Genomic database mining has been a very useful aid in the identification and retrieval of recently integrated Alu elements from the human genome. We analyzed Alu elements retrieved from the GenBank database and identified two new Alu subfamilies, Alu Yb9 and Alu Yc2, and further characterized Yc1 subfamily members. Some members of each of the three subfamilies have inserted in the human genome so recently that about a one-third of the analyzed elements are polymorphic for the presence/absence of the Alu repeat in diverse human populations. These newly identified Alu insertion polymorphisms will serve as identical-by-descent genetic markers for the study of human evolution and forensics. Three previously classified Alu Y elements linked with disease belong to the Yc1 subfamily, supporting the retroposition potential of this subfamily and demonstrating that the Alu Y subfamily currently has a very low amplification rate in the human genome.
ALU elements have been accumulating in the human genome throughout primate evolution, reaching a copy number of over a million per genome. However, most of these Alu copies are not identical and can be classified into several subfamilies (reviewed in Deininger and Batzer 1993). These different subfamilies of Alu elements were generated once mutations occurred within the “master” or “source” gene that actively retroposed at different rates and time periods of primate evolution (Deiningeret al. 1992). Currently, the Alu retroposition rate is reduced by 100-fold from its peak early in primate evolution (Shenet al. 1991). The vast majority of the Alu elements present in the human genome inserted before the radiation of extant humans and are therefore observed in all individuals in the human population. However, almost all of the recently integrated Alu elements in the human genome are restricted to several closely related “young” subfamilies, with the majority being Ya5 and Yb8 subfamily members (Batzer et al. 1994, 1995). Several of these new subfamilies appear to originate from an Alu element that fortuitously inserted into a favorable region of the genome capable of supporting Alu retroposition. Subsequent or concurrent mutations in the new source element(s) result in groups of elements that are identifiable as new subfamilies.
Collectively, the Alu Y, Ya5, Ya5a2, Ya8, and Yb8 subfamilies comprise <10% of the Alu elements present within the human genome, with the Ya5/8 and Yb8 subfamilies together accounting for <0.5% of all Alu elements. Although the human genome contains >1,000,000 copies of Alu (~10% of the genome; Smit 1996), <0.5% are polymorphic. Due to their recent evolutionary introduction into the human genome, many of the young Alu elements are polymorphic between individuals and/or populations. There is an inverse correlation between the age of the Alu subfamily and the percentage of polymorphic elements it contains. Identification of evolutionarily recent Alu subfamilies and their polymorphic insertions is useful for human population studies, forensics, and DNA fingerprinting for two reasons: (i) There is no apparent specific mechanism to remove newly inserted Alu repeats, making inserts identical by descent; and (ii) the Alu insertions have a known ancestral state (Batzer and Deininger 1991; Batzeret al. 1994).
The availability of large quantities of human genomic DNA sequence provided by the Human Genome Project facilitates genomic database mining for recently integrated Alu elements. Through this approach we were able to identify the youngest Alu subfamily reported to date, termed (Ya5a2), and determined that the majority of its members are Alu insertion polymorphisms (Royet al. 2000). We expanded our computational analyses to identify other Alu subfamilies derived from the Alu Y and Yb8 subfamilies. Here, we present the analysis of three of the most recently formed Alu subfamilies and demonstrate their utility for the study of human genomic diversity.
