M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6

. sc iencedi rec t . com

ava i lab le a t www

journa l homepage: www.elsevier .com/ locate /modo

Bruno negatively regulates germ cell-less expressionin a BRE-independent manner

Jocelyn Moore, Hong Han, Paul Lasko*

Developmental Biology Research Initiative (DBRI) and Department of Biology, McGill University, Montreal, Que., Canada

A R T I C L E I N F O

Article history:

Received 5 December 2008

Received in revised form

8 April 2009

Accepted 10 April 2009

Available online 22 April 2009

Keywords:

Drosophila

Germ cell-less

Bruno, Oogenesis

Germline

Regulation

Translation

RNA-binding

0925-4773/$ - see front matter � 2009 Elsevidoi:10.1016/j.mod.2009.04.002

* Corresponding author. Tel.: +1 514 398 64E-mail address: paul.lasko@mcgill.ca (P. L

A B S T R A C T

Mechanisms of post-transcriptional control are essential during Drosophila oogenesis and

embryogenesis to sequester gene products in discrete regions and ultimately achieve

embryonic asymmetry. Maternal germ cell-less (gcl) mRNA accumulates in the pole plasm

of the embryo before Gcl protein is detectable. gcl mRNA, but not Gcl protein, can also be

detected in somatic regions of the embryo, suggesting that gcl RNA is subject to transla-

tional control. We find that Gcl is expressed during oogenesis, and that it is regulated by

the translational repressor Bruno (Bru). Increased levels of Gcl are observed in the oocyte

when Bru level is reduced, and overexpression of Bru reduces Gcl expression. Consistently,

reduction of the maternal dosage of Bruno leads to ectopic Gcl expression in the embryo,

which, in turn, represses anterior huckebein (hkb) expression. Bru binds directly to the gcl

3’UTR in vitro, but, surprisingly, this binding is independent of a BRE (Bruno response ele-

ment)-like motif. This motif is also not required for in vivo repression of Gcl expression

during oogenesis or early embryogenesis. Bru binds the gcl 3’UTR via its C-terminal

domain, which includes RNA recognition motif 3 (RRM3), with little or no contribution from

the remainder of the protein. We conclude that repression by Bruno during oogenesis is

required to restrict Gcl expression in the early embryo and that Bru represses gcl expression

in a manner that involves RRM3 and a sequence unrelated to the BRE.

� 2009 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

The polarization of body axes in the Drosophila embryo and

the segregation of the germline both require that the expres-

sion of specific maternal transcripts is restricted to limited

temporal and spatial patterns. Multiple levels of post-tran-

scriptional regulation of localized transcripts underlie this

process, as has been characterized for oskar (osk) and nanos

(nos), two mRNAs required for posterior patterning and germ

cell development (Vardy and Orr-Weaver, 2007). These mRNAs

become enriched in the pole plasm during oogenesis by pro-

cesses that include biased localization along the microtubule

network and localized trapping by actin microfilaments (For-

rest and Gavis, 2003; Zimyanin et al., 2008). Restriction of

er Ireland Ltd. All rights

01; fax +1 514 398 5069.asko).

Osk and Nos proteins to the posterior pole is accomplished

primarily by complex translational regulation, whereby the

mRNA component present in the pole plasm is translationally

active, while the rest is translationally repressed (Wilhelm

and Smibert, 2005; Vardy and Orr-Weaver, 2007).

Bruno (Bru) is an RNA-binding protein with three RNA

recognition motifs (RRMs) that regulates the translation of

several germline mRNAs, including osk. Bru binds directly to

discrete sequence elements in the osk 3’UTR known as Bruno

response elements (BREs), using both the N-terminal and

C-terminal RRMs (Kim-Ha et al., 1995; Webster et al., 1997;

Snee et al., 2008). Bru negatively regulates osk translation

through two distinct mechanisms: first by blocking recogni-

tion of the 5’ cap structure by recruiting Cup, a eukaryotic

reserved.

504 M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6

initiation factor 4E (eIF4E) binding protein that competitively

inhibits recruitment of eIF4G (Nakamura et al. 2004), and sec-

ond, in a cap-independent manner by packaging osk RNA into

heavy particles that render osk inaccessible to the translational

machinery (Nakamura et al., 2004; Chekulaeva et al., 2006).

Three more targets of Bruno-mediated translational repression

are also known: gurken (grk), Sex lethal (Sxl) and Cyclin A (CycA).

Bruno binds directly to the grk 3’UTR, and Bru overexpression

leads to defects in dorsal-ventral polarity due to a reduction

of Gurken in the oocyte (Filardo and Ephrussi, 2003). Regulation

of Sxl translation by Bru is required in the germarium for cysto-

blast differentiation, and repression of CycA RNA is essential to

maintain mitotic quiescence during meiosis and to prevent

overproliferation of germ cells (Parisi et al., 2001; Sugimura

and Lilly, 2006; Wang and Lin, 2007).

Maternal germ cell-less (gcl) activity is required for germ cell

specification, although, unlike osk and nos, gcl is not involved

in posterior somatic patterning. Females lacking gcl function

produce viable embryos with few or no pole cells, but with

no detectable somatic defects (Robertson et al., 1999). Gcl

accumulates at the interior face of the nuclear membrane of

pole cell nuclei, where it is required to establish transcrip-

tional silencing in the germline (Robertson et al., 1999; Leath-

erman et al., 2002). Ectopic expression of Gcl in the anterior of

the embryo, achieved by fusing the gcl coding region to the bi-

coid (bcd) 3’UTR, results in reduced transcription of anterior

somatic genes, sisterlessA, tailless and huckebein (hkb), suggest-

ing a shift towards germline fate (Leatherman et al., 2002).

gcl RNA is first detected in the nurse cells at stage 5 of

oogenesis and is abundant in the oocyte at stage 10 (Jongens

et al., 1992; Nakamura et al., 2001). gcl RNA is also present

throughout the early embryo and it accumulates in the pole

plasm (Jongens et al., 1992). Gcl protein is not present outside

of the pole plasm, and is first detected in the pole plasm after

gcl RNA has accumulated there. Gcl is readily detectable at the

time of pole bud and pole cell formation, when it is associated

with the nuclear envelopes of nuclei that contact the pole

plasm (Jongens et al., 1992). The distributions of gcl RNA

and protein support the hypothesis that gcl mRNA is transla-

tionally repressed in the embryonic soma, and that this

repression is relieved by pole plasm components. This

hypothesis has, however, not heretofore been directly tested.

Here, we implicate Bruno in the translational regulation of

gcl. We determined that Gcl protein is expressed during

oogenesis, where it localizes to the nuclear envelope in both

nurse cells and the oocyte, and that Gcl expression in the

oocyte cytoplasm is regulated by Bru. This regulation is re-

quired to restrict Gcl expression to the germline in the early

embryo and Gcl protein is detectable outside of the germ

plasm in embryos from females heterozygous for arrest (aret,

which encodes Bru). These embryos also show reduced zygo-

tic hkb expression. We identified a sequence in the gcl 3’UTR

that matches the BRE consensus sequence and is conserved

among several Drosophila species. Recombinant Bru binds to

the gcl 3’UTR in vitro, but surprisingly, this binding is indepen-

dent of the gcl BRE and dependent upon a different region of

the 3’UTR. The gcl BRE is also not required to repress Gcl

expression during oogenesis or embryogenesis. In contrast

to Bru binding to osk, gcl is not bound by an N-terminal Bru

construct that includes RRM1 and 2. However, a C-terminal

Bru construct binds the gcl 3’UTR efficiently. We conclude that

Bruno represses gcl translation during oogenesis in order to

refine Gcl expression in early embryos and that this regula-

tion occurs through a novel interaction involving RRM3 and

operating independently of the BRE.

2. Results

2.1. Germ cell-less protein is expressed during oogenesis

Four independent polyclonal antibodies were raised

against a GST-Gcl fusion protein and used to characterize

Gcl expression during oogenesis. Pre-adsorbed aGcl antisera

identified a single band on immunoblots at the predicted

molecular weight of 65 kDa in protein extracts made from

wildtype ovaries, but not in protein extracts made from gclD

ovaries (Fig. 1A). Similar results were obtained with embry-

onic extracts (data not shown). Using these antisera, we de-

tected Gcl at the nuclear membrane of germline cells in the

germarium as early as stage 2b (staging according to King,

1970). Gcl accumulates in the oocyte cytoplasm (Fig. 1D and

D’) and colocalizes with nuclear pore complex proteins in

the nurse cells (Fig. 1B). In contrast, Gcl does not colocalize

with Vasa (Vas), a component of the perinuclear nuage (Lasko

and Ashburner, 1990), but is detected immediately adjacent to

it toward the interior of the nuclei (Fig. 1C). This indicates

that, as in cells in the early embryo, Gcl localizes predomi-

nantly at the interior of the nuclear envelope in nurse cells.

