Basement Membrane and Cell Integrity of Self-Tissues in Maintaining Drosophila Immunological Tolerance

The mechanism underlying immune system recognition of different types of pathogens has been extensively studied over the past few decades; however, the mechanism by which healthy self-tissue evades an attack by its own immune system is less well-understood. Here, we established an autoimmune model of melanotic mass formation in Drosophila by genetically disrupting the basement membrane. We found that the basement membrane endows otherwise susceptible target tissues with self-tolerance that prevents autoimmunity, and further demonstrated that laminin is a key component for both structural maintenance and the self-tolerance checkpoint function of the basement membrane. Moreover, we found that cell integrity, as determined by cell-cell interaction and apicobasal polarity, functions as a second discrete checkpoint. Target tissues became vulnerable to blood cell encapsulation and subsequent melanization only after loss of both the basement membrane and cell integrity.


Introduction
The discovery of Toll-like receptors and other categories of pattern recognition receptors has greatly enhanced our understanding of how the immune system recognizes different types of pathogens [1,2]; however, it is less clear why the immune system often turns its arsenal toward self-tissues.In fact, the same receptors that were originally found to bind specific types of pathogens are often involved in autoimmune diseases, making this issue more puzzling [3].To understand this process, it is imperative to molecularly define the notion of the ''immunological self''.
Autoimmune-like responses are also observed in invertebrates.In Drosophila, degenerating internal tissues are subjected to hemocyte (insect blood cell) encapsulation, in which large, flat lamellocytes wrap up the target tissues in layers and melanize them via the phenoloxidase cascade.This process is called melanotic mass formation [4][5][6].The same process occurs as part of the immunological defense against oversized pathogenic invaders, such as parasitoid wasp eggs, which are too large to be engulfed by the most abundant phagocytic hemocytes, the plasmatocytes [7,8].Currently, about 100 genes have been found to display melanotic masses upon mutation or overexpression (see FlyBase.org).These genes are seemingly unrelated, and the specific triggers of this autoimmune-like reaction are largely obscure.
More than 30 years ago, Rizki and Rizki reported that the basement membrane (BM) appeared to serve as a barrier against hemocyte attack of self-tissues in Drosophila [9][10][11].Whereas a same-species implant with the intact BM remained in the host, implants that had been mechanically damaged or pre-treated with collagenase to disrupt the BM triggered lamellocyte encapsulation [11].Moreover, undamaged implants from sibling species did not induce lamellocyte encapsulation, whereas undamaged implants from distantly related species did, suggesting that hemocytes may recognize the molecular architecture of the BM of its own species.This interesting study raises several important questions as to which molecular component of the BM is crucial for blocking melanotic mass formation of self-tissues, and whether the BM is the sole surface feature for self-tolerance.Furthermore, their experiments were carried out in a sensitized genetic background, tu(1)Sz ts , in which the hemocytes were marginally hyperactive at a permissive temperature, which made it unclear whether the mass formation is caused primarily by defects in immune cells or target tissues.These questions have never been probed with genetic tools, largely due to the essential nature of the genes encoding the BM components collagen IV and laminin [12][13][14][15][16].More recently, BM disruption was shown to act as a signal to recruit hemocytes to wound regions or to metastasizing tumors, providing further evidence for the BM-hemocyte relationship [17][18][19].
The BM is located on the basal side of epithelial tissues and serves multiple functions as a cell-supporting matrix, a tissue barrier, and ligands for cell surface receptors [20,21].The composition of the BM varies between tissue types, but in general, the BM contains the following four major components: collagen IV and laminin, which together form a meshwork, and the proteoglycan Perlecan and Nidogen, which function in the scaffold.The BM is maintained by evolutionarily conserved cell surface receptors, such as integrin and dystroglycan [22].
Drosophila melanogaster has two collagen IV genes, Cg25C (for a1 chain) and vkg (a2), and four laminin genes, wb (for laminin a1,2), LanA (a3,5), LanB1 (b), and LanB2 (c) [16,[23][24][25].Collagen IV is thought to exist mostly as Cg25C/Vkg heterotrimers, and LanB1 and LanB2 form the common core of the two laminin trimers, laminin W and laminin A. Thus, the mutant phenotypes of these genes are very similar in their own categories, and absence of one subunit is known to prevent BM incorporation of the other(s) [14,15,26].The major BM components are expressed and secreted predominantly by the fat body and hemocytes [12,26,27], although laminins are also expressed in various other tissues [16,27].
Here, we genetically removed each of the major BM components using RNA interference (RNAi) followed by careful immunohistochemical analysis and examined their roles in melanotic mass formation.We discovered that lamellocyte encapsulation may be blocked by two separate and discrete selftolerance checkpoints that operate in healthy target tissues.The first checkpoint involves laminin of the BM, and the second involves cell integrity as determined by cell-cell adhesion and apicobasal cell polarity.

