βint-ν expression in all types of the adult midgut epithelial cells

Although homozygous null mutants of βint-ν have been reported to be viable and fertile, βint-ν was found to be evolutionally conserved among Drosophila species (Figure S1). In addition, βint-ν is exclusively expressed in the embryonic and larval midgut including imaginal island, a progenitor of the adult midgut epithelia [24]. More recently, βint-ν mRNA was also detected from the adult midgut with RT-PCR (Lin et al., 2013). Therefore, we speculated an additional novel function of βint-ν responsible for keeping the adult midgut healthy. To examine this possibility, we performed the following experiments using female midguts in this study.

The adult midgut was composed of epithelial cells (ISCs, EBs, ECs, ees) and visceral muscles (VMs) (Figure 1A). First, we confirmed expression of βint-ν in the adult midgut with in situ hybridization experiment.The βint-ν mRNA was detectable in all types of epithelial cells including diploid cells (ISCs, EBs and ees) and polyploid cells (ECs) with antisense probe (Figure 1B) but not with sense probe (Figure 1C). In addition, βint-ν mRNA disappeared in the adult midgut epithelium by depletion of GATAe that activates expression of βint-ν [31] (to be presented elsewhere). Furthermore, distribution of βint-ν protein at the basal side of the epithelium was detected with immunostaining using two kinds of anti-βint-ν antibodies that had independently been generated (Figure 1D and 1E) [24], [30]. A fewer βint-ν protein was detected also at the visceral muscle (Figure 1D and 1E). Moreover, we also found that NP0319, a GAL4-enhancer trap construct inserted near the βint-ν locus, showed a strong expression of GFP driven by GAL4 specifically in ECs (Figure 1F). Collectively, these indicated that βint-ν was expressed in all types of the adult midgut epithelium.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. βint-ν is expressed in the adult midgut epithelium.

(A) The diagram of the adult midgut composed of ISCs, EBs, ECs, ees, and VMs. The cell-type specific GAL4 drivers, esg-GAL4, MyoIA-GAL4, and 24B-GAL4, used in this study are shown in parentheses. (B–C) Transcripts of βint-ν were detected with an antisense probe (magenta in B and B’) for βint-ν mRNA but not with a sense probe (magenta in C and C’) for it. “Pros” and “Arm” respectively indicate the ee nuclei and the outline of cells (green in B’ and C’). Small cells without Pros are ISC/EB. (D–E) βint-ν protein was detected with anti-βint-ν antibodies (green, D [30], E [24]). Basal visceral muscles were strongly marked with phalloidin staining (magenta in D’ and E’). Arrows indicate the distribution of βint-ν protein at the basal side. (F-F”) Expression of NP0319 (a βint-ν-GAL4, green), monitored with UAS-GFP, was detected in polyploid ECs and a subset of esg-lacZ-positive cells (yellow) but not the other esg-lacZ- and Pros-positive cells (magenta). Yellow and magenta arrows indicate examples of esg-lacZ- and Pros-positive cells. Nuclei were stained with DAPI (white in D’, E’, and F”).

https://doi.org/10.1371/journal.pone.0089387.g001

βint-ν mutants showed shortened and multilayered epithelium of adult midgut

To examine whether βint-ν has a function in maintenance of normal adult midgut, we observed the adult midgut morphology of homozygote flies for βint-ν², a null allele of βint-ν [29], and found that it was shorter than that of wild-type flies (Figure 2A). Both in wild-type and βint-ν² flies, the length gradually decreased with age, which was measured at 7, 14, and 28 days old (Figure 2B).Furthermore, the decrease rate was higher in βint-ν² flies from 14 to 28 days old (Figure 2C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Adult midgut of βint-ν mutants showed shortened and multilayered epithelium.

(A) The whole adult midgut of wild-type flies and βint-ν² homozygotes, stained with DAPI, at 7 and 28 days old. Broken lines were used to measure the midgut lengths. (B) The average of adult midgut length of wild-type flies and βint-ν² homozygotes. The error bars are standard error of means (S.E.M.). P-values were calculated by Student's t test, *p<0.05. (C) The βint-ν² adult midgut length relative to that of wild-type flies. (D) A cross-section view of the adult midgut of wild-type, βint-ν², and control (Act5C>GFP) flies, and a knockdown of βint-ν (Act5C>βint-ν RNAi) flies, stained with Rhodamine-phalloidin (magenta) and DAPI (white).

https://doi.org/10.1371/journal.pone.0089387.g002

Next, we focused on the posterior region of the midgut (PMG) of wild-type and βint-ν² flies at 7, 14, and 28 days old, because substantial shortening was observed in this region (Figure 2A–C). In 7, 14, and 28-day-old wild-type flies, the epithelium constantly showed a monolayer feature (Figure 2D). In 7-day-old βint-ν² flies, the epithelium was also monolayer, but mild and severe abnormal multilayers were observed at 14 and 28 days old, respectively (Figure 2D). When a double strand RNA construct against βint-ν (βint-ν RNAi) was expressed with the Act5C-GAL4 driver, similar defects were induced in 28-day-old flies (Figure 2D). A time period of the multilayered defect was consistent with the number of days when severe shortening of the midgut observed. Thus a multilayered epithelium caused by depletion of βint-ν might be responsible for the shortening of the adult midgut. Also, above results suggested that βint-ν is required for maintaining morphological homeostasis of the adult midgut.

