Naturally Occurring Transposable Elements Disrupt hsp70 Promoter
Function in Drosophila melanogaster
Daniel N. Lerman* and Martin E. Feder* *Committee on Evolutionary Biology and Department of Organismal Biology & Anatomy, The University of Chicago;
Committees on Genetics and Molecular Medicine and Department of Organismal Biology & Anatomy, The University of Chicago
Naturally occurring transposable element (TE) insertions that disrupt Drosophila promoters are correlated with modified
promoter function and are posited to play a significant role in regulatory evolution, but their phenotypes have not been
established directly. To establish the functional consequences of these TE insertions, we created constructs with either
TE-bearing or TE-lacking hsp70 promoters fused to a luciferase reporter gene and assayed luciferase luminescence
in transiently transfected Drosophila cells. Each of the four TEs reduces luciferase signal after heat shock and heat
inducibility of the hsp70 promoter. To test if the differences in hsp70 promoter activity are TE-sequence dependent, we
replaced each of the TEs with multiple intergenic sequences of equal length. These replacement insertions similarly
reduced luciferase signal, suggesting that the TEs affect hsp70 promoter function by altering promoter architecture.
These results are consistent with differences in Hsp70 expression levels, inducible thermotolerance, and fecundity
previously associated with the TEs. That two different varieties of TEs in two different hsp70 genes have common
effects suggests that TE insertion represents a general mechanism through which selection manipulates hsp70 gene
expression.
Although mobile genetic or transposable elements
(TEs) have had a massive and ancient impact on genome
organization and diversity (Kazazian 2004), the roles of
TEs in ongoing adaptive evolution in eukaryotes are less
clear, because transposition is typically both deleterious
and effectively marginalized by evolutionary and genetic
mechanisms (Franchini, Ganko, and McDonald 2004).
Recent discoveries, however, distinctively correlate transposition with adaptive microevolution (Franchini, Ganko,
and McDonald 2004). The components of this correlation
are as follows: (1) transposition into regulatory regions of
genes; (2) such genes segregating as alleles of TE-less
genes in natural populations; (3) allelic variation in the regulation of gene expression; (4) allelic variation in gene
expression affecting fitness, and thus exposing allelic frequency to natural selection; and (5) covariation of allelic
frequency with natural variation in the agent of selection.
Although each component has numerous examples, the
experimental verification of all components in a single system has been elusive. For instance, the best recent examples
(Kalendar et al. 2000; Daborn et al. 2002; Lerman et al.
2003; Schlenke and Begun 2004) are complete except
for demonstrating that the TEs themselves (rather than
sequence to which they are linked) encode the adaptive phenotype and how they do so. Here we supply this missing
link by experimentally excising or replacing naturally
occurring TEs in a common genetic background and showing that the expected changes in gene expression result.
The present study uses the Hsp70-encoding genes of
Drosophila melanogaster as a model system. In this species, Hsp70 is a molecular chaperone that contributes significantly to inducible thermotolerance, which is key to
1
Present address: Office of Science Policy & Planning, Office of the
Director, National Institutes of Health, 1 Center Drive, Bldg. 1, Rm. 218,
Bethesda, MD 20892.
Key words: transposable element, hsp70, promoter, transcriptional
regulation, gene expression.
E-mail: [email protected].
Mol. Biol. Evol. 22(3):776–783. 2005
doi:10.1093/molbev/msi063
Advance Access publication December 1, 2004
reproductive success in the face of natural thermal stress
(Feder et al. 1996; Feder, Blair, and Figueras 1997; Lerman
et al. 2003). Hsp70, however, plays numerous additional
roles in the cell, for which stringent regulation at low concentrations is essential (Lerman et al. 2003). Thus, repeatably and depending on circumstances, either high or low
levels of Hsp70 expression evolve in both natural and
experimental populations (Sorensen, Dahlgaard, and
Loeschcke 2001; Feder et al. 2002). These levels are inversely correlated with the frequencies of hsp70 alleles in
which distinct TEs have inserted in the proximal promoter,
suggesting selection for or against TE-bearing alleles as a
mechanism of action (Michalak et al. 2001; Zatsepina et al.
2001; Bettencourt et al. 2002; Lerman et al. 2003). Indeed,
alleles with a Jockey insertion are associated with reduced
hsp70 mRNA, and both Jockey and P-bearing alleles are
associated with reduced Hsp70 protein (Lerman et al.
2003). The natural or evolutionary variation in Hsp70 level
could be due to the transposable elements themselves, and/
or to sequence variation that is linked to the TEs but external to them. If the former, the TEs could affect Hsp70 level
via two non-exclusive mechanisms: First, diverse components of the transcriptional apparatus associate with the
hsp70 promoter, and the proper organization and spacing
of their binding sites are essential for full-strength transcription (O’Brien and Lis 1991; Lis and Wu 1993; Amin et al.
1994; Li et al. 1996; Mason and Lis 1997; Weber et al.