BM disruption induces melanotic mass formation
To systematically investigate the relationship between the BM and the melanotic mass phenotype, we disrupted the BM using genetic approaches.We knocked down genes for the two collagen IV subunits and the four laminin subunits individually via UAS-RNAi using ubiquitous (Act5C-GAL4), inducible (Hsp70-GAL4), and tissue-specific GAL4 drivers (HmlD-GAL4, FB-GAL4, and Cg-GAL4) (summarized in Table S1).Knockdown of any one of the six genes consistently induced black masses in the larvae with either Hsp70-GAL4 or Cg-GAL4 drivers (Figure 1A and 1B; Table S1).Knockdown of the genes for the BM receptor integrins (scb for aPS3 and mys for bPS), Dystroglycan (Dg), or its cytosolic adaptor Dystrophin (Dys) similarly induced black masses (Figure 1A and 1B; Table S1).Melanotic masses formed mainly in fat bodies and salivary glands.We analyzed the fat bodies of these larvae by immunostaining with the lamellocyte-specific L1 antibody.Pale brown-colored fat bodies (dissected in early stages of melanin deposition) from larvae in which collagen IV, laminin, or integrin had been knocked down by RNAi were encapsulated by a few lamellocytes (Figure 1C-F).Black nodules (dissected in late stages of melanin deposition) recovered from these larvae were also L1-positive (Figure S1A).Using confocal microscopy, we confirmed the complete disappearance or severe disruption of the BM in the fat bodies of these larvae (Figure 1G-K).This immune response did not appear to be caused by pathogen infection, as the larvae did not induce the antimicrobial peptide gene Attacin-A (Figure S1B; Text S1).Thus, these data indicate that BM loss induced melanotic mass formation.

Analysis of the BM in extant melanotic mass mutants
To determine whether loss of the BM is a general feature of the melanotic mass phenotype, we examined various genes that had been previously associated with the melanotic mass using mutant or RNAi-treated larvae.Because we were primarily interested in the target tissues as opposed to hemocytes, we first excluded mutants that might be classifiable as ''true blood cell tumors'', in which melanotic mass formation was due to hemocyte hyperactivation [4].The following four genes were analyzed for the BM: spag [6], krz [6], mRpS30 [28], and hyx [28].We found that mutant or RNAi-treated larvae for these genes commonly had disrupted BMs in the fat bodies and that the fat bodies were positive for L1 (Figure S1C and S1D).We also examined 30 other melanotic mass-associated genes; however, the RNAi-treated larvae did not reproduce black nodules with the available UAS transgenes and tissue-specific GAL4 drivers or exhibited early lethality with either ubiquitous or stronger GAL4 drivers, thus precluding further analysis (Table S2).We also analyzed hop Tum larvae, in which the JAK kinase Hopscotch is constitutively active and melanotic mass phenotype is dominant at restrictive temperatures (.25uC) [29].Although the melanotic phenotype of this mutant may fit the classification for the blood cell tumors [30] in that hemocyte numbers increase dramatically (see Figure 2A and 2B), we sought to determine its target tissues.At 25uC, only the collagen IV level decreased severely, while at the restrictive temperature (29uC), both collagen IV and laminin were absent, and numerous lamellocytes attached to the fat body and the salivary gland (Figure S1C and S1D).Altogether, these observations corroborated the evidence that BM-deficient tissues induce melanotic masses.