βint-ν in ECs restricts proliferation of ISCs

Some studies have reported that the pathological multilayering of the epithelium was accompanied by an increase in mitotic ISCs in the adult midgut [32]–[34]. Therefore, we speculated that ISC proliferation was also increased in βint-ν² adult midguts. To confirm this possibility, we counted the number of mitotic cells possessing phosphorylated histone H3 (pH3), a marker of M-phase chromosomes, in the whole midgut (Figure 3A–F). In wild-type flies, the average number of pH3-positive cells per midgut was 0.83±0.48, 7.58±2.42, and 19.25±4.92 at 4, 7, and 14 days old, respectively (Figure 3G). For βint-ν² homozygotes, the numbers were higher: 5.29±2.18, 20.55±3.72, and 44.62±7.24 at 4, 7, and 14 days old, respectively (Figure 3G). These averages were statistically different between wild-type and βint-ν² homozygote flies at each age. These results indicated that the normal gene product of βint-ν restricted the number of proliferating ISCs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. A function of βint-ν in ECs was involved in ISC proliferation.

(A–B) The PMG of wild-type (A–C) and βint-ν² homozygote (D–F) flies at 4 (A and D), 7 (B and E), and 14 days old (C and F) stained with anti-pH3 antibody (green) and DAPI (white). Arrows indicate pH3-positive cells. (G) The average number of pH3-positive cells in the whole midgut of wild-type flies and βint-ν² homozygotes at 4, 7, and 14 days old. (H and I) The βint-ν mRNA (magenta) of the PMG epithelium treated with βint-ν RNAi using MyoIA-GAL4 (H-H’’’) and esg-GAL4 (I-I’’’). H and I are hybridization signals (magenta). H’’’ and I’’’ are nuclear staining by DAPI (white). Most of the cells with small nuclei, which are outlined with broken lines (H’, H’’, I’ and I’’), are considered to be ISCs/EBs. The outside of the outlined area is the cytoplasm of EC. The EC-specific and ISC/EB-specific reduction in βint-ν mRNA can be seen in H and I, respectively. (J–L) The average number of pH3-positive cells in the whole midgut treated either with non-RNAi (+), βint-ν RNAi, and mys RNAi for 7 days with MyoIA-GAL4 (J), esg-GAL4 (K), and 24B-GAL4 (L), respectively. (M–O) The PMG with non-RNAi (M), βint-ν RNAi (N), and mys RNAi (O) treatment using MyoIA-GAL4 driver for 7days, stained with anti-pH3 (yellow), anti-GFP (green), anti-Arm, and anti-Pros (magenta) antibodies. (M) As previously reported, MyoIA-GAL4 drove expression of UAS-target (GFP in this Figure) specifically in ECs but neither in ISCs/EBs (marked by Arm localization to cell membrane), nor in ees, (Pros-positive). Right two panels in M are magnifications of boxed area in the left panel. (N and O) Compared with βint-ν RNAi (N), mys RNAi (O) caused more increase of mitotic ISCs detected with anti-pH3 (yellow) antibody. Nuclei were stained with DAPI (white). In G and J–L, the error bars are S.E.M. and P-values were calculated by Student’s t test, *p<0.05.

https://doi.org/10.1371/journal.pone.0089387.g003

The adult midgut was composed of ISCs, EBs, ECs, ees, and VMs (Figure 1A). Next, to confirm which cell types required a function of βint-ν in restricting the number of mitotic ISCs, we performed cell type-specific βint-ν RNAi with a stage-specific forced expression system TARGET [35]. For this analysis, we used the following GAL4 drivers: esg-GAL4 (ISC and EB), MyoIA-GAL4 (EC), and 24B-GAL4 (VM), combined with tub-GAL80^ts, a temperature-sensitive GAL4-repressor GAL80 (Figure 1A) [35]. The βint-ν RNAi induced by esg- and MyoIA-GAL4 caused a decrease of the βint-ν mRNA specifically in ECs and ISCs/EBs, respectively (Figure 3H, and 3I). When esg-GAL4 or 24B-GAL4 was used, we did not detect any statistically significant difference in the number of pH3-positive cells between the control and βint-ν RNAi-treated midguts (Figure 3K and 3L). By contrast, using MyoIA-GAL4 resulted in a statistically significant increase in the number of mitotic ISCs. (Figure 3J). In this condition, any disruption of the basement membrane (BM) marked by Vkg-GFP, a GFP fusion of collagen IV, cannot be detected (Figure S2) [8]. From these results, we concluded that a function of βint-ν in ECs was involved in restricting the number of mitotic ISCs.

βint-ν genetically interacts with Dl/N signalling factors in regulating ISC pool

One possible reason for the increase in the number of mitotic ISCs was expansion of the ISC pool. To confirm this hypothesis, we first observed ISCs specifically marked by an anti-Delta (Dl) antibody in βint-ν² flies. Compared with wild-type, Dl-positive diploid cells increased in βint-ν² homozygotes (Figure 4A and 4B). Furthermore, these ISCs were frequently juxtaposed (Figure 4B and 4B’ insets), which is unusual in wild-type (Figure 4A and 4A’ insets), suggesting the occurrence of symmetric division of ISC. In addition, localization of Dl to the plasma membrane rather than to the cytoplasm was predominantly observed in βint-ν² flies, which was similar to that in the ISC-like tumour in the neuralized mutant where symmetric division occurred exclusively by inactivating Dl/N signalling (Figure 4B and 4B’ arrows in insets compared with 4A and 4A’ arrows in insets) [16]. Next, to recognize ISCs more clearly in the overcrowded epithelial cells of the βint-ν² midgut, we also used the Dl-lacZ strain that expresses β-galactosidase localizing to the nuclei. Consequently, we observed an increase in the ISC number from early stage (7 days old) (Figure 4E and 4E’ compared with 4C and 4C’). Similarly, pairs of ISCs were frequently observed in βint-ν² homozygotes (Figure 4E and 4E’). In addition, a lower angle between the spindle body and BM, which has been speculated to indicate symmetric division of the ISCs, was observed more often in βint-ν² flies than in wild-type flies (Figure S3) [36]. These observations suggested that the frequent duplication of the ISCs caused an expansion of the ISC pool in βint-ν mutant.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Frequent ISC duplication was caused in βint-ν mutant.