MATERIALS AND METHODS
Computational analyses: Sequence alignments for the identification of Alu subfamilies were made using MegAlign software (DNAStar version 3.1.7 for Windows 3.2). Screening of the GenBank nonredundant (nr), the high throughput genome sequence (htgs), and the genomic survey sequence (gss) databases was performed using the advanced basic local alignment search tool 2.0 (BLAST; Altschulet al. 1990) available from the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/). Database searches for Yb8 consensus Alus showed a common single-base variant termed Yb9. The databases were searched for matches to the 289 bases of the Yb9 consensus sequence (as inferred from the previous Yb8 analysis) or the 281 bases of the Alu Y consensus with the expected value (real) set at −e 1.0e−150 and −e 1.0e−140, respectively, in the advanced BLAST options. Only Alu Yb9 elements with all nine diagnostic mutations were selected. A similar type of search procedure was performed with the Yc1 and Yc2 consensus sequences or with an oligonucleotide query sequence complementary to the subfamily diagnostic base positions. Only Alu Yc1/Yc2 elements with 100% identity to the oligonucleotide query sequences or entire subfamily-specific consensus sequence were utilized for further analysis. To estimate the copy numbers of the Yb9 subfamily we searched the draft sequence of the human genome (Landeret al. 2001), using a subfamily-specific probe that contained the Yb9-specific mutation as well as the insertion in the Yb8 subfamily. A complete list of the Alu elements identified from the GenBank search is available from M. A. Batzer or P. L. Deininger.
DNA samples: Human DNA samples from the European, African-American, Alaskan Native, Egyptian, and Asian population groups were isolated from peripheral blood lymphocytes (Ausubelet al. 1996) that were available from previous studies (Royet al. 1999).
Oligonucleotide primer design and PCR amplification: Flanking unique DNA sequences adjacent to each Alu repeat were used to design primers for the Yb9, Yc1, and Yc2 Alu elements (Table 1). PCR primers and reactions were performed as previously described (Royet al. 1999). The heterozygosity associated with each element was determined by the amplification of 20 individuals from each of four populations (African American, Alaskan Native or Asian, European, and Egyptian; 160 total chromosomes). The chromosomal location for elements identified from randomly sequenced anonymous large-insert clones was determined by PCR as previously described (Royet al. 1999).
The Alu Yb9, Yc1, and Yc2 subfamilies: Analysis of a set of 243 Yb8 Alu elements retrieved from the GenBank database allowed us to identify a putative subfamily containing all the known Yb8 diagnostic mutations plus one new mutation, which is referred to as Yb9 in compliance with the standard Alu subfamily nomenclature (Batzeret al. 1996). The Yb9 consensus sequence is shown in Figure 1. Searches from the nr, the htgs, and gss retrieved a total of 56 Yb9 elements. Of these, 25 elements were retrieved from the nr database (30.4% of the human genome at the time), giving an estimated size of 82 members for the Yb9 subfamily. This estimate is also in good agreement with a search of the draft human genomic sequence (Landeret al. 2001) that identified 79 perfect matches with a Yb9 subfamily-specific query sequence.
Using a different approach, we also retrieved one previously identified subfamily, Yc1 [formerly termed Sb0 (Jurka 1995)], and a new variant, Yc2. GenBank database searches for Alu Y elements that perfectly match the consensus sequence brought several Alu Y elements to our attention that share one or two specific mutations that differ from the Y consensus. Closer inspection facilitated the retrieval of the additional Alu subfamilies. BLAST searches using the consensus sequence for Alu Yc1 and Yc2 will also retrieve a large number of elements that are matches to the Alu Y subfamily as well, making the analysis of the elements identified in this manner impractical. Therefore, we selected only the elements of these subfamilies with 100% identity to the oligonucleotide query sequence that contained the subfamily-specific diagnostic bases. A total of 176 Yc1 (13 perfect matches to the entire subfamily consensus sequence) and 17 Yc2 (11 perfect matches to the entire subfamily consensus sequence) elements were retrieved. A count of all Yc1 elements retrieved by BLAST on a single initial search of the nr database yielded a total of 116 elements, giving an estimated copy number of 381 Yc1 elements in the human genome (the nr database contained 30.4% of the human genome sequence at the time of the search). Interestingly, three of the four elements previously classified as Alu Y elements linked to disease (Deininger and Batzer 1999) belong to the Alu Yc1 subfamily (Figure2): the de novo insertion in the C1 inhibitor gene (C1inh; Stoppa-Lyonnetet al. 1990), another de novo insertion in BRCA2 (BRCA2; Mikiet al. 1996), and glycerol kinase deficiency (GK; Zhanget al. 2000).