Melanotic mass formation in BM-deficient larvae is an autoimmune response against altered self
To see whether lamellocyte encapsulation of BM-deficient tissues was a normal hemocyte reaction to abnormal self-tissue or rather due to an abnormality in the hemocyte itself [4], we assessed the activation state of hemocytes first by counting the cells.The numbers of circulating hemocytes of some of the BMdeficient larvae were 2-2.5-fold higher than those of controls (Figure 2A); however, the numbers of lamellocytes and crystal cells, a third type of hemocytes that contain phenol oxidation enzymes as a crystal form in their cytosol [31], did not increase significantly in any of the cases (Figure 2A, S2A, and S2B; Text S1).This result was in stark contrast to the results for hop Tum (Figure 2A), Toll D , or cactus mutants, which harbor hyperactive hemocytes [5,30].To inhibit hemocyte hyperproliferation displayed by some of the BM-deficient larvae, we expressed a dominant-negative allele for the JAK/STAT pathway receptor Domeless (dome DCYT ) in mys RNAi larvae [18,32].Hemocyte numbers were restored to a normal level in these larvae; however, melanotic mass formation was not abrogated or reduced, indicating that the mass phenotype was not due to hemocyte hyperproliferation (Figure 2A).

Author Summary
Autoimmune diseases may be caused by failures in the immune system or by altered selfness in target tissues; however, which of these is more critical is controversial.To better understand such diseases, it is necessary to first define the molecular mechanisms that provide selftolerance to healthy tissues.As a model system, we used Drosophila melanotic mass formation, in which blood cells encapsulate degenerating self-tissues.By manipulating basement-membrane components specifically in target tissues, not in blood cells, we could elicit autoimmune responses against the altered self-tissues.Moreover, we found that at least two different checkpoints for selftolerance operate discretely in Drosophila tissues.This parallels mammalian immunity and provides etiological insight into certain autoimmune diseases in which structural abnormalities precede immune system pathology, such as Sjo ¨gren's syndrome and type I diabetes mellitus.
We then examined the larval hematopoietic lymph gland, as robust lamellocyte differentiation in the lymph gland is a common feature of blood cell tumors [28,[33][34][35][36].The lymph glands of wildtype larvae rarely contained the L1-positive lamellocytes [37] (Figure 2B).The lymph glands of collagen IV-or lamininknockdown larvae occasionally contained 1-5 L1-positive cells, while lymph glands of integrin-knockdown larvae had 5-20 of these cells.The observed levels of activation may be expected for melanotic mass-forming larvae, but the levels were significantly different from that of hop Tum in which the cortical zone of the lymph gland was filled with L1-positive cells [38] (Figure 2B).We then examined sessile hemocytes, another source of lamellocytes upon wasp egg infection [39].Collagen IV knockdown did not change the sessile hemocyte population, as analyzed by plasmatocyte-specific Eater-GFP (Figure S2C).Finally, we knocked down the BM genes using the fat body-specific FB-GAL4 [28] to determine whether gene manipulation at the target site only, and not in the hemocyte or in the hematopoietic organs, still induced melanotic masses.Knockdown using any of the available UAS-RNAi transgenes singly did not induce melanotic masses, presumably due low knockdown efficiencies with FB-GAL4 driver (Figure 2C); however, knockdown of various combinations of the transgenes induced melanotic masses specifically in the fat body in the absence of a sensitized genetic background [11].We obtained similar results following fat bodyspecific RNAi knockdown for integrins or Dystroglycan, which should act strictly in a cell-autonomous manner (Figure 2C).Circulating hemocytes produced collagen IV and laminin, as reported previously [12,26,27], but were not enclosed by a sheet of collagen IV or laminin outside of the cell (Figure S2D; Text S1), excluding the possibility that knockdown of these components in the fat body may have produced the mass phenotype by affecting the surface of the hemocyte rather than the target.Based on these results, we conclude that melanotic mass formation in BMdeficient larvae results from a normal immune response (operationally defined by FB-GAL4) against altered self and is not ascribed to a failure in the immune system.

