The Notch pathway regulates both the proliferation and differentiation of follicular cells in the panoistic ovary of Blattella germanica

The Notch pathway is an essential regulator of cell proliferation and differentiation during development. Its involvement in insect oogenesis has been examined in insect species with meroistic ovaries, and it is known to play a fundamental role in cell fate decisions and the induction of the mitosis-to-endocycle switch in follicular cells (FCs). This work reports the functions of the main components of the Notch pathway (Notch and its ligands Delta and Serrate) during oogenesis in Blattella germanica, a phylogenetically basal species with panoistic ovary. As is revealed by RNAi-based analyses, Notch and Delta were found to contribute towards maintaining the FCs in an immature, non-apoptotic state. This ancestral function of Notch appears in opposition to the induction of transition from mitosis to endocycle that Notch exerts in Drosophila melanogaster, a change in the Notch function that might be in agreement with the evolution of the insect ovary types. Notch was also shown to play an active role in inducing ovarian follicle elongation via the regulation of the cytoskeleton. In addition, Delta and Notch interactions were seen to determine the differentiation of the posterior population of FCs. Serrate levels were found to be Notch-dependent and are involved in the control of the FC programme, although they would appear to play no crucial role in panoistic ovary oogenesis.


Introduction
Cell-cell communication is a basic activity of multicellular organisms, and regulates one of the most important events of development: cell differentiation. Certainly, Notch (N) signalling would appear to be one of the most important pathways involved in determining cell fate [1][2][3].
The core integral proteins of the Notch pathway in insects are the Delta (Dl) and Serrate (Ser) ligands of N (known as Delta-like and Jagged, respectively, in mammals), the N receptor (Notch 1-4 in mammals) and the transcription factor suppressor of hairless (Su-H) (CSL in mammals). Dl and Ser localize to the membranes of signal-sending cells, whereas N is located in the membranes of signal-receiving cells. Su-H joins to the N intercellular domain (NICD) in the nucleus, triggering the transcription of multiple target genes [4,5].
Notch has a pleiotropic function during the development of Drosophila melanogaster [4,6], acting as a general developmental factor in the direction of cell fate choices. The N gene was first described in Drosophila in 1919 by Mohr [7]; its mutation produces females with notched wings. Notch signalling has since been thoroughly studied in the development of the nervous system of D. melanogaster embryos. During lateral inhibition, high levels of Dl provide a neural precursor that sends a signal through the N receptor to neighbouring cells, inducing them to follow an epidermal fate [8]. The absence of Dl or N leads to neuroblastic hyperplasia [9]. Dl/N signalling also has an inductive role in eye and wing development [10], and participates in the determination of cell fate during myogenesis [11]. Additionally, N signalling is involved in the segmentation of the crustacean Daphnia magna [12].
The Notch pathway is also reported to be an important regulator of insect oogenesis in three holometabolan species with meroistic ovaries-D. melanogaster [13], Apis mellifera [14] and Tribolium castaneum [15]-and in the development of the panoistic ovary of the hemimetabolan insect Blattella germanica [16]. In the germarium of D. melanogaster, N signalling is associated with the establishment of the germ stem cell niche that regulates the number of cap cells, the niche size and the subsequent proliferation of the germ cells [17]. In the egg chamber of D. melanogaster, Dl passes the signal from the germline cyst to N, located in the somatic cells, for the differentiation of cell populations that finally affect the polarity of the oocyte [2,18,19]. Notch and Dl mutants show an excess of posterior FCs owing to a change in cell fate; this also affects the anterior-posterior polarity of the oocyte [13]. The breakdown in the differentiation of the subpopulations of FCs has a dramatic effect on the microtubule cytoskeleton, impairing the formation of the anteriorposterior axis [20]. Notch mutations also cause hyperplasia in the polar cell precursor at the expense of stalk cells during early oogenesis, leading to the fusion of the egg chambers [13]. Finally, the expression of a constitutively active N or Dl ligand in the D. melanogaster egg chamber, or overexpression of Dl in the germline, gives rise to long stalk-like structures that include the stalk itself as well as polar and undifferentiated FCs [2]. A similar interaction between Dl and N may occur in A. mellifera oogenesis, in which both ligand and receptor are localized to the oocytes and FCs [14].
