Full text of DOI:10.1007/s002390010150

J Mol Evol (2001) 52:215–231
DOI: 10.1007/s002390010150
© Springer-Verlag New York Inc. 2001
Ikirara Insertions Reveal Five New Anopheles gambiae Transposable
Elements in Islands of Repetitious Sequence
Sharon R. Hill, Sheldon S. Leung, Nada L. Quercia, Dan Vasiliauskas, Jessica Yu, Ivan Pasic, Derek Leung,
Anne Tran, Patricia Romans
Department of Zoology, University of Toronto, 25 Harbord Street, Toronto, Ontario, Canada M5S 3G5
Received: 29 July 1999 / Accepted: 31 October 2000
Abstract. Characterization of Anopheles gambiae ge-
nomic clones containing Ikirara inverted repeats re-
vealed five novel sequences related to known transpos-
able elements (TEs). One TE is related to the mariner/
Tc1 superfamily of class II (DNA-to-DNA) transposons,
while four are related to class I (RNA-mediated transpo-
sition) elements. Crusoe, the class II element, is most
similar to the Caenorhabditis elegans transposon Tc1-
like TEs. Vash elements, represented twice in our clones,
are related to the Q/T1 family of A. gambiae non-LTR
retrotransposable elements. Guildenstern is a member of
the RT1 and RT2 non-LTR retrotransposon family. Al-
though RT1 and RT2 elements normally have a highly
stereotyped insertion preference for sequences within ri-
bosomal genes, Guildenstern is not located in ribosomal
sequence. JuanAg is the first anopheline member of the
mosquito non-LTR retrotransposon family of Juan ele-
ments that previously had included just the culicine ele-
ments JuanA and JuanC. Approximately 753 bp is miss-
ing from the central portion of the JuanAg reverse
transcriptase gene, where an Ikirara inverted repeat is
found in its stead. Ozymandias, the only LTR retrotrans-
poson found in the clones, is most similar to the Dro-
sophila melanogaster 412 element. Single Ikirara in-
verted repeats were also found adjacent to
nontransposable element repetitious sequences. Our
analysis suggests that the A. gambiae genome organiza-
tion could best be described as islands of short-period
interspersion repetitious DNA in a sea of long-period
interspersion, mostly unique sequence DNA.
Key words: Anopheles gambiae — Malaria vector
mosquito — Transposon — Ikirara — Crusoe — Vash
— Ozymandias — Guildenstern — JuanAg — Genome
organization
Introduction
Currently we have no efficient and stable means of trans-
ferring genes into the germline of the Anopheles gambiae
mosquito despite a single event of stable transformation
(Miller et al. 1987) and recent successes in A. stephensi
(Catteruccia et al. 2000) and the culicine mosquito Aedes
aegypti (Sarkar et al. 1997a; Coates et al. 1998a, b; Ja-
sinskiene et al. 1998; Thibault et al. 1999). mariner/Tc1
and hAT superfamily TEs successful in causing germline
transformation of other mosquito species (i.e., Mos1,
Hermes, and Minos) have not yet resulted in stable A.
gambiae germline transformation. Recently the Lepidop-
teran TE piggyBac has been demonstrated to transpose in
both Ae. aegypti and A. gambiae embryos (Lobo et al.
1999; Grossman et al. 2000) and therefore shows prom-
ise as a potential general mosquito transformation vector.
Nevertheless, the identification or assembly of other TEs
suitable for use as gene transfer vectors in malaria-
transmitting mosquitoes would facilitate vital in vivo
studies of gene expression, of mosquito/Plasmodium in-
teractions, of vector competence, and of the development
of genetic vector control strategies.
Correspondence to: Patricia Romans. email: victorie@zoo.utoronto.ca
Historically DNA-to-DNA TEs, class II elements ac-

216
cording to Finnegan’s (1992) classification, have proven
to be the most successful vectors for creating transgenic
insects. The most widely known and used class II ele-
ment is the Drosophila melanogaster P-element. How-
ever, P has very restricted activity in species other than
D. melanogaster (O’Brochta and Handler 1988;
O’Brochta et al. 1991; Handler et al. 1993) due to its
requirement for additional host-specific factors to carry
out transposition (Beall et al. 1994; Adams et al. 1997;
Hammond et al. 1997). Other class II elements unrelated
to P, including mariner, Tc1, Minos, Sleeping Beauty,
hobo, and Hermes, have proven useful as gene transfer
vectors. These elements are considered more promising
than P as candidates for utilization as general gene trans-
fer vectors in insects because they do not rely heavily on
host factors for efficient transposition (O’Brochta et al.
1994, 1996; Pinkerton et al. 1996; Vos et al. 1996; Han-
dler and Gomez 1996; Lampe et al. 1996; Ivics et al.
1997; Sarkar et al. 1997b; Coates et al. 1998a, b; Luo et
al. 1998; for review of hAT TEs see O’Brochta and At-
kinson 1996).
Different genomes contain varying proportions of
highly repetitious and moderately repetitious DNA se-
quences, both TE and non-TE. The euchromatic regions
of insect genomes have been classified as having either
long (D. melanogaster)- or short (Ae. aegypti)-period
patterns of DNA interspersion. Long-period intersper-
sion is defined by Britten and Davidson (1976) as long
stretches of repetitious sequence (>5.6 kb) alternating
with longer (>12 kb) stretches of single-copy DNA. The
short-period interspersion pattern consists of unique se-
quence DNAs (1–2 kb long) alternating with short (200-
to 600-bp) and moderately long (1 to 4 kb) repetitious
sequences. Large-genome organisms typically display
short-period DNA interspersion patterns and those with
smaller genomes typically display long-period intersper-
sion (Hoy 1994). The short-period Schistocerca gregaria
genome is 9.3 × 106 kb as cited by Hoy (1994), 2.1 times
the size of the maize genome [4.5 × 106 kb (White and
Doebley 1998)], which is 3.4 times larger than the Ae.
aegypti genome [1.32 × 106 kb (Rao and Rai 1987)]. The
Ae. aegypti genome is 5 times larger than the A. gambiae
genome [2.6 × 105 kb (Besansky and Powell 1992)],
which, in turn, is 1.6 times larger than the long-period D.
melanogaster genome [1.65 × 105 kb (Rudkin 1969)].
Repetitious sequences, both TE and non-TE, have
been suggested to be important in genome evolution.
These sequences have been implicated in local, regional,
and global regulation of gene expression and in mainte-
nance of chromosome structure (nucleoskeleton) (Ma-
thiopoulos et al. 1998, 1999). TEs specifically, can cause
genome reorganization, especially when their movement
is triggered by environmental stresses (White and Doe-
bley 1998). It is important to note that if the TEs em-
ployed in transformation are as promiscuous as mariner
(Weinberg 1998), or can be mobilized by endogenous
related elements, the resulting transgenic lines may be
unstable. Therefore it is desirable to investigate the dis-
tribution and mobility properties of all newly discovered
TEs. Elements found in any one species might be used
for a variety of gene identification techniques such as
enhancer and promoter trapping and transposon muta-
genesis in that species and for stable germline transfor-
mation in other species that lack them.
Recently, we reported the discovery of Ikirara1, a
potentially active class II TE in A. gambiae (Romans et
al. 1998; Leung and Romans 1998). Ikirara1 is a 610-bp
internally deleted TE with 216 bp perfectly matched in-
verted terminal repeats (IRs). The terminal nucleotides of
the IRs fit the Tc1/hAT consensus (Besansky et al. 1996).
The IRs are also flanked by TA, the canonical insertion
site duplication for mariner/Tc1 superfamily elements.
When an Ikirara1-based construct containing a lucifer-
ase reporter gene was transfected into A. gambiae MOS-
55 cells, it exhibited stereotyped excision from its parent
plasmid. The excisions appeared to have occurred by a
mechanism formally similar to those demonstrated for
mariner, Tc1, and Minos (Lampe et al. 1996; Vos et al.
1996; Arca et al. 1997), except that endonucleolytic
cleavage of the two DNA strands at the IR termini was
staggered by 15 bp rather than 2, 3, or 4.
In an attempt to characterize the Ikirara transposase
gene, we isolated five additional genomic clones con-
taining at least one Ikirara inverted repeat (I-IR). We did
not find any Ikirara elements unambiguously larger than
a single I-IR. However, by analyzing sequence from
these clones we found five new A. gambiae TEs. One TE
is clearly a member of the mariner/Tc1 superfamily of
class II elements, while the four others are related to
class I retrotransposons (RNA-to-DNA TEs). Character-
ization and phylogenetic analyses of these novel TEs are
reported here. I-IR insertions were also found in regions
of non-TE-related repetitious sequence. Parts of two host
genes were found in close association with TE and non-
TE repetitious sequences. These observations suggest a
model for A. gambiae genome organization that includes
“islands” of short-period DNA interspersion of repeti-
tious DNA within a “sea” of long-period interspersion,
primarily unique sequence, DNA.
Materials and Methods
Mosquito Species, Strain, and Rearing
Anopheles gambiae G3 mosquitoes were obtained from the National
Institutes of Health (Bethesda, MD) and Centers for Disease Control
(Atlanta, GA) and reared using standard conditions (Vanderberg and
Gwadz 1980). G3 is a standard laboratory strain in widespread use.
Ikirara Clone Isolation
The A. gambiae G3 genomic library in ␭EMBL3 (Romans 2001) was
screened for additional Ikirara-containing clones by conventional