1997). Hence disruption of promoter architecture by TE
insertion could reduce transcriptional activation (Xiao
and Lis 1988; Lee et al. 1992; Amin et al. 1994; Shopland
et al. 1995). Alternatively, variation in hsp70 transcription
could stem from introduction of novel regulatory elements
in the TEs themselves (Kazazian 2004). To characterize the
effect of the TEs and elucidate the mechanism by which
they affect Hsp70 levels, we placed hsp70 promoter alleles
with or without TE insertions in luciferase expression constructs, replaced TE insertions in these constructs with
intergenic DNA of equal length, and excised these sequences from constructs. Each TE reduced luciferase
signal, indicating that the TEs can be sufficient to reduce gene
transcription. Furthermore, each replacement insert similarly
reduced luciferase signal in all four hsp70 constructs,
Molecular Biology and Evolution vol. 22 no. 3 Ó Society for Molecular Biology and Evolution 2004; all rights reserved.
Transposable Elements and Promoter Function 777
suggesting that the corresponding TEs affect hsp70 promoter
function primarily by altering promoter architecture.
upstream of HSE4 and end 60 bp downstream of the transcription start site (fig. 1).
Materials and Methods
Fly Populations and hsp70Bb Promoter Sequencing
Creation of Replacement Inserts
We obtained promoter sequence from three Drosophila populations harboring hsp70Ba alleles with or without TEs in the proximal promoter (T32, collected in
subequatorial Chad and maintained at 30°–31°C (henceforth referred to as the T strain; Zatsepina et al. 2001);
Arv/Zim, founded from crosses of parents from California
and Zimbabwe (Lerman et al. 2003); and 1997 Evolution
Canyon, collected from the North-facing slope (NFS) of
Evolution Canyon, Israel, during August–September,
1997 [Michalak et al. 2001]) and a fourth population with
similar allelism in hsp70Bb (2000 Evolution Canyon, collected from both the North-facing and South-facing [SFS]
slopes of Evolution Canyon during the months of August
and September 2000). In the 2000 Evolution Canyon population, hsp70Bb-specific primers (table 1A in the Supplementary Material online) were used to amplify a fragment
(wild-type, 3,644 bp) containing the hsp70Bb promoter,
coding sequence, and 3# UTR, and sequenced as previously
described (Lerman et al. 2003).
Creation of Wild-type and TE-Bearing Reporter
Constructs
The hsp70Ba promoter region from TE-bearing
(TE1) and TE-lacking (TEÿ) flies from the T, Arv/Zim,
and 1997 Evolution Canyon (‘‘Canyon 97’’) populations
was amplified from single-fly DNA in a 100-ll reaction,
as above (see table 1B in the Supplementary Material
online). This amplification yields a ;900 bp (for wild-type;
2,099–2,354 bp for TE-disrupted promoters) product containing the promoter, 5# UTR, and first 190 bp of the
hsp70Ba coding sequence.
To create luciferase reporter constructs, hsp70 promoters were re-amplified from the polymerase chain reaction (PCR) products described above (Canyon 2000 whole
hsp70Bb gene and hsp70Ba promoter and partial coding
sequence for the other three populations) with primers
designed to introduce restriction sites to facilitate cloning
(table 1C in the Supplementary Material online). Reamplification of gene-specific PCR products ensured that
the wild-type and TE-disrupted promoter constructs were
derived from the same gene (rather than from different
hsp70 gene copies). A 2–3 lg portion of these new PCR
products was digested for 3 h at 37°C with 15 U each of
Kpn1 and Bgl2 in a 40-ll reaction. Then, 4 lg of the
pGL-3-Basic Vector (Promega; ;4.8 kb), which contains
the cDNA-encoding firefly luciferase (luc) but lacks
eukaryotic promoter or enhancer sequences, was also
digested with the same restriction enzymes. Digests were
purified with the QIAquick Gel-Extraction kit (Qiagen),
promoters were ligated into the pGL-3 vector, and maxi
preps were purified with the Qiagen Plasmid Maxi Kit,
yielding 300-ll purified plasmid at a concentration of
1–2 lg/ll. TE1 and TEÿ constructs were confirmed by
gel electrophoresis and sequencing. All constructs begin
To remove the TEs and facilitate their replacement
with random DNA sequences, two hsp70 promoter fragments were amplified from the TE-containing constructs
described above. The portion of the hsp70 promoter
upstream from the TE was amplified using the aforementioned Kpn1-containing upper primer and a lower primer
containing a Pac1 restriction site. This lower primer starts
just upstream of the TE, and was therefore specific for each
TE-containing promoter (table 1D in the Supplementary
Material online). The part of the hsp70 promoter downstream of the TE was amplified using the aforementioned
Bgl2-containing lower primer and gene-specific Pac1containing upper primers, which began just downstream
of the TE insertions (table 1D in the Supplementary
Material online).
A 2-lg portion of each of the amplified promoter fragments was digested for 3 h at 37°C with Pac1, and the
upstream and downstream promoter pieces (150 lg each)
were ligated to each other at the Pac1 site with 3 U T4
Ligase. Successful upstream-downstream ligants were
amplified with the aforementioned Kpn1- and Bgl2containing primers and confirmed by gel electrophoresis.
To create ‘‘Excision’’ reporter constructs, these ligants
were digested with Kpn1 and Bgl2, ligated into pGL-3
(cut with Kpn1 and Bgl2), and purified as above. Preparations were confirmed by DNA sequencing. The excision
constructs contained an 8-bp Pac1 site at the excision site,
which was not present in wild-type promoters (the insertion
of this site did not significantly reduce luciferase signal
in the Canyon 2000, Arv/Zim, and T populations (see
Results). The excision constructs differ from the TEcontaining constructs in that the former contain an 8-bp
Pac1 site where the TE was, and lack the 8 bp sequence
duplicated at the TE insertion site.
