GATA factors participate in tissue-specific immune responses in Drosophila larvae

Tissue-Specific Expression of Drosophila GATA Factors.

To determine whether Serpent might regulate Dpt-lacZ and Mtk-lacZ induction in the gut, RT-PCR assays were done to identify sites of the expression of all five GATA family members in the Drosophila genome (Fig. 2A). Expression of srp is primarily restricted to the fat body (lane 1), and there is little expression in the anterior or more posterior portions of the midgut (lanes 2 and 3). Thus, it is unlikely that srp mediates immunity gene expression in the gut.

Open in a separate window

Fig. 2.

Tissue-specific localization of Drosophila GATA family members. (A) Semiquantitative radioactive RT-PCRs performed with total RNA extracted from the indicated tissues. PV refers to proventriculus. A ribosomal protein gene (rp49) was used as a loading control. srp transcripts are most abundant in the fat body (lane 1). Transcripts for grn appear to be midgut-specific (lane 3), whereas dGATAe is detectable throughout the midgut (lanes 2 and 3) and the malpighian tubules (lane 4). dGATAd is detectable in every tissue tested (lanes 1–5). The three Rel factors dorsal, Dif, and Relish, are detectable in all tissues at varying levels. (B) In situ hybridizations of larval midguts with DIG-labeled dGATAd and dGATAe antisense probes. (Left) dGATAd expressed in the anterior midgut, fading along the gastric caecae. Two sharp domains with different staining intensities appear in the anterior proventriculus. (Right) dGATAe present at high levels in the anterior midgut, with reduced levels posterior to this region, along the gastric caecae, and in the proventriculus. (C) Close-up view of dGATAd and dGATAe RNA localization in the proventriculus. The domain of dGATAd expression (Left) apparently complements several enhancer-lacZ staining patterns, including Dpt-lacZ (Right). dGATAe is expressed in a graded fashion throughout this organ (Center).

Instead, there is strong expression of two other GATA factors in regions of the gut that show an immune response, dGATAe and Grain. dGATAe is expressed in three regions: the proventriculus plus anterior midgut (lane 2), middle and posterior midgut (lane 3), and malpighian tubules (lane 4), whereas grn is expressed solely in the middle plus posterior midgut (lane 3). It is therefore possible that dGATAe and Grain work together to induce immunity gene expression in posterior portions of the midgut, whereas dGATAe functions alone in the anterior midgut, proventriculus, and malphigian tubules. dGATAd is constitutively expressed in all tissues surveyed.

To visualize the spatial localization of dGATAd and dGATAe transcripts, in situ hybridizations with DIG-labeled antisense RNA probes were performed on dissected larval midguts (Fig. 2 B and C). dGATAd is expressed in two bands with different intensities around the anterior proventriculus (Fig. 2B Left). Additional staining appears in the anterior midgut. dGATAe is strongly expressed in the anterior midgut, with weaker expression throughout the proventriculus and gastric caecae (Fig. 2B Right). It is possible that dGATAd functions as a repressor to restrict gene expression to discrete regions of the gut. For example, ingestion of E. carotovora induces Dpt-lacZ within posterior regions of the proventriculus (Fig. 2C Right), even though dGATAe is expressed throughout this organ in both anterior and posterior regions (Fig. 2C Center). This expression appears to complement sites of dGATAd mRNA accumulation (Fig. 2C Left).

Ectopic Expression of dGATAe Leads to a Toll^10b Phenotype.

The preceding analysis raises the possibility that different GATA transcription factors mediate immune responses in different tissues. dGATAe was selected for further study because it is expressed in regions of the gut reacting to infection. The entire dGATAe protein coding sequence was placed under the control of yeast upstream activating sequences (UAS) and misexpressed by using different Gal4 drivers (Fig. 3).

Open in a separate window

Fig. 3.

dGATAe misexpression causes severe immunological defects. (A and B) Flies carrying an hsp70-GAL4 driver were mated with flies carrying a UAS-dGATAe cassette. The trachea (tr, A) and hindgut (hg, B) of the third-instar progeny become melanized 18 h after heat shock. Trachea melanization first appears in the posterior spiracles of most larvae, although in this example it has begun in the middle. All larvae die prior to pupation. (C–I) The P{GawB}c754 strain produces Gal4 in the fat body, blood, and salivary glands of second- and third-instar larvae. This strain was mated with UAS-dGATAe transgenic lines. Many of the offspring (20–50%) develop melanized spots on the cuticle (C–E) whereas a few (≈10%) exhibit small melanized regions within the hindgut. The cuticle lesions are typically oriented along the anterior–posterior axis, possibly in response to the crawling action of the larvae. (F) Blood smears stained with rhodamine-phalloidin (red) and DAPI (blue) reveal the presence of lamellocytes (LM), typically a response to parasite infection. (G–I) The fat body, normally a uniform monolayer of cells (G), becomes disorganized on dGATAe expression (H). Rhodamine–phalloidin (red) and DAPI (blue) staining of the tissue reveals enlarged cells and nuclei, similar to Toll^10b fat (I).