217
methods (Sambrook et al. 1989) using a plasmid subclone of the 154-bp
SacII–SspI fragment contained within each Ikirara1-IR (Romans et al.
1998). DNA was prepared from purified I-IR-containing bacte-
riophages and cut with restriction enzymes, and the presence of I-IR
sequences verified on Southern blots. Plasmid subclones in pBluescript
II KS- (Stratagene) were made from selected recombinant phages, as
needed to facilitate further analysis. All probes were labeled with ␣32P-
dCTP by random priming using an Oligolabelling Kit (Pharmacia).
Southern blots were prepared on maximum-strength Nytran (Schleicher
and Schuell) as described previously (Miller et al. 1987; Romans et al.
1991).
analyses to calculate relationships among elements and to describe the
results in the form of phylogenetic trees. We used the PHYLIP 3.5c
suite of computer programs (Felsenstein 1988) to calculate the NJ and
UPGMA trees and PAUP 3.1.1 (Swofford 1993) for the parsimony
analyses.
Results
Isolation and Analysis of Additional Ikirara Elements
DNA Sequencing and Analysis
Despite its long completely identical inverted terminal
repeats, Ikirara, at 610 bp in length, is too small to be the
source of an active transposase and contains no signifi-
cantly long protein coding regions (Romans et al. 1998).
PCR screening of genomic and cloned DNA was unsuc-
cessful in identifying other Ikirara elements, probably
due to the length of the IRs and to secondary structure
within them. We therefore searched for the Ikirara trans-
posase gene by recovering several additional I-IR-
containing clones from a genomic library of A. gambiae
G3 DNA in ␭EMBL3 (Romans 2001). Approximately
75,000 plaques were screened with a probe specific for
the central 154 bp of Ikirara1-IRs. DNA was prepared
from 51 plaque-purified positive clones and examined on
Southern blots. Initially the four clones ␭Ikirara25, 28,
29, and 38 displaying I-IR-hybridizing SacII fragments
in the range of approximately 1.5–5 kb were chosen for
further study. SacII cuts once within each Ikirara1-IR
but not within its internal sequence. This SacII fragment
size range was based on the sizes of autonomously active
mariner/Tc1 and hAT superfamily elements. Further, al-
though there are SacII sites close to the termini of Iki-
rara1-IRs (Romans et al. 1998), we expected SacII to cut
rarely within the TE because its target site is exclusively
G and C. This is because A. gambiae TEs were found to
be more AT-rich than host genes (Besansky 1993) and
the sequence internal to I-IRs is only 22% G+C (Romans
et al. 1998). The four recombinant phages were mapped
further and larger restriction fragments containing the
I-IR-hybridizing SacII fragments were subcloned into
plasmids restriction mapped, and partial DNA sequences
obtained. Restriction maps of these clones are shown in
Fig. 1.
Double-stranded sequencing template DNA was purified from plas-
mids using Qiagen miniprep or midiprep columns. Most sequencing
was performed manually using a Sequenase version 2.0 kit (United
States Biochemical Corporation/Amersham) or T7 sequencing kit
(Pharmacia), ␣35S-dATP, 1000 Ci/mmol (Amersham), and T3, T7, SK,
or M13/pUC forward/reverse primers. Additional 17- to 20-mer prim-
ers were predicted from the sequence and synthesized commercially
(Canadian Life Technologies/Gibco BRL or Hospital for Sick Chil-
dren). Sequence was resolved from standard 6% polyacrylamide urea
gels on Kodak BioMax X-ray film. The 3.4 kb HindIII subclone from
␭I15 was sequenced automatically by Bio S&T Inc. (Lachine, Quebec).
Initial sequence analysis was performed using the PC/Gene suite of
programs (IntelliGenetics). Additional analysis was performed using
BLAST (Altschul et al. 1990, 1997). Sequences were aligned using
CLUSTAL W (Thompson et al. 1994) and the alignments were opti-
mized manually using previous phylogenetic analyses (Xiong and
Eickbush 1988, 1990; Capy et al. 1996, 1997) as guides.
All DNA sequences completed on both strands were deposited in
GenBank under the following accession numbers: ␭Ikirara15,
AF170018; ␭Ikirara25, AF170019 and AF170020; ␭Ikirara28,
AF170021; ␭Ikirara29, AF170022; and ␭Ikirara38, AF170023. The
Ikirara1 accession number is U55049.
Transposable Element Phylogenetics
The following TE sequences were used in our analyses: Drosophila
melanogaster 412, X04132; D. melanogaster 1731, X07656; D. mela-
nogaster Bari-1, X67681; Zea mays Cin4, Y00086; D. melanogaster
copia, X02599; Crithidia fasciculata CRE 1, M33009; Bombyx mori
Dong, L08889; Mus musculus F, X57795; D. melanogaster G,
X06950; D. melanogaster gypsy, X03734; D. melanogaster HB1,
X10748; D. melanogaster I, Fawcett et al. (1986); Fusarium oxyspo-
rum impala, AF282722; Trypanosoma brucei Ingi, M33485-84,
M21448; D. funebris jockey, M38437; Caenorhabditis elegans ZK370,
M98552; Homo sapiens L1Hs, M15269; M. musculus L1Md, M29324;
Dugestia tigrina Mar, S35068; D. hydei Minos, X61695; Hyalophora
cecropia MLE, M63844; D. mauritiana Mos1, M14653; Anopheles
albimanus Q, U03849; A. gambiae Quetzal, L78801, L78802; B. mori
R1Bm, M19755; D. melanogaster R1Dm, X51968; B. mori R2Bm,
M16558; D. melanogaster R2Dm, X51967; Rous sarcoma virus RSV,
J00844; T. brucei SLACS, X17078; salmonid fish Sleeping Beauty
(SB10), (constructed), DS30090 (FTP.EBI.AC.UK in directory pub/
databases/embl/align); A. gambiae T1, M93689; Autographa califor-
nica Ta1, M32662; C. elegans Tc1, X01005; C. elegans Tc3, M77697;
C. briggsae TCb1, S01245; Euplotes crassus Tec1, L03359; Eptatretus
stouti Tes1, M93038; Nicotiana tabacum Tnt1, X13777; Salmo salar
TSS (consensus), L12206, L12207, L12208; Xenopus laevis Tx1,
M13092; Saccharomyces cerevisiae Ty1, X03840; S. cerevisiae Ty3,
M23367; and D. heteroneura Uhu, X17356.
The 0.7 kb BamHI–XhoI fragment found between the
degenerate and the perfect IRs in ␭Ikirara29 (I29) was
identified as containing TE-like sequence by BLAST
searches and manual alignments. The manual alignments
revealed a putative hAT-like DNA binding domain motif
of 10 amino acids (EGWGICKYCK) just 5Ј to a long
ORF also identified as TE-like by BLAST. The other 50
clones hybridizing the I-IR probe were rescreened for
hybridization to this fragment. We expected the probe to
hybridize weakly to all I-IR-positive clones since it con-
tains approximately 160 bp of the degenerate I-IR, and
strongly to any I-IR clone containing a sequence similar
to this I29 hAT-like DNA binding domain sequence. All
We used neighbor joining (NJ) (Saitou and Nei, 1987), unweighted
pair-group method using arithmetic average (UPGMA), and parsimony

Fig. 1. Restriction maps of relevant portions of six Ikirara inverted
repeat-containing clones in ␭EMBL3. Ikirara-IRs are shown as broad
arrows beneath the maps. The thinner, dotted arrows are degenerate
IIRs. Thick, gray-shaded arrows identify transposable element se-
quences and signify class or superfamily as follows: medium gray, class
I-LTR retrotransposable element; light gray, class I non-LTR retro-
transposable elements; dark gray, class II mariner/Tc1 transposable
element. Boxes containing arrows indicate regions of similarity to
known genes. The arrows indicate 5Ј-to-3Ј sequence orientation. Ver-
tically striped boxes indicate sequences similar to previously described
noncoding sequences. (B) Regions sequenced. Dark lines indicate that
both DNA strands were sequenced. GenBank accession numbers: Iki-
rara1, U55049; Ikirara15, AF170018; Ikirara25, AF170019 and
AF170020; Ikirara28, AF170021; Ikirara29, AF170022; Ikirara38,
AF170023. Stippled lines indicate single-strand sequences. Restriction
site abbreviations: B, BamHI; H, HindIII; K, KpnI; P, PstI; Pv, PvuII;
RI, EcoRI; RV, EcoRV; S, SalI; Sa, SacII; X, XhoI; Xba, XbaI. [S]
indicates SalI sites from the ␭ sequence.

