Pharmacological Inhibition of Calcineurin Affects Immune Responses in Drosophila
The medically important immunosuppressants FK506 and cyclosporin A inhibit calcineurin and thereby suppress the transcriptional activation of cytokines in mammals [2
]. Although the calcineurin-regulated NFAT (nuclear factor of activated T cells) transcription factors (NFAT1–4) are not conserved beyond vertebrates [4
], calcineurin can also target the distantly related NFκB proteins, which are conserved mediators of innate immunity [5
]. We tested the impact of cyclosporin A and FK506 on the immune response in Drosophila
larvae. Larvae were fed gram-negative bacteria (Erwinia carotovora carotovora
15, Ecc15), which promotes the activation of the NFκB-like transcription factor Relish [7
]. The calcineurin inhibitors did not affect bacterial intake by larvae (approximately 105
cfu/larva, data not shown). To assess the immune response to oral infection, we followed green fluorescent protein (GFP) expression from the Relish-dependent Diptericin
]. As detected by the fluorescence of larvae () and western blotting (), cy-closporin A and FK506 suppressed Dipt-GFP
induction. As a further test, we used northern blots to examine induction of four endogenous antimicrobial peptide (AMP) transcripts, three (Attacin, CecropinA
, and Diptericin
) whose expression depends on Relish [10
] and one (Defensin
) whose expression is not exclusively dependent on Relish but which is especially sensitive to an as yet incompletely characterized signal from hemocytes to the fat body [11
]. Cyclosporin A and FK506 attenuated the induction of these AMP transcripts, particularly that of Defensin (). Furthermore, the survival of larvae, which generally tolerate large oral doses of Ecc15, was compromised by treatment with calcineurin inhibitors (), whereas treatment with these inhibitors alone did not affect survival (data not shown).
Calcineurin Modulates Immune Responses In Vivo
Previous work showed that the inhibition of nitric oxide synthase (NOS) attenuates larval responses to infection and that nitric oxide (NO) donors are capable of inducing immune responses via Relish [3
]. Pharmacological inhibition of calcineurin suppressed the ability of NO donors (either sodium nitroprusside [SNP], , or S-nitroso-N-acetylpenicillamine [SNAP], data not shown) to induce Dipt-GFP
. This shows that calcineurin is required for a robust response to NO but does not preclude additional regulation, such as a feedback loop in which calcineurin also induces NO.
A Novel Calcium- and Calcineurin-Dependent Pathway Mediates NO Action on GFP-Relish in S2 Cells
Responses to gram-negative bacteria in Drosophila
are largely mediated by the activation of the NFκB-like transcription factor Relish through the Imd signaling cascade [14
]. Peptidoglycan, a component of bacterial cell walls, activates this cascade, and the downstream action of Ird5 kinase and Dredd caspase promote Relish cleavage and translocation from the cytoplasm to the nucleus (reviewed in [1
]). The treatment of Drosophila
Schneider cells (S2 cells) with a crude preparation of lipopolysaccharide [15
] (LPS), containing peptidoglycan (PGN) [16
], induces the rapid and robust cleavage of Relish in an Imd-pathway-dependent fashion [18
]. Similarly, LPS treatment of GFP-Relish-expressing S2 cells causes the translocation of GFP into the nuclei in over 95% of the cells () [19
], demonstrating that S2 cells efficiently transduce an immune signal to induce Relish translocation. In contrast, the transcriptional response downstream of Relish in culture cells is much less efficient, cell-line-dependent, and variable [16
], and it needs to be enhanced by the addition of the hormone ecdysone, through unknown means [19
]. Consequently, we have relied on the GFP-Relish translocation assay for our work in S2 cells and turned to in vivo studies for a direct test of the significance of the findings in cell culture without investigating the transcriptional response in culture.
SNAP Induces Nuclear Translocation of GFP-Relish through a Calcium- and Calcineurin-Dependent, Imd-Independent Pathway
The LPS-induced translocation of GFP-Relish was not affected by calcineurin inhibitors, a NOS inhibitor (L-NAME), or the NO scavenger PTIO (, Figure S1 in the Supplemental Data
available online). Thus, NOS and calcineurin are not integral components of the pathway mediating PGN-induced Relish translocation in S2 cells.
Because NO donors trigger an Imd response in larvae, we tested the response of S2 cells to this inducer. The NO donor SNAP induced the nuclear translocation of GFP-Relish (). As seen in larvae (), the S2 cell response to NO donors was inhibited by cyclosporin A and FK506 (). We used RNA interference (RNAi) to knock down Imd, Ird5, or Dredd expression (, left), blocking the response to LPS (, right). In contrast, the response to NO did not require Imd, but it did depend on the more proximal activators of Relish, Ird5, and Dredd. Thus, in S2 cells, NO and calcineurin operate independent of Imd and LPS but nonetheless impinge on Relish function. Hence, there are two pathways that induce of Relish translocation. In vivo, these pathways might function in concert to activate Relish, or different cells might utilize different pathways.
