2995
Development 125, 2995-3003 (1998)
Printed in Great Britain © The Company of Biologists Limited 1998
DEV5181
Cytoplasmic polyadenylation of Toll mRNA is required for dorsal-ventral
patterning in Drosophila embryogenesis
Jennifer A. Schisa and Sidney Strickland*
Department of Pharmacology and Program in Genetics, University at Stony Brook, Stony Brook, NY 11794-8651, USA
*Author for correspondence (e-mail: [email protected])
Accepted 18 May; published on WWW 9 July 1998
SUMMARY
Toll encodes a receptor that is critical for dorsal-ventral
patterning in the early Drosophila embryo. Previous data
have suggested that the accumulation of Toll protein in the
embryo temporally correlates with elongation of the poly
(A) tail of the message. Here, we demonstrate that Toll
mRNA is translationally activated by regulated
cytoplasmic polyadenylation. We also identify a 192
nucleotide regulatory element in the Toll 3′ UTR that is
necessary for robust translational activation of Toll mRNA
and also regulates polyadenylation. UV crosslinking
INTRODUCTION
In Drosophila, from late oogenesis to 1.5 hours of
embryogenesis, there is little to no transcription. Consequently,
regulation of gene expression to direct critical processes, such
as meiotic maturation and fertilization, must occur posttranscriptionally. Many mRNAs that are generated during
oogenesis and deposited into the growing oocyte are regulated
at the level of translation during embryogenesis. While several
mRNAs in the four Drosophila maternal patterning systems
(anterior, posterior, dorsal-ventral and terminal) are known to
be translationally regulated, the mechanisms of activation have
not been completely elucidated. To date, two different
mechanisms of activation have been observed in Drosophila.
Translational activation of nanos mRNA, the posterior
determinant, is regulated by localization of the mRNA (Gavis
and Lehmann, 1994). Alternatively, the anterior morphogen,
bicoid, is translationally regulated by cytoplasmic
polyadenylation (Sallés et al., 1994).
Cytoplasmic polyadenylation has been observed in a diverse
array of species, from starfish to mice (reviewed in Richter,
1996; Wickens et al., 1996). In mouse, Drosophila and
Xenopus, this mechanism is necessary for proper development
to occur (Gebauer et al., 1994; Sallés et al., 1994; Sheets et al.,
1995). Discrete cis elements in the 3′ untranslated regions (3′
UTRs) of maternal mRNAs regulated by this mechanism have
been identified in Xenopus and mouse. These elements have
been termed CPEs (cytoplasmic polyadenylation elements)
(Fox et al., 1989; McGrew et al., 1989) or ACEs (adenylation
control elements) (Huarte et al., 1992). These cis elements are
analyses suggest that two proteins bind specifically to the
192 nucleotide element. One or both of these proteins may
be factors that are required for translational regulation or
cytoplasmic polyadenylation. These studies demonstrate
that regulated polyadenylation plays a critical role in the
Drosophila dorsal-ventral patterning system.
Key words: Dorsal-ventral, Polyadenylation, Translational
regulation, Drosophila, Toll
both necessary and sufficient to direct polyadenylation and
translation of such maternal mRNAs. The nuclear
polyadenylation signal, AAUAAA, is also required for
cytoplasmic polyadenylation of mRNAs in both Xenopus and
mouse (McGrew et al., 1989; Fox et al., 1989; Vassalli et al.,
1989). The sequence elements that direct polyadenylation in
Drosophila have not yet been defined.
To understand better the importance of cytoplasmic
polyadenylation for patterning the early Drosophila embryo,
we examined the dorsal-ventral patterning system. Toll is one
of twelve maternal mRNAs required for proper dorsal-ventral
patterning. Toll encodes a transmembrane receptor protein that
mediates a signal transduction pathway, resulting in the
differential nuclear localization of the transcription factor
dorsal (Belvin and Anderson, 1996). Previous work has
suggested that Toll protein expression begins to accumulate
only prior to nuclear migration and reaches a peak at the late
syncytial blastoderm stage of embryogenesis (Gay and Keith,
1992), and that cytoplasmic polyadenylation may regulate
translation of Toll mRNA (Sallés et al., 1994).
In this report, we have investigated the role of
polyadenylation in the dorsal-ventral patterning system of
Drosophila. We demonstrate that polyadenylation of an
otherwise non-rescuing Toll RNA results in rescue of Toll
mutant embryos. We also identify sequences required for
translation and polyadenylation of Toll mRNA. Three regions
of the Toll 3′ UTR control polyadenylation. One region that
regulates the extent of polyadenylation correlates with the
region necessary for translational activation, suggesting that
the same sequences are required for both processes. We
2996 J. A. Schisa and S. Strickland
identify two proteins that bind specifically to the Toll 3′ UTR.
These proteins may be required to regulate polyadenylation
and translation of Toll mRNA. Our studies indicate that
cytoplasmic polyadenylation is necessary for proper dorsalventral patterning and support the idea that this mechanism is
vital for coordinated initiation of Drosophila development.
MATERIALS AND METHODS
Fly stocks
Canton-S flies were used for wild type. Toll embryos were obtained
from females transheterozygous for Df(3R)Tl 9QRX and Df(3R)ro
XB3, kindly provided by Dr Carl Hashimoto (Hashimoto et al., 1988).
Extracts and immunoblotting
Extracts from 0- to 3-hour old wild-type or Toll mutant embryos were
prepared (Driever and Nusslein-Volhard, 1988). Embryos were
collected at room temperature, dechorionated with chilled 50%
bleach, washed with 10 mM Tris (pH 7.4), 300 mM NaCl, 0.1%Triton
X-100, and allowed to settle to determine volume. Embryos were
frozen in liquid nitrogen and homogenized with 3 volumes 2× sample
buffer, 6 M urea. Extracts were incubated for 5 minutes at 95°C,
clarified by low-speed centrifugation at 4°C and stored at −80°C.
