Proc. Nati. Acad. Sci. USA
Vol. 89, pp. 683-687, January 1992
Cloning of Drosophila transcription factor Adf-1 reveals homology
to Myb oncoproteins
BRUCE P. ENGLAND*, ARIE ADMON, AND ROBERT TJIAN
Howard Hughes Medical Institute, Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720
Contributed by Robert Tjian, September 19, 1991
ing protein, Adf-1, is capable of activating transcription of the
alcohol dehydrogenase gene, Adh, and is implicated in the
transcriptional control of other developmentally regulated
genes. We have cloned thecDNA encoding Adf-1 by generating
specific DNA probes deduced from partial amino acid sequence
ofthe protein. Several cDNA clones encoding an extended open
reading frame were isolated from a phage A library. The
complete amino acid sequence of Adf-1 deduced from the
longest cDNA reveals structural similarities to the putative
helix-turn-helix DNA binding motif of Myb and Myb-related
proteins. DNA sequence analysis ofgenomicclones and North-
ern blot analysis of mRNA suggest that Adf-1 is a single-copy
gene encoding a 1.9-kb transcript. Purified recombinant Adf-1
expressed in Escherichia coli binds specifically to Adf-1 recog-
nition sites and activates transcription of a synthetic Adh
promoter in vitro in a manner indistinguishable from the
protein purified from Drosophila. Temporally staged Drosoph-
ila embryos immunochemically stained with affinity-purified
anti-Adf-1 antibodies indicate that Adf-1 protein is not detect-
able in very early embryos and does not appear to be mater-
nally inherited. During later stages of embryogenesis, Adf-1
appears to be expressed in the nucleus of most somatic cells in
the embryo with possibly higher concentrations found in some
The Drosophila sequence-specific DNA bind-
Recent advances in the biochemical and genetic analysis of
Drosophila melanogaster gene regulation have established
that transcriptional initiation is a primary mechanism con-
trolling patterns ofgene expression during development and
cellular differentiation. The cis regulatory DNA elements and
trans-activating proteins of many Drosophila genes have
subsequently been identified. One class of transcriptional
regulators, the sequence-specific DNA binding factors, have
been of particular interest because they are responsible in
large measure for governing the temporally programmed and
spatially restricted patterns of gene expression during Dro-
sophila embryogenesis (1). For example, the Drosophila
alcohol dehydrogenase gene, Adh, is transcribed in a tissue-
specific and temporally regulated manner during the life cycle
ofthe fly (2). The expression ofAdh in Drosophila is directed
in an unusual and complex manner by two tandemly arrayed
promoters, distal and proximal, that are 700 base pairs (bp)
apart. These two promoters are activated in different cell
types and are subject to different temporal programs (3, 4).
The cis-controlling DNA elements responsible for mediating
temporal and tissue-specific expression of Adh have been
mapped, both by introducing altered gene sequences into
Drosophila by P-element-mediated transformation (5-8) and
by transcription of mutant templates either in vitro or fol-
lowing transfection into Drosophila cells (9-11). A complex
assortment of distinct promoter and enhancer elements was
found to be required for regulated transcriptional initiation
from the distal and proximal Adh promoters.
In vitro transcription and DNA binding studies using
extracts derived from temporally staged Drosophila embryos
or established tissue culture cell lines have identified multiple
site-specific transcription factors that recognize and bind to
the cis-controlling elements of the Adh promoter (9, 12).
These studies have revealed that one of these transcription
factors, Adf-1, binds to and activates transcription from a
specific upstream recognition site in the distal Adh promoter.
Recently, we succeeded in purifying Adf-1 to homogeneity
and found that the purified protein also binds specifically to
DNA sequences in the promoters of several other develop-
mentally regulated Drosophila genes, including the Anten-
napedia (Antp) P1 promoter and the dopa decarboxylase gene
promoter (10). The next step toward achieving a more
comprehensive biochemical and molecular characterization
ofAdf-1 required cloning the gene encoding Adf-1. Here, we
report the isolation and sequencet of the gene encoding
Drosophila melanogaster Adf-1. In addition, we have raised
antibodies against this transcription factor and determined
the temporal and spatial patterns of Adf-1 expression during
different stages of Drosophila embryogenesis.
