Jumu is required for hemocyte phagocytosis and development
In a previous study, we showed that heterozygous jumuGE27806 mutant adults (containing a P-element at the 5’ UTR of jumu in a forward orientation) exhibit defects in defenses against both fungi and bacteria [32]. To further investigate the role of Jumu in the phagocytosis of invading microbes by hemocytes, heterozygous jumuGE27806/+ and Df(3R)Exel6157/+ (a small deficiency that deletes jumu) larvae were injected with fluorescent latex beads or pathogens. At 1 h postinfection, the circulating hemocytes were isolated, and the number of engulfing cells and phagocytosis indexes (i.e., the number of engulfed latex beads or bacteria per hemocyte) were determined. Nearly 90% of the hemocytes were able to engulf latex beads, with a phagocytosis index (PI) of 7.397 ± 0.4358 in w1118 third-instar larvae (Fig. 1a-c). However, the jumu heterozygous mutants showed a poor ability to engulf latex beads (Fig. 1a). Only 70 and 50% of the hemocytes were able to engulf latex beads, with PIs of 4.037 ± 0.3056 and 2.656 ± 0.1094, in Df(3R)Exel6157/+ and jumuGE27806/+, respectively (Fig. 1b and c). Next, we further tested the phagocytosis ability of jumuGE27806/+ mutants against B. bassiana, S. aureus or E. coli. Although the ratio of engulfing cells to total hemocytes was not obviously reduced during the phagocytosis of S. aureus and E. coli, the PIs for the three pathogens were reduced significantly in heterozygous jumuGE27806 compared with those in w1118 (Fig. 1d and e). Because all homozygous null jumu mutants died during embryogenesis, to achieve a more severe deficiency of jumu expression, we utilized double heterozygotes generated from crosses between Df(3R)Exel6157/+ and jumuGE27806/+. In the analysis of the phagocytosis of the latex beads and pathogens, compared with w1118, the double heterozygote displayed an obvious reduction in the ratio of engulfing cells to the total number of circulating hemocytes (Fig. 1b and d). However, the PI for the phagocytosis of latex beads was not reduced, and the PIs for the phagocytosis of pathogens were increased in jumuGE27806/Df(3R)Exel6157 (Fig. 1c and e). Similar to jumu heterozygous, most jumuGE27806/Df(3R)Exel6157 hemocytes with a normal size also displayed a poor phagocytosis ability (Fig. 1a, arrowhead). However, we found that approximately 20% of hemocytes were greatly enlarged to 4–6 times the size of control cells, and most of these enlarged hemocytes (diameter > 20 μm) exhibited a strong ability to engulf latex beads and pathogens, leading to the increased PIs (Fig. 1a, arrow). These phenomena indicate that Jumu may affect the size of hemocytes in an autonomous and dose-dependent manner. To verify this hypothesis, the transcription levels of jumu in jumu mutant third-instar larvae were quantified by real-time PCR. The jumu mRNA levels were reduced by approximately 50% in the heterozygotes and nearly by 15-fold in the double heterozygote mutants (Fig. 1f). We next knocked down jumu using the hemocyte-specific driver Hml-Gal4. Knockdown of jumu at 29 °C led to reduced numbers of engulfing cells and the generation of enlarged cells exhibiting strong phagocytosis in circulating hemocytes (Fig. 1g-j, arrows). These results show that the enlarged cells resulting from the loss of jumu can effectively engulf invading pathogens and latex beads. Therefore, we next investigated whether these cells could digest the engulfed microbes. The acidification of mature phagosomes allows the digestion of engulfed particles; thus, we detected the maturation of phagosomes in jumu mutant hemocytes using E. coli labeled with pHrodo, which is a pH-sensitive dye that fluoresces in the acidic environment of a mature phagosome when it fuses with lysosomes. We found that the enlarged hemocytes could engulf the pHrodo-E. coli in an acidified mature phagosome (Fig. 1i, arrows), and the phagocytosis ability for pHrodo-E. coli was similar to the phagocytosis of FITC-E. coli in the hemocytes of jumu heterozygotes or jumu double heterozygotes (Fig. 1m and n). Taken together, these results indicate that the loss of Jumu can decrease the uptake of pathogens and latex beads and disrupt the development of circulating hemocytes in a dose-dependent manner, but it does not affect the formation of mature phagosomes.