Laminin deficiency in the BM induces melanotic masses
To investigate which component of the BM is crucial in selftolerance, we individually knocked down each of the BM genes using various GAL4 drivers and analyzed BM integrity and the melanotic mass phenotype.Since melanotic masses in the BMdeficient larvae formed mainly in fat bodies and salivary glands, we focused on these two organs.In fat bodies, collagen IV knockdown with Hsp70-GAL4 reduced not only collagen IV (the fluorescence intensity was 5.23% of the control level) but also laminin in the BM (hereafter, BM laminin) effectively (33.46%; Figure 1H and 1K).Similarly, laminin knockdown (11.56%) nearly completely eliminated the BM collagen IV (0.91%; Figure 1I and 1K), indicating that these factors are structurally inter-dependent in this organ.In salivary glands, however, collagen IV knockdown with the same Hsp70-GAL4 eliminated only collagen IV (0.47%), while leaving 75.61% of laminin in the BM (Figure 3B and 3D), allowing for separation of the two components in this organ.Laminin knockdown (6.71%) effectively removed the BM collagen IV (18.02%) in the salivary gland (Figure 3C and 3D), as in the fat bodies.These results were consistent with the fact that laminin is the key component of BM assembly [14,20,21].More importantly, collagen IV knockdown induced melanotic masses only in the fat body and not in the salivary gland, whereas laminin knockdown induced melanotic masses in both organs (Figure 3E-G and L, the first four experiments; melanotic encapsulation often occurred regionally but not in the entire organs, which might be due to incomplete removal of the BM laminin).These results strongly suggest that BM laminin and possibly other factors that are tightly associated with laminin block melanotic mass formation, whereas collagen IV is not necessary for blocking this process as the collagen IVdeficient larvae did not form melanotic masses in the salivary gland (Figure 3F and 3L).

Laminin is the key element in the self-tolerance checkpoint
To further explore the self-tolerance checkpoint function of the BM components, we knocked down the collagen IV and laminin genes using HmlD-GAL4 and FB-GAL4, which are active in hemocytes and fat bodies, respectively [28,40].Laminin knockdown using these GAL4 drivers was inefficient.As for collagen IV knockdown, the changes in BM integrity induced by these two GAL4 drivers were generally the same as those with Hsp70-GAL4 (Figure 3H-K), except that collagen IV knockdown specifically eliminated the BM collagen IV but left considerable laminin at the BM even in the fat bodies (Figure S3A-D).These larvae did not form black masses, further demonstrating that collagen IV is dispensable in the BM for blocking melanotic masses (Figure 3L, the last four experiments).
We next examined Perlecan and Nidogen, the other two major components of the BM.Knockdown of the Perlecan-coding gene trol with Act5C-GAL4 did not induce black masses, indicating that Perlecan is not required in this process (Figure S3E; Table S1).Knockdown of Nidogen using available RNAi transgenes were unsuccessful; however, we found that Nidogen was completely absent in collagen IV-deficient, melanotic mass-free larvae, indicating that Nidogen is not required for blocking melanotic mass formation (Figure S3F; Table S1).Finally, we knocked down both trol and vkg, leaving laminin as the only major BM component, and found that melanotic masses were not formed (Figure 3M and 3N).Thus, these results indicate that the BM laminin was sufficient to block melanotic mass formation against self-tissue.