219
Table 1. Relationships between Ikirara inverted repeats (I-IR) as represented by pairwise percentage
nucleotide identity (top) and number of nucleotide substitutions/deletionsa/insertionsa to Ikirara1 216-bp
IRs (bottom)b
I1L
I1R
I38
I25
I28
I15
I29
I29D
I1L
I1R
I38
I25
I28
I15
I29
I29D
100
93.5
93.5
96.3
96.3
90.7
93.5
93.5
89.4
90.2
86.0
86.0
80.6
82.4
80.6
91.7
91.7
86.1
87.9
88.4
80.6
68.5
68.5
63.0
64.8
64.4
57.4
64.4
0/0/0
12/2/0
2/1/0
12/2/0
2/1/0
9/1/0
10/1/1
10/2/0
45/5/4
12/3/0
19/1/0
20/3/1
22/2/0
55/7/4
9/1/0
11/2/0
12/2/1
12/3/0
47/6/4
10/1/1
10/2/0
45/5/4
18/2/1
15/1/0
50/6/4
14/3/1
49/6/5
50/7/4
aEach deletion or insertion of adjacent base(s) has been regarded as a single mutational event.
bL and R indicate the left and right, 5Ј and 3Ј I-IRs, respectively.
50 ␭Ikirara clones hybridized the probe, while a non-
Ikirara-containing ␭ clone did not. ␭Ikirara15 hybrid-
ized very strongly, as did I29, while the other previously
selected clones all gave moderately strong signals.
Therefore I15 was added to our analysis (Fig. 1).
The Crusoe Region of Ikirara15
Upstream of the I-IR in I15 we found 1.1 kb, approxi-
mately 65%, of a novel class II mariner/Tc1 superfamily
TE which we have named Crusoe (Figs. 1 and 3A). A
BLAST search revealed that this element is related to the
270 bp of Tiang I sequence reported by Grossman et al.
(1999) (GenBank accession No. U89804). Tiang I was a
product of PCR amplification using degenerate primers
derived from Tc1. Over the 166 bp of common sequence,
Crusoe shares 91% nucleotide identity, 71.6% amino
acid identity, and 76.5% similarity to Tiang I. Since the
nucleotide identity is high while the predicted amino acid
similarity is moderate over what should be a highly con-
served region if these sequences were from two copies of
the same element, Crusoe is a Tiang family element, but
not Tiang I.
Crusoe is in opposite orientation to the I-IR. It is 1.1
kb of the 3Ј end of a Tc1-like element. The 3Ј IR appears
to be at least 240 bp long, as defined by three sequence
motifs: two 9 bp direct repeats 5Ј-CATCGCATT-3Ј
separated by 204 bp, a 5Ј-CAGTT-3Ј IR-terminal se-
quence conforming to the consensus C,A,G,T>G,G/T for
Tc1/hAT IR termini (Besansky et al. 1996), and the mari-
ner/Tc1 insertion site dinucleotide TA immediately ad-
jacent to the IR. The exact size of Crusoe IRs can be
determined only when a 5Ј end is isolated.
The Crusoe transposase gene includes the D,D35D/E
catalytic domain (Doak et al. 1994; Capy et al. 1996) and
the glycine-rich domain of a mariner/Tc1 superfamily
element, but it also contains six frameshift mutations
though no missense or translational termination muta-
tions (Fig. 3B). Analysis shows that Crusoe is more
closely related to Tc1 than to mariner. CLUSTAL W and
manual alignments of the known Crusoe1 transposase
sequence with Tc1 revealed 38% identity and 53% simi-
larity at the amino acid level (Rosenzweig et al. 1983)
(Table 2). Crusoe’s next closest relatives, Quetzal, a Tc1-
like element in Anopheles albimanus (Ke et al. 1996),
and Sleeping Beauty (SB10), a constructed Tc1-like el-
ement in salmonid fish (Ivics et al. 1997), are 36% iden-
Alignments of I-IRs sequenced from the ␭ clones re-
vealed a high degree of identity to those of Ikirara1 and
to each other (Table I and Fig. 2). They differed from
Ikirara1 by a small number of single-base changes and
single- or multiple-base deletions/insertions. Sequences
at the 5Ј (terminal) ends of the I-IRs are more highly
conserved than their 3Ј (internal) ends. I25 and I28 are
100% identical to I1 in the 5Ј 50 base pairs, while I15,
I29, and the degenerate I29-IR show a single base-pair
change, and I38 has three base-pair changes. At the I-IR
3Ј ends, only I25 shows 100% identity to I1 over 50 base
pairs and I38 is 98% identical. The 3Ј end of the I15-IR
is 84% conserved with I1. Of the final 7 base pairs,
however, only 3 base pairs are conserved. The degener-
ate I29-IR is truncated well before this sequence (not
shown). When sequences internal to the I-IRs were in-
cluded in the analysis, both Ikirara1-IRs and those of
I15, I25, and I38 could be aligned (Fig. 2). All of the
complete I-IRs contain a TA dinucleotide immediately 5Ј
to their termini. This is the TA observed at all insertion
sites of the mariner/Tc1 superfamily TEs. Sequences in-
ternal to the I28- and I38-IRs did not align well with
those of any other Ikirara. The reason for nonalignment
was that the sequences immediately adjacent to both of
these I-IRs are derived from other known repetitious se-
quences, MSQREPEL2 (Dotson et al. 1998) (GenBank
accession No. AF043433) in I28 and the TE we call
JuanAg in I38.
New Transposon-Like Elements
New TEs were identified in our sequences (see Materials
and Methods for their GenBank accession numbers) and
compared with the elements listed in the Materials and
Methods.

220
Fig. 2. Alignments of Ikirara inverted repeats and their internal sequences showing conserved internal sequences. Ikirara inverted repeat
sequences are in boldface. Asterisks indicate identical bases. Medium gray highlighting over the I38 internal sequence indicates the JuanAg
sequence. Light gray highlighting over the I28 internal sequence indicates the MSQREPEL2-G3 sequence.
tical, 52% similar, and 34% identical, 50% similar at the
amino acid level, respectively (Table 2). There is a 29%
amino acid identity and 41% similarity among all four
elements.
1035 bp long, and the sequenced 200-bp 3Ј untranslated
region.
The Vash Regions of ␭Ikirara15 and 29
A Southern blot of A. gambiae G3 genomic DNA was
hybridized with the 1.1 kb XhoI fragment containing
Crusoe. The probe hybridized to the expected 1.1 kb
XhoI fragment. However, many other fragments were
also observed. This result is expected, both if the genome
contains many Crusoe copies, and if the transposase se-
quences in the probe cross-hybridize with other related
elements. Southern blots probed with a PCR fragment of
the Crusoe IR suggested that there may be fewer than 25
copies of this TE in the genome (not shown). We believe
that a full-length Crusoe element should be approxi-
mately 1.7 kb in length, being composed of two IRs of
about 240 bp each, a transposase gene approximately
A comparison between the sequence directly adjacent
and 3Ј to the putative hAT-like DNA binding domain
motif (EGWGICKYCK) in I29 and the 0.7 kb hAT-like
DNA binding domain probe hybridizing sequence of the
region in I15 revealed 90% nucleotide and 87% amino
acid identity over 412 bp. In I15 this 412 bp is part of a
285 amino acid ORF extending 444 bp (148 amino acids)
5Ј to the region of sequence identity (Fig. 1). The se-
quence 3Ј to the ORF is also conserved in the two clones,
with 80.3% nucleotide identity over 56 bp.
BLAST analysis of the regions of I15 and I29 identity
revealed similarity between their ORFs and the A. gam-

221
Fig. 3. The Anopheles gambiae class II transposable element Crusoe
in ␭Ikirara15. A Restriction map of the I15 region containing the
Ikirara inverted repeat. The arrows and boxes below the restriction
map indicate the sequence content. Light gray indicates the sequence
from the Crusoe element. The triangles represent 9-bp direct repeats
within its 3Ј inverted repeat. B Comparison of Crusoe with related
elements from the mariner/Tc1 superfamily. Sequences in boldface are
characteristic of all mariner/Tc1 elements: paired-like domain with
leucine zipper (not present in the 5’-truncated Crusoe element); bipar-
tite nuclear localization signal (NLS); glycine-rich domain; DD35D/E,
the catalytic domain of mariner/Tc1 transposases. *, : and . indicate iden-
tity, conservative, and semi-conservative substitutions, respectively.
biae non-LTR retrotransposons Q and T1 (Besansky
1991; Besansky et al. 1994). Additional analyses of I15
were performed with an expanded 1180-bp region con-
taining the 0.96 kb XbaI–XhoI ␭ clone fragment (Fig. 1).
The 1180 bp region was aligned with the relevant 3Ј
portions of Q and T1 reverse transcriptase genes using
CLUSTAL W (Thompson et al. 1994) and additional
manual alignment (not shown). The three elements are
34.2% identical at the nucleotide level and 30.5% iden-
tical and 52.8% similar at the amino acid level (Table 2).
Thus, the non-LTR element with 5Ј truncated copies in
I15 and I29 is neither Q nor T1, although it is most
probably a member of this subfamily. We have named
this new element Vash.
Xiong and Eickbush (1988, 1990) and Garvey (1997)
identified sequence motifs in the 3Ј regions of class I
element reverse transcriptase genes. A number of these
motifs are general motifs found in LTR retrotransposons,
non-LTR retrotransposons, and retroviruses. Other mo-
tifs are identified as family specific. The Vash elements
in I15 and I29 display the non-LTR retrotransposon fam-
ily sequence motifs (alignments available from PR and
SRH on request). Southern blot hybridizations with
Vash1-region probes from I15 (fragments KpnI–XhoI