MATERIALS AND METHODS
HPLC Purification of Adf-1 and Peptide Sequencing. Adf-1
was affinity-purified from Drosophila embryos as described
(10). An estimated 2.4 gg of affinity-purified Adf-1 was
alkylated with 4-vinylpyridine and digested with trypsin, and
the partial amino acid sequence of several tryptic peptides
was determined-all as described (13).
Isolation and Analysis of Genomic and cDNA Clones. Oli-
godeoxynucleotide probes designed according to consider-
ations described previously (14, 15) using Drosophila codon
preferences (16) were prepared from the sequences of pep-
tides 4 and 5 (see Fig. 1). The sequence of probe 1 (peptide
4) was 5'-GTACTTGTTGTGCTGGGCCAGCAGCTGCAT-
GTCGGTCAGGTA-3' and the sequence of probe 2 (peptide
5) was 5'-CTGGTCCTCGGCIGAIACGGCCTGIGAI-
ACGTTGTTCTGGAAGAT-3' (I represents inosine). These
probes were used to screen a Drosophila genomic library in
AFIX (Stratagene). Two stringent washes were with 0.30 M
NaCl/0.03 M Tris, pH 7.5/5 mMEDTA/0.5% SDS for 30min
at 58°C (probe 1) or 54°C (probe 2).
To isolate Adf-1 cDNA clones, a high-specific-activity
single-stranded DNA probe was prepared from a subclone of
genomic clone 1. This probe was used to screen a phage Agtll
cDNA library prepared from mRNA isolated from 9- to
12-hr-old Drosophila embryos (17). The three largest clones
obtained (numbers 4, 19S, and 22) were cloned in both
Abbreviation: Adh, alcohol dehydrogenase.
*Present address: Affymax Research Institute, 4001 Miranda Ave.,
Palo Alto, CA 94304.
tThe sequence reported in this paper has been deposited in the
GenBank data base under the name DROADF1A (accession no.
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Proc. Natl. Acad. Sci. USA 89 (1992)
GSSACiGAGCTGGGACGTArCGTTA CCGTTGGCAGAGACGCGACTGAGAAAIAAAAI IAAAACGTCiGA
AAACAGCAGAGCGTGCGTTCGCGCCLAAATACTTAACAACAATTAGCAAACCG'AAGAAGT.AAAGTGAGTCTCG' LGGL C GuAGGAG"
AC-'CGCT;CG, . C C CC G T GAGAA
-C - T C... C *C AG
IT C AAG'C'LA
! A. .
C-. 3, ,^sSA- C,AT
E4 P> P
ASS GTG C AGA4 CA
; ; ;
SACS C GA
A7 CC AAC
IC CAA AAC AAC
AAG 'TAT' AA -A-
CA G ACACATCAC
GAGAACAGiAC CCAC'GGCIAC'CCA C'AT C CACAT CTACUACGGGT T TGGG."C
TAGACG, AACATCCCACAGCGTAAACCATAACGCGTAATGATAUTUCCbACG"~ CATGTAGACAAA4"A ATA'A~TTATAT.
C -'iiG kj ,
O C A
.T"-7A;'_-. -. A7.
GGA1'ACGAGTAAAATCGAAATAG-TCCGGCAACGGCSAG"GTTGSTAAACA'T'iTAAT - AT-AGTTGTGCAAT-.A
1ATT GTAAATTTT GACACGATT TAGTT GTAATTIG LC,AT G uG T f, T'ArtCCTI-TAAAAC-TAA GT
hlA4A l" <1TAAAA T -T''h'-
the consensus sequence is shown. The deduced Adf-1 peptide sequenceis shown below the DNAsequence.Adf-1tryptic peptidesare indicated
cDNA and amino acid sequence of Adf-1. The complete insert of phage A cDNA clone 19-S was sequencedon both strands, and
orientations into the EcoRI site ofpBluescript SK(+) (Strat-
agene), and the DNA sequences of the ends of these clones
were determined by using Sequenase (United States Bio-
chemical). The complete sequence of clone 19S was deter-
mined from a 5' and a 3' series of nested deletions prepared
with exonuclease III (18).