Four random DNA sequences used to replace each of
the naturally occurring TEs were obtained by amplification
(table 1E in the Supplementary Material online) of intergenic sequences from D. melanogaster chromosome 3R
(GenBank accession number AE003602). Designated by
letter, these are: A (positions 64618–66588 of the GenBank
sequence), B (248199–249655), C (253707–255437), and
D (262490–263956). These positions are the limits of the
respective intergenic sequences; each 3#-complimentary
primer (table 1E in the Supplementary Material online)
was designed so that the resulting amplicon would be
the same length as the TE it was replacing.
Intergenic PCR products and the excision luciferase
constructs described above were digested with Pac1 and
cleaned with Qiagen spin columns. Three of the four random DNA sequences eventually were successfully ligated
into the site from which each naturally occurring TE had
previously been excised, and ligants (henceforth ‘‘Replacement’’ constructs) were purified and confirmed by gel electrophoresis. Because the Replacement constructs were the
same length as the TEs they replaced, Replacement sequences from the same intergenic region were not identical
778 Lerman and Feder
A hsp70BaT
TCTCTCACTCTGTC
Jockey
TCTCTCACTCTGTC
+1
-306
HSE4
HSE3
HSE2
HSE1
TATA
luciferase
-107
Kpn1
Bgl2
B hsp70BaArv/Zim
GACATACT
P Element
GACATACT
+1
HSE4
-306
HSE3
HSE2
HSE1
TATA
luciferase
-97
Kpn1
Bgl2
C hsp70BaCanyon 97
P Element
CTCTGGCC
CTCTGGCC
+1
-306
HSE4
HSE3
HSE2
HSE1
luciferase
TATA
-184
Kpn1
Bgl2
D hsp70BbCanyon 2000
GCCGGAGT
GCCGGAGT
P Element
+1
-306
HSE4
HSE3
HSE2
Kpn1
HSE1
luciferase
TATA
-41
Bgl2
FIG. 1.—hsp70-luciferase reporter constructs. TE1 and TEÿ promoters from each fly population were amplified, digested with Kpn1 and Bgl2, and
ligated into the pGL3-basic vector as described in the text. TE insertion sites in the A. hsp70BaT; B. hsp70BaArv/Zim; C. hsp70BaCanyon 9; and D.
hsp70BbCanyon 2000 promoters. HSEs, TATA box, GAGA elements (striped boxes), and transcription start site are as shown. Duplicated host DNA
at TE insertion sites flank TEs as shown; wild-type (TEÿ) promoters contain a single copy of this sequence. Insertion site duplication in hsp70BaT
(A) results in an additional putative GAGA element (underlined in sequence), defined as the pentamer consensus sequence GAGAG (Omichinski
et al. 1997; Wilkins and Lis 1997). The last T in the duplicated hsp70BbCanyon 2000 sequence (D) is the first T of the TATA box. Constructs are
not drawn to scale.
(e.g., insert A for Canyon 97 was not identical to replacement sequence A for Arv/Zim, and so on).
Transient Transfection and Luciferase Assay
Drosophila Schneider S2 cells were cultured at 25°C
in Sheilds and Sang M3 medium (Sigma) containing 10%
heat-inactivated fetal bovine serum (Sigma); 3.5 6 105 cells
were seeded in 350 ll medium per well in 24-well plates the
day before transfection. Cells were transiently transfected
with Superfect Transfection Reagent (Qiagen). Then 1 lg
plasmid DNA was co-transfected with 0.05 lg pRL-CMV
(Promega) per well. The pRL-CMV vector contains the
cDNA encoding Renilla luciferase (Rluc) and the CMV
enhancer and early promoter.
After transfection, cells were incubated at 25°C for
6 h, transferred to 1.5-ml microcentrifuge tubes and placed
in thermostatted waterbaths at either 36.5°C (heat shock) or
25°C (control) for 5 min, transferred to a 25°C waterbath for
1 h, and harvested by centrifugation. Eight samples per promoter (i.e., TE1, TEÿ, Excision, Replacements) were subjected to heat shock, and four samples/promoter were
subjected to control treatment for a total of 72 separate
transfections for each of the populations (except in the T
strains, for which only two replacement inserts were created). Cells were washed in phosphate-buffered saline
(PBS), lysed in 150 ll Passive Lysis Buffer (Promega),
and stored at ÿ80°C overnight for analysis.
Luciferase signal was quantitated with the DualLuciferase Reporter Assay System (Promega), which allows
sequential measurement of firefly and Renilla luciferase
activity from the same sample of lysate. For each assay,
5 ll lysate (1 5 ll PBS) was analyzed in a manual luminometer programmed to perform a 10-s measurement. Firefly
Transposable Elements and Promoter Function 779
luciferase background was determined by measuring luminescence activity of nontransfected control cells. Before
quantitation of Renilla luminescent signal, firefly signal
was quenched; nonetheless, residual luminescence from
the firefly luciferase reaction can affect Renilla luminescence measurements. Accordingly, Renilla background
luminescence was determined by preparing a lysate of
heat-shocked cells transfected with the appropriate hsp70luc reporter vector only (i.e., no pRL-CMV) and measuring
firefly luciferase luminescence followed by apparent Renilla
luminescence. Backgrounds were subtracted from raw firefly and Renilla luminescence measurements, respectively,
and luciferase activity was expressed as the ratio of firefly/Renilla luminescence.