Starting at stage 8, Gcl becomes slightly enriched at the

oocyte cortex, particularly at the anterior and posterior poles,

which persists at the posterior at least until the end of stage

10 (Fig. 1D and D’). We also detected Gcl expression in

embryogenesis as previously described (Jongens et al., 1992).

All four independent antisera we produced gave essentially

identical results for Gcl expression in ovaries, as did a previ-

ously published antiserum (Jongens et al., 1992).

2.2. Gcl expression in the oocyte depends on Bruno

We assessed Gcl distribution in ovaries from females mu-

tant for osk (osk54 /Df(3R)p-XT103), vas (vas1) and tud (tudtux),

respectively (Lehmann and Nusslein-Volhard, 1986; Schup-

bach and Wieschaus, 1986, 1991; Thomson and Lasko, 2004).

osk and tud mutations had no effect on Gcl distribution during

oogenesis (osk54/Df(3R)p-XT103, Fig. 1G and G’ and data not

shown) indicating that Gcl accumulation is independent of

osk and tud. However, in vas1 oocytes Gcl frequently accumu-

lates in an irregular pattern at the anterior and lateral cortex

and cortical Gcl appears more prominent (Fig. 1H–J). As gcl

RNA localization is not significantly altered in egg chambers

from pole plasm mutants (Fig. 1K–M and data not shown), this

suggests that posterior enrichment of Gcl protein in stage

8–10 oocytes is independent of the canonical pole plasm

assembly pathway, but may depend specifically on vas for

maintenance at the lateral and anterior cortex.

We next investigated whether known regulators of transla-

tion affect Gcl expression during oogenesis. gcl RNA colocal-

izes with Me31B (Nakamura et al., 2001), but we observed no

change in Gcl expression in me31BD1 germline clones (data

Fig. 1 – Gcl is expressed during oogenesis. (A) Immunoblot of ovary extracts probed with aGcl antiserum. The same blot

probed with aeIF4a is shown as a loading control. A minor nonspecific band is present above the specific Gcl band in extracts

of ovaries from gclD and other females. Our antisera detect a single band in all extracts except those from gclD ovaries. (B,C)

Nurse cell nuclei, immunostained with aGcl (green), DAPI (blue) and either MAb414 (Nuclear Pore Complex, NPC, red) or aVas

(red). aGcl immunostain colocalizes with NPC (B), but not with Vas (C), indicating that Gcl localizes to the interior face of the

nuclear envelope. D-E, Egg chambers immunostained with aGcl (D, E, green or D’, monochrome) and stained with DAPI (D, E,

blue). Gcl is detected in the oocyte at the nuclear membrane (D’, arrows) in nurse cells and oocytes, and accumulates at the

oocyte cortex at the anterior and posterior (D’, arrowheads). (F–H) Stage 9 egg chambers from wildtype (F,F’), osk54/Df(3R)p-

XT103 (G,G’) or vas1 (H,H’) females immunostained with aGcl (green). Nuclei were labeled with DAPI (blue). Monochrome

panels (F’–H’) are projections of images collected at multiple confocal planes. (H, H’) Gcl accumulates in an irregular pattern at

the anterior (H’, arrow) and is reduced in the cytoplasm of vas1 oocytes (H’, arrowhead). (I,J) Magnification of the anterior

cortical region of the wt and vas1 oocytes in F and H, respectively. (K–M) RNA in situ hybridization to detect gcl mRNA in stage

8–9 egg chambers from wild-type (K), osk54/Df(3R)p-XT103 (L) or vas1 (M) females. Unlike Gcl protein distribution which is

affected by vas1, gcl RNA localization is not significantly altered in these pole plasm mutants; compare L and M with K.

M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6 505

506 M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6

not shown). However, we found that Gcl is expressed at abnor-

mally high levels between stages 8 and stage 10B in egg cham-

bers from mothers heterozygous for aretQB72 (Fig. 2A–F), which

encodes a truncated form of Bru lacking RRM3. Gcl overex-

pression is especially prominent in the oocyte in these egg

chambers. Gcl is also overexpressed in females heterozygous

for a deficiency including aret (Df(2L)esc-P3-0/+; Fig. 2C and F)

indicating that the phenotype results from a reduction in

Bru dosage and not from an uncontrolled second site mutation

on the aretQB72 chromosome. There is no significant change in

gcl RNA level or localization in aret mutants at mid-oogenesis

indicated by either gcl RNA in situ hybridization (Fig. 2G–I) or

Fig. 2 – Gcl expression in the oocyte depends upon Bru. (A–F) Lat

from females of the following genotypes: Oregon-R (A, A’, D, D’

deficiency includes aret), immunostained with aGcl (green). Nuc

are projections of images collected at multiple confocal planes.

mutants (B, C, E, F, arrowheads). (G–I) gcl RNA in situ hybridizati

and Df(2L)esc-P3-0/+ (I) females. Unlike Gcl protein, gcl RNA is n

with G). (J) Northern blot of total RNA from ovaries of the genot

methylene blue (MB) stained membrane is shown as a loading

transgene, but is similar in ovaries from wildtype females and a

aret-overexpressing (L,L’) females immunostained with aGcl (gr

Monochrome panels (K’, L’) are projections of images collected at

chambers when Bru is overexpressed (L, L’), particularly in the oo

9 egg chambers from control (M) or aret-overexpressing (N) fem

overexpression. O, P, Stage 9 egg chambers from control (O) or a

hybridization. Unlike Gcl protein, gcl RNA is not decreased whe

by Northern blot (Fig. 2J) that would account for the observed

increase in Gcl protein. In addition, there is no change in the

expression level or localization of Gcl in the germaria of either

aretQB72/Df(2L)esc-P3-0 or aretPA62/Df(2L)esc-P3-0 ovaries, which

arrest at an early stage of oogenesis (data not shown), indicat-

ing that Bru does not regulate gcl expression at this stage.

Overexpression of Bru using the Gal4–UAS system (Brand

and Perrimon, 1993; Snee et al., 2007) causes a reciprocal ef-

fect: Gcl expression is reduced in both nurse cells and oo-

cytes ( nosGal4::VP16; UAS-Bru, Fig. 2K–N). Consistent with

previously published results, Bru overexpression also re-

presses Osk expression in oocytes starting at stage 8

e stage 8 (A–C, A’–C’) and stage 10A (D–F, D’–F’) egg chambers

), aretQB72/+ (B, B’, E, E’) and Df(2L)esc-P3-0/+ (C, C’, F, F’; this

lei were labeled with DAPI (blue). Monochrome panels (A’-F’)

Gcl expression is increased in the oocyte cytoplasm in aret

on of stage 9 egg chambers from Oregon-R (G), aretQB72/+ (H),

ot increased in heterozygous aret oocytes (compare H and I

ypes indicated probed with a 32P-labeled gcl cDNA. The

control. gcl RNA increases upon overexpression from a gcl

ret mutants. K-L, Stage 9 egg chambers from control (K, K’) or

een) and aOsk (red). Nuclei were marked with DAPI (blue).

multiple confocal planes. Gcl expression is decreased in egg

cyte cytoplasm (L’, arrowhead, compare with K’). M, N, Stage

ales immunostained with aBru to visualize Bru

ret-overexpressing (P) females probed for gcl RNA by in situ

n Bru is overexpressed (compare P with O).