The FCs must enter, and later exit, the proliferative stage in a very precise manner, steps in which N signalling actively participates. The absence of N activity in D. melanogaster FCs leads to prolonged mitosis at the expense of the endocycle [19,21,22], whereas in T. castaneum the absence of N activity triggers premature entrance into the endocycle [15]. The interplay between the Notch and Hippo pathways during the oogenesis of D. melanogaster and B. germanica [16,23 -25] has also been studied. Unlike in D. melanogaster, in which the Hippo pathway promotes N signalling in the FCs [23], N expression is repressed by Hpo in B. germanica, triggering the mitosis-to-endocycle switch [16].
Information is still lacking, however, on the molecular mechanisms underlying the N signalling pathway, and how ligands and receptors interact during oogenesis in the ancestral panoistic ovary. Using the cockroach species B. germanica as a model, this work examines the role of N (BgN in this context) in the elongation of the ovarian follicle, and the part it plays in FC cycle status. Insights gained via RNAi-based analyses into the role of the canonical ligands Dl (BgDl) and Ser (BgSer) in the panoistic ovary of this cockroach are discussed.

BgN regulates ovarian follicle elongation
BgN mRNA was strongly expressed in the ovaries during the pre-vitellogenic stage, from when the insects began the sixth (last) nymphal instar until becoming 3-day-old adults. At this latter time, which coincides with the beginning of vitellogenesis, BgN expression halved and remained close to this level until oviposition (figure 1a). Via the detection of the NICD, it has previously been shown that N is present at varying levels in the different ovarian follicles of B. germanica ovarioles [16]. In this work, BgN labelling appeared in the cells located in the germarium, in the stalk cells between the basal and sub-basal ovarian follicles, and in the FCs of the basal ovarian follicle (figure 1b). BgN expression was increased in mitotically active cells.
The architecture of the F-actin microfilaments in the ovarian follicles of 0-day-old dsBgN-treated adult females appeared disorganized, and they stained poorly with phalloidin-TRITC (figure 1d,d 00 ). Alterations in the planar polarity of the FCs were apparent; most of the FCs grouped at the poles and were elongated with an anterior-posterior orientation, as revealed by their nuclear morphology (figure 1d 0 ). No BgN labelling was detected in the ovarian follicles (figure 1d,d 000 ). At higher magnification, the changes in the FCs were more evident. The FCs in the basal ovarian follicles of dsMock-treated females were small, mitotically active (figure 1e,e 0 ) and had no spaces between them, and the F-actin microfilaments were well distributed around the cell membranes (figure 1e 00 ). However, in the dsBgN-treated females, FCs were scarce, their nuclei showed different sizes and morphologies compared with the controls, and some had a picnotic appearance (figure 1f-f 00 ).
In 5-day-old dsBgN-treated adult females, the ovarian phenotype was more obviously affected (figure 2a). mRNA levels of BgN were depleted by some 98% (41.6-fold; P(H1) ¼ 0.0001; REST analysis; figure 2b). Those of the N ligands were affected too: BgSer was downregulated by 11.6-fold (P(H1) ¼ 0.0001; REST analysis), whereas BgDl was upregulated by 1.   rsob.royalsocietypublishing.org Open Biol. 6: 150197 sending signal complex that helps N reach the nucleus, was also upregulated 2.5-fold (P(H1) ¼ 0.0001; REST analysis; figure 2b). The depletion of BgN resulted in basal ovarian follicles 82% shorter than those of the controls ( p , 0.0001; figure 2a,c; see also [16]), and which maintained the spherical shape observed in 0-day-old dsBgN-treated females (figure 1f ). A typical feature seen in the insects of this age was the reduction (even the absence) of the stalk between the ovarian follicles (figure 2a,c,d), a consequence of the absence of differentiated stalk cells in both the anterior region of the basal ovarian follicle and the posterior region of the sub-basal ovarian follicles (figure 2c,g).
