Alu), is mediated by the LINE-1–encoded proteins, as are reverse-transcribed mRNAs (i.e., processed pseudogenes).⁹ LINE-1, Alu, SINE-R (short interspersed element of HERV origin)–VNTR (variable-number tandem repeat)–Alu (SVA), and processed pseudogenes account for a remarkable one third, or 1 billion bp, of human DNA.^1,9 There are more than 500,000 LINE-1 poly(A) retrotransposon sequences in each human genome, about 100 of which are active,^24,25 but only a small number (termed “hot LINE-1” sequences) cause most cases of LINE-1–mediated disease.²⁵

RETROTRANSPOSONS AS MUTAGENS

In the late 1980s, Kazazian and colleagues discovered that LINE-1 elements can cause disease. Of 240 boys with hemophilia A, an X-linked disorder, 2 had mutagenic LINE-1 insertions; each insertion disrupted an exon of F8 (on the X chromosome), which is the gene encoding coagulation factor VIII.²⁶ These LINE-1 insertions were not full length and therefore could not undergo further retrotransposition. The hypothesis was that both insertions were derived from full-length LINE-1 elements, and support for this hypothesis was obtained by showing that one full-length LINE-1 was present on chromosome 22 in the mother of one of the affected boys (Fig. 2). A portion of this full-length LINE-1 was identical to her son’s truncated LINE-1 insertion.¹¹ Pathogenic retrotransposition events mediated by LINE-1 have been reported in 130 persons with various diseases, including Duchenne’s muscular dystrophy, β-thalassemia trait, factor IX hemophilia, and cancers (both inherited and sporadic [i.e., occurring without a family history]).^13,27 LINE-1–mediated retrotransposition events account for approximately 1 in every 250 pathogenic human mutations²⁸; they occur in exons, introns, or regulatory regions and adversely affect gene expression.¹⁰ Moreover, the overexpression of Alu RNA in the cells of the retinal pigment epithelium has been linked to geographic atrophy, a form of age-related macular degeneration.²⁹

Figure 2. LINE-1 Retrotransposition in a Boy with Hemophilia A.

Panel A shows the family pedigree. The solid orange square denotes the affected boy, and the other symbols indicate unaffected family members. The boy’s mother is not a carrier of the insertion in exon 14 of F8 (the gene encoding factor VIII). In Panel B, the precursor LINE-1 element isolated from the mother’s DNA is shown above the insertion. The precursor was a full-length LINE-1 in chromosome 22. The smaller pathogenic insertion was identical in sequence to a portion of the precursor LINE-1. The brown arrow at the 5′ end of the precursor element denotes the transcription start site from the LINE-1 internal promoter. TSD denotes target-site duplication, and the light-gray box next to TSD indicates genomic DNA. The diagram in Panel A is adapted from Kazazian et al.,²⁶ and the diagram in Panel B is adapted from Dombroski et al.¹¹

MECHANISM OF LINE-1 RETROTRANSPOSITION

Retroviruses and HERVs “jump” by means of a rather complicated mechanism. First, their viral RNA is reverse-transcribed into DNA within a virus like particle in the cytoplasm of the cell. The full-length viral DNA is then transported into the nucleus and integrated into the genome.^8,30 In contrast, LINE-1 retrotransposition (also known as target-site–primed reverse transcription) occurs through a somewhat simpler process (Fig. 3).^31,32 Active LINE-1 elements contain two open reading frames, ORF1 and ORF2; an endonuclease encoded by ORF2 nicks one strand of DNA at the site of the new integration.³⁹ The reverse-transcriptase enzyme (also encoded by LINE-1 ORF2) then copies LINE-1 RNA, beginning at the chromosomal integration site. It is not yet clear how the second strand of LINE-1 DNA is made, although it probably involves the LINE-1 reverse transcriptase.⁴⁰

Figure 3. A Model of LINE-1 Retrotransposition.