(A–B) The PMG of 14-day-old wild-type flies (A and A’) and βint-ν² homozygotes (B and B’) stained with anti-Dl antibody (magenta in A, B, and insets). Dl was prominently observed in punctuates in wild-type flies (arrows in insets of A and A’). The diploid cells of βint-ν² homozygote predominantly showed Dl localization to the plasma membrane (arrows in insets of B and B’). (C–E) The PMG of seven-day-old Dl⁰⁵¹⁵¹/+ (C and C’), βint-ν²/+, Dl⁰⁵¹⁵¹/+ (D and D’), and βint-ν²/βint-ν², Dl⁰⁵¹⁵¹/+ (E and E’) flies stained with anti-β-gal antibody (magenta in C–E) and DAPI (white in C’, D’, and E’). (F) Either transient or ISC clones were labelled by the MARCM system. (G) An example of the PMG showing transient and ISC clones labelled by GFP (green). The ISC clones but not the transient clones contained Dl-positive ISC (magenta). (H) The ratio of transient (blue bar) and ISC (magenta bar) clones generated in control (FRT19A, tub-GAL80, hs-FLP/FRT19A; Act5C-GAL4, UAS-GFP^S65T/+) or βint-ν² (FRT19A, tub-GAL80, hs-FLP/FRT19A; βint-ν²/βint-ν²; Act5C-GAL4, UAS-GFP^S65T/+) midguts. (I) The ISC clone size in control and βint-ν² midguts at 5 days after clone induction. Each dot represents one clone. (J-L) The PMG of N^ts homozygotes (J), βint-ν² homozygotes (K), and N^ts and βint-ν² double homozygotes (L) stained with anti-Pros antibody (magenta). The ees marked by “Pros” increased more in the N^ts – βint-ν² double mutant than in the N^ts or βint-ν² single mutants. Nuclei were stained with DAPI (white in A’-E’, G, and J’-L’).

https://doi.org/10.1371/journal.pone.0089387.g004

To further confirm whether frequency of ISC self-renewal was increased in βint-ν mutants, we generated a mosaic analysis with a repressible cell marker (MARCM, [37]) clone in control and βint-ν mutant midguts. In this analysis, the committed EB or ISC respectively produced a visible transient clone (composed of a single cell) or an ISC clone (composed of multiple cells) (Figure 4F and 4G). As expected, the frequency of ISC clone generation in βint-ν² homozygotes was higher than that in control flies (Figure 4H), indicating that the ratio of ISC self-renewal to total ISC division increased. A slight promotion of ISC clone size was consistently observed in βint-ν² mutants (Figure 4I). The ISC pool expansion was also consistent with the midgut shortening and multilayered epithelia, since N RNAi or Su(H) mutation caused a similar ISC increase and morphological defects [10], [32].

Recently, lateral inhibition by Dl/N was found to be involved in balancing ISC self-renewal and EB commitment rates, which likely affected the ISC population size [38]. We thus tested a genetic interaction between βint-ν² and Dl⁰⁵¹⁵¹ (the above-mentioned Dl-lacZ strain, also known as a loss of function allele) and confirmed a further increase in the ISC population (Figure 4C–E), suggesting a possible relationship between βint-ν and Dl. If N signalling decreased in βint-ν mutants, the ee population was also expected to expand, since strong activation of N signalling in EBs is known to inhibit ee differentiation [16]. Although the homozygotes of N^l1N-ts1, a temperature sensitive allele of N, did not show conspicuous expansion of the ee population for 11-day-old flies cultured under a non-permissive condition (29°C) for N^l1N-ts1, the double homozygous mutants for βint-ν² and N^l1N-ts1 showed its considerable expansion (Figure 4J–L). This indicates that βint-ν genetically interacted with Dl/N signalling factors in choice of cell fate between the ECs and ees as well as between the ISCs and EBs.

Spread of EC misdifferentiation and ISC overproliferation from local to entire region in young βint-ν mutants

In some reports, ISC overproliferation is often associated with induction of misdifferentiated EC-like polyploid cells that are continuously expressing EB markers [12], [13],[16]. To determine whether these misdifferentiated EC-like cells appeared in βint-ν mutants, we examined the expression of ISC and EB markers, i.e. 10×STAT92E-GFP, Su(H)Gbe-lacZ, and a dual-phosphorylated form of ERK (dpERK), that had respectively been used as monitors for JAK/STAT, N, and EGF signalling activities. In control midguts of 7 day-old flies, 10×STAT92E-GFP (strong in ISCs and EBs and weak in ees and VMs), Su(H)Gbe-lacZ (specific to EBs), and dpERK (ISCs, EBs, and a few ECs) were detected as previously reported (Figure 5A) [12], [16], [39]. On the other hand, in all the 7-day-old βint-ν² homozygotes, their atypical expressions in the EC-like polyploid cells were observed although their levels were weaker than their typical expressions in the ISCs and/or EBs (Figure 5B). These aberrant differentiation was randomly induced in patches of cell populations along the midgut at 7 days old, and rapidly spread throughout the young βint-ν² midguts with age (Figure 5C–F). In addition, mitosis and duplication of the ISCs expressing a high level of Dl-lacZ were preferentially observed in these abnormal region at 7 days old (Figure 5G–I). A similar differentiation defect was also induced by βint-ν RNAi in the ECs but not in the ISCs/EBs (Figure. S4). In addition, no obvious differentiation defects were observed in 1-day-old young flies and 96 hours after pupation (Figure. S5). Collectively, these observations suggested that abnormal EC-like polyploid cells, related to the ISC overproliferation, were induced during the young adult stage of βint-ν² flies and that βint-ν was essential for maintaining normal adult midguts but not for developing midguts.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Misdifferentiated EC-like polyploid cells appear in young βint-ν mutants.