Expression of Adf-1 in Escherichia coli. As described pre-
viously for two other Drosophila transcription factors (13),
the coding portion ofAdf-1 cDNA clone 19S was placedin the
T7 RNA polymerase expression vector pET-3a (19), and the
resulting plasmid, pET-3a/Adf-1, was transformed into E.
coli strain BL21 (20). Adf-1 expression was induced, and
bacterial protein extracts were prepared as described (13).
Adf-1 was purified by passing the bacterial extract directly
over an Adf-1 affinity resin prepared and run as described
(10). DNase I footprinting reactions and in vitrotranscription
reactions with this purified recombinant Adf-1 were per-
formed as described (10).
Antibody Studies. Rabbits were immunized by subcutane-
ous injection of 100ggof recombinant Adf-1 suspended in
500 1ul of phosphate-buffered saline (PBS) and 500Al of
Freund's complete adjuvant. Rabbits were giventhree boost-
ers of 100ggof Adf-1 in incomplete adjuvant at 3-week
ifltervals. Antisera used for the experiments shown were
obtained 1 or 2 weeks after administration of the third
booster. To affinity-purify Adf-1 antibodies, 100Agof affin-
ity-purified, recombinant Adf-1 was coupled to 600 ,ud of
Affi-Gel 10 (Bio-Rad) according to the manufacturer's rec-
ommendations. Adf-1 antiserum Was diluted 3-fold with PBS,
applied to the affinity resin, and washed with PBS. Bound
antibodies were eluted with 100 mM glycine (pH 2.3) and
neutralized immediately after elution with 1/10th volume of
2 M Tris (pH 8.0). Immunohistochemical staining of whole
Drosophila embryos was performed as described (21).
RESULTS AND DISCUSSION
Isolation and Sequencing of Adf-i Genomic and cDNA
Clones. To clone the gene and cDNA coding for Adf-1, we
first determined the amino acid sequences of several Adf-1
tryptic peptides. The amino acid sequences of tryptic pep-
tides 4 and 5(see Fig. 1)were chosen togeneratethesynthetic
DNA probes used to screen aDrosophila genomic library.
Our initial screen identified twophageA clones that hybrid-
ized strongly to both Adf-1 probes. Partial DNAsequence
analysis confirmed that both of thegenomicclones isolated
contained thecoding sequencesfor the twopeptidesthat had
been used to generate the probes as well as the coding
sequence for peptide 1.
To isolate Adf-1 cDNAclones,aDNAprobewasprepared
from afragmentofgenomicclone 1 and used to screen aAgtll
cDNA library. We isolated eight cDNA clones that, by
restrictionanalysis,allappearedto be derived from the same
message (datanotshown). Sequence analysisof the ends of
the three longest cDNAclones.confirmed that they were
derived from the sametranscript.
A Northern blot ofembryomRNAhybridizedwith aprobe
made from thegenomicclone reveals asingleAdf-1message
of ':1.9 kilobases (kb) (datanot shown), and thelengthsof
seven of the eightcDNA clones isolated are in therangeof
1.5-1.8 kb. Therefore, it seemed likely that the largestof
these clones, at -1.8 kb, was anearly complete copyof the
Adf-1 mRNA. The complete DNA sequenceof this clone,
19S, was determined and is shown in Fig. 1. This cDNA
includes a long open readingframe that, when translated,
contains thesequencesofall ofthe Adf-1tryptic peptidesthat
had been determinedby peptide sequencing,thusproviding
strong evidence that we have indeed cloned the cDNA for
Adf-1. Between peptide 2, the most N-terminal of these
peptides,and the first in-frameupstream stopcodon(nucle-
otides 256-258) there isonly one methionine codon, which
we have designated as the initiation codon of the Adf-1
protein. Thepolypeptide shown has a calculated molecular
weight of 29.2 kDa, which is in close accord with Adf-l's
apparent molecular mass of 34 kDa based on SDS/PAGE
(see Fig. 3).