Because calcineurin requires and is activated by Ca2+
, we tested whether Ca2+
is involved in the response to NO. The cell-permeable Ca2+
chelator BAPTA-AM blocked GFP-Relish translocation in response to SNAP (), suggesting a requirement for Ca2+
. Thapsi-gargin promotes a rise in cytoplasmic Ca2+
by inhibiting a transporter, SERCA, responsible for the sequestration of Ca2+
in the sarcoplasmic or endoplasmic reticulum [21
]. The exposure of S2 cells to thapsigargin promoted GFP-Relish translocation. Similar results were obtained by the knockdown of SERCA
by RNAi, suggesting that a rise in Ca2+
is sufficient. The action of thapsigargin was blocked by FK506, suggesting critical involvement of calcineurin downstream of Ca2+
. The requirement for Ca2+
is consistent with the requirement for calcineurin, and the induction of GFP-Relish translocation by Ca2+
mobilization suggests that the activation of calcineurin is sufficient for the induction of the translocation.
The Calcineurin Gene CanA1 Is Required for NO-Induced GFP-Relish Translocation
genes encode isoforms of the catalytic calcineurin subunit; two are highly homologous (CanA-14F
: 94% aa identity) and one (CanA1
) is more diverged (76% identity to others). RNAi to CanA1
specifically targeted the cognate calcineurin isoform, as demonstrated by CanA1
RNAi knockdown of hemagglutinin (HA)-tagged CanA1 (CanA1-HA) but not of Pp2B-14D-HA (). CanA1
RNAi attenuated the response to SNAP (), as did a second nonover-lapping double-stranded RNA (dsRNA) targeting CanA1
(data not shown). In contrast, dsRNAs targeting CanA-14F
RNAs (Figure S2
) failed to suppress the response to SNAP. To further demonstrate CanA1 requirement, we showed that dominant-negative CanA1 [22
] (CanA1 DN) blocked the response to SNAP. These results suggest that CanA1 is uniquely required for the mediation of the NO effect on GFP-Relish in S2 cells.
CanA1 Is Required for NO-Induced GFP-Relish Translocation
We also tested the sufficiency of CanA1. The expression of CanA1-HA but not Pp2B-14D-HA promoted substantial nuclear localization of GFP-Relish. The expression of constitutively active CanA1, ΔCanA1-HA, which lacks its autoinhibitory domain [23
], promoted the robust nuclear localization of GFP-Relish. The expression of the calcineurin constructs was verified by western blot (). Thus, CanA1 is required for SNAP-induced nuclear translocation, and its activity is sufficient for the induction of this translocation.
In mammals, calcineurin dephosphorylates NFAT1–4 during immune induction and dephosphorylates the NFκB inhibitor, IκBβ, when activated by pharmacological stress [2
]. Relish is a compound NFκB/IκB protein containing a C-terminal IκB-like region. The IκB region of Relish includes a domain homologous to the PEST domain of IκB that is targeted by calcineurin. The deletion of the Relish PEST region resulted in constitutive nuclear localization and activity [18
]. We mutated serines within this region to alanines, which similarly yielded constitutive nuclear GFP-Relish (Figure S3
). Although we have not assessed the phosphorylation of Relish, these find-ings identify sequences required for the cytoplasmic localization that are candidate sites for the action of calcineurin. Additionally, coimmunoprecipitation experiments showed that HA-tagged CanA1 interacts with GFP-Relish (Figure S4
). These findings suggest that cal-cineurin acts directly on Relish, a paradigm of calcineurin action that has been established for other transcription factors [2
CanA1 Modulates Immune Responses In Vivo
To test the in vivo relevance of CanA1 involvement in immune responses, we knocked down CanA1 in Drosophila
by using RNAi foldback transgenes (CanA1RNAi
). Flies ubiquitously expressing either of two independent CanA1RNAi
transgenes (FB1 and FB2) exhibited no obvious defects or developmental delays. CanA1RNAi
knocked down CanA1
RNA but not that of Pp2B-14D
(, left) and greatly reduced the expression of HA-tagged CanA1 (ΔCanA1-HA; , right), showing CanA1RNAi
efficiency. Ubiquitous CanA1RNAi
(with Da-Gal4) also decreased NO-mediated induction of Dipt-GFP
in larvae (, bottom left, and data not shown), suggesting that CanA1 mediates the action of NO. Furthermore, ubiquitous CanA1RNAi
induction upon infection (, “U”). Similar results were obtained with the entomopathogenic bacteria Pseudomonas entomophila
]) (Figure S5
), extending our observations to another gram-negative bacteria.