Supernatants were separated by SDS-PAGE and electroblotted to
nitrocellulose (Amersham ECL). Efficiency of electroblotting was
determined by transfer of prestained molecular mass markers
(Biorad). The blot was blocked in Tris-buffered saline (TBS), 5% nonfat dry milk (Carnation) and incubated with polyclonal anti-Toll
antibodies (gift of Dr Carl Hashimoto). Antibodies were preadsorbed
by incubation with wild-type embryo extract that had been run out on
an SDS-PAGE gel and transferred to nitrocellulose. After washing,
the blot was probed with goat anti-rabbit antibodies conjugated to
HRP (Amersham). Antibody/antigen complexes were detected by
ECL according to the manufacturer’s protocol (Amersham). Total
protein was determined visually after incubating the blot in 5% India
ink in PBS.
Polyadenylation assay (PCR-PAT)
For poly (A) tests (PAT) of Toll, oocyte or embryo RNA was isolated
as described (Wreden et al., 1997) and subjected to PAT reverse
transcription (Sallés and Strickland, 1995). PCR was performed using
a reaction including [32P]α-dATP (NEN). The anchor primer used for
reverse transcription and a message-specific primer (5′GTATCAACTGTAATCTCACGCCCA) were used in the PCR
reaction. The expected size of the PCR product is 335 nt.
In vitro transcriptions
The template for transcription reactions was Toll∆LRR, kindly
provided by Dr Carl Hashimoto. This construct of Toll has a large
deletion of the leucine-rich repeats in the N terminus, lacking amino
acids 151-801 of the extracellular domain (Winans and Hashimoto,
1995). In vitro transcription reactions of full-length Toll∆LRR (for
polyadenylation assays) were performed using approximately 1 µg of
linearized template in a 50 µl reaction containing 1× Epicentre
transcription buffer, 10 mM DTT, 0.5 mM ATP, 0.5 mM CTP, 0.05
mM GTP, 0.5 mM CAP analog [m7G(5′)ppp(5′)G] (Ambion), 0.012
mM UTP, 100 µCi [32P]α-UTP (3000 mCi/mmol), 80 units RNasin
(Promega) and 72.5 units Sp6 polymerase (Epicentre). The reactions
were incubated for 2 hours at room temperature. RNAs were
resuspended at a concentration of between 1×106 and 1×107
cts/minute/µl (approximately 10-100 ng/ml).
Small 3′ UTR RNA fragments were transcribed as above.
Templates were generated by PCR and T7 RNA polymerase was used.
Reactions were incubated at 37°C for 1.5 hours.
Toll∆LRR in vitro transcriptions for translation assays were
performed as above except 0.5 mM UTP and 0.1 µCi [32P]α-UTP
(3000mCi/mmol), were included in the reaction. RNA was
resuspended in filtered water at 0.2 mg/ml.
In vitro polyadenylation reactions were performed as described
(Sallés et al., 1994).
In vivo assay for translation
Toll mutant embryos (0-30 minutes old) were injected ventrally at
30% egg-length from the posterior (Hashimoto et al., 1988). Embryos
were allowed to develop for 48 hours prior to fixation with
Hoyer’s/lactic acid (1:1) to examine structures of the larval cuticle.
Cuticle preparations were baked at 60°C for 2-3 hours and visualized
with dark-field optics. Embryos were scored for formation of ventral
structures, ventral denticle bands and Filzkörper.
In vivo assay for polyadenylation
Embryos were injected as above, allowed to develop for 1 hour,
removed from the coverslip and the halocarbon oil (series 700:
Halocarbon Products Corp.) was removed by a heptane wash.
Embryos were washed in 1× PBS/0.1% Triton X-100 and Dounce
homogenized in 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 0.5% SDS
and 10 mM EDTA (Gottlieb, 1992). Carrier tRNA (10 mg) and 1 ml
proteinase K (14 mg/ml) (Boehringer Mannheim) were added and the
mix was incubated at 65°C for 30 minutes. The solution was phenolchloroform (1:1) extracted and ethanol precipitated in 0.3 mM sodium
acetate. RNAs were analyzed on either a 3.5% acrylamide/8.3 M urea
gel or 0.8% agarose/6% formaldehyde gel. In general, 1,000-4,000
cts/minute (equivalent to 10-25 embryos) were loaded per lane. The
amount of polyadenylation added to each RNA fragment was
estimated by comparison of the electrophoretic mobility of the
uninjected and injected RNA samples.
Stability analysis
Stability of RNAs was determined by injection of radiolabeled RNA,
incubation for 1 hour, recovery of RNA from embryos and gel
electrophoresis of RNA from equal numbers of embryos.
Autoradiography and densitometric analysis was performed to
determine the relative stability of different injected RNAs.
UV cross-linking assay
RNA probes to identify binding proteins were prepared as descibed
for in vitro transcriptions except that 10 mM GTP was included and
no CAP analog was used. Templates were prepared by PCR, followed
by transcription with T7 RNA polymerase. Reactions were incubated
at 37°C for 1.5 hours.
Protein extracts were prepared from 0 to 3 hour embryos as
described (Smibert et al., 1996). The cross-linking assay was done
using 5 µl protein extract in a 10 µl reaction with tRNA in 1× binding
buffer. Probe was incubated with extract for 20 minutes on ice.
Crosslinking was performed in a Stratalinker at 9999 µJoules. RNase
digestion was done with 20 µg RNaseA at 37°C for 15 minutes.
Proteins were resolved on a 10% SDS-polyacrylamide gel.
Competition binding was assayed using the probe RNA prepared
with 1000× less radioactive label incorporated. As non-specific
competitor RNA, a fragment of RNA called the SRE or smaug
recognition element, was used (kindly provided by Dr Craig Smibert).
Competitor RNA was incubated with protein extracts for 10 minutes
on ice prior to addition of the probe.
