Diverse cis factors controlling Alu retrotransposition: what causes Alu elements to die?

Abstract

The human genome contains nearly 1.1 million Alu elements comprising roughly 11% of its total DNA content. Alu elements use a copy and paste retrotransposition mechanism that can result in de novo disease insertion alleles. There are nearly 900,000 old Alu elements from subfamilies S and J that appear to be almost completely inactive, and about 200,000 from subfamily Y or younger, which include a few thousand copies of the Ya5 subfamily which makes up the majority of current activity. Given the much higher copy number of the older Alu subfamilies, it is not known why all of the active Alu elements belong to the younger subfamilies. We present a systematic analysis evaluating the observed sequence variation in the different sections of an Alu element on retrotransposition. The length of the longest number of uninterrupted adenines in the A-tail, the degree of A-tail heterogeneity, the length of the 3' unique end after the A-tail and before the RNA polymerase III terminator, and random mutations found in the right monomer all modulate the retrotransposition efficiency. These changes occur over different evolutionary time frames. The combined impact of sequence changes in all of these regions explains why young Alus are currently causing disease through retrotransposition, and the old Alus have lost their ability to retrotranspose. We present a predictive model to evaluate the retrotransposition capability of individual Alu elements and successfully applied it to identify the first putative source element for a disease-causing Alu insertion in a patient with cystic fibrosis.

Figure 1.

Schematic of Alu structure and transcript. (A) Representation of a typical transcript of a genomic Alu element with functional segments defined as the left and right monomers, A-tail, and 3′ unique region. (B) Schematic of the Alu base construct used in this study with 7SL promoter-enhancer sequence upstream of the Alu and the necessary restriction enzymes used for construct manipulation. The neomycin selection cassette in reverse orientation with a self-splicing intron that allows for the retrotransposition studies is introduced after the Alu and before the A-tail. The selection cassette contains a Pol III terminator (TTTT) at the 3′ end.

Figure 2.

The length of homogeneous A-tail affects Alu retrotransposition efficiency. The retrotransposition activity of Alu elements with homogenous A-tail of different lengths driven by the endogenous ORF2p present in HeLa cells was evaluated. Columns represent the mean G418^R colonies (n = 3) normalized relative to the Alu with 30 homogenous As that was arbitrarily designated as 100% with the standard deviation shown as error bars. Results significantly different from the A30 control with P-values ≤ 0.0001 (Students paired t-test) are indicated by an asterisk (*).

Figure 3.

Older Alu elements contain shorter homogenous A-stretches within the A-tail region than younger Alus. Data mining histogram that shows the frequency of Alu elements in different bin sizes (Sx n = 276; Ya5 n = 206)

Figure 4.

Increased A-tail heterogeneity reduces Alu retrotransposition capability. HeLa cells were transiently transfected with the corresponding Alu-tagged constructs driven by endogenously or exogenously supplied ORF2p. The relative activity of Alu elements with slightly disrupted A-tails (A) or extremely disrupted A-tails (B) is shown. The Alu with 30 homogenous As (A30) was arbitrarily selected as 100%; the asterisk (*) indicates a significant difference from A30, P < 0.05 (Students paired t-test). (C) The evaluation of how specific base disruptions in the A-tail affect retrotransposition rate (T >> C > G) under exogenous conditions of ORF2p is shown. (*) Asterisk indicates a significant difference between A5C, A5G, and A3C relative to A3T, P < 0.05, P < 0.01, and P < 0.01, respectively (Students paired t-test); # indicates a significant difference between A5C and A5G, P < 0.01 (Students paired t-test).

Figure 5.

The length of the 3′ unique sequence drastically affects Alu retrotransposition capability. (A) Constructs with 30 homogenous As and variable 3′ unique region lengths (0–126 bp) were evaluated by transient transfection of HeLa cells. The activity of these Alu elements under both exogenous (black bars) and endogenous (white bars) conditions of ORF2p is shown. The A30-0, an Alu with 30 homogenous As immediately followed by a terminator (TTTTT), was arbitrarily selected as the 100%; the asterisk (*) indicates significant difference from A30-0, P < 0.01 exogenous, P < 0.03 endogenous (Students paired t-test). (B) Northern blot analysis of poly(A) selected RNA extracts was performed from cells transfected with the A30-0 control and the variable A30 3′ unique constructs. The unspliced (open arrowhead) and spliced (black arrow) neo-tagged Alu transcripts are indicated. The spliced Alu transcript from the variable length constructs were normalized to cyclophillin (C, loading control) and expressed relative to the A30-0 construct (designated as 1.00). The mean ± SD for the quantitation results for each construct are indicated below (n = 3). (C) Histogram of the length distribution of the 3′ unique sequence of young and old Alu elements. The distribution of the length of 3′ unique sequence (defined as the sequence between A-tail and the first four Ts in the 3′ genomic flank) of a subset of randomly selected AluSx (n = 289) and Ya5 (n = 227) families is shown as the frequency subdivided into bins of various sizes. (•) Alu elements containing a premature terminator within their internal dimeric sequence are included in this bin; Sx (n = 25), Ya5 (n = 1).

Figure 6.

Random right monomer mutations affect Alu retrotransposition capability. (A) Constructs of tagged Alu elements containing a randomly selected genomic AluSx right monomer with several mutations were evaluated by transient transfections in HeLa cells under exogenously supplied ORF2p conditions. The relative activity of these Alu elements is shown. The construct A30 that contains the consensus AluYa5 right monomer was designated as 100. (B) A Northern blot analysis of the steady-state level of poly(A) selected RNA from the constructs in the same order as A is shown. The unspliced (open arrowhead) and spliced (black arrow) neo-tagged Alu transcripts are indicated. The spliced Alu transcript from the variable length constructs were normalized to cyclophillin (C, loading control) and expressed relative to the A30 construct (designated as 1.00). The mean ± SD for the quantitation results for each construct are indicated below (n = 3).

Figure 7.

Identification of the candidate source element for a disease-causing Alu insert. The sequence of the disease-causing Alu compared to the candidate source element found on chromosome 14 in the genome. The Alu sequences share 100% identity within the body of the element. (A) Alu insertion disrupting Cftr gene, exon 17b (Chen et al. 2008). Flanking sequence is lower case. Inserted Alu sequence is italicized. Bold text indicates the target site duplication formed during the retrotransposition process. The box outlines the thymine mutation that is inferred to have transferred to the progeny sequence during retrotransposition. Underlined portions within the direct repeat indicate the transcription terminator. (B) The parent locus of the Cftr Alu insertion. With the flanking sequence, Alu, target site duplication, thymine residue, and terminator labeled as in A.

Figure 8.

Time frame of events influencing Alu element activity. (1) De novo Alu insertion acquiring new 5′ transcription regulators and 3′ unique sequence which dramatically affect future retrotransposition efficiency. In addition, epigenetic changes will impact transcriptional capability of the Alu insert. (2) Alu A-tail length shortening occurs rapidly after insertion to squelch future activity. (3) The quick introduction of heterogeneity in the A-tail through microsatellite expansion causing an additional reduction in retrotransposition efficiency. (4) Slower and random accumulation of random mutations throughout the element including mutations to the A and B box reducing Alu transcription efficiency, mutations in the A-tail increasing heterogeneity, and mutations throughout the left and right monomers reducing overall identity with a consensus sequence possibly affecting RNA stability and structure.