Fig. 3 – Ectopic Gcl expression perdures into embryogenesis. (A–D) Stage 5 embryos produced by Oregon-R (A,A’), P[gcl+]/+ (B,

B’), aretQB72/+ (C, C’) and aretQB72/+; P[gcl+]/+ (D,D’) immunostained with aGcl (A–D, green or A’–D’, monochrome). Nuclei were

marked with DAPI (A–D, blue). Inset images (A’–D’, upper right) are 4-fold magnifications of somatic nuclei. Gcl is present in

the somatic regions of embryos from females that overexpress gcl RNA from a transgene (B,B’) or from aret mutant females

(C,C’). Gcl is present at even higher levels in embryos from aret mutant females that also express gcl RNA from a transgene

(D,D’). (E–H) gcl RNA in situ hybridization of pre-blastoderm embryos produced by Oregon-R (E), P[gcl+]/+ (F), aretQB72/+ (G) and

aretQB72/+; P[gcl+]/+ (H). gcl RNA level increases upon expression from a transgene, but is not significantly altered by reduction

of maternal aret dosage (compare G with E and H with F). (I) Northern blot of total RNA from stage 1–5 embryos of various

genotypes probed with 32P-labeled gcl cDNA (upper panel). The same blot probed with a 32P-labeled RpS15A ribosomal protein

cDNA is shown as a loading control (lower panel). gcl RNA level increases upon expression from a transgene, but is not

significantly altered by reduction of maternal aret dosage. (J) Graph showing the relative levels of aGcl immunofluorescence in

the somatic regions of early embryos (measured from 5 to 9 embryos of each genotype), quantified from images shown in A–

D. Normalized values are shown at the top of each bar, *p < 0.0005, **p < 0.005. Somatic Gcl expression is elevated in embryos

from both females that overexpress gcl RNA from a transgene and aret mutant females, and this effect is additive. (K) Graph

showing somatic Gcl immunofluorescence normalized against RNA expression (measured from 3 Northern blots for each), as

determined by quantitative Northern blot. Normalized values are shown at the top of each bar, *p < 0.0005, **p < 0.005. Somatic

Gcl protein expression is elevated relative to RNA expression in embryos from aret mutant females. The data shown in J and

K are also presented in Table 1. L, M, hkb RNA distribution of stage 5 embryos produced by Oregon-R (L) or aretQB72/+ (M)

females, showing a reduction of anterior hkb in the latter. (N) Quantitative comparison of the anterior hkb RNA expression

domain, based on RNA in situ data taken from 83 to 134 embryos of each of the indicated genotypes. The data shown

graphically in N are also presented in Table 2.

M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6 507

508 M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6

(Fig. 2L). As observed with aret mutants, there is no change

in gcl RNA localization at this stage that would account for

the change in Gcl protein expression (Fig. 2O and P). In

addition, there is no observable change in Gcl level in egg

chambers that lack osk (osk54/Df(3R)p-XT103, Fig. 1G and

G’), so we conclude that the reduction in Gcl observed in

nosGal4::VP16; UAS-Bru egg chambers is due to a direct ef-

fect of Bru on gcl, rather than an indirect effect of reduced

Osk.

2.3. Ectopic Gcl expression perdures into embryogenesis

We examined embryos from aret heterozygous females for

disruption of Bru-mediated regulation of gcl expression in the

early embryo because aret homozygotes do not complete

oogenesis. We observed Gcl protein in the soma of embryos

from aretQB72/+ females, unlike in wildtype (Oregon-R, OrR) em-

bryos (Fig. 3A and C) in which Gcl expression is confined to

the pole cells. Expression of a genomic gcl transgene (P[gcl+],

previously published as hg130) (Jongens et al., 1994), induced

a similar increase in somatic Gcl (Fig. 3B, B’, J and Table 1).

These increases are statistically significant (p < 0.0005), and

are additive in that embryos from aretQB72/+; P[gcl+]/+ females

express somatic Gcl at 4-fold increase over background

(Fig. 3D, D’, J and Table 1). In these embryos, ectopic Gcl is

readily detectable at the nuclear envelopes of somatic nuclei

(Fig. 3D’, inset).

In contrast, RNA in situ hybridization indicated that gcl

RNA expression and localization are unaffected by loss of

one dose of aret (compare Fig. 3G–E and H–F). We hypothe-

sized therefore that the increase in somatic Gcl results from

an effect on translational control. To test this, we quantified

the Gcl signal around the somatic nuclei at the periphery of

several embryos, specifically excluding the pole cell region,

in order to isolate the effect of reducing Bru on somatic Gcl

expression. There is measurable somatic Gcl signal in the

soma of wildtype embryos, although it appears diffuse and

unlocalized (Fig. 3A’, inset), and the level is similar to back-

ground signal present in gclD embryos (see Fig. 5M and Table

3). However, we found that somatic Gcl increased by twofold

over background in embryos from aretQB72/+ females (Fig. 3J

and Table 1).

To further test whether increased somatic Gcl expression

is due to enhanced translation or increased mRNA levels,

Table 1 – gcl RNA and protein expression in aretQB72/+ embryossomatic nuclei (i.e., excluding pole cells) in stage 5 embryos aRpS15A RNA. Quantified signal was normalized against wildtycence was normalized against RNA signal, indicating that Gcl pand that this effect does not correlate to a similar increase in gc(*p < 0.0005, **p < 0.005).

Gcl signal ± SEM (N) Normalizedprotein

gcl/RRNA

OrR 75.95 ± 9.86 (9) 1.00 0.133

P[gcl+]/+ 150.85 ± 13.68 (6)* 1.99 0.256

aretQB72/+ 163.73 ± 13.11 (6)* 2.16 0.163

aretQB72/+; P[gcl+]/+ 331.04 ± 13.65 (5)** 4.36 0.340

we normalized somatic Gcl level against gcl RNA level as

determined by quantitative Northern blot (Fig. 3I). Based

on this analysis, ectopic Gcl in embryos from P[gcl+]/+ fe-

males is correlated to increased gcl RNA (Fig. 3K and Table

1). However, embryos from aretQB72/+ and aretQB72 /+;

P[gcl+]/+ females express approximately 1.7 times as much

protein (p < 0.005) per unit of RNA than wildtype or P[gcl+]/

+ embryos (Fig. 3K and Table 1). Thus, Gcl protein is specif-

ically overexpressed compared to the amount of gcl RNA

when Bru is reduced, consistent with a primary role for

Bru in repressing Gcl translation, as opposed to regulating

gcl RNA stability.

Anterior somatic expression of Gcl from a gcl-bcd3’UTR

transgene leads to repression of zygotic hkb transcription

(Jongens et al., 1994; Leatherman et al., 2002). Therefore,

repression of anterior hkb expression is a sensitive biologi-

cal assay for somatic Gcl. In accordance with the results de-

scribed above, we found that over 50% of embryos from

aretQB72/+ or P[gcl+ ]/+ females show reduced hkb RNA at

the anterior (Fig. 3L–N and Table 2), as do over 75% of em-

bryos from aretQB72 /+; P[gcl+ ]/+ females (p < 0.0001). Bru is

not present in early embryogenesis (Webster et al., 1997)

and thus we conclude that ectopic Gcl expression caused

by reduction of Bru repression during oogenesis perdures

into embryogenesis, and is sufficient to alter the expression

of hkb. Although hkb is an essential gene (Weigel et al.,

1990), embryonic lethality did not result from overexpress-

ing Gcl in this way, even in embryos that lacked paternal

hkb (hkb2/+) (data not shown). This may be because either

the reduction of hkb in aretQB72 /+; P[gcl+]/+ embryos, or a

specific reduction of hkb in its anterior domain, is not suf-

ficient to be lethal.

2.4. Recombinant Bruno binds directly to the gcl 3’UTR ina BRE-independent manner

The gcl 3’UTR contains a sequence similar to BREs present

in osk and grk (Fig. 4A, Kim-Ha et al., 1995; Filardo and

Ephrussi, 2003), with an imperfect BRE immediately up-

stream. Degenerate BREs are also present near the osk BREs.

We refer to the canonical BRE and the adjacent non-canonical

sequence as the ‘‘BRE-like’’ region (BRL, nucleotides 204–220

of the gcl 3’UTR). The gcl BRL is highly conserved among ele-

ven other Drosophila species (Fig. 4A), even in those where

. Quantification of aGcl immunofluorescence signal aroundnd of Northern blots of embryo extracts probed for gcl andpe signal for each category and protein immunofluores-rotein is overexpressed in embryos from aretQB72/+ femalesl RNA. p Values were determined by a Mann–Whitney U test

pS15A± SEM (N)

NormalizedRNA

Gclprotein/RNA

Normalizedprotein/RNA

5 ± 0.005 (3) 1.00 569.10 ± 73.88 1.00

1 ± 0.011 (3) 1.92 589.12 ± 53.43 1.04

7 ± 0.007 (3) 1.23 1000.33 ± 80.12** 1.76

5 ± 0.027 (3) 2.55 972.27 ± 40.10** 1.71

Table 3 – gcl RNA and protein expression in transgenic embryos. Quantification of aGcl immunofluorescence signal aroundsomatic nuclei (i.e., excluding pole cells) in stage 5 embryos and of Northern blots of embryo extracts probed for gcl andRpS15A RNA. Quantified signal was normalized against wildtype signal for each category and protein immunofluores-cence was normalized against RNA signal, indicating that deletion of the gcl BRL (P[gclDBRL]/+) does not cause an increase inGcl expression, but rather a subtle decrease when compared to either yw or P[gclcDNA]/+ embryos. P values weredetermined by a Mann–Whitney U test (compared to yw: *p < 0.0005, ***p < 0.05; compared to P[gclcDNA]/+: �p < 0.005).