LINE-1 RNA is transcribed from a promoter located within its 5′ UTR. The RNA is exported into the cytoplasm, where it undergoes translation. The LINE-1–encoded proteins ORF1p and ORF2p bind to the LINE-1–encoding RNA by a process known as cis-preference,^33,34 leading to the formation of a cytoplasmic complex.^35–38 Components of this complex (at least ORF2p and LINE-1 RNA) gain nuclear access, at which point the ORF2p endonuclease (EN) cleaves a single strand of chromosomal DNA at a consensus sequence (i.e., 5′-TTTT/A-3′), liberating a 3′ hydroxyl group that is used by ORF2p reverse transcriptase to copy LINE-1 RNA and integrate the resultant LINE-1 DNA into this new chromosomal location.^32,39 It is not known how the second DNA strand at the insertion site is cleaved and how second-strand LINE-1 DNA is synthesized, but the ORF2 protein probably mediates these processes. The ORF2 protein is also required for the retrotransposition of nonautonomous RNAs, such as cellular mRNAs (creating processed pseudogenes), SVA RNAs, and Alu RNAs.⁹ The ORF1 protein may aid in SVA and Alu retrotransposition.⁹ RNP denotes ribonucleoprotein particle. The diagram is adapted from Richardson et al.⁹

Knowledge of how LINE-1 elements retrotranspose has led to insight into how other retrotransposons use LINE-1–encoded proteins to move to new genomic locations. For example, the LINE-1 endonuclease and reverse-transcriptase activities of LINE-1 ORF2 are critical for the mobilization of Alu RNAs.⁴¹ In contrast, the proteins encoded by both LINE-1 ORF1 and LINE-1 ORF2 appear to be critical for the mobilization of SVA elements,^42,43 other noncoding RNAs,^44,45 and processed pseudogenes (Fig. 3).^33,34

ROLE OF LINE-1 AND ALU IN GENOMIC REARRANGEMENTS AND DISEASE

LINE-1–mediated retrotransposition not only creates insertional mutations but also results in other structural rearrangements in both LINE-1 itself and genomic DNA.^10,46–48 Such rearrangements can obstruct subsequent retrotransposition.^49,50 Indeed, only 7 of the 31 known disease-producing LINE-1 insertions are preserved, full-length LINE-1s.¹³ Examples involving genomic disruption of the sequence flanking the insertion site are a large deletion involving PDHX, leading to pyruvate dehydrogenase deficiency,⁵¹ and deletions affecting NF1, causing neurofibromatosis.⁵²

LINE-1–mediated retrotransposition generally requires the endonuclease activity encoded by its ORF2. However, on rare occasions, LINE-1 elements can act as chromosomal bandages by integrating at preexisting DNA lesions, such as dysfunctional telomeres, through a process known as endonuclease-independent (ENi) retrotransposition.^53,54 Indeed, an ENi-retrotransposition–mediated insertion of LINE-1 into EYA1, accompanied by a genomic DNA deletion, was found in a person with the branchiootorenal syndrome.⁵⁵

The abundance of retrotransposon-derived DNA provides ample templates for disease-producing DNA recombination events (Fig. 4). For example, unequal crossing-over events mediated by mispaired Alu elements in the low-density lipoprotein receptor gene have caused familial hypercholesterolemia and more than 70 cases of other diseases.^57,58 In addition to causing monogenic disease by disrupting single genes, Alu-mediated recombination events can generate copy-number variations in the human genome.⁵⁹ Nearly 500 Alu-mediated deletions are known to have occurred since the divergence of chimpanzees and humans.⁶⁰ Mispairing and unequal crossing over of LINE-1s, although less common than Alu-mediated recombination events, has resulted in sporadic cases of genetic disease.¹⁰ Increased use of whole-genome sequencing to diagnose disease will probably uncover more retrotransposon-mediated pathogenic events.

Figure 4. DNA-Level Recombination Events and Genome Instability.

Mispairing of Alu elements or LINE-1 elements, followed by crossing over and unequal nonallelic homologous recombination, leads to the deletion or duplication of the genomic sequence. Exons and introns (black lines) of the hypothetical gene are shown. The diagram is adapted from Hulme et al.⁵⁶

LINE-1–MEDIATED RETROTRANSPOSITION EVENTS

Retrotransposition events mediated by LINE-1 are estimated to occur, at a minimum, in 1 of 20 meioses for Alu, 1 of 20 to 200 meioses for LINE-1, and 1 of 900 meioses for SVA.⁵⁹ These events are the cause of a great deal of interindividual variation in the population. For example, the genomes of any two unrelated persons are likely to differ at about 300 LINE-1 insertion sites.⁶¹ Similarly, any two unrelated persons are likely to have many differences in Alu and SVA insertion sites; some of these differences probably affect gene expression.

Some disease-producing LINE-1–mediated insertions occurred many generations ago. For example, an SVA insertion into an intron of FCMD (the gene that, when mutated, causes Fukuyama-type congenital muscular dystrophy) causes mis-splicing and the production of a mutant form of the fukutin protein.⁶² This SVA insertion is specific to the Japanese population, causes most cases of Fukuyama-type congenital muscular dystrophy, and has been segregating within the population for many generations.