The depletion of BgN mRNA in 5-day-old dsBgN-treated adult females resulted in the absence of BgN protein labelling in all ovarian follicles (figure 2a). BgDl was present in the younger oocytes (from the sub-basal follicles to the germarium), and extended throughout the cytoplasm (figure 2c), but no labelling for BgDl was seen in the basal ovarian follicles (figure 2c,d ). The FCs at the poles were still mononucleated, and most of the nuclei maintained an elongated shape giving the cells an anterior -posterior orientation (figure 2d,e). The FCs located in the central region of the ovarian follicle, however, showed nuclei more rounded in shape. Finally, the majority of FCs in all locations showed a nucleus with a picnotic appearance (figure 2d ). Conversely, in 5-dayold dsMock-treated adult females, the FCs were binucleated and had entered the endocycle (figure 2f,f 0 ), and the F-actin microfilaments associated with the cell membranes had lateral extensions connecting the cells (figure 2f 00 ), which showed patency. In dsBgN-treated adult females, the F-actin microfilaments (figure 2c-e) were not associated with FC membranes, and were distributed like a thin mesh covering the oocyte surface. They appeared more concentrated, however, in certain areas of the follicular epithelium.
The absence of stalk cells, the changes in FC morphology observed and the modifications of F-actin distribution precluded the normal growth of ovarian follicles. The subbasal ovarian follicles of the 5-day-old dsBgN-treated adult females commonly showed overgrowth that broke the follicular epithelium to produce an external bulge hanging from the ovariole (figure 2a,h-j). In some cases, this bulge contained the oocyte nucleus (figure 2j ). No FCs were seen covering these structures, and F-actin microfilaments associated with the oocyte membrane were detected as spots spreading over its surface (figure 2i). Occasionally, the anomalous distribution of the F-actin microfilaments constrained the elongation of the sub-basal ovarian follicles (figure 2k). In the regions connecting two ovarian follicles, cell membranes emerging from the basal follicle appeared, extending and invading the sub-basal follicles (figure 2l ).
These results show that BgN is important for oocyte growth and maturation. BgN participates in ovarian follicle elongation, acting on the cytoskeleton network and regulating the anterior-posterior axis.

BgN maintains the follicular cells in a nonapoptotic stage
In the basal ovarian follicles of the dsMock-treated adult females, the FCs were mitotically active during pre-vitellogenesis (figure 3a,a 0 ). Cytokinesis was halted, and the endocycle begun when vitellogenesis started (occurring in 3-day-old adult females). From this moment, the FCs then became binucleated and polyploid (figures 2f and 3d). As seen in the 0-day-old dsBgN-treated adult females, the depletion of BgN resulted in the absence of mitosis in the FCs, as revealed by PH3 labelling ( figure 3b,b 0 ). Cyclin-E (BgCyc-E) levels were therefore checked by PCR to determine whether the FCs were entering the endocycle. mRNA levels of BgCyc-E were measured in 0-day-old (mitosis stage) and 5-day-old (endocycle stage) dsMock-treated and dsBgN-treated adult females, and 2.4-fold (P(H1) ¼ 0.0001; REST analysis) and 2.6-fold (P(H1) ¼ 0.036; REST analysis) upregulations recorded for these different dsBgN-treated females, respectively (figure 3e). Moreover, the morphology of the ovarian follicles in both the 0-and 5-day-old dsBgN-treated adult females suggested that they had become prematurely apoptotic. The expression of the effector caspase-1 (BgCasp-1) was therefore measured (figure 3e). In the ovaries of 0-day-old dsBgN-treated adult females it was found not to be affected, but in those of 5-dayold females its expression increased 3.2-fold (P(H1) ¼ 0.013; REST analysis). Caspase activity in the ovaries was localized using an anti-cleavage caspase-3 antibody. No labelling was detected in the ovarian follicles of dsMock-treated females (figure 3c,d). However, in 5-day-old dsBgN-treated females, half of the ovarioles (54.5%; n ¼ 22) showed extremely strong caspase-3 labelling in the basal ovarian follicles, the FCs and the ooplasm (figure 3f,g). It is worth noting that, among the ovarian follicles, only the basal type were labelled (figure 3h); no labelling of any younger follicles was seen. In the remaining dsBgN-treated ovarian follicles (45.5%; n ¼ 22), very faint caspase-3 staining (figure 3i,j) was detected in groups of FCs mainly in the anterior region of the follicle and the transition zone between the basal and sub-basal follicles. Moreover, in agreement with the negative results for BgCasp-1 expression in 0-day-old dsBgN-treated females, no caspase-3 labelling was detected at this age (not shown). Taken together, these results show that BgN regulates the cell cycle, maintaining, in the basal ovarian follicle, the FCs in a proliferative and non-apoptotic stage.