Several sets of S2 cells were used for transformations.
These differed systematically in their normalized luminescence signal (data not shown). Accordingly, luminescence
signals were compared only in transfected cells from the
same source cell population, and all comparisons within
a fly population (e.g., Arv/Zim etc.) were performed in a
single experiment.
Statistics
Statistical analyses used StatView 4.5 (Abacus Concepts). Factorial analyses of variance (ANOVA) tested
for group effect. Post hoc analyses tested for effect of promoter after heat-shock and control treatments, as well as for
effect of heat shock in TE1, TEÿ, Excision, and Replacement samples.
RESULTS
Sequencing
The Canyon 2000 NFS population is polymorphic for
a nonautonomous P element insertion in hsp70Bb promoter
(fig. 1; GenBank accession number AY370940). While the
other TEs described in Lerman et al. (2003) inserted
between HSEs 2 and 3 in the hsp70Ba promoter, this element is located between the TATA box and the first HSE
(41 bps upstream from the transcription start site) and disrupts the hsp70Bb gene promoter. This P element (1,050
bp) is smaller than those found in the Arv/Zim (1,383
bp) and Canyon 2000 (1,221 bp) populations. Like these
P elements, the Canyon 2000 P element has a large internal
deletion and is missing part of exon 1, all of exon 2, and part
of exon 3; these P elements are therefore similar to the type
II repressor-making KP element (Ohare and Rubin 1983;
Black et al. 1987; Rasmusson, Raymond, and Simmons
1993). The Canyon 2000 P element is at an allelic frequency of approximately 20% in the 2000 NFS population
and 0% in the 2000 SFS population.
Constructs, Controls, and Repeatability of
Luciferase Assays
Renilla luciferase activity was positively correlated with
the amount of transfected pRL-CMV. Luminescence in cells
transfected with the pGL3-Basic vector only (i.e., with no
promoter) was equal to firefly luciferase background measurements from nontransfected control cells or blank tubes after
both heat-shock and control treatments (data not shown).
Control (no heat shock) Treatment
In the absence of heat-shock luciferase luminescence
was low and unaffected by TE insertions in promoters
extracted from each of the four populations (i.e., posthoc comparisons are not significant [statistics not shown];
fig 2). Under these conditions, luciferase luminescence was
also low and did not differ among constructs in which the
TEs had been excised or replaced with the intergenic DNA
amplicons A-D (fig. 2).
Heat Inducibility (comparison of control and heat
shock treatments)
To determine the impact of the TE insertions on heat
inducibility, we compared luciferase levels for a given construct before and after heat shock. Heat shock increased
luciferase luminescence above that of non-heat-shocked
cells for all TEÿ promoters, but it did not significantly
increase luminescence for TE1 promoters (fig 2; table 1).
That is, the TEs reduced heat inducibility of these promoters. Heat inducibility of luciferase luminescence was
also abolished when TEs were replaced with several intergenic DNA sequences of similar length (fig 2; table 1).
Excising the TEs rescued wild-type heat inducibility in
three of the four hsp70 promoter constructs (Canyon
2000, Canyon 97, and Arv/Zim).
Luciferase Activity in Heat-Shocked Cells
Whereas above we compare each promoter variant to
itself under heat-shock or control conditions, here we compare promoter variants to one another, all after heat shock:
1. TE1 vs. TEÿ (wild-type). All constructs with TE1 promoters had lower luciferase luminescence than those
with TEÿ promoters (fig 2; table 1). This difference
in luciferase activity was similar for each of the TEs,
ranging from 52% (for Canyon 97 promoters) to 72%
(for Canyon 2000 promoters).
2. TE1 vs. Replacement. Luciferase signal did not differ
between any Replacement construct and its corresponding TE1 construct for any of the four fly populations
(fig 2; statistics not shown). Thus replacement of the
naturally occurring TEs with intergenic DNA amplicons
reduces luciferase luminescence to a similar extent as do
the naturally occurring TEs themselves.
3. TEÿ vs. Excision. TEÿ hsp70 promoters and their corresponding Excision constructs resulted in similar luciferase activity in the Canyon 2000, Arv/Zim, and T
populations. In the Canyon 97 samples, however, luciferase luminescence was 32% lower in the Excision constructs than in TEÿ (fig 2; table 1).
4. Replacement vs. Excision. For the Canyon 2000, Canyon 97, and Arv/Zim hsp70 promoters, hsp70-luc constructs with each of the three replacement insertions
expressed less luciferase than did the corresponding
Excision constructs. These differences in luminescence
were similar to those between TE1 and TEÿ constructs,
ranging from 44% (in Canyon 97 Insert A) to 92% (in
Canyon 2000 Insert D). For the T strain, only the D
insert had significantly lower luciferase levels than its
780 Lerman and Feder
FIG. 2.—hsp70 promoter activity. Drosophila S2 cells were transfected with TE1 and TEÿhsp70-luc reporter constructs and subjected to control
(‘‘Con’’) or heat-shock (‘‘HS’’) treatments, as described in the text. Firefly luciferase signal is expressed as the ratio of firefly to Renilla luciferase luminescence. A. T; B. Arv/Zim; C. Canyon 97; and D. Canyon 2000 populations.
corresponding Excision construct, although both C and
D Replacements had significantly lower luciferase levels
than the corresponding wild-type (TEÿ) construct (fig 2;
table 1; Replacement versus TEÿ post-hocs not shown).