LINE-1–mediated retrotransposition events occur in somatic cells of the early human embryo^63,64 and can occur in human embryonic stem cells.⁶⁵ Similarly, studies in transgenic mice and rats revealed that engineered human LINE-1 retrotransposition was more prevalent in the blastocyst and morula stages of early development than during spermatogenesis.⁶⁶

What other somatic cells are susceptible to LINE-1–mediated insertions? Human LINE-1 can retrotranspose in various regions of the brain in transgenic mice and in neural progenitor cells differentiated from embryonic stem cells in humans.^67,68 Studies of DNA derived from either bulk brain tissues or single neurons have confirmed that LINE-1 can retrotranspose in the brain.^69–71 It remains unclear whether brain-cell types differ in their capacity for LINE-1 retrotransposition and whether somatic insertions influence behavior and susceptibility to psychiatric disorders.^14,72 There have been no detailed studies of somatic retrotransposition in human tissues other than the brain and the gastrointestinal tract.

LINE-1 retrotransposition occurs in various cancers.^73–83 Studies have shown that somatic LINE-1 insertions primarily occur in epithelial cancers (e.g., those in the gastrointestinal tract); that the number of LINE-1 insertions varies among epithelial tumors (with some having >50 somatic insertions and others having none); that certain clonal somatic insertions in esophageal and gastric tumors are present at low frequencies in normal tissue, suggesting that a normal cell harboring a somatic LINE-1 insertion may be clonally expanded in the cancer^78,79,81; and that some types of cancers (e.g., certain hematopoietic and brain cancers) lack somatic LINE-1 insertions.

Whether LINE-1 insertions represent “driver” or “passenger” mutations in cancers is unclear. Most somatic LINE-1 insertions found in epithelial cancers probably represent passenger mutations.⁸⁴ However, some somatic LINE-1 insertions, such as those reported to inactivate the tumor-suppressor genes APC^27,85 and PTEN,⁷⁷ probably promote tumorigenesis.

DEFENSES AGAINST MOBILE ELEMENTS

The ability to replicate is critical for the evolutionary survival of retrotransposons. Because the resultant insertions can act as mutagens, evidence supporting the evolution of mechanisms to combat retrotransposition comes as no surprise.⁸⁶

Methylation of DNA at CpG dinucleotides (sites at which a cytosine precedes a guanine in the DNA sequence) restricts retrotransposon expression in both the germline and soma. Gene-knockout experiments in mice have shown that disruption of a gene encoding an enzyme that aids in the methylation of CpG motifs⁸⁷ leads to a loss of CpG methylation, derepression of LINE-1s and certain endogenous retroviruses in the male germline, and meiotic arrest, resulting in male infertility. Whether the derepression of retrotransposon expression causes meiotic catastrophe or leads to increased retrotransposition requires further study. Small noncoding RNAs are also known to affect CpG methylation and transcriptional silencing of retrotransposons in the male germline.^86,88

Certain zinc-finger proteins can recruit protein complexes to sequences that reside within retrotransposons, leading to the deposition of repressive histone modifications and transcriptional silencing of retrotransposons.^89–92 For example, zinc-finger protein 91 (ZNF91) was found to direct the deposition of repressive histone modifications on human SVA elements, and zinc-finger protein 93 (ZNF93) has a similar effect on older LINE-1 subfamily members.⁹³ An active LINE-1 has a deletion that allows it to escape ZNF93-mediated repression.⁹³

Various cytoplasmic pathways influence the stability of retrotransposon RNAs, their translation, or both, and certain proteins inhibit retroviral replication and retrotransposon mobility.^9,94,95 Some of these proteins colocalize with LINE-1 RNAs in cytoplasmic stress granules, suggesting that sequestration to stress granules may play a role in the degradation or translational repression of LINE-1 RNAs.^96,97 Finally, since full-length LINE-1 RNAs contain functional splice sites and polyadenylation signal sequences, proteins involved in RNA processing may inhibit LINE-1 retrotransposition.^98,99

CONCLUSIONS

Since the discovery nearly 30 years ago that LINE-1 insertions can cause human disease, research has yielded insights into how LINE-1 mobilizes, disrupts genes, and causes disease. Many questions remain to be addressed. Does somatic retrotransposition play an important role in the development of certain cancers? Can somatic retrotransposition in the brain affect human behavior? How often are sequences within retrotransposons coopted by the host for functional purposes? These are just a few areas for future research. Although we have focused on the harmful effects of retrotransposons in this review, it is clear that transposable elements are not just weeds in the garden. They provide the fertile soil that is the fodder for the evolution of genomes.¹⁰⁰

Glossary

Accessory protein

A protein that is not required for viral or retrotransposon replication but that plays an indirect regulatory role in the function of a virus or retrotransposon.