Loss of cell integrity correlates with melanotic mass formation
As an independent approach to determine BM function, we removed the BM by overexpressing Matrix metalloproteinase 2 (Mmp2).Mmp2 expression in the salivary gland severely disrupted BM integrity, as reported previously [18] (Figure S4A and S4B), but contrary to our expectations, the larvae did not form melanotic masses.Hemocytes attached to the salivary gland, indicating that hemocyte access to the target cells was not blocked (Figure S4C and S4D).We noticed that the salivary gland cells of these larvae looked similar to those of wild-type larvae (Figure 4A and 4B vs. 4E and 4F), whereas the cells of laminin-knockdown larvae were dissociative and round (Figure 4G and 4H), indicative of the loss of cell-cell adhesion.We further examined whether these melanotic mass-containing larvae had defects in cell polarity using apicobasal cell polarity markers [41].In wild-type larvae, Cora and Dlg localized to the lateral and basal sides of cells (Figure 4A and 4B).A similar pattern was observed in vkg trol RNAi larvae (Figure 4C  and 4D).In ptc.Mmp2 larvae in which Mmp2 is expressed in the salivary gland [18], the lumen often twisted as it expanded, and cell arrangement occasionally became abnormal (Figure 4F).Dlg diffused throughout the membrane, albeit weakly (Figure 4F).Nevertheless, Cora was still excluded from the apical side (Figure 4E), indicating that apicobasal polarity was partially maintained, perhaps due to the remains of the disrupted BM on the cell surface (Figure S4B).In contrast, the salivary gland cells of laminin-knockdown larvae displayed complete loss of apicobasal polarity.Cora and Dlg were localized throughout the cell membrane and were more often lost from the membrane in these larvae (Figure 4G and 4H).Cell-cell contacts were also severely disrupted.These results strongly suggest that loss of cell integrity, in addition to the loss of the BM, may be required for melanotic mass formation.In this report, we will subsequently use the term ''cell integrity'' to refer to the cellular aspects involving cell-cell adhesion and cell polarity.

Cell integrity and the BM are two discrete checkpoints for self-tolerance
To examine the possible role of cell integrity as another checkpoint, we used melanotic mass-free AB1.mys-i larvae.Integrin knockdown with the salivary gland driver AB1-GAL4 [42] did not disrupt the BM but resulted in detachment of the BM from the salivary gland tissue (Figure 4K).The cells lost both cell polarity and adhesion properties, and as a result, the BM appeared as a sack containing sticky balls (Figure 4I-K).As expected, hemocytes were not detected on the surface of the salivary gland in these larvae (Figure 4L).To explore this phenotype in more depth, we first mechanically sheared the salivary gland BM by pinching the AB1.mys-i,GFP larva at its anterior side with forceps (Figure 5A and 5B).After two days, these larvae developed black masses in the salivary gland at a rate that was 12-fold higher than that of the pinched control larvae (Figure 5C).Melanized salivary glands dissected from the wounded larvae were positive for L1 (Figure 5D).Second, we enzymatically disrupted the salivary gland BM of ptc.mys-i larvae by overexpressing Mmp2, as a means to more specifically manipulate the larvae.These larvae developed black masses, and the salivary glands dissected from the larvae were positive for L1 (Figure 5E-G).In these experiments, ptc-GAL4 and mys-i 27735 were used instead of the previously used AB1-GAL4 and mys-i 33642 because the latter combination with UAS-Mmp2 caused severe growth retardation of salivary glands.The reproducibility of the knockdown phenotypes was confirmed using mys-i 27735 (Figure S5).Third, we tried to wear out the BM by reducing levels of collagen IV in mys-i larvae.Black masses formed only in the fat bodies of FB.vkg-i, mys-i larvae (Figure 2C): but due to the additional AB1-GAL4 driver, a few of FB+AB1.vkg-i,mys-i larvae developed black masses also in the salivary glands, and again, the salivary glands of those larvae were indicate salivary gland and fat body, respectively.(M, N) Confocal images of the salivary-gland BMs of vkg G454 /+ (M) and vkg G454 UAS-vkg-i/+ +; Hsp70-GAL4/UAS-trol-i (N) larvae after visualization of collagen IV using Vkg-GFP (green), Perlecan (Trol) using anti-Trol antibodies (purple), Nidogen using anti-Ndg antibodies (cyan), laminin using anti-LanB2 antibodies (red), and nuclei using DAPI (blue).Scale bar: 50 mm (A-C, H-J, M, N) and 200 mm (E-G).doi:10.1371/journal.pgen.1004683.g003positive for L1 (Figure 5J).Neither mys-i nor vkg-i alone formed black masses in the salivary gland with the same FB+AB1 GAL4 drivers (Figure 5H and 5I).Taken together, our data demonstrate that cell integrity is an additional and discrete checkpoint for tolerance to self-tissues.We sought to define cell integrity in this system by knocking down genes known to be involved in cell-cell adhesion and cell polarity.Knockdown of either scrib, dlg, cora, FasIII, shg, or arm together with Mmp2 overexpression, however, did not induce melanotic mass formation, indicating that loss of any of these components at least singly did not affect cells sufficiently for disrupting the cell-integrity checkpoint function.