Expression of Notch ligands in the ovary of
Blattella germanica  rsob.royalsocietypublishing.org Open Biol. 6: 150197 however, it was detected close to the membrane. Later, in 5day-old sixth-instar nymphs (figure 4d,e), it was observed in a more basal location in the FCs of all the ovarian follicles. A patchy distribution was seen for the entire follicular epithelium, with many cells showing no BgDl labelling at all (figure 4e). Remarkably, in both 0-and 5-day-old sixth-instar nymphs, no BgDl labelling was seen in the FCs at either the posterior or anterior pole in any developing ovarian follicle throughout the vitellarium ( figure 4c,d).
In 0-day-old dsMock-treated adult females, the absence of BgDl labelling at the poles of the basal ovarian follicles was apparent ( figure 4f, arrows)   rsob.royalsocietypublishing.org Open Biol. 6: 150197 no BgDl was detected in the most anterior FCs, or in the stalk cells in contact with them. Similar observations were made at the posterior pole in those cells close to the pedicel (figure 4h). However, in the younger oocytes in the vitellarium and germarium, and indeed in some germ stem cells, BgDl appeared to be distributed throughout the cytoplasm (figure 4i). In 5-day-old dsMock-treated adult females, labelling in the FCs of the basal ovarian follicle was weak, but spread over the cytoplasm and nuclear membrane (figure 4j ). During post-vitellogenesis, in 7-day-old dsMock-treated adult females, the decline in BgDl was more obvious, until becoming practically undetectable just before oviposition (figure 4k).

Depletion of BgDl and BgSer prevents the proper maturation of the ovarian follicle
To determine whether the relationship between the receptor (BgN) and their canonical ligands (BgDl and BgSer) is conserved in the panoistic ovaries of B. germanica, BgDl and BgSer were depleted by injecting the corresponding dsRNAs into 6-day-old sixth-instar nymphs and observing the phenotype in the adult stage.
In the ovaries of 5-day-old dsBgDl-treated adult females, BgDl mRNA levels were significantly reduced (1.5-fold; P(H1) ¼ 0.0001; REST analysis), whereas the expression of BgN and BgSer showed a trend towards decreasing (figure 5a). Conversely, in dsBgSer-treated adult females of the same age, the levels of BgSer mRNA in the ovaries were strongly reduced (4.5-fold; P(H1) ¼ 0.0001; REST analysis; figure 5a) but the expressions of BgDl and BgN were unaffected. Further, all other target genes of the Notch pathway were affected by the depletion of the ligands (electronic supplementary material, figure S2). In dsBgDl-treated females, again 5-day-old, the size of the basal ovarian follicles showed notable intra-individual variation. Compared with dsMock-treated females (1.55 + 0.08 mm; n ¼ 12), some ovarian follicles partially grew (1.14 + 0.19 mm (n ¼ 7); p ¼ 0.0006; Mann-Whitney test; figure 5b), and some were smaller (0.45 + 0.06 mm (n ¼ 14); p , 0.0001, Mann-Whitney test; figure 5c). These smaller ovarian follicles showed a fragile appearance, were closely connected and appeared tangled with the tunica propria.