RESULTS
Toll mRNA is polyadenylated concomitant with its
translational activation during early embryogenesis
To begin to examine whether cytoplasmic polyadenylation
regulates translation during dorsal-ventral axis formation, we
Polyadenylation of Toll mRNA regulates translation 2997
focused on the Toll gene, one of twelve genes required for
dorsal-ventral patterning. It had been previously suggested that
Toll mRNA is translationally regulated in early development.
Toll mRNA is maternally provided and is found in the ovary
and early embryo (Hashimoto et al., 1988). Toll protein does
not accumulate in the embryo, however, until the syncytial
blastoderm stage, as shown by whole-mount immunohistochemical analyses (Gay and Keith, 1989). We used
western analysis to further examine the temporal pattern of Toll
protein expression. Toll protein was not detectable in ovaries
(data not shown) and protein levels substantially increased
during the first 2 hours of embryogenesis. Toll protein
expression peaked at the syncytial blastoderm stage (Fig. 1A),
consistent with whole-mount staining results. These results are
consistent with translational activation of Toll mRNA during
early embryogenesis or with Toll protein instability during the
first hour of development. In the context of our RNA injection
experiments (Figs 2, 3, and data not shown), we observe no
evidence for differential protein stability during the first few
hours of embryogenesis. Therefore, we favor the interpretation
that Toll mRNA is translationally activated during early
development as has been surmised for several maternally
deposited mRNAs.
Polyadenylation was previously implicated in regulating the
translation of Toll mRNA from studies utilizing a reverse
transcription-PCR assay, referred to as the PCR poly (A) test
A.
wt
Tl−
B.
Length of
poly(A) tail
Fig. 1. Temporal correlation between translational activation and
poly (A) tail elongation of Toll mRNA. (A) Western blot analysis of
Toll protein from wild-type or Toll embryos. Toll embryos were
collected from females that were homozygous Toll deficient. Tollspecific bands were detected in wild type but not in the Toll extracts
and migrated at the expected size (Hashimoto et al., 1991). At 0-1
hours of embryogenesis, only low levels of Toll protein were
expressed. By 1-2 hours of development, expression levels increased
substantially. Protein levels then decreased at 2-3 and 3-4 hours. This
blot was stained with India ink after probing; equal amounts of total
protein were loaded among the lanes (DNS). (B) PAT analysis of Toll
mRNA was performed upon developmentally staged wild-type
embryos and ovaries. By 1-1.5 hours of embryogenesis, the Toll
mRNA was elongated to its maximum length of 240 bases.
or PAT (Sallés et al., 1994; Sallés and Strickland, 1995). In this
assay, differences in the size of the PCR products reflect
differences in the length of the poly (A) tail of the mRNA. We
extended this analysis to ensure that the timing of
polyadenylation correlated with the timing of translational
activation as observed by Western analysis. RNA from wildtype ovaries and embryos was collected at several time points
during early embryogenesis and the poly (A) status of Toll
mRNA was examined. Toll mRNA had a poly (A) tail in the
oocyte of approximately 50 A’s (Fig. 1B). The poly (A) tail
was elongated during the first 2 hours of embryogenesis,
peaking at approximately 240 A’s. The timing of the onset of
polyadenylation (immediately after fertilization) and of
maximal elongation (approximately 2 hours postfertilization)
was consistent with polyadenylation as a mechanism of
translational activation of Toll.
The polyadenylation profile of Toll was comparable to that
seen in a subset of maternal mRNAs that are known to be
translationally activated at a similar time as Toll. One of these
maternal mRNAs, bicoid, is critical for anterior patterning of
the embryo (Frohnhöfer and Nüsslein-Volhard, 1986) and is
translationally regulated by polyadenylation (Sallés et al.,
1994). Thus, the temporal correlation between translational
activation and polyadenylation of Toll mRNA indicated a
possible direct regulatory role for polyadenylation.
Toll is translationally regulated by polyadenylation
Toll mutant embryos have a dorsalized phenotype lacking all
ventrally derived structures (compare Toll− to wild-type
embryo in Fig. 2B). The injection of Toll mRNA into the
ventral region of Toll mutant embryos results in rescue of the
mutant phenotype at the point of injection (Hashimoto et al.,
1988). This injection assay thus provides a means to analyze
translation of injected RNAs, as rescue of ventral structures
occurs only if Toll protein is generated. Therefore, to determine
if polyadenylation promotes translational activation of Toll
mRNA, we analyzed various injected Toll transcripts for their
ability to rescue Toll mutant embryos.
The construct of Toll used for these studies, ∆LRR, has a
deletion of the leucine-rich repeats in the extracellular region
of the protein (Winans and Hashimoto, 1995). This cDNA is
approximately 2 kb smaller than full-length Toll, thus
facilitating in vitro transcription, and its endogenous 5′ and 3′
UTRs remain intact. The UTR integrity was important as
sequences that regulate translation and polyadenylation often
reside in the 3′ UTRs of mRNAs (Jackson and Standart, 1990).
Injection of ∆LRR RNA into any region of Toll mutant embryos
results in a constitutively active Toll protein that rescues the
dorsalized phenotype at the point of injection (Winans and
Hashimoto, 1995).