Discussion
While noncoding regulatory variation is thought to
underlie important phenotypic changes, distinguishing
the impact of specific nucleotide variants from those of
linked loci is often difficult (Rockman and Wray 2002).
Here we tested the functional consequences of naturally
occurring TE insertions that disrupt Drosophila hsp70 promoters. To mitigate the confounding influence of linkage
disequilibrium, we measured luciferase luminescence in
Drosphila S2 cells transfected with hsp70–luc reporter constructs. These experiments clearly demonstrate that the each
of the TEs reduces hsp70 promoter activity, and they suggest that the differences in Hsp70 expression, thermotolerance, and reproductive success previously associated with
these insertions (Lerman et al 2003) are caused by TE disruption of the hsp70 promoters. Furthermore, replacement
of the TEs with intergenic sequences establishes that disruption of hsp promoter architecture is sufficient for
reduced hsp transcription
Transposition events are frequent at both hsp70 gene
clusters (Zatsepina et al. 2001; Bettencourt et al. 2002;
Maside, Bartolome, and Charlesworth 2002), and, as the
present study shows, TEs have interrupted the proximal promoters of two hsp70 genes. Lerman et al. (2003) hypothesized that the distinctive open chromatin conformation of
hsp70 promoters facilitates their transpositional disruption.
While high levels of Hsp70 are important for tolerance
of extreme temperatures (see Introduction), high levels of
Hsp70 can be deleterious at moderately high but non-lethal
temperatures (Krebs and Feder 1997a; 1997b; Bettencourt,
Feder, and Cavicchi 1999; Zatsepina et al. 2001), presumably as a result of over-activation of its multiple functions in
the cell (Lerman et al. 2003). Thus Lerman et al. (2003)
hypothesized that selection maintains the TEs to reduce
Hsp70 levels in Drosophila in non-stressful environments.
The reduction in hsp70BaP allele frequency on the hotter
and more variable south-facing slope (SFS; 1.2%) compared with the NFS (33.9%) in the 1997 Evolution Canyon
Transposable Elements and Promoter Function 781
Table 1
Analysis of Variance of Normalized Luminescence Measurements
Source
df
MS
F-value
P value
Canyon 2000
Residual
Post-hoc comparisons after HS:
11
60
1.11 6 104
6.63 6 102
TE-absent vs. TE-present
TE-absent vs. Excision
Excision vs. Insert A
Excision vs. Insert C
Excision vs. Insert D
HS vs. Con TE-absent
HS vs. Con TE-present
HS vs. Con Excision
HS vs. Con, Insert A
HS vs. Con, Insert C
4.24 6 105
2.57 6 104
TE-absent vs. TE-present
TE-absent vs. Excision
Excision vs. Insert A
Excision vs. Insert C
Excision vs. Insert D
HS vs. Con TE-absent
HS vs. Con TE-present
HS vs. Con Excision
HS vs. Con, Insert A
HS vs. Con, Insert C
HS vs. Con, Insert D
3.18 6 105
4.39 6 104
TE-absent vs. TE-present
TE-absent vs. Excision
Excision vs. Insert A
Excision vs. Insert B
Excision vs. Insert D
HS vs. Con TE-absent
HS vs. Con TE-present
HS vs. Con Excision
HS vs. Con, Insert A
HS vs. Con, Insert B
HS vs. Con, Insert D
3.49 6 104
4.18 6 103
TE-absent vs. TE-present
TE-absent vs. Excision
Excision vs. Insert C
Excision vs. Insert D
HS vs. Con TE-absent
HS vs. Con TE-present
HS vs. Con Excision
HS vs. Con, Insert C
HS vs. Con, Insert D
16.709
,0.0001
16.491
,0.0001*
0.1204
,0.0001*
,0.0001*
,0.0001*
,0.0001*
0.1156
,0.0001*
0.8266
0.4364
,0.0001
7.236
,0.0001*
0.0002*
0.0005*
,0.0001*
,0.0001*
,0.0001*
0.0039
0.0003*
0.0634
0.4891
0.3255
,0.0001
8.350
0.0001*
0.6877
,0.0001*
0.0001*
,0.0001*
,0.0001*
0.1552
0.0004*
0.6204
0.3200
0.6182
,0.0001
Post-hoc heat inducibility
Comparisons:
Canyon 97
Residual
Post-hoc comparisons after HS:
11
59
Post-hoc heat inducibility
Comparisons:
Arv/Zim
Residual
Post-hoc comparisons after HS:
11
60
Post-hoc heat inducibility
Comparisons:
T strain
Residual
Post-hoc comparisons after HS:
Post-hoc heat inducibility
Comparisons
9
50
0.0001*
0.1231
0.0122
0.0001*
,0.0001*
0.2181
0.0022
0.1729
0.5935
ANOVAs performed on luciferase measurements (firefly luc/Rluc luminescence) from the four groups depicted in figure 2. Bonferroni/Dunn post-hoc analyses tested
pairwise comparisons as shown (*denotes significance).
populations is consistent with this hypothesis (Michalak
et al. 2001), as is the lower hsp70BbP allele frequency for
the SFS population (0%) than for the NFS (20%) population
in the Evolution Canyon 2000 collection. An alternative
hypothesis, that the absence of the hsp70BbP allele in the
Canyon 2000 SFS population results from drift/founder
effect, requires data on gene flow between these populations
for testing.