(A–B) The PMG epithelial cells of 7-day-old control (A) and βint-ν² (B) flies, marked with 10×STAT92E-GFP (green), Su(H)Gbe-lacZ (cyan), anti-dpERK antibody (magenta), and DAPI staining (white). (C-E) Examples of the PMG of control (Su(H)Gbe-lacZ/+, 10×STAT92E-GFP/+) and βint-ν² (Su(H)Gbe-lacZ/+, βint-ν²/βint-ν², 10×STAT92E-GFP/+) flies showing 10×STAT92E-GFP (green) and Su(H)Gbe-lacZ (cyan). The broken lines and percentages respectively indicate the outline of the region showing misdifferentiated EC-like polyploid cells expressing 10×STAT92E-GFP+Su(H)Gbe-lacZ and the ratio of its area (measuring/calculation is detailed in the “Materials and methods” section). (F) Ratio of the abnormal epithelial area as indicated by broken lines in C–E for each of the control flies and βint-ν² homozygotes at 4, 7, and 11 days old. Each blue and orange circle represents a datum derived from a single PMG. Dotted lines and values indicate averages. (G) The local midgut epithelium with a patch of the misdifferentiated EC-like cells marked by 10×STAT92E-GFP (green) in 7-day-old βint-ν² flies (βint-ν²/βint-ν², 10×STAT-GFP/+). In this region, pH3-positive ISC (magenta arrows) were predominantly contained. (H and I) The PMG of 7-day-old control (10×STAT-GFP/Dl⁰⁵¹⁵¹) and βint-ν² (βint-ν²/βint-ν², 10×STAT-GFP/Dl⁰⁵¹⁵¹) flies. The right panels are magnifications of the boxed areas in the left panels. Magenta staining in H and I represent pH3-positive and Dl-lacZ-positive cells, respectively.

https://doi.org/10.1371/journal.pone.0089387.g005

Local up-regulation of Upds was induced in young βint-ν mutants

In a stressed/damaged condition, Upds and Vn (ligands of JAK/STAT and EGF signalling pathways, respectively) are up-regulated in the ECs and VMs, which non-autonomously promotes ISC proliferation and differentiation [12], [13], [40]. In this study, we found that they were also up-regulated in the ECs/ees and VMs of βint-ν mutants on the basis of the expression of the reporter construct lines upd-lacZ or vn-lacZ (Figure 6A and B). The regions where this up-regulation was found corresponds to those in which the misdifferentiated EC-like cells were appeared (Figure 6A and B). In addition, we detected hyperactivation of JAK/STAT signalling in the VMs that surrounded the misdifferentiated ECs and up-regulated vn-lacZ (Figure 5B brackets). Combining these findings with information in previous reports, we speculated that in βint-ν² homozygotes the Upds secreted from defective EC-like cells promoted a local ISC proliferation and misdifferentiation of ECs by the direct pathway (from ECs into ISCs/EBs) and indirect pathway (from ECs into ISCs via VMs) where JAK/STAT signalling was activated by Upds from the ECs). In the latter case, JAK/STAT in the VM cells was known to induce an expression of vn that promoted ISC proliferation [41].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Induction of upd in epithelium and vn in VMs is initiated around the defective areas.

(A–B) The PMG of 7-day-old βint-ν² homozygotes carrying 10×STAT92E-GFP and upd-lacZ (A) or vn-lacZ (B). (A) The upd-lacZ (magenta in A and A”) was up-regulated in the polyploid EC-like (white arrows in A” and A”’) and the Pros-positive ee (yellow arrows in A” and A”’) cells weakly expressing 10×STAT92E-GFP but not in the cells strongly expressing it (i.e. ISCs/EBs, cyan arrows in A’’ and A’’’). (B) The vn-lacZ (magenta in B) and 10×STAT92E-GFP (green in B’) were up-regulated in the VMs, marked by phalloidin (magenta in B’), which surrounds the defective EC-like cell area (B”’). Brackets (in A, A’ and B-B”’) and squares (in A’’ and A’’’) indicate the regions of defective cells and high magnification (in A and A’), respectively. Nuclei were stained with DAPI (white in A’, A’’’, B”, and B”’). The genotypes in A-A”’ are upd-lacZ/+, βint-ν²/βint-ν², and 10×STAT92E-GFP/+, and those in B-B”’ are βint-ν²/βint-ν² and 10×STAT92E-GFP/vn-lacZ.

https://doi.org/10.1371/journal.pone.0089387.g006

βint-ν mutant phenotypes were suppressed by blockade of JNK signalling

It has been found that JNK signalling, known as a stress/damage sensor, is dependent on the defect levels in the ECs. It promotes ISC proliferation and differentiation to rapidly regenerate the epithelium via interleukin-6-like secreted factors Unpaired (Upds) [19], [42], [43]. It has also been reported that hyperactivation or inactivation of JNK signaling in ISCs/EBs affected proliferation of ISCs that resulted in reducing shortened lifespan [44]. From these reports and our result on βint-ν mutant midguts showing up-regulation of upd-lacZ, we speculated that activation of JNK signalling accompanied the up-regulation of Upds. To monitor JNK signalling activity, we used puc^E69 (puc-lacZ), a lacZ reporter line [45]. As expected, βint-ν² homozygotes showed a high level activation of JNK signalling in the EC-like polyploid cells and Pros-positive ee cells around the regions where the misdifferentiated EC-like cells were induced (Figure 7A and B). In these regions, a local increase in the cell number (Figure 7A”), two layered epithelia (arrows in Figure 7Ba and Bb), and an accumulation of F-actin (Figure 7B’ and B”) were also detected clearly in βint-ν² homozygotes carrying puc^E69. Since puc^E69 is a loss of function allele of puc that encodes a negative regulator of JNK signalling, heterozygosity of this allele results in a slightly high level of JNK signalling. Accordingly, the above phenotypes were probably enhanced when compared with those observed in normal puc condition.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. βint-ν mutant phenotypes were suppressed by inactivation of JNK signalling.