222
Table 2. New transposable elements identified in Anopheles gambiae Ikirara inverted repeat-containing clones and their relationships to other
known elements
Transposon
class
No. amino acids
scored
Most similar
elements
% amino acid
identity
% amino acid
similarity
Element
Clone
Crusoe
␭I15
II
215
Tc1 (C. elegans)
38
36
34
20
53
52
50
39
Quetzal (A. albimanus)
Sleeping Beauty (salmonids)
mariner (D. mauritiana)
Ozmandias
Vash1
␭I15
␭I15
I LTR
618
400
412 (D. melanogaster)
18
33
I non-LTR
Q (A. gambiae)
T1 (A. gambiae)
51
43
71
62
Vash2
␭I29
␭I29
␭I38
I non-LTR
I non-LTR
I non-LTR
140
999
329
Q (A. gambiae)
T1 (A. gambiae)
31
28
56
46
Guildenstern
JuanAg
RT1 (A. gambiae)
RT2 (A. gambiae)
44
44
59
58
JuanA (Ae. aegypti)
JuanC (C. pipiens)
31
25
48
43
and BamHI–XhoI; Fig. 1) revealed that there are approxi-
mately 20 Vash-like elements in the A. gambiae genome
(data not shown). The 3Ј UTRs of Q and T1 end with
eight tandem repeats of the trinucleotide TAA and four
tandem repeats of the pentanucleotide GAAAT, respec-
tively. These 3Ј tandem repeats are found 95 and 206
nucleotides downstream from their respective reverse
transcriptase gene termination codons. The Vash 3Ј
UTRs contain TAAAC pentanucleotide tandem repeats
65 bp downstream from the putative ORF 2 termination
codon. In Vash1 (I15) there are four perfect repeats
and two imperfect repeats, TAA(TAAAC)1TAAC
(TAAAC)3, while in Vash2 (I29) there are eight perfect
and two imperfect repeats, TAA(TAAAC)4TTAAC
(TAAAC)4.
the Juan family of non-LTR mosquito retrotransposons
(Mouches et al. 1991, 1992; Agarwal et al. 1993). Align-
ments with the other two Juan family elements reveal
that this element, JuanAg1, is incomplete. JuanAg1 en-
tirely lacks a polymerase ORF, and approximately 1.5 kb
at the 5Ј end of the reverse transcriptase ORF is also
deleted. Approximately 750 bp is also missing from the
central portion of the reverse transcriptase gene where
the I38 perfect I-IR is found. One possible explanation of
this is that an imprecise Ikirara excision event left the
JuanAg deletion and one I-IR behind. The conceptual
translation of the remaining 987 bp at the 3Ј end of the
reverse transcriptase gene reveals 21% amino acid iden-
tity and 43.5% similarity to JuanA and JuanC together
and 31 and 25% identity to these elements individually
(Table 2). JuanA and JuanC are 67% identical to each
other at the amino acid level, and therefore, JuanAg is
not closely related to these culicine elements. We suggest
that an ancestral Juan element was present in primordial
mosquitoes and that the observed differences in Juan
sequences reflect mosquito phylogeny.
The Guildenstern Region of Ikirara29
To the right of the degenerate I-IR in I29 (Fig. 1) we
found another 5Ј-truncated class I TE that shows 44%
amino acid identity and 58% similarity to retrotrans-
posases from the A. gambiae ribosomal gene sequence-
specific elements RT1 and RT2 (Besansky et al. 1992)
(Table 2). We have obtained 2997 bp of the expected 7–8
kb from this new non-LTR retrotransposable element,
Guildenstern. Since Guildenstern is immediately 3Ј to
the I29 SalI junction with the ␭ sequence, this element
may not be 5Ј-truncated in the genome. There is no se-
quence identity 3Ј to this element that would suggest its
incorporation into a ribosomal gene. In fact, Guilden-
stern1 terminates prematurely, 31 bp away from the de-
generate I-IR, the Vash2 sequence, and a PIF1 DNA
helicase-like gene (Foury and Lahaye 1987; see below).
The Ozymandias Region of Ikirara15
We analyzed the sequence of the entire region between
the I-IR and Vash in I15 (Fig. 1). BLAST analysis
showed that approximately 3 kb downstream of the I-IR,
in the opposite orientation, is the 3Ј end of an LTR ret-
rotransposon reverse transcriptase gene. An 1850-bp re-
gion showed 40% nucleotide identity to the D. melano-
gaster 412 element (Yuki et al. 1986) (GenBank
accession no. X04132), with 18% amino acid identity
and 33% similarity (Table 2). The regions identified as
conserved among LTR retrotransposons by Xiong and
Eickbush (1988) and Garvey (1997) are also highly con-
served in this 412-like sequence, with 81% identity and
90.4% similarity at the amino acid level. Therefore we
The JuanAg Region of Ikirara38
Surrounding the I38 perfect I-IR is 987 bp of sequence
we have identified as part of an A. gambiae member of

223
believe that this region of I15 contains part of a novel A.
gambiae LTR retrotransposable element which we call
Ozymandias. Southern blot analysis using the I15 1.11
kb HindIII–XhoI fragment as a probe revealed that there
are approximately 20 Ozymandias copies in the genome.
Since only the 3Ј end of the reverse transcriptase gene is
present in Ozymandias1, we are unable to identify the
LTR, even if it is present in the clone.
Evolutionary Relationships Among Transposons
Neighbor-joining (NJ), unweighted pair group method
using arithmetic average (UPGMA), and parsimony
analyses were used along with traditional bootstrapping
methods to construct phylogenetic trees relating the
newly identified elements to previously identified TEs.
For each analysis we used 100 repetitions to construct
consensus trees. The trees generated using all three meth-
ods maintained similar overall organizations. Class I and
class II elements clustered separately and the family clas-
sifications were maintained with few exceptions. How-
ever, the relationships both within and between the fami-
lies differed greatly among the trees described by the
different methods. Since our data set is composed of TEs
that are thought to evolve at different rates, and that are
also known to contain few conserved regions, we believe
that the NJ method is the most appropriate analysis to use
to describe their phylogenetic relationships (see Discus-
sion). Therefore we have presented only the results of
our NJ analyses.
Fig. 4. Unrooted phylogenetic tree of the reverse transcriptase se-
quences of non-LTR class I retrotransposable elements based on dis-
tance matrix analysis using the neighbor-joining method [PHYLIP 3.5c
(Felsenstein 1988)]. The numbers at each node represent the bootstrap
values for 100 repetitions. The elements’ host organisms and GenBank
accession numbers are given under Materials and Methods.
Non-LTR Retrotransposable Elements
NJ analysis of the three new non-LTR elements Vash,
Guildenstern, and JuanA, was based on the amino acid
sequence alignments (available from PR and SRH). The
Drosophila LTR element 412 was included in the data
set as an outgroup. The phylogenetic tree describing the
possible evolutionary relationships among these ele-
ments is shown in Fig. 4. The jockey and Juan-like ele-
ments form families before clustering together. The RT
elements, RT1, RT2, and Guildenstern, group with the R1
elements, R1Bm and R1Dm, before clustering together
with the jockey and Juan-like groups. This large group of
non-LTR retrotransposable elements clusters with an-
other large group containing the R2-like, L1-like, Cin4,
Tx1, Dong, Crx1, and Slacs groups of TEs. The T1-like
elements and I appear basal to these two large groups.
Vash, Q, and T1 clearly form a subfamily of non-LTR
elements not closely related to the other groups dis-
cerned. Basal bootstrap values were generally low.
scribed by Capy et al. (1996, 1997). (Alignments avail-
able from and P.R. and S.R.H.)
NJ analysis based on this alignment predicts the phy-
logenetic tree in Fig. 5 using RSV as an outgroup. The
Tc1-like elements Tc1, TCb1, Quetzal, SB10, TSS, Tc3,
HB1, and Crusoe cluster together, while the mariner-like
elements, mariner1, MLE, Mos1, and ZK370, form an-
other cluster. Tec1, impala, Minos, and Bari-1 are basal
to the mariner-like elements. These elements also appear
to be more closely related to the mariner-like elements
than to the Tc1-like elements. All of the class II elements
cluster together before clustering with the LTR elements
with the exception of Uhu. Uhu clusters with Ty3 and is
basal to both gypsy and the other class II elements.
The LTR elements are described as basal to the mari-
ner/Tc1 elements. In this tree the gypsy-like elements
appear to be most closely related to the class II elements,
followed by the 412-like elements and then the copia-
like elements.
mariner/Tc1-Like and LTR Retrotransposable Elements
Amino acid sequences from the transposase genes of
mariner/Tc1-like and LTR elements were aligned in the
Block B1 and Block B2 D35D/E-containing regions de-