Full-strength transcription of heat-shock genes
requires the orderly interaction of the transcriptional holoenzyme/polymerase complex, the heat-shock transcription
factor (HSF), and conserved elements in the heat-shock
promoter. hsp70 genes normally have an open chromatin
conformation (Costlow and Lis 1984), and TFIID, RNA
polymerase II (Pol II), and the GAGA factor (GFA) all
associate with the hsp70 promoter prior to heat shock
(Lis and Wu 1993). While this constitutive binding and
pre-assembly primes hsp70 genes for rapid response to heat
shock, it is not sufficient for hsp70 induction because the
Pol II complex pauses 20–40 nucleotides downstream of
the transcription start site (Weber et al. 1997). Resumption
of elongation by Pol II, the rate-limiting step in hsp70 transcription (O’Brien and Lis 1991), requires cooperative
binding of HSF trimers to HSEs (Amin et al. 1994).
Heat-shock transcription factor and the TATA binding protein also interact cooperatively in vivo (Mason and Lis
1997), and GFA further stabilizes HSF-HSE engagement
(Li et al. 1996; Mason and Lis 1997). Hence, disruption
782 Lerman and Feder
of promoter architecture, by TE insertion or any other
means, might reduce transcriptional activation by altering
cooperative protein-protein interactions at the hsp70 promoter. Prior work supports this hypothesis. Although HSEs
1 and 2 are sufficient for some hsp70 transcription, fullstrength transcription requires all four (Xiao and Lis
1988; Lee et al. 1992; Shopland et al. 1995). The spacing
between and stereoalignment of HSEs is also critical. Insertion of spacers of 1–5 and 11–14 nucleotides between HSEs
1 and 2 reduces promoter activity by 90%, whereas spacers
of 6–10 and 16–18 nucleotides largely restore promoter
activity (Amin et al. 1994).
As we show, all of the TEs affect hsp70 promoter
function, despite differences in insertion site, length, and
sequence. This effect was not a foregone conclusion, however, as the experimental manipulations of hsp70 promoter
structure discussed above differ from the disruptions caused
by the naturally occurring TEs in several important ways.
For example, Amin et al. (1994) used constructs that contained only the minimal hsp70 promoter—i.e., HSEs 1 and
2. Thus, changing the spacing between HSEs 2 and 3,
where three of the TEs have inserted, was previously unexamined. In addition, while the maximum spacer length previously tested was 25 bp, the TEs described here are all
much larger (.1 kb). Finally, the TEs are natural—not
anthropogenic—mutagens.
In addition, that the TE insertions would decrease,
rather than increase, expression of adjacent genes was
unforeseen. Indeed, other TEs thought to play an adaptive
role in regulatory evolution may increase gene expression.
For example, Daborn et al. (2002) demonstrate that an
Accord insertion in the 5# end of the D. melanogaster cytochrome P450 gene Cyp6g1 is associated with Cyp6g1 overtranscription, which confers insecticide resistance.
Schlenke and Begun (2004) show that a 4,803-bp Doc
insertion in the 5# flanking region of the D. simulans
Cyp6g1 gene is also correlated with increased transcriptional abundance. Yet despite differences in insertion site,
length, and sequence, all the TEs described here similarly
reduced hsp70 promoter function. This reduction, and the
absence of TE insertions that increase Hsp70 expression,
may result from selection against high levels of Hsp70,
which can be harmful in environments with infrequent
extreme stress (Lerman et al. 2003).
Particular sequence elements within each of the TEs
could in principle also be responsible for the impact of
TEs on promoter activity. Some TEs carry regulatory motifs
that alter expression of nearby genes. For example, insertion of the gypsy element, which contains an insulator that
can disrupt regulatory interactions, is responsible for many
spontaneous mutations in Drosophila (Modolell, Bender,
and Meselson 1983; Geyer and Corces 1992; Hogga et al.
2001). Further, TE1 and TEÿ promoters contain differences in addition to the TE that may contribute to the differences observed in promoter activity. For example,
duplicated host DNA flanks each of the TE insertion sites.
In the T population, this duplication results in an additional
putative GAGA element, which may alter promoter activity, and the other insertion site duplications may also contain putative regulatory sequences. In addition, the Canyon
2000 and Arv/Zim populations contain single-nucleotide
polymorphisms (SNPs; two and four SNPs, respectively)
that are linked to the P element insertion. Although such
noncoding regulatory variation is thought to underlie
important phenotypic changes, distinguishing the impact
of specific nucleotide variants from those of linked loci
is often difficult (Rockman and Wray 2002). Indeed, differences in luciferase luminescence observed between TE1
and TEÿ promoters might result from these SNPs, rather
than from the TEs that they are linked to.