Jumu regulates phagocytosis by modulating the expression of NimC1
The first step in phagocytosis is the recognition of microbes through a receptor on phagocytes. NimC1 and Eater have been suggested to be the phagocytosis receptors required for the phagocytosis of E. coli and S. aureus [13, 39], both of which are also markers of plasmatocytes [40]. In this study, we found that the knockdown of Eater and NimC1 also reduced the phagocytosis of the latex beads (Fig. 2a-e). Thus, we first asked whether the loss of jumu causes a deficiency of NimC1 or Eater and affects the phagocytosis of hemocytes. We found that more than 90% of the w1118 circulating hemocytes show positive staining for NimC1 and display a strong capacity to engulf the latex beads, while 30 and 70% of the hemocytes could not be marked by NimC1 in Df(3R)Exel6157/+ and jumuGE27806/+, respectively, and these hemocytes exhibited a reduced phagocytosis ability compared with that of the NimC1+ hemocytes (Fig. 2f-h, k and l; Additional file 2: Figure S1a-c). A previous study reported that multiple fly stocks and transgenic lines exhibit variations in NimC1 expression due to naturally occurring deletions and insertions at the NimC1 locus [41]. Thus, to ensure that the variations in the NimC1 expression levels in the jumu mutants were not attributed to this potential factor, we examined the expression of the NimC1 gene using RT-PCR as previously described (variation of NimC1 expression in Drosophila stocks and transgenic strains) and found that only the appropriately sized products of the NimC1-RA and NimC1-RB genes are present in the jumu mutants (Additional file 2: Figure S1d). However, we found that all circulating hemocytes in the jumu mutants could be marked by Eater-GFP and an anti-H2 antibody (pan hemocyte marker) (Additional file 2: Figure S1e-j). Similar to the heterozygous jumu mutants, only approximately 50% of the jumuGE27806/Df(3R)Exel6157 hemocytes could be marked by NimC1 and displayed a strong capacity to engulf the latex beads (Fig. 2i, arrow, k and l; Additional file 2: Figure S1 k). However, we found that some NimC1− hemocytes also exhibited a strong phagocytosis ability, and in terms of their shape, some large hemocytes in jumuGE27806/Df(3R)Exel6157 had the appearance of lamellocytes (Fig. 2i, asterisks). Thus, we next evaluated the lamellocytes of jumu mutants using an anti-L1 antibody. Few lamellocytes could be detected in w1118, Df(3R)Exel6157/+ and jumuGE27806/+(data not shown). However, we found that more than 10% of the hemocytes in jumuGE27806/Df(3R)Exel6157 could be marked by L1, and most of these cells could effectively engulf the latex beads (Fig. 2j, arrow and k; Additional file 2: Figure S1 l, arrow). These results suggest that NimC1 levels affect the hemocyte-dependent phagocytosis of latex beads and that lamellocytes also have the ability to engulf latex beads. To further investigate whether jumu regulates the expression of NimC1 in plasmatocytes in a cell-autonomous manner, we detected the level of NimC1 in the hemocytes of Hml > GFP > jumu RNAi (Hml-Gal4 begins to be expressed in second instar circulating cells), and Gcm > jumu RNAi (Gcm-Gal4 begins to be expressed in embryonic circulating cells). We detected multinucleate cells and lamellocytes among the circulating hemocytes (Fig. 2m-o; Additional file 2: Figure S1 m and n), and the immunostaining signal of NimC1 was reduced in Hml > GFP > jumu RNAi and Gcm > jumu RNAi compared with that observed in the control (Fig. 2p, q and t; Additional file 2: Figure S1o and p). The results of immunostaining against NimC1 and L1 showed that the round enlarged multinucleate cells caused by jumu deficiency are mainly plasmatocytes and lamellocytes (Fig. 2i, j and n; Additional file 