224
MSQREPEL2 has been found repeated three times be-
tween A. gambiae Suakoko (SUA) strain putative pupal-
specific cuticular protein genes cp2a and cp2b (Dotson et
al. 1998) (GenBank accession no. AF043433). Our 120
bp sequence shows a high identity to only the 3Ј 41 bp
of the MSQREPEL2 repeat unit. The 100 bp tandem
repeat in I15 also shows a high identity to this 41 bp
of MSQREPEL2. We have identified five new
MSQREPEL2-like sequences in association with other
A. gambiae SUA strain genes using BLAST searches.
They are found between ANCHYM2 and ANCHYM1,
between ANTRYP3 and ANTRYP2, upstream of white,
5Ј to the class II element clone Pegasus-27, and in the
third intron of the guanylyl cyclase ␤-subunit gene
(GCS␤). Based on our alignments (Fig. 7), we suggest
that MSQREPEL2 sequences should be extended to in-
clude at least 50 bp 5Ј and exactly 25 bp 3Ј to the 58 bp
repeat. We therefore designate our 100 bp tandem repeat
in I15 and its counterpart in I28 as MSQREPEL2-G3, a
G3 strain repetitious sequence related to the SUA repeat
MSQREPEL2.
We have identified a new microsatellite sequence
640 bp 3Ј to Vash1 in I15 (Fig. 8A). This microsatellite
is d(CT)5 + d(CT)25. In addition, approximately 1,720 bp
3Ј to Vash1 is a d(GT) microsatellite-containing se-
quence previously identified in A. gambiae SUA strain
DNA by Zheng et al. (1996) (GenBank accession no.
Z72039). Over the 183 bp sequenced (Zheng et al. 1996;
Zheng, unpublished sequence) there is 65% nucleotide
identity with the I15 sequence (Fig. 8B). The microsat-
ellite in I15, however, lacks 8 bp of sequence found in
the middle of the Suakoko microsatellite, CCTGTGTG,
resulting in a d(TG)9 microsatellite rather than a d(GT)5
+ d(GT)6 microsatellite. The G3 strain sequence in I15
also shows several small insertions, deletions, transi-
tions, and transversions.
Fig. 5. Unrooted phylogenetic tree of mariner/Tc1 class II and LTR
class I retrotransposable elements based on distance matrix analysis
using the neighbor-joining method [PHYLIP 3.5c (Felsenstein 1988)].
The numbers at each node represent bootstrap values for 100 repeti-
tions. Rous sarcoma virus (RSV) was used as the outgroup. The ele-
ments’ host organisms and GenBank accession numbers are given un-
der Materials and Methods.
There is no microsatellite 3Ј to Vash2 in I29 similar to
that found 3Ј to Vash1 in I15. However, 10 bp 3Ј to
Vash2, and 152 bp 5Ј to the perfect 3Ј I-IR, there is a 693
bp region 32.2% identical to the Sacchromyces cerevi-
siae PIF1 DNA helicase gene (Foury and Lahaye 1987)
and 23% identical and 43.7% similar at the amino acid
level to the C. elegans homologue of PIF1 (GenBank
accession no. AF040649). In yeast, PIF1 DNA helicase
is an inhibitor of de novo telomere formation and of
telomere elongation (Schulz and Zakian 1994; Lahaye et
al. 1993).
The HindIII–PstI region at the left end of the I25 map
in Fig. 1 is 66.5% identical at the nucleotide level and
67% identical and 80% similar over a 93 amino acid
region to the C. elegans dynamin-like gene (GenBank
accession no. U61944). This DNA is 82% identical to
sequence from a D. melanogaster P1 clone, DS02782,
(GenBank accession no. AC002444), from chromosome
region 2L, 23A, which has not been characterized fur-
Nontransposable Element Sequences in Ikirara15, -25,
-28, and -29
We found no evidence of TE-like sequences in I25 other
than the I-IR, though we did find other repetitious DNA
sequences. To the right of the I-IR, i.e., 3Ј to Ikirara (see
Fig. 1), is a region 74% identical to a short A. gambiae
sequence previously found distal to the 826 bp middle
repeats in the subtelomeric region of chromosome 2L
(Biessmann et al. 1998) (Fig. 6). To the left of the I25-IR
is a short, 48 bp region closely resembling the context
sequence 5Ј to the Pegasus2 Tc1-related element (Be-
sansky et al. 1996). This sequence may be repeated else-
where in the genome.
I28 contains an interesting 120 bp sequence just in-
ternal to the I-IR (Fig. 2), which shows a high nucleotide
identity to sequences associated with several known A.
gambiae genes. A similar 58 bp sequence named

225
Fig. 6. Comparison of the
sequence 3Ј to the I-IR in
␭Ikirara25 with the novel middle
repetitive sequence (dark gray box)
found distal to the subtelomeric
satellite DNA of chromosome 2L in
the Anopheles gambiae pink eye
(PE) strain (Biessmann et al. 1998).
A Restriction maps of relevant
regions of ␭Ikirara25 and ␾PE-R2.
The restriction map of ␾PE-R2 has
been inverted relative to Fig. 6 of
Biessmann et al. (1998) to facilitate
sequence comparison. Alternating
black and white boxes represent 7
of the 14 826 bp satellite repeats
found in ␾PE-R2. The medium gray
box indicates a partial satellite
repeat (1–689 bp). The light gray
box indicates the sequence distal to
the subtelomeric repeats. Thick
black lines with 121 and 120 bp
labels indicate the sequences
aligned in B. The arrow represents
the Ikirara IR. B Alignment of
Ikirara25 and the subtelomeric
novel middle repetitive sequence
beginning at 1329 bp distal to the
partial satellite repeat from ␾PE-R2.
Nucleotide identity is 74% over the
126 bp consensus sequence.
ther. We suggest that this A. gambiae gene is not dyna-
min itself, because its alignments with C. elegans and D.
melanogaster dynamin genes gave only 24% amino acid
identity and 44% similarity and 25% identity and 44%
similarity, respectively. Dynamin is a GTPase involved
in the formation of endocytotic vesicles from clathrin-
coated pits (Bottomley et al. 1999; Benmerah et al.
1999). The function of the dynamin-like gene is un-
known. Thus, the Ikirara-IRs in both I25 and I29 appear
to be inserted in chromosomal regions of normal, host
protein-encoding genes.
G/C-richness of host genes (Besansky, 1993). These
studies were initiated prior to the Ikirara excision tests
reported by Leung and Romans (1998), which strongly
suggest that Ikirara1 is mariner/Tc1-like in behavior
and, by extension, that a complete Ikirara should be
between 1.3 and 3 kb in size.
The differences between the inverted repeats of Iki-
rara1 and the other clones (Table 1) might be attributed
to Ikirara1 having been a more recent insertion than the
I-IRs in ␭Ikirara15, -25, -28, -29, and -38. Alternatively,
these other elements were inserted into regions more
subject to the accumulation of mutations, possibly be-
cause these regions already contained noncoding, repeti-
tious, and/or TE sequences.
Discussion
Among these elements, sequences internal within the
I-IRs are more heterogeneous than the IR terminal se-
quences. It is possible these I-IR sequences are less cru-
cial for transposition than are the termini, as has been
found for the Mutator (Mu) family of elements in maize
(Walbot 1991). The I-IR termini may be recognized by
the Ikirara transposase, a related transposase, and/or host
factors. The requirement for specific sequences in IR
termini of other TEs has been well documented. Current
transposition models for TEs including P, Tc1, Tc3, and
Minos propose that the first step of excision involves the
synapsis of IRs enabling formation of the nucleoprotein
Ikirara Clones
The original screen for a complete, or at least more com-
plete, Ikirara element was designed based on two as-
sumptions: first, the complete element would be between
1.5 and 5 kb based on the sizes of known hAT-like and
mariner/Tc1-like elements; second, the restriction en-
zyme SacII (CCGC/GG) would not cut elsewhere in a
complete element since Anopheles gambiae TE se-
quences are A/T-rich in comparison with the relative

226
Fig. 7. Anopheles gambiae G3 strain I15 and I28 MSQREPEL2-G3
repetitious sequences compared with A. gambiae Suakoko strain
MSQREPEL2 repeats. MSQREPEL2 sequence locations are as fol-
lows: Pup1 and Pup2, both between cp2a and cp2b (Dotson et al. 1998)
(GenBank accession No. AF043433); Anchym, between ANCHYM2
and ANCHYM1 (Mu¨ller et al., 1993) (GenBank accession No.
Z32645); GCS␤, soluble guanylyl cyclase ␤-subunit (Caccone et al.
1999); Peg-27, upstream of the Pegasus-27 inverted repeat element
(Besansky et al. 1996); white, upstream of the white gene (Besansky et
al. 1995); ALU1, AluI repeats (Hill and Crampton 1994); Antryp, be-
tween ANTRYP3 and ANTRYP2 of the trypsin cluster (Mu¨ller et al.
1993). I15 repeats are in boldface. Medium gray shading indicates the
original MSQREPEL2 sequence; light gray shading indicates the se-
quence we believe should also be included in MSQREPEL2.
complex involved in transposition. In the case of P, DN-
ase I footprinting has shown that a non-P-encoded in-
verted repeat binding protein (IRBP) binds specifically
to the first 16 bp of the 31 bp IR (Rio and Rubin 1988;
Beall et al. 1994). It is believed that one of the potential
roles for IRBP is to align the termini precisely for proper
cleavage. Interestingly, P-element IRs can be classified
as either 5Ј or 3Ј because the transposase binding sites,
though they are not actually part of the IRs, are asym-
metrical, 40 and 52 bp from the sites of cleavage (Kauf-
man et al. 1989). However, the I-IRs are much longer
and there is no evidence that they might be functionally
distinct from each other. The Tc1 transposase is the only
protein required for transposition (Vos et al. 1996).
DNase I footprinting has shown that Tc1 transposase
protects base pairs 6 through 29 of the 54 bp IRs (Vos et
al. 1993). Furthermore, both in vitro and in vivo experi-
ments have demonstrated that the Tc1 transposase is re-
sponsible for the DNA cleavage required for excision
(Vos et al. 1993, 1996). Application of this model to
Ikirara leads to the prediction that 5Ј sequences in the
I-IRs must be maintained to permit synapsis of the se-
quences in which DNA cleavage occurs and, potentially,
to maintain a binding site for a host factor. In this con-
text, we emphasize that SacII sites begin at base pair 26
in all I-IRs and form part of a conserved predicted stem–

