Apoptosis in Hemocytes Induces a Shift in Effector Mechanisms in the Drosophila Immune System and Leads to a Pro-Inflammatory State

Abstract

Apart from their role in cellular immunity via phagocytosis and encapsulation, Drosophila hemocytes release soluble factors such as antimicrobial peptides, and cytokines to induce humoral responses. In addition, they participate in coagulation and wounding, and in development. To assess their role during infection with entomopathogenic nematodes, we depleted plasmatocytes and crystal cells, the two classes of hemocytes present in naïve larvae by expressing proapoptotic proteins in order to produce hemocyte-free (Hml-apo, originally called Hemoless) larvae. Surprisingly, we found that Hml-apo larvae are still resistant to nematode infections. When further elucidating the immune status of Hml-apo larvae, we observe a shift in immune effector pathways including massive lamellocyte differentiation and induction of Toll- as well as repression of imd signaling. This leads to a pro-inflammatory state, characterized by the appearance of melanotic nodules in the hemolymph and to strong developmental defects including pupal lethality and leg defects in escapers. Further analysis suggests that most of the phenotypes we observe in Hml-apo larvae are alleviated by administration of antibiotics and by changing the food source indicating that they are mediated through the microbiota. Biochemical evidence identifies nitric oxide as a key phylogenetically conserved regulator in this process. Finally we show that the nitric oxide donor L-arginine similarly modifies the response against an early stage of tumor development in fly larvae.

Fig 1. Plasmatocytes and lamellocytes attach to nematode-inflicted wounds.

(A-L): Dissected gut from non-infected (A-C) and infected (D-L) late 3^rd instar vkg-GFP larva as control were visualized using DAPI (A, D, G, J, M, P,), pan hemocyte Hemese (B, E), plasmatocyte-specific Nimrod- (H) and lamellocyte-specific L2- (K) antibodies (the vkg-GFP, signal is not shown, but compare (14)). (B, E) Hemese staining was not observed in non-infected guts (B) whereas nematode-inflicted wounds showed Hemese, Nimrod and L2 staining (E, H, K). (N, Q) GFP-positive crystal cells (green) were not observed at the wound site (N), but they were detected at other places in the gut (Q). Bright field exposure (C, F, I, L, O and S) shows the extent of the wound due to melanization. Arrows indicate hemocytes, plasmatocytes, lamellocytes and crystal cells in E, H, K and Q respectively. Scale bars represent 20 μm.

Fig 2. Drosophila larvae lacking hemocytes are not more susceptible to nematode infections.

Hemocytes were depleted using two Hid insertion lines (viable-V or lethal-L) and two Hemolectin Gal4 driver lines. To visualize hemocytes, UAS-eGFP combined with hml-Gal4 was employed. Hml-Gal4,UAS-eGFP driven UAS-hid expression in both insertion lines successfully eliminated hemocytes (see S2 and S3 Figs). To maximize Hid expression we also used hml-Gal4 without UAS-eGFP. However, none of the crosses between Gal4 driver lines and Hid responder lines showed significantly increased mortality compared to the positive control (vx.Bc Imd see [13, 18]). The vertical axis shows normalized mortality and the negative control was set to 1. Data presented are means ± SD; t test: * p<0.05; **p<0.01.

Fig 3. Induction of apoptosis in plasmatocytes and crystal cells triggers lamellocyte differentiation.

Hemocyte preparations from 3^rd instar larvae were analyzed under the epi-fluorescence microscope. (A-I) Both Hid- and Grim-expressing samples showed massive lamellocyte differentiation (F and I) whereas control samples showed none (C). In addition when apoptotic cell bodies were included in the counts, an increase in counts was observed (using DAPI staining) in Hid- and Grim-expressing samples (D and G) compared to controls (A). (J, K) Some GFP-positive hemocytes were still detectable after Hid (J) or Grim (K) expression (the scale bars correspond to 50 μm). (L) Quantification of total cell (including apoptotic bodies) and lamellocyte numbers (hemocyte counts were determined within a defined area, see Material and Methods for details). Significantly higher numbers of total cells/cell fragments and lamellocytes were detected in Hid or Grim samples compared to controls. (M, N) Quantification of lamellocyte numbers per larva at 25° (M) and 29°C (N). Black arrow indicates lamellocytes (F and I), and green arrows indicate GFP-positive live hemocytes (J and K). HFP: hml-Gal4,UAS-eGFP. Data represent means ± SD; t test: **p<0.01, *** p<0.001.

Fig 4. Lamellocyte differentiation in the lymph gland of Hml-apo larvae.

Dissected fixed lymph glands from 3^rd instar larvae were stained with DAPI and the early lamellocyte-specific antibody L1. (B and G) Control (HFP/+) lymph glands show the presence of some plasmatocytes and/or crystal cells (B) whereas no GFP signal was observed in Hid-expressing lymph gland (G), which confirms effective elimination through apoptosis. Strong and extensive L1 staining was observed in the lymph gland of Hid-expressing larvae (H) whereas no L1 staining was found in the control (C). HFP: hml-Gal4,UAS-eGFP. The scale bars represent 50 μm.

Fig 5. Melanotic masses are formed in Hid- and Grim-expressing larvae.