Discussion
Based on the results of these studies, we propose that BM laminin on target tissues functions as a crucial component not only in BM assembly but as a tolerance checkpoint to self-tissues (Figure 6).As a self-tolerance checkpoint, the BM may either serve as a physical barrier or provide an inhibitory ligand for hemocyte receptors.We speculate that the latter is the case for several reasons.First, the heterospecific implantation experiments described above suggest that the hemocyte recognizes the BM structure of its own species [11].Second, insect hemocytes are known to encapsulate a wide range of foreign materials, from parasitoid wasp eggs to synthetic beads, when injected into the hemocoel [37,43,44].Encapsulation of parasites is faster than encapsulation of non-parasitic, heterospecific implants [discussed in 11].Thus, hemocytes must be equipped with various cell surface receptors, including some as activating receptors with different binding spectra for pathogen-associated molecular patterns, and others as inhibitory receptors with narrow binding specificities to self-tissues.More specifically, laminin-coating of Sephadex beads inhibit melanotic encapsulation of the beads in mosquito hemocoel [43].The outer surface of the Plasmodium oocyst appears to bind to mosquito-derived laminin upon passage through the midgut epithelium of the mosquito [45], suggesting that the insect laminin may indeed serve as an inhibitory ligand to hemocytes of its own species.
Furthermore, we have identified cell integrity as another checkpoint for self-tolerance.Self-tissues must lose both the BM and cell integrity in order to be subjected to lamellocyte encapsulation (Figure 6).The two checkpoints are indeed discrete and experimentally separable.In our experimental system, BM integrity or cell integrity could be disrupted in the salivary gland selectively via ptc.Mmp2 or AB1.mys-i, respectively.The salivary glands of these larvae are subjected to melanotic encapsulation if the tissues fail to acquire the state of self-tolerance from the other, remaining checkpoint.Therefore, multiple failures on sequential self-tolerance checkpoints elicit autoimmune responses, which is analogous to the mechanism of self-tolerance in the mammalian adaptive immune system [46].In this respect, BM laminin is unique in that knockdown of laminin disrupted both BM integrity and cell integrity simultaneously.During pharyngeal tube formation of the C. elegans embryo, laminin is required for establishment of cell polarity of a group of precursor cells called the double plate [47].Since this process occurs before the tissue BM is formed and the mutant phenotype is not shared by collagen IV or perlecan mutants, the authors concluded that this function of laminin is distinguished from its role in the BM.Our results indicate that the Drosophila laminin functions similarly thus is unique among the BM components.
Since cell integrity was identified as a discrete self-tolerance checkpoint, it is interesting that target tissue is not subjected to encapsulation during wound healing or developmental processes in which BM integrity is disrupted temporarily.Upon tissue damage, circulating plasmatocytes attach to the wound region as early as 5 min [17], presumably to help remodel the tissue, including the disrupted BM, as hemocytes also produce BM components [12,27] (Figure 3I).According to our model, as long as cell integrity remains intact, this checkpoint would warrant selftolerance and block lamellocyte encapsulation in these cases, and thus, wound repair would proceed safely (as in Figure 5C).Cancer metastasis is a more complex problem, as the hallmarks of cancer include BM degradation and loss of epithelial polarity [48,49].Additional work will be required to probe this situation, but it is plausible that cell integrity may exert its checkpoint function normally through inhibitory cytokines, and prior to successful metastasis, cancer cells may have to obtain the ability to secret such cytokines independently of the state of cell integrity.
While our results were obtained in the invertebrate Drosophila system, we think these findings have relevance to mammalian autoimmunity.In type I diabetes in humans and in a mouse model, leukocyte infiltration and subsequent b cell destruction have been shown to correlate with disruption of the peri-islet BM, suggesting that the BM may protect islets from autoimmune attack [50,51].In Sjo ¨gren's syndrome, which mainly affects the exocrine tissues such as tear and salivary glands, increased degradation of the basal lamina is observed in labial salivary glands, and changes in laminin composition of the salivary acini correlates well with disease progression [52,53].A clear causal relationship is still lacking, but it is tempting to speculate that at least some of the autoimmune diseases may be caused by alterations in BM integrity and/or cell polarity and loss of these features may be recognized by the immune system as ''missing self'', similar to the way natural killer cells distinguish between self and nonself [54,55].