Alu

A short (approximately 280 bp) element of which 1.1 million copies are present in the human genome. Some Alu elements move to new genomic locations by means of the protein encoded by the second open reading frame (ORF2) of LINE-1.

Autonomous transposable elements

Elements that encode proteins that are required for the mobility of the transposable elements throughout the genome.

Copy-number variation

Interindividual genetic variation caused by either the increase or decrease of a block of DNA sequence within the genome.

CpG dinucleotides

Sites at which a cytosine precedes a guanine in the DNA sequence and DNA methylation occurs at the cytosine in the human genome.

DNA transposon

A DNA mobile element containing transposase, an enzyme that allows the transposon to cut itself out of one genomic site and insert itself into another site.

Exon

The protein-coding and untranslated-region (UTR) sequences of a messenger RNA (mRNA). Exons remain in the mRNA after splicing.

Human endogenous retrovirus (HERV)

A retrovirus-like sequence accounting for approximately 8% of the human genome. These DNAs are immobile retrotransposons and usually contain defective envelope genes. Over evolutionary time, sequences within HERVs have been “exapted,” or co-opted, for functional use in humans (e.g., placental syncytins).

Histone

An octameric protein complex that packages DNA into structures called nucleosomes.

Intron

A sequence that resides between exons in the precursor of mRNA and is spliced out of the RNA.

Long interspersed element 1 (LINE-1 or L1)

Repeated DNA elements that are present at approximately 500,000 copies in the human genome and account for approximately 17% of the human genome. Some autonomous LINE-1 elements can retrotranspose. The LINE-1–encoded proteins can also retrotranspose nonautonomous elements, such as Alu and SVA. LINE-1 retrotransposons are unable to move from cell to cell.

LINE-1 endonuclease

An enzyme encoded by LINE-1 that can cut one strand of a DNA double helix at a consensus sequence at the insertion site for a retrotransposition event.

Mispairing and unequal crossing over

Mispairing of chromosomal homologues when two homologous sequences are in proximity on the chromosome. For example, Alu1 is close to Alu2. Mispairing leads Alu1 on one chromosome 1 to pair with Alu2 on the other chromosome 1. Recombination of Alu1 and Alu2 results in unequal crossing over, leading to deletion on one chromosomal homologue and duplication on the other.

Nonautonomous retrotransposon

A transposable element that does not encode proteins but instead relies on the proteins encoded by an autonomous transposable element (LINE-1) to move to a new genomic location.

Polyadenylation signal sequence

A specific signal sequence (5′-AAUAAA-3′) near the back end (3′ end) of an RNA that signals cleavage of the RNA roughly 20 nucleotides downstream of the signal. After cleavage of the RNA, a poly(A) tail is added at the cleavage site.

Processed pseudogene

A reverse-transcribed copy of an mRNA that is inserted back into the genome with the use of LINE-1 proteins. Processed pseudogene sequences lack the introns present in precursor mRNAs.

Retrotransposon

A piece of DNA that can be transcribed into RNA and then reverse-transcribed into complementary DNA (cDNA), with the cDNA copy then reinserted into the genome at a new location by a copy-and-paste process. Human retrotransposons are LINE-1, Alu, and SVA.

Retrovirus

A virus that encodes gag, pol, and env proteins. The gag protein forms a capsule around the viral RNA. The pol protein contains reverse-transcriptase and integrase activity. The env protein aids the virus in entering and exiting cells. Retroviruses and HERVs have similar sequences, but the HERV env protein is usually defective, precluding its movement from one cell to another.

Reverse transcriptase

An enzyme that copies RNA back into DNA. Some LINE-1 elements and HERVs encode reverse-transcriptase activity.

SINE-R–VNTR–Alu (SVA)

A primate-specific retrotransposon present in approximately 2700 copies in the human genome that can retrotranspose with the help of LINE-1–derived proteins. SINE-R denotes short interspersed element of HERV origin, and VNTR variable-number tandem repeat.

Target-site–primed reverse transcription

The mechanism by which LINE and SINE retrotransposons, as well as processed pseudogenes, move to new genomic locations.

Transposable elements

Segments of DNA that can move from one location in DNA to another.

Untranslated region (UTR)

A sequence in mRNA or LINE-1 RNA that is not translated into protein.