Thecytologicallocus oftheAdf-J genewas determinedby
hybridizing a biotinylated Adf-l genomic probeto salivary
gland polytene chromosome squashes and visualizingthe
annealedprobe by biotin-specifichistochemicalstaining (22).
The Adf-l clone hybridized to a single cytological locus,
Proc. Natl. Acad. Sci. USA 89 (1992)
42C2-7 (data not shown). No previously identified Drosoph-
ila genes have been mapped to this interval (23).
Comparison of the Adf-1 sequence with available se-
quences in the GenBank and National Biomedical Research
Foundation-Protein Identification Resource databases re-
vealed potentially interesting similarities between Adf-1 and
other proteins. First, the peptide sequence between amino
acid residues 44 and 63 has some homology to a bacterial
helix-turn-helix DNA-binding domain. Although this homol-
ogy appears somewhat weak, when it is analyzed by a
statistical method designed to evaluate the similarity of any
protein sequence to the phage A Cro type helix-turn-helix
domain (24), we find that this segment of Adf-1 is rated as
more similar to the Cro-type binding domain than are the
homeodomains of the products of Ubx, en, and ftz. These
Drosophila homeodomains form a helix-turn-helix-type
structure (25, 26) but show almost no similarity to the
potential helix-turn-helix domain of Adf-1. Second, it has
been proposed that the DNA-binding domains of the Myb
oncoproteins and other related proteins also adopt a helix-
turn-helix tertiary structure (27). In contrast to the homeo-
domain, the proposed Myb helix-turn-helix domain shows
significant similarities to the potential helix-turn-helix do-
main of Adf-1 (Fig. 2). Although a gap must be inserted
between Adf-1 residues Leu-52 and Gly-53 to optimize the
alignment of Adf-1 with the Myb sequences, this same gap
must be introduced into both the bacterial helix-turn-helix
consensus and the homeodomain consensus to align them
with Myb sequences (27). It is interesting to note that the best
alignment was observed between Adf-1 and REB1, a Myb-
related yeast transcription factor involved in transcription of
rRNA (30). Although the amount of amino acid identity
between Adf-1 and the prokaryotic helix-turn-helix proteins
and the Myb family ofproteins is somewhat limited, it occurs
primarily at highly conserved positions that have been iden-
tified as important for maintaining a helix-turn-helix struc-
ture by virtue of hydrophobic interactions with the protein
Another noteworthy feature ofthe Adf-1 protein sequence
is the large number of glutamine residues in the protein
(11.1%) and the presence ofa polyglutamine run interrupted
by a single alanine residue from amino acid 120 through
amino acid 127. Glutamine-rich protein segments have been
shown to direct transcriptional activation by the human
transcription factors Spl and Oct-2 (31, 32). Mutational
analysis will be required to test whether the glutamine-rich
sequence of Adf-1 serves the same function.
Q TrWK Q IE
v P E QKa]TK
K S L R D K F
DNA Binding and Transcription Properties of Recombinant
Adf-1. To produce large amounts of Adf-1 for biochemical
studies, we expressed the Adf-1 gene in E. coli using a
bacteriophage 17 RNA polymerase expression system (20).
Upon induction with isopropyl f3-D-thiogalactoside, a new
polypeptide of -34 kDa appeared in crude lysates of cells
carrying the Adf-1 expression plasmid but not in control
lysates from cells bearing the vector alone. Crude extracts
from the induced bacteria were passed directly over an Adf-1
DNA-binding affinity column and were eluted with a high-salt
buffer, yielding a single polypeptide with a mobility on
SDS/PAGE identical to that of Drosophila Adf-1 (Fig. 3A).
The purified recombinant Adf-1 was tested for the char-
acteristic Adf-1 DNA-binding activity in a DNase I footprint
assay and was found to bind to theAdh distal promoter Adf-1
site with a pattern identical to that of the native Drosophila
protein (Fig. 3B). Since recombinant Adf-1 appeared to bind
to its recognition element with the same specificity as the
authentic Drosophila protein, we next tested its ability to
activate transcription in an Adf-1 site-dependent manner.
