ABSTRACT
Expression of the
[beta]
1 tubulin
gene of
Drosophila melanogaster
is under complex developmental control. For high levels of transcription in the embryonic central nervous system (CNS) different modules dispersed over 3 kb have to co-operate. Combination of a core promoter with either far upstream localized
enhancer elements or, alternatively, with an enhancer from the intron results
in expression limited to only a few neuronal cells. Cooperation of all three
modules, however, leads to high level expression in most neuronal cells of the
CNS. In the intron, we identified a 6 bp core element which is essential for
transcription in the CNS, as well as an 8 bp element required for maternal
expression. Interestingly, both motifs are quite similar, with CAAAAT as the
CNS core and CAAAAAT as the maternal enhancer core. Specific binding of
proteins from nuclear extracts to the CNS-specific element could be demonstrated. We suggest that the
[beta]
1 tubulin
gene represents an ideal marker gene to elucidate connections between pro-neural or neurogenic genes and downstream target genes throughout the CNS.
Formation of the
Drosophila
central nervous system (CNS) occurs from a stereospecific set of stem cells
called neuroblasts, which comprise ~20% of the neuro-ectodermal region (for reviews see
1
,
2
). Genetic data show that these cells are singled out by a lateral inhibitory
process. Their further competence to adopt a neural fate is mediated by the
expression of the pro-neural genes, which include
daughterless
(
da
), the genes of the
achaete
-
scute
complex (
AS-C
) and the
ventral nervous system defective
gene (
vnd
) (
3
-
7
). Ongoing differentiation of the CNS is marked by the expression of a group of
genes referred to as pan-neural genes (
8
). According to mutant phenotypes, these genes probably play a role in
establishing neuronal cell type identity. This would explain why in mutants for
prospero
, a gene encoding a nuclear protein containing a homeodomain-like sequence, incorrectly specified progeny of certain neuroblasts are
produced, which show pathfinding defects (
9
,
10
).
Concerning regulation of the transcription of neural-specific genes, little information at the molecular level is available. At
least partially, activation of the proneural genes is directed by pair-rule and dorso-ventral polarity genes. Furthermore, their activity is controlled by the lateral inhibition pathway and restricted to
the neural precursors. Basically, at least in the peripheral nervous system
(PNS), repressing activities seem to play a major role, as in mutants for
hairy
,
extramacrochaete
and
pannier
ectopic expression of the
AS-C
genes is observed, which leads to the formation of supernumerous sensory organs
(
1
1
-
13
). Kramatschek and Campos-Ortega (
1
4
) showed that the neurogenic genes
E(spl)
and
HLH-m5
can be activated by the pro-neural gene products.
Few target genes of the neurogenic and pro-neural gene products expressed throughout the CNS have been identified so
far. By analyzing the promotor of the neuron-specifically expressed gene
elav
, Yao and White (
15
) described a 333 bp region as sufficient for CNS-specific expression. However, no distinct elements have been identified
and no transactivators for
elav
could be shown. Surprisingly, Ip
et al.
(
1
6
) recently presented data showing that expression of the
snail
gene in the CNS is not dependent on the pro-neural genes, demonstrating that distinct pathways must co-exist in order to define neuronal identity.