In the basal ovarian follicles that grew partially, BgDl was detected in some FCs (figure 5d). In those FCs lacking BgDl labelling, the distribution of F-actin microfilaments appeared modified; they accumulated in the cytoplasm (figure 5e 0 ,i 00 ). In general, these basal ovarian follicles did not show patency (figure 5e,e 0 ) and not all the FCs were binucleated (figure 5e); this contrasts with that seen in dsMock-treated females, in which all the FCs were binucleated (figure 4j ). Moreover, the nuclei showed condensed chromatin, suggesting that these cells might be dying prematurely, as also indicated by caspase-3 labelling ( figure 5i). Further, in the smaller basal ovarian follicles of 5-day-old dsBgDl-treated females, the FCs were smaller, had an irregular shape, and were mononucleated (figure 5f ). The F-actin microfilaments appeared randomly distributed (figure 5f 0 ), affecting the planar polarity of the epithelium and leaving large spaces between the cells. In addition, the oocyte membrane was generally detached from the FCs (figure 5g). Despite the severe defects of this phenotype, low levels of BgDl were detectable in some FCs (figure 5g). Although the morphology of these ovarian follicles suggests they were dying (figure 5g), the number of caspase-3 labelled FCs was more or less the same as that seen in the oocytes that partially grew ( figure 5i,j).
In 5-day-old dsBgSer-treated adult females, the mRNA levels of BgSer were dramatically depleted. However, the basal follicles reached a similar length (1.6 + 0.15 mm (n ¼ 13); p ¼ 0.42; Mann -Whitney test) to those of dsMocktreated females, and the morphological phenotype was the same as that of these controls. However, the FCs were significantly shorter (20.8 + 0.37 mm (n ¼ 27); p , 0.001; Mann -Whitney test) than in the dsMock-treated females (25.4 + 0.6 mm (n ¼ 37); electronic supplementary material, figure S3), and some FCs in the basal ovarian follicles of the dsBgSer-treated females showed clear caspase-3 labelling, indicating them to be apoptotic (electronic supplementary material, figure S4).

Unlike that seen with BgN depletion, RNAi of BgDl
and BgSer revealed the stalk structure in the basal ovarian follicles It was expected that depleting the ligands of BgN would also lead to the absence of the stalk that resulted from BgN depletion (figures 1 and 2). However, and although the mRNA expression of BgDl and BgSer was depleted, the stalk formed in a manner similar to that seen for the dsMock-treated females; indeed, sometimes it was even longer ( figure 6b-d).
With the aim of enhancing the expected interference, both dsBgDl and dsBgSer were injected at the same time in 0-dayold adult females, and follicle development observed 5 days later. In the ovaries of these double-treated females, the expression of BgN showed a trend to reduce by about half, BgDl levels were not affected and BgSer levels showed notable variation (figure 6a and electronic supplementary material, figure S4). In half of the treated specimens BgSer expression was significantly downregulated (P(H1) ¼ 0.0001; REST analysis), whereas in the other half it was increased (P(H1) ¼ 0.0001; REST analysis, electronic supplementary material, figure S5). However, the same ratio of strong-to-mild phenotypes was observed as for dsBgDl treatment alone (not shown), and the stalk was similar-indeed, it was sometimes even longer than those seen in controls (figure 6e). These results show that reducing BgDl mRNA levels by half is not enough to endanger the formation of the stalk structure, and that the BgSer ligand is not necessary for its production.

The Notch pathway is involved in the differentiation of follicular cell populations at the poles
The absence of BgDl protein in the FCs at the basal ovarian follicle poles (figure 4), together with the changes in the distribution of these FCs observed after BgN depletion (figure 1), suggests that the Notch pathway is involved in the differentiation of specific FC populations. Moreover, BgN appears to be still active in FCs after BgHpo depletion, as was assessed by the strong NICD labelling in the posterior FCs (figure 7a; see also [16]). Further, these FCs showed differences in shape and pattern of distribution compared with dsMock-treated controls ( figure 7b,c, dashed area). In rsob.royalsocietypublishing.org Open Biol. 6: 150197 some groups of FCs, the cytoskeletal proteins b-tubulin (figure 7b) and F-actin (figure 7c 00 ) were almost absent. The nuclei of these cells were also larger and they showed different DNA staining compared with controls (figure 7b 0 ,c 000 ). In the band of cells surrounding the posterior FCs, BgDl protein was highly overexpressed (figure 7c,c 0 ). Remarkably, BgDl was detected in the posterior FCs; this ligand was not usually detected in these cells.