Gcl signal ± SEM (N) Normalizedprotein

gcl/RpS15ARNA ± SEM (N)

NormalizedRNA

Gcl Protein/RNA Normalizedprotein/RNA

yw 139.28 ± 8.51 (16) 1.00 0.1472 ± 0.042 (4) 1.00 946.33 ± 57.79 1.00

gclD 139.37 ± 7.55 (11) – – – – –

P[gclcDNA]/+ 268.15 ± 16.78 (29)* 1.93 0.2556 ± 0.044 (12) 1.74 1049.23 ± 65.67 1.11

P[gclDBRL]/+ 274.17 ± 16.52 (32)* 1.97 0.3545 ± 0.057 (12) 2.41 773.36 ± 46.59 ***,� 0.82

Table 2 – Anterior hkb RNA expression pattern in aretQB72/+ embryos. Quantification of anterior staining patterns of hkb instage 5 embryos, as determined by hkb RNA in situ hybridization, indicating that anterior hkb staining is diminished inembryos from females that express P[gcl+] and in embryos from aretQB72/+ females, and that this effect is statisticallysignificant (p < 0.001). P values were determined by a chi-square contingency test.

Fly line Anterior hkb expression N

Strong (%) Reduced/irregular (%) Severely reduced (%)

OrR 67.5 31.7 0.8 120

P[gcl+]/+ 47.0 51.8 1.2 83

aretQB72/+ 48.4 49.5 2.1 95

aretQB72/+; P[gcl+]/+ 23.3 66.2 10.5 134

p < 0.0001.

M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6 509

much of the gcl 3’UTR has diverged substantially from D.

melanogaster.

We used an in vitro UV crosslinking assay to determine

whether Bru can bind to the gcl 3’UTR and to its BRL. Radiola-

beled gcl 3’UTR RNA probe (gcl 3’UTR, Fig. 4B and C) was bound

by a single protein from bacterial extracts expressing GST-Bru

that migrates on SDS–PAGE at the predicted molecular weight

of GST-Bru (96 kDa) and comigrates with a band detected by

aBru on immunoblots (Fig. 4C and data not shown). The same

band was recovered with a radiolabeled fragment of the osk

3’UTR that contains 4 BREs (oskAB, Fig. 4C). Control lysates

that express GST alone did not contain the 96 kDa protein.

Two non-specific, faster migrating bands are detected in all

samples, including those prepared with GST lysate alone

(Fig. 4C–F).

We determined whether GST-Bru binding could be com-

peted by various unlabeled RNAs. GST-Bru binding to 32P-gcl

3’UTR RNA is competed efficiently by unlabeled oskAB RNA

(Fig. 4D and E), and with unlabeled gcl 3’UTR RNA, but not

with an unrelated unlabeled RNA (a-tubulin84B (atub) 3’UTR,

Fig. 4D and E). Surprisingly, an unlabeled RNA oligomer con-

taining the gcl BRL region (gclBRL) was not able to compete

GST-Bru binding to 32P-gcl 3’UTR RNA (Fig. 4D), while unla-

beled gcl 3’UTR RNA with the BRL deleted (gclDBRL, Fig. 4E)

competed with GST-Bru binding to 32P-gcl 3’UTR RNA as effec-

tively as the full-length gcl 3’UTR. Taken together, these data

indicate that Bru binds to the gcl 3’UTR independent of the

BRL.

Unlike gcl, all other Bru targets contain multiple (2–6) BREs

(Kim-Ha et al., 1995; Filardo and Ephrussi, 2003; Sugimura and

Lilly, 2006; Wang and Lin, 2007). We tested the ability of GST-

Bru to bind to multiple tandem copies of the gcl BRL. Unla-

beled RNA oligomers including two or four copies of the gcl

BRL are able to compete GST-Bru binding to the full length

gcl 3’UTR (gcl2XBRL, gcl4XBRL, Fig. 4F), indicating that the gcl

BRL is competent for Bru binding when present in multiple

copies.

2.5. The gcl BRL is not required for translationalrepression

A transgene that encodes the wildtype gcl cDNA (P[gclcDNA])

and a modification of this transgene with the BRL deleted

(P[gclDBRL], Fig. 5A) were used to investigate whether the gcl

BRL is required to maintain translational repression of gcl

in vivo. Gcl expression from the P[gclcDNA] transgene in a gclD

background largely recapitulates the wildtype pattern during

oogenesis (Fig. 5B and D). Gcl expression in gclD; P[gclDBRL] egg

chambers is also very similar to wildtype expression, consis-

tent with our in vitro experiments (Fig. 5C and E). Ectopic Gcl

expression in the follicle cells seen in Fig. 5B and E occurs

with several transgenes, including P[gclcDNA], suggesting that

it is 3’UTR independent and may be an effect of the Hsp83

promoter.

We observed no difference in Gcl expression between

P[gclcDNA] and P[gclDBRL] comparable to that seen between

wildtype and aretQB72/+ oocytes, or between nosGal4::VP16;

UAS-Bru and nosGal4::VP16 oocytes. We therefore conclude

that the gcl BRL is not required to maintain translational

repression of gcl in the oocyte. We did see a transient small in-

Fig. 4 – GST-Bru binds to the gcl 3’UTR. (A) Schematic diagram of the gcl cDNA, indicating the position of the BRE-like region

(BRL). The BRL contains a BRE consensus sequence (boxed in red) and an adjacent region of similar sequence (boxed in blue),

and is conserved among all twelve Drosophila species with sequenced genomes. (B) Schematic diagram of RNA probes used in

UV-crosslinking experiments. The gcl3’UTR and gclDBRL probes (shown in red) span the entire gcl 3’UTR and the gclBRL probe

(shown in green) spans the BRL and immediately adjacent region. (C) 32P-labeled oskAB and gcl3’UTR probes crosslinked to

bacterial lysates expressing GST or GST-Bru, and separated by SDS–PAGE. GST-Bru, but not GST, can be crosslinked to both

radiolabelled probes. (D–F) 32P-labeled gcl3’UTR RNA crosslinked to bacterial lysates expressing GST-Bru and separated by

SDS–PAGE. Competitor RNAs (as indicated) were added at up to 500X molar excess before addition of 32P-gcl 3’UTR probes.

Binding of 32P-labeled gcl3’UTR RNA by GST-Bru is competed by unlabeled oskAB, gcl3’UTR and gclDBRL, but not by single-copy

gclBRL or atub3’UTR RNA oligos (D, E). The gcl BRL is, however, able to compete with 32P-labeled gcl3’UTR RNA when present in

multiple tandem copies (F).

510 M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6

crease in Gcl in the P[gclDBRL] oocyte cytoplasm at late stage 8,

but this was not sustained through stage 10B (Fig. 5E), and Gcl

accumulated at similar levels at the oocyte nuclear mem-

brane in late stage 8 or stage 10B whether or not the BRL

was present (Fig. 5B–E). P[gclDBRL] fully rescued gcl RNA expres-

sion and accumulation at the posterior pole (Fig. 5F and H–K).

Embryos from gclD; P[gclDBRL] females formed an average of

22.5 ± 0.6 pole cells at syncytial blastoderm stage, which is

comparable to the average of 19.0 ± 0.6 pole cells formed in

embryos from gclD; P[gclcDNA] females, indicating that the

BRL sequence is not required for the activity of gcl in pole cell

specification.

We measured the level of ectopic Gcl expression induced

in the soma of wildtype embryos that also express either

P[gclcDNA] or P[gclDBRL], and found, in both cases, a similar in-

crease to that observed in P[gcl+]/+ embryos (Fig. 5L–O and Ta-

ble 3). P[gclDBRL]/+ embryos also do not show an increase in the

Gcl protein/RNA ratio compared to either P[gclcDNA]/+ or wild-

type embryos (Fig. 5G and P; Table 3); in fact, a small decrease

was observed.