227
Fig. 8. Microsatellite sequences downstream of Vash1 in Ikirara15. A Novel d(CT) microsatellite sequence 640 bp downstream of Vash1. B
Alignment of d(TG) microsatellite-containing sequence 1720 bp downstream of Vash1 with A. gambiae Suakoko strain microsatellite AGH101
(Zheng et al. 1996; Zheng, unpublished sequence).
loop secondary structure expected to be highly stable,
⌬G 25°C of −8.4 kcal (Romans et al. 1998).
phylogeny for TEs that does not presuppose a molecular
clock. TEs contain few identified conserved regions,
suggesting that TE sequences are highly variable. The
two distance matrix-based methods for phylogeny con-
struction, NJ and UPGMA, are relatively unaffected by
highly variable characters because the variation is buff-
ered by the averaging of character values.
New Elements vs Ikirara
Ikirara IRs cannot be parts of the same TE as any of the
Vash, Guildenstern, JuanAg, or Ozymandias sequences.
This is because all known features of Ikirara, including
the stereotyped excision pattern (Leung and Romans
1998), suggest that it is a mariner/Tc1 superfamily class
II element, not a class I retrotransposon. Further, I-IR
hybridization to Southern blots of genomic DNA also
used for hybridization with parts of these new elements
showed that the I-IRs do not colocalize to the same DNA
fragments as these elements. Crusoe is not part of Ikirara
due to its location upstream of the I-IR and to the pres-
ence of its own probable 3Ј inverted repeat, a sequence
that is clearly different from the 216 bp I-IR.
All 10.25 kb of I15 DNA 5Ј to Crusoe has been se-
quenced. This DNA contains no trace of a relatively
degenerate Crusoe 5Ј end, or of a second I-IR, inverted
relative to the first. Further, the Ikirara element can be
interpreted as containing relics of Ozymandias and
Vash1 insertions. Therefore it is possible that more Iki-
rara sequence and the rest of the Crusoe element could
be present on the adjacent, presently uncloned genomic
fragment.
Non-LTR Retrotransposable Elements
Our phylogeny describes the closest relationships be-
tween the three new non-LTR retrotransposable elements
and previously described TEs as moderately to strongly
probable, bootstrap values >60 and >80, respectively.
Vash is a member of the T1 family of retroposons (Be-
sansky 1991; Besansky et al. 1994). JuanAg is an ele-
ment in the Juan family (Mouches et al. 1991, 1992;
Agarwal et al. 1993), and Guildenstern is a member of
the RT subfamily of elements (Besansky et al. 1992).
Therefore Vash, JuanAg, and Guildenstern are related to
previously described subfamilies but represent new ad-
ditions to them.
Comparisons between Fig. 4 and previously published
phylogenies (Xiong and Eickbush 1988, 1990) reveal
consistent patterns. Families and subfamilies are shown
to cluster together and the general relationships between
families are maintained with one exception, the I element
family (I). As described by Xiong and Eickbush (1988),
I is most closely related to Ingi and positioned with Ingi
between the R1–G cluster and the R2–L1 cluster. Figure
4 describes the I element as having diverged basally to
both of these groups. It is possible that the addition of our
new elements to the phylogeny has revealed the previ-
Evolutionary Relationships of TEs Found in Ikirara
IR-Containing Clones
TEs probably do not evolve in direct proportion to time.
We therefore chose the NJ method to construct a possible

228
ously suggested relationship between I and Ingi to be
spurious. Unfortunately, to date the reliability of all phy-
logenies attempting to relate I, Ingi, and indeed all non-
LTR retrotransposable elements suffers from low basal
bootstrap values.
Anopheles gambiae Genome Organization
Different genomes contain varying proportions of highly
repetitious and moderately repetitious DNA. These se-
quences, both TE and non-TE, have been suggested to be
important in genome evolution. They have been impli-
cated in local, regional, and global regulation of gene
expression and in maintenance of chromosome structure
(nucleoskeleton). TEs specifically can cause genome re-
organization, especially when their movement is trig-
gered by environmental stresses (White and Doebley
1998). The first Ikirara element was found in the 3-kb
intergene region between two genes in the vitellogenin
(Vg) gene tandem array, 2 kb upstream of a Vg gene
transcriptional start site (Romans et al. 1998; Romans,
unpublished). It is interesting that this cluster provides an
example of TE insertions being associated with gene
duplication. There are three tandem Vg genes and one
additional nontandem gene close by in genomes lacking
the Ikirara1 insertion and four tandem genes plus the
nontandem gene in genomes containing Ikirara1. Possi-
bly the best-known example of this phenomenon is the
spontaneous Bar eye mutation in Drosophila, a small,
but cytologically visible chromosomal duplication which
may have arisen by recombination between two roo el-
ements (for further description see Lindsley and Zimm
1992).
Recently, as large regions of eukaryotic genomes
have been sequenced, in some cases whole genomes,
intergene regions have come under increasing scrutiny.
The 240-kb region surrounding Adh1 in maize is com-
posed of 150 kb of LTR retrotransposons, as well as solo
LTRs, MITES (miniature inverted-repeat transposable
elements), and other repetitious sequences (White and
Doebley 1998). For every 10 kb of genomic sequence,
there are two to seven TEs. White and Doebley also
found that approximately one-third of 376 nuclear gene
sequences previously reported contained at least one TE
and that many TEs were within 1 kb of a transcriptional
start site. Based on this analysis they described genic
(euchromatic) regions of the maize genome as expressed
genes in a sea of repetitious sequences.
mariner/Tc1 Superfamily and LTR Retrotransposons
The basal bootstrap values are as low for mariner/Tc1
and LTR retrotransposons (Fig. 5) as they are for non-
LTR retrotransposable elements (Fig. 4). The mariner-
like elements and Tc1-like elements form separate clus-
ters similar to those described by Capy et al. (1996). Our
association of Tec1, impala, Minos, and Bari-1 with the
mariner-like elements and the position of Uhu diverging
basally to class II TEs differs from those results. Capy et
al. (1996) argued that these elements are more Tc1-like
because of their positions in the tree resulting from NJ
analysis. From their NJ results they concluded that all
elements containing the motif YSDLNPIE are Tc1-like.
This relationship is not borne out in Fig. 5. One differ-
ence between analyses is that we did not include the
Capy et al. (1996) Block A containing the first glutamate
residue of D,D35D/E in our alignments because the
length of sequence between Block A and Block B is not
fixed and therefore the definition of Block A may be
imprecise. Furthermore, Block A is short, only 10 amino
acids, with 1 or 2 informative amino acids in addition to
D defining each major grouping.
We combined the LTR retrotransposable elements
and the class II element phylogenies from Capy et al.
(1996) and included four new elements in the align-
ments, Crusoe, Ozymandias, Quetzal, and Sleeping
Beauty (S10), none of which is very closely related to
any element included in the original alignment. The LTR
retrotransposable element relationships in Fig. 5 do re-
flect those described in the summary tree described by
Capy et al. (1996). Our tree suggests that Ty and copia-
like elements may be more closely related to retroviruses
than to the other LTR retrotransposable elements, while
gypsy-like elements are more closely related to the class
II TEs.
Is the A. gambiae genome primarily long-period in-
terspersion, or should it be described as having some
form of intermediate periodicity? Of 28.5 kb, the total
amount of DNA we sequenced from the five new A.
gambiae ␭ clones containing I-IRs, 12 kb (40%) is com-
posed of TE sequences. This percentage is even higher in
the 11 kb of contiguous sequence obtained from I15,
50% of 11.7 kb. In addition to TEs, we also found nu-
merous other forms of repetitious sequence in our clones.
These include the MSQREPEL2-G3 sequences in I15
and I28, the middle repetitive sequence in I25 similar to
subtelomeric middle repeats, the 49-bp sequence in I25
similar to the sequence flanking Pegasus-2, and the two-
microsatellite sequences in I15. The discovery of
TE phylogenies, including ours, have been derived
from a mixture of autonomous, nonautonomous, and
consensus TE sequences. Perhaps as a consequence, TEs
are thought to have relatively few conserved regions, and
even these regions can be highly variable among ele-
ments in the same family and class. We suggest that if
these phylogenies were constructed solely from autono-
mously active or active consensus sequences (e.g., Sleep-
ing Beauty), the numbers of identifiable conserved re-
gions might increase. In any case, additional conserved
regions of sequence should be sought and added to these
phylogenetic analyses to strengthen the trees.