These results show that BgDl interacts with BgN in the differentiation of a specific type of posterior FCs.

Discussion
In the meroistic polytrophic ovary of D. melanogaster, the activation of N via Dl signalling triggers the switch from mitosis  rsob.royalsocietypublishing.org Open Biol. 6: 150197 to the endocycle [19,21]. However, an ancestral role for the Notch pathway has recently been revealed in the meroistic telotrophic ovary of T. castaneum [15] and in the panoistic ovary of B. germanica [16]. In these less modified ovarian models, low level N signalling in the FCs is crucial in triggering the transition from mitosis to endocycle. In B. germanica, N activity is under the control of Hpo [16], as the depletion of BgHpo results in the maintenance of N activity, which keeps the mitotic programme of FCs along the first gonadotrophic cycle. This prolonged mitotic state was counteracted when N was depleted in these females with low levels of BgHpo, thus confirming the ancestral role of N in FC mitotic-endocycle switch [16]. The expression of N during oogenesis in B. germanica [16] is similar to that seen in A. mellifera [14] and T. castaneum [15], occurring both in germ and in somatic cells. BgN labelling disappears when the follicles have left the germarium, only to appear once more in oocytes that are ready to mature [16]. This is similar to that described for the A. mellifera ovary [14], but different to that seen in D. melanogaster, in which FC expression of N is maintained until oogenesis is complete [26].
As in the meroistic telotrophic ovary of T. castaneum [15], the present results show N signalling to be needed in the panoistic ovary of B. germanica in order to maintain FCs mitotically active, and to keep them in an immature-like state. The entrance of the FCs into the endocycle usually requires the arrest of factors promoting mitosis, such as Cyclin-B, plus the phosphorylation of histone-3 and the upregulation of endocycling components such as Cyclin-E and/or Cyclin-D [27]. However, the depletion of BgN causes mitosis in the FCs to cease, preventing entry into the endocycle, as seen by the small size of the cells and their very small nuclei. Further, although Cyclin-E is overexpressed, strong caspase-1 expression suggests that these cells prematurely began programmed cell death; this was further confirmed by the presence of caspase-3 in the basal ovarian follicles of 5-dayold dsBgN-treated adult females. Remarkably, the apoptotic marker is only detected in the basal follicle and in stalk cells, where BgN is usually expressed. Whether this occurs because they fail to enter the endocycle, or because the FCs remain sensitive to other developmental signals when BgN is depleted, remains to be clarified. However, it has been described that the Notch pathway modulates the response to apoptosis in many developing organisms, connecting through crosstalk with other signalling pathways [28]. Moreover, in D. melanogaster oogenesis, N has an indirect role repressing apoptosis in the FCs after stage 6 [29]. Thus, decreasing BgN levels are required for FCs to transition to the endocycle, whereas N signalling is necessary for the survival of the maturing ovarian follicle.
The disorganization of the F-actin microfilaments after almost complete BgN depletion hinders the proper elongation of the follicle, resulting in the formation of spherical ovarian follicles. A similar phenotype has been observed in the egg chambers of D. melanogaster mutants for Fat2 cadherin; these mutants produce eggs spherical in shape and which have defects in the planar polar organization of the F-actin microfilaments, preventing egg elongation [30]. In addition, the interplay between the Notch and EGFR pathways maintains the anteriorposterior axis of D. melanogaster egg chambers; N is required at both poles for the correct expression of anterior-posterior markers [2,5,31]. The same has recently been described in B. germanica [32], suggesting the involvement of BgEGFR in the planar polarity deficiencies of FCs seen after BgN depletion.
