ABSTRACT
All
Drosophila
alcohol dehydrogenase (
Adh
) genes sequenced to date contain two small introns within the coding region.
These are conserved in location and, to some extent, in sequence between the
various species analyzed. To determine if these introns play a role in
Adh
gene expression, derivatives of the
Drosophila affinidisjuncta
Adh
gene lacking one or both introns were constructed and analyzed by germline and
trans- ient transformation of
Drosophila melanogaster
. Removal of both introns lowered expression, whether measured by enzyme
activity or by RNA levels. The decrease was seen in both germline transformed
and transiently transformed larvae, with the effect being larger for germline
transformants. Similar decreases (averaging 5-fold) were also seen at the embryonic and adult stages for germline
transformants. Nuclear run-off transcription with nuclei from germline transformed embryos indicated
that the reduction in RNA levels is due to decreased transcription. However,
LacZ
fusion constructs designed to test for the presence of a classical enhancer in
the introns provided no evidence for such a mechanism. Removal of each intron
individually resulted in more complex phenotypes. The introns have smaller,
additive effects on expression in adults. In larvae, removal of the upstream
intron significantly increases RNA levels but modestly decreases enzyme
activity. Removal of the downstream intron lowers expression in both germline
and transiently transformed larvae, but also increases position effects in
germline transformants. Therefore, the small introns are clearly needed for
optimal transcription of this
Adh
gene, but multiple mechanisms are involved.
The
Adh
genes from many
Drosophila
species contain two promoters (
1
,
2
). Primary transcripts originating from the upstream (distal) promoter contain
three introns: a large distal- specific intron (intron 1) in the non-translated leader and two smaller introns (introns 2 and 3) within
the coding region. Transcripts arising from the downstream (proximal) promoter
carry only the two small introns. The positions and approximate sizes of these
two smaller introns are strongly conserved among the
Adh
genes from various species (
3
). Therefore, they could conceivably contribute to conserved aspects of
Adh
gene expression, such as transcription, in the larval and adult fat bodies.
Though there is precedent for transcriptional regulatory elements within introns
(
4
-
8
), no prior study has shown that the small introns (<100 bp) within the coding region of
Drosophila
Adh
genes are needed for normal expression. Shen and co-workers addressed this for the
Drosophila
melanogaster
Adh
gene, by employing a transient transformation system that measured gene
expression predominantly in the larval midgut (
9
). Their results suggest that the small introns of this gene are not essential
for normal expression. On the other hand, phylogenetic analyses hint that there
may be important roles for the small introns. Sullivan and colleagues (
10
) point out that rates of intron sequence divergence between some
Drosophila
species are less than expected relative to silent substitution rates.
Furthermore, co-variational analysis suggests that a conserved stem-loop structure is formed by sequences within the second intron of
the
Adh
genes from 10 diverse species of
Drosophila
, including
Drosophila
affinidisjuncta
(
11
).
To determine if the small introns within the coding region of the
D.affinidisjuncta
gene are needed for normal transcription, genes lacking one or both of these
small introns were constructed and tested by both germline and transient
transformation. The gene lacking both introns is expressed at a significantly
lower level than the unaltered gene. The decrease in expression was most
pronounced in germline transformed embryos, larvae and adults, but was also
seen in the fat body of transiently transformed larvae. Both introns contribute
to expression, as deletions of each intron individually have distinct yet
relatively modest effects. In germline transformed larvae, an unexpected
increase in RNA levels resulted from deletion of intron 2, which contains the
putative stem-loop structure. Deletion of the third intron caused increased
susceptibility to chromosomal position effects in germline transformed larvae.
The observed decrease in expression upon deletion of both introns is due to
lowered transcription as shown by nuclear run-off transcription. However, gene fusions provided no evidence for the
presence of a DNA enhancer in the introns. Taken together, these results
suggest that the small introns are needed for normal transcription but that
mechanisms other than a classical DNA enhancer are involved.