Using an Adf-1-depleted extract described (10), in vitro
transcription reactions were carried out with DNA templates
containing one or three copies of an Adf-1 binding site.
Addition of recombinant Adf-1 stimulated in vitro transcrip-
tion to approximately the same extent (5- to 8-fold) as was
previously observed with Drosophila Adf-1 (Fig. 4). These
results establish that the cDNA clone we isolated codes for
an Adf-J product that is active for both DNA binding and
promoter selective transcription. The demonstration that
purified recombinant Adf-1 has the same biochemical prop-
erties as the purified Drosophila protein confirms the finding
that the 34-kDa Adf-1 protein is the essential component
required for Adf-1 site-dependent transcriptional activation.
- <- Adf-l
( 13 920
13 9 25 0
W A E I G K T L
KM W v
I A R Y L N G R
K RW S V I A K H L K G R I G K Q C R
N Q V A K I A K R L P G R
N R W A E I A K L L P G R I
- G R M P X D C R D R W R N Y V K C G
I I G P H F K D R L E Q Q V Q Q R V A K V L N P
Kt V I A N Y L P N R -0 D V Q C Q
- G K Q C R E R W H N H
R It1 Q K V L N P _
L7 N P
RW H N H L N P E
K N HWNS
N H W N S
I D N A
M K K
. M R R K
repeats of Myb proteins and the Myb-related region of yeast tran-
scription factor REB-1. The Adf-1 sequence from amino acid residue
42 to residue 67 is shown aligned with parts of the repeats of
Drosophila c-Myb (Dm) (28) and mouse c-Myb (MO) (29) and to
yeast REB1 (30). Residues identical to the corresponding Adf-1
residue are indicated by shading. Boxed residues in the Adf-1
sequence indicate amino acid identity with the majority of the
sequences shown. Dashes indicate gaps inserted into the sequences
to optimize alignment. The numbers 1, 2, or 3 after the Myb protein
abbreviations indicate first, second, or third Myb repeat, respec-
tively. The helix-turn-helix motifshown is that previously proposed
for Myb proteins (27).
Comparison of Adf-1 with portions of the DNA-binding
amide gel comparing purified E. coli-expressed Adf-1 with Adf-1
purified from Drosophila embryos. Lanes: 1, molecular weight
standards, 100 ng of protein per band; 2, 5 1.l of affinity-purified
embryo Adf-1 estimated at 2 DNase I footprint units/hl; 3 and 4, 1
and 5 Al of a fraction of affinity-purified embryo Adf-1 of undeter-
mined activity; 5, 1Alof affinity-purified E. coli-expressed Adf-1
estimated at 20 DNase I footprintunits/Al.(B) DNase I footprints of
purified E. coli expressed Adf-1 and purified Drosophila embryo
Adf-1. Protein samples shown in lanes 2 and 5 ofA were diluted by
factors of 50 and 500, respectively (to estimated concentrations of
0.32 and 0.14 pAg/ml, respectively), and assayed for Adf-1-
characteristic footprinting on the Adh distal promoter. Lanes: 1, 6,
and 11, DNase I digestion pattern in the absence ofadded Adf-1; 2-5,
DNase I footprint due to increasing amounts of E. coli-expressed
Adf-1, as indicated at the top; 7-10, DNase I footprint due to
increasing amounts of embryo Adf-1, as indicated at the top.
Expression of Adf-1 in E. coli. (A) SDS/8% polyacryl-
Proc. Natl. Acad. Sci. USA 89 (1992)
I... LC0li A\dF-'.