Although BgDl and BgSer are the canonical ligands involved in the activation of the Notch pathway, other noncanonical ligands might participate in signalling [5,33]. In the  Figure 6. Depletion of BgDl and BgSer did not impede stalk formation. (a) mRNA expression levels of BgN, BgDl and BgSer in ovaries from 5-day-old dsBgDl þ dsBgSer-treated females. Data represent values normalized against the control (reference value ¼ 1) (n ¼ 3). (b-e) Detail of the stalk structure connecting the basal (BOF) and the sub-basal (sBOF) ovarian follicles in 5-day-old dsMock-treated (b), dsBgDl-treated (c), dsBgSer-treated (d) and dsBgDl þ dsBgSer-treated (e) females. The stalk was always formed in treated females and sometimes it is even longer than in dsMock-treated females. The F-actin microfilaments were stained with phalloidin-TRITC (green), and DNA with DAPI (blue). The posterior pole of the basal follicle is oriented towards the bottom left. Scale bars, 50 mm. rsob.royalsocietypublishing.org Open Biol. 6: 150197 ovary of B. germanica, the above canonical ligands showed complementary mRNA expression patterns, suggesting different ligand activity occurs at different times. In D. melanogaster oogenesis, Ser plays no role in the induction of N activity [10], the opposite to that seen in spermatogenesis, in which N is activated by Ser in the embryonic gonad [34]. In this work, the modulation of this ligand could not be confirmed, because no dramatic phenotypes were obtained after BgSer depletion, despite the drastic fall in the transcription of BgSer mRNA. However, BgSer is expressed in the ovary, and this expression is significantly altered when BgN levels are modified by dsBgN, dsBgHpo or dsBgEGFR treatment [16,32]. After dsBgSer treatment, the FCs were shorter, suggesting a functional action in the follicular epithelium of panoistic ovary. Thus, in the ovary of B. germanica, BgN appears to regulate the expression of BgSer, but the ligand does not affect the expression of the receptor.
The expression of Dl protein has been noted restricted to the germline cells in those insect species in which the Notch pathway has been studied [13][14][15]26]. In these species, Dl sends a signal from the oocyte which activates N in the FCs, thus controlling both the cell cycle and cell differentiation [19]. In this work, however, BgDl was detected in both germ and somatic cell lines, and was clearly absent in the posterior FCs. Further, after the activation of the Notch pathway resulting from BgHpo silencing, BgDl was overexpressed even in the posterior FCs, suggesting that somatic N-DI signalling determines the phenotype of the posterior FC population. It is known that N signalling induced by the ligands can be achieved by activating or inhibiting the receptor, depending on the developmental context [3,6].
The role of the Notch pathway in stalk formation is conserved from panoistic to telotrophic and through to polytrophic meroistic ovary types [15,16,19,21]. In the case of D. melanogaster and T. castaneum, the phenotypes resulting from N and Dl depletion are similar [2,13], though they are more severe for Dl depletion in T. castaneum [15]. However, neither the depletion of BgDl, nor of BgSer, nor even the simultaneous depletion of both, led to the absence of the stalk seen with BgN depletion. It is possible that small amounts of BgDl are enough to trigger a positive response of the pathway, as suggested by the different degrees to which the ovarioles are affected in dsBgDl-treated adult females. Thus, subtle changes in Dl levels may lead to differential responses in the follicles. This in turn suggests that non-canonical proteins are involved in the transmission of the signal, or that N acts independently of the ligand signal, as occurs in the D. melanogaster ovary when the Dl and Ser ligands are removed [35].
Notch signalling in the panoistic ovary of B. germanica allows for the proper development of the ovarian follicle by regulating both the proliferation and differentiation of somatic cells. The present results reveal the role of the Notch pathway in maintaining the proliferative and nonapoptotic state of FCs, as well in the differentiation of the rsob.royalsocietypublishing.org Open Biol. 6: 150197