2.6. Recombinant Bru binds the gcl A region of the gcl3’UTR

As there are no other sequences in the gcl 3’UTR that are

obvious candidates for Bru binding, 32P-labeled probes cover-

ing overlapping regions of the gcl 3’UTR were used to map the

Bru binding site (Fig. 6A). GST-Bru bound to radiolabeled gcl A

(nucleotides 1–210 of the gcl 3’ UTR) with similar affinity as to

the full-length gcl 3’ UTR (Fig. 6B). gcl A excludes the canonical

BRE sequence of the BRL. GST-Bru did not bind with high

affinity to probes that correspond to other regions of the gcl

3’UTR (gcl B-D and gcl BRL, Fig. 6B). GST-Bru binding to the

gcl 3’UTR could be efficiently competed with unlabeled gcl A,

but not with unlabeled RNAs that cover the rest of the gcl

3’UTR (Fig. 6C). We attempted to further subdivide gcl A, and

also observed GST-Bru binding to nucleotides 1–130 (gcl A1,

Fig. 6D), and, with lower affinity, to nucleotides 59–160 (gcl

A3). GST-Bru binding to the 3’ UTR is abolished upon gcl A

deletion, reduced when gcl A1 is deleted, and not affected

when gcl A2 is deleted (nucleotides 123–290). Taken together,

Fig. 5 – The gcl BRL is not required for normal Gcl expression during oogenesis and embryogenesis. (A) Schematic diagram of

P[gclcDNA] and P[gclDBRL] transgenic constructs. (B–E) Late stage 8 (B,C) and stage 10B (D,E) egg chambers from gclD; P[gclcDNA]

(B,D) or gclD; P[gclDBRL] (C,E) females immunostained with aGcl (monochrome, projections of images taken at multiple confocal

planes). Deletion of the BRL has no detectable effect on expression of Gcl protein from the transgene. (F) Northern blot of total

RNA from ovaries from wildtype or gclD females or from two independent transgenic lines carrying P[gclcDNA] or P[gclDBRL] in a

gclD background, probed with 32P-labeled gcl cDNA (upper panel). gcl RNA expression is rescued by P[gclcDNA] and P[gclDBRL]

transgenes. G, Northern blot of total RNA from stage 1 to 5 embryos produced from from wildtype or gclD females or from two

independent transgenic lines carrying P[gclcDNA] or P[gclDBRL] in a wildtype background probed with 32P-labeled gcl cDNA,

indicating that gcl RNA is overexpressed in yw embryos that express P[gclcDNA] and P[gclDBRL] transgenes (upper panel). For

both F and G the same blot probed with 32P-labeled RpS15A ribosomal protein cDNA is shown as a loading control (lower

panels). H-K, gcl RNA in situ hybridization of pre-blastoderm embryos produced by Oregon-R (H), gclD (I), gclD; P[gclcDNA] (J), or

gclD; P[gclDBRL] (K) females indicating that P[gclcDNA] and P[gclDBRL] transgenes equally rescue gcl RNA expression and

localization in embryos. (L–O) Stage 5 embryos produced by yw (L), gclD (M), yw; P[gclcDNA] (N), or yw; P[gclDBRL] (O) females

immunostained for aGcl (monochrome). Inset images (upper right) are 4-fold magnifications of somatic nuclei. (P) Graph

showing somatic Gcl immunofluorescence measured from 11 to 32 embryos of each normalized against RNA expression, as

determined by quantitative Northern blots (4–12 for each genotype). Normalized values are shown at the top of each bar,*p < 0.0005, ***p < 0.05, �p < 0.005 relative to P[gclcDNA]/+. Gcl protein expression is slightly reduced relative to RNA expression in

yw; P[gclDBRL] embryos. The data shown in p are also presented in Table 3.

M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6 511

Fig. 6 – Bru binds to a novel sequence in the gcl 3’UTR. (A) Schematic diagram of RNA probes used in UV crosslinking

experiments. The first nucleotide downstream of the gcl stop codon corresponds to nucleotide 1. The gcl3’UTR and gclDBRL

probes (shown in red) span the entire gcl 3’UTR and the gclBRL probe (shown in green) spans the BRL and immediately

adjacent region. The gclA-D series of probes is shown in blue and gclA1-A3 probes, which subdivide gclA, are shown in brown.

(B) 32P-labeled probes covering the gcl 3’UTR crosslinked to bacterial lysate expressing GST-Bru and separated by SDS–PAGE.

GST-Bru binds to gclA RNA with much greater affinity than to gclB, C, or D, or gclBRL. C, 32P-labeled gcl 3’UTR RNA crosslinked to

bacterial lysate expressing GST-Bru and separated by SDS–PAGE. Competitor RNAs were added up to 500· molar excess before

addition of 32P-gcl 3’UTR probes. GST-Bru binding to 32P-labeled gcl3’UTR RNA probe is competed by unlabelled gclA, but not

significantly by unlabelled gclB, (C) or D RNA. (D_ 32P-labeled probes including or deleting part or all of the gclA region

crosslinked to bacterial lysate expressing GST-Bru and separated by SDS–PAGE. GST-Bru binds to the gclA1 RNA probe with

greater affinity than to gclA2 or A3, although some binding is observed to gclA3. Binding to the full length 3’UTR is diminished

by deletion of regions gclA and A1, but not by deletion of A2. (E) Sequence alignment of a segment of the gclA1 region that is

related to the Nanos response element and other predicted Pumilio binding motifs. (F) 32P-labeled probes crosslinked to GST,

GST-Bru, GST-BruN or GST-BruC and separated by SDS–PAGE. * indicates a non-specific band present in all lysates. 32P-labeled

gcl3’UTR and gclA probes are bound by GST-Bru and GST-BruC, but not by GST-BruN.

512 M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6

M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6 513

these results indicate that Bru interacts with the gcl 3’UTR

through an association with its first 130 nucleotides, a region

which lacks any sequence that resembles a BRE. This region

contains a U-rich sequence that is well-conserved among se-

ven Drosophila species (Fig. 6E). The UUGUAAAUU motif at the

3’ end of the conserved element is also well conserved among

more divergent species (i.e., Drosophila persimilis, willistoni,

mojavensis, virilis and grimshawii). This motif contains a Pumi-

lio binding motif that was derived from in vitro selection

(UGUAAAU, Gerber et al., 2006). The sequence GUUGU-(N)7-

UUGUA is closely related to the consensus for a Nos response

element (NRE, Zamore et al., 1999), and further suggests this

region may function in RNA-protein interactions.

Bru binds to the osk 3’UTR via two separate protein do-

mains that include the two N-terminal and the C-terminal

RNA recognition motifs (RRMs), respectively (Snee et al.,

2008). Radiolabeled RNA probes of the oskAB region, gcl

3’UTR, gclA region and atub 3’UTRs were used in UV crosslink-

ing assays with truncated Bru constructs to test whether Bru

binds gcl via the same or a different mechanism. GST-BruN

contains the first 429 amino acids of Bru, including the first

two RRMs, and GST-BruC contains amino acids 422–604,

including RRM3. While the gcl 3’UTR and gclA probes are effi-

ciently bound by full length GST-Bru, and by GST-BruC, they

are not bound by GST-BruN (Fig. 6F). In contrast, oskAB is effi-

ciently bound by both GST-BruN and GST-BruC, confirming

previously published results (Snee et al., 2008). These data

indicate that Bru does not require RRM1 and 2 to bind to the

gcl 3’UTR and suggest that unlike for osk, Bru binding to gcl ap-

pears to be mediated by RRM3 alone.

3. Discussion

The data presented here establish that Bru represses

expression of gcl outside the pole plasm during mid-oogene-

sis, and that perturbation of this regulation prevents restric-

tion of Gcl to the embryonic germline and affects zygotic

gene expression in early embryogenesis. Substantial evidence

has implicated Bru as a translational repressor of target

mRNAs, and our data support a similar role for Bru in regulat-

ing gcl. A caveat regarding this conclusion is that, while the

histochemically stained in situ hybridizations we used re-

vealed no major obvious changes in gcl mRNA localization

or level in aret mutant oocytes compared with wild type, they

do not permit us to exclude more subtle changes, as histo-

chemical staining does not show the boundary of the nurse

cells and oocyte very clearly, nor is it as sensitive as fluores-

cent methods of detection. Interestingly, Bru-mediated

repression of gcl mRNA proceeds through a BRE-independent

mechanism, and Bru can bind efficiently to the gcl 3’UTR se-

quence using only RRM3. These observations challenge previ-

ously determined criteria for Bru target mRNAs and suggest

that there could be more potential Bru targets than previously

predicted (Wang and Lin, 2007).