Culture conditions and heat shock induction
All flies were maintained at 25uC unless otherwise indicated on standard cornmeal and agar media.For RNAi knockdown using Hsp70-GAL4, a single heat pulse of 30 min at 37uC was applied at 24, 48, 72, or 96 h after egg laying (AEL).After counting larval black nodules, 48 h and 96 h time points were chosen as the two optimal conditions for later experiments as 24 h yielded a lower rate of melanotic mass induction and 72 h resulted in a higher rate of lethality.For imaging the BM, a single heat-shock pulse was applied at 48 h AEL for Hsp70.vkg-i, and double pulses were applied at 48 h and 96 h AEL for Hsp70.LanB2-i 104013 ;+/ LanB2 MI03747 for stronger knockdown.

Counting melanotic masses
Larval density and stage were tightly controlled during culture due to the fact that melanotic mass formation was affected by overcrowding growth conditions.Ten virgin females of GAL4 strains were crossed to 5-7 males carrying different UAS-RNA-i strains.After 2 days, eggs were collected for 6 h (approximately 40-60 eggs).Melanotic masses were counted at the wandering third instar stage by rotating the larvae under the dissection microscope.The percentages of larvae containing at least one melanotic mass in their bodies were determined after counting 3 or more vials (n$150) for each genotype or developmental time point.

Counting hemocytes
Wandering third instar larvae were washed in PBS and dried.A single larva was bled on a silicone pad in a 12-ml drop of PBS by ripping the epidermis with two fine forceps.The PBS/hemocyte drop was swirled gently using a micropipette tip and mounted on the Neubauer hemocytometer for counting.The total hemocytes were counted, and the lamellocytes were counted based on their characteristic morphology.For each genotype, at least 15 larvae were analyzed.

Wounding
In situ wounding was performed as described previously [18] with minor modifications.Third instar larvae were immersed in water on a silicone pad.Under the GFP/dissection microscope, larvae were gently pinched at the salivary glands (with help of AB1.GFP) using a pair of forceps (Fine Science Tools, Cat.No. 11295-00).Larvae were transferred to fresh cornmeal-agar media and incubated at 25uC.Dead or melanized larvae that were identified within the first 3 h were removed.After 48 h of wounding, non-pupariated larvae were dissected and analyzed for melanotic mass formation on the salivary gland.