229
Besansky NJ (1991) A retrotransposable element from the mosquito
Anopheles gambiae. Mol Cell Biol 10:863–871
Besansky NJ (1993) Codon usage patterns in chromosomal and retro-
poson genes of the mosquito Anopheles gambiae. Insect Mol Biol
1:171–178
Besansky NJ, Powell JR (1992) Reassociation kinetics of Anopheles
gambiae (Diptera: Culicidae) DNA. J Med Entomol 29:125–128
Besansky NJ, Paskewitz SM, Hamm DM, Collins FH (1992) Distinct
families of site-specific retrotransposons occupy identical positions
in the rRNA genes of Anopheles gambiae. Mol Cell Biol 12:5102–
5110
Besansky NJ, Bedell JA, Mukabayire O (1994) Q: A new retrotrans-
poson from the mosquito Anopheles gambiae. Insect Mol Biol 3:
49–56
Besansky NJ, Bedell JA, Benedict MQ, Mukabayire O, Hilfiker D,
Collins FH (1995) Cloning and characterization of the white gene
from Anopheles gambiae. Insect Mol Biol 4:217–231
Besansky NJ, Mukabayire O, Bedell JA, Lusz H (1996) Pegasus, a
small terminal inverted repeat transposable element found in the
white gene of Anopheles gambiae. Genetica 98:119–129
Biessmann H, Kobeski F, Walter MF, Kasravi A, Roth CW (1998)
DNA organization and length polymorphism at the 2L telomeric
region of Anopheles gambiae. Insect Mol Biol 7:83–93
Bottomley MJ, Surdo PL, Driscoll PC (1999) How dynamin sets
vesicles PHree! Curr Biol 9:R301–304
Britten RJ, Davidson EH. (1976) DNA sequence arrangement and pre-
liminary evidence on its evolution. Fed Proc 35:2151–2157
Caccone A, Garcia BA, Mathiopoulos KD, Min GS, Moriyama EN,
Powell JR (1999) Characterization of the soluble guanylyl cyclase
␤-subunit gene mosquito Anopheles gambiae. Insect Mol Biol 8:
23–30
Capy P, Vitalis R, Langin T, Higuet D, Bazin C (1996) Relationships
between transposable elements based upon the integrase-
transposase domains: Is there a common ancestor? J Mol Evol
42:359–368
Capy P, Langin T, Higuet D, Maurer P, Bazin C (1997) Do the inte-
grases of LTR-retrotransposons and class II element transposases
have a common ancestor? Genetica 100:63–72
MSQREPEL2 sequences in close proximity to A. gam-
biae SUA strain genes, within 2 kb of the transcriptional
start sites for white, cp2b, ANCHYM1, and ANCHYM2,
and within the third intron of GCS␤, is also suggestive of
a shorter-period interspersion pattern in what would pre-
viously have been called long-period regions.
Based on the number of TE and non-TE repetitious
sequences we have discovered in relatively short
stretches of A. gambiae genomic DNA, and our analysis
of longer genomic regions primarily in gene clusters in-
cluding the trypsins (Mu¨ller et al. 1993), the putative
pupal-specific cuticular protein genes (Dotson et al.
1998), the vitellogenin genes [GenBank accession No.
AF281078 (Romans, unpublished)], 40 kb of the Diphe-
nol oxidase-A2 gene subcluster within the larger Dopa
decarboxylase (Ddc) gene cluster, and over 60 kb sur-
rounding Ddc itself [GenBank accession Nos. AF042732
and AF063021 (Romans, unpublished)], we suggest that
the maize-like genome organization occurs in the A.
gambiae genome as “islands” in the sea of a more Dro-
sophila-like, generally long-period interspersion genome
containing few, if any, repetitious sequences.
Acknowledgments. S.R.H. and P.R. dedicate this paper to Al Downe,
who fostered our interest in mosquitoes. We are grateful to Mel Green,
Colin Malcolm, Rita Avancini, and Clare Garvey for discussion, to
Biao Zheng for sharing his unpublished AGH101 microsatellite se-
quence, and to Agustina Ponvia for technical assistance. This investi-
gation received financial support from the UNDP/World Bank/WHO
Special Programme for Research and Training in Tropical Diseases
(TDR). Additional support was provided by grants to P.R. from the
Natural Sciences and Engineering Research Council of Canada and
from the J.P. Bickell Foundation for a DNA workstation.
Catteruccia F, Nolan T, Loukeris TG, Blass C, Savakis C, Kafatos FC,
Crisanti A (2000) Stable germline transformation of the malaria
mosquito Anopheles stephensi. Nature 405:959–962
References
Coates CJ, Jasinskiene N, Miyashiro L, James AA (1998a) mariner
transposition and transformation of the yellow fever mosquito, Ae-
des aegypti. Proc Natl Acad Sci USA 95:3745–3751
Coates CJ, Jasinskiene N, Pott GB, James AA (1998b) Promoter-
directed expression of recombinant fire-fly luciferase in salivary
glands of Hermes transformed Aedes aegypti. Gene 226:317–325
Doak TG, Doeder FP, John CL, Herrick G (1994) A proposed super-
family of transposase genes: Transposon-like elements in ciliated
protozoa and a common “D35E” motif. Proc Natl Acad Sci USA
91:942–946
Dotson EM, Cornel AJ, Willis JH, Collins FH (1998) A family of
pupal-specific cuticular protein genes in the mosquito Anopheles
gambiae. Insect Biochem Mol Biol 28:459–472
Fawcett DH, Lister CK, Kellett E, Finnegan DJ (1986) Transposable
elements controlling I-R hybrid dysgenesis in D. melanogaster are
similar to mammalian LINEs. Cell 47:1007–1015
Adams MD, Tarng RS, Rio DC (1997) The alternative splicing factor
PSI regulates P-element third intron splicing in vivo. Genes Dev
11:129–138
Agarwal M, Bensaadi N, Salvado JC, Campbell K, Mouches C (1993)
Characterization and genetic organization of full-length copies of a
LINE retroposon family dispersed in the genome of Culex pipiens
mosquitoes. Insect Biochem Mol Biol 23:621–629
Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic
local alignment search tool. J Mol Biol 215:403–410
Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W,
Lipman DJ (1997) Gapped BLAST and PSI-BLAST: A new gen-
eration of protein database search programs. Nucleic Acids Res
25:3389–3402
Arca B, Zabalou S, Loukeris TG, Savakis C (1997) Mobilization of a
Minos transposon in Drosophila melanogaster chromosomes and
chromatid repair by heteroduplex formation. Genetics 145:267–279
Felsenstein J (1988) Phylogenies from molecular sequences: inference
and reliability. Annu Rev Genet 22:521–565
Finnegan DJ (1992) Transposable elements. Curr Opin Genet Dev
2:861–867
Foury F, Lahaye A (1987) Cloning and sequencing of the PIF gene
involved in repair and recombination of yeast mitochondrial DNA.
EMBO J 6:1441–1449
Beall EL, Admon A, Rio DC (1994) A Drosophila protein homologous
to the human p70 Ku autoimmune antigen interacts with the P
transposable element inverted repeats. Proc Natl Acad Sci USA
91:12681–12685
Benmerah A, Bayrou M, Cerf-Bensussan N, Dautry-Varsat A (1999)
Inhibition of clathrin-coated pit assembly by an Eps15 mutant. J
Cell Sci 112:1303–1311
Garvey CF (1997) A study of transposable elements in the malaria
vector Anopheles stephensi including an analysis of the Dox-A2
region. Ph.D. thesis QMWC, University of London, London