Bru-mediated repression is mainly directed at gcl mRNA in

the oocyte cytoplasm. Ectopic expression of Gcl in aret hetero-

zygous egg chambers, combined with reduction of Gcl expres-

sion in egg chambers that overexpress Bru, implicates aret in

a genetic relationship with gcl. The reciprocal phenotypes, in

addition to the observation that recombinant Bru binds di-

rectly to the gcl 3’UTR strongly suggest that Bru acts directly

on gcl mRNA and does not act indirectly through another

Bru target, such as osk. Consistent with this hypothesis, there

is no change in Gcl expression in egg chambers from osk mu-

tant females.

Misexpression of Gcl observed in aret heterozygous egg

chambers perdures into embryogenesis, indicating that Bru-

mediated regulation is required to refine Gcl expression in

the early embryo. Thus, as is the case with other pole plasm

components, such as osk and nos, restriction of Gcl expression

to the pole plasm in the early embryo is achieved in part by

repression of somatic expression (reviewed in Vardy and

Orr-Weaver, 2007). Analysis of the complete role of Bru in

embryonic patterning has been confounded by its early role

in oogenesis, as embryos cannot be recovered from aret

homozygotes. In embryos from aret heterozygotes, ectopic

Gcl results in suppression of the anterior domain of zygotic

hkb expression, but this effect is not severe enough to cause

embryonic lethality. It is possible that complete loss of Bru-

mediated regulation of gcl and other targets would produce

a stronger phenotype. Alternatively, Bru may be partially

redundant, and gcl expression may also be repressed by addi-

tional factors during oogenesis and embryogenesis.

Although all previously identified Bru targets contain be-

tween two and six BREs, Bru-binding and regulation has been

mapped to the BRE only for osk and Sxl (Kim-Ha et al., 1995;

Webster et al., 1997; Wang and Lin, 2007). While Bru binds

to both grk and CycA 3’UTRs and regulates translation of both

RNAs, neither binding nor regulation has been specifically

linked to the BRE for either target (Filardo and Ephrussi,

2003; Sugimura and Lilly, 2006). As neither Bru binding to gcl

RNA in vitro, nor translational repression of gcl in vivo, require

the gcl BRL, this suggests that Bru-mediated regulation could

act through multiple target sequences, and leads us to specu-

late that there are more Bru-target RNAs than previously rec-

ognized (Wang and Lin, 2007). Indeed, the gclA1 region

contains a U-rich element that resembles predicted binding

sites for translational repressor Pumilio (Zamore et al., 1999;

Gerber et al., 2006), but could conceivably serve as a substitute

for a BRE. This U-rich sequence is conserved in the gcl 3’UTR

among seven Drosophila genomes, and the terminal motif

(UUGUAAAUU) is conserved among more divergent species

as well. Surprisingly, we found that Bru is able to bind the

gcl BRL when it is present in either two or four tandem copies.

Consistent with this result, reduction of copies of the osk BREs

decreases Bru binding (data not shown and Kim-Ha et al.,

1995). This leads us to conclude that a functional BRE must

contain more than one copy of the element, perhaps because

cooperative interactions stabilize Bru binding. The sequence

conservation of the gcl BRL among many Drosophila species

suggests that it is functional, and the modest decrease in

Gcl expression observed in embryos that express P[gclDBRL]

may indicate that the BRL has a Bru-independent positive reg-

ulatory function.

Bru was originally identified by its ability to bind to the osk

3’UTR through RRM3, but recent evidence indicates that the

N-terminal portion of Bru also contributes to osk binding

(Kim-Ha et al., 1995; Snee et al., 2008). In contrast, our data

indicate that a region containing RRM3 alone mediates Bru

514 M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6

binding to gcl, which is also consistent with our in vivo data

obtained from aretQB72, an allele that produces a form of Bru

that lacks RRM3. However, as Gcl overexpression is somewhat

stronger in Df(2L)esc-P3-0/+ oocytes (in which the aret gene is

completely deleted), we cannot exclude the possibility that

other portions of Bru may also act to repress Gcl expression.

Other proteins that contain three RRMs in a similar organiza-

tion to Bru, such as Sxl, PABP and HuD, form structures that

are hypothesized to bind short RNA sequence elements by

contact with conserved residues in both RRM1 and RRM2

(Deo et al., 1999; Handa et al., 1999; Wang and Tanaka Hall,

2001). Vertebrate Bruno-like proteins conserve the organiza-

tion and much of the sequence of the three RRM domains

found in Drosophila Bru, but bind to U-rich elements (UREs)

via RRM1 and RRM2, or by a motif in the divergent linker re-

gion between RRM2 and RRM3 (Good et al., 2000; Suzuki

et al., 2002; Delaunay et al., 2004). In each of these cases,

the role of RRM3 is unclear. Bru binding to gcl via RRM3 may

enable the use of a novel target binding site and contact to

gcl via RRM3 could also leave RRM1 and RRM2 free to contact

other RNAs, and/or leave other domains available for protein–

protein interactions (Lunde et al., 2007). A precedent for this

type of multiple contact binding is polypyrimidine tract bind-

ing protein (PTB), a multi-RRM containing protein where each

RRM binds to RNA with a different sequence specificity

(Oberstrass et al., 2005). Bru regulation of gcl could also in-

volve different protein partners from those required for regu-

lation of targets that require RRM1 and RRM2 for the Bru-BRE

interaction. The alternative RNA-protein complex created by

binding to gcl could thus enable a Bru-mediated mechanism

of translational repression that is distinct from those used

for repression of osk.

4. Experimental procedures

4.1. Fly stocks

gclD and P[gcl+] (previously hg130) fly stocks were gifts from

Tom Jongens (Jongens et al., 1994; Robertson et al., 1999);

aretQB72/CyO and aretPA62/CyO flies were obtained from Haifan

Lin (Webster et al., 1997); tudtux/CyO-hs-hid flies were obtained

from Travis Thomson (Thomson and Lasko, 2004); osk54 and

P[UASBru] were obtained from Paul Macdonald (Lehmann

and Nusslein-Volhard, 1986; Snee et al., 2007); and the nos-

GAL4::VP16 driver was provided by Ruth Lehmann (Van Doren

et al., 1998). vas1, Df(2L)esc-P3–0 and Df(3R)p-XT103 were ob-

tained from the Bloomington Stock Centre Bloomington, Indi-

ana, USA (Schupbach and Wieschaus, 1986; Lehmann and

Nusslein-Volhard, 1991).

4.2. Transgenic fly lines

P[gclDBRL] was constructed using the Invitrogen GeneTailor

kit with P[gclcDNA] (a gift from Tom Jongens and Aron Jaffe) as a

template. The resulting cDNA was recombined into pCasper4

converted to a destination vector (Invitrogen, Carlsbad, CA,

USA). P[gclcDNA] and P[gclDBRL] transgenic fly stocks were estab-

lished by standard microinjection and genetic methods.

Transgene expression was analyzed by Northern blotting.

4.3. Northern blot

Total RNA was extracted from ovaries or 0–4 h embryos

using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and fur-

ther purified by low pH phenol extraction and ethanol precipi-

tation. RNA was separated on a denaturing 1% agarose gel

(formaldehyde/MOPS), transferred to a nylon membrane by os-

motic transfer and crosslinked to the membrane using a Strat-

alinker (Stratagene, La Jolla, CA, USA). Membranes were probed

with a 32P-labeled gcl or RpS15A cDNA probe using ExpressHyb

Hybridization solution (Clontech Laboratories, Inc., Mountain

View, CA, USA). Pre-hybridization, hybridization, and washing

followed standard protocols. Signal was detected using a Storm

Phosphorimager and quantified using ImageQuant software

(GE Healthcare, Piscataway, NJ, USA).

4.4. Production of polyclonal antibodies

A plasmid expressing full-length Gcl fused to glutathione-

S-transferase (GST) was constructed by recombination of

pDEST15 with pENTR/TEV/D-TOPO (Invitrogen, Carlsbad,

California, USA) containing the gcl coding region (1710 nucle-

otides, CG8411-RA). GST-Gcl protein was expressed in BL21-A1

E. coli cells (Invitrogen, Carlsbad, California, USA) and purified

by separation on a polyacrylamide gel and electroelution.

Polyclonal antibodies were raised in rabbits by standard

methods. Anti-Gcl antibodies were purified before use by

incubating overnight at 4 �C with an equal volume of fixed

ovaries or embryos from gclD females.