Our strategy was to first identify a region of the mRNA that
was required to rescue the Toll mutant phenotype. We could
then determine the effect of poly (A) tail length on the ability
of the RNA to rescue. Since polyadenylation and translation
regulatory sequences are often located in the 3′ UTRs of
mRNAs (reviewed in Richter, 1996; Wickens et al., 1996), we
first truncated 1126 nt from the 3′ UTR (∆LRRXba). This
mRNA was unable to rescue the mutant phenotype (Fig. 2). To
investigate the role of poly (A) tail length in the ability to
rescue, we in vitro polyadenylated ∆LRRXba RNA prior to
injection. ∆LRRXba+A50 RNA contained ~50 A residues,
2998 J. A. Schisa and S. Strickland
Fig. 2. Toll is translationally regulated by
polyadenylation. (A) Translation of injected RNAs was
determined by the ability to rescue the Toll mutant
phenoype. An RNA truncated from the 3′ end by 1126
nt was not able to rescue the mutant phenotype
(LRRXba). Addition of 50 A’s did not affect the ability
to rescue. Addition of 200 A’s restored the ability to
rescue with 66% efficiency compared to full-length
LRR RNA. To determine polyadenylation, in vitro
transcribed radiolabelled RNAs were injected into
embryos, recovered and analyzed on denaturing gels to
assess changes in mobility, indicative of changes in the
poly (A) tail length. Only the full-length RNA with an
intact 3′ UTR was polyadenylated. Stability of RNAs
was determined by RNA injection, recovery, gel
electrophoresis of RNA from equal numbers of
embryos, autoradiography, and densitometric analysis.
The non-rescuing RNAs, LRRXba+ A50 and LRRXba,
were more stable than the rescuing RNA,
LRRXba+A200 (nd was not determined). (B) Cuticle
preparations of embryos: Wild-type embryo; Toll
embryo with dorsalized phenotype; Toll embryo
injected with LRRXba where no rescue of ventral
structures was observed; Toll embryo injected with
LRRXba+ A200 where ventral structures were rescued as
assayed by formation of ventral denticle bands or
Filzkörper. This embryo showed rescue of ventral
denticle bands (arrows); other injected embryos also
showed rescue of Filzkörper.
A.
B.
approximately the number on the endogenous, translationally
quiescent, oocyte mRNA. ∆LRRXba+A200 RNA contained ~200
A residues, which approximates the number on the
endogenous, translationally active, embryonic mRNA.
Addition of 50 A’s to ∆LRRXba had little effect on the ability
of the RNA to rescue. However, addition of the longer poly (A)
tail resulted in rescue of ventral structures in 30% of injected
embryos (Fig. 2).
One possibility for the difference in the three RNAs’ ability
to rescue ventral structures was differential stability of the
RNAs. Therefore, we determined the relative stability of the
three RNAs by injecting radiolabeled RNAs, recovering the
RNA after injection, and analyzing the amount of radioactivity
in RNA from equal numbers of injected embryos. The
truncated RNAs that did not rescue (∆LRRXba and
∆LRRXba+A50) were slightly more stable than the RNA that did
rescue (Fig. 2A); therefore, differential stability can not explain
the differential rescuing ability of the RNAs. It is interesting
to note that the least stable RNA was most efficient in ability
to rescue. A possible explanation for the decreased stability of
this RNA is that RNA degradation is often coupled to
translation; in mouse oocytes, mRNAs that are not translated
can remain stably in the cytoplasm for long periods of time
(Strickland et al., 1988).
We conclude from these experiments that the addition of a
long poly (A) tail can compensate efficiently for deletion of a
large region of the 3′ UTR (Fig. 2A, compare ∆LRRfull length
rescues 45% of embryos and ∆LRRXba+A200 rescues 30% of
embryos). We extrapolate from the rescue assay that
polyadenylation promotes the translation of Toll mRNA, as
ventral structure formation cannot occur in the absence of Toll
protein in these embryos. These data indicate that the Toll 3′
UTR contains sequences normally necessary for translation. In
addition, the length of the poly (A) tail of the Toll mRNA
regulates the ability of the RNA to rescue ventral structures.
The most parsimonious explanation is that the deleted
sequences normally direct polyadenylation (Fig. 4) which is
necessary for robust translation. In the absence of this sequence
information, a long poly (A) tail is sufficient to restore
translatability to the RNA.
From these data, we would expect that injected Toll RNA
containing an intact 3′ UTR would be polyadenylated and
injected truncated RNAs that can not rescue would not be
polyadenylated. We determined the polyadenylation status of
the abovementioned RNAs and found that only full-length
RNA was polyadenylated (Fig. 2A), further correlating
polyadenylation with the ability to rescue via translation of
Toll.
A 192 nt region of the Toll 3′ UTR is necessary for
rescue
To understand better the mechanism of translational activation
of Toll mRNA, we wanted to identify cis sequences necessary
for translational activation. We again used our rescue assay as
a measure of translation. We prepared a series of RNAs with
successive truncations from the 3′ end of ∆LRR RNA, injected
these RNAs into Toll mutant embryos and scored for rescue of
ventral structures. Two truncated RNAs proved to be most
informative in defining a region necessary for rescue. Deletion
of 492 nt from the 3′ end of the RNA (∆LRRSty) resulted in
formation of ventral structures in 16 of 32 injected embryos
(Fig. 3). Deletion of 684 nt however (∆LRRSfu), resulted in no
Polyadenylation of Toll mRNA regulates translation 2999
rescue of ventral structures in 77 injected embryos. These data
suggested the 192 nt region of the 3′ UTR between the
restriction sites SfuI and StyI is necessary for rescue. To further
investigate this possibility, we generated a construct of ∆LRR
with a discrete deletion of the 192 nt region between the SfuI
and StyI restriction sites. We found this RNA resulted in rescue
of ventral structures in only 6.2% of injected embryos. This
efficiency of rescue is significantly lower than the 45% rescue
observed with ∆LRRSma or with ∆LRRSty (Fig. 3A). The reason
for the differences in rescue with the different mutated RNAs
was not due to differential stability (Fig. 3A). This result
indicated that the 192 nt region is not absolutely required for
Toll RNA to be translated; however, the dramatic decrease in
efficiency of rescue suggested that these sequences normally
contribute to translation of Toll mRNA.