Constitutive dGATAe expression was achieved by using a Gal4 driver under the control of heat shock hsp70 5′ regulatory sequences. When hsp70-Gal4;UAS-dGATAe strains are raised at room temperature, all progeny die during the first larval instar, indicating leaky expression from the hsp70 regulatory sequences. When raised at 18°C, the larvae develop normally. Heat shock was applied for 30 min at 30°C, and larvae were subsequently transferred to room temperature for 18 h. Third-instar larvae treated in this fashion displayed a number of abnormalities (Fig. 3 A and B), including melanization of the trachea and hindgut. All of the larvae died prior to pupation.

To determine the site of the ectopic dGATAe expression causing the mutant phenotype, a Gal4 driver was used that causes selective misexpression in the developing fat body, salivary glands, and blood. The P{GawB}c754 driver induces such expression as early as the second instar stage (16). The resulting climbing larvae display a consistent syndrome of defects. The cuticle is extensively spotted and scarred (Fig. 3 C–E), and there are melanized portions of the hindgut and excessive lamellocytes (Fig. 3F, LM). These observations suggest that the ectopic expression of dGATAe causes a systemic immune response in the absence of infection.

The fat body expressing dGATAe displays a number of abnormalities. The normal fat body is a single-layered tissue (Fig. 3G), but in response to dGATAe expression it exhibits buckling and regions composed of multiple cell layers (Fig. 3H). In addition, DAPI staining reveals nuclei that are larger than normal, suggesting an additional round of endoreplication. The general appearance of the fat body is similar to that seen in Toll^10b mutants (Fig. 3I). The abnormal fat body morphology, combined with tissue melanization and lamellocytes, raises the possibility that ectopic dGATAe expression induces constitutive Toll signaling, as seen in Toll^10b mutants.

Ectopic dGATAe Does Not Induce Toll Signaling.

RT-PCR assays were performed with RNA extracted from fat bodies of P{GawB}c754;UAS-dGATAe and Toll^10b larvae (Fig. 4). As documented previously (17), constitutive Toll signaling augments the steady-state mRNA levels of different components of the Toll pathway, including the Spatzle ligand, the Toll receptor itself, the Cactus inhibitor, and all three Rel-containing transcription factors, Dorsal, Dif, and Relish (Fig. 4; compare lane 4 with lane 1). None of these genes is up-regulated in either one of two lines containing the UAS-dGATAe fusion gene inserted in different regions of the genome and driven by P{GawB}c754 (Fig. 4; compare lanes 2 and 3 with lane 1). This is not because dGATAe is only transiently expressed compared with Toll^10b, because P{GawB}c754;UAS-Toll^10b larvae also show heightened levels of Toll signaling components (data not shown). The only obvious link between the two genotypes is the constitutive expression of Drs in the absence of infection. These results suggest that the phenotypes are only superficially similar and arise from distinct underlying causes.

Open in a separate window

Fig. 4.

dGATAe causes a Toll^10b phenotype without up-regulation of pathway components. (A) Total RNA was purified from third-instar larval fat and used in RT-PCR assays with primers specific to the indicated genes. Lane 1, yw control; lanes 2 and 3, offspring of two independent UAS-GATAe lines mated with P{GawB}c754 flies; lane 4, Toll^10b. Results were quantified, normalized to rp49, and the relative intensities reported above each band. (B) Larvae bearing the Mtk-lacZ transgene were fed Erwinia and compared with larvae bearing mutant versions of the Mtk regulatory region. The leftmost panels depict Mtk-lacZ. Constitutive lacZ expression is seen in the proventriculus (PV) and malpighian tubules (MPT) with inducible expression in the anterior midgut (AMG). The center panels show the Mtk regulatory DNA bearing three mutated GATA sites. All lacZ expression is abolished. The rightmost panels show the Mtk regulatory DNA bearing three mutated Rel sites. Constitutive proventriculus expression and enhanced malpighian tubules expression occur, but inducible anterior midgut expression disappears.