About one-half of the 56 total Yb9 elements (29) shared 100% nucleotide identity with the subfamily consensus sequence. To get an approximation of the age of the Yb9 subfamily, we evaluated the number of non-CpG mutations present within the different Alu elements as previously described (Royet al. 2000). A total of 19 CpG mutations, 25 non-CpG mutations, and two 5′ truncations occurred within the 56 Alu Yb9 subfamily members identified. Using a neutral rate of evolution for primate intervening DNA sequences of 0.15% per million years (Miyamotoet al. 1987) and the non-CpG mutation density of 0.1908% (25/13,104 bases using only non-CpG bases) within the 56 Yb9 Alu elements yield an estimated average age of 1.27 million years (myr). The age for the Yb9 subfamily members is predicted at a 95% confidence level in the range of 0.8–1.8 myr, given that the mutations were random and fit a binomial distribution. No analysis can be made for the Yc1 and Yc2 Alu elements, because only subfamily members with perfect identity to the subfamily consensus sequence or one mismatch were isolated from the database using one of the database screening procedures.
Phylogenetic distribution and human genomic diversity of the new subfamilies: Amplification of the Yb9, Yc1, and Yc2 elements from nonhuman primate genomes facilitated the analysis of the phylogenetic distribution of these elements, using PCR and the oligonucleotide primers in Table 1. Almost all of the elements evaluated were absent from the genomes of the nonhuman primates, suggesting that these elements dispersed and were fixed in the human genome after the human and African ape divergence.
We performed a PCR analysis on a panel of human DNA samples to determine the levels of human diversity associated with the Alu elements from these new subfamilies, using the oligonucleotide primers shown in Table 1. The panel consists of 20 individuals of European origin, African-Americans, Asians, and Egyptians for a total of 80 individuals (160 chromosomes). We were able to analyze 28 out of the 56 Yb9 elements, 97 out of 176 Yc1 elements, and 8 out of 17 Yc2 Alu elements, using this approach. Several factors did not allow for analysis of all the elements. Mainly, we were unable to design appropriate primers due to insufficient flanking unique DNA sequences or because the element analyzed resided within another type of repeat as described previously (Batzeret al. 1991). The Alu elements were classified as fixed present and high, intermediate, or low frequency insertion polymorphisms (see Table 1 for definitions). In general, we observed that approximately one-fourth to one-third of the elements analyzed had some degree of insertion polymorphism (Yb9 with 10/28, Yc1 with 24/97, and Yc2 with 3/8). The population-specific genotypes and levels of heterozygosity for each element are shown in Table 2. The high proportion of polymorphic elements in these Alu subfamilies is in good agreement with our previous observations, indicating that these subfamilies are very recent in origin and still actively retroposing within the human genome.
From our subset of AluYb8 and Y elements, we were able to retrieve three Alu subfamilies termed Yb9, Yc1, and Yc2. A schematic of the evolutionary relationship of these subfamilies with the previously defined Alu subfamilies is shown in Figure 3. Alu subfamilies arise as a result of mutations occurring in an existing master element or new source elements capable of significant amplification. In this case, the new subfamilies are presumably examples of Alu subfamilies that may have originated from the rare instances when an Alu element fortuitously becomes both transcriptionally and retropositionally active, therefore allowing it to be another Alu source gene.
The young Alu subfamilies are currently active with respect to retroposition, whereas the older Alu subfamilies typically are not. The old Alu subfamilies (Sx, J, and Sg1), which comprise the vast majority (>1,000,000 copies) of the Alu elements present in the human genome, appear completely inactive as none of their members have been associated with de novo Alu inserts that result in human diseases (Table 3). When noting the ratio of reported Alu insertions associated with diseases and the estimated size of the Alu subfamily, the younger subfamilies Ya5, Yb8, and Yc1 currently appear to be ~1000 times more active than the Alu Y subfamily with 7/2640, 3/1852, and 3/400 compared to 1/200,000 (Table 3). The Alu Ya5a2 subfamily appears to have even a higher current retroposition rate (1/40), but the very young age and small size of the subfamily may be an influencing factor. In general, two independent observations support the current mobility of these young Alu subfamilies within the human genome. First, there are examples of Alu inserts that have caused disease that belong to these young subfamilies. Second, the subfamilies have a high proportion of Alu insertion polymorphisms between individuals/populations (Table 3), indicating the recent proliferative/amplification activity of these Alu elements in the human genome.