The
D.affinidisjuncta
stock S36G1 was used as the source of the genomic and cDNAs for cloning (
12
). All
D.melanogaster
stocks were maintained as previously described (
13
). The
Adh
null stock,
Adh
fn6
cn
;
ry
506
, was used as a host for both P element-mediated germline and transient transformation. Unless otherwise
specified, transformants were harvested as embryos (aged 13-19 h), feeding third instar larvae (aged 7-8 days post-hatching) or as adults (aged 4-8 days post-eclosion).
The cDNA clone pADH3A (
14
) and the 5.4 kb
D.affinidisjuncta
fragment (-2832/+2618), abbreviated Aff, were used to construct intron-deleted genes (Fig.
1
). Sequence identity between the coding regions of the genomic and cDNA clones
was confirmed by dideoxynucleotide sequencing (
14
; M.Brennan, unpublished data, GenBank accession no. for cDNA U63563). Other
genes were constructed by standard methods (
15
) using available restriction enzyme cleavage sites. Unless otherwise stated,
restriction endonucleases and other DNA modifying enzymes were obtained from
New England Biolabs and used according to the supplier's recommendations. To
produce plasmids for germline transformation, genes were inserted into Carnegie
20 as described (
16
,
17
). A heat-shock promoter-
LacZ
fusion gene, SP73Lac1 (kindly provided by T.Abel), served as the reporter to
assess enhancer function. Briefly, this construct consists of the
D.melanogaster hsp70
promoter (-43 to +265) fused in-frame to the
Escherichia coli
LacZ
gene inserted into the
Sma
I site of the polylinker of the vector pSP73 (
18
). Fragments of the
D.affinidisjuncta
Adh
gene tested for enhancer function were inserted in the sense orientation
between the
Xba
I and
Xho
I sites of the polylinker immediately downstream of the
LacZ
gene. All plasmids were purified by CsCl density gradient centrifugation prior
to injection.
To produce germline transformants, plasmids were injected into pre-blastoderm embryos and homozygous stocks were established from transformed
progeny as described previously (
13
,
17
,
19
-
21
). Only transformed stocks carrying single intact transposons in autosomal
locations (as determined by Southern analysis) were analyzed (
22
). Determination of alcohol dehydrogenase enzyme (ADH) specific activity for
germline transformants was as described (
23
,
24
). RNA levels were determined by RNase protection assays with the probes and
methods detailed previously (
25
). Quantification of RNA levels was by scanning densitometry of autoradiograms
or by phosphorimaging (Molecular Dynamics PhosphorImager) and, unless otherwise
specified, all
Adh
values were corrected by normalization to the actin control (
25
).
Transient transformation of the larval fat body, including measurement of ADH
activity and [beta]-galactosidase activities, was as described previously (
26
). Unless otherwise specified, the
vermillion
-
LacZ
fusion gene (pTUF1.1
v-LacZ
) served as an internal control for injection efficiency (
27
).
Nuclear run-off transcription reactions were carried out by a modification of
previously described methods (
28
,
29
). Embryos (aged 13-19 h) were collected on nylon mesh and rinsed with chilled water (10-15oC). They were dechorionated for 90 s in a 2-fold dilution of commercial bleach at room temperature
and rinsed quickly in chilled wash solution buffer (0.7% w/v NaCl, 0.04% v/v
Triton X-100) using 10 ml/g embryos. They were then rinsed in chilled water (10-15oC), blotted dry and weighed. Unless otherwise specified, all
subsequent steps were performed at 0-4oC. Typical yields were 50-150 mg embryos from four half-pint bottles of adult flies. Volumes, where specified,
are those used for 100 mg embryos.