To further investigate the possibility that dGATAe acts independent of Toll signaling, we tested the midgut response of the Mtk 5′ regulatory sequence bearing mutations in the GATA and Rel binding sites (Fig. 4B). The intact Mtk regulatory region drives constitutive lacZ expression in the proventriculus and malpighian tubules. Upon infection, additional lacZ expression occurs in the anterior midgut (Fig. 4B, leftmost panels). Mutation of the three consensus GATA sites within this sequence abolishes lacZ expression in every tissue (Fig. 4B, center panels). Mutation of the three consensus Rel sites, however, only disrupts inducible anterior midgut expression (Fig. 4B, rightmost panels). The expression in the proventriculus remains, and the malpighian tubule expression becomes stronger, suggesting Rel factor-dependent and -independent domains of gene activity.

Erwinia Infections and lacZ Staining.

50 ml of E. carotovora was grown overnight, pelleted, and mashed into 4 g of standard fly food. Third-instar larvae were gently stirred in and left for one hour. Afterward, the entire mixture was transferred onto apple juice plates to sit overnight at room temperature. Larvae were then dissected, fixed, and stained as described in ref. 10. Three or more independent lines were compared for each construct.

RNA Isolation and RT-PCR.

For RNA isolation dissected larval tissues were collected on ice. Samples were homogenized in a few drops of 1× PBS using a motorized pestle, then 1 ml of TRIzol (catalog no. 15596-026; Invitrogen, Carlsbad, CA) was added, and the samples were processed according to the manufacturer's instructions. The total RNA pellet was resuspended in RNase-free water. For RT-PCRs, the Access RT-PCR system (catalog no. A1250; Promega, Madison, WI) was used according to the manufacturer's instructions with the addition of 2 μCi [α-³²P]dCTP per reaction (1 Ci = 37 GBq). The final products were resolved on a 6% acrylamide (29:1 acyl/bis) gel with 0.5× TBE and quantified by using a Storm phosphoimager (Molecular Dynamics, Sunnyvale, CA). The primer pairs were as follows: pnr, 5′-ATGTACCACAGTAGCGCCGTTG-3′/5′-CGCCAAACTGGAAGTCCATGGCGCTCT-3′; srp, 5′-ATGCGGAACAACTTTGCGTTC-3′/5′-GCTGCTGCTGCTGATGGTGATGCAGTT-3′; grn, 5′-ATGGATATGACCTCAACAGCGG-3′/5′-GGGCATGCGGGATGTGTGCTGATGATA-3′; dGATAd, 5′-ATGAATAATGTACCACATAAGTTTCG-3′/5′-CGGGCACAGGCACGGAAACGGGTAT3′; dGATAe, 5′-ATGGTCTGCAAAACTATCTCACCG-3′/5′-TTCGCTGACGCCCGCTTGGCCCGTCT-3′; dl, 5′-ATGTTTCCGAACCAGAACAA-3′/5′-TCTTGCAGCCCTCCTTGCCAAC-3′; Dif, 5′-ATGTTTGAGGAGGCTTTCGG-3′/5′-GAACCGGCGGTGCGACCCTCGC-3′; Rel, 5′-ATGAATCAGTACTACGACCTG-3′/5′-ACGGTGGCACAGTGGCCGGAGC-3′; spz, 5′-CCAAGTATCGGCCACCACAATCCCCAGC-3′/5′-CCCTCAAGCCCTTTTTTGGGTACACCAG-3′; Tl, 5′-ATGAGTCGACTAAAGGCCGCTTCCGAG-3′/5′-AACCCTGTCGACCTCACCGATCCGCAAC-3′; cact, 5′-ATGCCGAGCCCAACAAAAGCAGCGG-3′/5′-GCTGATCCTTATCCTGTTCCTCGCTATC-3′; Drs, 5′-CCGTGAGAACCTTTTCCAATATGATGATGCAG-3′/5′-aTTGCAGCATAGAATATGTGTAAGTAGTGGAGAGC-3′; rp49, 5′-TCCGCCCAGCATACAGGCCCAAGATCGT-3′/5′-TTACTCGTTCTCTTGAGAACGCAGGCG-3′.

In Situ Hybridization.

Midgut tissue from third-instar larvae was fixed and hybridized with digoxigenin-UTP antisense RNA probes as described for Drosophila embryos in ref. 27. The probes correspond to nucleotides 2380–2640 for dGATAd, nucleotides 1930–2230 for dGATAe, nucleotides 242–602 for CG5096, nucleotides 329–689 for CG18480, nucleotides 235–595 for Tsp42Ep, and nucleotides 178–538 for Tsp29Fa.

We thank Tony Ip for helpful feedback and for providing UAS-Toll^10b flies, Karen Vranizan and Vivian Peng for aiding in microarray hybridization and analysis, and Patrick O'Farrell for providing the Erwinia strain. This work was supported by a National Institutes of Health postdoctoral fellowship (to K.S.) and National Institutes of Health Grant R01 GM46638 (to M.L.).