The [beta]
1 tubulin
gene of
D.melanogaster
is expressed maternally as well as zygotically. Transcription during
embryogenesis is restricted to the CNS and PNS and to the attachment sites of
the somatic muscles (
17
). While the [beta]1 protein is present at high levels in all tissues until hatching of the
larvae, the maternal [beta]1 mRNA disappears around stage 9. Zygotic transcription in the CNS starts
at stage 10, reaches its maximum level at stage 12 and decreases after stage 13
to very low levels at stage 16. In a previous report we showed that expression
of the [beta]
1 tubulin
gene in the CNS is dependent on the presence of the 5'-part of the intron (
17
). By sequence comparison with the
D.hydei
[beta]
1 tubulin
gene (Buttgereit, unpublished results) we identified two conserved elements in
the intron, which were termed IE1 and IE2. Deletion analysis revealed that IE1
is solely necessary for expression in the CNS during embryogenesis. By shifting
its position to -2.3 kb upstream of the start site we present evidence that this element
acts as a classical enhancer. Based on its zygotic function, the enhancer is
called the
The vector used for cloning was -2.2/BS, comprising [beta]
1 tubulin
gene sequences from -2348 to +448 in BlueskriptII KS. The clone was cut with
Nar
I (+160) and
Hin
dIII (+473, polylinker derived from pUC19), the 5.2 kb fragment isolated and
synthetic oligonucleotides were inserted. For construct [beta]1-1, the sequence used (MutD1, 56 bp) was 5'-CGCCAAGGTA AGTTTTCCCA TTTGCATTTT CCATTCATTT TCCAGTACTG
GATCCA, deleting IE1 and IE2. For [beta]1-2 the oligonucleotide MutD2 comprised 129 bp, 5'-CGCCAAGGTA AGTTTTCCCA TTTGCATTTT CCATTCATTT TCCATCCAGA
GTACAGTTGC
CAAAATGGCG TCATTTTTGC
ACGACTTCGC TGT
CCATGTG GCAAAAATT
T GTATTGAGTA CTGGATCCA (IE1 and IE2 are in bold). For constructs [beta]1-3 and [beta]1-4, the following oligonucleotides were used: OlIE1 with
100 bp comprising only IE1, which is in bold, 5'-CGCCAAGGTA GTTTTTCCCA TTTGCATTTT CCATTCATTT TCCATCCAGA GTACAGTTGC
CAAAATGGCG TCATTTTTGC
ACGACTTAGT ACTGGATCCA); OlIE2 (129 bp), 5'-CGCCAAG- GTA AGTTTTCCCA TTTGCATTTT CCATTCATTT TCCATCCAGA GTACAGTTGC
CTGATCGAAC TTGTACTAGT ACGACTTCGC TGT
CCATGTG GCAAAAATT
T GTATTGAGTA CTGGATCCA, comprising IE2 only (bold). IE1 was mutagenized
randomly. For construct [beta]1-7, the oligonucleotide SPL (55 bp), 5'-CGCCAAGGTA AGTTT
CAAAA TGGCGTCATT TTTGC
ACGAC TTAGTACTGG ATCCA, comprising IE1 (bold) was used. The bases in front of IE1 (+175 to
+220) were deleted, but the splice site is conserved. In order to test the
relevance of both IE1 and IE2, construct [beta]1-9, with the intron to +275 bp, was constructed. In contrast to [beta]1-2, IE1 and IE2 (bold) were both mutagenized by randomly
shuffling bases in oligonucleotide IE1/2Mut (129 bp), 5'-CGCCAAGGTA AGTTTTCCCA TTTGCATTTT CCATTCATTT TCCATCCAGA GTACAGTTGC
CGCTGTGGCG TTATCTCTAT
ACGACTTCGC TGT
CCATGTG GCCACATGG
T GTATTGA- GTA CTGGATCCA. For construct [beta]1-8, the clone [beta]1-1/BS, which does not contain IE1 and IE2, was
used. The DNA was cut with
Sac
I at -2.4 kb and the oligonucleotide EB1 (73 bp), 5'-CAGTTGC
CAA AATGGCGTCA TTTTTGC
ACG ACT- TCGCTGT C
CATGTGGCA AAAAT
TTGTA TTGGATCCGA GCT, comprising IE1 and IE2 (bold) was inserted. For construct [beta]1-10 the oligonucleotide Neuro (55 bp), 5'-CGCCAAGGTA AGTTT
AGCTC A
GGCGT
CATT TTTGC
ACGAC TTAGTACTGG ATCCA, which mutagenizes the CNE, was inserted. To obtain construct [beta]1-11, the oligonucleotide Mater (55 bp), 5'-CGCCAAGGTA AGTTT
CAAAA T
GGCGT
ACGT CAGT-
CACGAC TTAGTACTGG ATCCA, which mutagenizes the second motif (ME), was used.
For all oligonucleotides only one strand is shown. The double-stranded oligonucleotides had at their ends the corresponding restriction half-sites as adaptors for cloning and usually an additional
Bam
HI site for identification of positive clones. The fragments were excised from
Bluescript with
Eco
RI and
Sca
I and ligated to the adapter fragment
Sca
I-
Xba
I from PF3 (
17
), containing the 3' splice site from the [beta]
1 tubulin
gene and the codons to amino acid 23 fused in-frame to the
Escherichia coli
lacZ
gene, in the P-element transformation vector pW8 (
1
8
) cut with
Eco
RI and
Xba
I. All constructs were checked by sequencing of the intron or upstream region,
respectively.