(A-B) The PMG of 7-day-old βint-ν² homozygotes carrying 10×STAT92E-GFP and puc-lacZ (βint-ν²/βint-ν², 10×STAT92E-GFP/puc^E69). Nuclei were stained with DAPI (white). (A) In the region of the defective misdifferentiated ECs (brackets in A-A”), puc-lacZ (magenta) was up-regulated (A and A’) and an increase in cell number was observed (A”). (A”’ and A””) High magnifications of the boxed area in A. The up-regulation of puc-lacZ expression (magenta) was induced in the polyploid EC-like cells (white arrow) and Pros-positive ee cells (yellow arrow) but not in the other diploid cells (cyan arrow). (B-B”) The F-actin (yellow) was accumulated at the outlines of cells showing high level activation of JNK signalling (B’ and B”), in which two-layered epithelium was observed (arrows in Ba and Bb). B” is a high magnification of the boxed area in B’, and Ba and Bb are Z-sections indicated by broken lines in B. (C–D) The PMG of βint-ν² homozygotes with a bsk^DN–expressing mosaic obtained with a FLP-out system (hs-flp/UAS-bsk^DN, βint-ν²/βint-ν², 10×STAT92E-GFP/GAL4-Act5C(FRT.CD2), UAS-RFP). (C) Misdifferentiated cells (with both 10×STAT92E-GFP and polyploid nucleus) and signs of overproliferation (white arrowheads) were observed in the regions without bsk^DN–expressing EC-like polyploid cells (upper left area). (D) No defects were caused in the regions where bsk^DN was expressed in the polyploid ECs. The ISC/EB-like diploid cells, circumscribed by ECs expressing bsk^DN, showed a normal aspect in appearance, irrespective of the presence (green-magenta arrowheads) or absence (green arrowheads) of bsk^DN expression.

https://doi.org/10.1371/journal.pone.0089387.g007

To further examine whether the excess ISC proliferation and EC misdifferentiation in βint-ν² homozygotes were dependent on JNK signalling, we performed mosaic suppression of JNK singalling by forced expression of bsk^DN, a dominant negative form of bsk, in βint-ν² homozygotesby the FLP-out system [46]–[48]. Although signs of overproliferation and misdifferentiation were obvious in the regions without bsk^DN, they were not observed in the region with bsk^DN in most of the ECs (Figure 7C, out of 10 midguts observed). Remarkably, all of the ISC- and EB-like diploid cells with or without bsk^DN showed normal aspects in the regions surrounded by ECs expressing bsk^DN (green and green+magenta arrowheads in Figure 6C and D). This suggested that the overproliferation of ISCs and misdifferentiation of ECs in βint-ν mutants depended on up-regulation of JNK signalling in the ECs.

Local defects in young βint-ν² mutants were not directly related to cell death

Previous reports showed that activation of JNK signalling induced apoptosis in the ECs [42]. Therefore, we performed TUNEL staining for the βint-ν² mutant midguts. Although a slight increase in the dying ECs was detected in the βint-ν² mutants, compared with wild type, at 14 days old (Figure S6A and B), no common cell death processes (e.g. fragmented nuclei, cell blebbing, sloughing of cells into the lumen) were obtained with cross-section analyses (Figure. 2D and data not shown). In addition, no TUNEL-positive cells were observed in the local defective regions with ectopic JAK/STAT activation in the βint-ν² midgut (Figure S6C). Thus the slight increase in cell death observed at 14 day old was probably a secondary effect in the βint-ν² midgut.

βint-ν represses senescence of the adult midgut epithelium

A few studies have reported that βint-ν was involved in the embryonic endoderm migration and phagocytosis for apoptotic cells [29], [30], [49]. In this study, we found a novel function of βint-ν in maintaining homeostasis of the adult midgut epithelium. The adult midgut of βint-ν mutants exhibited a shortened and multilayered epithelium, which became more severe with aging (Figure 1). Deterioration of these defects are probably due to increases in the ISCs and misdifferentiated EC-like cells (Figures 2–4). Genetic manipulation for several signalling pathways, bacterial infections, or spontaneous aging have also been reported to cause excess ISC proliferation and abnormal cell differentiation accompanied by shortening and multilayering of the epithelium. [10], [40], [41]. Thus, maintaining the morphology of the Drosophila adult midgut is considered to be closely linked to proper ISC proliferation and differentiation. These processes are known to be regulated by several signalling pathways including Notch, JAK/STAT, EGF, and JNK signalling. The βint-ν also inhibits misregulation of these signalling pathways induced by hyperactivation of JNK signalling as discussed below (Figure 8A) to maintain normal ISC proliferation and differentiation that ensures homeostasis of the midgut morphology and cell type proportion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. A model for homeostasis disruption in βint-ν mutant.