2: Figure S1 k-p). We found that the overexpression of jumu can reduce the generation of multinucleate cells and lamellocytes and increase the expression of NimC1 in the hemocytes of Hml > GFP > jumu RNAi (Fig. 2r and t). Moreover, the overexpression of the NimC1 gene could rescue the expression level of the NimC1 protein and phagocytosis but could not recuse the enlarged multinucleate cells in the hemocytes of Hml > GFP > jumu RNAi (Fig. 2s-y). We next detected transcription levels of the NimC1 gene and the other seven receptor genes in circulating hemocytes. The knockdown of the jumu gene under the control of Hml-Gal4 led to decreased transcription levels of NimC1, Dscam, peste and draper (Fig. 2z). Similar to Hml > GFP > jumu RNAi, the jumu heterozygous hemocytes also displayed a reduced expression of these genes (Additional file 2: Figure S1q). It has been suggested that Dscam, peste and draper are phagocytic recognition receptors and required for microbial phagocytosis [39]. Thus, the decreased mRNA levels of these receptors might also strengthen the phagocytic defects in jumu mutants. Taken together, these results suggest that the loss of jumu can decrease the phagocytosis of hemocytes by reducing NimC1 protein levels, reduce the transcription levels of Dscam, peste and draper in circulating hemocytes and induce the generation of lamellocytes.
Loss of jumu and NimC1 in hemocytes cause defects of filopodia
It has been suggested that phagocytosis also requires dynamic rearrangement of the plasma membrane, along with actin-dependent cytoskeletal remodeling, after activation of the signaling pathway by phagocytosis receptors [15, 42, 43]. Thus, we investigated whether the loss of jumu also affected the reorganization of the actin cytoskeleton and, consequently, phagocytosis. Phalloidin staining revealed numerous filopodia extending from the periphery of w1118 hemocytes (Fig. 3a). However, jumu heterozygous and jumuGE27806/Df(3R)Exel6157 displayed obvious reductions in the number and length of filopodia but did not affect the formation of lamellipodia (Fig. 3b-d, i and j). We found that the number of filopodia increased, but their length was shorter in w1118 hemocytes after the phagocytosis of latex beads (Fig. 3e, i and j). However, the injection of latex beads did not induce an increase in the number or changes in the length of filopodia in the hemocytes of jumu mutants but did increase the number of filopodia in the enlarged round cells of the jumuGE27806/Df(3R)Exel6157 mutant (Fig. 3f-j). In addition, we found that jumuGE27806/Df(3R)Exel6157 lamellocytes exhibited large lamellipodia at the periphery under normal conditions, but after the phagocytosis of latex beads, the lamellocytes also extended numerous longer filopodia from the plasma membrane (Fig. 3d and h, asterisk). Similar to the jumu mutants, jumu RNAi under the control of Hml-Gal4 also led to reduction in the number and length of filopodia in circulating hemocytes, and overexpressing jumu could recuse the phenotypes of filopodia observed in Hml > GFP > jumu RNAi (Fig. 3k -m, q and r). The above results suggested that loss of jumu likely decreases phagocytosis by reducing NimC1 expression. Therefore, we next asked whether the phagocytosis receptors also affected the formation of filopodia. Phalloidin staining showed that deficiency of NimC1 also caused obvious reduction in filopodium formation (Fig. 3n, q and r). Moreover, the overexpression of NimC1 increased the number and length of the filopodia in circulating hemocytes and rescued the phenotypes of the filopodia observed in Hml > GFP > jumu RNAi (Fig. 3o-r). Taken together, these results suggest that Jumu may maintain the normal filopodium formation by modulating the expression of NimC1.