:1 Internal Control
Adf-1. Twenty-five-microliter transcription reaction mixtures con-
tained 50 ng (33 fmol) ofthe indicated distal promoter template DNA,
200 ng of pADHP.a5'-55 internal control DNA, 9.5 ul of 3x
Adf-1-depleted embryo nuclear extract (20 mg ofprotein per ml), and
purified E. coli-expressed Adf-1 as indicated. The concentration of
the Adf-1 used was estimated to be 7
diagrams indicate sequences protected by Adf-1 in DNase I footprint
reactions. Lanes: 1-4, bands marked "Distal Transcripts" are due to
transcription from the -46 distal promoter template in the presence
of0, 1, 2, and 3
transcription from the -86 template in the presence ofthe same four
amounts of Adf-1; 9-12, transcription from the -86+2 template
(which contains two inserted Adf-1 binding sites in addition to the
normal distal promoter site) in the presence ofthe same fouramounts
ofAdf-1. Internal control bands in lanes 1-9 are due to transcription
from the pADHP.A5'-55 internal control promoter included in each
In vitro transcriptional activation by E. coli-expressed
Lg/ml. Ovals in template
p1A of Adf-1 as indicated at the top of the lanes; 5-8,
Antibody Studies of Adf-1 Expression During Embryogen-
esis. Earlier work suggested that the level of Adf-1 activity
during embryogenesis might be correlated with the transient
period of Adh distal promoter activity observed midway
through embryogenesis (12). Further work has shown that
this embryonic Adh distal promoter activity is restricted to
the developing larval fat body (3). Therefore, we were
interested in obtaining antibodies directed against Adf-1 to
study both the levels of Adf-1 during Drosophila embryo-
genesis and the embryonic tissues in which it is expressed.
Rabbits were immunized with SDS/PAGE-purified recom-
binant Adf-1, and antibodies ofhigh titer and specificity were
obtained (data not shown). To further increase the specificity
of these antibodies, they were affinity-purified by passage
over a resin of purified Adf-1 coupled to agarose beads. The
afflinity-purified anti-Adf-1 antibodies were shown by West-
ern blotting to still bind specifically to Adf-1 when present in
a crude nuclear extract from Drosophila Kccells (data not
To examine the spatial pattern of Adf-1 expression during
embryogenesis, we immunostained a collection of whole
Drosophila embryos from 0 to 14 hr old. Fig. SA shows the
pattern ofAdf-1 expression over the time course ofembryo-
genesis from the cellular blastoderm stage, =z2.5 hr after
fertilization (embryo 1), to the stage just prior to dorsal
closure, -11 hr after fertilization (embryo 7). Adf-1 appears
to be distributed in most cells of these embryos with the
genesis. Embryos (0-14 hr old) were dechorionated, fixed, devitel-
linized, and then incubated with a 1:4000 dilution of affinity-purified
anti-Adf-1 antibodies followed by 3,3'-diaminobenzidine staining.
(A) Time course of Adf-1 staining. Individual embryos from repre-
sentative stages are shown. All embryos (with the possible exception
ofembryo 1) are oriented with the posterior end toward the bottom
and the ventral side to the left. Embryos: 1, embryo at stage 5 of
Campos-Ortega and Hartenstein (33),
stage 6-7, -3 hr afterfertilization; 3, stage 9, -4 hr afterfertilization;
4, stage 10, -5 hr after fertilization; 5, stage 12, -8.5 hr after
fertilization; 6, stage 13, -10 hr after fertilization; 7, stage 14, -11
hr after fertilization. (B) Optical section through gastrulating embryo
(stage 6), -3 hr after fertilization. The embryo is oriented with the
posterior to the left and ventral surface to the bottom. Nuclear
localization ofAdf-1 is most clearly seen along the ventral side. VN,
possible vitellophage nucleus staining positively for Adf-1; PC, pole
cells unstained for Adf-1. (C) Optical section through a germ band-
extended embryo (stage 9), U5 hr after fertilization. Posterior end is
to the right, and the ventral surface is to the bottom. Me, mesoderm;
Ec, ectoderm; AM, anterior midgut primordium; PM, posterior
Immunohistochemical staining of Adf-1 during embryo-
2.5 hr after fertilization; 2,
possible exception of the amnioserosa (unstained, dorsally
located region in embryos 5, 6, and 7). However, since Adf-1
Proc. Natl. Acad. Sci. USA 89 (1992)
is localized in the nucleus (as shown below), this lack of
staining may not reflect an absence of Adf-1 in the amnio-
serosa but rather the relative paucity of nuclei in this thin
monolayer. The increase in stain intensity observed with
embryos of increasing age may be due to an increase in the
intranuclear concentration of Adf-1, to cellular proliferation
or to a combination of these two factors. However, the
intranuclear concentration of Adf-1 was observed to defi-
nitely increase over the time between the syncytial blasto-
derm stage and gastrulation (data not shown).