Nuclear extracts were prepared according to Dignam
et al
. (
19
) with modifications. Embryos were collected on apple juice/agar plates,
dechorionized with 50% Clorix, washed with 0.7% NaCl, 0.1% Triton X-100 and suspended in homogenization buffer (10 mM HEPES-KOH, pH 7,6, 10 mM KCl, 1.5 mM MgCl
2
, 0.1 mM EDTA, 0.25 M sucrose, 0.5 mM DTE, 0.5 mM leupeptin). They were homogenized with
a motor-driven glass/Teflon homogenizer with a tight fitting pestle at 600 r.p.m.
Intact embroys and cell debris were pelleted by centrifugation for 5 min at 800
r.p.m. The supernatant was loaded onto a cushion of 30% glycerol in
homogenization buffer and nuclei were collected by centrifugation at 6000
r.p.m. for 20 min. The volume of the nuclear pellet was estimated and one
volume of nuclear extraction buffer was added (20 mM HEPES-KOH, pH 7.6, 420 mM KCl, 1.5 mM MgCl
2
, 0.1 mM EDTA, 0.5 mM DTE, 0.5 mM leupeptin). The suspension was stirred on ice
for 30 min, the nuclei pelleted by centrifugation at 13 000 r.p.m. and the
supernatant dialyzed twice against 200 volumes buffer A (20 mM HEPES-KOH, pH 7.6, 100 mM KCl, 5 mM MgCl
2
, 0.1 mM EDTA, 0.5 mM DTE, 2 [mu]g/ml leupeptin). To remove precipitated material, the extract was
centrifuged for 10 min at 13 000 r.p.m. and quick-frozen in small aliquots in liquid N
2
. Aliquots were stored at -80oC without significant loss of activity for at least 12 months.
Standard binding reactions contained up to 30 [mu]g nuclear extract, 6 fmol
32
P-labeled oligonucleotide and 200-500 ng dI[middot]dC or dA[middot]dT as non-specific competitor in 25-50 [mu]l. Specific competition was performed in the range 5- to 500-fold excess. The gels
were run in 1* TBE at 200 V for 2-3 h, dried and autoradiographed using Kodak X-Omat AR films. The oligonucleotides used for binding were AGTTGCCAAA ATGGCGTCAT TTTTGCACGA CT and, as non-specific competitor, an oligonucleotide comprising a SP1 binding site, CTTGGTGGGG GCGGGGCCTA
AGCTG.
Standard binding reactions were pipetted onto Parafilmtm and exposed to UV light at 254 nm for 1 min at 4000 [mu]W/cm
2
using a Stratagenetm 2400 crosslinker. Samples were denatured and run on 10% SDS-PAGE. Gels were dried and autoradiographed using Kodak X-Omat AR films.
Previously we have shown that [beta]1 tubulin protein is present ubiquitously in the embryo and can be
detected using an anti-[beta]1 tubulin antibody in most tissues until hatching (
17
).
In situ
hybridization to embryo sections using
3
H-labeled probes indicated that the distribution of the mRNA, however, is
not homogeneous, but is present at higher levels in the CNS (
20
). By whole mount
in situ
hybridization using digoxygenin-labeled probes we extended these data (Fig.
1
). While maternal mRNA leads to a strong homogeneous staining of early embryos (Fig.
1
A), the mRNA outside the CNS decays to undetectable levels after stage 9, while in
the ventral neurogenic region three rows of cells show expression of the [beta]
1 tubulin
gene at stage 10 (Fig.
1
C). These cells represent neuroblasts exclusively, as double staining with an
antibody against a glial-specific marker, repo (
21
), reveals no overlapping patterns (data not shown). In addition, [beta]1 tubulin mRNA is detected in cells of the PNS. The mRNA levels in the CNS
remain high until stage 12, after which they start to decay. Beginning at stage
13, the attachments of the somatic musculature, the apodemes, show expression
of the [beta]
1 tubulin
gene. At stage 15, PNS and apodemes contain high levels of [beta]1 tubulin mRNA, while it is hardly detectable in the CNS. In contrast, the [beta]1 tubulin protein is present at high levels in these tissues at all
times (
17
).