To distinguish between these alternative impacts of
TEs, spatial disruption of the promoter versus addition
of novel regulatory motifs—and to remove the potentially
confounding affect of the insertion site duplications and
SNPs—we created the Excision and Replacement luc constructs. All three intergenic replacement inserts in the Canyon 2000, Canyon 97, and Arv/Zim hsp70-luc promoters
and one of the two inserts in the T promoter significantly
reduced luciferase signal compared with both the wild-type
(TEÿ) and Excision constructs. These reductions were
comparable to those observed in the TE1 constructs.
The intergenic inserts also reduced heat inducibility of
the hsp70-luc promoters, as the TEs themselves did. Taken
together, 10 of 11 replacement inserts significantly reduced
luciferase luminescence, as did all four TEs, indicating that
the reductions in promoter activity caused by the TE insertions did not result primarily from TE-specific sequence, insertion site duplications, or from SNPs linked
to the TEs. Rather, these results suggest that disruption
of hsp promoter architecture is sufficient for reduced hsp
transcription.
We have characterized naturally occurring TE insertions that disrupt the Drosophila hsp70 promoter and show
that each of these insertions reduces hsp70 promoter activity. These TEs, moreover, reduce Hsp70 expression levels
and affect organismal fitness measurements (Lerman et al.
2003). This study is the first direct demonstration that specific natural nucleotide polymorphisms underlie differences
in Hsp70 expression. More broadly, these results suggest
that TEs can underlie adaptive regulatory evolution in natural populations.
Acknowledgments
We thank Harinder Singh and his laboratory for use of
his luminometer, and David Rand for providing the Arv/
Zim populations. This work was supported by National Science Foundation awards IBN99-86158, IBN02-06582, and
IBN03-16627, and by a Howard Hughes Medical Institute
Predoctoral Fellowship (D.N.L.).
Literature Cited
Amin, J., M. Fernandez, J. Ananthan, J. T. Lis, and R. Voellmy.
1994. Cooperative binding of heat shock transcription factor
to the hsp70 promoter in vivo and in vitro. J. Biol. Chem.
269:4804–4811.
Bettencourt, B. R., M. E. Feder, and S. Cavicchi. 1999. Experimental evolution of Hsp70 expression and thermotolerance
in Drosophila melanogaster. Evolution 53:484–492.
Bettencourt, B. R., I. Kim, A. A. Hoffmann, and M. E. Feder.
2002. Response to natural and laboratory selection at the Drosophila hsp70 genes. Evolution 56:1796–1801.
Transposable Elements and Promoter Function 783
Black, D. M., M. S. Jackson, M. G. Kidwell, and G. A. Dover.
1987. Kp Elements repress P-induced hybrid dysgenesis in
Drosophila melanogaster. EMBO J. 6:4125–4135.
Costlow, N., and J. T. Lis. 1984. High-resolution mapping of
DNase I-hypersensitive sites of Drosophila heat shock genes
in Drosophila melanogaster and Saccharomyces cerevisiae.
Mol. Cell Biol. 4:1853–1863.
Daborn, P. J., J. L. Yen, M. R. Bogwitz, G. Le Goff, E. Feil, S.
Jeffers, N. Tijet, T. Perry, D. Heckel, P. Batterham, R. et al.
2002. A single P450 allele associated with insecticide resistance in Drosophila. Science 297:2253–2256.
Feder, M. E., T. B. C. Bedford, D. R. Albright, and P. Michalak.
2002. Evolvability of Hsp70 expression under artificial selection
for inducible thermotolerance in independent populations of
Drosophila melanogaster. Physiol. Biochem. Zool. 75:325–334.
Feder, M. E., N. Blair, and H. Figueras. 1997. Natural thermal
stress and heat-shock protein expression in Drosophila larvae
and pupae. Funct. Ecol. 11:90–100.
Feder, M. E., N. V. Cartaño, L. Milos, R. A. Krebs, and S. L.
Lindquist. 1996. Effect of engineering hsp70 copy number
on Hsp70 expression and tolerance of ecologically relevant
heat shock in larvae and pupae of Drosophila melanogaster.
J. Exp. Biol. 199:1837–1844.
Franchini, L. F., E. W. Ganko, and J. F. McDonald. 2004. Retrotransposon-gene associations are widespread among D. melanogaster populations. Mol. Biol. Evol. 21:1323–1331.
Geyer, P. K., and V. G. Corces. 1992. DNA position-specific
repression of transcription by a Drosophila zinc finger protein.
Genes Dev. 6:1865–1873.
Hogga, I., J. Mihaly, S. Barges, and F. Karch. 2001. Replacement
of Fab-7 by the gypsy or scs insulator disrupts long-distance
regulatory interactions in the Abd-B gene of the bithorax complex. Mol. Cell 8:1145–1151.
Kalendar, R., J. Tanskanen, S. Immonen, E. Nevo, and A. H.
Schulman. 2000. Genome evolution of wild barley (Hordeum
spontaneum) by BARE-1 retrotransposon dynamics in
response to sharp microclimatic divergence. Proc. Natl. Acad.
Sci. USA 97:6603–6607.
Kazazian, H. H. 2004. Mobile elements: drivers of genome evolution. Science 303:1626–1632.
