Functional analysis of the nuage, a unique germline organelle, in drosophila melanogaster 5

77

filamentous-actin (F-actin)-coated GRK spheres at the anterior-dorsal region of D.
melanogaster oocytes (Kugler et al., 2009).




















Figure 3.1.15 KRIMP domains display distinct functions. Using nosgal4VP16 to drive
germline-specific expression, the expression level of fusion proteins appears to be
variable at different stages of oogenesis. The minimal expression of KRIMP-NT appears
100%(n=29)
0%(n=36)
100%(n=26) 0%(n=29)
95.5%(n=22)

8.3%(n=12)
94.7%(n=19)
31.4%(n=35)
15.6%(n=32)
78.1%(n=32)
78

sufficient to rescue the perinuclear localisation of AGO3 and MAEL, the expression level
and anterior-dorsal localisation of GRK, and karyosome compaction of the oocyte; but
not precocious osk mRNA translation in krimp mutant ovary. On the other hand, KRIMP-
CT only restores the timely expression of OSK protein (green) in krimp mutant egg
chamber that overexpresses KRIMP-CT (red), but the remaining mutant phenotypes are
not rescued. Although GRK localises to the anterior-dorsal region of the oocyte in krimp
mutant ovary habouring KRIMP-CT, GRK protein appears to be deposited as spheroid
aggregates. The number of rescued ovarioles for each phenotype is expressed as a
percentage of the total number of ovarioles counted. Bar is 10 µm. n, sample
size.

Since KRIMP-NT appeared sufficient to localise AGO3 to the perinuclear region of
germline cells, co-IP was performed to examine for the interaction between both proteins
in vitro. Surprisingly, when AGO3-HA was pulled down, KRIMP-NT-MYC was
undetectable by immunoblotting (Figure 3.1.16a). A similar experiment was performed
with KRIMP-CT and no observable interaction was detected (Figure 3.1.16b).

a
79


















Figure 3.1.16 KRIMP-NT and CT do not interact with AGO3 in vitro. Fusion
proteins are synthesized using T
NT® Coupled Rabbit Reticulocyte Lysate System
(Promega). MYC-tagged (a) KRIMP-NT and (b) KRIMP-CT do not co-
immunoprecipitate in the presence of HA-tagged AGO3.

The in vitro co-IP data therefore suggest that the presence of the CCCH-type zinc finger
motif and coiled-coil domain on KRIMP-NT or TUD domain on KRIMP-CT are not
sufficient to promote KRIMP direct interaction with AGO3 in vitro. Although MYC and
HA have a protein size of only 1 kilodalton (kDa), it remains possible that fusion proteins
have altered physical properties and/or interaction modules. Alternatively, certain
bridging factors may be expressed sufficiently in the presence of KRIMP-NT to promote
AGO3 perinuclear localisation in the ovary but not in vitro.

b
80

As described earlier, retroelements were de-repressed in krimp mutant (Figure 3.1.13). To
determine which of the motif and/or domains contribute to retroelement repression, semi-
quantitative RT-PCR was performed on total RNAs prepared from krimp mutant ovary
expressing either the NT or CT transgene. The expression of KRIMP-NT in krimp mutant
rescued the expression of HeT-A, TAHRE, I-element, and mst40, while the expression of
KRIMP-CT rescued the expression of all examined retroelements, except HeT-A (Figure
3.1.17).

Figure 3.1.17 KRIMP-NT restores retroelement repression. Semi-quantitative RT-
PCR of retroelements HeT-A, TAHRE, I-element, and mst40 indicates that the expression
of KRIMP-NT in krimp mutant ovary rescues retroelement de-repression. The expression
of KRIMP-CT in krimp mutant ovary restores all retroelement repression, except HeT-A.

Taking all the observations together, the tudor domain is sufficient to ensure the timely
expression of OSK protein, while the coiled-coil domain and/or CCCH-type zinc finger
81

mediates KRIMP genetic interaction with AGO3 and MAEL, as well as regulates oocyte
polarity. All of the modules on KRIMP appear to participate in retroelement repression,
either singly or in combination, to different extents.

3.2 Nuage mediates piRNA-dependent retroelement silencing
3.2.1 Nuage components mediate retroelement silencing
Some nuage components such as SPN-E, AUB, and ARMI have been reported to
participate in RNAi to regulate polarity during ovary development and/or retroelement
silencing in D. melanogaster (Cook et al., 2004; Vagin et al., 2006). Since retrolements
were de-repressed in krimp mutant and KRIMP appeared to interact genetically and
physically with other nuage components, the expression of different subtypes of
retroelements was examined using semi-quantitative RT-PCR in some nuage mutant
ovaries. Retrolements that were examined include mst40, LINEs/non-LTRs HeT-A, TART
and I-element (Aravin et al., 2003), and the euchromatic LTR retrotransposon roo
(Bowen, 2001).