During the deletion analysis of the [beta]
1 tubulin
gene intron we found that sequences required for high level expression in the
CNS reside within the first 270 bp of the intron (
17
). This conclusion was drawn from the differences in the expression patterns
between strains transgenic for W[beta]1K and W[beta]1C and the results with [beta]1-1. Sequence comparison with the [beta]
1 tubulin
gene of
D.hydei
revealed two blocks of homology in this region (Fig.
2
; Buttgereit, unpublished results), which were named IE1 and IE2, respectively.
IE1 shows 19 out of 20 bases identity, IE2 16 out of 16. The distance between the two blocks is different, with 13 bases in
D.melanogaster
and only four bases in
D.hydei
. Two closely related motifs are present in IE1: the sequence CAAAAT at +221 to
+225 and the sequence ATTTTTGC at +231 to + 238. The second motif occurs a
second time in IE2 on the complementary strand, from +261 to +268 (Fig.
2
). The palindromic orientation of the second motif is reminescent of specific
recognition elements for DNA binding proteins. To address the question whether
these two homologies indeed represent
cis
-regulatory elements, deletions in the 5'-part of the intron were made, keeping the bases to the splice
site and spacing of the elements as in the reference construct W[beta]1K (
17
). In construct [beta]1-1 (Fig.
3
), both elements were deleted, leaving only 49 bp of 5' intron sequence. As in comparison with W[beta]1K, which contains both IE1 and IE2, an additional 169 bp from +277
to +446 were deleted due to the cloning strategy used, a control construct, [beta]1-2, including both boxes but deleting the same 169 bp as in [beta]1-1, was analyzed. Both constructs were cloned in front of
a
lacZ
gene, inserted into the P-element vector pW8 and transgenic fly strains were established. Embryos were collected and analyzed for reporter gene expression using a polyclonal anti-[beta]-galactosidase antibody. Strains transformed with [beta]1-2 show strong maternal expression as well as
high level expression in the CNS, as does W[beta]1K (Fig.
3
D). This result shows that no essential elements for maternal or zygotic
expression during embryogenesis are located in the 169 bp removed. Deletion of
both boxes in [beta]1-1 (Fig.
3
C), however, results in a strong reduction of maternal expression and the number
of reporter gene-positive cells in the CNS is greatly reduced. This expression corresponds
to the pattern observed in strains transgenic for construct W[beta]1C (Fig.
3
B), which completely lacks the intron (
17
), indicating that IE1 and IE2 are the only sequences necessary for CNS and
maternal expression. These data furthermore prove that the bases at the 3'-end of the intron, present in W[beta]1K and [beta]1-1, for example, are not required for CNS-specific expression. However, in contrast
to W[beta]1K and W[beta]1C, expression of the reporter gene is observed in the PNS with both
[beta]1-1 and [beta]1-2. Thus, putative silencing elements acting in the PNS
may be present in the 169 bp removed. To exclude the possibility that the
deleted bases of IE1 and IE2 simply disturb the DNA topology in this region
leading to the reduction of enhancer activity observed, both elements were
mutagenized by shuffling the nucleotides randomly ([beta]1-9), keeping the number of bases constant. Analysis of transgenic fly
strains yielded the same result as with [beta]1-1, in that very weak maternal and extremely reduced CNS expression
were observed (Table
1
). Removal of the remaining bases 5' to IE1 up to the splice site ([beta]1-7) revealed no differences in the expression pattern compared
to [beta]1-2 (Table
1
), indicating that no further sequences are relevant in this region. Thus, IE1
and IE2 are the only
cis
-acting sequences identified in the 5' part of the intron.
To narrow down the activating sequences, both boxes were tested individually for
their potential neural system-specific activity. As for [beta]1-1 and [beta]1-2, synthetic oligonucleotides comprising either
IE1 and flanking sequences ([beta]1-3, Fig.
4
A) or IE2 and flanking sequences ([beta]1-4, Fig.
4
B) were used. Analysis of the [beta]1-3 transgenic strains show maternal as well as strong expression in
the CNS. Thus, the IE1 element is capable of driving both modes of expression
(Fig.