(A) A schematic diagram of the βint-ν signalling pathway for regulating ISC proliferation and differentiation. βint-ν inhibits up-regulation of JNK signalling in young flies cultured under normal condition. The up-regulation of JNK signalling induced upd transcription as well as F-actin accumulation. Upds in turn promoted ISC proliferation (duplication) and differentiation. In addition, accumulation of F-actin related to activation of JNK signalling. (B) A possible model for progression of homeostasis disruption of the adult midgut epithelium in βint-ν mutants. 1) Locally and weakly stressed/damaged cells are produced. 2) JNK signalling is activated. 3) Upds and F-actin are up-regulated, which promotes ISC proliferation (duplication) and differentiation as shown in D. 4) Retention of JNK–active cells and F-actin-mediated multilayering results in a further increase in Upds. 5) Expansion of epithelium with a high turnover rate from the local defective region (step 4) together with additional local stress/damage (step 1) cause an entire and chronic high level activation of the JNK-Upds pathway for additional defects such as midgut shortening.

https://doi.org/10.1371/journal.pone.0089387.g008

Mys and βint-ν probably have different roles in the adult midgut epithelium

Via integrins, the ISCs adhere to the VMs that provide niche factors such as Wg, Vn, and Dilp3, promoting the ISC proliferation [15], [36], [50]. The other Drosophila integrin β subunit gene, mys, is required for maintenance of germline and follicle stem cells to anchor the niche [25], [26]. Most recently, Goulas et al. reported that mys was also involved in asymmetric ISC division [27]. In addition, mys RNAi in ISCs and ECs respectively resulted in an increase in and no effect on ISC proliferation [27]. However, we obtained a different result, i.e. that mys RNAi in the ISCs/EBs with esg-GAL4 resulted in decreasing the ISCs/EBs (Figure S7A, S7C and 3K) and delaying growth of the ISC-like tumour induced by N RNAi (Figure S7F compared with S7D) as most recently reported by Lin et al [28]. These suggested that, unlike βint-ν, mys was required for maintaining the ISCs. By contrast, mys RNAi in the ECs with MyoIA-GAL4 caused an increase in mitotic ISCs and multilayered epithelia, which were more severe than that caused by βint-ν RNAi (Figure 3J, 3M-O and data not shown). Thus, according to our results, in the ISC proliferation/maintenance, mys played opposite roles between cell types, that is, a role in the ISCs/EBs essential for maintaining the ISCs and a role in the ECs essential for suppressing the ISC proliferation. On the other hand, βint-ν phenotypically showed its function only in the ECs for the ISC proliferation (Figures 2 and 3). Consistently, the βint-ν mutants did not show any alteration of Mys localization in the midgut (Figure S8). In tumour progression of vertebrates, some integrins such as αvβ3 and α6β4 enhance the progression, whereas others such as α5β1 inhibit it [21]. It is thought that one possible reason for these differences is that the expression levels in each integrin are varied with cell types.

Role of JNK activation to suppress midgut scenescence

In young βint-ν mutants, hyperactivation of JNK signalling was induced in the EC-like and ee cells just around the local defects with ISC proliferation and EC misdifferentiation, which were suppressed by inactivation of JNK in the ECs (Figure 7). It has been hypothesized that, in response to local tissue damage, activation of JNK signalling in the ECs non-autonomously promotes ISC proliferation and differentiation via Upds, which also induces vn expression through JAK/STAT activation in the VMs [41]. Consistently with this finding, in our observation, up-regulation of upd in the EC-like and ee cells and vn expression in the VMs also occurred in the defective region (Figure 6). Therefore, we can interpret that the normal βint-ν gene product inhibits hyperactivation of the JNK-Upds pathway, which nonautonomously induces ISC overproliferation and EC misdifferentiation via direct (ECs into ISCs/EBs) and indirect (mediated with VMs) pathways (Figure 8A).

For example, overproliferation, misdifferentiation, and high cell turnover state were provoked together in patches of cell populations in the midgut challenged by Ecc15, a non-lethal strain of the phytobacteria Erwinia carotovora [43]. In this case, clustered apoptotic cells accompanied by highly activated JNK signalling in the ISCs/EBs/ECs were also induced to improve the midgut shrinkage, an acute symptom of bacterial infection [40], [43]. Thus, the local, transient, and high level activation of JNK was important in regenerating the midgut epithelium.In the case of young βint-ν mutants, similar local defects such as high cell turnover and high level (but not transient) JNK activation in the EC-like /ee cells (Figure 5 and 7) were observed, which led us to the idea that the state of βint-ν mutant midgut mimics the stressed/damaged conditions such as infection by bacteria (Figure 8B). However, unlike the case of bacterial infection, the defects in the young βint-ν mutants rapidly spread throughout the midgut and the midgut shortening was never restored later (Figure 2). On the other hand, the normally aged wild type midgut has also been found to show a high cell turnover with chronic high level activation of JNK signalling in the ECs and ees throughout the midgut [10]. Also in aged midgut, esg-GAL4 was induced in misdifferentiated EC-like cells like the ectopic expression of 10×STAT92E-GFP in young βint-ν mutant midgut [44]. In this condition, artificial activation of JNK signalling accelerated overproliferation of ISCs, which accompanied short lifespan [44]. Thus, the symptoms of the βint-ν mutant midgut were more similar to those of an aged midgut than a bacterially infected midgut, and βint-ν probably plays two roles through its inhibiting of JNK. One is suppressing random causes of the local high cell turnover. The other is inhibiting the spread of the defects, which subsequently leads to chronic disorganization of the epithelial cell turnover (Figure. 8B).

In the local defects, accumulation of F-actin, which is related to JNK signalling, was observed at the cell-cell boundaries (Figure 7B). We are considering that this is also related to the local increase of ISC proliferation (data not shown). Additionally, abnormally two-layered epithelia were observed only in the local areas with F-actin accumulation (Figure 7B), suggesting that this leads to inhibition of epithelial elimination or promotion of EC misdifferentiation. Furthermore, Hippo signalling might be involved in mediating JNK signalling activity and F-actin dynamics in the midgut defects of βint-ν mutant, since the Hippo pathway functions as a known sensor of cell adhesion and cytoskeletal integrity in the ECs to regulate ISC proliferation and promote it to a stress-response condition via Upds [18], [19], [34], [51]–[53].