Figure 2 .
Figure 2. Melanotic mass formation in the BM-deficient larvae is a normal immune response against altered self.(A) Numbers of circulating hemocytes (blue+red) and lamellocytes (red) in each larva were counted in the various genotypes.Cg-GAL4 drives gene expression in hemocytes and the fat body.hop Tum was raised at 29uC.Error bars represent SEM.***p,0.001by Student's t-test.Percentages of larvae containing melanotic masses: Cg.mys-i, 27.38% (n = 141/515); Cg.mys-i dome DCYT , 32.00% (n = 72/225).(B) Lamellocyte differentiation in the lymph gland was analyzed by staining with anti-L1 antibodies (red) in the larvae of the indicated genotypes.The nuclei were stained with DAPI (blue).For the vkg and LanB2 knockdowns, two classes were observed, the frequency of which plus an example is shown.The control panel represents Oregon R. hop Tum was raised at 25uC.Scale bar: 50 mm.(C) Melanotic mass formation after knockdown of genes for collagen IV, laminin, and integrin in various combinations using the fat body-specific FB-GAL4.For each genotype, n.150.doi:10.1371/journal.pgen.1004683.g002

Figure 4 .
Figure 4. Cell integrity is severely disrupted specifically in BM-deficient, melanotic mass-containing larvae.(A-J) Confocal images of larval salivary glands from the indicated genotypes illustrate defects in apicobasal cell polarity and cell-cell adhesion.Cora and Dlg were stained with anti-Cora (red; A, C, E, G, I) and anti-Dlg (red; B, D, F, H, J), respectively.F-actin and the nuclei were stained with phalloidin-FITC (green) and DAPI (blue), respectively.ptc-GAL4 is active in the salivary gland and various other tissues.AB1-GAL4 is active in the salivary gland.The control was Oregon R. Asterisks mark the lumen.(K) Confocal images of the salivary-gland BMs of AB1.mys-i larvae showing BM collagen IV (Vkg-GFP in green) and BM laminin (anti-LanB2 in red).(L) Confocal images of the salivary gland revealed that the organ surface is negative for the pan-hemocyte marker Hemese (Hem, red).Nuclei were stained with DAPI (blue).Scale bar: 50 mm (A-J, K) and 100 mm (L).doi:10.1371/journal.pgen.1004683.g004

Figure S1
Figure S1 Analysis of melanotic mass formation in BM-deficient larvae and in extant mutants.(A) Immunostaining of black masses recovered from Hsp70.vkg-i,Hsp70.LanB2-i, and Hsp70.mys-i larvae.Lamellocytes were visualized following staining with anti-L1 antibody (red) and phalloidin-FITC (green).The nuclei were stained with DAPI (blue).(B) Induction of the antimicrobial peptide gene Attacin-A was analyzed by real-time PCR in Hsp70.vkg-ilarvae.Rp49 was used as a loading control.Error bars represent standard deviations (SD).(C) Confocal images of the BM of the larval fat bodies of the indicated genotypes.The control represents Cg-GAL4 only.Collagen IV and laminin were stained with anti-Col IV (green) and anti-LanB1 (red), respectively.The nuclei were stained with DAPI (blue).(D) Immunostaining of the larval fat bodies or salivary glands of the indicated genotypes.Lamellocytes were stained with anti-L1 antibodies (red) and phalloidin-FITC (green).The nuclei were stained with DAPI (blue).Scale bar: 50 mm (A, C) and 100 mm (D).(TIF) Figure S2 Analysis of the activation state of hemocytes in the BM-deficient larvae.(A, B) Numbers of crystal cells were counted in larvae of the indicated genotypes.Serrate (Ser) was used as a positive control [64].Error bars represent SEM.(C) Sessile hemocytes were analyzed using the plasmatocyte-specific