Alu elements that are polymorphic for insertion presence/absence have previously proven useful for the study of human population genetics and forensics (Batzer et al. 1991, 1994; Pernaet al. 1992; Novicket al. 1993; Hammer 1994; Tishkoffet al. 1996; Stonekinget al. 1997; Majumderet al. 1999; Comaset al. 2000; Jordeet al. 2000; Watkinset al. 2001). The identification of very young Alu subfamilies with a high proportion of polymorphic members provides new sources of Alu insertion polymorphisms for the study of human population genetics. However, it is important to note that an exhaustive analysis of these small subfamilies will only generate a relatively small number of new Alu insertion polymorphisms.
Master element vs. source gene: Alu elements have been proposed to fit an evolutionary model where the copies arose from “master” genes (Deininger and Slagel 1988; Labuda and Striker 1989; Shenet al. 1991; Deiningeret al. 1992). A master gene can be defined as an element that is highly active during a long period, therefore generating a lot of copies of itself. However, we demonstrated that recently inserted Alu elements (de novo) belong to a variety of Alu subfamilies, indicating the simultaneous presence of multiple active elements in the human genome. These active elements that have a low rate of amplification and are only active for a very short period of time should not be classified as master genes. To distinguish between them, we suggest the use of the nomenclature of “master gene” when referring to the highly active genes for long evolutionary periods of time, like the Alu element that generated the majority (>90%) of the Alu elements currently present in the genome today. For those copies, or daughters, that acquired the ability to retropose we propose the use of the term “source genes.” However, some of the elements classified as source genes may be potential master genes, and only the progression of time will allow the appropriate distinction to be made.
Evolutionary reduction in the Alu retroposition rate: Our data indicate the existence of several currently active Alu elements that belong to different subfamilies within the human genome. However, the present amplification rate of Alu elements has drastically decreased from when it reached its peak between 35 and 60 million years ago (mostly Sx subfamily). The majority of the Alu elements present in the genome of extant humans inserted during this peak amplification period. There are multiple reasons that could explain the reduction in the amplification rate of Alu elements. First, mutations within or near the master Alu element could reduce its retroposition activity or even totally abolish it by a variety of mechanisms (Deininger and Batzer 1993; Schmid 1996). Alternatively, mutations within the master gene or in the LINE elements that affect the ability to “parasitize” LINE element-encoded enzymes necessary for retroposition could also reduce the Alu amplification rate. Furthermore, the host may have also evolved cellular mechanisms to reduce Alu proliferation. Finally, the availability of suitable genomic “insertion sites” may be reduced, since most evolutionarily neutral or positive sites are presumably already “filled” with different types of preexisting repeats. Alternatively, new Alu insertions may result in unacceptable local levels of unequal homologous recombination (Deininger and Batzer 1999).
AMR was supported by a Brown Foundation fellowship from the Tulane Cancer Center. This research was supported by National Institutes of Health RO1 GM45668 (P.L.D.); Department of the Army DAMD17-98-1-8119 to (P.L.D. and M.A.B.); Louisiana Board of Regents Millennium Trust Health Excellence Fund HEF (2000-05)-05 and HEF (2000-05)-01 (M.A.B. and P.L.D.); and award 1999-IJ-CX-K009 from the Office of Justice Programs, National Institute of Justice, Department of Justice (M.A.B.). Points of view in this document are those of the authors and do not necessarily represent the official position of the U.S. Department of Justice.
Communicating editor: Y.-X. Fu
- Received February 24, 2001.
- Accepted June 8, 2001.
- Copyright © 2001 by the Genetics Society of America