Examination at higher magnification of Adf-1-stained em-
bryos from the gastrulation [stage 6 of Campos-Ortega and
Hartenstein (33)] and early germ-band extended (stage 9)
stages confirms the observation that Adf-1 is distributed
among almost all embryonic cells (Fig. 5 B and C). The
nuclear localization of Adf-1 is most apparent along the
ventral side of the optical sagittal section of the gastrulating
embryo (Fig. 5B). The weak staining scattered throughout the
yolk in this figure (indicated by an arrow) is probably due to
Adf-1 in the vitellophage nuclei. The only nuclei that do not
appear to contain appreciable Adf-1 at this stage are those in
the pole cells, the germ-line primordia. Fig. SC shows the
pattern ofAdf-1 staining after the formation ofthe three germ
layers: ectoderm, mesoderm, and endoderm. Again, Adf-1 is
present in all three cell types in comparable amounts. The
striking double stripe pattern of Adf-1 staining most clearly
seen along the ventral side is due to the regular arrangement
of the nuclei of the ectoderm on the exterior and the nuclei
of the mesoderm on the interior of the embryo. The anterior
and posterior midgut primordia, indicated by arrows, are
endodermal cells and also stain strongly for Adf-1.
TheAdh distal promoter is active in embryos from 10 to 15
hr after fertilization, and this activity is restricted to the
developing larval fat body (3). In contrast, the antibody data
shows that Adf-1 is expressed at apparently constant levels
from around 4 or 5 hr after fertilization onward, and this
expression is distributed throughout the developing embryo.
Therefore, it is unlikely that Adf-1 is the only factor required
for the pattern of Adh distal transcription observed during
embryogenesis. This conclusion is supported by the fact that
Adh genes lacking the distal promoter Adf-1 site are never-
theless transcribed according to the correct temporal pro-
gram during embryogenesis (8). However, these mutant Adh
genes lacking the Adf-1 site are expressed at a much lower
level than wild-type, demonstrating that Adf-1 plays an
important role in regulating the level of Adh transcription.
We cannot at present explain the difference between the
quantitative changes in Adf-1 footprinting activity during
embryogenesis observed previously (12) and the more con-
stant levels of Adf-1 protein observed over the same time
period. This difference could reflect the presence ofan activity
in early and late embryo extracts that modulates the binding of
Adf-1 to DNA. Alternatively, there could be a temporally
programmed change in the affinity ofAdf-1 for its binding site,
controlled by some means such as protein modification.
Since Adf-1 binds to and may regulate transcription from
the dopa decarboxylase gene promoter and the Antennapedia
gene P1 promoter as well as theAdh distal promoter, it is not
surprising that Adf-1 is found in many cells ofthe developing
embryo. After all, these three genes have very different
spatial and temporal patterns of embryonic expression. It
remains to be determined just how widespread Adf-1-
regulated promoters are in the Drosophila genome and what
role, ifany, Adf-1 plays in the orchestration ofdevelopmental
gene expression. Perhaps Adf-1 acts in a manner analogous
to the mammalian factor, Spl, which is found to be essential
for directing transcription ofmany genes, including cell-type
specific genes, but is not itselfcell-type specific. Instead, Spl
appears to be a ubiquitous transcription factor that can act in
conjunction with other tissue-specific enhancer proteins to
specify unique transcription programs.
The best test of the role of Adf-1 in developmentally
regulated gene expression would be to assess the transcrip-
tional effect ofmutations in the Adf-1 gene. Such mutants, in
conjunction with the Adf-1 clones, antibodies, and purified
protein already available, and the powerful genetic and
biochemical techniques available to Drosophila biologists,
would make Adf-1 a useful model for addressing important
questions regarding eukaryotic transcriptional regulation.