Polyadenylation regulatory elements in Toll 3′ UTR
To further correlate polyadenylation with translation of Toll
mRNA, we examined polyadenylation in vivo. We injected in
vitro transcribed, radiolabeled RNAs into the ventral region of
embryos. Total RNA was recovered 1 hour after injection and
resolved by denaturing gel electrophoresis. The timing of the
recovery of the RNA was 1.5-2 hours post-egg deposition,
when maximal elongation of the poly (A) tail is observed by
PAT analysis (Fig. 1).
As previously mentioned, we observed that full-length
∆LRR RNA underwent poly (A) tail elongation, whereas
∆LRRXba, which does not contain the 192 nt element, did not
result in any discernible elongation (summarized in Fig. 2).
However, quantitative assessment of rather modest changes in
apparent size (50-150 nt) relative to the large size of the ∆LRR
RNA (3.3 kb) was difficult. We therefore analyzed RNAs
corresponding to small regions of the 3′ UTR. A region of the
3′ UTR that was sufficient for rescue (XbaI-StyI) was
polyadenylated efficiently (Fig. 4). When the 192 nt region
from that RNA was removed (leaving XbaI-SfuI), the RNA
received only modest poly (A) addition, approximately half as
much as XbaI-StyI (Fig. 4). These data indicated the 192 nt
region contributes to the regulation of the extent of
polyadenylation. To determine if the 192 nt region was
sufficient to direct polyadenylation, we injected the SfuI-StyI
RNA fragment; however, we observed no poly (A) tail
elongation (Fig. 4). Therefore, this 192 nt region is not
sufficient to direct polyadenylation and other sequences must
contribute such as those directly upstream of the SfuI site.
Using this approach, we identified other cis elements in the
3′ UTR that contain polyadenylation regulatory sequences
(summarized in Fig. 4A). Sequences between the XbaI and
HindIII sites and between the NheI site and the 3′ end-directed
modest polyadenylation, similar to that observed with XbaISfuI above. No primary sequence homology was found
between these elements in an attempt to identify a Drosophila
CPE. The addition of the 192 nt region to either the upstream
or downstream sequences resulted in significantly more
adenylation, and we know from previous experiments with
∆LRRXba (Fig. 2) that the extent of adenylation is critical in
promoting translation. To test if an additive mechanism was
involved in promoting efficient polyadenylation, i.e. whether
several regions of the 3′ UTR were required together, the 192
nt region (SfuI-StyI) was deleted from the 3′ UTR. This RNA
A.
B.
Fig. 3. 192 nt in the Toll 3′ UTR is necessary for efficient
translation. In vitro transcribed RNAs were injected into Toll
embryos and assayed for the ability to rescue the mutant
phenotype as described in Fig. 2. (A) An RNA lacking 492 nt,
LRRSty, rescued the mutant phenotype. This RNA rescued
with the same or better efficiency as full-length RNA
(LRRSma). An RNA with a larger deletion from the 3′ end,
LRRSfu, no longer rescued ventral structures. RNA with a
small deletion of the 192 nt region resulted in a greatly
decreased ability to rescue compared to full-length RNA. The
stability of the various RNAs was determined as described in
Fig. 2. No significant differences in stability were observed.
(B) Cuticle preparations of injected embryos: uninjected Toll
embryo has dorsalized phenotype; Toll embryo injected with
full-length RNA (LRRSma) showed rescue of ventral
structures, ventral denticle bands (arrows) and Filzkörper
(asterisk); Toll embryo injected with LRRSty also showed
rescue of ventral structures, ventral denticle bands (arrows)
and partial Filzkörper (asterisk); LRRSfu did not rescue ventral
structures.
3000 J. A. Schisa and S. Strickland
exhibited only modest poly (A)
tail elongation, approximately the
amount observed with XbaI-SfuI.
The reduction in polyadenylation
efficiency of this deletion
compared to full-length may
explain the inefficient rescue
observed in Fig. 3. Thus, it
appears the 192 nt region is
critical
both
for
optimal
polyadenylation and also for
robust translation.
From previous work in the
laboratory, it was known that
embryos
from
females
homozygous for the cortex
mutation (referred to as cortex
embryos) have defects in
polyadenylation of Toll and bicoid
mRNAs as assayed by the PCR
polyadenylation assay (Lieberfarb
et al., 1996). We were interested
to examine Toll polyadenylation
in cortex embryos by our injection
and retrieval assay. Several Toll 3′
UTR RNA fragments were
injected into cortex embryos.
Each of the fragments that was
polyadenylated in wild-type
embryos showed a severe
reduction in poly (A) elongation
in cortex embryos (Fig. 4). These
results are consistent with
previously obtained results and
further suggest a role for cortex in
regulating polyadenylation of Toll
RNA.
Trans-acting factors
regulating polyadenylation
To understand better how
polyadenylation
regulates
translation, UV cross-linking
studies were utilized to identify
factors that bind to the Toll 3′
UTR. Several proteins from 0-3
hour embryo extracts were
observed to bind to the 192 nt
RNA fragment. To determine if
any proteins were binding this
region specifically, the 192 nt
region was used as a probe and
competitor RNA was included in
the binding reaction. Two
proteins, of apparent molecular
mass 89 kDa and 101 kDa,
specifically bound to the 192 nt
RNA fragment. Binding of both
of these proteins was greatly
reduced in the presence of excess
specific competitor RNA but not
A.
Xba
Hind
Sfu
Sty
Nhe
Poly (A) tail elongation
in wild-type
Poly (A) tail elongation
in cortex
++
nd
+
nd
--
nd
+
nd
--
nd
++
--
+
--
--
--
+
--
++
nd
+
nd
B.