4.5. Immunoblots

Immunoblots were performed using ovaries from well-fed,

3–5 day old, mated females or embryos collected for 0–2, 2–4

and 0–4 h after egg lay. The tissue sample was crushed directly

into 3· SDS–PAGE loading buffer immediately after collection.

Samples were separated by 8% SDS–PAGE, transferred to a

nitrocellulose membrane, probed using a standard protocol

and detected by chemiluminescence. Anti-Gcl antibodies (rab-

bit) were used at 1/2000 and anti-eIF4A antibodies (rabbit) were

used at 1/10,000. HRP-conjugated goat anti-rabbit antibodies

(GE Healthcare, Piscataway, NJ, USA) were used at 1/1000.

4.6. Immunostain and RNA in situ hybridization

Ovaries and embryos were immunostained according to

published protocols (Kobayashi et al., 1999; Findley et al.,

2003). Anti-Gcl (aGcl) antibodies (rabbit) were used at 1/2000,

anti-Osk (aOsk) antibodies (rat) were used at 1/2000, MAb414

(Nuclear Pore Complex, mouse, Covance Research Products,

Inc., Berkeley, CA, USA) was used at 1/2000, anti-Vas (aVas)

antibodies (rat) were used at 1/2000, and anti-Bruno (aBru)

antibodies (rabbit) were used at 1/2000 (Webster et al., 1997).

Alexa Fluor-conjugated anti-rabbit and anti-rat secondary

antibodies (Molecular Probes, Eugene, Oregon, USA) were de-

pleted against Or-R ovaries or embryos and used at 1/500.

Samples for immunofluorescence were incubated with 1 lg/

mL DAPI and rinsed several times before mounting on slides

using SlowFade Antifade kit (Molecular Probes, Eugene, Ore-

gon, USA) and visualized using an LSM 510 confocal micro-

M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6 515

scope (Carl Zeiss, Inc., Jena, Germany). RNA in situ hybridiza-

tion protocols for embryos and ovaries were as described

(Kobayashi et al., 1999). DIG-labeled probes were generated

by in vitro transcription from gcl and hkb cDNA templates

(CG8411 and RE60512, respectively) using the DIG RNA label-

ing kit (Roche Applied Science, Penzberg, Germany). P values

for quantification of RNA in situ hybridization samples were

calculated using a chi-square contingency test.

4.7. Quantitative analysis

Quantitative data of immunofluorescence was collected

using Volocity software (Improvision, Waltham, MA, USA)

from images collected in a single plane of focus by confocal

microscopy. The immunofluorescent signal around 50–80 so-

matic nuclei at the periphery of five to 15 whole mount em-

bryos (syncytial blastoderm stage) immunostained with

aGcl, was quantified and averaged for each embryo. Back-

ground values from outside the sample periphery were sub-

tracted from each sample. Averages, p values and standard

error were calculated based on whole embryo data, rather

than individual data points, to prevent bias due to embryo

size. Quantitative data was collected from Northern blots

using ImageQuant software (GE Healthcare, Piscataway, NJ,

USA). Signals from samples in three to six Northern blots

probed with radiolabeled gcl and RpS15A cDNA probes were

quantified. Background values from peripheral membrane

positions and the gclD sample were subtracted and gcl signal

was normalized against RpS15A signal for each sample. p Val-

ues were calculated for all quantitative samples using a non-

parametric rank sum test (Mann–Whitney U test).

4.8. UV crosslink assay

UV crosslink assays were performed as in Kim-Ha et al.

(1995), with minor modifications. All samples contained

10 lg each of heparin and yeast tRNA as a blocking agent

and the protein source in all cases was bacterial extract

expressing GST fusion proteins from pDEST15-derived vec-

tors. Protein extracts were prepared from BL21-A1 bacterial

cultures expressing GST (25 pmol), GST-Bru (2 pmol, contain-

ing nucleotides 330–2138 of CG3176-RA), GST-Bru-N (10 pmol,

containing nucleotides 330–1617 of CG3176-RA) or GST-Bru-C

(25 pmol, containing nucleotides 1596–2138 of CG3167-RA).

Cultures were grown and protein expression induced as de-

scribed above. Cells were recovered by centrifugation, lysed

and the soluble lysate was used in crosslinking reactions.

Radiolabeled RNA probes were transcribed in vitro from

DNA templates composed of nucleotides 1–501 (gcl 3’UTR),

1–210 (gcl A), 123–390 (gcl B), 190–310 (gcl C), 290–501 (gcl D),

1–130 gcl A1), 123–210 (gcl A2), 59–160 (gcl A3), 210–501

(gclDA), 130–501 (gclDA1), and 1–501 with 123–210 deleted

(gclDA2) of the gcl 3’UTR (1939–2440, 1939–2149, 2062–2329,

2129–2249 and 2229–2440 of CG8411-RA); 117–314 of the oskar

3’UTR (osk AB, 1953–2156 of CG10901-RA) and nucleotides

1561–1813 of CG1913-RA (a-tub 3’UTR). Probe concentration

was adjusted to 50 fmol per reaction. Unlabeled RNA oligo-

mers were transcribed in vitro from DNA templates of nucle-

otides 1–501 (gcl 3’UTR), 1–501 with 200–226 deleted,

(gclDBRL), 1–210 (gcl A), 123–290 (gcl B), 190–310 (gcl C),

290–501 (gcl D) and 173–320 (gcl BRL) of the gcl 3’UTR,

(1939–2440, 1939–2149, 2062–2329, 2129–2249,2229–2440 and

2112–2259 of CG8411-RA; 117–320 (osk AB) of the oskar

3’UTR (1953–2156 of CG10901-RA) and nucleotides

1561–1813 of CG1913-RA (a-tub 3’UTR). Unlabeled oligomer

concentration was adjusted to 5, 15 or 25 pmol per reaction.

4.9. Sequence analysis

All Drosophila sequences originated from FlyBase, Version

FB2008_06 (http://www.flybase.org, Wilson et al., 2008). Non-

melanogaster sequences were identified using FlyBlast (http://

www.flybase.org/blast, Drosophila 12 Genomes Consortium,

2007; Wilson et al., 2008). Sequence alignments were per-

formed using ClustalW2 (http://www.clustal.org; Larkin

et al., 2007).

Acknowledgements

We thank Tom Jongens and Aron Jaffe for the contribution of an

unpublished vector (pCasper4gclcDNA), as well as published fly

lines and antibodies. We also thank P. Macdonald, R. Lehmann,

H. Lin and T. Thomson for fly lines. Thanks to B. Hu for technical

assistance, to B. Leung for advice regarding statistical analysis

and to C. Gamberi, C. Laplante, J. Kugler, G. Tettweiler and T.

Thomson for critical reading of this manuscript. This work

was supported by a grant to P.L. from the Natural Sciences

and Engineering Research Council and from grant

R01HD036631 from the National Institute of Child Health and

Development. J.M. was supported by fellowships from the

Canadian Institutes of Health Research and from McGill

University.

R E F E R E N C E S

Brand, A.H., Perrimon, N., 1993. Targeted gene expression as ameans of altering cell fates and generating dominantphenotypes. Development 118, 401–415.

Chekulaeva, M., Hentze, M.W., Ephrussi, A., 2006. Bruno acts as adual repressor of oskar translation, promoting mRNAoligomerization and formation of silencing particles. Cell 124,521–533.

Delaunay, J., Le Mee, G., Ezzeddine, N., Labesse, G., Terzian, C.,Capri, M., Ait-Ahmed, O., 2004. The Drosophila Bruno paralogueBru-3 specifically binds the EDEN translational repressionelement. Nucleic Acids Res. 32, 3070–3082.

Deo, R.C., Bonanno, J.B., Sonenberg, N., Burley, S.K., 1999.Recognition of polyadenylate RNA by the poly(A)-bindingprotein. Cell 98, 835–845.

Drosophila 12 Genomes Consortium, 2007. Evolution of genesand genomes on the Drosophila phylogeny. Nature 450, 203–218.

Filardo, P., Ephrussi, A., 2003. Bruno regulates gurken duringDrosophila oogenesis. Mech. Dev. 120, 289–297.

Findley, S.D., Tamanaha, M., Clegg, N.J., Ruohola-Baker, H., 2003.Maelstrom, a Drosophila spindle-class gene, encodes a proteinthat colocalizes with Vasa and RDE1/AGO1 homolog,Aubergine, in nuage. Development 130, 859–871.