We thank G. Dailey and U. Heberlein for assistance with the
purification of sufficient Adf-1 protein for sequencing; T. Laverty for
performing the chromosome in situ hybridizations; N. Patelforadvice
on embryo staining and microscopy; K. Moses and K. Zinn for
providingphage A libraries; J. Heilig forembryo mRNA; G. Peterson,
T. Hoey, and C. Hart forcomments on the manuscript; and K. Ronan
forhelpwiththe manuscriptpreparation. Thisworkwasfunded in part
by a grant to R.T. from the National Institutes of Health.
Biggin, M. D. & Tjian, R. (1989) Trends Genet. 5, 377-383.
Sofer, W. & Martin, P. F. (1987) Annu. Rev. Genet. 21,
Lockett, T. J. & Ashburner, M. (1989) Dev. Biol. 134, 430-437.
Savakis, C., Ashburner, M. & Willis, J. H. (1986) Dev. Biol.
Fischer, J. & Maniatis, T. (1988) Cell 53, 451-461.
Corbin, V. & Maniatis, T. (1990) Genetics 124, 637-646.
Corbin, V. & Maniatis, T. (1989) Genes Dev. 3, 2191-2200.
Heberlein, U. A. (1987) Dissertation (University of California,
Heberlein, U., England, B. & Tjian, R. (1985) Cell 41, 965-977.
England, B. P., Heberlein, U. & Tjian, R. (1990)J. Biol. Chem.
Benyajati, C., Ayer, S., McKeon, J., Ewel, A. & Huang, J.
(1987) Nucleic Acids Res. 15, 7903-7920.
Heberlein, U. & Tjian, R. (1988) Nature (London) 331, 410-
Perkins, K. K., Admon, A., Patel, N. & Tjian, R. (1990) Genes
Dev. 4, 822-834.
Lathe, R. (1985) J. Mol. Biol. 183, 1-12.
Martin, F. H. & Castro, M. M. (1985) Nucleic Acids Res. 13,
Streck, R. D., MacGaffey, J. E. & Beckendorf, S. K. (1986)
EMBO J. 5, 3615-3623.
Zinn, K., McAllister, L. & Goodman, C. S. (1988) Cell 53,
Henikoff, S. (1987) Methods Enzymol. 155, 156-165.
Rosenberg, A. H., Lade, B. N., Chui, D. S., Lin, S. W.,
Dunn, J. J. & Studier, F. W. (1987) Gene 56, 125-135.
Studier, F. W. & Moffatt, B. A. (1986) J. Mol. Biol. 189,
Patel, N. H., Martin-Blanco, E., Coleman, K. G., Poole, S. J.,
Ellis, M. C., Kornberg, T. B. & Goodman, C. S. (1989) Cell58,
Shapiro, R. A., Wakimoto, B. T., Subers, E. M. & Nathanson,
N. M. (1989) Proc. Natl. Acad. Sci. USA 86, 9039-9043.
Cockburn, A. (1986) Drosoph. Inf. Serv. 64, 89-158.
Dodd, I. B. & Egan, J. B. (1987) J. Mol. Biol. 194, 557-564.
Qian, Y. Q., Billeter, M., Otting, G., Muller, M., Gehring,
W. J. & Wuthrich, K. (1989) Cell 59, 573-580.
Kissinger, C. R., Liu, B. S., Martin-Blanco, E., Kornberg,
T. B. & Pabo, C. 0. (1990) Cell 63, 579-590.
Frampton, J., Leutz, A., Gibson, T. J. & Graf, T. (1989) Nature
(London) 342, 134 (lett.).
Katzen, A. L., Kornberg, T. B. & Bishop, J. M. (1985) Cell41,
Gonda, T. J., Gough, N. M., Dunn, A. R. & de Blaquiere, J.
(1985) EMBO J. 4, 2003-2008.
Ju, Q., Morrow, B. E. & Warner, J. R. (1990) Mol. Cell. Biol.
Courey, A. J., Holtzman, D. A., Jackson, S. P. & Tjian, R.
(1989) Cell 59, 827-836.
32. Tanaka, M. & Herr, W. (1990) Cell 60, 375-386.
Campos-Ortega, J. A. & Hartenstein,V.(1985)TheEmbryonic
Development ofDrosophila Melanogaster (Springer, Berlin).