Fig. 4. Multiple elements in the Toll 3′ UTR regulate polyadenylation. (A) A summary of all
injected RNA fragments indicates multiple polyadenylation regulatory elements. Multiple
experiments were performed for each fragment injected into wild-type embryos. The amount of
poly (A) added is indicated by ++ for approximately 120-160 A’s added, + for 50-120 A’s added
and – for <50 A’s added. Two small regions (XbaI-HindIII and NheI-3′ end) were each sufficient
to direct some polyadenylation and one (SfuI-StyI) contributed to the extent of poly (A) added but
was not by itself sufficient. (B) In vitro transcribed radiolabeled RNAs corresponding to small
regions of the 3′ UTR were injected and recovered as described in Fig. 2. An RNA containing
sequences upstream of and including the 192 nt element (fragment XbaI-StyI) was able to promote
efficient polyadenylation. An aliquot of the injected and recovered RNA was subjected to
treatment with oligo(dT) and RNaseH. This RNA was the same apparent size as uninjected RNA,
indicating the injected RNA had undergone addition of poly (A) (DNS). Deletion of the 192 nt
region from this fragment (fragment XbaI-SfuI) reduced the amount of polyadenylation; however,
the 192 nt region alone was not sufficent to direct polyadenylation (SfuI-StyI). Sequences
downstream of the 192 nt region can also direct polyadenylation as shown with fragments SfuI-3′
end, StyI-3′ end, and NheI-3′ end. Each of these fragments was also injected into cortex mutant
embryos and each showed a significant reduction of elongation in cortex embryos.
Polyadenylation of Toll mRNA regulates translation 3001
A
1
2
B
3
4
5
Fig. 5. Proteins binding to the 192 nt element of the Toll mRNA 3′
UTR. UV cross-linking assays were used to determine binding of
proteins to a radiolabeled RNA containing the 192 nt Toll 3′ UTR
element. Competition binding experiments were performed to
demonstrate specificity of binding. Increasing amounts of tracelabeled competitor RNAs (10× and 100× molar excess over probe)
were included in the binding reactions. A specific competitor A
(lanes 2,3) or a non-specific competitor B, (SRE: see Materials and
Methods), (lanes 4,5) was included. The binding of two proteins, of
apparent molecular mass 89 kDa and 101 kDa, was lost in the
presence of the specific competitor only. The positions at which
molecular mass markers migrated are indicated.
affected in the presence of a non-specific competitor RNA. To
examine further if these proteins were promiscuously binding
AU-rich stretches of RNA, we also examined binding with four
other fragments of the Toll 3′ UTR and with an antisense 192
probe. None of the other probes bound proteins of the same
apparent molecular mass as those shown to be specifically
bound by the sense 192nt RNA (89 kDa and 101 kDa – data
not shown). Interestingly, neither of the other two regions of
the 3′ UTR that regulate polyadenylation (Fig. 4) bound the
192 nt region-binding proteins. This result suggests that the
192 nt element recruits a specific factor(s) to the Toll 3′ UTR
and this factor(s) may be integral in directing the proper extent
of poly (A) added to Toll mRNA such that translation is
stimulated.
DISCUSSION
Polyadenylation stimulates translation of Toll mRNA
Maternal mRNAs in many diverse species are translationally
regulated by cytoplasmic polyadenylation. To date, only one
other Drosophila mRNA (bicoid) has been shown directly to
be regulated via this mechanism, and there have been strong
suggestions for others (Sallés et al., 1994; Wreden et al.,
1997). Our studies have investigated the role of
polyadenylation in establishing dorsal-ventral polarity.
Truncation of a large region of the 3′ end of the Toll 3′ UTR
resulted in an inability of the RNA to be translated. The
addition of a short poly (A) tail was not sufficient to promote
translation; however, a long poly (A) tail was able to stimulate
translation to nearly 30% efficiency as compared to the fulllength RNA. These data indicate that the length of the poly
(A) tail is a determining factor in predicting whether a Toll
RNA can be translated.
Organization of the Toll mRNA polyadenylation
signal
Our initial findings indicated a correlation between
polyadenylation and translation. Only RNAs that were
translated in our in vivo assay underwent significant
polyadenylation after injection and recovery. This suggested
that sequences directing translation and polyadenylation were
the same. We set out to identify sequences necessary for
polyadenylation of Toll mRNA, initially searching for a
vertebrate CPE-like sequence. Unlike polyadenylation
elements in mouse or Xenopus mRNAs that are discrete and
comprise two small cis elements in the 3′ UTR, sequences
regulating polyadenylation of Toll appear more complex.
Multiple regions of the 3′ UTR, with no apparent sequence
homology, can direct polyadenylation. Perhaps most
unexpected was the finding that while the 192 nt region is
required for robust levels of rescue in our in vivo
translation assay, this region is not sufficient to direct
polyadenylation. The 192 nt region is necessary for efficient
polyadenylation of RNA and we presume this is linked to
the requirement for robust translation. We imagine some
factor(s) may be binding this region of the 3′ UTR and
functioning to enhance poly (A) tail elongation. Identification
of such binding proteins is the next step in better
understanding how polyadenylation regulates translation of
Drosophila mRNAs.
Vertebrate versus invertebrate polyadenylation
regulatory elements
One of the required polyadenylation cis elements in mouse
and Xenopus is the hexanucleotide AAUAAA (nuclear
polyadenylation signal). The presumptive nuclear
polyadenylation signal of Toll does not appear necessary for
cytoplasmic polyadenylation or translation of Toll mRNA
(Fig. 3). This element also appears to be dispensable
regarding regulated translation of Drosophila bicoid mRNA
(Sallés et al., 1994). Thus, there appear to be at least some
differences among species so far as to which elements are
required for cytoplasmic polyadenylation. Interestingly, the
Xenopus polyadenylation machinery recognizes sequences in
the bicoid 3′ UTR and polyadenylates this RNA in a
developmentally regulated manner (Verrotti et al., 1996).
However, in initial reciprocal experiments, a Drosophila
embryo did not recognize polyadenylation elements in an
injected
Xenopus
RNA
(C.
Wreden,
personal
communication). Thus, it is far from clear the extent of
conservation of polyadenylation signals; the further
identification of trans factors in a variety of species will likely
aid in answering these questions.