Jumu affects the expression of the proteins associated with the formation of actin filopodia
A previous study suggested that Ena, Fascin, Rho1 and Profilin participate in the process of filopodium formation, and the loss of these factors can lead to defects in the number and length of filopodia [20, 24, 25, 44]. Thus, we asked whether the absence of filopodia in jumu mutants was related to these factors. We first determined the expression levels of Ena, Fascin, Rho1 and Profilin in hemocytes. We found that the Ena and Fascin signals were dramatically decreased, but the levels of Rho1 and Profilin were slightly increased in jumu mutants compared with those in w1118 (Additional file 2: Figure S2a-d). Similarly, the reduced Ena and Fascin signals were also shown in Hml > GFP > jumu RNAi, but the expression levels of Rho1 and Profilin remained unchanged (Fig. 4a-c; Additional file 2: Figure S3a-c). In addition, we knocked down ena and fascin in hemocytes and observed an absence or shortening of filopodia, and knockdown of ena in particular resulted in hemocytes that rarely extended spiky protrusions (Fig. 4e-j). The above results suggest that loss of jumu can cause a decrease in NimC1 levels, and loss of NimC1 also leads to defective filopodia. Thus, we next asked whether jumu indirectly affects the expression of Ena and Fascin by regulating the expression of NimC1. However, we found that the expression levels of Ena, Fascin, Rho1 and Profilin proteins were not reduced in NimC1 knockdown hemocytes (Fig. 4a-c; Additional file 2: Figure S3a-c). Next, we detected the transcription levels of ena, fascin, rho1 and profilin and found that knockdown of jumu reduced the mRNA level of fascin (Fig. 4d). Moreover, we observed the subcellular localization of the four proteins. Immunostaining showed that loss of jumu did not markedly change the subcellular localization of Fascin, Rho1 and Profilin, except for a slight defect of Ena subcellular localization at the tips of filopodia and lamellipodia (Fig. 4m-n’; Additional file 2: Figure S2e-h). The Ena signal was primarily enriched at the tips of filopodia in control circulating hemocytes but was decreased at the tips of filopodia and lamellipodia in jumu-deficient hemocytes (Fig. 4 m-n’). Fascin and Rho1 localized to the lamellipodia and filopodia in control and jumu mutant hemocytes Additional file 2: Figure S2f and g). Profilin was distributed throughout the cell but was rarely observed at the leading edge of lamellipodia and filopodia (Additional file 2: Figure S2 h). Interestingly, we found that loss of NimC1 also caused an obviously alteration in the subcellular localization of Ena at the tips of filopodia and lamellipodia (Fig. 4o and o’). This result demonstrates that the mechanism whereby Jumu regulates the levels of Ena and Fascin is independent of NimC1, but Jumu may affect the subcellular localization of Ena by regulating the expression of NimC1.
To further identify the relationship between cell spreading and phagocytosis, we evaluated the phagocytosis of latex beads after knockdown of ena and fascin. The phagocytosis ability was reduced in the ena knockdown hemocytes; however, compared with that in the control, the knockdown of fascin in the hemocytes increased the phagocytosis of latex beads (Fig. 4h-l). Similar to Fascin, a previous study showed that loss of profilin suppresses the formation of filipodia but also causes increased phagocytosis, and the authors suggested that the loss of profilin may change the balance between elongation and filament branching and lead to greater membrane ruffling, thereby increasing phagocytosis indirectly [44]. These results suggest that the proteins associated with the formation of actin filopodia regulate phagocytosis by different manners.