(A-B’’’) Different patterns of melanotic masses were found in both Hid- (A-A’) and Grim-expressing larvae (B-B’’’). (C-E) The melanotic mass of the area marked in B’’’ was visualized at higher magnification (C-E). GFP-positive hemocytes were observed within the melanotic mass (D) indicating hemocyte origin. (F-H) Melanotic masses from Hid-expressing larvae also displayed a GFP signal (G) as in (D). (G). Of note, Hid expression was found stronger than Grim. (I) Positive correlation between larval frequency (melanotic spot) and lamellocyte numbers in different Hid and Grim lines (Spearman correlation, P value = 0.0108).

Fig 6. Constitutive Drosomycin expression signaling is up- and imd-dependent expression is down-regulated in hemocyte depleted larvae.

Compared to controls, induction of Drosomycin was observed in both Hid- and Grim-expressing larvae whereas imd dependent Diptericin was downregulated. CecropinA1, which is regulated with input from both pathways (Toll and imd) showed an intermediate pattern although this was non-significant in Grim-expressing lines. The relative expression level (the ratio of pro-apoptotic samples/control) is shown as log2 mean from at least 3 independent triplicates ± SD; t test: * p<0.05; **p<0.01.

Fig 7. Antibiotic treatment, different fly medium and inhibition of apoptosis can rescue eclosion, and melanization defects after hemocyte depletion.

(A) Both Hid and Grim lines showed significantly higher pupal lethality (measured as a drop in eclosure rate in %) than controls; antibiotic treatment rescued Hid and Grim induced lethality. Lethality of Hid-expressing larvae was also affected by using different fly media. SF—standard fly food (potato source). DIM—Drosophila instant medium (see S2 Table for the composition of the food, both parental lines were homozygous). (B) Coexpression of UAS-grim28.2 with UAS-p35 (caspase inhibitor) in the same larva rescued pupal lethality. Dashed lines indicate the expected frequency of eclosing flies for the crosses (25 and 50% respectively). (C) A higher melanotic spot frequency was found in adults in both Hid- and Grim-expressing lines compared to controls. Antibiotic treatment rescued melanotic mass formation in Hid- but not in Grim-expressing lines. The melanotic spot frequency was compared between Hid, Grim and controls using standard fly food (bracket 1), the influence of the food source (bracket 2, no significant differences) and the antibiotic treatment, (bracket 3, significant only for Hid-expressing larvae). Data presented are means ± SD; t test: * p<0.05; **p<0.01.

Fig 8. Hml-apo flies show a defective leg phenotype, which can be rescued by blocking apoptosis, pharmacological inhibition of NOS and antibiotic treatment.

(A) Control adults (HFP/+) where legs were normal. (B-C’) Both Hid- and Grim-expressing adults showed defective legs ranging from shortened leg segments (mild phenotype, B and C) to the complete absence of a leg (strong phenotype, B’ and C’). In both cases, phenotypes were most pronounced for the 3^rd leg pair. Arrows indicate the defective leg phenotype (B, B’, C and C’). (D) Inhibiting apoptosis by co-expressing UAS-grim28.2 with UAS-p35 rescued the defective leg phenotype. (E-G’’’) shows isolated legs including (E) normal control adult legs (E, HFP/+) and defective legs in both Hid and Grim lines (F, G, and G’) in non-treated conditions. (E”) Feeding L-NAME (a pharmacological inhibitor of NOS) to 3^rd instar larvae, rescued the leg defects in both Hid and Grim adult flies (F” and G’’’) while feeding D-NAME (inactive isomer of NAME) did not (F’ and G”). (H) Quantification of the defective leg phenotype in apoptotic adults, and after rescue with antibiotic treatment and upon co-expressing UAS-grim28.2 with UAS-p35. Defective legs were found in both Hid and Grim lines. Hid lines showed a stronger phenotype (such as a complete absence of legs, bracket 1) and a higher frequency of defective legs than Grim adults. Using a different food source (DIM) rescued the defective leg phenotype in Hid- but not Grim-expressing flies (bracket 2) and the same was true for antibiotics treatment (bracket 3). Blocking apoptosis using UAS-p35 also rescued the defective leg phenotype (right part). (I) Quantification of defective leg penetrance after treatment (D-NAME and L-NAME) compared to non-treated flies. Data presented are means ± SD; t test: * p<0.05; **p<0.01 (n = 81, 120 and 62 for controls, 189, 114 and 62 for Hid-expressing flies and 206, 120 and 72 for Grim-expressing flies).

Fig 9. Standard fly medium supplemented with arginine (NOS substrate) increased melanization and lamellocytes in pro-tumor model.

Dominant active Ras (Ras^V12)-expressing larvae showed increased penetrance of melanization and lamellocyte formation (expressed as the relative frequency of lamellocytes) upon feeding the NO donor L-arginine.

Fig 10. Model for the effects of apoptosis in hemocytes and the contribution of NO.

(A-B) Our observations suggest that Hml-apo Drosophila larvae are not fully immune-deficient, but show instead a shift towards different responses (lamellocyte differentiation, appearance of melanotic masses, Toll activation) leading to developmental defects (reduction in eclosure and the defective leg phenotype). Nitric oxide acts as a key regulator explaining the rescue we observe using L-NAME. NO levels are influenced by the microbiota, which affects gut NO production (hence the rescue by antibiotics treatment and the influence of the food source) and by NO production in hemocytes, which we block using p35 (B).