4
A). Strains harboring [beta]1-4 also show strong maternal expression, but significantly reduced
expression in the CNS, as does [beta]1-1 in which both boxes are deleted (compare Figs
4
B and
3
C). Thus, IE2 is not capable of enhancing expression in the CNS, but elevates
maternal transcription. As in IE1 the two motifs CAAAAT and ATTTTTGC are found,
while only the second motif is repeated in IE2, the palindromic structure of
the second element is not required for its function. Thus, either the first
motif in IE1 alone represents the neuronal enhancer or the imperfect palindrome
found in IE1 (CAAAT---N
6
---ATTTTTG) fulfils this function. No further regulatory
sequences are located in these elements, as mutagenesis of both motifs in IE1
and the motif in IE2 together in construct [beta]1-9 (Table
1
) completely abolishes maternal as well as CNS expression.
To further test the regulatory potential of the motifs in IE1, both were
individually mutagenized. In construct [beta]1-10, the first motif CAAAAT was mutagenized to AGCTCA. Lines carrying
this construct show intermediate levels of maternal expression, while
expression in the CNS is reduced to very low levels, as in [beta]1-1 (Fig.
4
C). Conversely, in construct [beta]1-11, the second motif ATTTTTGC was changed to ACGTCAGT. Strains
transgenic for [beta]1-11 reveal no maternal expression, while expression in the CNS is
comparable to [beta]1-3 (Fig.
4
D). These results show, although both motifs are very similar, that their
specificity
in vivo
is distinct, as the first motif exclusively drives neuronal expression, while the second motif, which is repeated in IE2, is restricted to the
enhancement of maternal expression. Therefore, the first motif of the IE1
element was named CNE (
To check a further criterion for enhancer function of the CNE, a synthetic
oligonucleotide comprising both IE1 and IE2 was cloned upstream, at -2348 bp, in the construct [beta]1-1, which lacks both IE1 and IE2 in the intron, resulting in construct [beta]1-8 (Fig.
5
). Expression monitored by anti-[beta]-galactosidase antibody staining showed no detectable
differences to [beta]1-2, demonstrating that the CNE can also exert its enhancer activity
when placed upstream of the promotor (Fig.
5
). However, in contrast to classical enhancers, its activating potential is
dependent on additional sequences of the [beta]
1 tubulin
gene promoter. This was shown by combination of the intron lacking only 0.15 kb
at the 3'-end with the heterologous hsp70 promoter (
17
; construct WHI), where no CNS-specific or maternal expression is observed (
17
). The 0.15 kb omitted in WHI themselves are not capable of driving expression in the CNS as
they are present in [beta]1-1, which definitively shows reduced expression, as does W[beta]1C.
Having characterized the enhancer function of the CNE, the identification of
proteins that specifically interact with this element became of interest.
Nuclear extracts from embryos staged at 6-14 h were prepared in order to reduce the amount of maternally retained components and tested for specific interaction with a double-stranded oligonucleotide comprising the CNE and ME elements (Fig.
6
A). Several bands were detected, but competition experiments revealed that only
one specific retarded signal, complex 1, resulted from specific binding to IE1.
The second complex is only competed out in the presence of both IE1 and IE2,
indicating that it contains different proteins. When the CNE was mutagenized
according to the strategy used for [beta]1-7, no specific binding was observed (not shown). These results
demonstrate that the core motifs identified indeed represent specific DNA
binding elements. As both motifs of IE1 are very similar in their sequence, in
addition to the gel retardation assays the interaction of proteins with these
motifs was analyzed in UV cross-linking experiments. Standard binding reactions using the double-stranded oligonucleotide Mater (see Materials and Methods) were
exposed to UV light and analyzed by 10% SDS-PAGE. A specific interaction of a 71 kDa protein was exclusively observed with the oligonucleotide Mater (Fig.
6
B).
The [beta]
1 tubulin
gene of
D.melanogaster
is transcribed during oogenesis as well as zygotically in the nervous system
and the attachments of the somatic musculature, the apodemes. The gene is
subject of a complex network of transcriptional regulation during embryogenesis. Deletion analysis revealed separate elements driving expression in the CNS, the chordotonal organs and the attachment sites
of the somatic musculature (
17
,
22
). While the intron elements necessary for expression in the chordotonal organs
and apodemes also act independently of the [beta]
1 tubulin
gene promoter (
17
,
22
), the situation for expression in the CNS is different. We could show that for
complete expression at high levels at least three different modules have to
interact: upstream sequences between -2348 and -1136, promoter-proximal elements and sequences from the 5'-part of the intron. None of these modules alone
is capable of interacting with a heterologous promoter to drive expression in
the CNS.