Fate asymmetry of the ISC offsprings, ISC self-renewal, and EB commitment are required for ensuring a constant ISC pool. The fate choice is known to be regulated through lateral inhibition by Dl/N signalling between the neighboring ISC siblings, and its ratio is also affected by proximate differentiated cells such as the damaged ECs [38]. Our study indicated that frequent ISC duplication was induced in βint-ν mutants (Figures 4 and 5), which suggested that the normal βint-ν gene product in the ECs nonautonomously affected Dl/N interaction between the ISCs through the JNK-Upds pathway (Figure 8A). Furthermore, the high level of Dl-lacZ expression and change in D1 localization observed in βint-ν mutants suggested that the expression level and/or D1 localization might be involved in the fate choice of the ISC offsprings (Figure 4).In previous reports, aging and hyperactivation of JNK signalling also affected N signalling and ISCs proliferation [10], [44]. Future studies should provide more understanding of the molecular mechanisms for switching between fate symmetry/asymmetry of the ISC offsprings in response to various midgut conditions.

Integrins are possibly involved in age-related disease caused by disruption of tissue homeostasis

In young βint-ν mutants, hyperactivity of JNK-Upds signalling led to chronic hyperplasia of the ISCs throughout the midgut. As discussed above, this symptom was similar to that of aged phenotypes of the midgut [10], [54], [55]. Therefore, we speculated that βint-ν probably inhibits premature aging of the Drosophila adult midgut. There is much evidence suggesting that aging creates conditions conducive to the incidence of neoplastic diseases such as cancer [56]. In mammals, the cytokine interleukin-6 and its downstream effector STAT3, a homologous pathway of the Drosophila Upds/JAK/STAT signalling, are involved in inducing intestinal inflammation and cancer [57], [58]. Together with Wnt signalling, JNK signalling activation also leads to tumour formation in the vertebrate intestine [59]. Furthermore, moderate activation of JNK resulted in an increased tolerance to stresses and extended lifespan but its excessive or chronic activation was implicated in the occurrence of pathologies including cancer [60]. Although the relationship between integrin dysfunction and progression of aging in mammals remains unclear, our results indicate that the chronic activation of JNK and subsequent elevation of cytokines caused by integrin dysfunction were likely involved in inducing or affecting age-related diseases. These findings provide an insight into a possible etiology of age-related diseases such as colorectal carcinoma that may be affected by dysfunction of the integrin function as seen in the above-mentioned case of integrin α5β1 [21].

Fly strains

We used w¹¹¹⁸ as a wild-type strain, as well as βint-ν², which is a null allele of βint-ν [29]. The following mutant and reporter lines were used: N ^1N-ts1 (a temperature sensitive allele of N [61], Dl⁰⁵¹⁵¹ (Dl-lacZ), PD (upd-lacZ), vn¹⁰⁵⁶⁷ (vn-lacZ), puc^E69 (puc-lacZ), and esg^k00606 (esg-lacZ) (lacZ enhancer trap lines of Delta, unpaired, vein, puckered, and escargot, respectively [10], [45], [62]–[65], Su(H)Gbe-lacZ and 10×STAT92E-GFP (reporter lines of Notch and JAK/STAT signalling activity, respectively [66], [67], and vkg-GFP (a protein trap line expressing green fluorescent protein (GFP) fused to Viking, the Drosophila Collagen IV [8]). The GAL4 lines used were NP6267 (esg-GAL4) [68], NP1 (MyoIA-GAL4) [12], how^24B (24B-GAL4) [69], NP0319 (103585), Ay-GAL4 [70], and Act5C-GAL4 (FRT.CD2) [71]. The following UAS strains were used: UAS-GFP^S65Tand UAS-RFP.nls obtained from Bloomington Drosophila Stock Center (BDSC), UAS-bsk^DN (a dominant negative form of basket, the Drosophila JNK [46]), UAS-βint-ν RNAi (#40895 from Vienna Drosophila RNAi Center (VDRC)), UAS-mys RNAi (#29619 from VDRC), and UAS-N RNAi (#100002 from VDRC). The hs-FLP provides FLP recombinase under the heat-shocked condition. The tub-GAL80^ts ubiquitously overexpresses a temperature-sensitive GAL80 protein [35]. We obtained the tub-GAL80, hs-FLP, w^*, FRT19A and the FRT19A strains used for MARCM analysis from BDSC.

Flies were cultured in standard medium and at appropriate temperature and female individuals were observed in all the experiments in this study.