Overexpression of jumu induces enhanced cell spreading and large numbers of filopodia
Next, we asked whether Jumu would be sufficient to induce the formation of lamellipodia and filopodia. As expected, numerous filopodia extended radially throughout the lamellipodia, and the number and length of filopodia as well as the area of the lamellipodia were obviously increased in hemocytes overexpressing jumu (Fig. 5a, b, e-g). However, despite inducing an increased lamellipodia area and filopodia number, the overexpression of jumu did not sufficiently enhance the phagocytosis ability of hemocytes. The PIs for the latex beads and pathogens did not differ in Hml > GFP > UAS-jumu compared with those in the control (Fig. 5h). First, we determined whether the overexpression of jumu induces numerous filopodia by increasing the expression of NimC1. However, although the loss of jumu reduced the expression of NimC1, the overexpression of jumu is insufficient to increase the expression level of NimC1 (Additional file 2: Figure S4a-d). Next, we detected the expression of Ena, Fascin, Rho1 and Profilin and found that in contrast to the loss of jumu, the overexpression of jumu caused an increase in the expression levels of Ena and Fascin and a decrease in the expression levels of Profilin and Rho1 (Fig. 5i-l). To further evaluate the role of jumu in the regulation of Ena, Fascin, Rho1 and Profilin, we transfected the full-length jumu gene into S2 cells and then detected the expression levels of these four proteins through Western blotting. Similar to the in vivo experiments, compared with the control cells, the overexpression of jumu in the S2 cells caused increases of 90 and 60% in the protein levels of Ena and Fascin, respectively, and a 50% reduction in the Profilin expression level; however, the Rho1 expression level was not reduced (Fig. 5m). The overexpression of jumu did not affect the subcellular localization of Ena, Fascin, Profilin and Rho1, but an enhanced enrichment of Ena was observed at edges of the lamellipodia and filopodia (Additional file 2: Figure S4e). To further examine whether the increases in lamellipodia and filopodia in jumu-overexpressing hemocytes were caused by the increases in Ena and Fascin, we knocked down ena or fascin in Hml > GFP > UAS-jumu hemocytes. Phalloidin staining showed that knockdown of ena or fascin efficiently inhibited the elongation of lamellipodia and filopodia in Hml > GFP > UAS-jumu hemocytes (Fig. 5c and d). Taken together, these results indicated that Jumu is sufficient to promote the formation of lamellipodia and filopodia by increasing the expression levels of Ena and Fascin.
Jumu maintains proper hemocyte division by regulating the cell cycle and cytokinesis
We detected enlarged and multinucleate cells among the circulating hemocytes in jumuGE27806/Df(3R)Exel6157 and Hml > GFP > jumu RNAi (Figs. 1a and 2n). In addition, a previous study showed that heterozygous jumuGE27806/+ larvae display an increase in the number of circulating hemocytes [32]. Thus, we detected the number of circulating hemocytes in jumuGE27806/Df(3R)Exel6157 third-instar larvae. However, we found that the number of circulating hemocytes was not increased in jumuGE27806/Df(3R)Exel6157 larvae (Fig. 6a). We speculated that this phenotype may be associated with enlarged multinucleate hemocytes in jumuGE27806/Df(3R)Exel6157. These phenomena indicate that Jumu may affect the number and size of hemocytes in a dose-dependent manner. To further verify this possibility, we next knocked down jumu using the ubiquitous driver da-Gal4 or the hemocyte-specific driver Hml-Gal4 under different temperatures to control the levels of jumu and subsequently determined the number of hemocytes. RNAi knockdown of jumu at 25 °C led to an increase in the number of hemocytes compared with the control; however, the severe deficiency of jumu caused by knockdown at 29 °C led to reduced numbers of hemocytes compared with the number of hemocytes observed following jumu RNAi at 25 °C, similar to the observation in the jumu double heterozygotes (Fig. 6b).