By deletion analysis we could show that the intron of the gene is essentiel for
maternal as well as neuronal expression (
17
). Sequence comparison with the [beta]
1 tubulin
gene from the distantly related species
D.hydei
revealed the presence of two conserved sequence blocks in the proximal 5'-part of the intron. The elements were termed IE1 (20 bp) and IE2
(14 bp). In this report we present evidence for the neural enhancer activity of
a 6 bp core element localized in IE1. Zygotically, this enhancer is specific
for the neural system, because no transcription is observed in apodemes. In
total, two separate motifs were detected in IE1: the 5'-located core of 6 bp, CAAAAT, which is essential for neuronal
expression in the CNS, and a 3'-located 8 bp motif, ATTTTTGC, which occurs a second time in the
inverted orientation in IE2 and is required for maternal expression. Mutational
analysis reveals that distortion of the 5' core sequence reduces expression in the CNS, as does complete deletion
of IE1. Therefore, this sequence was termed the
Having defined the CNE
in vivo
, we were interested in identifying transactivating proteins that might bind to
sequences in IE1.
In vitro
binding studies using EMSA assays reveal the presence of nuclear proteins
interacting with this element. A problem with these studies arises from the
close relationship of the CNE and ME. As their relative affinities for
interacting factors
in vitro
may be very similar, although
in vivo
sufficient for specific regulation, the identification of such factors in gel
retardation assays may be quite difficult. Possible cofactors which are
relevant
in vivo
may not function equivalently
in vitro
, or only with reduced efficiency. In UV cross-linking experiments, a 71 kDa protein showed a specific interaction with
the CNE.
Concerning possible transactivators, only a few candidates are known. The
products of the pro-neural gene class, as well as the E(spl) and HLH-5m proteins, which represent bHLH transcription factors, recognize a consensus motif CANNTG (E-box) or CACNAG (N-box), which is not found in the CNE, only the variant
CAAAATG at the 5'-end of the motif. Analysis of the [beta]1 tubulin mRNA distribution in a mutant for a pro-neural gene belonging to the bHLH family,
daughterless
, revealed no significant reduction in CNS expression (Buttgereit, data not
shown). Thus, the bHLH genes are unlikely to be involved in direct activation
of [beta]
1 tubulin
gene expression. However, due to the high maternal mRNA level in early
embryogenesis stages, which overlaps with the onset of zygotic expression, this
observation remains preliminary.
For two other genes, elements for expression throughout the CNS have been
mapped. In the case of the
elav
gene, a 333 bp promoter fragment could be identified which is capable of
driving high level expression in the CNS (
15
). This is different from the situation in the [beta]
1 tubulin
gene, where the interaction of several widely dispersed, distinct motifs are
essential for activation. For the
snail
gene it was shown (
1
6
) that activation in the CNS, at least, is not dependent on the pathway
initiated by the genes of the
achaete
-
scute
complex. In addition, elements for the CNS and PNS could be separated. Sequence
comparison of the IE1 and IE2 elements revealed no homologies to either gene,
indicating that the CNE represents a new DNA binding motif. Further analysis of
the CNE and the specifically interacting proteins may yield insight into the
regulatory cascades leading to the differentiation of the embryonic
D.melanogaster
CNS. So, for the first time, the differential activity of two closely related
enhancer motifs active either during oogenesis or embryogenesis has been
demonstrated.
This work was supported by the Deutsche Forschungsgemeinschaft (Bu816/1-2). Special thanks to Sabina Kerl for excellent technical assistance and
to A.Paululat and E.Kothe for critical reading of the manuscript. The work was
done in the laboratory of Renate Renkawitz-Pohl in Marburg. Her permant readiness for critical discussion was very
stimulating for the progress of the experiments.
+
Present address: MLP Finanzdienstleistungen, Seidlstraße 4, 80335 München, Germany
REFERENCES
Return