A comparison of the 3′ UTR of bicoid and the 192 nt region
of Toll revealed one small region of homology. However, the
significance of this homology is not yet known. Interestingly,
deletion of a vertebrate CPE-like region of the bicoid 3′ UTR
had no effect on polyadenylation (Sallés et al., 1994). CPE-like
sequences are also not necessary for polyadenylation of
ribonucleotide reductase mRNA in the clam (Standart and
Dale, 1993). These data are consistent with the idea that the
regulatory sequences of this evolutionarily conserved
mechanism are not completely conserved between vertebrates
and invertebrates.
3002 J. A. Schisa and S. Strickland
Recognition of the translational control element by
trans-acting factors
We have demonstrated that two proteins specifically interact
with the 192 nt region of the Toll 3′ UTR. As this region of the
3′UTR was shown to be necessary for efficient polyadenylation
and efficient translation, these proteins may be factors
regulating these processes. It will be informative to determine
the identity of these proteins to aid in understanding how
polyadenylation promotes translational activation. One
candidate protein might be a homolog of the Xenopus CPEB
(CPE-binding protein). A Drosphila protein, orb, has been
identified that has homology to CPEB (Lantz et al., 1992). The
major defect in severe orb alleles is lack of localization of
several mRNAs during oogenesis (Christersen and McKearin,
1994; Lantz et al., 1994). The orb protein may have a role in
polyadenylation of mRNAs during oogenesis (P. Schedl,
personal communication). One of the proteins that we have
identified, of apparent molecular mass 101 kDa, may be orb
which is 97-99 kDa (Lantz et al., 1992). Examination of Toll
polyadenylation in orb mutants or immunoprecipitation of the
cross-linked proteins with an orb antibody may be useful in
determining whether any of the Toll-binding proteins are orb.
We plan to identify and characterize these proteins to gain
insight into the mechanism of Drosophila polyadenylation. It
will be interesting to see if either of these proteins binds 3′
UTRs of other maternal mRNAs regulated by polyadenylation,
such as bicoid.
Why is translational regulation of Toll mRNA
important for dorsal-ventral patterning?
A question that remains unanswered is why the embryo
temporally regulates expression of Toll. One possibility is that
while ventral-specific activation of the ligand of Toll, spatzle,
acts as the spatial restictor of the dorsal-ventral signal, Toll
functions as a temporal modulator. Toll is not spatially
restricted at any point of development; rather it is expressed
ubiquitously in the plasma membrane (Hashimoto et al., 1991).
If Toll protein was generated earlier in oogenesis or
embryogenesis, there could be adverse developmental
consequences. For instance, Toll is known to have more than
one functional ligand (Lemaitre et al., 1995). If such a ligand
was present in the oocyte or early embryo and Toll was also
expressed at this time, there could be deleterious consequences
to the embryo. Alternatively, as Toll is a transmembrane
protein, expression during oogenesis could result in the protein
being trapped in the nurse cells and not deposited into the
oocyte. Some maternal mRNAs are known to be degraded soon
after translation and our data suggests that translated Toll RNA
is slightly less stable than truncated, untranslated forms of Toll
RNA (Fig. 2A). Thus if Toll protein was translated but not
transported to the oocyte, there might not be sufficient mRNA
remaining in the oocyte and a lethal Toll null phenotype would
result.
Recent work examining the role of Toll in regulating
synaptic initiation of the motoneuron, RP3, found that
heterochronic misexpression of Toll in muscles leads to
delayed synaptic initiation. The authors concluded that both
temporal and spatial control of Toll expression in muscle cells
is critical for its role in development of the embryonic
musculature (Rose et al., 1997). Although many studies of the
dorsal-ventral pathway have focused on understanding how
Toll is activated in a spatially restricted manner, the importance
of its temporal regulation may also prove to be a critical
regulatory aspect.
We thank Dr Carl Hashimoto for providing fly stocks, Toll antibody,
the LRR cDNA and generous and invaluable expertise. We thank Dr
Christopher Wreden for technical advice and comments on the
manuscript, Dr Craig Smibert for advice on crosslinking and
providing cDNAs, Drs Stella Tsirka and Fernando Sallés and Emily
Harms for critical reading of the manuscript, Drs Paul Schedl and
Christopher Wreden for permission to cite unpublished data, and Drs
Peter Gergen, Joanne Engebrecht and William Theurkauf, and
members of the Strickland laboratory for helpful discussion. This
work was supported by a grant from the National Institutes of Health
to S. S. (GM51584) and a predoctoral training grant in genetics
(GM07964).
REFERENCES
Belvin, M. P. and Anderson, K.V. (1996). A conserved signaling pathway:
the Drosophila Toll-dorsal pathway. Ann. Rev. Cell Dev. Biol. 12, 393-416.
Christersen, L. B. and McKearin, D. M. (1994). orb is required for
anteroposterior and dorsoventral patterning during Drosophila oogenesis.
Genes Dev. 8, 614-628.
Driever, W. and Nüsslein-Volhard, C. (1988). A gradient of bicoid protein
in Drosophila embryos. Cell 54, 83-93.
Fox, C. A., Sheets, M. D. and Wickens, M. (1989). Poly (A) addition during
maturation of frog oocytes: distinct nuclear and cytoplasmic activities and
regulation by sequence UUUUUAU. Genes Dev. 3, 2151-2162.
Frohnhöfer, H. G. and Nüsslein-Volhard, C. (1986). Organization of anterior
pattern in the Drosophila embryo by the maternal gene bicoid. Nature 324,
120-125.
Gavis, E. R. and Lehmann, R. (1994). Translational regulation of nanos by
RNA localization. Nature 369, 315-318.
Gay, N. J. and Keith, F. J. (1992). Regulation of translation and proteolysis
during the development of embryonic dorso-ventral polarity in Drosophila:
Homology of easter proteinase with Limulus proclotting enzyme and
translational activation of Toll receptor synthesis. Biochim. Biophys. Acta
1132, 290-296.