It has been suggested that a defect in cytokinesis or DNA overreplication can lead to enlarged cells [19, 45]. Moreover, the above results showed that the loss of jumu affects actin-dependent cytoskeletal remodeling; therefore, we speculate that the loss of jumu might also cause a defect in microtubule cytoskeleton rearrangement during mitosis. To investigate whether the enlarged hemocytes caused by the loss of jumu are due to these causes, third-instar larvae hemocytes were stained with antibodies against Tubulin and phospho-histone H3 (PH3) to visualize microtubules and cell mitosis. The hemocytes that were not undergoing mitosis showed a similar microtubule cytoskeletal organization in w1118 and jumuGE27806/Df(3R)Exel6157 (Fig. 6c and d). We found that less than 1% of circulating w1118 hemocytes were undergoing mitosis (PH3+ cells), most of which displayed clear spindles, especially during metaphase, and nuclear division accompanied cytokinesis during anaphase and telophase (Fig. 6c1-c3 and e). However, more than 3% of circulating jumuGE27806/Df(3R)Exel6157 hemocytes were PH3+ cells, nearly half of which did not display spindles or show signs of nuclear division associated with cytokinesis, and most defective cells were larger and multinucleated (Fig. 6d1-e). Next, we examined whether the phagocytic deficit observed in jumu lacking hemocytes is a secondary consequence of the defects in mitosis. We found that compared with the PH3-negative cells in w1118, the PH3-positive cells have an obviously reduced phagocytosis ability (Fig. 6f, h). However, the phagocytosis ability of the PH3-positive cells is not reduced in jumuGE27806/Df(3R)Exel6157, although some normally sized PH3-positive cells showed a reduced phagocytosis of latex beads, and the enlarged PH3-positive cells have a stronger phagocytosis ability (Fig. 6g, h). This result suggests that the phagocytic deficit of the hemocytes in jumuGE27806/Df(3R)Exel6157 is not attributed to a mitotic deficit. Moreover, the Hml > GFP > jumu RNAi hemocytes displayed a mitotic phenotype similar to that observed in jumuGE27806/Df(3R)Exel6157 (Fig. 6i-j2). We investigated whether the loss of jumu could also cause DNA overreplication in hemocytes. To investigate this possibility, we detected cells in the S phase through BrdU incorporation assays. However, the incorporation of BrdU was not increased in the jumuGE27806/Df(3R)Exel6157 third-instar larvae hemocytes (Additional file 2: Figure S5a-c), and we found that the BrdU-positive cells and BrdU-negative cells in w1118 and jumuGE27806/Df(3R)Exel6157 hemocytes had a similar phagocytosis ability (Additional file 2: Figure S5d-f’). Moreover, the TUNEL staining showed that the jumuGE27806/Df(3R)Exel6157 circulating hemocytes did not display an increase in apoptotic cells compared with the number in w1118 (Additional file 2: Figure S5 g and h). Taken together, these results suggest that compared with jumu heterozygotes, a severe deficiency in Jumu levels can induce hemocyte mitosis but inhibit the formation of spindles and cytokinesis, leading to the generation of enlarged hemocytes with multiple nuclei and a reduced number of circulating hemocytes in jumuGE27806/Df(3R)Exel6157.
Similar to cells lacking jumu, hemocytes expressing dominant-negative Rho1 (Rho1N19) or Rac1 (Rac1 DN) were enlarged and multinucleated (Fig. 6l and m) [19]. We investigated whether the enlargement of hemocytes resulting from the loss of jumu was related to the inactivation of Rho1 and Rac1. However, the expression of constitutively active Rho1 (Rho1V14) and Rac1 did not rescue the enlarged cell size of the Hml > jumu RNAi hemocytes (Fig. 6n and o), suggesting that jumu regulates hemocyte size in a Rho1- and Rac1-independent manner.