CanA1 Modulates Immune Responses In Vivo
Because infection induces NOS in the gut and hemocytes appear to be required for the response of the fat body [3
], we used a hemocyte-specific Gal4 driver to express CanA1RNAi
and tested the effect on fat-body responses to infection. Hemocyte CanA1RNAi
suppressed the induction of Dipt-GFP fluorescence in the fat body (, “H”). In contrast, fat-body CanA1RNAi
did not (, “F”). Similarly, western blots on Dipt-GFP
larvae showed that ubiquitous and hemocyte CanA1RNAi
, but not fat-body CanA1RNAi
, decreased the induction of GFP expression.
We also assessed the CanA1RNAi influence on the induction of endogenous AMP transcripts (Attacin A, CecropinA1, Diptericin, and Defensin) by northern blots. Ubiquitous CanA1RNAi reduced the transcription of all AMPs (, left), but in particular that of Defensin. Hemocyte CanA1RNAi (, middle) reduced the expression of all AMPs as effectively as did ubiquitous expression. As in the other assays, fat-body CanA1RNAi had no effect (, right). These findings show that calcineurin contributes to the immune response in larvae and functions in hemocytes to make a nonautonomous contribution to the induction of AMPs in the fat body.
Whereas the modulatory input of CanA1 has a major impact on the level of the induction of antimicrobial peptide genes in the larvae, preliminary analyses suggest that it only has a modest influence on similar responses in the adult. Although ubiquitous CanA1RNAi reduced the resistance of adults to P.e. infection by either feeding or by septic injury, we did not detect a change in the induction of Dipt-GFP. Perhaps the larva relies more heavily on a particular pathway of signal transduction that uses NO and calcineurin.
To test whether calcineurin is sufficient for immune induction in larvae, we expressed a CanA1 gain-of-function transgene (ΔCanA1). Because the ubiquitous expression of ΔCanA1 is lethal (data not shown), we examined the consequences of ΔCanA1 expression in hemocytes. ΔCanA1 expression in hemocytes of third-instar larvae promoted the nuclear localization of GFP-Relish (, left), as it did in S2 cells. However, ΔCanA1 expression in hemocytes did not induce Dipt-GFP in the fat body (data not shown). This negative result might not be meaningful because the ubiquitous expression of ΔCanA1 is lethal, and ΔCanA1 could disturb hemocyte function. However, we suspect that the result could have biological significance, perhaps reflecting the specialization of the signaling pathway for a response to acute rather than chronic induction.
We also expressed ΔCanA1 in the fat body, where it did induce Dipt-LacZ (, right), showing that calcineurin is sufficient in these cells. However, it is not clear whether ΔCanA1 acts entirely cell autonomously because Dipt-LacZ induction in the fat body is variegated, whereas fat-body Gal4 expression is uniform (data not shown). This gain-of-function experiment suggests that ectopically activated CanA1 can activate Relish in the fat body, whereas the CanA1RNAi experiments suggest that this ability does not make an important contribution to the normal response and that the fat-body response depends on hemocytes. The difference between these results suggests that CanA1 is not significantly activated in the fat body in response to infection. Accordingly, AMP induction in the fat body requires hemocyte-associated CanA1, but ectopic activation of CanA1 in fat body can bypass this requirement.
Altogether, our findings show that calcineurin contributes to innate immune responses and conveys an NO signal that activates AMP production in the Drosophila
larva. The bacteria (Ecc15) fed to the larvae remain con-fined to the gut [26
] but nonetheless induce responses in the fat body. Because infection induces NOS in the gut, NO produced in the gut might signal to hemocytes, which then induce responses in the fat body [3
]. This proposal is supported by previous demonstrations that NOS contributes to immune induction and that Domino
mutant larvae, which have a severe reduction of hemocytes (among other defects), fail to induce Dipt
in response to NO or to natural infection [3
]. Furthermore, psidin
gene function in hemocytes promotes fat-body expression of AMPs [11
]. Our demonstration that CanA1 is required in hemocytes for the immune response in the fat body provides further support of this proposal, illustrated in a model (). In this model, the response of the hemocyte to NO is independent of Imd, like the response of S2 cells, whereas the robust induction of AMPs in downstream tissues requires Imd. Consequently, Imd acts downstream of NO to induce AMPs in larvae [3
]. These findings argue that tissue-to-tissue signaling plays a role in a natural infection model in larvae and that CanA1 participates in this signaling.