Forrest, K.M., Gavis, E.R., 2003. Live imaging of endogenous RNAreveals a diffusion and entrapment mechanism for nanosmRNA localization in Drosophila. Curr. Biol. 13, 1159–1168.

516 M E C H A N I S M S O F D E V E L O P M E N T 1 2 6 ( 2 0 0 9 ) 5 0 3 – 5 1 6

Gerber, A.P., Luschnig, S., Krasnow, M.A., Brown, P.O., Herschlag,D., 2006. Genome-wide identification of mRNAs associatedwith the translational regulator PUMILIO in Drosophilamelanogaster. Proc. Natl. Acad. Sci. USA 103, 4487–4492.

Good, P.J., Chen, Q., Warner, S.J., Herring, D.C., 2000. A family ofhuman RNA-binding proteins related to the Drosophila Brunotranslational regulator. J. Biol. Chem. 275, 28583–28592.

Handa, N., Nureki, O., Kurimoto, K., Kim, I., Sakamoto, H.,Shimura, Y., Muto, Y., Yokoyama, S., 1999. Structural basis forrecognition of the tra mRNA precursor by the Sex-lethalprotein. Nature 398, 579–585.

Jongens, T.A., Ackerman, L.D., Swedlow, J.R., Jan, L.Y., Jan, Y.N.,1994. Germ cell-less encodes a cell type-specific nuclear pore-associated protein and functions early in the germ-cellspecification pathway of Drosophila. Genes Dev. 8, 2123–2136.

Jongens, T.A., Hay, B., Jan, L.Y., Jan, Y.N., 1992. The germ cell-lessgene product: a posteriorly localized component necessary forgerm cell development in Drosophila. Cell 70, 569–584.

Kim-Ha, J., Kerr, K., Macdonald, P.M., 1995. Translationalregulation of oskar mRNA by bruno, an ovarian RNA-bindingprotein, is essential. Cell 81, 403–412.

King, R.C., 1970. Ovarian Development in Drosophila melanogaster.Academic Press, New York.

Kobayashi, S., Amikura, R., Nakamura, A., Lasko, P.F., 1999.Techniques for analyzing protein and RNA distribution inDrosophila ovaries and embryos at structural andultrastructural resolution. In: Richter, J. (Ed.), Advances inMolecular Biology: A Comparative Methods Approach to theStudy of Oocytes and Embryos. Oxford University Press,Oxford, pp. 426–445.

Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R.,McGettigan, P.A., McWilliam, H., Valentin, F., Wallace, I.M.,Wilm, A., Lopez, R., Thompson, J.D., Gibson, T.J., Higgins, D.G.,2007. Clustal W and Clustal X version 2.0. Bioinformatics 23,2947–2948.

Lasko, P.F., Ashburner, M., 1990. Posterior localization of vasaprotein correlates with, but is not sufficient for, pole celldevelopment. Genes Dev. 4, 905–921.

Leatherman, J.L., Levin, L., Boero, J., Jongens, T.A., 2002. Germ cell-less acts to repress transcription during the establishment ofthe Drosophila germ cell lineage. Curr. Biol. 12, 1681–1685.

Lehmann, R., Nusslein-Volhard, C., 1986. Abdominalsegmentation, pole cell formation, and embryonic polarityrequire the localized activity of oskar, a maternal gene inDrosophila. Cell 47, 141–152.

Lehmann, R., Nusslein-Volhard, C., 1991. The maternal gene nanoshas a central role in posterior pattern formation of theDrosophila embryo. Development 112, 679–691.

Lunde, B.M., Moore, C., Varani, G., 2007. RNA-binding proteins:modular design for efficient function. Nat. Rev. Mol. Cell. Biol.8, 479–490.

Nakamura, A., Amikura, R., Hanyu, K., Kobayashi, S., 2001. Me31Bsilences translation of oocyte-localizing RNAs through theformation of cytoplasmic RNP complex during Drosophilaoogenesis. Development 128, 3233–3242.

Nakamura, A., Sato, K., Hanyu-Nakamura, K., 2004. Drosophila cup isan eIF4E binding protein that associates with Bruno andregulates oskar mRNA translation in oogenesis. Dev. Cell 6,69–78.

Oberstrass, F.C., Auweter, S.D., Erat, M., Hargous, Y., Henning, A.,Wenter, P., Reymond, L., Amir-Ahmady, B., Pitsch, S., Black,D.L., Allain, F.H.T., 2005. Structure of PTB bound to RNA:specific binding and implications for splicing regulation.Science 309, 2054–2057.

Parisi, M.J., Deng, W., Wang, Z., Lin, H., 2001. The arrest gene isrequired for germline cyst formation during Drosophilaoogenesis. Genesis 29, 196–209.

Robertson, S.E., Dockendorff, T.C., Leatherman, J.L., Faulkner, D.L.,Jongens, T.A., 1999. Germ cell-less is required only during theestablishment of the germ cell lineage of Drosophila and hasactivities which are dependent and independent of itslocalization to the nuclear envelope. Dev. Biol. 215, 288–297.

Schupbach, T., Wieschaus, E., 1986. Germline autonomy ofmaternal-effect mutations altering the embryonic bodypattern of Drosophila. Dev. Biol. 113, 443–448.

Schupbach, T., Wieschaus, E., 1991. Female sterile mutations onthe second chromosome of Drosophila melanogaster. II.Mutations blocking oogenesis or altering egg morphology.Genetics 129, 1119–1136.

Snee, M., Benz, D., Jen, J., Macdonald, P.M., 2008. Two distinctdomains of Bruno bind specifically to the oskar mRNA. RNABiol. 5, 1–9.

Snee, M.J., Harrison, D., Yan, N., Macdonald, P.M., 2007. A latephase of Oskar accumulation is crucial for posteriorpatterning of the Drosophila embryo, and is blocked by ectopicexpression of Bruno. Differentiation 75, 246–255.

Sugimura, I., Lilly, M.A., 2006. Bruno inhibits the expression ofmitotic cyclins during the prophase I meiotic arrest ofDrosophila oocytes. Dev. Cell 10, 127–135.

Suzuki, H., Jin, Y., Otani, H., Yasuda, K., Inoue, K., 2002. Regulationof alternative splicing of alpha-actinin transcript by Bruno-likeproteins. Genes Cells 7, 133–141.

Thomson, T., Lasko, P., 2004. Drosophila tudor is essential for polargranule assembly and pole cell specification, but not forposterior patterning. Genesis 40, 164–170.

Van Doren, M., Williamson, A.L., Lehmann, R., 1998. Regulation ofzygotic gene expression in Drosophila primordial germ cells.Curr. Biol. 8, 243–246.

Vardy, L., Orr-Weaver, T.L., 2007. Regulating translation ofmaternal messages: multiple repression mechanisms. TrendsCell Biol. 17, 547–554.

Wang, X., Tanaka Hall, T.M., 2001. Structural basis for recognitionof AU-rich element RNA by the HuD protein. Nat. Struct. Mol.Biol. 8, 141–145.

Wang, Z., Lin, H., 2007. Sex-lethal is a target of Bruno-mediatedtranslational repression in promoting the differentiation ofstem cell progeny during Drosophila oogenesis. Dev. Biol. 302,160–168.

Webster, P.J., Liang, L., Berg, C.A., Lasko, P., Macdonald, P.M., 1997.Translational repressor bruno plays multiple roles indevelopment and is widely conserved. Genes Dev. 11, 2510–2521.

Weigel, D., Jurgens, G., Klingler, M., Jackle, H., 1990. Two gap genesmediate maternal terminal pattern information in Drosophila.Science 248, 495–498.

Wilhelm, J.E., Smibert, C.A., 2005. Mechanisms of translationalregulation in Drosophila. Biol. Cell 97, 235–252.

Wilson, R.J., Goodman, J.L., Strelets, V.B., the FlyBase consortium,,2008. FlyBase: integration and improvements to query tools.Nucleic Acids Res. 36, D588–D593.

Zamore, P.D., Bartel, D.P., Lehmann, R., Williamson, J.R., 1999. ThePumilio–RNA interaction: A single RNA-binding domainmonomer recognizes a bipartite target sequence. Biochemistry38, 596–604.

Zimyanin, V.L., Belaya, K., Pecreaux, J., Gilchrist, M.J., Clark, A.,Davis, I., St Johnston, D., 2008. In vivo imaging of oskar mRNAtransport reveals the mechanism of posterior localization. Cell134, 843–853.