Gebauer, F., Xu, W., Cooper, G. M. and Richter, J. D. (1994). Translational
control by cytoplasmic polyadenylation of c-mos mRNA is necessary for
oocyte maturation in the mouse. EMBO J.13, 5712-5720.
Gottlieb, E. (1992). The 3′ untranslated region of localized maternal messages
contains a conserved motif involved in mRNA localization. Proc. Natl.
Acad. Sci. USA. 89, 7164-7168.
Hashimoto, C., Gerttula, S. and Anderson, K. V. (1991). Plasma membrane
localization of the Toll protein in the syncytial Drosophila embryo:
importance of transmembrane signaling for dorsal-ventral pattern formation.
Development 111, 1021-1028.
Hashimoto, C., Hudson, K. L. and Anderson, K. V. (1988). The Toll gene
of Drosophila, required for dorsal-ventral embryonic polarity, appears to
encode a transmembrane protein. Cell 52, 269-279.
Huarte, J., Stutz, A., O’Connell, M., Gubler, P., Belin, D., Darrow, A.,
Strickland, S. and Vassalli, J.-D. (1992). Transient translational silencing
by reversible mRNA deadenylation. Cell 69, 1021-1030.
Jackson, R. and Standart, N. (1990). Do the poly (A) tail and 3′ untranslated
region control mRNA translation? Cell 62, 15-24.
Lantz, V., Ambrosio, L. and Shedl, P. (1992). The Drosophila orb gene is
predicted to encode sex-specific germline RNA-binding protein and has
localized transcripts in ovaries and early embryos. Development 115, 75-88.
Lantz, V., Chang, J. S., Horabin, J. I., Bopp, D. and Shedl, P. (1994). The
Drosophila orb RNA-binding protein is required for the formation of the
egg chamber and the establishment of polarity. Genes Dev. 8, 598-613.
Lemaitre, B., Meister, M., Govind, S., Georgel, P. and Steward, R. (1995).
Functional analysis and regulation of nuclear import of dorsal during the
immune response in Drosophila. EMBO J. 14, 536-545.
Lieberfarb, M. E., Chu, T., Wreden, C., Theurkauf, W., Gergen, J. P. and
Strickland, S. (1996). Mutations that perturb poly (A)-dependent maternal
Polyadenylation of Toll mRNA regulates translation 3003
mRNA activation block the initiation of development. Development 122,
579-588.
McGrew, L. L., Dworkin-Rastl, E., Dworkin, M. B. and Richter, J. D.
(1989). Poly (A) elongation during Xenopus oocyte maturation is required
for translational recruitment and is mediated by a short sequence element.
Genes Dev. 3, 803-815.
Richter, J. D. (1996). Dynamics of poly (A) addition and removal during
development. In Translational Control (eds. J. W. B. Hershey, M. B.
Mathews and N. Sonenberg), pp. 481-503. Cold Spring Harbor, New York:
Cold Spring Harbor Laboratory Press.
Rose, D., Zhu, X., Kose, H., Cho, B. H. and Chiba, A. (1997). Toll, a muscle
cell surface molecule, locally inhibits synaptic initiation of the RP3
motoneuron growth cone in Drosophila. Development 124, 1561-1571.
Sallés, F. J., Lieberfarb, M. E., Wreden, C., Gergen, J. P. and Strickland,
S. (1994). Coordinate initiation of Drosophila development is regulated by
polyadenylation of maternal messenger RNAs. Science 266, 1996-1999.
Sallés, F. J. and Strickland, S. (1995). Rapid and sensitive analysis of mRNA
polyadenylation states by PCR. PCR Meth. Applic. 4, 317-321.
Sheets, M. D., Wu, M. and Wickens, M. (1995). Polyadenylation of c-mos
mRNA as a control point in Xenopus meiotic maturation. Nature 374, 511516.
Smibert, C., Wilson, J. E., Kerr, K. and MacDonald, P. M. (1996). smaug
protein represses translation of unlocalized nanos mRNA in the Drosophila
embryo. Genes Dev. 10, 2600-2609.
Standart, N. and Dale, M. (1993). Regulated polyadenylation of clam
maternal mRNAs in vitro. Dev. Genet. 14, 492-499.
Strickland, S., Huarte, J., Belin, D., Vassalli, A., Rickles, R. J. and Vassalli,
J.-D. (1988). Antisense RNA directed against the 3′ UTR noncoding region
prevents dormant mRNA activation in mouse oocytes. Science 241, 680684.
Vassalli, J.-D., Huarte, J., Belin, D., Gubler, P., Vassalli, A., O’Connell, M.,
Parton, L., Rickles, R. and Strickland, S. (1989). Regulated
polyadenylation controls mRNA translation during meiotic maturation of
mouse oocytes. Genes Dev. 3, 2163-2171.
Verrotti, A. C., Thompson, S. R., Wreden, C., Strickland, S. and Wickens,
M. (1996). Evolutionary conservation of sequence elements controlling
cytoplasmic polyadenylation. Proc. Natl. Acad. Sci. USA 93, 9027-9032.
Wickens, M., Kimble, J. and Strickland, S. (1996). Translational control of
developmental decisions. In Translational Control (eds. J. W. B. Hershey,
M. B. Mathews and N. Sonenberg), pp. 411-450. Cold Spring Harbor, New
York: Cold Spring Harbor Laboratory Press.
Winans, K. A. and Hashimoto, C. (1995). Ventralization of the Drosophila
embryo by deletion of extracellular leucine-rich repeats in the Toll protein.
Mol. Bio. Cell 6, 587-596.
Wreden, C., Verrotti, A. C., Schisa, J. A., Lieberfarb, M. and Strickland,
S. (1997). Nanos and pumilio establish embryonic polarity by promoting
posterior deadenylation of hunchback mRNA. Development 124, 30153023.