The above result shows that the loss of jumu causes an increased number of circulating hemocytes in the M phase and accelerates the cell cycle process. Thus, we investigated whether Jumu deficiency affects the expression of Cyclins. We detected the mRNA levels of CycA, CycB, CycD and CycE using real-time PCR and found that the knockdown of jumu reduces the CycA level and increases the CycB and CycD levels (Fig. 6p). Previously, we analyzed the gene expression profiles of larval circulating hemocytes with overexpression of jumu using the GeneChip Drosophila Genome 2.0 Array and found four genes, pav, png, bam and piwi, which were significantly upregulated (> 5-fold), participated in cell cycle and cell division according to gene ontology analysis (unpublished data). Moreover, a previous study showed that the RNAi of pav in the S2R or Kc cell could result in enlarged and multinucleate cells [46]. Therefore, we next detected the expression of pav, png, bam and piwi in jumu-deficient hemocytes. Quantitative RT-PCR indicated that the expression of pav and bam were not changed, but the transcription levels of png and piwi were significantly downregulated (Fig. 6p). Moreover, a similar change in the mRNA level of these genes was observed in the jumu mutant (Additional file 2: Figure S5i). These findings suggest that Jumu may control the cell cycle and mitosis process by affecting the expression of Cyclin genes, png and piwi.
Knockdown of jumu induces the generation of lamellocytes via activation of the toll pathway in hemocytes
A previous study suggested that activation of the Toll pathway in hemocytes or fat body can induce lamellocyte formation [47]. Moreover, in a recent study, we demonstrated that loss of jumu in the entire lymph gland leads to the generation of lamellocytes through activation of Dif [34]. Thus, we next questioned whether the generation of lamellocytes in the circulating hemocytes of jumuGE27806/Df(3R)Exel6157 was related to the Toll pathway. We found that Dorsal and Dif rarely exhibited nuclear localization in hemocytes and the fat body of w1118 but were obviously enriched in the nuclei of most of the jumuGE27806/Df(3R)Exel6157 hemocytes and the fat body (Fig. 7). Next, to further investigate whether Jumu tissue autonomously participates in the activation of the Toll pathway in hemocytes and the fat body, we knocked down jumu in the fat body and hemocytes using ppl-Gal4 and Hml-Gal4, respectively. The RNAi knockdown of jumu only in the fat body induced the activation of Dorsal in the fat body but did not cause the activation of Dorsal in circulating hemocytes or the generation of lamellocytes (Fig. 8a-d). This result suggests that lamellocyte formation caused by loss of jumu does not depend on activation of Toll signaling in the fat body. To further investigate whether jumu cells autonomously control the activation of Dorsal in the fat body, we used the MARCM technique to analyze the localization of Dorsal in GFP-marked clones. The clones expressing jumu RNAi (GFP+) did not show an increased nuclear enrichment of Dorsal compared with that in the wild-type clones (GFP−), and some wild-type clones showed activation of Dorsal (Fig. 8e-f”), suggesting that jumu noncell-autonomously affects the activation of the Toll pathway in the fat body. Moreover, the hemocytes of Hml > GFP > jumu RNAi exhibited nuclear enrichment of Dorsal and Dif (Fig. 8g-j’). Moreover, loss of Dif effectively inhibited the nuclear enrichment of Dorsal and Dif and reduced the generation of lamellocytes but did not rescue the enlarged cell phenotype in Hml > GFP > jumu RNAi (Fig. 8k-m’). In addition to the Toll pathway, Janus Kinase/signal transducer and activator of transcription (JAK/STAT) and c-Jun N-terminal kinase (JNK) pathway activation in hemocytes also promote lamellocyte formation [48]. Thus, we next detected activation of the JAK/STAT and JNK pathways in jumu-deficient hemocytes by assessing the expression of target genes. However, we found that the transcription levels of JAK/STAT target genes hop and Stat92E and JNK target genes puc and bsk were not increased in jumu-deficient hemocytes (Additional file 2: Figure S6). Taken together, these results indicate that severe deficiency of jumu induces the generation of lamellocytes through activation of the Toll pathway in circulating hemocytes.