Embryos were homogenized in a 7 ml Dounce homogenizer (Wheaton Scientific)
containing 1 ml ice-cold buffer I [15 mM HEPES (K
+
), pH 7.6, 10 mM KCl, 5 mM MgCl
2
, 0.1 mM EDTA, 0.5 mM EGTA, 350 mM sucrose, 1 mM dithiothreitol, 1 mM sodium
bisulfite, 1 mM benzamidine, 0.2 mM phenylmethylsulfonyl fluoride] using the B
pestle. Homogenates were filtered through one layer of Miracloth (Calbiochem)
and the debris retained by the Miracloth was rinsed with an additional 0.2 ml
buffer I. The nuclei were transferred to a 15 ml Corex tube and centrifuged in
a Sorvall SS-34 rotor at 9200 r.p.m. (10 000
g
) for 15 min. The supernatant was carefully decanted from the loose pellet of
nuclei and the nuclei were suspended, avoiding both the hard yellow yolk pellet
and lipid deposits, in 1 ml chilled reaction buffer (5 mM Tris-HCl, pH 8, 5 mM MgCl
2
, 0.3 M KCl). The nuclei were transferred to a 15 ml Corex tube and dispersed by
gentle pipetting through a 1-200 [mu]l pipette tip. They were centrifuged in a Sorvall SS-34 rotor at 9200 r.p.m. for 15 min and then resuspended in 0.1
ml freshly prepared reaction buffer containing 1 mM each of ATP, CTP and GTP by
gentle pipetting. The reaction mixture was transferred to a 1.5 ml Eppendorf
centrifuge tube, 5 [mu]l [[alpha]-
32
P]UTP (ICN, 760 Ci/mmol, 10 mCi/ml) was added and the tube was agitated at 25oC for 30 min.
To prepare RNA, 150 [mu]l HSB buffer (0.5 M NaCl, 50 mM MgCl
2
, 2 mM CaCl
2
, 10 mM Tris-HCl, pH 7.4) containing 0.12 U/[mu]l RNase-free DNase I (Stratagene) were added to the reaction mixture
followed by incubation at 30oC for 5 min. Digestion was stopped by adding 50 [mu]l SDS/Tris buffer (5% SDS, 0.5 M Tris-HCl, pH 7.4, 0.125 M EDTA), 2.5 [mu]l proteinase K (20 mg/ml) and 10 [mu]l yeast tRNA (10 mg/ml). This was incubated at 42oC for 60 min followed by two extractions with an
equal volume of phenol/chloroform/isoamyl alcohol (24:24:1 v/v/v) and ethanol
precipitation. Nucleic acids were recovered by centrifugation, dried under
vacuum and suspended in 750 [mu]l DNase I/elution buffer (2.5 mM MgCl
2
, 0.5 mM CaCl
2
, 0.5% SDS, 2.5 mM EDTA, 10 mM HEPES, pH 7.5). Then, 188 [mu]l 1 M NaOH were added and the mixture was placed on ice for 10 min. The
reaction was quenched with 375 [mu]l 1 M HEPES (free acid) in a 15 ml Corex tube followed by precipitation with
sodium acetate (0.33 M) and 2.5 vol cold ethanol. Nucleic acids were recovered
by centrifugation for 30 min at 10 000 r.p.m. in a HB-4 rotor and dissolved in 400 [mu]l TES (10 mM Tris-HCl, pH 7.4, 10 mM EDTA, 0.2% SDS). RNA was precipitated with
0.075 vol 4 M NaCl and 2.5 vol ethanol, recovered by centrifugation, dried
under vacuum and dissolved in 250 [mu]l TES. Incorporation of
32
P was determined by scintillation counting; typical incorporation was 4-8 * 10
6
c.p.m./ml. Samples were then precipitated with 0.5 vol 7.5 M ammonium acetate
and 2.5 vol cold ethanol. RNA was dissolved in 250 [mu]l TES. An additional 250 [mu]l TES solution containing 0.6 M NaCl (1* TES/NaCl) was then added. The total volume of solution was
hybridized to DNA immobilized on filter strips with gentle agitation for 36 h
at 65oC.
Plasmids were linearized and bound to nitrocellulose as described (
29
). Three plasmids were used as controls: pUC19, actin 5C and rp49. The actin 5C
plasmid, a gift of S.Tobin, is a 3.7 kb
Eco
RI-
Hin
dIII genomic fragment carrying exons 1 and 2 of the actin 5C gene (
30
) inserted into pGEM-1 (Promega). The rp49 (ribosomal protein-49) plasmid, HR0.6, has been described elsewhere (
31
). The cDNA-containing
Adh
plasmid used was pADH3A (
14
). Membranes were blocked with 1* TES/NaCl containing 10* Denhardt's by gentle agitation for 3 h at 65oC, followed by a similar incubation with prehybridization
buffer [1* TES/NaCl solution, 10* Denhardt's, 50 [mu]g/ml sonicated, single-stranded calf thymus DNA, 10 [mu]g/ml poly(A) (Boehringer Mannheim) and 0.1% w/v
sodium pyrophosphate].