HeT-A and I-element were de-repressed in krimp and mael mutants (Figure 3.2.1a),
similar to what has been reported previously in vas, aub, and armi mutants (Savitsky et
al., 2006; Vagin et al., 2004; Vagin et al., 2006). TART was de-repressed only in mael
and aub mutants, whereas mst40 de-repression was only observed in krimp and mael
mutants (Figure 3.2.1a). This difference may have resulted from the dissimilarities

between the oogenesis progression defects among the different nuage component
mutants. Alternatively, this may indicate differences in element specificity among the
82

nuage components. As for roo, no significant de-repression was observed for all
examined nuage component mutants (Figure 3.2.1a and c). The semi-quantitative RT-
PCR results were confirmed using quantitative RT-PCR (Figure 3.2.1b-c).















Figure 3.2.1 LINEs are de-repressed in the nuage component mutants. (a) Semi-
quantitative RT-PCR of the retroelements HeT-A, I-element, TART, mst40, and roo in the
nuage component mutants. (b-c) Quantitative RT-PCR of the retroelements in the nuage
component mutants armi, aub, krimp, and mael. cDNAs that are previously employed for
the semi-quantitative RT-PCR in (a) are used for quantitative PCR, in the presence of
iQ™ SYBR® Green Supermix (Bio-Rad). A mock reaction without reverse transcriptase
is also performed for each RNA sample. All results are normalised with respect to adh.
(b) HeT-A is significantly de-repressed in the nuage component mutants krimp and mael,

and RNA silencing component mutants aub and armi. (c) I-element is significantly de-
a
b
c
d
83

repressed in the nuage component mutants krimp and mael, and RNA silencing
component mutants aub and armi. TART is significantly de-repressed in mael mutant.
Significant de-repression of mst40 is observed in krimp and mael mutants. No roo de-
repression is observed in the nuage component mutants. Error bars indicate the standard
deviation between the triplicates of each sample. * p < 0.01, ** p < 0.05, n = 3. (d) In
tud
1
mutants, all the examined retroelements are repressed to similar extents as in the
control. RT-PCR was performed using Superscript III One-step RT-PCR kit (Invitrogen).


In one of the nuage component mutants tud, HeT-A, I-element, and TART appeared to be
repressed to similar extents as the control ovary (Figure 3.2.1d). This suggests that TUD
performs a different role at the nuage site. Indeed, a recent study has suggested that TUD
aids in the association of piRNAs with AUB and AGO3 in an arginine methylation-
dependent manner (Nishida et al., 2009).

The nuage components, at least for those that were examined, appeared to exhibit
different regulatory specificities for different retroelement subtypes. The LINE family
elements HeT-A, I-element, and TART, as well as mst40, are located predominantly in the
heterochromatin regions of the chromosome, whereas roo is primarily a euchromatic
retrotransposon (Aravin et al., 2003; Bowen, 2001). Defective silencing of LINEs/non-
LTRs among the examined nuage components therefore implies a common role in

maintaining the silenced state of the heterochromatic retroelements.

Vagin et al (2004 and 2006) have shown that unlike the nuage components SPN-E, VAS,
and AUB, the conventional dicing enzymes (Dicer-1) DCR-1 and DCR-2 are not
involved in the silencing of retroelements in the Drosophila female germline. Indeed,
immunostaining for the nuage proteins in dcr-1 and dcr-2 mutants indicated normal
84

perinuclear localisation of VAS, AUB, KRIMP, and MAEL (Figure 3.2.2). This suggests
that the conventional dicing enzymes and nuage components function independently, and
distinct RNA silencing machineries are probably utilised to regulate retroelement
expression in the germline.














Figure 3.2.2 Nuage localisation is unaffected in the conventional dicing enzyme
mutants dcr-1 and dcr-2. Immunotstaining of VAS, AUB, KRIMP, and MAEL in dcr-1
mutant clones and dcr-2 homozygous mutants indicates normal perinuclear localisation.
Bar is 10 µm.

3.2.2 Nuage components regulate the production of piRNAs
A previous report has linked the upregulation of retroelements in aub and armi mutants to
the failure in piRNA production (Vagin et al., 2006). To determine if the production of
piRNAs is compromised in the nuage component mutants, the levels of roo, I-element,
Protein localisation of
AUB KRIMP MAEL
VAS
85

and HeT-A piRNAs were analysed. Interestingly, all of the examined piRNAs were
reduced in spn-E, vas, krimp, and mael mutants (Figure 3.2.3). This indicates that the
nuage regulates the production of piRNAs. Although roo, I-element, and HeT-A piRNAs
were reduced in the nuage component mutants, only I-element and HeT-A, which belong
to the LINE family, were significantly de-repressed (Figure 3.2.1). This implies that the
amount of roo piRNAs that is produced in the nuage component mutants is sufficient to
ensure repression. Alternatively, the expression of roo is regulated by mechanism(s)
other than piRNA-mediated silencing. However, it remains possible that the reduction of
other LTR-derived piRNAs in the nuage component mutants may affect the repression of
their respective transcripts.

Figure 3.2.3 Production of piRNAs is defective in the nuage component mutants.
PAGE northern analysis of roo, I-element and HeT-A piRNAs. The production of roo, I-
element and HeT-A piRNAs is reduced in the nuage component mutants.