Krebs, R. A., and M. E. Feder. 1997a. Natural variation in the
expression of the heat-shock protein Hsp70 in a population
of Drosophila melanogaster, and its correlation with tolerance
of ecologically relevant thermal stress. Evolution 51:173–179.
Krebs, R. A., and M. E. Feder. 1997b. Deleterious consequences
of Hsp70 overexpression in Drosophila melanogaster larvae.
Cell Stress Chaperones 2:60–71.
Lee, H., K. W. Kraus, M. F. Wolfner, and J. T. Lis. 1992. DNA
sequence requirements for generating paused polymerase at the
start of hsp70. Genes Dev. 6:284–295.
Lerman, D. N., P. Michalak, A. B. Helin, B. R. Bettencourt, and
M. E. Feder. 2003. Modification of heat-shock gene expression
in Drosophila melanogaster populations via transposable elements. Mol. Biol. Evol. 20:135–144.
Li, B., J. A. Weber, Y. Chen, A. L. Greenleaf, and D. S. Gilmour.
1996. Analyses of promoter-proximal pausing by RNA polymerase II on the hsp70 heat shock gene promoter in a Drosophila nuclear extract. Mol. Cell. Biol. 6:5433–5443.
Lis, J., and C. Wu. 1993. Protein traffic on the heat shock promoter: parking, stalling, and trucking along. Cell 74:1–4.
Maside, X., C. Bartolome, and B. Charlesworth. 2002. S-element
insertions are associated with the evolution of the hsp70 genes
in Drosophila melanogaster. Curr. Biol. 12:1686–1691.
Mason, P. B., and J. T. Lis. 1997. Cooperative and competitive
protein interactions at the Hsp70 promoter. J. Biol. Chem.
272:33227–33233.
Michalak, P., I. Minkov, A. Helin, D. N. Lerman, B. R. Bettencourt,
M. E. Feder, A. B. Korol, and E. Nevo. 2001. Genetic
evidence for adaptation-driven incipient speciation of Drosophila melanogaster along a microclimatic contrast in
‘‘Evolution Canyon,’’ Israel. Proc. Natl. Acad. Sci. USA 98:
13195–13200.
Modolell, J., W. Bender, and M. Meselson. 1983. Drosophila melanogaster mutations suppressible by the suppressor of hairywing are insertions of a 7.3-kilobase mobile element. Proc. Natl.
Acad. Sci. USA–Biol. Sci. 80:1678–1682.
O’Brien, T., and J. T. Lis. 1991. RNA polymerase II pauses at the
5# end of the transcriptionally induced Drosophila hsp70 gene.
Mol. Cell. Biol. 11:5285–5290.
Ohare, K., and G. M. Rubin. 1983. Structures of P-transposable
elements and their sites of insertion and excision in the Drosophila melanogaster genome. Cell 34:25–35.
Omichinski, J. G., P. V. Pedone, G. Felsenfeld, A. M. Gronenborn,
and G. M. Clore. 1997. The solution structure of a specific
GAGA factor-DNA complex reveals a modular binding mode.
Nat. Struct. Biol. 4:122–132.
Rasmusson, K. E., J. D. Raymond, and M. J. Simmons. 1993.
Repression of hybrid dysgenesis in Drosophila melanogaster
by individual naturally-occurring P-elements. Genetics
133:605–622.
Rockman, M., and G. Wray. 2002. Abundant raw material for cisregulatory evolution in humans. Mol. Biol. Evol. 19:1991–
2004.
Schlenke, T. A., and D. J. Begun. 2004. Strong selective sweep
associated with a transposon insertion in Drosophila simulans.
Proc. Natl. Acad. Sci. USA 101:1626–1631.
Shopland, L. S., K. Hirayoshi, M. Fernandes, and J. T. Lis.
1995. HSF access to heat-shock elements in vivo depends
critically on promoter architecture defined by GAGA factor,
TFIID and RNA polymerase II binding sites. Genes Dev.
9:2756–2769.
Sorensen, J. G., J. Dahlgaard, and V. Loeschcke. 2001. Genetic
variation in thermal tolerance among natural populations of
Drosophila buzzatii: down regulation of Hsp70 expression
and variation in heat stress resistance traits. Funct. Ecol.
15:289–296.
Weber, J. A., D. J. Taxman, Q. Lu, and D. S. Gilmour. 1997.
Molecular architecture of the hsp70 promoter after deletion
of the TATA box or the upstream regulation region. Mol. Cell.
Biol. 17:3799–3808.
Wilkins, R. C., and J. T. Lis. 1997. Dynamics of potentiation and
activation: GAGA factor and its role in heat shock gene regulation. Nucleic Acids Res. 25:3963–3968.
Xiao, H., and J. T. Lis. 1988. Germline transformation used to
define key features of heat-shock response elements. Science
239:1139–1142.
Zatsepina, O. G., V. V. Velikodvorskaia, V. B. Molodtsov,
D. Garbuz, D. N. Lerman, B. R. Bettencourt, M. E. Feder,
and M. B. Evgenev. 2001. A Drosophila melanogaster strain
from sub-equatorial Africa has exceptional thermotolerance
but decreased Hsp70 expression. J. Exp. Biol. 204:1869–1881.
Claudia Schmidt-Dannert, Associate Editor
Accepted November 25, 2004