Following hybridization, membranes were washed with 2* SSC (1* SSC = 150 mM NaCl, 15 mM sodium citrate, pH 7.0) for 1 h at 65oC and then incubated in 2* SSC containing RNase A (10 [mu]g/ml) at 37oC for 30 min. Membranes were washed with 2* SSC at 37oC for 1 h, followed by washes in
1* and 0.5* SSC for 10 min each at 65oC. Autoradiograms were prepared and signal strength was
determined by scanning densitometry.
Four to five different germline transformed lines were analyzed for each
construct. Enzyme activity and RNase protection values for germline transformed
larvae and adults were obtained from three independent preparations. The mean
value for the transformed line was then treated as a single determination such
that
n
for each gene analyzed by germline transformation was equal to the number of
independent lines. For transient transformation, five samples, each consisting
of 10 dissected larvae, were analyzed for each construct. Mean values for
different genes were compared by the two sample Student's
t
-test for two-tailed hypotheses (
32
).
To determine if the two small introns common to both distal and proximal
transcripts are critical to expression of the
D.affinidisjuncta
gene, constructs lacking one or both introns were used for P element-mediated and transient transformation of
D.melanogaster
. As a starting construct we used a 5.4 kb genomic fragment from
D.affinidisjuncta
(Aff, Fig.
1
) described previously (
17
,
26
).
In germline transformed larvae, the construct lacking both small introns, [Delta]Ints, shows a 3- to 13-fold (mean 5-fold) drop in enzyme activity and a similar 2- to 7-fold (mean 3-fold) drop in RNA levels (Fig.
2
). The same gene, when analyzed by transient transformation, shows a 2-fold drop in enzyme activity (Fig.
2
). As described below, to determine if both introns are involved in the effects
observed, constructs lacking either of the small introns were similarly
analyzed by germline and transient transformation.
To determine if the two small introns are critical to expression of the
Adh
gene in adults, enzyme activities and RNA levels of adult germline
transformants carrying the same constructs were analyzed. Loss of both introns
results in a 4- to 14-fold (mean 7-fold) decrease in enzyme activity and a 3- to 10-fold (mean 4-fold) drop in total
Adh
RNA levels. The latter can be partitioned into a 3- to 13-fold (mean 5-fold) drop in distal transcript levels and a 2- to 6-fold (mean 3-fold) drop in proximal transcript levels
(Fig.
3
).
Deletion of the two small introns lowers
Adh
RNA levels 3- to 5-fold in both larvae and adults. This drop in steady-state levels of RNA is due either to decreased transcription
or to enhanced degradation of transcripts from the altered genes. The former
possibility would be expected if the introns contain elements that contribute
to transcription by any of several possible mechanisms, including one or more
DNA or RNA enhancers or altered chromatin structure. However, it is conceivable
that loss of the two small introns, along with the intronic splicing signals,
affects the ability of these altered transcripts to be properly processed
(capping, proper 3'-end generation and polyadenylation). This, in turn, may hinder any
of a number of processes that could affect RNA stability, such as export from
the nucleus, subcellular localization or ribosome targeting (
33
).
In order to differentiate between effects on transcription and stability,
transcription rates for the intron-deleted gene ([Delta]Ints) must be compared with that of the intron-containing gene by nuclear run-off transcription. The previously described experiments
involved larval and adult flies. However, nuclear run-off experiments in
Drosophila
are more conveniently accomplished using embryos, because the methods for
obtaining nuclei from embryos are relatively well established.