Immunostaining and in situ hybridization

A dissected adult midgut at an appropriate adult stage was fixed with 4% paraformaldehyde. After it was washed with phosphate buffered saline containing TritonX-100, immunostaining was performed with the following primary antibodies: rat anti-βint-ν (1∶200) [30], rabbit anti-βint-ν (1∶200) [24], rat anti-GFP (Nacalai Tesque, 1∶200), rabbit anti-RFP (Clontech, 1∶200), chick anti-β-galactosidase (β-gal) (abcam, 1∶200), rabbit anti-phospho-histone H3 (pH3) (Upstate Biotech, 1∶200), mouse anti-dpERK (Invitrogen), mouse anti-γ-tubulin (GeneTex, 1∶100), mouse anti-Prospero (DSHB, 1∶100), mouse anti-Armadillo (DSHB, 1∶50), mouse anti-Myospheroid (DSHB, 1∶100), and mouse anti-Delta (DSHB, 1∶50). Secondary antibodies used were Cy3- or DyLight649-conjugated anti-mouse IgG (Jackson ImmunoResearch, 1∶200), Alexa488-conjugated anti-rat IgG (Jackson Immuno Research, 1∶200), Cy2- or Alexa Fluor555-conjugated anti-rabbit IgG (Jackson ImmunoResearch, 1∶200), Alexa647-conjugated anti-rabbit IgG (Jackson ImmunoResearch, 1∶200), and DyLight649-conjugated anti-chick IgY (Jackson ImmunoResearch, 1∶200). Rhodamine-conjugated and Alexa Fluor 647-conjugated phalloidin (Molecular Probe, 1∶100) were used to stain F-actin. Nuclei were stained with 4′6-diamidino-2-phenylindole (DAPI, SIGMA) or TOTO-3 (molecular probes).

In situ hybridization was performed by standard protocol with anti-Digoxygenin labelled with alkaline-phosphatase (Roche, 1∶7000) and Fast Red (Roche). We used cDNAs of βint-ν as templates to perform antisense and sense RNA probe synthesis with a DIG RNA labelling kit (Roche).

The stained midgut was mounted in 80% glycerol and analysed with Nikon Digital Eclipse C1 and C1Si confocal microscopes (Nikon). The images were processed with EZ-C1 3.90 Free Viewer software (Nikon).

TUNEL assay

The dissected adult midgut was fixed with 4% paraformaldehyde and cell death was detected with an ApopTag Kit (Millipore Corporation, Billerica, MA, USA).

Experiments with temperature shift

When knockdown and overexpression experiments with a TARGET system [35] were performed, adult flies of a given genotype that were carrying tub-GAL80^ts were raised at 18°C, a permissive temperature. One-day-old flies were placed and cultured in a non-permissive condition (29°C) for an appropriate period before dissection.

We used the FLPout technique [47], [70] to raise and culture other adult flies at 18°C until they were 7 days old. These flies were heat-shocked at 37°C for 20 minutes and cultured at 25°C for an appropriate number of days.

We raised N ^1N-ts1/N ^1N-ts1, βint-ν²/βint-ν², and N ^1N-ts1/N ^1N-ts1, βint-ν²/βint-ν² females at 18°C until eclosion and cultured them at 29°C for 11 days.

In MARCM clone generation [37], control (FRT19A, tub-GAL80, hs-FLP/FRT19A, Act5C-GAL4, UAS-GFP^S65T/+) and βint-ν² (FRT19A, tub-GAL80, hs-FLP/FRT19A, βint-ν²/βint-ν², Act5C-GAL4, UAS-GFP^S65T/+) flies were raised and cultured at 25°C until they were 7 days old, heat-shocked 4 times at 37°C for 2 hours and additionally cultured for 5 days.

Measurements, cell counts, and statistical analysis

We dissected midguts out from five wild-type flies and five βint-ν² homozygote flies, at each age, in order to measure their lengths. After DAPI staining, we used confocal microscopes to obtain images of the whole midguts, then used ImageJ (NIH) software to process and measure them.

After preparing the whole stained midguts with an anti-pH3 antibody, a fluorescence microscope was used to count the pH3-positive cells (mitotic cells) in them.

To measure the angle between spindle and basal surface, 14 (wild-type) and 20 (βint-ν² homozygote) 7-day-old mitotic cells from prometaphase to anaphase in the posterior midgut (PMG) were stained with anti-γ-tublin and anti-pH3 antibodies and phalloidin.

To calculate the ratio of areas showing misdifferentiated cells (polyploid nuclei, 10×STAT92E-GFP, and Su(H)Gbe-lacZ-positive EC) in the PMG, control (Su(H)Gbe-lacZ/+, 10×STAT92E-GFP/+) and βint-ν² homozygote (Su(H)Gbe-lacZ/+, βint-ν²/βint-ν², 10×STAT92E-GFP/+) flies at each age were dissected, then stained with anti-GFP and anti-β-gal antibodies. One-side images of them were then obtained with the 20×objective lens of a confocal microscope, with a focus on the boundary between the midgut and hindgut, as shown in Fig. 4C–E. The total differentiation defect area was measured with ImageJ (NIH) software. The average ± S.D. of the total area obtained was 94653±17209dpi (control at 4 days old, n = 10), 103802±22616dpi (control at 7 days old, n = 11), 135312±32651dpi (control at 11 days old, n = 10), 104397±12674dpi (βint-ν² at 4 days old, n = 11), 90862±33413dpi (βint-ν² at 7 days old, n = 11), and 105521±17078dpi (βint-ν² at 11 days old, n = 12) with resolution set to 1024×1024 dpi.

The Student's t-test was used to calculate P-values as needed. We also used the following software: EZ-C1 3.90 FreeViewer (Nikon), Photoshop (Adobe), and Illustrator (Adobe).

Phylogenetic analysis

The predicted amino acid sequences encoded by the other Drosophila species's βint-ν and mys orthologous genes that were derived from the 12 Drosophila genome projects [72], [73] were retrieved from GenBank. However, the D. simulans and D. persimilis genes were not analyzed since the sequence data of C-terminal region including highly-conserved transmembrane domain of the D simulans's βint-ν and D. persimilis's mys were not obtained from the database even with BLAST analysis. Using the available predicted amino acid sequences, we constructed a multiple alignment by the program MAFFT (http://www.genome.jp/) using the default parameters. Based on this alignment, the phylogenetic tree was conducted using MEGA6 program by Neighbor-joining method. The bootstrap test was performed with 1000 replicates. The result obviously indicated the presence of two distinct group, βint-ν and mys (Figure S1).