230
Grossman GL, Cornel AJ, Rafferty CS, Robertson HM, Collins FH
(1999) Tsessebe, Topi and Tiang: Three distinct Tc1-like transpos-
able elements in the malaria vector, Anopheles gambiae. Genetica
105:69–80
Grossman GL, Rafferty CS, Fraser MJ, Benedict MQ (2000) The pig-
gyBac element is capable of precise excision and transposition in
cells and embryos of the mosquito, Anopheles gambiae. Insect Bio-
chem Mol Biol 30:909–914
LINE retroposon dispersed in the genome of three non-sibling Ae-
des species. Gene 120:183–190
Mu¨ller H-M, Crampton JM, della Torre A, Sinden R, Crisanti A (1993)
Members of a trypsin gene family in Anopheles gambiae are in-
duced in the gut by blood meal. EMBO J 12:2891–2900
O’Brochta DA, Atkinson PW (1996) Transposable elements and gene
transformation in non-drosophilid insects. Insect Biochem Mol Biol
26:739–753
Hammond LE, Rudner DZ, Kanaar R, Rio DC (1997) Mutations in the
hrp48 gene, which encodes a Drosophila heterogeneous nuclear
ribonucleoprotein particle protein, cause lethality and developmen-
tal defects and affect P-element third-intron splicing in vivo. Mol
Cell Biol 17:7260–7267
Handler AM, Gomez SP (1996). The hobo transposable element ex-
cises and has related elements in tephritid species. Genetics 143:
1339–1347
Handler AM, Gomez SP, O’Brochta DA (1993) A functional analysis
of the P-element gene-transfer vector in insects. Arch Insect Bio-
chem Physiol 22:373–384
Hill SM, Crampton JM (1994) Synthetic DNA probes to identify mem-
bers of the Anopheles gambiae complex and to distinguish the two
major vectors of malaria within the complex., An. gambiae s.s. and
An. arabiensis. Am J Trop Med Hyg 50:312–321
O’Brochta DA, Handler AM (1988) Mobility of P elements in dro-
sophilids and nondrosophilids. Proc Natl Acad Sci USA 85:6052–
6056
O’Brochta DA, Gomez SP, Handler AM (1991) P element excision in
Drosophila melanogaster and related drosophilids. Mol Gen Genet
225:387–394
O’Brochta DA, Warren WD, Saville DJ, Atkinson PW (1994) Inter-
plasmid transposition of Drosophila hobo elements in non-
drosophilid insects. Mol Gen Genet 244:9–14
O’Brochta DA, Warren WD, Saville DJ, Atkinson PW (1996) Hermes,
a functional non-Drosophilid insect gene vector from Musca do-
mestica. Genetics 142:907–914
Pinkerton AC, O’Brochta DA, Atkinson PW (1996) Mobility of hAT
transposable elements in the Old World bollworm, Helicoverpa
armigera. Insect Mol Biol 5:223–227
Hoy MA (1994) Insect molecular genetics, Academic Press, San Diego
Ivics Z, Hackett PB, Plasterk RIH, Izsvak Z (1997) Molecular recon-
struction of Sleeping Beauty, a Tc1-like transposon from fish, and
its transposition in human cells. Cell 91:501–510
Jasinskiene N, Coates CJ, Benedict MQ, Cornel AJ, Rafferty CS, James
AA, Collins FH (1998) Stable transformation of the yellow fever
mosquito, Aedes aegypti with the Hermes element from the house-
fly. Proc Natl Acad Sci USA 95:3743–3747
Kaufman PD, Doll RF, Rio DC (1989) Drosophila P element trans-
posase recognizes internal P element DNA sequences. Cell 59:359–
371
Ke Z, Grossman GL, Cornel AJ, Collins FH (1996) Quetzal: A trans-
poson of the Tc1 family in Anopheles albimanus. Genetica 98:141–
147
Lahaye A, Leterme S, Foury F (1993) PIF1 DNA helicase from Sac-
charomyces cerevisiae. Biochemical characterization of the en-
zyme. J Biol Chem 268:26155–26161
Lampe DJ, Churchill ME, Robertson HM (1996) A purified mariner
transposase is sufficient to mediate transposition in vitro. EMBO J
15:5470–5479
Leung SS, Romans P (1998) Excisions of the Ikirara1 transposon in an
Anopheles gambiae cell line. Insect Mol Biol 7:241–248
Lindsley DL, Zimm GG (1992) The genome of Drosophila melano-
gaster, Academic Press, San Diego
Lobo N, Li X, Fraser MJ Jr. (1999) Transposition of the piggyBac
element in embryos of Drosophila melanogaster, Aedes aegypti and
Trichoplusia ni. Mol Gen Genet 261:803–810
Rao PN and Rai KS (1987) Inter and intraspecific variation in nuclear
DNA content in Aedes mosquitoes. Heredity 59:523–258
Rio DC, Rubin GM (1988) Identification and purification of a Dro-
sophila protein that binds to the terminal 31-base-pair inverted
repeats of the P transposable element. Proc Natl Acad Sci USA
85:8929–8933
Romans P (2001) The Diphenol oxidase-A2 gene and its surrounding
region are highly similar in the evolutionarily distant Diptera,
Anopheles gambiae and Drosophila melanogaster (submitted for
publication)
Romans P, Seeley DC, Kew Y, Gwadz RW (1991) Use of a restriction
fragment length polymorphism (RFLP) as a genetic marker in
crosses of Anopheles gambiae (Diptera: Culicidae): Independent
assortment of a diphenol oxidase RFLP and an esterase locus. J
Med Entomol 28:147–151
Romans P, Bhattacharyya RK, Colavita A (1998) Ikirara, a novel
transposon family from the malaria vector mosquito, Anopheles
gambiae. Insect Mol Biol 7:1–10
Rosenzweig B, Liao L, Hirsch D (1983) Sequence of the C. elegans
transposable element Tc1. Nucleic Acids Res 11:4201–4209
Rudkin GT (1969) Non replicating DNA in Drosophila. Genetics 61
(Suppl):227–238
Saitou N, Nei M (1987) The neighbor-joining method: a new method
for reconstructing phylogenetic trees. Mol Biol Evol 4:406–425
Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning: A labo-
ratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold
Spring Harbor, NY
Luo G, Ivics Z, Izsvak Z, Bradley A (1998) Chromosomal transposition
of a Tc1/mariner-like element in mouse embryonic stem cells. Proc
Natl Acad Sci USA 95:10769–10773
Mathiopoulos KD, della Torre A, Predazzi V, Petrarca V, Coluzzi M
(1998) Cloning of inversion breakpoints in the Anopheles gambiae
complex traces a transposable element at the inversion junction.
Proc Natl Acad Sci USA 95:12444–12449
Mathiopoulos KD, della Torre A, Santolamazza F, Predazzi V, Petrarca
V, Coluzzi M (1999) Are chromosomal inversions induced by
transposable elements? A paradigm from the malaria mosquito
Anopheles gambiae. Parassitologia 41:119–123
Miller LH, Sakai RK, Romans P, Gwadz RW, Kantoff P, Coon HG
(1987) Stable integration and expression of a bacterial gene in the
mosquito Anopheles gambiae. Science 237:779–781
Sarkar A, Yardley K, Atkinson PW, James AA, O’Brochta DA (1997a)
Transposition of the Hermes element in embryos of the vector
mosquito, Aedes aegypti. Insect Biochem Mol Bio 27:359–363
Sarkar A, Coates CJ, Whyard S, Willhoeft U, Atkinson PW, O’Brochta
DA (1997b) The Hermes element from Musca domestica can trans-
pose in four families of cyclorrhaphan flies. Genetica 99:15–29
Schulz VP, Zakian VA (1994) The Saccharomyces PIF1 DNA helicase
inhibits telomere elongation and de novo telomere formation. Cell
76:145–155
Swofford DL (1993) PAUP: Phylogenetic analysis using parsimony,
version 3.1.1 Illinois Natural History Survey, Urbana
Thibault ST, Luu T, Vann N, Miller TA (1999) Precise excision and
transposition of piggyBac in pink bollworm embryos. Insect Mol
Biol 8:119–123
Mouches C, Agarwal M, Campbell K, Lemieux L, Abadon M (1991)
Sequence of a truncated LINE-like retroposon dispersed in the ge-
nome of Culex mosquitoes. Gene 106:279–280
Thompson JD, Higgins DG, Gibson TJ (1994) CLUSTAL W: Improv-
ing the sensitivity of progressive multiple sequence alignment
through sequence weighting, position-specific gap penalties and
weight matrix choice. Nucleic Acids Res 22:4673–4680
Mouches C, Bensaadi N, Salvado JC (1992) Characterization of a

231
Vanderberg JP, Gwadz RW (1980) The transmission by mosquitoes of
Plasmodium in the laboratory. In: Krier JP (ed) Malaria: pathology,
vector studies, and culture, Vol 2. Academic Press, New York, pp
153–234
Vos JC, van Luenen HGAAM, Plasterk RH (1993) Characterization of
the Caenorhabditis elegans Tc1 transposase in vivo and in vitro.
Genes Dev 7:1244–1253
Vos JC, De Baere I, Plasterk RH (1996) Transposase is the only nem-
atode protein required for in vitro transposition of Tc1. Genes Dev
10:755–761
Walbot V (1991) The Mutator transposable element family of maize.
Genet Eng (NY) 13:1–37
White S, Doebley J (1998) Of genes and genomes and the origin of
maize. Trends Genet 14:327–4332
Xiong Y, Eickbush TH (1988) Similarity of reverse transcriptase-like
sequences of viruses, transposable elements, and mitochondrial in-
trons. Mol Biol Evol 5:675–690
Xiong Y, Eickbush TH (1990) Origin and evolution of retroelements
based upon their reverse transcriptase sequences. EMBO J 9:3353–
3362
Yuki S, Inouye S, Ishimura S, Siago K (1986) Nucleotide sequence
characterization of a Drosophila retrotransposon, 412. Eur J Bio-
chem 158:403–410
Zheng L, Benedict MQ, Comel AJ, Collins FH, Kafatos FC (1996) An
integrated genetic map of the African human malaria vector mos-
quito, Anopheles gambiae. Genetics 143:941–952
Weinberg ES (1998) Harnessing horizontal gene transfer. Curr Biol
8:R244–R247