To determine the developmental profile of
Adh
transcription in
D.melanogaster
embryos carrying the unaltered
D.affinidisjuncta Adh
gene, RNA was isolated from staged, transformed embryos and analyzed by RNase
protection. The signals were scanned densitometrically and the results are
shown in Figure
4
. Low levels of both distal and proximal transcripts are first detected at ~1 h following egg laying. These remain relatively unchanged to ~7 h, whereupon distal transcripts increase, followed by a similar
increase in proximal transcripts at ~10 h. Levels of distal RNA peak at ~10 h, whereas levels of proximal RNA continue to increase until ~17 h and then plateau. This profile is remarkably similar to that
determined by others for the
D.melanogaster
gene (
34
,
35
).
Prior to this study, no regulatory role had been assigned to the small introns
of any
Adh
gene in
Drosophila
. The present study shows that the small introns of the
D.affinidisjuncta
gene individually have modest but significant effects on expression and that
removing both introns lowers expression an average of 5- to 10-fold in embryos, larvae and adults. Expression from both the distal
and proximal promoters is affected by the presence of both introns. Nuclear run-off transcription demonstrated that the two small introns clearly
influence transcription. However, multiple mechanisms are likely to be
involved.
The effect on transcription is probably not due to the presence of a classical
enhancer in one or both small introns. The reporter construct that we used to
test for the presence of a DNA enhancer is known to be responsive to
D.melanogaster
Adh
enhancers (
18
). However, we cannot rule out the possibility that the small introns contain an
enhancer that fails to activate the
hsp70
promoter. Furthermore, although we have retained relative positioning and
orientation within the intron-containing fragment tested, it is possible that a sub-element of a large enhancer lies outside the tested region (
36
).
One possible mechanism by which lack of intervening sequences could interfere
with transcription of this gene involves torsional stress. Intronic sequences
may serve in part to relieve such forces inherent in the transcription process
(
37
). The relatively greater influence of the introns when the genes are assayed by
incorporation into the chromosomes rather than by transient transfection is
consistent with a systematic difference in torsional stresses between the
chromosomal and plasmid environments. Other possible mechanisms are discussed
below in the context of removing each of the small introns individually.
The RNA/activity ratios (relative RNA levels compared with relative enzyme
activities) for the individual transformed stocks carrying the unaltered gene,
Aff, show little deviation from 1. In contrast, RNA levels in larvae for stocks
carrying [Delta]I(2) exceed enzyme activities consistently by at least 2-fold. This may reflect ordered splicing of the primary transcript.
Evidence from several systems indicates that splicing of introns proceeds in an
ordered fashion such that the presence of one intron facilitates removal of
another (
38
-
40
). If splicing of the
D.affinidisjuncta
Adh
primary transcript also proceeds in an ordered fashion, deletion of the second
intron may affect removal of the downstream third intron. Consistent with this
possibility, the second intron contains 5' and 3' splice sites that more precisely match the consensus for small
introns in
Drosophila
(
3
,
41
). Thus, the splicing machinery may interact less efficiently with primary
transcripts lacking intron 2, which could lead to their accumulation in the
nucleus or the export of incompletely processed RNAs to the cytoplasm (
33
,
40
). Given that the RNase protection assay measures only the 5'-termini of the
Adh
transcripts and we isolated total cellular RNA, this may account for inflated
RNA levels for [Delta]I(2) larval transformants accompanied by low to normal levels of enzyme
activity. Alternatively, if the [Delta]I(2) transcripts are transported normally but not translated efficiently,
perhaps due to poor polyadenylation or capping (
33
,
42
-
44
), this could explain the altered relationship between RNA levels and enzyme
activities, although it would not account for the inflated RNA levels.
A possible explanation for the relative importance of the second intron in
larvae is suggested by the work of Stephan and Kirby (
11
), which provides evidence for a stem-loop that is conserved across 10 species of
Drosophila
, including
D.affinidisjuncta
. In the
D.affinidisjuncta
gene, the potential stem-loop is positioned toward the 5'-end of the 81 nt intron, with 13 nt preceding it. Such RNA
secondary structures have been shown to be involved in splicing (
45
-
47
).
Results for a naturally occurring sequence variant of the
D.melanogaster
gene may parallel these findings. Laurie and Stam (
8
) demonstrated that deletion of sequences, containing a potential hairpin
structure (
48
), within the large distal-specific intron of the
Fast
allele of
Adh
in
D.melanogaster
, causes a 15-20% higher protein level in ADH
Fast
adults compared with ADH
Slow
adults. However, they found no statistically significant difference between
levels of RNA for the two genes, suggesting that secondary structure in the
primary transcript affects translational efficiency.
Secondary structure within primary transcripts can influence transcription as
well (
49
). In HIV-1 transcripts, a stem-loop referred to as TAR interacts with a viral protein, Tat, to
affect transcription. The manner in which this occurs has yet to be resolved.
The ability of the TAR-Tat complex to activate the formation of subsequent transcription
complexes led Sharp and Marciniak to designate such elements `RNA enhancers' (
50
). Others have suggested that in the absence of Tat, TAR serves as a
transcription terminator (
51
). In the present study, the increase in RNA for [Delta]I(2) in larvae is consistent with loss of a potential transcription
terminator similar to that described above. However, there may be more than one
mechanism at work, since RNA levels for [Delta]Ints, which also lacks the potential stem-loop terminator in the second intron, fail to show such an
increase.
Adults carrying [Delta]I(2) do not show inflated RNA levels. Perhaps the upstream distal intron
facilitates recognition of the primary transcript by the splicing or export
machinery in adults. It is not clear whether this differs from processing of
the primary transcript of the
D.melanogaster
Adh
gene. In this case, mutations at either the donor or acceptor site of the
second intron result in accumulation of a large transcript, presumably carrying
the distal intron, in adults (
52
). We have observed a similar accumulation of transcripts containing the distal
intron in adults carrying the
D.affinidisjuncta
gene lacking both small introns, but not in those carrying either gene lacking
one of the two small introns (unpublished results). Large and small introns in
Drosophila
have somewhat different 5' and 3' consensus splice sites (
41
) and may, in fact, be recognized and removed by somewhat different mechanisms (
53
,
54
). Possibly, at least one of the small introns is needed for efficient splicing
of the larger upstream intron.
In contrast to the systematically inflated values for the ratio of RNA to enzyme
activity for larvae from stocks carrying [Delta]I(2), larvae from stocks carrying [Delta]Ints and [Delta]I(3) exhibit extreme ranges of RNA levels and enzyme
activities along with poor correlation between the two measures of gene
expression. This variation, due to a position effect in germline transformants,
is seen only in larvae. In larvae, the range of enzyme activities for [Delta]Ints is +-156% of the mean value and that for [Delta]I(3) is +-115%. In contrast, these values for Aff and [Delta]I(2) are +-38 and +-37% respectively.
Similarly, the ranges of RNA levels are relatively large for genes lacking the
third intron [[Delta]Ints +-169% and [Delta]I(3) +-138% versus Aff +-24% and [Delta]I(2) +-42%]. The mechanism by
which intron 3 buffers against position effects in larvae is unclear.
Interestingly, studies of linkage disequilibria suggest the existence of a stem-loop structure in the corresponding intron of the
D.pseudoobscura
gene (
55
). So it is possible that an RNA enhancer is involved. Alternatively, by analogy
with the
D.melanogaster
gene, nucleosomal positioning is likely to be important for transcription (
56
,
57
) and this could be affected by the presence of the third intron.
In summary, the two small introns of the
D.affinidisjuncta
Adh
gene clearly affect its transcriptional efficiency. Therefore, such small
introns may be more important in gene expression than generally assumed. The
introns do not contain a classical DNA enhancer, at least for larvae. They may,
however, contain one or more RNA enhancers. Additionally, in larvae removal of
the second intron likely results in altered RNA processing or nuclear export,
while removal of the third intron causes increased sensitivity to chromosomal
position effects.
We thank T.Maniatis and T.Abel for providing the SP70Lac1 construct, S.Tobin for
the actin 5C plasmid and L.Searles for providing pTUF1.1
V-LacZ
. We also thank our colleague J.Hu for providing
Adh
plasmids used for some constructs. This work was supported by NIH grant R01-GM34961 and a University of Louisville Medical Research Fund grant to MDB.
REFERENCES
Return